Skip to contents

Drug-specific validation vignettes (one per published model) are published on the package website at https://nlmixr2.github.io/nlmixr2lib/articles/. The name column links to the vignette when one exists for that model. Models are grouped by their location in inst/modeldb/: general PK/PD templates first, then specific drug models, then models curated from the DDMoRe Foundation library.

General

name description
A1pi (Tortorici 2017) Empirical disease-progression model of intravenous alpha-1 proteinase inhibitor (A1-PI) augmentation therapy in alpha-1 antitrypsin deficiency (Tortorici 2017). Combines a sequential dose-exposure regression that predicts the per-subject median trough serum A1-PI from average dose rate (mg/day, supplied as DOSE covariate), baseline body weight, and baseline endogenous A1-PI, with a piecewise-linear-in-time exposure-response model that predicts CT lung density at total lung capacity from the dose-derived A1-PI exposure and baseline FEV1, with a slope transition at 720 days separating the RAPID-RCT and RAPID-OLE study phases.
AAA (Sherer 2012) Hierarchical Bayesian disease-progression model of abdominal aortic aneurysm (AAA) diameter in men with small (30-49 mm) AAA identified during ultrasound screening (Sherer 2012). The expected AAA diameter follows the ODE dY/dt = beta1 + beta2 * (Y - beta0) with Y(0) = beta0, so the growth rate changes at a constant rate with change in size; the three individual-level parameters (baseline size beta0, baseline growth rate beta1, and constant first derivative of growth rate with size beta2) are drawn from a multivariate normal distribution with a full 3x3 covariance. Covariate effects: baseline AAA diameter on all three parameters, log10 plasma D-dimer on beta1 and beta2, and a diabetes-mellitus binary on beta2. Disease-progression model with no drug dosing.
ABPM (Sheng 2013) Cyclic-fluctuation (circadian rhythm) model for 24-h ambulatory blood pressure monitoring (24-h ABPM) in Chinese patients with mild-to-moderate essential hypertension during the placebo run-in period of four antihypertensive drug clinical trials (Sheng 2013). Predicts systolic blood pressure (SBP) and diastolic blood pressure (DBP) simultaneously as the sum of (a) a rhythm-adjusted 24-h mean and (b) two cosine harmonics with periods 0.5 day (12-h harmonic) and 1 day (24-h harmonic). The two phase-shift parameters PHS1 (12-h harmonic) and PHS2 (24-h harmonic) are shared across SBP and DBP because the authors found the per-output phase-shift estimates were close enough to combine; the four amplitudes (one per output x harmonic) and the two baselines (one per output) remain output-specific. No drug input and no compartments – a baseline-BP circadian model intended for combination with a drug PK/PD model when simulating antihypertensive trials.
Albumin (Bisaso 2014) Semi-mechanistic disease-progression model for plasma albumin in Ugandan adults co-infected with HIV (on efavirenz-based ART) with or without active TB (on rifampicin-based anti-TB co-treatment). Indirect-response form with zero-order hepatic albumin secretion that follows a Verhulst logistic transition from baseline rate Q0 to steady-state rate Qss, and first-order plasma elimination at a fixed literature rate.
Alzheimer (Delor 2013) Disease-progression model (no drug input) for Alzheimer’s disease (AD) progression on the Clinical Dementia Rating scale - Sum of Boxes (CDR-SOB, 0-18 score) over time, fit by Delor et al. (2013) to 2,700 CDR-SOB observations from 380 mild cognitive impairment (MCI) plus 180 AD patients in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database with up to 4 years of follow-up. The model embeds an individual disease-onset time (DOT) and a logit-domain disease trajectory A(1) with a smooth-step activation (T^30 / (DOT^30 + T^30)) that turns the disease progression on at the subject’s DOT; a two-component mixture model assigns each subject to either a fast-progression branch (rate plus accelerating term alphaA(1)) or a slow-progression branch (rate alone, alpha = 0). A study-entry ‘placebo’ term PL(1 - exp(-KPL*(t - T_ENTRY))) captures an early drop or delay observed after enrollment. Baseline CDR-SOB and ADAS-cog explain most of the DOT variability; baseline SCORE_MMSE modifies alpha; baseline CDR-SOB, SCORE_FAQ, and normalized hippocampal volume (SCORE_RHPNM) modify the mixture probability. The published mixture probability and the residual-error scale-domain choice are clarified in the validation vignette’s Assumptions and deviations section. No drug input is dosed in this model.
Asp8232 (Hoefman 2021) Mechanism-based exposure-response (PD-only) model for the vascular adhesion protein-1 (VAP-1) inhibitor ASP8232 in adults with diabetic kidney disease. Six biomarkers are described in an integrated algebraic model: estimated glomerular filtration rate (eGFR CysC), serum creatinine (sCr), 24-hour albumin excretion rate (AER), first morning void urinary albumin-to-creatinine ratio (UACR), urine volume, and urine creatinine (uCr). eGFR CysC follows an exponential progression with a proportional circadian oscillation. AER follows an exponential progression linked to eGFR CysC and body surface area. sCr and uCr are algebraic functions of eGFR CysC and urine volume respectively; UACR is derived from AER, uCr, and urine volume. Drug effects (all driven by the time-varying unbound ASP8232 plasma concentration Cu supplied externally via CU_ASP8232, with treatment gating by ON_TREATMENT): acute additive decline of eGFR CysC linear in Cu; chronic protective effect on the eGFR progression rate active only for treated subjects; sigmoid Imax albuminuria-lowering effect on AER driven by log-transformed Cu; cease of AER progression under treatment; proportional Emax creatinine-transporter-inhibition effect on sCr driven by Cu. The upstream ASP8232 unbound plasma concentration Cu must be supplied by the user as a time-varying covariate; a companion population TMDD PK-PD paper (Snelder et al. 2021, doi:10.1007/s10928-020-09717-w) provides the Cu(t) source at steady state Cu = 125.58 nM for 40 mg qd oral ASP8232 in a typical DKD subject.
AtopicDermatitis qsp (Miyano 2022) QSP + MBMA. Systems-level mathematical model of atopic dermatitis (AD) pathogenesis with drug-effect modules for nine biologics (dupilumab, lebrikizumab, tralokinumab, secukinumab, fezakinumab, nemolizumab, tezepelumab, GBR 830, and recombinant IFN-gamma). Fourteen ODEs describe skin barrier integrity, infiltrated pathogens, four helper T-cell subsets (Th1, Th2, Th17, Th22), seven cytokines (IL-4, IL-13, IL-17A, IL-22, IL-31, IFN-gamma, TSLP), and OX40L; the EASI efficacy endpoint is an algebraic function of skin barrier and pathogens. 51 log-normally distributed patient-specific parameters define virtual-patient heterogeneity; model-based meta-analysis across 9 drug trials tuned the 51 mu and 51 sigma distribution parameters. Drug effects enter as multiplicative inhibition of the effective cytokine concentration (r_inhibit = 0.99 for antibodies, or 0.99 x e_a2 = 0.436 for tralokinumab) or as an additive rIFNg concentration (c_rifng = 210). Time in weeks.
Baicalein (Xiang 2018) In vitro (RAW264.7 mouse macrophage cell line). Semi-mechanism-based cellular pharmacodynamic model for the anti-inflammatory effect of baicalein on LPS-induced cytokine and iNOS/NO release in RAW264.7 mouse macrophages. Four indirect-response states arranged in a TNF-alpha -> {IL-6, iNOS -> NO} cascade: (1) TNF-alpha indirect response with LPS-stimulated zero-order production and baicalein’s log-linear inhibition f(Bai) = alpha * log(C_Bai + 1) on the production rate; (2) IL-6 indirect response with delayed TNF-alpha drive (lag tau1); (3) iNOS indirect response with delayed TNF-alpha drive (lag tau2) and elimination held at zero (paper choice for the post-12.5 h plateau); (4) NO indirect response with iNOS^delta amplification and elimination held at zero. Baicalein concentration enters as a static covariate (CONC_BAI_UM); LPS is constant at 1 ug/mL throughout the experiment and is absorbed into the kinTNF zero-order production rate (no explicit LPS state). Tau1 and tau2 are encoded via single-compartment delay states (mean transit time = tau) because rxode2 does not provide a native delay-differential-equation solver; see the validation vignette Errata for the impulse-response implications. Typical-value-only mechanism: no IIV, no residual error is reported in the source (CV% in Table 1 are RSE / precision of estimate, not BSV).
Bmd fracture (Plan 2012) Bayesian joint disease-progression model linking postmenopausal bone mineral density (BMD) decline to a repeated time-to-fracture hazard, fit by Plan et al. (PAGE 2012 conference poster) to the 2005-2008 NHANES dataset (1605 postmenopausal women, 204 fracture events total; one femoral-neck BMD measure per subject and 0-5 fracture events each). The BMD time-course is piecewise linear in years since final menstrual period (FMP), with three annualized percent-change slopes: menopause transition (t in [-1, 2] yr), early postmenopausal (t in [2, 5] yr), and long-term post-FMP (t > 5 yr). The fracture hazard is exponential (Weibull shape alpha = 1 in the final model) with a BMD-centred log-linear modulation h(t) = exp(theta_h * (1 + theta_bmd * (BMD(t) - 0.8))). The cumulative-hazard ODE d/dt(cumhaz) <- h accumulates from t = 0 (FMP), and the survival probability sur = exp(-cumhaz) is exposed as a derived output. BMI, NHANES ethnicity, and age at FMP were carried as ‘centered covariate effects on all parameters’ in the source Bayesian fit but the per-parameter coefficient magnitudes are not reported in the poster; they are recorded in covariatesDataExcluded for provenance. See validation vignette Errata for the slope-unit interpretation: the printed source equation BMD(t) = b + sum(slope * piece) is additive in linear g/cm^2 only if the slopes are interpreted as %/yr-of-baseline (the Greendale 2012 convention cited as reference [5] in the poster); the model() block does the /100 conversion explicitly.
Bmi (Oniki 2018) Population prediction model for body mass index (BMI, kg/m^2) in 341 elderly Japanese health-screening participants (Oniki 2018). BMI is parameterised as the typical value at age 70.8 years for a male non-T/T reference subject, multiplied by a power-of-age scalar and a female sex multiplier, with an additive shift for carriers of the DsbA-L (GSTK1) rs1917760 -1308G>T T/T genotype (Eq. 1 of the source paper). Log-normal between-subject variability acts on the typical BMI, and log-normal (exponential) residual error is added on the linear scale. The fit was performed with NONMEM 7.2.0 $PRED METHOD=COND INTER (s010 control stream); MINIMIZATION SUCCESSFUL, OBJ 1454.644. There is no drug input. Companion BMI-driven NAFLD-risk model in Oniki_2018_nafld_risk. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Netterberg_2018_breast_cancer_FN.html">Breast cancer FN biomarkers (Netterberg 2018)</a> </td> <td style="text-align:left;"> Joint turnover model for serum interleukin-6 (IL-6) and C-reactive protein (CRP) in early breast cancer patients (n=49) receiving adjuvant FEC and docetaxel chemotherapy. Each biomarker follows a first-order indirect-response form (d/dt(BioM) = Rin_BioM * (1 + g(t)) - kout_BioM * BioM) with the Netterberg 2018 Eq. 1 Lorentzian surge function g(t) = SA / ((t - PT)/SW)^4 + 1 driving the post-dose elevation. CRP production is additionally stimulated by the relative change from baseline of IL-6 (RCFB_IL6) through a linear Slope term. Surge activation in each cycle is gated by the binary covariates MIX_ELEV_IL6 / MIX_ELEV_CRP, whose probabilities Pelevation,IL-6 (63.4%) and Pelevation,CRP (44.3%) are the NONMEM mixture-model frequencies the paper estimated. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Netterberg_2018_breast_cancer_FN.html">Breast cancer FN tte atFN (Netterberg 2018)</a> </td> <td style="text-align:left;"> Parametric time-to-event (TTE) submodel for febrile neutropenia (FN) in early breast cancer patients (n=49) receiving adjuvant FEC and docetaxel chemotherapy: the 'when-FN-occurs' variant whose hazard depends on the instantaneous log of normalized CRP. The hazard form is h(t) = h0 * exp(beta4 * LN_CRP(t)) where LN_CRP(t) = log(CRP(t) / CRP0) (Netterberg 2018 Eq. 3c). The underlying joint IL-6 / CRP biomarker turnover model (see Netterberg_2018_breast_cancer_FN_biomarkers) is embedded inline so this submodel can be simulated standalone. The hazard is fixed to 0 for t < 84 h (3.5 days) per the paper. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Netterberg_2018_breast_cancer_FN.html">Breast cancer FN tte prechemo (Netterberg 2018)</a> </td> <td style="text-align:left;"> Parametric time-to-event (TTE) submodel for febrile neutropenia (FN) in early breast cancer patients (n=49) receiving adjuvant FEC and docetaxel chemotherapy: the 'prior-to-chemotherapy' variant whose hazard depends only on baseline-available covariates. Time-constant exponential baseline hazard h0 with a log-linear age effect h(t) = h0 * exp(beta1 * (AGE - 54)) (Netterberg 2018 Eq. 3a). The hazard is fixed to 0 for t < 84 h (3.5 days) per the paper: no patient experienced grade 3 neutropenia before that time. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Netterberg_2018_breast_cancer_FN.html">Breast cancer FN tte preFN (Netterberg 2018)</a> </td> <td style="text-align:left;"> Parametric time-to-event (TTE) submodel for febrile neutropenia (FN) in early breast cancer patients (n=49) receiving adjuvant FEC and docetaxel chemotherapy: the 'prior-to-FN' variant whose hazard combines a baseline-available baseline ANC (NEUT) effect and a time-varying IL-6 effect routed through a first-order effect compartment. The hazard form is h(t) = h0 * exp(beta2 * CE_LN_IL6(t) + beta3 * (NEUT - 3.53)) (Netterberg 2018 Eq. 3b). The underlying joint IL-6 / CRP biomarker turnover model (see Netterberg_2018_breast_cancer_FN_biomarkers) is embedded inline so this submodel can be simulated standalone. The hazard is fixed to 0 for t < 84 h (3.5 days) per the paper. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/EudyByrne_2021_buprenorphine.html">Buprenorphine (EudyByrne 2021)</a> </td> <td style="text-align:left;"> Indirect-response pharmacodynamic model of sublingual buprenorphine effect on MOTHER NAS severity scores in neonates with neonatal opioid withdrawal syndrome (NOWS). Natural postnatal-age-driven withdrawal decay (NOWSMAX, NOWSM) sets a time-varying baseline NAS score via Kin * (1 + NOWST); buprenorphine central-compartment concentration Cbuprenorphine stimulates NAS-score elimination (Kout) through a Hill Emax term (EMAX * C / (EC50 + C) + 1). The model has NO PK ODE; the buprenorphine concentration Cbuprenorphine (ng/mL) is a required time-varying input covariate carried in the event table. Upstream buprenorphine PK from Moore et al. 2018 (Clin Pharmacol Ther 103:1029-1037; DOI 10.1002/cpt.1064) is not packaged in nlmixr2lib at extraction time; users needing to simulate the PD from a dose regimen must supply concentrations from an external PK source. The initial NAS score is set to its drug-free quasi-steady state, NOWS0 = Kin * (1 + NOWST) / Kout, at the baseline PNA. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Walsh_2024_buprenorphine_opioidBlockade.html">Buprenorphine cows (Walsh 2024)</a> </td> <td style="text-align:left;"> Bounded-integer maximum-inhibition (Imax) exposure-response model of the Clinical Opiate Withdrawal Scale (COWS) total score under subcutaneous long-acting buprenorphine (CAM2038 Q1W) in non-treatment-seeking adults with moderate-to-severe opioid use disorder. The driver is the time-matched buprenorphine plasma concentration CP_BPN_NGML. The COWS total score (0-44, 45 ordered categories) is modelled with the bounded-integer approach of Ueckert (2021) in the numerically-stable implementation the authors cite: the prediction lives on the scale of the standard-normal quantile function and is mapped onto the score range by the normal CDF. Structure (Supplementary Methods, 'Model development: COWS score', third and final variant): IPRED(Cp) = BASEb.int - (BASEb.int - LBASEb.int) * Cp^gamma / (IC50^gamma + Cp^gamma), with gamma fixed to 1 and IC90 estimated in place of IC50 (IC50 = IC90 / 9 for gamma = 1). The drug effect therefore drags the latent score from a pre-treatment baseline down to a lower asymptote rather than to zero. Back-transforming the two asymptotes reproduces clinically sensible scores: 45 * pnorm(-0.887) = 8.4 (mild withdrawal, pre-treatment) and 45 * pnorm(-2.11) = 0.78 (essentially no withdrawal, fully blocked). Encoded here as probitNorm(addSd, 0, 45), the continuous relaxation of the bounded-integer likelihood -- the integer flooring step is not applied (see vignette Errata). The model has NO PK ODE and no compartments: it is a static algebraic exposure-response relationship, and CP_BPN_NGML is a required time-varying input covariate supplied per observation record. The CAM2038 popPK model is not reported in this paper and is not packaged in nlmixr2lib at extraction time. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Walsh_2024_buprenorphine_opioidBlockade.html">Buprenorphine desireToUse (Walsh 2024)</a> </td> <td style="text-align:left;"> Maximum-inhibition (Imax) exposure-response model of the desire to use visual analog scale (VAS) score under subcutaneous long-acting buprenorphine (CAM2038 Q1W) in non-treatment-seeking adults with moderate-to-severe opioid use disorder. The endpoint is the pre-hydromorphone-challenge desire to use score on a 100 mm unipolar VAS; the driver is the time-matched buprenorphine plasma concentration CP_BPN_NGML. Structure (Supplementary Methods, 'Model development: Desire to use VAS score'): IPRED = BASE * (1 - Imax * Cp^gamma / (IC50^gamma + Cp^gamma)), with gamma fixed to 1 and the typical Imax fixed to 1. Unlike the sister drug-liking model this one is NOT direct: a delay in the onset of effect was observed in many participants and is modelled by making Imax an exponential function of time since treatment start, Imax = TVImax * (1 - exp(-kt * time)), parameterised through an onset half-life of 0.330 day. The baseline is logit-transformed onto the range -1 to 101 and its IIV carries a Box-Cox transformation (shape -12.9) to capture the heavily skewed baseline distribution. Predictions are logit-transformed onto the same -1 to 101 range with an additive residual error on that logit scale, encoded as logitNorm(addSd, -1, 101). The model has NO PK ODE and no compartments: it is a static algebraic exposure-response relationship in Cp with an explicit time term, and CP_BPN_NGML is a required time-varying input covariate supplied per observation record. The CAM2038 popPK model is not reported in this paper and is not packaged in nlmixr2lib at extraction time. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Walsh_2024_buprenorphine_opioidBlockade.html">Buprenorphine drugLiking (Walsh 2024)</a> </td> <td style="text-align:left;"> Direct maximum-inhibition (Imax) exposure-response model of opioid drug-liking blockade by subcutaneous long-acting buprenorphine (CAM2038 Q1W) in non-treatment-seeking adults with moderate-to-severe opioid use disorder. The endpoint is the period-corrected drug liking maximum-effect (Emax) bipolar visual analog scale (VAS) score measured after an intramuscular hydromorphone 18 mg challenge; the driver is the time-matched buprenorphine plasma concentration CP_BPN_NGML sampled 60 min before each challenge. Structure (paper Methods equation): VAS(Cp) = BASE * (1 - Imax * Cp^gamma / (IC50^gamma + Cp^gamma)), with Imax and gamma both fixed to 1, so the relationship reduces to VAS = BASE * IC50 / (IC50 + Cp) and IC90 = 9 * IC50 = 0.675 ng/mL (reproducing the paper's independently reported IC90). Predictions are logit-transformed onto the range -1 to 52 with an additive residual error on that logit scale, encoded as logitNorm(addSd, -1, 52) (Supplementary Methods, RUV equations). The model has NO PK ODE and no compartments: it is a static algebraic exposure-response relationship, and CP_BPN_NGML is a required time-varying input covariate supplied per observation record. Users who need to drive it from a CAM2038 dose regimen must supply buprenorphine concentrations from an external PK source; the CAM2038 popPK model is not reported in this paper and is not packaged in nlmixr2lib at extraction time. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fu_2023_cardiovascular.html">Cardiovascular qsp (Fu 2023)</a> </td> <td style="text-align:left;"> QSP (rat). Cardiovascular systems (CVS) hemodynamic model developed by Snelder et al. (2013, 2014) and re-used as a fixed system by Fu et al. (2023) for a stochastic-simulation and re-estimation (SSE) identifiability analysis. The model links five interdependent hemodynamic variables in rats -- heart rate (HR, beats/min), stroke volume (SV, mL/beat), total peripheral resistance (TPR, mmHg*min/mL), cardiac output (CO = HR*SV, mL/min) and mean arterial pressure (MAP = CO*TPR, mmHg) -- through three coupled indirect-response ODEs on HR, SV-turnover (SVT), and TPR, a shared MAP-negative-feedback term on the production rate of each state, cosine circadian rhythms on HR and TPR, and a paper-specific direct HR-on-SV coupling (SV = SVT*(1 - HRSV*log(HR/BSLHR))). Drug PK is a hypothetical one-compartment IV-bolus model with k = 0.173/h (half-life 4 h). In the on-disk NONMEM control stream, the drug effect (Emax model on the drug amount) acts on HR-production, matching the mode-of-action-on-HR scenario of Fu 2023 Section 2.3.3; the vignette shows how to relocate the effect to SV or TPR for the SSE identifiability sweep. All 12 system-specific parameters (baselines, k_out rate constants, MAP-feedback constant, HR-on-SV coupling constant, four circadian rhythm parameters) and all 3 IIV variances (on the three baselines) are FIXED to the Snelder-derived rat estimates transcribed in Fu 2023 Supplemental Materials S1. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Minucci_2024_CART_qsp.html">CART qsp (Minucci 2024)</a> </td> <td style="text-align:left;"> QSP (cellular kinetic / pharmacodynamic). Multi-scale semi-mechanistic model for CD19-targeted CAR T-cell therapy of B-cell non-Hodgkin lymphoma. Six T-cell state compartments (infused, effector, memory) for CD8+ and CD4+ phenotypes plus tumor / B cells and endogenous lymphocytes. Effective activation and killing rates are proportional to the CAR fraction bound to CD19, computed as a Hill-type function of tumor burden under the paper's constant-receptor-density assumption. Per-patient parameters (body weight, CD8 fraction of drug product, initial tumor burden, number of activated-T-cell divisions, memory-cell fraction) come from the fits to n=13 relapsed/refractory NHL patients in Ying et al. 2021. Calibrated to CAR T CK, B-cell aplasia, and structural drug-product characteristics. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Venisse_2008_candida_albicans.html">Caspofungin (Venisse 2008)</a> </td> <td style="text-align:left;"> In vitro (Candida albicans, ATCC 3153). Mechanism-based PK-PD model of caspofungin fungicidal activity in a dynamic broth-renewal flask. Candida population dynamics follow Mouton-type saturable exponential growth with a maximum carrying capacity Nmax minus a first-order natural-death rate Kd minus a fixed broth-renewal rate Ke (set by the experimenters to 0.231 1/h, equal to the drug 1-compartment kel and yielding a 3 h elimination half-life). Caspofungin adds a sigmoid-Emax (Hill=1) Emax * C / (EC50 + C) death-stimulation term to the natural Kd, producing the concentration-dependent CFU decay followed by regrowth seen in Figure 3B. Drug concentration is the central/Vc concentration from a one-compartment PK with V and CL fixed to the apparatus geometry (V = 0.4 L bulk volume of the glass flask, CL = 1.54 mL/min consistent with t1/2 = 3 h). Candida growth/death/Nmax parameters were estimated from all six experiments (3 fluconazole + 3 caspofungin) in a single Nonmem run, so the Kg / Kd / Nmax values reproduced here are the joint typical-value estimates shared with the companion model Venisse_2008_fluconazole. Interindividual variability was reported on Nmax (225% CV exponential) and on Emax (65% CV exponential); the residual error was reported as exponential at 96% CV on the linear CFU/mL scale, encoded here as an additive SD on natural-log CFU/mL via sigma_ln = sqrt(log(CV^2 + 1)). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Leshinsky_2017_caspofungin_cat.html">Caspofungin cat (Leshinsky 2017)</a> </td> <td style="text-align:left;"> Preclinical (cat). Two-compartment population PK model with first-order linear elimination from the central compartment for intravenous caspofungin acetate in healthy adult cats (Leshinsky 2017) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bulitta_2009_ceftazidime.html">Ceftazidime (Bulitta 2009)</a> </td> <td style="text-align:left;"> In vitro (Pseudomonas aeruginosa PAO1). Mechanism-based PD model (model D) for the pronounced inoculum effect of ceftazidime, fit to static time-kill experiments at initial inocula 10^6 / 10^7 / 10^8 CFU/mL with ceftazidime 0-128 mg/L. Two co-existing bacterial subpopulations (genotypically susceptible and resistant) each follow a two-state life-cycle (S1 -> S2 -> 2 S1) with a shared slow S1 -> S2 generation step (k12) and a shared fast doubling step (k21 fixed at 50 /h). A logistic-saturation replication-efficiency factor Rep gates the doubling step toward the carrying capacity CFUmax. Ceftazidime stimulates an autolysin effect (alys_s / alys_r) that decreases the per-doubling success probability via (1 - alys); the resistant population has a smaller maximum autolysin effect (Smax_r < Smax_s). Freely diffusible signal molecules (csig1 / csig2) synthesised by all viable bacteria mediate quorum-sensing-like phenotypic tolerance at high inocula: high csig1 suppresses the autolysin loss (the inoculum effect) and prolongs the generation time (drug-conditional Inhk12 factor). Ceftazidime broth concentration cb degrades first-order at fixed half-life 45.9 h (Viaene 1973). The model has no human PK component; the experimental drug exposure is supplied as a dose into cb at time zero, and the bacterial / signal initial conditions are derived from the inoculum log10_cfuo parameter. Random effects (eta) are NOT estimated in the NONMEM analysis -- the paper reports the additive log10-scale residual SD only (between-run variability was negligible). The packaged model is therefore intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bulitta_2010_colistin.html">Colistin PAO1 (Bulitta 2010)</a> </td> <td style="text-align:left;"> In vitro (Pseudomonas aeruginosa PAO1; NONMEM VI primary fit). Mechanism-based population pharmacodynamic model for colistin static time-kill experiments across initial inocula of 10^4-10^9 CFU/mL. The bacterial system has three pre-existing subpopulations (susceptible / intermediate / least-susceptible) with different second-order killing rate constants by colistin, plus a lag compartment that initially holds the susceptible cells (no growth or natural death, but subject to colistin killing and signal-molecule inhibition). A first-order rate constant klag transfers cells from the lag compartment into the replicating susceptible compartment. Each replicating subpopulation grows via a saturable Michaelis-Menten-style function parameterised by POPmax (would-be plateau without signal-molecule inhibition) and a per-subpopulation low-density growth half-life t1/2(kg,low CFU); cells die first-order at kd = 0.3 /h. A receptor-occupancy submodel encodes competitive displacement of Mg2+/Ca2+ from outer-membrane LPS sites by colistin (Eq. 1) followed by a steep Hill function (Hill = 10 fixed) that maps the fraction of receptors not occupied by cations into the effective colistin concentration at the target site (Eq. 2). Signal molecules tracking CFUALL with first-order kinetics (kdeg) inhibit both bacterial replication (ImaxRep) and colistin killing (ImaxKill) via a Hill-1 function with the same IC50 (Eqs. 4, 5). Drug concentration Ccolistin (mg/L) and cation concentration Ccations (umol/L) are external time-varying covariates supplied from the data; the model has no human PK. Random effects (eta) are NOT included: the paper reports CV 24% IIV on the three growth half-lives only (jointly via difference-in-VGmax), reflecting between-experiment replicate variability; the packaged model is intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bulitta_2010_colistin.html">Colistin URMC1 (Bulitta 2010)</a> </td> <td style="text-align:left;"> In vitro (Pseudomonas aeruginosa URMC1; S-ADAPT fit). Mechanism-based population pharmacodynamic model for colistin static time-kill experiments against the clinical P. aeruginosa isolate URMC1, used in Bulitta 2010 alongside PAO1 and URMC2 to test whether the structural model generalises beyond the reference strain. Identical structure to Bulitta_2010_colistin_PAO1: three pre-existing bacterial subpopulations (S / I / R) with different second-order colistin killing rate constants; a lag compartment that holds susceptible cells at t = 0 and transfers them into the replicating susceptible compartment at rate klag; a saturable Michaelis-Menten-style growth function parameterised by POPmax and per-subpopulation low-density growth half-lives; first-order natural death at kd = 0.3 /h; receptor-occupancy submodel for Mg2+/Ca2+ competition followed by a Hill-10 mapping into effective colistin at the target site; and a signal-molecule compartment tracking CFUALL whose Hill-1 inhibition of replication and killing produces the observed inoculum effect. Drug concentration Ccolistin (mg/L) and cation concentration Ccations (umol/L) are external time-varying covariates from the data. Random effects (eta) are NOT included -- URMC1 is reported in the S-ADAPT column of Table 1 without IIV. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bulitta_2010_colistin.html">Colistin URMC2 (Bulitta 2010)</a> </td> <td style="text-align:left;"> In vitro (Pseudomonas aeruginosa URMC2; S-ADAPT fit). Mechanism-based population pharmacodynamic model for colistin static time-kill experiments against the clinical P. aeruginosa isolate URMC2, the second of two clinical strains used in Bulitta 2010 to externally qualify the structural model developed against PAO1. Identical structure to Bulitta_2010_colistin_PAO1 / _URMC1 -- three pre-existing subpopulations, lag compartment, saturable growth, first-order natural death, receptor-occupancy submodel with Hill-10 mapping to effective colistin, and a signal-molecule compartment producing the inoculum effect. Drug concentration Ccolistin (mg/L) and cation concentration Ccations (umol/L) are external time-varying covariates from the data. Random effects (eta) are NOT included -- URMC2 is reported in the S-ADAPT column of Table 1 without IIV. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Barnett_2018_coproporphyrin_I.html">Coproporphyrin I (Barnett 2018)</a> </td> <td style="text-align:left;"> Semi-mechanistic turnover model for the endogenous OATP1B-substrate biomarker coproporphyrin I (CPI) in healthy adult males (Barnett 2018), with simultaneous plasma and urine outputs and competitive rifampicin inhibition of biliary CPI clearance. CPI is produced at a zero-order synthesis rate ksyn, distributed in volume Vcpi, and eliminated via biliary clearance CLb,CPI (the dominant route, ~88% of total CL under baseline conditions) and renal clearance CLr,CPI. Rifampicin inhibits CLb,CPI competitively through KiCPI driven by the instantaneous plasma rifampicin concentration; CLr,CPI and ksyn are unaffected. A binary RIF-coadministration covariate additionally captures a paper-reported ~50% reduction in Vcpi during the rifampicin phase (Barnett 2018 Table 1). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yoshida_2018_coproporphyrin_I_GDC0810.html">Coproporphyrin I GDC0810 (Yoshida 2018)</a> </td> <td style="text-align:left;"> One-compartment endogenous turnover model for the OATP1B-substrate biomarker coproporphyrin I (CPI) in healthy adults (Yoshida 2018, GDC-0810-CPI calibration). CPI is produced at a zero-order synthesis rate Ksyn = kdeg * Baseline and eliminated as a single first-order pool whose overall rate constant kdeg is decomposed into a non-hepatic fraction fNH (held fixed at 12.9 %, unaffected by inhibitor) and a hepatic fraction 1 - fNH (competitively inhibited by the OATP1B perpetrator via Ki,u). The perpetrator portal-vein unbound concentration enters as a time-varying covariate CP_GDC_UM (umol/L); setting CP_GDC_UM = 0 collapses the model to the inhibitor-free steady state Baseline. This file encodes the GDC-0810-CPI calibration (Table 2 right column) with IIV on Baseline (18.2 %CV) and Ki,u (30.1 %CV); a sibling file Yoshida_2018_coproporphyrin_I_rifampin encodes the rifampin calibration with its own Ki,u, kdeg, and no IIV. The original fit used a Y. Chen et al. in-house PBPK model for GDC-0810 portal-vein concentrations (personal communication, not on disk and not in the nlmixr2lib registry), so downstream users must supply CP_GDC_UM externally. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yoshida_2018_coproporphyrin_I_rifampin.html">Coproporphyrin I rifampin (Yoshida 2018)</a> </td> <td style="text-align:left;"> One-compartment endogenous turnover model for the OATP1B-substrate biomarker coproporphyrin I (CPI) in healthy adults (Yoshida 2018, rifampin-CPI calibration). CPI is produced at a zero-order synthesis rate Ksyn = kdeg * Baseline and eliminated as a single first-order pool whose overall rate constant kdeg is decomposed into a non-hepatic fraction fNH (unaffected by inhibitor) and a hepatic fraction 1 - fNH (competitively inhibited by the OATP1B perpetrator via Ki,u). The perpetrator portal-vein unbound concentration enters as a time-varying covariate CP_RIF_UM (umol/L); setting CP_RIF_UM = 0 collapses the model to the inhibitor-free steady state Baseline. This file encodes the rifampin-CPI calibration (Table 2 left column; no IIV reported); a sibling file Yoshida_2018_coproporphyrin_I_GDC0810 encodes the GDC-0810 calibration with its own Ki,u, kdeg, and IIV structure. The original fit used a Simcyp v16r1 default single-dose rifampin model for the portal-vein concentration profile; that PBPK output is not reproducible from on-disk sources and the paper itself documents an approximately 5-fold sensitivity of the estimated Ki,u to the choice of perpetrator-PK model, so downstream users must supply CP_RIF_UM externally and treat the calibrated Ki,u as conditional on that choice. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Rao_2023_covid19_qsp.html">Covid19 qsp (Rao 2023)</a> </td> <td style="text-align:left;"> QSP. Quantitative systems pharmacology model of SARS-CoV-2 within-host viral dynamics coupled to the innate and adaptive immune response and alveolar tissue damage, with pluggable anti-SARS-CoV-2 neutralizing-antibody (nAb) and antiviral treatment arms. 54 ODE states spanning an alveolar and a plasma compartment: virus, healthy/infected/damaged alveolar type I and type II cells, dendritic cells, macrophages, neutrophils, CD4+ Th1/Th17/Treg and CD8+ cytotoxic T cells, eleven cytokines (TNFa, IL-6, IL-1b, IFN-b, IFN-g, IL-2, IL-12, IL-17, IL-10, TGF-b, GM-CSF), the biomarkers surfactant protein D, ferritin and C-reactive protein, a two-compartment PK model for each of the two nAbs in a cocktail, and an endogenous antibody response. The distinguishing feature of this release is log-sensing (Weber-Fechner) activation of the innate immune cells by virus, infected cells and damaged cells, which keeps the immune response sensitive over orders of magnitude of viral load. Calibrated to a curated set of observational COVID-19 studies and to the Blaze-1 Ph3 (bamlanivimab + etesevimab), REGEN-COV Ph2/Ph3 (casirivimab + imdevimab) and molnupiravir Ph2 outpatient randomized controlled trials. All parameters are hand-calibrated literature / assumed constants rather than maximum-likelihood estimates; no fitted IIV or residual error (the published heterogeneity is generated by Latin-hypercube resampling of the parameters, not by an omega matrix). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dodds_2021_covid19_viralkinetic.html">Covid19 viralkinetic (Dodds 2021)</a> </td> <td style="text-align:left;"> Mechanistic viral kinetic (target-cell-limited with eclipse phase) model of SARS-CoV-2 infection dynamics adapted by Dodds et al. (2021) from the target-cell-limited viral kinetics framework of Goncalves et al. (2020) that characterized nasopharyngeal viral load in 13 hospitalized COVID-19 patients. Four ODE states: uninfected epithelial target cells 'target' (T; #/mL), latently (eclipse) infected cells 'eclipse' (I1; #/mL), productively infected cells 'infected' (I2; #/mL) and free virions 'virus' (V; copies/mL). Rate constants (all /day) are: k for eclipse to productive transition, delta for productively-infected cell death, rho for virion production per productively-infected cell, c for virion clearance, and beta (mL/#/day) for target-cell infection. beta is derived from the basic reproductive ratio R0 via beta = R0 * delta * c / (T0 * (rho - R0 * delta)). All structural parameters are FIXED to the consensus values reported in Dodds 2021 Section 2.1 (T0 = 1.33e5 #/mL, V0 = 0.1 #/mL, k = 3 /day, delta = 0.60 /day, rho = 22.7 /day, c = 10 /day, R0 = 8.6). Simulation-time drug interventions are supported as Heaviside step modifiers applied at time t_intervention: fractional inhibition on beta, k, and rho (imax_*, 0-1) and multiplicative stimulation on delta and c (smax_*, >=0), all defaulting to fixed(0) so the untreated natural-history dynamics are the baseline behaviour. The model reproduces the untreated viral trajectory peaking ~9 days post-infection used by Dodds 2021 to explore single- and combination-target antiviral effects on viral load AUC, duration of viral shedding (>= 100 copies/mL) and epithelial cells infected. This is a deterministic simulation model (no IIV, no residual error, no PK compartment): drug potencies must be supplied per scenario by overriding imax_*, smax_*, and t_intervention. The eclipse-phase compartment is declared paper-specific via paper_specific_compartments; the productive compartment reuses the existing canonical 'infected' registered by Wang_2018_daclatasvir_asunaprevir. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dogra_2023_covid19vaccine.html">Covid19vaccine (Dogra 2023)</a> </td> <td style="text-align:left;"> QSP. Mechanistic ODE model (17 states) of the adaptive immune response to COVID-19 mRNA vaccination in healthy and immunocompromised adults. States include naive and activated antigen-presenting cells (apc, apc_active); naive and effector CD4+ T-cells (cd4, cd4_effector); naive and effector CD8+ T-cells (cd8, cd8_effector); naive and activated B-cells (bcell, bcell_active); antibody-secreting plasma cells (plasma_cell); neutralizing antibody titer (antibody, U/mL); type-I and type-II interferons and interleukin-6 (ifn1, ifn2, il6, pg/mL); and a target-cell-limited SARS-CoV-2 infection subsystem (healthy_cell, infected_cell, virus). Antigen from mRNA vaccine doses enters via a nanoparticle depot (depot_antigen); the immune-status modulation factor f_immune in [0,1] scales the naive-cell homeostasis of CD4, CD8, and B cells to represent healthy (f_immune=1), mildly (~0.75), or severely (~0.55) immunocompromised subjects. Deterministic typical-value mechanism (no IIV, no residual error) calibrated jointly against Lucas 2020 infection kinetics (n=80) and Collier 2021 healthy-adult (n=31) and Peeters 2022 cancer-patient vaccine kinetics (n=63 chemotherapy, n=16 immunotherapy); validated against Bayart 2021 (n=158) and Papazisis 2022 (n=110). Parameter values are the healthy-adult vaccine calibration (Table 1 VCH column); IC (infection-calibration) alternates for the six calibration-dependent parameters are documented in comments. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Desikan_2022_covidVaccine.html">CovidVaccine (Desikan 2022)</a> </td> <td style="text-align:left;"> QSP (immunodynamics, deterministic). Two-antigen, three-epitope, multi-B-cell / antibody model of the humoral recall response to a two-strain vaccine sequence (ancestral Wuhan strain WU and a novel variant OM = SARS-CoV-2 Omicron BA.1; the framework is generic and is also validated against H5N1 influenza-vaccine data in Figure 5 of Desikan et al. 2022). Each antigen carries a conserved epitope class (C, shared across WU and OM) and a variant-unique epitope class (W on WU, O on OM). Fourteen ordinary differential equations track: two free antigens (agW, agV), six antibody-bound antigen intermediates (agWw / agWc / agWwc / agVv / agVc / agVvc, where the subscripts denote which epitope class carries a bound antibody), three B-cell populations producing antibodies specific to each epitope class (Bc, Bw, Bv), and three corresponding antibody pools (Ac, Aw, Av). B-cell proliferation follows a saturating dose-response in the amount of cognate antigen exposed (free-epitope form); previously primed (memory) B cells can additionally be recruited by non-homotypic antigen at fraction f_b of the homotypic rate, which provides the original-antigenic-sin (OAS) mechanism. Reported observables (used to reproduce all figures) are the multiplicity-weighted antibody titres against each virus variant: Ab_WU = m_conserved*Ac + n_unique*Aw and Ab_OM = m_conserved*Ac + n_unique*Av, with m:n = 1:5 (Table 1). All ten model constants are FIXED at Table 1 values; there is no IIV and no residual error (deterministic simulation is the intended use). Vaccinations enter as bolus dose events on the appropriate free-antigen compartment (agW for WU-vaccinations; agV for OM-vaccinations); at the FIRST WU vaccination the naive WU-unique B-cell and antibody pools are seeded with n_unique units on Bw and Aw, and at the FIRST OM vaccination the naive OM-unique pools are seeded with n_unique units on Bv and Av. Conserved-epitope pools Bc and Ac start at their naive m_conserved units regardless of which vaccine is given first. See the validation vignette (Desikan_2022_covidVaccine) for reference vaccination schedules and reproductions of Figures 2A, 3, 4A, 4B and 5. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ribba_2022_ctdna_tumor_size.html">Ctdna (Ribba 2022)</a> </td> <td style="text-align:left;"> Stein bi-exponential population model of the circulating tumor DNA (ctDNA) time course in advanced non-small cell lung cancer (NSCLC), fit by Ribba et al. (Roche) to the atezolizumab arm of the phase III OAK study (n = 46 participants with serial ctDNA). ctDNA was measured with the Roche AVENIO next-generation-sequencing panel and reported as average mutant molecules per millilitre of plasma (MMPM) at four cycles (baseline, ~21, ~42 and ~63 days). The data were base-10 log transformed before fitting, so the modeled quantity is log10(MMPM) and the Stein baseline y0 is log10 of the observed baseline MMPM, supplied as a per-subject regressor rather than estimated. The model decomposes the ctDNA signal into an exponentially growing and an exponentially decaying fraction; the population nadir time log(kse/kge)/(kge+kse) = 63.6 days = 9.1 weeks underpins the paper's conclusion that sampling at cycle 2 (21 days) is too early to capture the maximal ctDNA response. Inter-individual variability on both rate constants is close to 100% CV, which the authors highlight as the central challenge for ctDNA-informed decision making. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ribba_2022_ctdna_tumor_size.html">Ctdna sld joint (Ribba 2022)</a> </td> <td style="text-align:left;"> Joint Stein bi-exponential model of circulating tumor DNA (ctDNA) and tumor size (sum of the longest diameters, SLD) in advanced non-small cell lung cancer, fit by Ribba et al. (Roche) to the phase III OAK study. Two Stein sub-systems run in parallel -- one on SLD in mm, one on base-10 log-transformed ctDNA in MMPM -- and are coupled through a single estimated link parameter zeta: the ctDNA decay rate is not free but is set to zeta times the SLD decay rate, encoding the hypothesis that both biomarkers are driven by one underlying treatment-response process. Every other population parameter (both growth rates, the SLD decay rate, and their between-subject variances) is FIXED to the value obtained from the two independent single-endpoint fits, Ribba_2022_sld.R and Ribba_2022_ctdna.R; only zeta, its variance, and the two constant residual-error terms were estimated. The estimated zeta of 1.94 means ctDNA falls about twice as fast as tumor size. The authors present the resulting individual fits as illustration only and state explicitly that the parameters were not estimated with sufficient precision for predictive use; this model is packaged for structural reproduction of Ribba 2022 Figure 2F (right), not for prediction. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Harun_2019_cysticFibrosis.html">CysticFibrosis (Harun 2019)</a> </td> <td style="text-align:left;"> Non-linear mixed-effects disease-progression model of forced expiratory volume in 1 second (FEV1) percent predicted versus age in children with cystic fibrosis (Harun 2019). The model describes the typical sigmoid-Emax decline of FEV1% predicted from a baseline at age 5 years to an asymptote, with covariate effects of BMI z-score and severe air trapping at age 5 on the baseline, and time-varying hospitalisation due to pulmonary exacerbation on the maximum drop and the half-effect age. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2018_daclatasvir_asunaprevir.html">Daclatasvir asunaprevir (Wang 2018)</a> </td> <td style="text-align:left;"> MBMA PK + mechanistic HCV viral-dynamic (VD) model for daclatasvir (DCV, NS5A inhibitor) and asunaprevir (ASV, NS3/4A protease inhibitor) combination therapy in adults with genotype-1 chronic hepatitis C (Wang 2018). PK was developed by a model-based meta-analysis of arm-mean concentration data pooled from 26 trials (1067 concentration records, DCV 198 subjects in 30 arms / 7 trials, ASV 290 subjects in 35 arms / 11 trials); each drug uses an independent 2-compartment disposition model with inter-arm variability (IAV) encoded as study-arm-level etas (not between-subject variability). DCV absorption is first-order; ASV absorption is simultaneous zero- plus first-order with formulation-dependent fraction FK absorbed via the zero-order route (FK=0.184 for capsule/tablet, 0.334 for suspension/solution). The shared viral dynamics is a Neumann-style three-state target-cell model (uninfected target cells `target` T, productively infected cells `infected` I, free virions `virus` V) with most system constants (Tmax, d, R0, delta) FIXED to literature values from Neumann et al 1998; virion clearance c and production p are estimated. Each drug acts via its own effect compartment with a sigmoid-Emax inhibition of virion production (Emax=1) and an empirical exponentially time-increasing IC50 capturing the emergence of drug-resistant variants (Kr coefficient). Genotype subtype (GT1A vs GT1B) modifies IC50 by a fixed scaling factor (SCL_DCV=0.18, SCL_ASV=0.30, both GT1B/GT1A ratio) and modifies the DCV resistance rate (Kr_DCV=0.43 /day for GT1A vs 0.13 /day for GT1B). Combination efficacy follows the Bliss-additive form ECOMB/(1-ECOMB) = EDCV/(1-EDCV) + EASV/(1-EASV) (Eq 13). The model is intended for simulating arm-mean PK and viral-load trajectories under DCV monotherapy, ASV monotherapy, or DCV+ASV combination regimens; downstream NCA-style summaries (Cmax, Tmax, AUC) reproduce the published per-dose-group PK profiles, and viral-load trajectories reproduce the published biphasic decline and resistance-driven rebound shapes. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vasalou_2023_dendriticNanoparticle.html">DendriticNanoparticle dog (Vasalou 2023)</a> </td> <td style="text-align:left;"> Preclinical (beagle dog, male). PBPK (semi-mechanistic dendritic-nanoparticle biodistribution). Dual-species model carrying nanoparticle-conjugated API and nanoparticle-released (free) API through four compartments each -- blood, liver, spleen and a lumped 'other'/rest compartment (8 ODEs). Structure is identical to the mouse anchor; species-specific parameters were obtained PROSPECTIVELY by scaling the mouse fit, not by re-fitting dog data: blood and rest volumes and the saturable-binding parameters by the blood-protein-binding ratio (Eqs 22 and 24), blood/rest flow by a BW^0.7 power law (Eq 23), and the conjugated-API extravasation rates by organ blood-flow ratios (Eq 25). This is the species the paper's sensitivity analysis was run on, because the dog dataset was the most complete. Deterministic typical-value simulator: no IIV and no residual error are reported. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vasalou_2023_dendriticNanoparticle.html">DendriticNanoparticle human (Vasalou 2023)</a> </td> <td style="text-align:left;"> PBPK (semi-mechanistic dendritic-nanoparticle biodistribution), human first-in-human PROJECTION -- no human data were fitted. Dual-species model carrying nanoparticle-conjugated API and nanoparticle-released (free) API through four compartments each -- blood, liver, spleen and a lumped 'other'/rest compartment (8 ODEs). Structure is identical to the mouse anchor; human parameters were obtained entirely by prospective scaling of the mouse fit (Eqs 22-25), with the released-API clearance set to 0.32 L/h/kg from the in-vitro human hepatocyte Clint of 15.5 uL/min/1e6 cells via the well-stirred model. Intended as a translational / clinical-projection tool; the simulated dose-normalised profiles overlay the observed mouse, rat and dog data (Fig 9). Deterministic typical-value simulator: no IIV and no residual error are reported. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vasalou_2023_dendriticNanoparticle.html">DendriticNanoparticle mouse (Vasalou 2023)</a> </td> <td style="text-align:left;"> Preclinical (mouse, SCID CB-17). PBPK (semi-mechanistic dendritic-nanoparticle biodistribution). Dual-species model carrying nanoparticle-conjugated API and nanoparticle-released (free) API through four compartments each -- blood, liver, spleen and a lumped 'other'/rest compartment (8 ODEs). Conjugated API extravasates into liver and spleen at fitted rates and releases free API by first-order, compartment-specific rate constants (t50 = 5.5 h blood, 43 h liver, 110 h spleen); free API distributes by blood flow with saturable liver and spleen partition coefficients and is cleared from the liver. No separate nanoparticle clearance term. This is the mouse parameterisation, the anchor that was fitted to data; the rat, dog and human files scale it per Eqs 22-25. Deterministic typical-value simulator: no IIV and no residual error are reported. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vasalou_2023_dendriticNanoparticle.html">DendriticNanoparticle rat (Vasalou 2023)</a> </td> <td style="text-align:left;"> Preclinical (rat, male Han Wistar). PBPK (semi-mechanistic dendritic-nanoparticle biodistribution). Dual-species model carrying nanoparticle-conjugated API and nanoparticle-released (free) API through four compartments each -- blood, liver, spleen and a lumped 'other'/rest compartment (8 ODEs). Structure is identical to the mouse anchor; species-specific parameters were obtained PROSPECTIVELY by scaling the mouse fit, not by re-fitting rat data: blood and rest volumes and the saturable-binding parameters by the blood-protein-binding ratio (Eqs 22 and 24), blood/rest flow by a BW^0.7 power law (Eq 23), and the conjugated-API extravasation rates by organ blood-flow ratios (Eq 25). Deterministic typical-value simulator: no IIV and no residual error are reported. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gilkey_2015_DiRnanoparticle.html">DiRnanoparticle (Gilkey 2015)</a> </td> <td style="text-align:left;"> Preclinical (mouse, BALB/c, 4-6 weeks). PBPK model for fluorescently labeled (DiR) block-copolymer nanoparticles in mice, developed as a surrogate model for dexamethasone-encapsulated nanoparticles in pediatric acute lymphoblastic leukemia therapy. Five compartments: plasma, liver, spleen, kidneys, and a virtual 'other' compartment introduced to close the mass balance for ~50% of injected dose that experimental imaging could not account for in the four sampled organs. Single 100 uL IV bolus of 5 ug/mL DiR-NPs; the model treats plasma initial concentration as 5 ug/mL per paper convention (rather than 0.5 ug dose distributed into 1.7 mL plasma volume) -- see vignette Assumptions section. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hajjar_2018_DMD_6MWT.html">DMD 6MWT (Hajjar 2018)</a> </td> <td style="text-align:left;"> Latent variable disease-progression model for the six-minute walk test (6MWT, meters) in healthy boys and boys with Duchenne muscular dystrophy (DMD), fit by Hajjar et al. (ACoP9 2018 poster T-011) to publicly available individual-level longitudinal natural-history 6MWT data from 16 healthy controls and 219 DMD patients. The 6MWT is modelled as a one-compartment indirect-response state (walkDist, meters): a zero-order production rate KIN feeds the state and a first-order dissipation rate KOUT removes it. A change point at subject age MTIME (1.75 years) switches KIN from 0 to its non-zero value, encoding the developmental lag before toddlers can walk a measurable distance in six minutes. A latent exponential disease process DIS = ALPHA*exp(BETA*age) stimulates the dissipation rate for DMD subjects only; healthy subjects fix ALPHA = BETA = 0 so the disease term vanishes. The DIS_DMD covariate (1 = DMD, 0 = healthy) additionally multiplies KIN by KCOV (0.63) for DMD subjects so the two populations share KOUT but have separate KIN. Between-subject variability is exponential on KOUT (DMD subjects only, per the source NONMEM control stream), on KIN (both populations), and on ALPHA and BETA (DMD subjects only). Residual error is additive on the 6MWT scale. The model has no drug input; the source poster frames it as a simulation tool for designing future DMD efficacy trials. Time is age in years (the integration variable; the source poster reports KOUT and KIN in per-month units for human readability, see vignette Errata for the unit-conversion step). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hamuro_2017_DMD_6MWT.html">DMD 6MWT (Hamuro 2017)</a> </td> <td style="text-align:left;"> Natural-history disease-progression model for the six-minute walk test (6MWT, meters) in ambulatory boys with Duchenne muscular dystrophy (DMD) on stable corticosteroids (Hamuro 2017). The 6MWT vs subject age is modelled as the minimum of two simultaneously estimated linear lines (Phoenix NLME 'min' function): a developmental line with a positive slope (improvement) and a disease-induced line with a negative slope (decline). The two lines intersect at the population-typical age of maximum 6MWT (10 years). Exponential between-subject variability is estimated on both slopes (the intercepts have no IIV); residual error is additive. Disease-progression model with no drug dosing. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mody_2023_doxorubicin_dexrazoxane.html">Doxorubicin dexrazoxane (Mody 2023)</a> </td> <td style="text-align:left;"> In vitro (AC16 immortalized human cardiomyocyte cell line) toxicodynamic (TD) model of doxorubicin (DOX) induced cardiotoxicity and dexrazoxane (DEX) cardioprotection, linked to clinical human pharmacokinetics for the in vitro to in vivo translation of drug-drug interaction and Q3W / Q1W dosing-regimen optimization. Clinical DOX PK is a 3-compartment mammillary model with linear elimination and 15-min IV infusion (parameters from Kontny et al. 2013, ref [30], reproduced inline in Mody 2023 Table 2 for a 1.8 m^2 body surface area typical subject). Clinical DEX PK is a 2-compartment mammillary model with linear elimination and 15-min IV infusion, originally fitted to phase 1 data reported by Earhart et al. 1982 Cancer Res 42(12):5255-61 (Mody 2023 Table 2). The TD model on AC16 has (1) an exponential-growth baseline for the untreated cell viability R, (2) a linear DOX growth-inhibition slope S_DOX on kg, (3) a linear DEX growth-inhibition slope S_DEX on kg, (4) a Hill (capacity-limited) DOX stimulation-of-death signal K_DOX = Kmax * CDOX / (KC50 + CDOX) delayed through three transit compartments K1..K3 with rate 1/tau_DOX, and (5) a Hill (capacity-limited) DEX inhibition of the DOX-death signal K_DEX = Imax * CDEX / (IC50 + CDEX). The DEX inhibition enters the transit-chain source term as a subtraction on Kmax (see Assumptions and deviations in the vignette). No IIV or residual error is fitted (Table 1 reports %RSE on point estimates only); an arbitrary 10% CV IIV is added at simulation time per the paper's Methods. Placeholder additive residual SDs are encoded as fixed() so the model parses cleanly. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wen_2016_enrofloxacin_Pmultocida.html">Enrofloxacin MIC0p01 (Wen 2016)</a> </td> <td style="text-align:left;"> In vitro (Pasteurella multocida bovine-lung isolate, enrofloxacin MIC = 0.01 ug/mL). Additive inhibitory sigmoidal Emax pharmacodynamic model for the rate of growth/decline of the bacterial population under constant exposure to enrofloxacin. Wen 2016 Eq (1) parameterises the rate of change of log10(CFU/mL) as E(C) = E0 - Emax * C^H / (EC50^H + C^H), with C the enrofloxacin concentration in the broth. The packaged model encodes the bacterial density bact (linear CFU/mL) with d/dt(bact) = ln(10) * E(C) * bact so a constant drug exposure yields the linear log10-CFU-vs-time slope that Wen 2016 fit in Phoenix NLME. This is the most-susceptible isolate (concentration-dependent PD per the paper Discussion): Emax = 1.64, EC50 = 0.11 ug/mL (10x MIC), Hill = 0.97, E0 = 0.32 log10(CFU/mL)/h. The enrofloxacin concentration is an external time-varying covariate Cenrofloxacin (no PK component); the experimental design holds it constant for 24 h at 0, 0.5, 0.75, 1, 2, 3, 5, or 10 multiples of the isolate MIC. Random effects are NOT included -- the paper added 'the experiment was added as a random effect' (Methods, Fitting PD model) but did not report the variance, so the packaged model is intended for typical-value / parametric-uncertainty simulation matching the 1,000-Monte-Carlo simulations in Figs 2 and 3. The packaged additive residual SD on log10(CFU/mL) is FIXED at 0 because Wen 2016 did not report a density-scale residual SD (their residual was on the growth-rate regression). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wen_2016_enrofloxacin_Pmultocida.html">Enrofloxacin MIC1p5 (Wen 2016)</a> </td> <td style="text-align:left;"> In vitro (Pasteurella multocida bovine-nasal-swab isolate, enrofloxacin MIC = 1.5 ug/mL). Additive inhibitory sigmoidal Emax pharmacodynamic model for the rate of growth/decline of the bacterial population under constant exposure to enrofloxacin. Wen 2016 Eq (1) parameterises the rate of change of log10(CFU/mL) as E(C) = E0 - Emax * C^H / (EC50^H + C^H), with C the enrofloxacin concentration in the broth. The packaged model encodes the bacterial density bact (linear CFU/mL) with d/dt(bact) = ln(10) * E(C) * bact so a constant drug exposure yields the linear log10-CFU-vs-time slope that Wen 2016 fit in Phoenix NLME. This is a less-susceptible isolate (time-dependent PD per the paper Discussion): Emax = 0.81, EC50 = 2.13 ug/mL (close to the MIC), Hill = 3.05, E0 = 0.31 log10(CFU/mL)/h. The enrofloxacin concentration is an external time-varying covariate Cenrofloxacin (no PK component); the experimental design holds it constant for 24 h at 0, 0.5, 0.75, 1, 2, 3, 5, or 10 multiples of the isolate MIC. Random effects are NOT included -- the paper added 'the experiment was added as a random effect' (Methods, Fitting PD model) but did not report the variance, so the packaged model is intended for typical-value / parametric-uncertainty simulation matching the 1,000-Monte-Carlo simulations in Figs 2 and 3. The packaged additive residual SD on log10(CFU/mL) is FIXED at 0 because Wen 2016 did not report a density-scale residual SD (their residual was on the growth-rate regression). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wen_2016_enrofloxacin_Pmultocida.html">Enrofloxacin MIC2p0 (Wen 2016)</a> </td> <td style="text-align:left;"> In vitro (Pasteurella multocida bovine-nasal-swab isolate, enrofloxacin MIC = 2.0 ug/mL). Additive inhibitory sigmoidal Emax pharmacodynamic model for the rate of growth/decline of the bacterial population under constant exposure to enrofloxacin. Wen 2016 Eq (1) parameterises the rate of change of log10(CFU/mL) as E(C) = E0 - Emax * C^H / (EC50^H + C^H), with C the enrofloxacin concentration in the broth. The packaged model encodes the bacterial density bact (linear CFU/mL) with d/dt(bact) = ln(10) * E(C) * bact so a constant drug exposure yields the linear log10-CFU-vs-time slope that Wen 2016 fit in Phoenix NLME. This is the least-susceptible isolate (time-dependent PD per the paper Discussion): Emax = 0.69, EC50 = 1.60 ug/mL (below the MIC), Hill = 4.37, E0 = 0.29 log10(CFU/mL)/h. The enrofloxacin concentration is an external time-varying covariate Cenrofloxacin (no PK component); the experimental design holds it constant for 24 h at 0, 0.5, 0.75, 1, 2, 3, 5, or 10 multiples of the isolate MIC. Random effects are NOT included -- the paper added 'the experiment was added as a random effect' (Methods, Fitting PD model) but did not report the variance, so the packaged model is intended for typical-value / parametric-uncertainty simulation matching the 1,000-Monte-Carlo simulations in Figs 2 and 3. The packaged additive residual SD on log10(CFU/mL) is FIXED at 0 because Wen 2016 did not report a density-scale residual SD (their residual was on the growth-rate regression). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tetschke_2018_erythropoiesis.html">Erythropoiesis (Tetschke 2018)</a> </td> <td style="text-align:left;"> Three-compartment population mixed-effects model for human erythropoiesis (red blood cell regeneration after a phlebotomy / blood donation) in healthy adults </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wu_2014_FEV1_asthma.html">FEV1 asthma (Wu 2014)</a> </td> <td style="text-align:left;"> Longitudinal nonlinear mixed-effects disease-progression model of pre-bronchodilator forced expiratory volume in 1 second (FEV1) in children with asthma, fit to 1,041 participants in the Childhood Asthma Management Program (CAMP) study (Wu 2014). The structural model is an exponential function of age and height with a race-stratified intercept and an additive treatment-arm effect for inhaled budesonide on the linear-scale FEV1; nedocromil shows no significant effect and is pooled into the placebo reference. The model has no PK component -- the inhaled-corticosteroid effect is encoded as a per-subject binary treatment-arm indicator (CONMED_BUDESONIDE). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Venisse_2008_candida_albicans.html">Fluconazole (Venisse 2008)</a> </td> <td style="text-align:left;"> In vitro (Candida albicans, ATCC 3153). Mechanism-based PK-PD model of fluconazole fungistatic activity in a dynamic broth-renewal flask. Candida population dynamics follow Mouton-type saturable exponential growth with a maximum carrying capacity Nmax minus a first-order natural-death rate Kd minus a fixed broth-renewal rate Ke (set by the experimenters to 0.231 1/h, equal to the drug 1-compartment kel and yielding a 3 h elimination half-life). Fluconazole multiplicatively inhibits growth via a sigmoid-Emax (Hill=1) Imax * C / (IC50 + C) term where Imax < 1 represents the maximum fractional reduction of the growth rate. Drug concentration is the central/Vc concentration from a one-compartment PK with V and CL fixed to the apparatus geometry (V = 0.4 L bulk volume of the glass flask, CL = 1.54 mL/min consistent with t1/2 = 3 h). Candida growth/death/Nmax parameters were estimated from all six experiments (3 fluconazole + 3 caspofungin) in a single Nonmem run, so the Kg / Kd / Nmax values reproduced here are the joint typical-value estimates shared with the companion model Venisse_2008_caspofungin. Interindividual variability was reported on Nmax (225% CV exponential); the residual error was reported as exponential at 197% CV on the linear CFU/mL scale, encoded here as an additive SD on natural-log CFU/mL via sigma_ln = sqrt(log(CV^2 + 1)). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bonner_2015_gastric_emptying.html">Gastric emptying (Bonner 2015)</a> </td> <td style="text-align:left;"> Population meta-analysis of the postprandial gastric emptying (GE) time course in humans, pooled across 49 published studies of 1457 subjects spanning 28 weeks gestational age (very-low-birth- weight preterm neonates) through adults (Bonner 2015). Structural model is a double Weibull mixture on the fraction of test meal remaining in the stomach vs. time-after-ingestion (Eq. 1: GE = (100 - PR) * exp(-(t/gamma1)^beta1) + PR * exp(-(t/gamma2)^beta2)). The five test meal categories (aqueous, breast milk, formula, semi-solid, solid) are the only significant covariate; postnatal age and gestational age were tested and found NOT significant. The five reported theta_meal coefficients enter as multiplicative scalings on the fast-phase Weibull scale gamma1, with aqueous as the renormalised reference (theta_meal / theta_Aqueous). This attachment point, the natural- percent reading of PR, and the CV-percent reading of the Table 3 omega^2 column are documented best-effort interpretations because the paper's Methods do not specify which parameter the meal-type theta modifies, the PR reporting scale, or the omega^2 unit convention, and neither the main paper nor the two supplementary material files contain the NONMEM control stream. See the validation vignette's Errata for the full ambiguity list and the operator-approved (2026-07-24) best-effort choices. Not a drug PK model -- no dose, no plasma compartment, no drug PK observation; the single state is the percentage of test meal remaining, initialised to 100 at t = 0 by construction. Reported meal-type mean gastric residence times (MGRT, Eq. 4) from the paper's 1000-individual simulations: 45 min (aqueous), 57 min (breast milk), 64 min (formula), 87 min (semi-solid), 98 min (solid). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Guiastrennec_2016_gastric_emptying.html">Gastric emptying (Guiastrennec 2016)</a> </td> <td style="text-align:left;"> Mechanism-based model of postprandial gastric emptying (GE), cholecystokinin (CCK) plasma kinetics, and gallbladder emptying (GBE) in response to caloric intake (Guiastrennec 2016). Acetaminophen (1.5 g oral) is used as a gastric-emptying marker: it shares the stomach -> upper-small-intestine transit rate KG with the nutrients, and is absorbed by KA into a two-compartment paracetamol disposition (CL, VC, Q, VP) where it is observed as plasma Cc. KG is gated by a Hill-type onset function in time-after-dose (sigmoidicity SIG; T50OGTT 15.7 min for glucose-only drinks, T50Fat 23.1 min for fat- containing drinks; T50 = 0 for water) and inhibited by a linear feedback of caloric content in the upper SI (SLPCAL = -0.0173 1/kcal, 40.7% stronger in females via SEXF). Per-nutrient amounts (g of fat, protein, carbohydrate) are tracked through three parallel signal tracks: an 'upper SI' track that drives the GE caloric feedback (stomach -> upper_si via KG, drains via KUL with a fixed saturable MM absorption RAMAX/KM scaled from glucose), a 'duodenum' track that drives both the CCK-fast (CCKF) Emax stimulus and the GBE Emax stimulus (stomach -> duodenum via KG, drains via fixed KDJ), and an 'upper jejunum' track that drives the CCK-late (CCKL) linear stimulus (duodenum -> upper_jejunum via KDJ, drains via fixed KJI). Both downstream tracks run in parallel from a single stomach (literal reading of Figure 1; the joint-topology choice is documented in the validation vignette's Errata). CCKF and CCKL are encoded as paired precursor + plasma indirect-response models (KRF stimulated by duodenal-signal Emax; KRL stimulated by jejunal-signal linear; DIS_DIAB depresses carbohydrate potency POTcarbC by -81.1%). GBE is an indirect-response model where the duodenal-nutrient signal increases the gallbladder release rate KRB (Emax with S50_BILE_eff that scales positively with AGE at +2.15%/yr above reference 58 y, and a WT-driven baseline gallbladder volume BASEBILE_eff at +1.19%/kg above reference 88 kg). No recirculation of emptied bile. Observed outputs are paracetamol plasma concentration (Cc, uM), total CCK plasma concentration (TCCK = CCKF + CCKL, pM), and gallbladder volume (GVol, mL); each carries its own residual-error component as reported in the paper's Table 3. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wu_2024_gfr_maturation.html">Gfr maturation (Wu 2024)</a> </td> <td style="text-align:left;"> Glomerular filtration rate (GFR) maturation model for preterm and term-born individuals from birth to 18 years of age (Wu 2024 Pharm Res). The model simultaneously characterises inulin clearance values (assumed equal to GFR; Eq. 1) and serum creatinine concentrations (assumed at quasi-steady state with synthesis = production / GFR; Eq. 2). GFR at birth GFRbirth is a linear function of birthweight and postnatal maturation follows a sigmoidal Emax (Hill) function of postnatal age (PNA, days; Eq. 3 and Eq. 4) with GFRmax allometrically scaled to current weight (exponent fCW) and PNA50 power-scaled to gestational age (exponent GAPNA50) so the rate of postnatal maturation depends on the degree of prematurity. Creatinine synthesis rate uses the Pierce 2021 (Kidney Int) age- and sex-dependent k with height and BSA scaling (Table S1 of the supplement). Time axis is PNA in days; the model has no drug compartments and no dosing -- it predicts GFR (mL/min, observed as inulin clearance) and Scr (mg/dL) at each user-supplied observation time given the subject covariates. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lin_2020_glasdegib_AML_overall_survival.html">Glasdegib decitabine (Lin 2020)</a> </td> <td style="text-align:left;"> Parametric exponential time-to-event model for overall survival in the exploratory pooled BRIGHT AML 1003 Phase 1b + Phase 2 analysis (Lin 2020): older adults with newly diagnosed AML or high-risk MDS ineligible for intensive chemotherapy across three treatment arms (LDAC alone, glasdegib + LDAC, glasdegib + decitabine; n = 162). Hazard h(t) = lambda * (1 - theta_gl_ldac * GL_LDAC_arm) * (1 - theta_gl_dec * GL_DEC_arm), where lambda is the daily death hazard for the LDAC alone reference arm and the two arm indicators encode the proportional hazard reduction associated with each glasdegib-containing combination. Treatment arm was the only covariate retained after SCM backward elimination (alpha < 0.001) from a full model that also included log-transformed baseline percentage of bone marrow blasts and prior hypomethylating-agent use. Original NONMEM run was fit with no estimated IIV and no residual error (likelihood-based parametric TTE). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lin_2020_glasdegib_AML_overall_survival.html">Glasdegib exposure (Lin 2020)</a> </td> <td style="text-align:left;"> Parametric exponential time-to-event model for overall survival in the exposure-response analysis subset of BRIGHT AML 1003 Phase 2 (Lin 2020): older adults with newly diagnosed AML ineligible for intensive chemotherapy who were randomised to glasdegib 100 mg QD + LDAC and had glasdegib PK data available (n = 75). Hazard h(t) = lambda. Seven glasdegib exposure metrics (first-dose Cmax, end-of-cycle-1 Cmax, end-of-cycle-1 Cmin, cycle-1 cumulative AUC, cycle-1 Cavg, average AUC over dosing interval, overall Cavg; raw scale and log-transformed) and baseline ECOG performance status and cytogenetic risk were tested in SCM forward inclusion (alpha < 0.05); ECOG and cytogenetic risk entered the full model but no covariates survived backward elimination (alpha < 0.001). The final exposure-response model is therefore an intercept-only exponential survival on the glasdegib + LDAC subset. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lin_2020_glasdegib_AML_overall_survival.html">Glasdegib treatment (Lin 2020)</a> </td> <td style="text-align:left;"> Parametric exponential time-to-event model for overall survival in older adults (>= 55 y) with newly diagnosed AML who were ineligible for intensive chemotherapy in BRIGHT AML 1003 Phase 2 (Lin 2020). Treatment-response analysis: hazard h(t) = lambda * (1 + theta_ldac_alone * LDAC_alone), with lambda the daily death hazard for glasdegib + LDAC (reference) and theta_ldac_alone the proportional increase in hazard for LDAC alone. Treatment arm was the only covariate retained after SCM backward elimination (alpha < 0.001); demographics, baseline safety labs, and disease characteristics were not significant. Original NONMEM run was fit with no estimated IIV and no residual error (likelihood-based parametric TTE). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hong_2013_glucose_insulin.html">Glucose insulin HGC (Hong 2013)</a> </td> <td style="text-align:left;"> Integrated glucose-insulin population pharmacodynamic model for the hyperglycemic glucose clamp (HGC) test in adults with type 2 diabetes mellitus (Hong 2013). Built on the Silber 2007 glucose-insulin framework with two modifications: (1) endogenous glucose production GP is constant and derived at steady state from baseline glucose (GCss) and insulin (ICss) so that feedback control of GP is suppressed in DIS_DIAB (Silber's GP-feedback estimate was close to zero in DIS_DIAB); (2) the first-phase insulin response is captured by an empirical Gaussian secretion pulse (amplitude Amplitude, peak time Tsec fixed at 3.54 min, width Tdur fixed at 1.76 min) rather than a NONMEM bolus. The second-phase insulin response rises linearly with time through a constant gamma multiplied by the elevation of glucose above its steady state (rectified at zero). Glucose is eliminated via insulin-independent clearance (CLG) and insulin-dependent clearance (CLGI_HGC) gated by an effect-compartment insulin concentration ICE (rate constant kIE). Insulin follows first-order elimination (CLI / VI). Palosuran 125 mg b.i.d. was the investigational drug; the paper concludes it has no clinically meaningful effect on insulin secretion or sensitivity, so the published final estimates set the palosuran treatment effect to zero -- the structural model below is the drug-free glucose-insulin homeostasis model in DIS_DIAB. VI is fixed at 6.09 L (Silber 2007 DIS_DIAB literature value) because the within-study estimate (0.52 L) was not physiologically meaningful and biased CLI. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hong_2013_glucose_insulin.html">Glucose insulin MTT (Hong 2013)</a> </td> <td style="text-align:left;"> Integrated glucose-insulin population pharmacodynamic model for the meal tolerance test (MTT) in adults with type 2 diabetes mellitus (Hong 2013). Companion model to Hong_2013_glucose_insulin_HGC: shares the glucose-disposition and effect-compartment structure but replaces the intravenous glucose-clamp input with an oral-meal absorption arm and replaces the biphasic insulin-secretion equation with a power-function plus incretin Emax form. Glucose absorption from the meal is modelled as a continuous Savic 2007 analytical transit compartment chain (rxode2 transit(n, mtt, BIO)) with non-integer n = 0.781, mean transit time MTT = 62.5 min, and a 'bioavailability' BIO (0.252) multiplied by a dummy meal dose of 100 g of glucose. The incretin effect (insulin secretion triggered by oral glucose intake via GLP-1) is captured by an Emax function of the instantaneous absorption rate ABSG: Emax = 2.02, ABSG50 = 14.8 mg/min FIXED to the Jauslin 2007 literature value because Emax / ABSG50 were correlated. Insulin secretion has no first-phase pulse (in contrast to the HGC model); the second-phase response is a power function of (glucose_concentration / glucose_baseline) with exponent IPRG = 3.06, modulated by the incretin factor. The disposition parameters CLG, VG, CLI, VI, kIE are FIXED at the HGC point estimates per the paper's Results section, on the assumption that after accounting for the incretin effect the disposition parameters should not markedly differ between HGC and MTT. Palosuran 125 mg b.i.d. was the investigational drug and was found to have no clinically meaningful effect, so the published final estimates set the palosuran treatment effect to zero -- the structural model below is the drug-free glucose-insulin homeostasis model in DIS_DIAB under MTT conditions. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Denti_2010_glucoseMinimal.html">GlucoseMinimal (Denti 2010)</a> </td> <td style="text-align:left;"> Bergman glucose minimal model fitted to insulin-modified IVGTT data from 204 healthy adults aged 18-87 years (Mayo Clinic cohort, Basu 2003/2006); population NLME implementation with mean-centred covariate effects of age, sex, visceral abdominal fat (CT-derived), percent total body fat (DEXA-derived), basal glucose, and basal insulin on the four mechanistic parameters glucose effectiveness (SG), insulin sensitivity (SI), insulin-action rate constant (P2), and apparent glucose distribution volume per kg body mass (VOL). Plasma insulin is supplied as a time-varying error-free forcing function (regressor INS). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/You_2013_hCG_GTN.html">HCG GTN (You 2013)</a> </td> <td style="text-align:left;"> Population kinetic biomarker decline model for serum human chorionic gonadotrophin (hCG) during 8-day methotrexate (MTX) therapy in low-risk gestational trophoblastic neoplasia (GTN). hCG decays mono-exponentially from an initial amplitude hCG0 toward a non-zero asymptote hCGres (the modelled residual production attributable to MTX-insensitive tumour cells). The same kinetic structure recasts as an indirect-response / turnover model with kout = K and zero-order production kin = K * hCGres, so the canonical lkout / lrbase names are used. No drug PK is modelled (the fixed 8-day MTX/folinic-acid regimen is implicit). Inter-individual variability: Box-Cox transformed eta on the hCG0 amplitude (shape -0.207) and log-normal etas on lkout and lrbase. Proportional residual error. M3 censoring of BLQ titres (< 2 IU/L) used in the original fit but is not reproduced here -- the packaged model is intended for typical-value + IIV simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhou_2024_HER2breastCancer_qsp.html">HER2breastCancer qsp (Zhou 2024)</a> </td> <td style="text-align:left;"> QSP. Quantitative systems pharmacology model of HER2-positive metastatic breast cancer coupling the ErbB receptor network to PI3K/AKT, Ras/MAPK and BTK signal transduction and to tumour growth, with five clinically approved therapeutics built in: the antibody-drug conjugates T-DM1 and T-DXd, the tyrosine kinase inhibitors lapatinib and pyrotinib, and capecitabine (with its 5'DFCR / 5'DFUR / 5-FU metabolite cascade). 54 ODE states and 79 reactions: 18 PK states (ADC, naked antibody, released payload, drug-antibody ratio, and the two TKI and capecitabine disposition chains) plus 36 tumour-cell states (EGFR / HER2 / HER3 / HER4 monomers, EGF- and NRG1-driven homo- and heterodimers and their phosphorylated forms, Raf / ERK / PI3K / AKT / BTK phospho-cycles, an AKT-driven negative feedback on HER3 synthesis, receptor- mediated ADC binding, internalisation and payload release, and a logistic tumour-growth state). Default parameterisation is the in vivo SKBR3 mouse xenograft with T-DM1; in vitro and T-DXd, BT-474, NCI-N87 and ZR-75-1 alternates are given in the per-parameter comments and in population$notes. All parameters are literature constants or the authors’ pattern-search calibration estimates; the paper reports no standard errors, no inter-individual variability and no residual-error model, so the model is deterministic.
HIV HCV CD4 recovery (Majekodunmi 2017) Longitudinal disease-progression / immune-reconstitution model for age-standardised CD4 T-cell counts (z-scores) in HIV-infected children receiving antiretroviral therapy (ART), with HIV/HCV coinfection slowing the recovery rate (Majekodunmi 2017, fit to 401 children – 355 HIV monoinfected and 46 HIV/HCV coinfected – from the Ukraine Paediatric HIV Cohort Study and the EPPICC HIV/HCV coinfection study across 8 European countries). The age-standardised CD4 z score is modelled as an asymptotic recovery curve z(t) = asy + (int - asy) * exp(-krec * t), where t is duration on ART, int is the pre-ART z score, asy is the long-term z score, and krec (the paper’s symbol c) is the per-subject recovery rate (1/year; ln(2)/krec is the time to half the total recovery from int to asy; renamed from the paper’s c to avoid shadowing R’s built-in c() combine function). Age at start of ART (centred at 4.3 years – the all-cohort median per Table 2 footnote) shifts both asy and int (younger children start higher and reach higher long-term levels). EPPICC enrollment country shifts int with Ukraine as the implicit reference. HCV coinfection is a multiplicative fractional reduction on krec: krec_coinf = krec_mono * (1 + e_hcv_pos_krec * HCV_POS) with e_hcv_pos_krec = -0.77, so coinfected children recover at krec = 1.55 * 0.23 = 0.357 /year (half-time ~2 years) versus 1.55 /year for monoinfected (half-time ~0.45 year). Disease-progression model with no drug dosing – the population were on combination ART (most commonly lamivudine + zidovudine + lopinavir/ritonavir) and the model’s t = 0 is the time of ART initiation.
igg kim 2006 Immunoglobulin G (IgG) model for nonlinear metabolism in healthy subjects
Imipenem tobramycin (Landersdorfer 2018) In vitro (carbapenem-resistant Acinetobacter baumannii FADDI-AB034). Mechanism-based pharmacodynamic model for an imipenem-plus-tobramycin combination in a 7-day hollow-fiber infection model (HFIM). Three pre-existing bacterial subpopulations differing in imipenem and tobramycin susceptibility (population 1 IPM-S/TOB-S, population 2 IPM-R/TOB-I, population 3 IPM-I/TOB-R) each follow a Bulitta two-state life-cycle growth model (S1 -> S2 -> 2*S1 with replication rate k21 fixed and per-subpopulation mean generation time 1/k12). Logistic carrying-capacity attenuation applies to the replicating step. Imipenem kills with a sigmoidal Hill function (Hill = 3 fixed) and tobramycin with an Emax function; mechanistic synergy is encoded as a discrete 70-fold reduction of the imipenem KC50 against population 3 when the tobramycin concentration meets or exceeds 1.15 mg/L (i.e. tobramycin permeabilizing the outer bacterial membrane toward imipenem). Imipenem and tobramycin concentrations are external time-varying inputs (covariates Cipm and Ctob); the model contains no human PK component.
Imipenem tobramycin (Yadav 2017) Preclinical (mouse, neutropenic murine thigh infection model; clinical Pseudomonas aeruginosa isolate FADDI-PA088). Mechanism-based pharmacodynamic model for an imipenem-plus-tobramycin combination with humanized dosing schemes. Two pre-existing bacterial subpopulations differing in imipenem and tobramycin susceptibility (population 1 IPM-S/TOB-R, population 2 IPM-I/TOB-R) each follow a Bulitta two-state life-cycle growth model (S1 -> S2 -> 2S1 with replication rate k21 fixed; mean generation time 1/k12 shared across the two subpopulations). A plateau factor PLAT = 1 - CFUall/(CFUall + CFUmax) attenuates the replication step so the no-drug viable count plateaus at CFUmax. Imipenem kills with a sigmoidal Hill function whose effective KC50 is multiplied by the mechanistic-synergy factor OM_effect = 1 - Imax,OM,TOB Ctob / (Ctob + IC50,OM,TOB) (i.e. tobramycin permeabilizing the outer bacterial membrane toward imipenem). Both subpopulations are tobramycin-resistant, so tobramycin has no direct killing term in this model; its only role is to lower the effective imipenem KC50 via OM_effect. Imipenem and tobramycin concentrations are external time-varying inputs (covariates Cipm and Ctob); the model contains no rodent PK component (the paper imported the murine one-compartment PK of imipenem and tobramycin from external references and the PK parameter values are not reported in the source on disk).
Immunogenicity qsp (Chen 2014) QSP (multiscale immunogenicity). Chen 2014 theoretical mechanistic model of antidrug-antibody (ADA) formation against therapeutic proteins: subcellular antigen presentation on dendritic cells, cellular T-helper and B-cell activation / proliferation / differentiation, and whole-body antigen and ADA disposition. Encoded with 1 T-epitope (j=1), 1 lumped effective binding MHC-II (representing 3 human MHC-II binders of the paper demo, all sharing the same affinity), 1 lumped ADA affinity class (i=1) and 1-compartment antigen PK – deterministic simulation with human parameter set. See vignette for the paper’s polyclonal 17-clone / 6-MHC-II full form and the reasons for the reduction.
indirect 0cpt transitEx Two compartment PK model with Michealis-Menten clearance using differential equations
indirect 1cpt inhi kin One compartment indirect response model with inhibition of kin.
indirect 1cpt inhi kin CLV One compartment indirect response model with inhibition of kin.
indirect 1cpt inhi kin r0rmaxcrmax One compartment indirect response model with inhibition of kin.
indirect 1cpt inhi kout One compartment indirect response model with inhibition of kout.
indirect 1cpt inhi kout CLV One compartment indirect response model with inhibition of kout.
indirect 1cpt inhi kout r0rmaxcrmax One compartment indirect response model with inhibition of kout.
indirect 1cpt stim kin One compartment indirect response model with stimulation of kin.Parameterized using rate cosntants
indirect 1cpt stim kin CLV One compartment indirect response model with stimulation of kin.
indirect 1cpt stim kin r0rmaxcrmax One compartment indirect response model with stimulation of kin.
indirect 1cpt stim kout One compartment indirect response model with stimulation of kout.Parameterized using rate cosntants
indirect 1cpt stim kout CLV One compartment indirect response model with stimulation of kout.
indirect 1cpt stim kout r0rmaxcrmax One compartment indirect response model with stimulation of kout.
indirect circ 1cpt inhi kin kin t One compartment indirect response model with inhibition of kin and circadian kin_t.
indirect circ 1cpt inhi kin kout t One compartment indirect response model with inhibition of kin and circadian kin_t.
indirect circ 1cpt inhi kout kin t One compartment indirect response model with inhibition of kout and circadian kin_t.
indirect circ 1cpt inhi kout kout t One compartment indirect response model with inhibition of kout and circadian kin_t.
indirect circ 1cpt stim kin kin t One compartment indirect response model with stimulation of kin and circadian kin_t.Parameterized using rate cosntants
indirect circ 1cpt stim kin kout t One compartment indirect response model with stimulation of kin and circadian kout_t.Parameterized using rate constants
indirect circ 1cpt stim kout kin t One compartment indirect response model with stimulation of kout and circadian kin_t.Parameterized using rate cosntants
indirect circ 1cpt stim kout kout t One compartment indirect response model with stimulation of kout and circadian kout_t.Parameterized using rate cosntants
indirect prec 1cpt inhi CLV One compartment precursor-dependent indirect response model with inhibition of drug response. Parameterized with clearance and volume. (effect).
indirect prec 1cpt inhi r0rmaxcrmax One compartment precursor-dependent indirect response model with inhibition of drug response (effect).
indirect prec 1cpt stim CLV One compartment precursor-dependent indirect response model with inhibition of drug response (effect). Parameterized with clearance and volume
indirect prec 1cpt stim r0rmaxcrmax One compartment precursor-dependent indirect response model with inhibition of drug response (effect). Parameterized with clearance and volume
Interleukin6 cyp pbpk (Willemin 2024) PBPK (reduced from Simcyp Simulator V21). Interleukin-6 (IL-6) disposition driving concentration- and time-dependent modulation of five hepatic cytochrome P450 activities (CYP1A2, 2C9, 2C19, 3A4, 3A5), developed to assess the drug-interaction risk created by the transient IL-6 elevation of cytokine release syndrome after teclistamab (BCMAxCD3 bispecific antibody) step-up and first treatment dosing in MajesTEC-1. Teclistamab itself is never modelled: because IL-6 is endogenous, its appearance is represented by a hand-calibrated series of six zero-order IV IL-6 infusions whose rates were adjusted to recover the observed MajesTEC-1 IL-6 profile, so the model is an IL-6 exposure driver rather than a teclistamab PK model. IL-6 is described as a one-compartment IV model (Vss 0.43 L/kg, CLiv 1 L/h) reduced from the Simcyp minimal-PBPK topology. Each CYP activity follows the Machavaram enzyme-turnover equation d(E)/dt = kdeg * (1 + (Indmax - 1) * [IL-6] / (IndC50 + [IL-6]) - E) with activity relative to an untreated baseline of 1, suppressing activity for CYP2C9, 2C19, 3A4 and 3A5 and inducing it for CYP1A2. The downstream victim-drug exposure ratios (caffeine, s-warfarin, omeprazole, midazolam, cyclosporine, simvastatin) are NOT part of this model: those used proprietary Simcyp V21 compound files whose in vivo dispositions cannot be reconstructed from the published inputs.
Interleukin6 cyp3a pbpk (Chen 2024) PBPK (reduced from Simcyp Simulator V17). Interleukin-6 (IL-6) disposition driving concentration- and time-dependent suppression of CYP3A enzyme activity in liver and gut, developed to assess the CYP3A drug-drug interaction risk created by the transient IL-6 elevation that follows mosunetuzumab (CD3/CD20 bispecific antibody) dosing. IL-6 is described as a one-compartment intravenous model (CL 2.8 L/h determined top-down; Vss 0.43 L/kg) reduced from the Simcyp minimal-PBPK-plus-single-adjusting-compartment topology, and is driven by hypothetical zero-order IL-6 infusions rather than by mosunetuzumab pharmacokinetics (mosunetuzumab itself is never modelled). CYP3A activity in each tissue follows the Machavaram enzyme-turnover equation d(ENZact)/dt = kdeg * ENZ0 * [1 + (Emin - 1) * [IL-6] / (EC50 + [IL-6])] - kdeg * ENZact with in vitro suppression constants EC50 43.7 pg/mL and Emin 0.217 (geometric means over five human hepatocyte donors, Dickmann 2011) and Simcyp library degradation rate constants kdeg 0.0193/h (liver) and 0.03/h (gut). The tissue-specific kdeg values are what separate the liver and gut responses to a transient IL-6 pulse. The downstream midazolam and simvastatin exposure-ratio predictions are NOT part of this model: those used unmodified proprietary Simcyp V17 compound files whose in vivo clearances cannot be reconstructed from the published inputs.
Intratarget microdosing pbpk (Aoki 2024) PBPK-PKRO (simplified whole-body, permeability-limited target tissue). Drug-agnostic small-molecule model used by Aoki, Rowland & Sugiyama (2024) to ask when Intra-Target Microdosing (ITM) – direct delivery of a Phase-0 microdose into the target tissue – can engage a target as well as a systemically administered therapeutic dose. Plasma exchanges with three perfusion-limited well-stirred tissues (muscle, skin, adipose), with a permeability-limited liver (hepatic extracellular space in series with the hepatocyte), and with a permeability-limited target tissue (extracellular space in series with the intracellular space). Both the liver and the target tissue take drug up by saturable active transport plus passive diffusion, return it by passive diffusion, and metabolise it by a saturable intracellular route; renal clearance removes drug from plasma. Inside the target tissue the unbound drug binds a finite receptor pool of total amount rtot, so fractional receptor occupancy (RO) – the paper’s measure of target engagement – is an explicit model output. The same structure serves both routes: an intravenous dose enters central, an ITM dose enters the target-tissue extracellular space. This is a Monte Carlo simulation study, not a fit: the paper draws 10,000 ‘virtual compounds’ from the log-normal parameter distributions of Table 2 and scores each one for ITM success (24-h average RO >= 60%). The values below are therefore the published MEDIANS of those distributions (Supplementary Table S1, Pctl. 50), i.e. a reference typical virtual compound rather than any single compound; every parameter is fixed() because none was estimated from data. The vignette reproduces the nine fully specified example compounds of Supplementary Table S2 and their published outcomes. See the vignette Errata for three published-table discrepancies, each settled by the authors’ own code.
Iron hepcidin (Angeli 2016) Joint turnover model of serum iron and serum hepcidin during the menstrual cycle in healthy non-menopausal women; both molecules follow first-order turnover with a menses-induced increase in elimination (kloss) shared across iron and hepcidin and a delayed post-menses rebound in synthesis (krelI, krelH) starting on day 2 of the cycle, with serum iron multiplicatively modulating hepcidin synthesis around the iron baseline.
Iv1cmt bql (Beal 2001) Methodology reference. One-compartment IV-bolus generative toy model from Beal 2001 (the M1-M7 BQL methods paper). Typical CL is fixed at 0.693 and Vd at 1 by the author so the time unit equals one half-life; a unit dose is administered at time 0. Encoded as the SI1 population simulation model with geometric SD 0.2 on both CL and Vd and homogeneous residual variance 0.02.
Likert pain (Schindler 2017) Population PD model of 11-point Likert pain score (0-10) in adults with painful distal diabetic neuropathy (placebo arm of three phase III studies), using the minimal continuous-time Markov model (mCTMM) form of Schindler & Karlsson 2017. Eleven ordered-categorical states (scores 0-10) evolve via ODEs (Eq. 6) with transfer rate constants (Eqs. 8-9) parameterised through a mean equilibration time (MET) and the steady-state score probabilities Pss,k. Steady-state cumulative probabilities Pss(Y >= k) follow a proportional-odds model driven by a linear effect of time (on logit scale, accounting for placebo effect) and a shift for concomitant acetaminophen use (Eq. 13). MET is itself time-varying with a linear slope on t (in weeks), consistent with the time-dependent increase in score stability reported in the paper. The model has no PK and no drug-exposure input – the placebo effect is modelled directly on the score-probability logit.
Linezolid meropenem vancomycin (Wicha 2017) In vitro (MSSA ATCC 29213). Semimechanistic time-kill pharmacodynamic model of linezolid, meropenem, and vancomycin against methicillin-susceptible Staphylococcus aureus. Bacterial life cycle has three states: growing (gro), replicating (repl), and persisting (pers). LZD inhibits the GRO->REP transition (bacteriostatic via krep) and induces a replication-independent killing rate kdeath_lzd on growing bacteria. MER and VAN, as cell wall-active antibiotics, impair successful doubling at the REP->GRO transition; the joint MER+VAN action is encoded as a modified Bliss-independence term that includes the paradoxical Eagle-effect self-inhibition of MER at high concentrations and the VAN Emax cap. Drug-unsusceptible persisters are generated during replication at rates kper_mer * E_MER and kper_van * E_VAN, then die at kdeath_per. An adaptive-resistance submodel (Tam 2005) inflates the effective EC50 of MER and of VAN over time via fractional ARon states; subinhibitory VAN concentrations inhibit the MER-adaption rate (monodirectional VAN-on-MER PD interaction). MER and VAN solution concentrations decay first-order due to chemical degradation in growth medium (rates fixed from HPLC measurement); LZD is stable. The model is in-vitro PD only – there is no human PK component; drug exposures are static dosing at t = 0. Random effects (eta) are NOT present: the paper reports replicate-only experimental variability and uses an additive residual error on log10(CFU/mL).
M3M3FBS (Grzesk 2016) Preclinical (rat tail artery, ex vivo). Sigmoidal Emax concentration-response (CRC) model of perfusion pressure in the isolated, perfused male Wistar rat tail artery, parameterised for four mechanistically distinct vasoactive agonists (phenylephrine, arg-vasopressin, mastoparan-7, Bay K8644) in the absence or presence of the phospholipase-C activator m-3M3FBS (1e-5 M/L pretreatment). AGONIST_CODE (1..4) selects the agonist; M3M3FBS_PRESENT (0/1) selects the control vs +m-3M3FBS (EC50, Emax) sub-pair. CONC_AGONIST_M is the applied agonist concentration (M/L). The model is static (no time evolution, no PK), encoding only the per-condition concentration- response relationship reported in Tables I and II. OUT-OF-SCOPE caveats (see vignette Errata): the source paper is an ex-vivo concentration-response study that does NOT report a Hill coefficient, IIV / between-subject variability in the nlmixr2 sense, or a residual-error structure. The Hill coefficient is fixed to 1 as the structural minimum, and the residual SD is a placeholder (propSd fixed at 0.10) so the model is nlmixr2-fit-compatible.
MAb mouse pbpk (Liu 2023) PBPK (whole-body, 15-tissue Shah & Betts 2012 platform). Preclinical (mouse, female BALB/c, 28 g). Inter-antibody variability in monoclonal antibody plasma PK, with the antibody-specific pinocytosis-uptake coefficient F1 driven by a sigmoidal function of the antibody’s heparin-chromatography retention time (HEPARIN_RT) and the convective-transport coefficient F2 fixed at the panel median. Fitted simultaneously to IV plasma profiles of 53 aglycosylated human IgG1z SEFL2.2 antibodies dosed at 2 mg/kg.
MAb PBPK (Shah 2012) Shah and Betts 2012 platform PBPK model for monoclonal antibody plasma and tissue disposition - human (71 kg) parameter set, non-target-binding (nonspecific) mAb. 15 tissues x 6 sub-compartments (vascular plasma, vascular blood cells, endosomal unbound mAb, endosomal FcRn-bound mAb, endosomal free FcRn, interstitial) plus central plasma, central blood cells, and lymph node, totalling 93 ODE states. FcRn-mediated recycling is implemented in every tissue endosomal space.
MAb popPBPK iv (Kumar 2024) popPBPK (whole-body). Kumar 2024 population physiologically based pharmacokinetic model characterizing INTER-ANTIBODY variability in the clinical PK of monoclonal antibodies with linear disposition after INTRAVENOUS administration. The structure and the human (71 kg male) physiology are inherited unchanged from the Shah and Betts 2012 platform PBPK model (see Shah_2012_mAb_PBPK.R): 15 tissues x 6 sub-compartments (vascular plasma, vascular blood cells, endosomal unbound mAb, endosomal FcRn-bound mAb, endosomal free FcRn, interstitial) plus central plasma, central blood cells and a lymph node, totalling 93 ODE states, with FcRn-mediated recycling in every tissue endosomal space. Kumar 2024 re-estimated two of the four Shah and Betts system parameters - pinocytic uptake per unit endosomal space (CLup) and endosomal degradation of FcRn-unbound mAb (kdeg) - by fitting the model to multiple-dose-level IV profiles of 43-44 mAbs in Monolix 2021R1, treating EACH ANTIBODY as an individual. The random effects on CLup and kdeg are therefore INTER-ANTIBODY variability, not between-subject variability: sampling them draws a new hypothetical antibody, not a new patient. Kumar 2024 also reduced the central plasma and blood-cell volumes to 1412 and 1155 mL (total blood volume minus the vascular volume held in the tissues), which is the one physiological difference from the Shah and Betts parameterization. All remaining system parameters (FcRn concentration, FcRn on/off rates, recycled fraction FR, reflection coefficients, organ volumes and flows) are FIXED at the Shah and Betts 2012 human values.
MAb popPBPK sc (Kumar 2024) popPBPK (whole-body). Kumar 2024 population physiologically based pharmacokinetic model characterizing INTER-ANTIBODY variability in the clinical PK of monoclonal antibodies with linear disposition after SUBCUTANEOUS administration. This is the intravenous model (Kumar_2024_mAb_popPBPK_iv.R) with the skin bifurcated into a small injection-site ‘subcutaneous’ compartment (2.25 mL interstitial space, sized to a typical ~2 mL SC injection) and a ‘rest of skin’ compartment, giving 16 tissues x 6 sub-compartments plus central plasma, central blood cells and a lymph node - 99 ODE states. The SC dose is administered into the interstitial space of the subcutaneous compartment (cmt = ‘is_subcutaneous’), NOT into an empirical first-order depot. Two absorption parameters are added there: a first-order local degradation rate kSC representing pre-systemic catabolism at the injection site, and a scaling factor S_LU multiplying the convective lymphatic flow that carries mAb from the SC interstitium to the lymph node. Both were estimated in Monolix 2021R1 by fitting to the SC profiles of the 16 mAbs that also had IV data, with the antibody-specific CLup and kdeg values fixed as regressors from the IV fit. All four random effects (CLup, kdeg, kSC, S_LU) are INTER-ANTIBODY variability, not between-subject variability: sampling them draws a new hypothetical antibody, not a new patient. Physiology is the Shah and Betts 2012 human (71 kg male) parameter set, with the skin and subcutaneous rows taken from Kumar 2024 Table 1.
Mab7E3 (Cao 2013) Preclinical mPBPK model for the murine anti-platelet IgG1 mAb 7E3 in mice (Cao 2013 Model A; clearance from plasma)
Mab8C2 (Cao 2013) Preclinical mPBPK model for the murine anti-topotecan IgG1 mAb 8C2 in mice (Cao 2013 Model A; clearance from plasma)
Meropenem gentamicin ciprofloxacin (Sadouki 2025) In-vitro static-time-kill pharmacodynamic model for two- and three-way combinations of meropenem, gentamicin, and ciprofloxacin against Escherichia coli NCTC 12,241. Logistic bacterial growth (knet, Bmax) is killed by an Emax-Hill function for each drug; emergence of regrowth is captured by a time-decay term parameterised by BETA (loss of effect) and TAU (time-shape). Meropenem chemical degradation in CAMHB at 37.5 C is embedded as a first-order decay of the meropenem solution concentration. Drug-drug interactions are encoded as: a fixed -1 categorical shift on BETA whenever a 2- or 3-way combination is present (so the regrowth term reverses sign and effect is sustained), proportional reductions of -0.353 and -0.576 in ciprofloxacin IC50 in the presence of meropenem and gentamicin respectively (synergy on potency), and concentration-dependent Emax shifts of BETA for gentamicin and ciprofloxacin. The model is in-vitro PD only – there is no human or animal PK component; bacterial counts (CFU/mL) are observed on log scale.
MethionineMetabolismCycle (Han 2018) Preclinical (rat). Seven-compartment mechanistic methionine metabolism cycle (MMC) model in Zucker Diabetic Fatty (ZDF) rats vs non-diabetic controls; predicts plasma methionine and homocysteine after IV methionine.
MRNA translation (Frohlich 2018) QSP (mechanistic, single-cell, in vitro). Two-state gene-expression model with ribosomal rate-limited translation (model ii of Frohlich et al. 2018) for eGFP / d2eGFP fluorescent reporter expression after mRNA transfection in HuH7 hepatoma cells, fit by multi-experiment nonlinear mixed-effects modelling (MEMOIR + AMICI).
MRNA vaccine QSP (Selvaggio 2021) QSP. Preclinical (rhesus macaque, calibration). Multicompartmental quantitative systems pharmacology model of mRNA vaccine immunogenicity: kinetics of a lipid nanoparticle (LNP) carrying mRNA at the injection site, uptake and antigen expression by dendritic cells (myeloid + plasmacytoid), monocytes and neutrophils, migration to the draining lymph node, T-cell and B-cell adaptive immune response, and antibody production. Adapted and extended from the Chen 2014 immunogenicity backbone (adaptive immune response retained unvaried; Selvaggio 2021 supplement ‘Model equations’). Innate-cell recruitment / uptake / migration parameters fitted to Liang 2017 rhesus macaque data (fluorescent-protein-encoding modified mRNA-LNP). Time in days; LNP in number of molecules; cell states in cell counts; antibody in pmole. Delivered as a mean-field extraction with a single mean-affinity B-cell subgroup (Chen 2014 encodes 17 affinity-subclones for the BCR distribution; the single-subgroup extraction with mean Ka is a documented deviation, see vignette Assumptions) and an explicit low-antibody closure for the receptor-competition equation (paper Eq for total antigen requires a Newton solve on Ag_f per timestep; not natively expressible in rxode2).
MRNALNP (Parhiz 2024) Preclinical (mouse, C57BL/6, ~25 g). Whole-body PBPK / luciferase-expression model for systemically administered firefly-luciferase mRNA delivered in lipid nanoparticles (bare, untargeted parameterization). Six anatomical regions (blood, lung, heart, kidney, spleen, liver) plus portal-organ and carcass remainder vasculature; each of the five major tissues additionally tracks an intracellular LNP pool, a translatable mRNA pool, and a luciferase signal. The luciferase observation uses the bare-LNP homogenate-assay parameter set (LU/mg protein); the vignette documents how to switch to the bare-LNP BLI, IgG-coated, and PECAM-targeted parameterizations from paper Tables 1-3.
MultipleSclerosis (VelezdeMendizabal 2013) Discrete count disease-dynamics model for the number of contrast-enhancing lesions (CELs) per monthly T1-weighted post-contrast MRI in relapsing-remitting multiple sclerosis. Negative binomial with first- and second-order Markov dependence on the previous monthly CEL counts (‘NB nested MAK2 with steroid effects’ from Velez de Mendizabal 2013): per-month expected CEL count lambda = lambda_0 + PDV * theta_pdv_eff + PPDV * theta_ppdv, where PDV and PPDV are the observed CEL counts at the previous and second-previous monthly MRIs and theta_pdv_eff is the typical theta_PDV (= 0.447) in months with no corticosteroid administration and theta_PDV_S (= 0.145, about 67 percent lower) in months in which the patient received corticosteroids for a clinical relapse. The source publication uses a negative-binomial likelihood with mean lambda and overdispersion OVDP (variance = lambda * (1 + OVDP * lambda)); following the ddmore/Plan_2012_pain.R and ddmore/Schoemaker_2018_levetiracetam.R precedents, the nlmixr2 observation is declared as Poisson(lambda) for fitting-API compatibility, and OVDP is exposed as a derived model variable so a downstream user can post-process Poisson samples through a negative-binomial correction if they need the full source dispersion. No drug PK is dosed in this model; the corticosteroid effect enters as a binary per-record covariate.
Nafld risk (Oniki 2018) Population BMI-driven risk-prediction model for nonalcoholic fatty liver disease (NAFLD) prevalence in 341 elderly Japanese health-screening participants (Oniki 2018). The logit of the probability of NAFLD is a fixed baseline floor (-5) plus a sigmoidal-Emax function of (BMI - 17) with Hill exponent 3.43 (Eqs. 2-4 of the source paper). Covariates female sex, HDL-C and LDL-C act on the Emax of the logit (logit_max); the PNPLA3 rs738409 heterozygote and homozygote indicators and HbA1c act on the (BMI50 - 17) half-saturation offset. BMI is clamped to [17, 30] kg/m^2 before entering the sigmoidal term (s011 dataset column BMI_A). No between-subject random effect is estimated (ETA1 FIX 0) and the Bernoulli-likelihood fit (METHOD=COND LAPLACE LIKELIHOOD) reports no $SIGMA residual. The fit was performed with NONMEM 7.2.0 (s011 control stream); MINIMIZATION SUCCESSFUL HOWEVER, PROBLEMS OCCURRED WITH THE MINIMIZATION, with R and S matrices algorithmically singular; OBJ 1193.995. The published 95% CIs are bootstrap-derived (984/1000 successful runs) per the source supplement. Companion BMI prediction model in Oniki_2018_bmi.
oncology sdm lobo 2002 Signal transduction model for delayed concentration effects on cancer cell growth
oncology xenograft simeoni 2004 Oncology tumor growth model in xenograft models
Osteoporosis placebo qsp (Berkhout 2015) QSP. Systems-pharmacology disease-progression model of the placebo (calcium 500 mg/day) arm of postmenopausal osteoporosis. Applies the reduced Lemaire bone-remodeling core (Schmidt 2011 / Post 2013) to 470 women from the placebo arm of the EPIC study, evaluated over 4 years with baseline years-since-menopause (YSM) spanning 0.5-27 yr. The mechanistic core has two dimensionless ODE states (relative active osteoblast y, relative active osteoclast z, both = 1 at menopause onset) driven by (i) an exponential estrogen-loss disease progression function f(t) = exp(-k_estrogen * t) and (ii) a placebo (calcium) inhibition factor PCa acting from t_start (subject-specific YSM at study entry) as a delayed-onset / slow-offset dip: PCa = 1 - PLAC * (1 - exp(-k_Ca_onset * (t - t_start))) * exp(-k_Ca_offset * (t - t_start)). The core drives four biomarkers: algebraic transducers for urine N-telopeptide (NTX = NTX0 * z^q_NTX) and serum bone-specific alkaline phosphatase (BSAP = BSAP0 * (1 + k_BSAP0) * y^q_BSAP; k_BSAP0 rescales the fixed tibolone-study U/L baseline to the EPIC ng/mL assay), and two indirect-response ODEs for lumbar-spine and total-hip BMD driven by osteoblast (production coefficient D_AOB) and osteoclast (degradation coefficient D_AOC). BMD baselines are BMI-corrected around the cohort median BMI of 25.34114 kg/m^2. IIV is a 2 x 2 block on (NTX0, BSAP0) plus a 2 x 2 block on (BMD_LS_0, BMD_TH_0); residual error is log-normal on every biomarker. The 5 system parameters (z_s, k_B, k_estrogen, D_A, b) are fixed to the tibolone-study (Post 2013) values.
OvarianCancer ca125 (Wilbaux 2014) K-PD joint model of tumour size and CA-125 kinetics during platinum-based chemotherapy in recurrent ovarian cancer (CALYPSO phase III trial; carboplatin + paclitaxel or carboplatin + pegylated liposomal doxorubicin pooled)
Oxaliplatin platelet dynamics (PerezRuixo 2015) Semi-mechanistic PD-only platelet-dynamics model for peritoneal carcinomatosis (PC) patients treated with cytoreductive surgery (CRS) alone or CRS followed by hyperthermic intraperitoneal oxaliplatin (HIO). Extends the Harker (2000) cytokinetic model with (i) a self-renewing megakaryocyte progenitor pool (prol) coupled to a (PLT0/PLT)^gamma feedback on the proliferation rate, (ii) a five-compartment Friberg-Bulitta transit chain for platelet aging (transit1 -> transit2 -> … -> transit5) with a random-destruction rate ks on each transit compartment, (iii) a fixed release factor eta = 4000 platelets per maturing megakaryocyte, (iv) a transient post-surgery stimulation of prol proliferation with maximum effect SPmax that attenuates first-order with rate kp starting at surgery time t = 0, and (v) a power-function drug effect E_drug = alpha * CP_OXA_MGL^beta that inhibits prol proliferation. The paper does NOT develop an oxaliplatin PK model of its own; CP_OXA_MGL is obtained from empirical Bayes estimates of an upstream popPK model (Perez-Ruixo 2013 Cancer Chemother Pharmacol 71:693-704, cited as reference 15 – not packaged in nlmixr2lib), and is consumed here as an exogenous time-varying covariate column (oxaliplatin plasma concentration in mg/L). Prior splenectomy multiplies ks by Phi = 0.475, prolonging platelet lifespan from 3.23 to 7.78 days. Age, body surface area, sex, total proteins and HIO carrier solution were tested and dropped. The paper additionally models platelet transfusions as a bolus of TRF0 = 255 x 10^9/L platelet-count-equivalent decaying first-order with rate kt = 0.104/h (t1/2 = 6.66 h); this transfusion sub-model is NOT retained in the packaged model file (only 5 / 80 patients received transfusions in the source cohort, kt has the highest RSE 53.8% in the model, and the paper itself notes ‘this finding should be interpreted with caution given the limited number of patients who received transfusions’). Downstream users who need transfusion simulation can layer an exogenous decay on the observation.
Paclitaxel everolimus dasatinib (Ande 2018) In vitro (JIMT-1 HER2-positive trastuzumab-resistant breast cancer cell line) three-dimensional and dynamic (3DD) BelloCell-bioreactor PK/PD model for the sequential triple combination of paclitaxel (PAC) + everolimus (EVE) + dasatinib (DAS). PAC follows a two-compartment mammillary PK model (CL/Q/Vc=Vp_paper/Vp=Vt_paper) driven by a 3-h IV infusion. EVE and DAS share a single combined concentration covariate (their molar concentrations are equal throughout the experiment) constructed inside model() as a step-function-plus-exponential-decline: 0 nM for t<24 h, 50 nM during 24<=t<96 h, then exponential decline with rate kde from t>=96 h derived from the BelloCell perfusion mechanics (0.53 mL/min / 500 mL). PAC and DE concentrations jointly stimulate caspase-3 production via two paper-named slope parameters (spac, sde) entering a 5-state Lobo-Balthasar signal-transduction chain (4 transit relays + 1 active caspase-3 pool) with a Mager-Jusko-style feedback loop in which the production rate of the first transit is divided by the active caspase-3 level. The change in active caspase-3 from baseline drives the kill term in a modified Simeoni (2004) tumor-growth model (exponential-then-linear growth with psi=20 FIXED) on JIMT-1 cell count. No IIV or residual error is reported in the paper (Table 3 reports % RSE on point estimates only); a small fixed additive residual error is included per output as a placeholder so the model is fittable, with values flagged in the vignette Assumptions and deviations section.
ParathyroidHormone (Ahn 2014) Semi-mechanistic indirect-response PD model of parathyroid hormone (PTH) suppression after oral calcium intake in healthy adults; absorbed but unobserved ionized calcium inhibits PTH secretion via an Emax (fixed to 1) negative-feedback term, with a parallel homeostatic indirect-response model for observed plasma ionized calcium.
Parkinson progression (Lee 2011) Disease-progression model for the change-from-baseline in total Unified Parkinson’s Disease Rating Scale (UPDRS) score over study time in early Parkinson’s disease, fit by Lee and Gobburu (2011) as the worked example of a Bayesian disease-drug-trial modeling methodology paper. The predicted change from baseline is the linear-disease-progression slope minus an asymptotic short-term symptomatic-effect dip: mu = slope * time - symeff * (1 - exp(-ke0 * time)). Both slope and symptomatic effect have separate placebo (beta0, gamma0) and active-drug (beta1, gamma1) terms switched on by a binary ON_TREATMENT indicator (0 = placebo, 1 = active drug arm). Between-subject variability is additive on the linear slope (etaslope) and on the linear symptomatic effect (etasymeff), with the two etas assumed independent in the source paper. Residual error is additive on the UPDRS scale. No drug PK input – the drug effect is encoded purely as a treatment-arm indicator. Time is in 4-week months (the units of the model’s parameters as fit; the source paper’s Table II column header ‘Delta UPDRS score/week’ appears to be a typo for ‘/month’ – see the validation vignette’s Assumptions and deviations section for the math that resolves this from the paper’s own published placebo prediction at week 26). Default parameter values in ini() are the TEMPO study rasagiline (1 or 2 mg/day pooled) arm, Table II non-informative-prior Bayesian posterior means. The validation vignette additionally documents the ELLDOPA study (carbidopa/levodopa) Table II fit and the TEMPO Table III power-prior sensitivity at weight parameter alpha_0 in {0.1, 0.5, 1.0}.
ParkinsonPathogenesis (Menon 2024) QSP. Combined ROS + proteasome-sequestration mechanistic model of Parkinson’s disease pathogenesis in dopaminergic neurons, from the Menon-Bakshi-Krishnan 2024 CPT:PSP Perspective. Five ODEs describe the intracellular time-evolution (hours) of misfolded alpha-synuclein monomer (alpha_syn), the larger alpha-syn aggregate species F (fibrils and oligomers lumped, fagg), free proteasome components (pfree), the proteasome-aggregate complex CP (pcomplex), and cytoplasmic reactive oxygen species (ros). The ROS module (positive feedback: misfolded alpha-syn drives cytoplasmic ROS via mitochondrial + DA-vesicle leak; ROS drives further alpha-syn misfolding) is combined with the proteasomal module (mutual inhibition: proteasome degrades aggregates, aggregates sequester proteasome). Combining these two nonlinear feedback modules yields the emergent landscape mapped in the paper: monostable or bistable alpha-syn (from the ROS positive feedback) x steady-state or oscillatory proteasome (supercritical or subcritical Hopf, from the proteasome negative feedback). All concentrations are non-dimensionalised (scale factor 50 pM); time is in hours. The basal cytoplasmic ROS production rate k0 is the bifurcation-input parameter the paper varies to demonstrate pulse- and step-triggered switching to the diseased state (elevated misfolded alpha-syn, sustained or transient proteasome-sequestration oscillations, timescale ~1 year). Default ini() values are the Figure 2(A) scenario (supercritical Hopf in the proteasomal module, no overlap with the bistable regime); the vignette walks through Figures 1 (individual modules), 2 (combined landscape), and 2(I) / S2 (proteasome-availability intervention). The proteasome-degradation rate kP is not tabulated in the paper’s parameter tables and is inferred here from the steady-state balance at healthy [P]_ss = 1 (see vignette Assumptions and deviations). This is a mechanistic ODE model without drug input, without IIV, and without residual error (typical-value simulation). A companion DA compartmentalization variant is packaged as Menon_2024_parkinsonPathogenesis_DA.
ParkinsonPathogenesis DA (Menon 2024) QSP. Dopamine compartmentalization + ROS-feedback variant of the Menon-Bakshi-Krishnan 2024 Parkinson’s-disease pathogenesis model. Four ODEs describe the intracellular time-evolution (hours) of misfolded alpha-synuclein monomer (alpha_syn), cytoplasmic reactive oxygen species (ros), vesicular dopamine (da_ves), and cytoplasmic dopamine (da_cyto). This variant REPLACES the proteasomal-module of the combined model (Menon_2024_parkinsonPathogenesis) with an explicit three-compartment DA metabolism representation (vesicular / cytoplasmic / extracellular; extracellular DA fixed as a boundary condition D_ex). Cytoplasmic DA contributes to the ROS-production term (parameter kDA), which now enters the same Hill-activated ROS-positive-feedback that drives alpha-synuclein misfolding. Misfolded alpha-synuclein aggregates in turn drive DA leakage from vesicles into the cytoplasm (parameter tr2 applied through the same Hill activation). The paper uses this variant to demonstrate that inhibiting DA transport from extracellular space into the cytoplasm (parameter tr3) raises the bistable-switching threshold in alpha_syn – a potential DA-transporter-targeting therapeutic mechanism. Default ini() values are the Figure S8(B) scenario (constant k_synth DA production, tr3 = 40000 h^-1 baseline high-transport regime). The alternative Figure S8(C) scenario with extracellular-DA-mediated feedback inhibition of DA synthesis (k_synth = alpha/(beta + D_ex)) is documented in the vignette. DA concentrations are in uM (paper: ‘DA concentrations are in uM’, Supinfo2 Section 1.6); alpha_syn and ROS remain dimensionless (scale factor 50 pM). This is a mechanistic ODE model without drug input, without IIV, and without residual error. Companion combined-model variant is packaged as Menon_2024_parkinsonPathogenesis.
Pdl1 nsclc mbma (Franzese 2026) MBMA. Sequential two-stage model-based meta-analysis (MBMA) of Objective Response Rate (ORR), Overall Survival (OS), and Progression-Free Survival (PFS) for programmed cell death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1) inhibitors in metastatic non-small cell lung cancer (mNSCLC). Franzese 2026 assembled 114 studies (46 unique treatments) from the Certara CODEX PD-(L)1 Solid Tumor (PD1ST) database. ORR is modeled by mixed-effects logistic regression with treatment-specific intercepts, an average PD-L1 expression effect (quadratic for PD-1 monotherapy, linear for PD-L1 monotherapy and any PD-(L)1 combination), a first-line-therapy effect, and a squamous-histology effect on chemotherapy (Franzese 2026 Table 1 and Table S4). OS and PFS are modeled by mixed-effects semi-parametric proportional hazards on monthly discrete hazard intervals: predicted ORR drives the log(HR) through five treatment-category-specific slopes (PD-(L)1 monotherapy, PD-(L)1 + chemotherapy, PD-(L)1 + other, chemotherapy only, other), with additional per-arm covariate effects for ECOG-PS-0 (OS on non-chemotherapy; PFS globally), squamous histology on chemotherapy, an Asian-race interaction on the OS ORR slope, and a PD-(L)1-monotherapy hazard-intercept shift in the PFS model. The reference baseline hazards are the 75 monthly OS conditional death probabilities (Franzese 2026 Table S2) and the 46 (grouped) monthly PFS conditional event probabilities plus 30 chemotherapy time-dependent baseline-hazard shifts (Franzese 2026 Table S2). All parameter values including the discrete-time baseline hazards are wrapped in fixed() because the model is a downstream user of the published fit, not a re-estimation of it. Simulation scope: per-arm ORR (dimensionless proportion), per-arm S_OS(t) and S_PFS(t) survival curves at monthly resolution over the 1-75 month window supported by the paper. The model is intended for reproducing published head-to-head trial simulations (Figures 1-5); it is NOT suitable for individual-subject trajectory simulation because both endpoints operate at the study-strata-arm level. Random effects are between-study-strata and NOT between-subject (see MBMA discipline in Franzese 2026 Methods 2.3.1 and 2.3.2).
Phenylalanine (Kaufman 1999) Mechanistic model of human phenylalanine metabolism (PAH hydroxylation with substrate activation, transaminase, net protein degradation) in normal subjects, PKU heterozygotes, and phenylketonuric patients
phenylalanine charbonneau 2021 Phenylalanine model for absorption and metabolism in healthy subjects and patients with PKU
Pioglitazone (Stringer 2015) Population pharmacodynamic cascading indirect-response model for fasting plasma glucose (FPG) and glycosylated hemoglobin (HbA1c) in Japanese type 2 diabetes mellitus (T2DM) patients receiving pioglitazone or a non-thiazolidinedione oral glucose-lowering drug background over 2.5-4 years. FPG follows a zero-order production / first-order elimination turnover with a proportional disease- progression term on production; a time-driven Emax model stimulates FPG elimination to capture titration-related exposure ramp-up. HbA1c is driven by a power function of FPG plus a separate FPG-independent zero-order input with a linear-in-time disease- progression component. No PK data were collected; the drug effect is treatment-cohort- and time-driven, distinguished by different Emax and ET50 values in the pioglitazone vs control cohorts (Stringer 2015 Diabetes Technology and Therapeutics).
PK 1cmt One compartment PK model with linear clearance
PK 1cmt des One compartment PK model with linear clearance using differential equations
PK 1cmt tmdd full One-compartment TMDD archetype with explicit drug-target binding (Mager & Jusko 2001 full model)
PK 1cmt tmdd mm One-compartment TMDD archetype, Michaelis-Menten (MM) approximation (Gibiansky et al. 2008)
PK 1cmt tmdd qss One-compartment TMDD archetype, quasi-steady-state (QSS) approximation (Gibiansky et al. 2008)
PK 2cmt Two compartment PK model with linear clearance
PK 2cmt des Two compartment PK model with linear clearance using differential equations
PK 2cmt mAb Davda 2014 Two compartment PK model with linear clearance for average monoclonal antibodies (Davda 2014)
PK 2cmt no depot Two compartment PK model with linear clearance using differential equations
PK 2cmt tdcl des Two compartment PK model with time-dependent clearance using differential equations (structured like nivolumab PK model)
PK 2cmt tmdd mm Two-compartment TMDD archetype, Michaelis-Menten (MM) approximation (Gibiansky et al. 2008)
PK 2cmt tmdd qss Two-compartment TMDD archetype, quasi-steady-state (QSS) approximation (Gibiansky et al. 2008)
PK 3cmt Three compartment PK model with linear clearance
PK 3cmt des Three compartment PK model with linear clearance using differential equations
PK double sim 01 PK double absorption model with simultaneous zero order and first order absorptions
PK double sim 10 PK double absorption model with simultaneous first order and zero order absorptions
PK double sim 11 PK double absorption model with simultaneous first order absorptions
PSA mCRPC (Desmee 2015) Mechanistic joint biomarker-survival model for prostate-specific antigen (PSA) kinetics under chemotherapy in metastatic castration-resistant prostate cancer (mCRPC). PSA is produced by a proliferating prostate-cell compartment and eliminated by a first-order process; chemotherapy blocks cell proliferation at a per-subject effectiveness eps until an individual escape time Tesc. The Weibull-baseline overall-survival hazard is log-linear in the current PSA value. This is a published-simulation-study model: parameter VALUES are pre-specified (inspired by one arm of the Tannock 2013 VENICE phase III trial), not estimated from real-data fits.
Radiation neutropenia (Harrold 2020) Preclinical (rhesus macaque). Semi-mechanistic ANC response model linking acute radiation injury to neutropenia, with a coupled time-to-event survival sub-model linking the ANC time course to overall survival in nonhuman primates exposed to 750 cGy whole-body irradiation on day 0 and treated with placebo or G-CSF (filgrastim or pegfilgrastim). Three components: (a) a K-PD radiation injury compartment (depot_kpd) that decays exponentially at k_PD,e and drives mitotic-cell killing via a power-law sensitivity coefficient k_kill = k_PD,kill * RAD^gamma (Eq. 2); (b) a Friberg-style granulopoiesis chain (precursor1 = N_SM stem cell -> precursor2 = N_MT mitotic -> precursor3 = N_PM1 -> precursor4 = N_PM2 -> circ = ANC) where the radiation kill effect operates on the mitotic compartment via the additive loss term -k_kill * precursor2 (Eqs. 3-7); and (c) a Box-Cox-transformed-ANCe-driven time-varying hazard model for overall survival (Eqs. 9-11), with study-specific parameter sets (Table III) selected by the binary indicator STUDY_HARROLD_PEG (0 = filgrastim pivotal study, 1 = pegfilgrastim pivotal study). The ANC response parameters (Table II) were jointly fit on the combined placebo cohorts (n = 45 NHPs); the OS sub-model parameters (Table III) were fit separately to each study because an unexplained study effect remained after covariate screening (Discussion p. 7).
Remikiren (Weber 1993) Population pharmacodynamic Hill / Emax model for the orally active renin inhibitor remikiren (Ro 42-5892) in patients with mild-to-moderate essential hypertension. Inhibition of the angiotensin I production rate (APR; the net inhibition of plasma renin activity corrected for changes in immunoreactive renin) is described as a saturable function of observed plasma remikiren concentration via the Hill equation with the Hill coefficient fixed at 1. PD-only model: plasma remikiren concentration is supplied as a time-varying covariate CP_REM_NGML (ng/mL). The source publication characterised remikiren PK with model-independent NCA only (Cmax / tmax / AUC0-t in Table 1) and did not develop a structural population PK model, so the PD model has no coupled PK component. Population: 144 patients with mild-to-moderate essential hypertension across three multi-dose clinical pharmacology studies (oral solution or 100 mg capsules; 100-800 mg po qd for 8 days).
RheumatoidArthritis (Wojciechowski 2015) Disease-progression model of 28-joint disease activity score (DAS28) over time since initiation of triple disease-modifying anti-rheumatic drug (DMARD) therapy (methotrexate + sulfasalazine + hydroxychloroquine) in early rheumatoid arthritis. Wojciechowski 2015 fit DAS28 from initiation until 60 weeks in 263 patients at the Royal Adelaide Hospital Early Arthritis Clinic. The structural model is an exponential decay of logit-transformed DAS28 toward a new treated steady state, additive to the (logit) baseline: DAS28_logit(t) = BASE * (1 + e_conmed_steroid_base * CONMED_STEROID) * (AGE/57)^e_age_base + EX1 * (1 + e_conmed_steroid_fu_ex1 * CONMED_STEROID_FU) * (1 - exp(-EX2 * (1 + e_smoke_ex2 * SMOKE) * t)). The natural-scale DAS28 (0-9.2 plausible range) is recovered by DAS28 = 9.2 / (1 + exp(-DAS28_logit)) – a logit back-transform with upper bound 9.2 corresponding to 28 swollen and tender joint counts, 100 mm visual analogue scale for patient global assessment, and 120 mm/h ESR. Random effects on BASE and EX1 are additive on the linear (logit-domain) scale; EX2 carries log-normal IIV; all three etas are correlated via a full 3x3 OMEGA BLOCK (Table 2 off-diagonal entries reported as correlations). Combined proportional + additive residual error is applied to the predicted logit-DAS28 (paper Equation 3). The model has no PK component – DMARD doses were titrated to disease severity in a treat-to-target protocol and were not retained as covariates. CONMED_STEROID is time-varying per record (1 at clinic visits where intramuscular corticosteroids were administered) and CONMED_STEROID_FU is time-fixed per subject (1 if the subject received any systemic corticosteroid – i.m. or oral – at any point during the 60-week follow-up window).
Sld (Ribba 2022) Stein bi-exponential population model of the tumor-size (sum of the longest diameters, SLD) time course in advanced non-small cell lung cancer (NSCLC), fit by Ribba et al. (Roche) to the phase III OAK study. This is the companion tumor-size fit to Ribba_2022_ctdna.R: the authors applied the same Stein structural model to SLD independently of the ctDNA data in order to test whether the two decay-rate constants correlate (they found r = 0.45, Figure 2F bottom left), which motivated the coupled model Ribba_2022_ctdna_sld_joint.R. Unlike the ctDNA fit, SLD was modeled on its natural millimetre scale with a combined additive-plus-proportional residual-error model, and the fit pooled both OAK arms (atezolizumab and docetaxel) with treatment arm as a categorical covariate so that individual parameters could subsequently be read out for the atezolizumab arm alone. The baseline SLD is supplied as a per-subject regressor rather than estimated.
SLE mbma (Goteti 2024) MBMA. Bayesian latent-variable disease-trajectory model-based meta-analysis (DTM-MBMA) of composite systemic lupus erythematosus (SLE) endpoints across 25 studies, 81 study arms, and 16 SLE treatments (Goteti 2024 Table 1). A uni-dimensional latent SLE disease activity theta(t) is modelled as a mono-exponential approach from a baseline mu_ijk to a long-term level mu_ijk + delta_ijk with time-constant log(2)/lam (Goteti 2024 Eq 1). The paper reparametrizes mu and delta in terms of the more-observable theta_wk4 (latent SLE at week 4) and delta_wk24 (increment from week 4 to week 24), separately for the SRI (SRI-4/5/6) and BICLA endpoints. Mean baseline SLEDAI (centered at 10.5) modifies both mu (via beta_sledai_mu) and delta (via beta_sledai_delta). Continuous-dose Emax dose-response was fit for six drugs whose regimen had three or more doses at the same route and frequency: anifrolumab (i.v. q4w), belimumab (i.v. q2wx3+q4w), CC-220 (oral qd), epratuzumab (i.v. q2w), lulizumab pegol (s.c. q2w), and sifalimumab (i.v. q2w). All 24 other drug-dose-regimen combinations from Table 4 (ALX-0061, atacicept, baricitinib, the s.c. q.w. belimumab arm, the i.v. q.w. epratuzumab arm, evobrutinib, fenebrutinib, IPP-201101, the s.c. q.w. lulizumab pegol arm, lupuzor, PF-04236921, tabalumab, ustekinumab) were modelled as discrete per-regimen additive effects on the latent-scale delta multiplier (1 + xi_ij); the operator passes the reference discrete-effect value through the CONMED_XI_DISCRETE covariate (documented lookup in the vignette). SRI-4/5/6 responder probabilities are modelled as a generalized proportional-odds transform of the latent SLE activity (thresholds c_SRI5 = 0.586, c_SRI6 = 0.693), and BICLA responder probability is the logistic transform of the latent SLE activity directly. Between-study variability is encoded as study-level etas on theta_wk4 and delta_wk24 for both endpoints (no subject-level IIV); the simulation scope is aggregate study-arm-mean responder probability, not individual patient trajectories. All Table 3 point estimates are the posterior mean under the reference Bayesian fit (model 106-bm-1 in Table S1 of the supplement).
Statins ezetimibe mbma (Vargo 2014) MBMA. Literature-based meta-analysis dose-response model for percent change in low-density lipoprotein cholesterol (LDL-C) from baseline for six statins (atorvastatin, fluvastatin, lovastatin, pravastatin, rosuvastatin, simvastatin), ezetimibe, and statin-plus-ezetimibe combination therapy in adult dyslipidemia. Operates at the study-arm level over 245 trials (1,267 study-arm data points, 106,808 patients). Algebraic Emax-Hill (sigmoid) dose-response with statin-specific ED50 and shared sigmoidicity n=0.417 across statins; ezetimibe sigmoidicity is fixed to 1. Statin Emax depends on study-arm baseline LDL-C, baseline triglycerides, percentage with coronary heart disease (CHD), and binary cohort indicators for acute coronary syndrome (DIS_ACS) and heterozygous familial hypercholesterolemia (HeFH). Combination therapy is modelled via a sub-additive interaction coefficient gamma=0.523 (at maximal monotherapy effect the combined LDL-C reduction is about 7 percent smaller than the sum of the two monotherapies). Fluvastatin and lovastatin twice-daily and extended-release formulations multiply the statin ED50 by a fixed ratio (0.645 for fluvastatin; 0.59 for lovastatin). Between-study variances for Emax and ED50 were fixed to zero in the source paper, so the model has no eta IIV; the residual SD describes study-arm-mean variability and the suitable simulation scope is study-arm-mean percent change in LDL-C, not individual-subject concentrations.
Sunitinib dbp (Hansson 2013) Indirect-response model of sunitinib-induced increase in diastolic blood pressure (dBP) in adults with imatinib-resistant gastrointestinal stromal tumours (GIST). The dBP state turns over via a stimulated zero-order production rate Kin and a first-order removal rate Kout (= 1 / MRT), where the per-cycle drug-exposure summary AUC = DOSE / CLI linearly stimulates Kin via a slope factor dBP_slope. Kin is parameterised as dBP0 * Kout so the dBP steady state without drug equals dBP0. A separate higher baseline dBP0_placebo is recorded as a placebo-arm typical value (Results: ‘this group had a significantly higher baseline dBP (dBP0) when estimated separately’). The PD model has no PK ODE; sunitinib exposure enters as the AUC summary computed from time-varying DOSE and per-subject CLI.
Sunitinib fatigue (Schindler 2017) Population PD model of NCI-CTCAE fatigue grade (0 / 1 / 2 / 3+) in adults with imatinib-resistant gastrointestinal stromal tumours (GIST) on sunitinib, using the minimal continuous-time Markov model (mCTMM) form of Schindler & Karlsson 2017. Four ordered-categorical states (grades 0, 1, 2, 3+) evolve via ODEs (Eq. 6) with transfer rate constants (Eqs. 8-9) parameterised through a single mean equilibration time (MET) and the steady-state score probabilities Pss,k. Steady-state cumulative probabilities Pss(Y >= k) follow a proportional-odds model (Eq. 10) driven by the relative change from baseline over time in soluble vascular endothelial growth factor receptor 3 (sVEGFR-3), a sunitinib pharmacodynamic biomarker. sVEGFR-3 is generated inline as a one-compartment indirect-response turnover with simple-Imax inhibition of Kin, driven by a per-cycle drug-exposure summary AUC = DOSE / CLI (following the same convention used in the sister Hansson_2013_sunitinib_hfs and Hansson_2013a_sunitinib model files); the sVEGFR-3 upstream parameters BAS_SVEGFR3, MRT_SVEGFR3, and EC50_SVEGFR3 are consumed as data covariates. The model has no PK ODE. Outputs are the four marginal state probabilities pscore0-pscore3, the cumulative probabilities pge1-pge3, and the expected grade.
Sunitinib hfs (Hansson 2013) Population PD model of hand-foot syndrome (HFS; NCI-CTC grades 0 / 1 / 2 / 3+) in adults with imatinib-resistant gastrointestinal stromal tumours (GIST) on sunitinib. A first-order Markov + proportional-odds (PO) likelihood describes the per-visit HFS-grade transition probabilities conditional on the previous grade. The cumulative-logit baselines per starting state are shifted by a starting-state-specific slope on a delayed sVEGFR-3 relative-change signal (effect-compartment-smoothed by ke0 = 0.347/h per Hansson 2013 e85). sVEGFR-3 itself is simulated in-model as a one-compartment indirect-response turnover driven by the per-cycle drug-exposure summary AUC = DOSE / CLI. The PD model has no PK ODE and consumes individual posthoc upstream-PD parameters (BAS_SVEGFR3, MRT_SVEGFR3, EC50_SVEGFR3) and posthoc upstream-PK clearance (CLI) as data covariates. Random effects are diagonal across the four per-state baseline logits; the >=3 state IIV is fixed to zero (NE in Hansson 2013 Table 3).
Sunitinib hfs (Schindler 2017) Population PD model of NCI-CTCAE hand-foot syndrome (HFS) grade (0 / 1 / 2 / 3) in adults with imatinib-resistant gastrointestinal stromal tumours (GIST) on sunitinib, using the minimal continuous-time Markov model (mCTMM) form of Schindler & Karlsson 2017. Four ordered-categorical states (grades 0-3) evolve via ODEs (Eq. 6) with transfer rate constants (Eqs. 8-9) parameterised through a single mean equilibration time (MET) and the steady-state score probabilities Pss,k. Steady-state cumulative probabilities Pss(Y >= k) follow a proportional-odds model (Eq. 10) driven by the relative change from baseline over time in soluble vascular endothelial growth factor receptor 3 (sVEGFR-3), a sunitinib pharmacodynamic biomarker. sVEGFR-3 is generated inline as a one-compartment indirect-response turnover with simple-Imax inhibition of Kin, driven by a per-cycle drug-exposure summary AUC = DOSE / CLI (following the same convention used in the sister Hansson_2013_sunitinib_hfs and Hansson_2013a_sunitinib model files); the sVEGFR-3 upstream parameters BAS_SVEGFR3, MRT_SVEGFR3, and EC50_SVEGFR3 are consumed as data covariates. The model has no PK ODE. Outputs are the four marginal state probabilities pscore0-pscore3, the cumulative probabilities pge1-pge3, and the expected grade. Grade 4 was never observed in the source cohort, so the state space is 0-3.
Sunitinib myelosuppression (Hansson 2013) Semi-physiological Friberg-Karlsson myelosuppression model for sunitinib in adults with imatinib-resistant gastrointestinal stromal tumours (GIST). Absolute neutrophil count (ANC) is described by a self-renewing proliferating progenitor pool, three transit compartments reflecting cell maturation, and a circulating-neutrophil pool, with an Emax drug-effect function driven by the relative change in soluble VEGFR-3 from baseline (sVEGFR-3 REL) inhibiting proliferation and a (ANC0/circ)^gamma feedback term. sVEGFR-3 itself is simulated in-model as a one-compartment indirect-response turnover with simple-Imax inhibition of Kin by the per-cycle drug-exposure summary AUC = DOSE / CLI. The PD model has no PK ODE and consumes individual posthoc upstream-PD parameters (BAS_SVEGFR3, MRT_SVEGFR3, EC50_SVEGFR3) plus posthoc upstream-PK clearance (CLI) and a Japanese-cohort indicator (RACE_JAPANESE) as data covariates. The Japanese-cohort indicator switches the typical baseline ANC0 between 4.94 (non-Japanese) and 3.69 (10^9/L) per Hansson 2013 Table 2.
Sunitinib os (Hansson 2013) Weibull time-to-event model for overall survival (OS) in adults with imatinib-resistant gastrointestinal stromal tumours (GIST) on sunitinib. The hazard function is h(t) = lambda * alpha * (lambda * t)^(alpha - 1) * exp(beta_anc * ANC + beta_dbprel * DBP_REL + beta_tumor * TUMSZ), where the three log-linear hazard modulators are the time-varying absolute neutrophil count ANC(t) (from the upstream Hansson 2013 myelosuppression model), the relative change in diastolic blood pressure from baseline DBP_REL(t) (from the upstream Hansson 2013 dBP indirect-response model), and the time-fixed baseline tumor size TUMSZ in mm (paper Table 2 ‘beta3 Tumor base’). All three modulators are consumed as data covariates. Time units inside the model are hours; the source paper reports the Weibull lambda in per-week units, converted to per-hour inside ini() so the parameter values match Table 2 row ‘lambda (/week)’. A separate Weibull censoring distribution (lambdacens, alphacens) is exposed as a derived output for use in simulation-based dropout.
T2DM WHIG (Choy 2016) Semi-mechanistic disease-progression model for type 2 diabetes (WHIG: Weight, HbA1c, Insulin, Glucose). Body-weight turnover under diet+exercise and placebo drives insulin sensitivity, which together with beta-cell function drives a fasting serum insulin (FSI) and fasting plasma glucose (FPG) homeostatic feedback (steady-state quadratic). FPG and an additional postprandial-glucose contribution feed a three-compartment transit chain producing total HbA1c. Built from the placebo arm of NCT00236600 (181 obese newly-diagnosed adults with DIS_DIAB on diet+exercise for 67 weeks).
TB MTP GPDI invitro (Clewe 2018) In vitro (M. tuberculosis Beijing VN 2002-1585, BE1585 strain). Multistate Tuberculosis Pharmacometric (MTP) model linked to the General Pharmacodynamic Interaction (GPDI) model describing CFU/mL dynamics during 6-day time-kill assays of M. tuberculosis exposed to static rifampicin / isoniazid / ethambutol concentrations alone or in duo / trio combinations. Three bacterial states (fast-multiplying F, slow-multiplying S, non-multiplying N) follow MTP natural-growth dynamics with the transfer rates kFSlin, kSF, kFN, kSN, kNS fixed from Clewe 2016 JAC dkv416; growth is exponential (kG) rather than logistic. Drug effects per Table 1: rifampicin inhibits F growth (sigmoidal Emax) and kills F, S (Emax) and N (linear); isoniazid kills F and S via sigmoidal Emax with a linear isoniazid-induced adaptive-resistance loop (AR_off / AR_on) modulating EC50 on F-kill and S-kill; ethambutol kills F (sigmoidal Emax) and S (linear). The thirteen quantified GPDI interactions (Table 3) enter as multiplicative modifications of each modulated drug’s EC50 using the canonical Emax-style interaction term 1 + INT * C_modifier / (EC50_modifier + C_modifier) per Table 3 footer, with the modifier’s mono EC50 acting as the half-saturation anchor (least complex baseline per the paper’s Materials and methods). Combination effects on each bacterial state are pooled by the Bliss Independence criterion. Drug concentrations are static covariates (no PK structure – the in vitro experiment fixes nominal concentrations for the 6-day window). This is the in vitro twin of Chen 2017 in vivo mouse MTP-GPDI; see modellib(‘Chen_2017_TB_MTP_GPDI_mouse’) for the same MTP-GPDI framework applied to TB-infected BALB/c mice with explicit PK.
TB MTP GPDI mouse (Chen 2017) Preclinical (BALB/c mouse). Multistate Tuberculosis Pharmacometric (MTP) model linked to General Pharmacodynamic Interaction (GPDI) model describing CFU/lungs dynamics in M. tuberculosis Beijing VN 2002-1585 infected BALB/c mice receiving oral monotherapy or combination therapy with rifampicin, isoniazid, ethambutol, and pyrazinamide. Four parallel population PK models (1-cmt RIF/INH/PZA, 2-cmt EMB) drive concentration-dependent drug effects on three bacterial states (fast-multiplying F, slow-multiplying S, non-multiplying N). Rifampicin and isoniazid interact antagonistically on the killing of S and N (INT_S_RH = 4.49; INT_N_RH = 0.32); rifampicin and ethambutol interact synergistically on the killing of N (INT_N_RE = -0.15). The natural-growth transfer rates kSF, kFN, kSN, kNS are fixed from the in vitro MTP fit of Clewe 2016; the absorption rate constants for isoniazid, ethambutol, and pyrazinamide are fixed from the upstream mouse popPK of Chen 2016.
TB multistate (Lin 2024) Pharmacometric five-state multistate model for long-term treatment outcomes in adults with drug-resistant pulmonary tuberculosis treated with bedaquiline on top of a multidrug background regimen (pooled TMC207-C208 and TMC207-C209 phase IIb trials). Five ODE states carry the marginal state-occupancy probability of active TB (s_activeTb, the initial state), sputum-culture conversion (s_converted), recurrent TB (s_recurrentTb), study dropout (s_dropout, absorbing) and death (s_death, absorbing). Patients move from active TB to converted and on to recurrent TB, and may return from recurrent TB to converted; dropout and death compete from every transient state. The conversion hazard lambda12 follows a symmetric surge function of time (amplitude SA12, peak time PT12, surge width SW12) that peaks at 11.4 weeks; recurrence (lambda23), re-conversion (lambda32), dropout (lambda14, lambda24 = lambda34) and death-from-converted (lambda25) are constant, while death from active or recurrent TB (lambda15 = lambda35) follows a Weibull hazard with increasing risk over time (shape 1.96). Faster model-derived bacterial clearance through week 2 raises the surge amplitude and moves the peak earlier; a longer baseline MGIT time-to-positivity (lower bacterial burden) moves the peak earlier and widens the surge; XDR-TB lowers the surge amplitude by 46%. Recurrence is higher in men and rises with the model-derived mycobacterial load at the end of the 24-week treatment period (applied only after week 26). Dropout is higher in the C208 study and in younger patients; death risk falls with higher baseline body weight. There are no drug-dosing events and bedaquiline exposure does not enter this model directly – it reaches the model through the two model-derived covariates MBL_HL_WK2 and MBL_END, which are secondary metrics of the upstream mycobacterial-load model shipped as modellib(‘Svensson_2017_bedaquiline’). The source analysis was fitted in NONMEM with the exact likelihood on the observed categorical state and has no between-subject variability (OMEGA fixed to 0) and no residual-error model.
Testosterone (GonzalezSales 2015) Stretched-cosine model of the endogenous (baseline) circadian rhythm of serum testosterone in adult hypogonadal men (Gonzalez-Sales 2015 AAPS J). T(t) = Base + Amplitude * cos(pi * f(t; tacro, tnadir)) where f is the piecewise-linear phase function that is 0 at the peak time tacro, 1 at the nadir time tnadir, and 2 at the next peak; the descending arm lasts L1 = (tnadir - tacro) mod 24 hours and the ascending arm lasts L2 = 24 - L1 hours, allowing an asymmetric (stretched) cycle. Baseline is reduced by 2.40% per decade with age (centred on the pooled median 49.9 y) and elevated by 8.09% during winter and spring (SEMESTER = 1) relative to summer and fall (SEMESTER = 0). IIV on Base enters via a Box-Cox-transformed normal eta (lambda = -1.93; Petersson 2009 form). Pure typical-value + IIV simulation model (no exogenous drug); time is clock time in hours after midnight.
TGD OS gastric (Terranova 2022) Joint tumor growth dynamics (TGD) + overall survival (OS) disease-progression model for advanced gastric cancer / gastroesophageal junction cancer (GC/GEJC) developed on the JAVELIN Gastric 100 phase III trial by Terranova et al. (Merck KGaA / Merck Institute of Pharmacometrics / EMD Serono / Metrum Research Group). N = 499 patients randomized 1:1 to avelumab 10 mg/kg every 2 weeks (n = 249) or continued chemotherapy (n = 250) maintenance after 12 weeks of induction chemotherapy. The paper’s primary finding is NEGATIVE: no treatment effect on OS or TGD was identified, consistent with the primary JAVELIN Gastric 100 analysis. Machine learning (random forests + SIDEScreen + Boruta / permutation / random-splits variable-importance methods) was used to screen 89 (OS) / 52 (TGD) candidate covariates, then covariates surviving the ML screen were incorporated into two parametric sub-models: (1) a Gompertzian TGD model on tumor size measured by sum of longest diameters (SLD, mm) with the Vaghi 2020 Kd = slope*Kg + intercept reduced-parameter reformulation, and (2) a log-logistic parametric time-to-event OS model with an accelerated-failure-time link to time-invariant covariates on the log-median-OS scale, a proportional-hazards link to time-varying laboratory covariates, and a per-arm shape parameter that captures the crossing of hazard curves between avelumab and chemotherapy. There is NO drug PK sub-model – avelumab exposure is not carried explicitly; the treatment arm is encoded as the TRT categorical covariate (1 = chemotherapy = reference, 2 = avelumab). Extract-scope note: no drug PK; the two ODE outputs are tumor (SLD, mm) with proportional residual error and sur (survival probability), driven off d/dt(cumhaz) = hazard.
Tgi bertalanffy (Yates 2023) Theoretical (illustrative; no data fitted). Von Bertalanffy tumor-growth law (dV/dt = k * V^(2/3) - kd * V) with an Emax drug effect on the proliferation term, driven by a one-compartment i.v. bolus PK model. Case study 3 of Yates and Mistry (2023): an explicit cell-death term lets tumor size plateau, and the resulting dose-response is log-linear in shape.
Tgi exponential (Yates 2023) Theoretical (illustrative; no data fitted). Exponential tumor-growth law with a sigmoidal Emax drug effect on the growth rate, driven by a one-compartment i.v. bolus PK model. Case study 1 of Yates and Mistry (2023), used to show that the location (ED50) and steepness of a dose-response curve are set jointly by potency (EC50), efficacy (Emax) and the drug’s elimination rate constant a = CL/Vd.
Tgi mayneord (Yates 2023) Theoretical (illustrative; no data fitted). Mayneord linear-radial tumor-growth law (dV/dt = k * V^(2/3)) with an Emax drug effect on the radial growth rate, driven by a one-compartment i.v. bolus PK model. Case study 2 of Yates and Mistry (2023): a sub-exponential growth law yields a log-linear rather than sigmoidal dose-response shape.
tgi no sat expo One-compartment TGI model with exponential tumor growth, without saturation.
tgi no sat Koch One compartment TGI model with with exponential tumor growth, without saturation.
tgi no sat linear One compartment TGI model with with linear tumor growth, without saturation.
tgi no sat powerLaw One compartment TGI model with with exponential tumor growth, without saturation.
tgi sat genLogistic One compartment TGI model with tumor growth proportional to tumor size through a generalized logistic function, with saturation.
tgi sat genVonBertalanffy One compartment TGI model where tumor growth is limited by a loss term, with saturation.
tgi sat Gompertz One compartment TGI model with tumor growth proportional to tumor size through a generalized logistic function, with saturation.
tgi sat logistic One compartment TGI model with with exponential tumor growth that decelerates linearly, with saturation.
tgi sat VonBertalanffy One compartment TGI model where tumor growth is limited by a loss term, with saturation.
Tumorsize OS nsclc (Struemper 2025) Joint tumor-size (TS) / overall-survival (OS) framework model for non-small cell lung cancer (NSCLC), developed by Struemper et al. (GSK) on pooled individual-level data from 786 participants across seven GSK-sponsored clinical trials (INDUCE-1, INDUCE-2, Entree Lung Part 2, GARNET, AMBER, LUNG 037, PERLA) spanning immunotherapy, chemotherapy, and combinations thereof. The TS sub-model is the bi-exponential Stein model with per-treatment-arm typical tumor-growth (kge) and tumor-shrinkage (kse) rates selected at simulation / fit time via the TRT categorical covariate (12 levels; see covariateData[[TRT]] for the integer coding). The OS sub-model is an accelerated failure time (AFT) log-normal survival with a treatment-agnostic link to individual TS parameters: tumor growth rate kge enters via a fixed Emax function, baseline TS and time-to-tumor-growth (TTG) enter linearly, and three baseline laboratory covariates (albumin, total protein, neutrophil-to-lymphocyte ratio) are additive on the log-time scale. TS is observed in mm (sum of longest diameters of target lesions per RECIST 1.1); survival is reported on the day scale in the source paper and converted to weeks inside model() so the entire model runs on a single weeks time axis.
UricAcid (Aksenov 2018) Semi-mechanistic exposure-response model of uric acid disposition in adults with drug effects of xanthine-oxidase inhibitors (allopurinol via oxypurinol, febuxostat) and a uricosuric URAT1 inhibitor (lesinurad)
WHIG T2DM (Duong 2016) Semi-mechanistic weight-HbA1c-insulin-glucose (WHIG) disease-progression model of placebo-treated adults with type 2 diabetes mellitus (T2DM) (Duong 2016 BJCP). Structural model is the Choy 2016 WHIG framework: a body-weight turnover compartment, a steady-state closed-form solution for the fasting-serum-insulin (FSI) / fasting-plasma-glucose (FPG) homeostatic feedback pair, and three transit compartments for glycated haemoglobin (HbA1c). Insulin sensitivity (IS) is driven by change in body weight (dWGT); ScaleEFS = 0.0458 per kg scales the IS-vs-weight sensitivity. baseline b-cell function (b0, logit) is shifted between Study 1 (newly diagnosed obese T2DM, b0 = -0.298 -> BF0 ~ 57 pct) and Studies 2 and 3 (advanced T2DM, b0 = 0.677 -> BF0 ~ 34 pct) via the STUDY_1 covariate. Placebo treatment effects are step functions active at t > 0 (EFDE, diet-and-exercise + placebo run-in) and at t > placebo-run-in phase (EFPL, Study 1 only; the placebo effect at the treatment phase was removed for Studies 2 and 3 because it was not significant). rB (rate of b-cell function loss) and MTT (HbA1c mean transit time) were fixed to the Choy 2016 upstream WHIG estimates (0.209 per year and 38.9 days) because the shorter trial duration of Studies 2 and 3 rendered them poorly identifiable in the joint fit; these fixed values are re-estimated for the pooled 3-study population. The FSI / FPG closed-form solution is the quadratic root of the linked steady-state ODEs (Choy 2016 Eqs 11 and 12): C * FSI^2 + 3.5 * A * C * FSI - 35.1 * A = 0 where A = EFB * B * 7.8 (K_in,FSI / K_out,FSI = 7.8 is the healthy FSI_ss anchor) and C = EFS * IS0 (35.1 = K_in,FPG / K_out,FPG is the healthy FPG_ss * FSI_ss anchor). The 3.5 mmol/L subtracted from FPG is the physiological FPG floor for insulin secretion (Choy 2016 Methods).

Specific Drugs

name description
5hydroxysaxagliptin rat (Wang 2024) Preclinical (rat). Two-compartment intravenous PK model for the active saxagliptin metabolite 5-hydroxy saxagliptin (5-OH SAX), directly linked to a sigmoidal Emax model WITH a baseline term E0 for plasma dipeptidyl peptidase-4 (DPP-4) inhibition, in streptozotocin plus high-fat-diet induced type 2 diabetic male Sprague-Dawley rats. Wang 2024 Eq 4-6 write the disposition as dA1/dt = (CL2/V2)A2 - (CL2/V)A1 - (CL/V)A1 and dA2/dt = (CL2/V)A1 - (CL2/V2)A2, i.e. CL2 is a conventional inter-compartmental clearance (k12 = CL2/V, k21 = CL2/V2) and is encoded here as lq. The PD model is Wang 2024 Eq 7, E = E0 + Emax C^Gam / (EC50^Gam + C^Gam), applied directly to the plasma 5-OH SAX concentration with no effect compartment; the paper states explicitly that 5-OH SAX binds DPP-4 immediately and that no hysteresis was observed. This model is the metabolite half of Wang 2024 and is the model the authors used to attribute part of the total DPP-4 inhibition seen after oral saxagliptin to the metabolite; the parent-drug half is modellib(‘Wang_2024_saxagliptin_rat’). The fitted potencies satisfy EC50_5OHSAX = 0.46 * EC50_SAX, the paper’s central finding. All volumes and clearances are body-weight normalised (mL/kg, mL/h/kg), so ODE state amounts are in ng per kg body weight and the 0.5 mg/kg intravenous dose is supplied as amt = 5e5 into central. Parameters are the arithmetic means of three individually fitted rats (WinNonlin 8.1, naive individual fitting); no population model and therefore no OMEGA was estimated, so the model carries no eta terms and is intended for typical-value simulation. The PK residual error was described as additive but its magnitude was never reported, so addSd is FIXED at 0; the PD residual SD (stdev0) IS reported and is used. E0 is kept on the linear scale because the individual estimates in Supplementary Table S19-b include a negative value (-3.85%).
8C2 (Abuqayyas 2012) Preclinical (mouse). Two-compartment IV bolus population PK model of 8C2, a murine IgG1 anti-topotecan monoclonal antibody, in C57BL/6 wild-type, Fc-gamma-RI/RIII knockout, and Fc-gamma-RIIb knockout mice (Abuqayyas 2012). Data were pooled across all three strains and three IV bolus dose levels (0.04, 0.1, 0.4 mg/kg); strain was screened and found not statistically significant on any structural parameter, so no covariate effects are retained. Structural PK parameters were reported per kg body weight (L/kg and L/day/kg), which corresponds to linear body-weight scaling with a fixed exponent of 1.
Abacavir (Archary 2019) Two-compartment population PK model for abacavir in severely malnourished HIV-infected children (Archary 2019); CL/F steps up between day 1 and day 14 of antiretroviral treatment and bioavailability is 31% higher in the early-ART arm
Abacavir (Chandasana 2024b) Two-compartment oral population PK model with first-order absorption and elimination for abacavir in children with HIV-1 weighing 6 to 40 kg, applied without re-estimation to the ABC/DTG/3TC fixed-dose combination (dispersible tablet and tablet) in IMPAACT 2019 (Chandasana 2024)
Abacavir (Jullien 2005) Two-compartment population PK model for abacavir in HIV-infected adults (Jullien 2005); apparent clearance scales with body weight via an estimated power exponent, Q/F is fixed when BW is added to the model
Abacavir (Tikiso 2021) Two-compartment population PK model for oral abacavir in HIV-infected African children (Tikiso 2021), with a Savic 2007-style analytical transit-compartment chain feeding a first-order absorption depot, allometric body-weight scaling on disposition (0.75 on CL/Q, 1 on Vc/Vp at 70 kg), sigmoidal Hill-type maturation of CL on postmenstrual age, and multiplicative covariate effects of efavirenz co-medication on CL, rifampicin + super-boosted lopinavir/ritonavir co-medication on F, fixed-dose-combination tablet formulation on MTT, and a time-decaying malnutrition effect on F and CL.
Abacavir (Zhao 2012) Two-compartment population PK model for oral abacavir in HIV-infected infants and toddlers (Zhao 2012) developed on the PENTA 15 crossover trial of 8 mg/kg twice-daily vs 16 mg/kg once-daily dosing; CL/F scales with body weight via an estimated power exponent (1.14) referenced to the population median weight of 12 kg, and inter-occasion variability on CL/F is multiplexed by the binary OCC indicator across the BID (occasion 1) and QD (occasion 2) study phases.
Abacavir (Zhao 2013) Two-compartment population PK model for oral abacavir in HIV-infected infants, toddlers, and children (Zhao 2013); body weight is the only retained covariate (allometric on CL/F and V1/F with estimated exponents and reference weight 17.6 kg).
Abatacept (Gandhi 2021) Two-compartment population PK model for abatacept (CTLA4-Ig Fc-fusion) pooled across adults with rheumatoid arthritis and patients aged 2-17 years with polyarticular juvenile idiopathic arthritis (Gandhi 2021), with first-order SC absorption, zero-order IV infusion support, first-order linear elimination, logit-scale SC bioavailability with disease/age/weight covariates, and a KA parameterisation that enforces KA > k_el.
Abatacept (Li 2019) Two-compartment population PK model for abatacept (CTLA4-Ig Fc-fusion) in adults with rheumatoid arthritis (Li 2019), with first-order SC absorption, zero-order IV infusion support, first-order linear elimination, logit-scale SC bioavailability, full-block IIV on CL/VC/Q/VP, and a KA parameterisation that enforces KA > k_el.
Abatacept (Lon 2013) Two-compartment population PK model with linear elimination and short-term zero-order SC absorption for abatacept (CTLA-4Ig Fc-fusion) in male Lewis rats with collagen-induced arthritis (Lon 2013).
Abatacept (Takahashi 2023) Two-compartment IV population PK model for abatacept (CTLA4-Ig Fc-fusion) pooled across 685 adult/pediatric patients with rheumatoid arthritis or polyarticular juvenile idiopathic arthritis and adult/pediatric patients receiving allogeneic hematopoietic cell transplantation in the ABA2 trial (Takahashi 2023). Linear elimination, allometric weight scaling on CL/Vc/Vp/Q with estimated exponents, and a three-level cohort categorical (RA/JIA reference, ABA2 HLA 7/8, ABA2 HLA 8/8) on CL and Vc.
Abatacept (Zhong 2026) Two-compartment population PK model for abatacept (CTLA4-Ig Fc-fusion) pooled across 9 phase 2/3 studies (Zhong 2026): adults with rheumatoid arthritis, patients aged 2-17 years with polyarticular juvenile idiopathic arthritis, and patients aged 6+ years with hematologic malignancies receiving HLA-matched unrelated-donor HSCT (the ABA2 trial). Final model has zero-order IV infusion, first-order SC absorption, first-order linear elimination, additive plus proportional residual error, allometric weight on CL/VC/VP, hepatic (AST) and renal (cGFR) markers on CL, sex on CL and VC, two HSCT cohort indicators (7-of-8 and 8-of-8 HLA-matched URD) on CL/VC, and a logit-scale SC bioavailability sub-model with weight, age, and pJIA-disease covariates fixed to a previously developed internal JIA PPK model (values match Gandhi 2021).
ABT 102 (Othman 2013) Population PK/PD model of body-temperature effects of ABT-102, a TRPV1 antagonist, in 108 healthy adult volunteers across three phase 1 trials (Othman 2013). PK is a one-compartment model with one transit absorption compartment, first-order elimination, and formulation-dependent absorption lag (0.3 h solution, 0.6 h solid dispersion) and relative bioavailability (40% solution vs solid-dispersion reference); PK parameter values are taken from the upstream popPK analysis (Othman 2012, J Clin Pharmacol). The PD layer models body temperature as the additive sum of (a) a measurement-type-dependent baseline (oral thermometer vs core ingestible capsule), (b) a 24-h circadian rhythm (cosine in time with measurement-type-dependent amplitude and a shared 7.6-h phase shift), and (c) an Emax drug effect on plasma concentration with time-driven exponential tolerance (Emax decays with half-life T50 = 28 h). Two parallel outputs (BT_oral, BT_core) are produced with measurement-type-dependent additive residual error; for a given subject only one output is realised (oral thermometer subjects use BT_oral; core ingestible-capsule subjects use BT_core).
Acetaminophen (vanRongen 2016) Parent-and-metabolites population PK model for intravenous acetaminophen (paracetamol) and its glucuronide, sulphate, and CYP2E1-oxidation (cysteine + mercapturate) metabolites in morbidly obese and non-obese adults (van Rongen 2016). One-compartment plasma disposition for parent acetaminophen with four parallel elimination pathways from the central compartment (glucuronidation, sulphation, CYP2E1 oxidation, and unchanged renal); one-compartment plasma disposition for glucuronide and cysteine + mercapturate metabolites each fed via a single-transit-compartment delay; two-compartment plasma disposition for sulphate (central + peripheral, fixed equal volumes 5.66 L each). Lean body weight (LBW; Janmahasatian et al. 2005 equation) enters as a power-law covariate on parent V, all three formation clearances, the CYP2E1 transit rate constant, and glucuronide elimination CL. Total body weight enters on the glucuronide volume of distribution.
Acetaminophen rat pbpk (Westerhout 2012) PBPK (semi-mechanistic, regional brain) population PK model for acetaminophen (paracetamol) in 24 male Wistar WU rats (225-275 g), developed to investigate regional brain distribution kinetics with simultaneous microdialysis sampling in striatum (brain extracellular fluid), lateral ventricle (CSF_LV), and cisterna magna (CSF_CM) after a 10-min intravenous infusion of 15 mg/kg acetaminophen (Westerhout et al. 2012, AAPS J). Seven physiological compartments: plasma plus peripheral tissue plus brain extracellular fluid (with the brain intracellular space volume added per paper text page 5) plus four anatomically distinct CSF subcompartments (lateral ventricle, third and fourth ventricle combined, cisterna magna, subarachnoid space). Brain compartment volumes (V_pl, V_ICS, V_ECF, V_LV, V_TFV, V_CM, V_SAS) and bulk fluid flows (Q_ECF, Q_CSF) are fixed to literature physiological values for a 250-g rat; plasma-to-region and region- to-plasma clearances are estimated. The plasma-to-third-fourth- ventricle clearances CL15 / CL51 are structurally assumed equal to the plasma-to-lateral-ventricle clearances CL14 / CL41 (paper Results). An enterohepatic-recirculation continuous input F_abs * DOSE adds drug back to plasma to capture the apparent plateau after t = 120 min. The model is fitted to unbound plasma concentrations (plasma concentrations corrected to free fraction fu_p = 0.805); fu_p is documented in population metadata but does not enter the structural ODEs (see vignette Assumptions and deviations).
Acyclovir (Zeng 2009) One-compartment population PK model with first-order absorption for acyclovir in 43 children and young people (age 0.8-19.9 years; weight 7.3-70.2 kg) with malignancy, after intravenous acyclovir (5 mg/kg q8h, 1 h infusion) or oral valacyclovir prodrug (10 mg/kg q12h), developed in NONMEM v5.1.1 (FOCE-I) from 1216 plasma observations. Structural model: first-order absorption (ka) from a depot with bioavailability F (oral valacyclovir delivered as systemic acyclovir), one-compartment disposition with first-order elimination. Allometric body-weight scaling on CL (fixed exponent 0.75) and V (fixed exponent 1) referenced to the cohort median 19.6 kg; CL additionally varies with creatinine clearance via a power function (CRCL/106.7 mL/min/1.73 m2)FAC. Inter-individual variability is diagonal on CL, V, ka, and F. Residual error is a combined exponential (proportional after linearization) + additive model. Inter-occasion variability on CL (19.2% CV) and V (30.4% CV) reported by Zeng 2009 Table 3 is NOT encoded structurally here (per the Andrews 2017 / Brooks 2021 tacrolimus precedent) – the source paper does not define an operational occasion column for the model-library use case.
Adalimumab (Drweesh 2026) One-compartment population PK model with first-order subcutaneous absorption and linear elimination for adalimumab originator (Humira) and biosimilars (Amgevita, Hyrimoz) in adults with inflammatory bowel disease and other autoimmune disorders, fit to multicenter therapeutic-drug-monitoring trough data from Saudi Arabia and Qatar (Drweesh 2026). Structural backbone (V/F, IIV variances, residual error) inherited from Marquez-Megias 2023 because Drweesh 2026 reports only ka (fixed) and the typical clearance value.
Adalimumab (Marcantonio 2022) QSP. One-compartment monospecific anti-ligand mechanistic PKPD model of adalimumab-TNF-alpha binding in adults with rheumatoid arthritis (Marcantonio 2022 Early Feasibility Assessment, Case Study 1). Drug administered SC (first-order absorption) or IV, binds soluble TNF-alpha (monovalent, effective valency = 1 per Tran 2017 / Lim 2018), and blocks the reversible TNF:TNFR interaction. All species are eliminated by first-order processes; the drug-target-ligand complex eliminates at the same rate as the free drug; the ligand-receptor complex eliminates at the same rate as the free receptor (paper Model Assumptions). All parameters are FIXED from paper Table 2 (no fitting, no IIV, no residual error reported). Initial conditions for L1 (free TNF), R1 (free TNFR), L1R1 (TNF-TNFR complex) and S1 (soluble shed receptor) are derived analytically from the reported steady-state concentrations and the quasi-equilibrium ligand-receptor binding balance. Simulation output activity (paper active_1_central) is the L1R1 complex amount; target inhibition is computed as 1 - L1R1(t)/L1R1(0). Run in the Applied BioMath Assess Monospecific Anti-Ligand model; the ODE system encoded here reproduces the Assess reaction network verbatim (see Supplementary Material Data Sheet 2, one_compartment_anti_ligand.pdf).
Adalimumab (Marquez-Megias 2023) One-compartment population PK model with first-order subcutaneous absorption and linear elimination for adalimumab in adults with inflammatory bowel disease, with albumin and anti-drug-antibody covariates on apparent clearance (Marquez-Megias 2023)
Adalimumab (Nader 2017) One-compartment population PK model with first-order subcutaneous absorption and linear elimination for adalimumab in adults with moderate-to-severe hidradenitis suppurativa, with baseline body weight and baseline C-reactive protein power effects on apparent clearance, a time-delayed anti-adalimumab-antibody effect on apparent clearance, and baseline body weight on apparent volume of distribution (Nader 2017).
Adalimumab (Zheng 2012) Preclinical (Gottingen minipig). Two-compartment population PK model for adalimumab with first-order SC absorption and linear central clearance in Gottingen minipigs, per Zheng 2012 minipig IV+SC study (Table 1 population mean parameter estimates).
ADC human qsp (Scheuher 2023) QSP. Human platform model for HER2-targeting antibody-drug conjugates in HER2+ metastatic breast cancer (T-DM1 default; T-DXd variant via parameter overrides). Extends the mouse model with: (i) HER2 receptor sinks on normal cells in the central and peripheral compartments (with binding, endocytosis, recycling, degradation); (ii) soluble HER2 (sHER2) shed from cell-surface HER2 into central + peripheral + tumor compartments, with reversible binding to ADC and Ab and its own turnover; and (iii) larger physiologic volumes (3 L central, 13 L peripheral for a 70 kg adult). Mouse-derived TGI parameters (kkill_max, kc50, tau, n_Hill) are carried over from N87 xenograft fits. Amounts in nmol; concentrations amount/volume.
ADC invitro qsp (Scheuher 2023) QSP. In vitro cellular ADC processing model for HER2-targeting antibody-drug conjugates (T-DM1 default; T-DXd variant supported via parameter overrides). 13 ODE states describing extracellular ADC / Ab / free payload, cell-surface HER2 with reversible ADC and antibody binding, endosomal HER2 species, endosomal / cytosolic payload, and cytosolic intracellular target (tubulin for DM1, TOPO-1 for DXd). Amounts in nmol per paper Modeling Convention; concentrations are amount/volume. Parameter set defaults to SK-BR-3 cell line with T-DM1 (Erickson 2012 in vitro incubation).
ADC mouse qsp (Scheuher 2023) QSP. Mouse in vivo platform model for HER2-targeting antibody-drug conjugates (T-DM1 default, N87 tumor xenograft; T-DXd variant supported via parameter overrides). Combines: (i) mouse plasma PK for ADC / naked antibody / free payload in central + peripheral compartments; (ii) mechanistic tumor uptake via a Krogh cylinder + surface exchange model; (iii) intracellular ADC processing (HER2 binding, endocytosis, recycling, degradation, endosomal payload release, cytosol transport, tubulin binding); and (iv) tumor growth inhibition via a Simeoni-style 4-stage transit-chain cascade with Hill-type kill on the proliferating stage. Amounts in nmol; concentrations amount/volume.
Adecatumumab (Cao 2013) Second-generation minimal physiologically-based PK (mPBPK) model for adecatumumab in adults (Cao 2013 Model A; clearance from plasma)
Aditoprim pbpk (Mi 2024) Veterinary (pig). PBPK (whole-body, seven-compartment, Berkeley Madonna 8.3.23.0) for the diaminopyrimidine aditoprim (ADP) after intramuscular injection in swine, developed to optimise the dosage regimen against Streptococcus suis and to predict the withdrawal interval for edible tissues (Mi et al. 2024, Front Pharmacol 15:1378034). A single well-mixed blood pool, liver, kidney, muscle, fat and a lumped rest-of-body compartment are perfusion-limited and well stirred; liver, kidney, muscle and fat are carried as separate states because they are the edible tissues against which the maximum residue limits are set. Intramuscular absorption is a two-step process: Frac = 92% of the injected dose is immediately available at the injection site and is absorbed into the venous mixing pool at first-order rate Kim, while the remaining 8% is released from a slow depot at first-order rate Kdiss into that same injection-site pool. Elimination is renal (from kidney) plus hepatic metabolism (from liver). Plasma protein binding rescales the free plasma concentration only – it does not gate tissue perfusion, which is driven by TOTAL arterial concentration in the published code. The seven parameters that Mi 2024 Table 3 carried through the 1000-animal Monte Carlo pop-PBPK analysis carry between-animal variability here; every other parameter is a fixed physiological or fitted chemical-specific constant. The model was hand-calibrated in Berkeley Madonna against digitised literature data, so no standard errors and no residual-error model are reported – the propSd term is a placeholder. Several Table 2 / supplement-code discrepancies are reproduced as coded; see the vignette Errata.
Aditoprim pbpkpd (Mi 2024) Veterinary (pig) + in vitro (Streptococcus suis ATCC 49619). Integrated PBPK/PD model for the diaminopyrimidine aditoprim (ADP) after intramuscular injection in swine (Mi et al. 2024, Front Pharmacol 15:1378034). The disposition layer is the seven-compartment whole-body PBPK model of modellib(‘Mi_2024_aditoprim_pbpk’); the effect layer is a two-subpopulation semi-mechanistic pharmacodynamic model fitted in Monolix 2018R1 to in vitro time-killing curves against S. suis ATCC 49619 (MIC 0.5 ug/mL). Susceptible (S) and resistant (R) subpopulations share one logistic growth term capped at the stationary-phase count Bmax and are killed by a common sigmoid Emax function that differs only in potency (EC50_R = 1.63 mg/L is 2.4-fold EC50_S = 0.685 mg/L), so resistant organisms are selected under sub-optimal exposure. The starting inoculum is split by the fixed mutation frequency MF = 1e-3. The two layers are coupled one-way: the UNBOUND plasma concentration predicted by the PBPK model replaces the static broth concentration Cstastic of the time-kill experiment, and the bacteria do not feed back on the drug. The unbound (not total) concentration is the driver – that is what the published code computes as CAfree and is the only reading that reproduces all four dose levels of Mi 2024 Figure 6; see the vignette. Used to show that 20 mg/kg every 12 h for 3 days eradicates both subpopulations while 5 mg/kg selects for resistance. Neither layer reports between-animal or between-replicate random effects on the PD parameters, so the PD block is typical-value only and propSd is a placeholder.
Aflibercept (Thai 2011) Mechanism-based population PK model for free and bound aflibercept (anti-VEGF Fc-fusion ‘trap’ protein; VEGF-Trap) in healthy adult male subjects (Thai 2011 BJCP). Two-compartment disposition of free aflibercept with linear elimination from the central compartment plus Michaelis-Menten binding of free aflibercept to VEGF occurring in the peripheral (tissue) compartment, producing a one-compartment bound-aflibercept species that is eliminated by first-order internalisation (kint). This is the second Michaelis-Menten approximation of the TMDD model of Gibiansky et al. (irreversible binding), with the bound complex carried as an explicit state. The bound-aflibercept volume of distribution Vb is fixed equal to the central volume Vc for identifiability. Pooled data from two phase 1 single-dose IV-infusion studies in healthy males (1, 2, 4 mg/kg over 1 h). No covariates were tested or retained in the final model.
Agomelatine (Xie 2019) A semiphysiological population pharmacokinetic model of agomelatine and its metabolites in Chinese healthy volunteers
Albendazole (PillayFuentesLorente 2024) Joint population PK model for the two main albendazole metabolites after a single oral 400 mg albendazole dose (with or without co-administered ivermectin) in adolescents infected with Trichuris trichiura in Tanzania and Cote d’Ivoire: Savic transit-compartment absorption feeding a two-compartment albendazole sulfoxide disposition that converts quantitatively to a one-compartment albendazole sulfone, with a study-population (country) effect on both apparent clearances
AlbuminNanoparticles mouse pbpk (Kutumova 2024) Preclinical (mouse, BALB/c, male, 6-8 weeks, 0.024 kg). PBPK (whole-body, BioUML) model for the biodistribution of intravenous albumin nanoparticles (ANP, ~120 nm, zeta ~ -30 mV) during induction of and recovery from lipopolysaccharide-induced acute lung injury. Seven compartments: venous and arterial plasma plus four mononuclear phagocyte system organs (lungs, spleen, liver, kidneys) and a rest-of-body compartment. Each organ carries a membrane-limited triple of capillary blood, tissue interstitium, and an internalised (phagocytosed) pool, with time-dependent Hill-function endocytic uptake and first-order exocytic release; the liver additionally phagocytoses directly from capillary blood (Kupffer cells) and excretes into bile, and the kidneys excrete into urine. The permeability (PAC) and distribution (P) coefficients are cohort-specific: STUDY_LPS30M / STUDY_LPS6H / STUDY_LPS24H select the ANP-after-LPS arm, with all three zero giving the LPS-naive control. The observed quantity is organ total radiant efficiency per luminous area (TRE), obtained from the simulated tissue concentration through a single fitted scale factor k. Physiologic blood flows, organ volumes, and capillary blood fractions are inherited from the Cheng 2020 nanoparticle PBPK framework, which the authors cite as their source for these values.
Alemtuzumab (Mould 2007) Two-compartment population PK model with Michaelis-Menten elimination for alemtuzumab in B-cell chronic lymphocytic leukaemia (Mould 2007)
Alemtuzumab wbc (Mould 2007) Coupled population PK-PD model for alemtuzumab in B-cell chronic lymphocytic leukaemia (Mould 2007): the two-compartment Michaelis-Menten PK from Mould 2007 Table 2 driven by the simulated WBC state via Vmax = TVVmax * (WBC/10)^0.194, joined to an indirect-response model on WBC (stimulation of Kout by alemtuzumab; Mould 2007 Table 3). WBC is a state variable initialised per subject at Kin/Kout.
Alirocumab (Djebli 2017) (Djebli 2017 writes the drug’s linear catabolic clearance as CLL; it is stored here under the canonical name lcl.) Quasi-steady-state target-mediated drug disposition (TMDD-QSS) population PK model for alirocumab and total PCSK9 in healthy adults and adults with hypercholesterolemia (Djebli 2017, final model on expanded data set n=2870). Two-compartment disposition with first-order SC absorption (lag time and bioavailability), linear catabolic clearance from central, and PCSK9 binding / complex internalization described by QSS algebra; allometric weight scaling on CLL, Q, and Vc plus a statin-coadministration effect on CLL.
Alirocumab (Martinez 2019) Two-compartment population PK model for alirocumab in healthy volunteers and adults with hypercholesterolemia (Martinez 2019, Part I), with first-order SC absorption (with lag time), linear plus Michaelis-Menten (target-mediated) elimination from the central compartment, and logit-transformed bioavailability.
Allopurinol (Wright 2013) Sequential parent-metabolite population pharmacokinetic model for oral and intravenous allopurinol and its principal long-lived active metabolite oxypurinol in patients with chronic gout and healthy volunteers, pooled from three Christchurch (New Zealand) clinical studies (dose-escalation, furosemide-interaction, vitamin-C interaction) and two digitally-extracted historic intravenous-allopurinol studies. Structural model: 2-compartment first-order absorption disposition for allopurinol (parent) with 80 percent molar conversion to a 1-compartment oxypurinol (metabolite). Fat-free-mass (FFM) allometric scaling on all clearance parameters (exponent 0.75) and volume parameters (exponent 1.0) referenced at FFM = 70 kg. Oxypurinol clearance is decomposed into non-renal (0.178 L/h/70 kg FFM) plus renal (0.777 L/h/70 kg FFM per creatinine clearance of 6 L/h) components, with a 39 percent reduction of the renal component in patients on concomitant diuretic therapy. All allopurinol structural parameters, allometric exponents, oral bioavailability, and the molar conversion fraction were held FIXED during the sequential fit; oxypurinol structural parameters, IIV, and residual error were estimated. Wright 2013 Table 2 reports residual epsilon components under a ‘(CV%)’ column label whose numeric values are interpreted here as SD on fraction scale (see the model’s validation vignette Assumptions and deviations section for the rationale).
Allopurinol (Wright 2016) One-compartment population PK-PD model for allopurinol (via the active metabolite oxypurinol) and plasma urate in adults with gout (Wright 2016 BJCP). Oxypurinol disposition is a one-compartment first-order absorption / first-order elimination model with Ka fixed at 1.09 1/h; apparent oxypurinol clearance (CL/F_oxy) is allometrically scaled on fat-free mass (Janmahasatian formula, exponent fixed at 0.75) and power-scaled on Cockcroft-Gault creatinine clearance standardised to 70 kg, with a multiplicative reduction when a thiazide or loop diuretic is coadministered; apparent volume (V/F_oxy) is allometrically scaled on total body weight (exponent fixed at 1.0) and shares its IIV with CL/F_oxy via a fixed fractional scaler (Bonate 2006 fractional-effect parameterisation). Plasma urate is described by a direct-effect sigmoidal Emax inhibition of urate production on top of a baseline urate U0 that is power-scaled on renal function and multiplicatively higher with concomitant diuretic. The dose entered into the model is allopurinol oral mg; the implicit 1:1 molar conversion to oxypurinol is absorbed into the apparent CL/F_oxy and V/F_oxy. Cc is oxypurinol concentration (umol/L) and Eurate is plasma urate (mmol/L).
Alvespimycin (Aregbe 2012) Three-compartment population PK model for the heat shock protein 90 inhibitor 17-DMAG (alvespimycin, NSC 707545) given as a 1 h IV infusion to adult patients with advanced solid tumors (Aregbe 2012), with first-order elimination, log-normal IIV on CL/Q3/V1/V2/V3, and between-occasion variability on Q2 and V1 multiplexed by an OCC indicator across up to five daily dosing occasions.
Amatuximab (Gupta 2016) Two-compartment population PK model with parallel linear and Michaelis-Menten elimination for amatuximab in patients with advanced cancers / malignant pleural mesothelioma (Gupta 2016)
Amg221 (Gibbs 2011) Semi-mechanistic population PK/PD model of subcutaneous adipose 11-beta-hydroxysteroid dehydrogenase type 1 (11-beta-HSD1) activity after single oral doses of AMG 221, a selective 11-beta-HSD1 inhibitor, in healthy obese adults (Gibbs 2011). Two-compartment PK with first-order absorption and lag time; dose-dependent bioavailability reduced at 3 mg vs 30/100 mg; adipose-tissue effect compartment linked to plasma by a first-order equilibration rate constant (keo) and to the tissue-concentration observation by an adipose-plasma density correction factor (kpp); direct-effect Imax inhibitory model on 11-beta-HSD1 activity driven by the effect-site plasma-equivalent AMG 221 concentration.
Amifampridine (Thakkar 2017) Joint parent-metabolite population PK + fractional-Emax PD model for 3,4-diaminopyridine (3,4-DAP, amifampridine) free base and its N-acetyl metabolite 3-Ac DAP in 49 adults with Lambert-Eaton myasthenia (Thakkar 2017). Two-compartment parent + one-compartment metabolite with Fm fixed to 1 (all parent clearance forms metabolite). Body weight is allometrically scaled on CL/F and CLm/F3ACDAP (exponent 0.75 fixed) and linearly on Vp/F (exponent 1 fixed), all with reference weight 82 kg. Serum creatinine acts on CLm/F3ACDAP through (0.8/SCR)^0.7 with median SCR 0.8 mg/dL. The PD submodel describes the Triple Timed Up and Go (3TUG) score in seconds via a fractional-inhibitory Emax equation Effect = E0 * (1 - Emax * Cp / (EC50 + Cp)) where Cp is the parent 3,4-DAP plasma concentration in ng/mL.
Amikacin (Delattre 2010) Two-compartment IV population PK model for amikacin in critically ill adult patients with severe sepsis or septic shock during the first 24 hours of antibiotic treatment (Delattre 2010)
Amikacin (Tod 1998) Two-compartment intravenous population PK model for amikacin in febrile, severely neutropenic adults with hematological malignancies (Tod 1998); clearance modeled as the sum of a non-renal intercept and a Cockcroft-Gault-like renal component with sex-stratified slope coefficient (males theta_1, females theta_2), age-correction factor (theta_3 - AGE/100), and Cockcroft-Gault-like renal-function ratio (WT / CREAT). Power-variance residual-error model.
Aminocaproic acid (Stricker 2015) Two-compartment IV population PK model for epsilon-aminocaproic acid (EACA) in infants undergoing craniofacial reconstruction and adolescents undergoing posterior spinal fusion surgery (Stricker 2015)
AminocaproicAcid (Stricker 2013) Two-compartment IV population PK model for epsilon-aminocaproic acid (EACA) in infants (2-24 months) undergoing craniofacial reconstruction surgery. Allometric scaling on body weight (reference 8.82 kg; fixed exponents 0.75 on CL and Q, 1.0 on V1 and V2), an asymptotically increasing post-natal age maturation effect on clearance (age50 = 7.36 weeks), and binary intra-operative-period multipliers on CL (0.89) and V1 (0.80) capturing the composite effect of anaesthesia, blood loss, and surgical fluid management. Parameter values from Stricker 2013 Table 4.
Amivantamab (Marcantonio 2022) QSP. Two-compartment bispecific anti-receptor x anti-receptor mechanistic PKPD model of amivantamab-(EGFR, c-Met) binding in adults with EGFR-exon-20-mutant non-small-cell lung cancer (Marcantonio 2022 Early Feasibility Assessment, Case Study 2). Amivantamab has one arm for EGFR (Kd 1.4 nM) and one arm for c-Met (Kd 0.04 nM) with valency 1 per target (per Jarantow 2015). Model tracks membrane EGFR (R1), membrane c-Met (R2), and soluble c-Met (S2) at 5.9 nM in both central and peripheral compartments. Soluble EGFR is NOT modelled (shed_css_1 = 0 in JSON). Drug states: free drug (Ab_00) plus five populated bound forms Ab_R1, Ab_R2, Ab_S2, Ab_R1R2, Ab_R1S2 per compartment. Since each arm is monovalent, the states with double- target arm binding (Ab_R1R1, Ab_R2R2, Ab_S2S2) do not populate and are omitted. Parameters FIXED from paper Tables 3 (target-side EGFR/c-Met) and 4 (drug-side) plus the Assess run file JSON; disease and toxicity compartments omitted per paper Case Study 2 text.
Amlitelimab (Tiraboschi 2025) Two-compartment population PK model for amlitelimab (anti-OX40L mAb) in adults, with parallel first-order and Michaelis-Menten (TMDD) elimination, SC absorption with lag time, allometric body-weight scaling, and SCORE_EASI / albumin covariate effects (Tiraboschi 2025)
Amlodipine valsartan (Heo 2016) Joint two-drug population PK/PD model for the antihypertensive interaction of amlodipine (calcium-channel blocker, unsuffixed parent) and valsartan (angiotensin-II receptor blocker, sibling-drug suffix _val) on systolic (SBP) and diastolic (DBP) blood pressure in healthy adult Korean volunteers receiving a single-dose fixed-dose- combination tablet of amlodipine 10 mg + valsartan 160 mg. Each drug has a two-compartment popPK model with zero-order absorption (duration D1) and theory-based allometric weight scaling on CL, Q (exponent 0.75) and V1, V2 (exponent 1) at a reference weight of 70 kg. PD uses an effect-compartment Imax model on BP: amlodipine drives two separate effect compartments (SBP-side and DBP-side, different Keqs); valsartan drives one shared effect compartment (single Keq). Imax is fixed at 0.164 (the valsartan monotherapy estimate) and applies to all four drug/endpoint arms. Combined therapy uses a proportional interaction term (Heo 2016 Eq 8): BP = BSL * (1 - PD_amlo - PD_val - alpha * PD_amlo * PD_val) with PD_x = Imax * Ce_x / (IC50_x + Ce_x). alpha < 0 = infra-additive, alpha = 0 = additive, alpha > 0 = synergistic. Estimated alpha = -0.171 for SBP and -0.0312 for DBP (both infra-additive). Combined- therapy baselines (BSL_SBP = 117 mmHg, BSL_DBP = 72.8 mmHg) and alpha are estimated on the 48-subject FDC dataset; monotherapy IC50s and Keqs are fixed at their monotherapy point estimates (Tables 1 and 2).
Amodiaquine (Ali 2018) Joint parent-metabolite population PK model for oral amodiaquine and its CYP2C8-derived active metabolite desethylamodiaquine in adults and children with uncomplicated Plasmodium malaria, pooled across five WWARN cohorts (Burkina Faso, Ghana, Kenya, Uganda, Thailand). Two-transit absorption (NN = 2) into a 2-compartment amodiaquine disposition model with complete in-vivo conversion (with MW correction) to a 3-compartment desethylamodiaquine disposition model. Allometric body-weight scaling on CL/Q (exponent 0.75) and Vc/Vp (exponent 1.0) referenced at WT = 50 kg; sigmoidal postmenstrual-age maturation on both amodiaquine and desethylamodiaquine clearance; 22.4% lower bioavailability on the first daily dose relative to subsequent doses.
Amoxicillin (Fournier 2018) Two-compartment IV population PK model for amoxicillin in adult ICU burn patients hospitalized at a Swiss tertiary-care centre, with Cockcroft-Gault creatinine clearance as a linear covariate on CL (centered at 110 mL/min) and body weight as a linear (allometric exponent 1) covariate on the central volume V1 (centered at 70 kg) (Fournier 2018).
Amoxicillin (Keij 2023) One-compartment population PK model with first-order absorption for oral and intravenous amoxicillin in preterm and term neonates treated for possible serious bacterial infection, pooled from three studies (RAIN, Maastricht/Pullen, and SATT). Clearance carries fixed allometric weight scaling (exponent 0.75, reference 70 kg) plus two estimated maturation power terms on postnatal age (reference 6.8 days) and gestational age (reference 35.8 weeks); central volume carries fixed linear weight scaling (exponent 1.00, reference 70 kg). Oral bioavailability is 87.3% and absorption is slow (Ka 0.085 1/h), so oral profiles are absorption-rate-limited. Interindividual variability is on clearance only; residual error is combined additive plus proportional (Keij 2023).
Amoxicillin (Muller 2008) Three-compartment population PK model for intravenous amoxicillin in pregnant women before, during and immediately after labour, with labour-state binary indicators reducing the peripheral volume V2 during active labour (-13.7%) and the immediate postpartum period (-29.5%) relative to before labour (Muller 2008).
Amoxicillin (Tang 2019) Two-compartment population PK model with first-order elimination for intravenous amoxicillin in Chinese neonates and young infants (Tang 2019). Current weight enters as a fixed allometric power on both volumes (exponent 1) and on CL and Q (exponent 0.75); CL is further modulated by a maturation factor F_age that is the product of two power functions of gestational age and postnatal age. Interindividual variability is estimated on the peripheral volume V2 and on CL only; residual variability follows an exponential model (proportional in linear space).
Amphenmulin chicken (Wang 2024) Preclinical (broiler chicken). One-compartment intravenous PK model for amphenmulin, a novel pleuromutilin derivative, in healthy yellow-feathered broiler chickens given a single 20 mg/kg body-weight dose. Wang 2024 analysed the plasma concentration-time data non-compartmentally in Phoenix 8.1 and reported no structural compartmental model, but the paper states that a one-compartment open model with first-order elimination was the structure used to drive the companion in-vitro dynamic system, and the intravenous NCA parameters in Table 1 determine such a model completely and self-consistently: CL = 1.17 L/h/kg and Vss = 3.64 L/kg give kel = 0.321 1/h (Table 1 T1/2Ke = 2.13 h implies 0.325 1/h) and MRT = Vss/CL = 3.11 h (Table 1 MRT = 3.13 h). ONLY the intravenous route is packaged. The oral and intramuscular routes reported in Table 1 are deliberately excluded because they cannot be represented by any one-compartment model built on these intravenous parameters: the published oral Cmax of 0.73 ug/mL is 2.26 times F*Dose/Vss = 0.323 ug/mL, which is the absolute upper bound on the concentration a one-compartment model can reach after an oral dose of 20 mg/kg at F = 5.88%, and the published oral Cmax/AUC ratio of 0.658 1/h exceeds kel = 0.321 1/h, which a one-compartment model also cannot produce. The intramuscular route is feasible in magnitude but Wang 2024 reports no intramuscular absorption rate constant, and its Tmax of 0.38 h is incompatible with its own terminal rate constant of 0.13 1/h (flip-flop absorption at 0.13 1/h would place Tmax at 4.7 h). See the vignette Assumptions and deviations section. Wang 2024 reports mean and SD of the individual NCA estimates but no population model, so no between-subject variability is encoded and the residual error is held at zero for typical-value simulation.
Amphenmulin killrate (Wang 2024) In vitro (Mycoplasma gallisepticum strain S6). Concentration-dependent kill-rate PK/PD model for amphenmulin, a novel pleuromutilin derivative, coupling the one-compartment first-order-elimination PK of Wang 2024’s in-vitro dynamic system to the sigmoid Emax kill-rate relationship fitted to the static time-kill curves. Wang 2024 Materials and Methods (Static time-kill curves) parameterises the kill rate as E = E0 + Emax * Ce^N / (EC50^N + Ce^N), where E is the kill rate in log10(CFU/mL) per hour (POSITIVE = killing), E0 is the corresponding rate of change in the untreated control (NEGATIVE, i.e. net growth), Ce is the amphenmulin concentration and Emax is an INCREMENT above E0 rather than the asymptote, so the saturating kill rate is E0 + Emax. Note that Wang 2024’s Results text calls Emax ‘the maximum kill rate’; the printed equation, which matches the Phoenix Sigmoid Emax model the authors used, is taken as authoritative. Parameters are Wang 2024 Table 2, row 0-24 h, which the paper identifies as the optimal fit (R = 0.9936) and plots in Figure 4b: Emax = 0.1261 1/h, EC50 = 0.0325 ug/mL, E0 = -0.0093 1/h, Hill N = 0.9238. The bacterial density bact (linear CFU/mL) is integrated as d/dt(bact) = -ln(10) * E * bact so that log10(bact) falls at exactly E log10(CFU/mL) per hour, reproducing the paper’s fitted kill rate exactly at any constant concentration. The PK is the paper’s in-vitro dynamic model (Materials and Methods, In vitro dynamic model): first-order elimination C = C0 * exp(-kt) with k set from the chicken intravenous half-life T1/2Ke = 2.13 h, giving k = 0.3254 1/h, realised in the apparatus as a 1.63 mL/min flow through the 300 mL reaction chamber (1.6360/300 = 0.326 1/h). IMPORTANT PROVENANCE NOTE: Wang 2024 fitted the kill-rate parameters to STATIC concentrations and ran the dynamic apparatus as a separate experiment; the two were not fitted jointly. This model composes them, which is the paper’s stated PK/PD integration written as an ODE system. Setting the elimination rate to zero recovers the static experiment and reproduces Table 2 exactly. Wang 2024 reports no between-subject variability and no residual error magnitude for either component, so no eta parameters are present and both residual SDs are held at zero for typical-value simulation.
Amphenmulin pkpd index (Wang 2024) In vitro (Mycoplasma gallisepticum strain S6). Inhibitory sigmoid Emax PK/PD-index model for the anti-mycoplasma effect of amphenmulin, a novel pleuromutilin derivative, against M. gallisepticum strain S6 in Wang 2024’s in-vitro dynamic model. Wang 2024 Materials and Methods (Integration and modeling of pharmacokinetics/pharmacodynamics) parameterises the effect over a 24 h dosing interval as E = E0 - (E0 - Emax) * Ce^N / (EC50^N + Ce^N), where E is the SIGNED change in M. gallisepticum counts over 24 h of cultivation in log10(CFU/mL) (NEGATIVE = bacterial reduction), E0 is the corresponding change in the untreated control, Emax is the maximal (most negative) achievable change, and Ce is a PK/PD index. The packaged model uses the paper’s best-correlating index AUC24h/MIC (R = 0.9657, versus 0.8995 for Cmax/MIC), formed as the per-interval covariate AUC_AMPH divided by the parameter mic. Parameters from Wang 2024 Table 3, AUC24h/MIC column: Emax = -2.4214 log10(CFU/mL), E0 = -0.3845 log10(CFU/mL), EC50 = 1199.4720 h, Hill N = 3.1997. Note the sign convention differs from the companion Wang_2024_amphenmulin_killrate model, whose Table 2 parameters are kill RATES with positive meaning killing; here the effect is a signed change with negative meaning killing. Substituting Table 3 into the equation at the paper’s headline target of AUC24h/MIC = 904.05 h returns E = -0.97 log10(CFU/mL) against the stated 1 log10 reduction, confirming the sign convention and the equation form to within the rounding of the four-significant-figure parameters. The bacterial density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * (E / 24) * bact so that log10(bact) changes by exactly E across each 24 h interval, reproducing the paper’s per-interval model exactly at every 24 h boundary. There is no PK compartment: exposure enters as the externally supplied per-interval AUC_AMPH, because Wang 2024 derived the AUC of the in-vitro dynamic system non-compartmentally in Phoenix. Wang 2024 reports no between-subject variability and no residual error magnitude, so no eta parameters are present and addSd is held at zero for typical-value simulation.
AmphotericinB liposomal (Hong 2006) Two-compartment population PK model for liposomal amphotericin B (AmBisome) in 39 pediatric oncology patients receiving 1-h IV infusions (Hong 2006). Clearance and central volume scale exponentially with body weight centered at the cohort-median 21 kg; the paper additionally reports substantial between-occasion variability on CL and V1 that is encoded here as IIV on fixed-at-1 multiplicative anchors (Bellanti 2015 IOV-as-IIV pattern).
AmphotericinB liposomal calbicans (Beredaki 2024) In vitro (two Candida albicans isolates; one-compartment dilution PK/PD model). Liposomal amphotericin B (L-AMB) in vitro PK/PD model for the Candida albicans arm that Beredaki 2024 used to VALIDATE the in vitro system against a published neutropenic-mouse model of disseminated candidiasis. Beredaki 2024 simulated MOUSE L-AMB pharmacokinetics in the internal compartment of a dilution model, re-spiking to a constant target peak once daily with an average half-life of 8 h (range 7-11 h; target 11 h). The 48 h change in log10 CFU/mL from the starting inoculum was then related to the PK/PD index Cmax/MIC with the sigmoidal variable-slope Emax model E = Emax * EI^n / (EI50^n + EI^n) (Methods, ‘PK/PD analysis’), where E is the REDUCTION in log10 CFU/mL relative to the drug-free control and the MIC is the CLSI M27 amphotericin B deoxycholate MIC. Beredaki 2024 printed the equation and its goodness of fit (R^2 = 0.91) but no coefficients, so e0, lemax, lec50 and lhill were recovered by digitising the fitted curve in Figure 2; the recovered coefficients place the stasis exposure at 2.06 Cmax/MIC against the 2.1 (0.5-3.9) printed in Figure 2, which in turn brackets the 1.6-3.8 Cmax/MIC required for stasis in mouse kidneys. The fungal density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * ((e0 - kill48) / 48) * bact so that log10(bact) changes by exactly (e0 - kill48) across the paper’s 48 h observation window, reproducing the endpoint model exactly at the time the paper actually fitted it. The Candida auris models this arm validates are Beredaki_2024_amphotericinB_liposomal_cauris48h and Beredaki_2024_amphotericinB_liposomal_cauris24h.
AmphotericinB liposomal cauris24h (Beredaki 2024) In vitro (four Candida auris clinical isolates; one-compartment dilution PK/PD model). Liposomal amphotericin B (L-AMB) in vitro PK/PD model for the 24 h exposure-response relationship. Beredaki 2024 simulated human L-AMB pharmacokinetics in the internal compartment of a dilution model, re-spiking to a constant target peak once daily for 48 h with an average half-life of 10 h (range 5-12 h; target 9 h). The 24 h change in log10 CFU/mL from the starting inoculum was then related to the PK/PD index Cmax/MIC with the sigmoidal variable-slope Emax model E = Emax * EI^n / (EI50^n + EI^n) (Methods, ‘PK/PD analysis’), where E is the REDUCTION in log10 CFU/mL relative to the drug-free control and the MIC is the CLSI M27 amphotericin B deoxycholate MIC. Beredaki 2024 printed the equation and the fitted curves (Figure 5, R^2 = 0.86) but no coefficients, so e0, lemax, lec50 and lhill were recovered by digitising the 24 h curve in Figure 5; the recovered coefficients place the stasis exposure at 5.02 Cmax/MIC against the 5 (3-7) reported in the Figure 5 legend. The fungal density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * ((e0 - kill24) / 24) * bact so that log10(bact) changes by exactly (e0 - kill24) across the paper’s 24 h observation window, reproducing the endpoint model exactly at the time the paper actually fitted it. The companion 48 h parameterisation – the one the paper carries into its Monte Carlo target-attainment analysis – is Beredaki_2024_amphotericinB_liposomal_cauris48h, and the Candida albicans validation arm is Beredaki_2024_amphotericinB_liposomal_calbicans.
AmphotericinB liposomal cauris48h (Beredaki 2024) In vitro (four Candida auris clinical isolates; one-compartment dilution PK/PD model). Liposomal amphotericin B (L-AMB) in vitro PK/PD model for the 48 h exposure-response relationship. Beredaki 2024 simulated human L-AMB pharmacokinetics in the internal compartment of a dilution model, re-spiking to a constant target peak once daily for 48 h with an average half-life of 10 h (range 5-12 h; target 9 h). The 48 h change in log10 CFU/mL from the starting inoculum was then related to the PK/PD index Cmax/MIC with the sigmoidal variable-slope Emax model E = Emax * EI^n / (EI50^n + EI^n) (Methods, ‘PK/PD analysis’), where E is the REDUCTION in log10 CFU/mL relative to the drug-free control and the MIC is the CLSI M27 amphotericin B deoxycholate MIC. Beredaki 2024 printed the equation and its goodness of fit (R^2 = 0.86) but no coefficients, so e0, lemax, lec50 and lhill were recovered by digitising the fitted 48 h curve in Figure 5; the recovered coefficients place the stasis exposure at 9.03 Cmax/MIC against the 9 (6-14) reported in Figure 5. The fungal density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * ((e0 - kill48) / 48) * bact so that log10(bact) changes by exactly (e0 - kill48) across the paper’s 48 h observation window, reproducing the endpoint model exactly at the time the paper actually fitted it. The companion 24 h parameterisation is Beredaki_2024_amphotericinB_liposomal_cauris24h and the Candida albicans validation arm is Beredaki_2024_amphotericinB_liposomal_calbicans.
Ampicillin (Tremoulet 2014) One-compartment IV population PK model for ampicillin in preterm and term neonates (Tremoulet 2014; opportunistic POPS / PTN study). Clearance is allometrically scaled linearly to body weight and modulated by a serum-creatinine power factor (0.6/SCR)^0.428 and a postmenstrual-age power factor (PMA/37)^1.34. Central volume scales linearly with body weight (0.399 L/kg). Inter-individual variability is supported on CL only; residual variability is proportional.
Ampicillin sulbactam (Soto 2014) Joint two-compartment population PK model for ampicillin and sulbactam in 47 Japanese adults with moderate or severe community-acquired pneumonia receiving 30-minute IV infusions of 3 g ampicillin/sulbactam (2:1) every 6 hours (Soto 2014). Both drugs are fitted simultaneously via the NONMEM L2 data item; a single common Cockcroft-Gault CLcr power effect (0.701) is applied to CL of both drugs, and CL random effects are correlated across drugs (rho = 0.858). Body weight is a fixed linear allometric scalar on peripheral volume V2 for both drugs. Ampicillin uses the unsuffixed canonical compartment / parameter set; sulbactam carries the sibling-drug suffix _sbt throughout.
Amyloid (vanMaanen 2025) Population exposure-response model of brain amyloid plaque burden (Centiloid, CL) in patients with mild cognitive impairment or mild dementia due to Alzheimer’s disease. A single indirect-response (turnover) ODE describes the natural plaque time course (zero-order formation Kin, first-order elimination Kout). The BACE1 inhibitor verubecestat inhibits plaque formation via a fixed-Imax AUC-driven sigmoid on Kin (Inh_verub, AUC50 in uMh of verubecestat AUC over 24 h at steady state; Imax fixed to 1 per Table 1). Each of four anti-A-beta monoclonal antibodies (aducanumab, donanemab, gantenerumab, lecanemab) stimulates plaque elimination via a linear AUC/molecular-weight term on Kout (Stim_mAb = slope_mAb (AUC_mAb / MW_mAb), with distinct slopes per mAb). AUC of each drug is a time-varying model input (covariate); the paper’s simulations set each AUC to its steady-state value during dosing periods and to 0 during washout. Baseline plaque burden is a per-subject covariate used as the plaque initial condition. Fit by NONMEM FOCE-I to 370 individual verubecestat amyloid PET measurements from 188 aMCI due to AD subjects (phase 3 APECS, NCT01953601) pooled with 120 summary-level PET measurements from four anti-A-beta mAb trial programmes; external validation on 1506 amyloid-PET measurements from 521 amyloid-positive ADNI subjects with up to 10 years of follow-up. No IIV is encoded (paper: ‘no additional interindividual variability was included’; summary-level mAb data constrained the model to central-tendency predictions).
Anakinra (Urien 2013) One-compartment population pharmacokinetic model for subcutaneous anakinra (recombinant nonglycosylated human IL-1 receptor antagonist) in 87 children and adolescents (8 months to 21 years, 4.3 to 83 kg) treated for systemic-onset juvenile idiopathic arthritis (SJIA) and diverse autoinflammatory syndromes (Urien 2013). First-order absorption (Ka) into a single central compartment with first-order elimination; apparent clearance CL/F and apparent volume V/F are allometrically scaled to body weight with estimated power exponents (0.47 on CL/F and 0.76 on V/F, reference 70 kg). Inter-individual variability is reported on CL/F and between-occasion variability on V/F; no other covariate effect (age, sex, co-administered anti-inflammatory drugs) was retained.
Angiotensin (BuchwalderCsajka 1999) Population pharmacodynamic dose-response model of the peak systolic (SBP) and diastolic (DBP) blood pressure increase elicited by a single intravenous bolus of exogenous angiotensin (used as a pharmacologic probe / ‘challenge’) in 228 healthy male volunteers across 13 phase I trials of antihypertensive drugs acting on the renin-angiotensin system. The final structural form is the molecular-weight-corrected Emax model E = Emax * D / (D + ED50) (Buchwalder-Csajka 1999 Table 1 last row), where D = DOSE_AGT_UG is the angiotensin challenge dose in ug already expressed as angiotensin II equivalents (multiply an angiotensin I dose by Q = 0.78 in data preparation; the paper’s text reports Q = 0.78 as the molar-weight ratio), and Emax / ED50 are estimated separately for SBP and DBP. This is a purely algebraic snapshot model: no PK, no time course, no ODEs. Each observation row in the event dataset carries one DOSE_AGT_UG value (the dose given just before the peak was sampled) and yields one peak BP increase. The model is suitable for simulating the peak BP response to a single angiotensin bolus during dose-finding and placebo-period segments of an angiotensin-challenge phase I protocol; it is NOT a model of the antihypertensive drugs whose trials supplied the data.
Anifrolumab (Almquist 2022) Two-compartment QSS-TMDD population PK model for anifrolumab (anti-IFNAR1 IgG1-kappa) in healthy volunteers and adults with systemic lupus erythematosus (Almquist 2022): linear plus quasi-steady-state target-mediated elimination via a dynamic IFNAR1 receptor pool, time-varying linear clearance (Emax-on-time), and IFNGS-high/low and body-weight covariate effects.
Anrukinzumab (Hua 2015) Two-compartment population PK model for anrukinzumab (anti-IL-13 IgG1 monoclonal antibody) with first-order SC absorption and linear elimination, pooling healthy volunteers, mild-to-moderate asthma, moderate-to-severe asthma, and ulcerative colitis patients (Hua 2015)
Anthracycline troponinT (deVriesSchultink 2018) Kinetic-pharmacodynamic (K-PD) direct-effect model for serum high-sensitive cardiac troponin T (hs-TnT) in early-breast-cancer patients receiving an adjuvant anthracycline regimen (de Vries Schultink 2018). The dose enters a virtual K-PD body amount compartment with first-order elimination at rate kel; the linear direct effect TRP = TRP0 * (1 + SLOPE * Aant) raises serum troponin T above a population baseline TRP0 in proportion to the current K-PD amount. The proportional SLOPE is anthracycline-type dependent: epirubicin produces a 0.524-fold smaller effect than the doxorubicin reference, encoded jointly by the CONMED_DOXORUBICIN and CONMED_EPIRUBICIN indicators. No other covariates retained. Companion file deVriesSchultink_2018_trastuzumab_LVEF.R consumes the per-subject peak troponin T from this model as a covariate.
Anti tryptase (Rymut 2023) Mechanistic population PK/PD model for the anti-tryptase IgG4 monoclonal antibody MTPS9579A in healthy adults and adults with moderate-to-severe asthma (Rymut 2023). Two-compartment serum disposition with first-order SC absorption and allometric weight scaling on linear CL and central volume; quasi-equilibrium (QE) TMDD describes saturable binding of MTPS9579A to total monomeric serum tryptase; a mechanistic airway interstitial-fluid (ISF) compartment receives free mAb and mAb-monomer complex from the systemic circulation through lymph flow with vascular reflection coefficients; in the ISF, tryptase is secreted as the active tetramer (target_isf), spontaneously dissociates into inactive monomers (monomer_isf), and is rapidly disrupted by bound mAb (kbreak); free MTPS9579A binds tetramer and monomer with the same KD. Estimated systemic TMDD parameters come from a NONMEM 7.4.3 SAEM fit to 106 healthy Phase 1 subjects (Table 1); fixed mechanistic ISF parameters come from in vitro / physiological literature and from a healthy-subject visual fit of the upper-airway biodistribution coefficient at 3% (Methods, Table S2, Figure S2).
Antithrombin (DeJongh 2014) Two-compartment population PK model with linear elimination from the central compartment for recombinant human antithrombin (rhAT) in adults with hereditary antithrombin deficiency (HD) undergoing perioperative or peripartum treatment (DeJongh 2014). Delivery (pregnancy) status approximately doubles CL and Vss versus non-pregnant surgery patients. Baseline AT activity is modelled as a subject-level endogenous parameter (Dansirikul-type baseline handling) that adds to the drug-derived AT activity in the observation.
Antithrombin (Moffett 2017) One-compartment population PK model with exponential residual error for intravenous human (plasma-derived) antithrombin (AT) concentrate in paediatric patients (Moffett 2017). Structural clearance uses an allometric weight arm with additive concurrent-UFH-dose potentiation; volume of distribution uses allometric weight scaled by a power of the per-subject baseline AT activity level. The observed AT activity level is the sum of dose-derived AT concentration plus the endogenous baseline (Dansirikul B3 handling of endogenous baseline).
Apitolisib human (Moein 2024) Integrated PK-PD-efficacy model for orally administered apitolisib (GDC-0980), a dual PI3K/mTOR inhibitor, in patients with advanced solid tumors or non-Hodgkin’s lymphoma (two phase 1 dose-escalation studies, NCT00854152 and NCT00854126). Three sequentially-fitted layers in one ODE system: (1) two-compartment first-order-absorption population PK with an absorption lag (individual estimates carried in and held fixed); (2) an indirect-response model in which apitolisib inhibits the production rate of phosphorylated Akt (pAkt) measured in PLATELET-RICH PLASMA as a surrogate for tumor target modulation, expressed as %pAkt relative to a drug-free baseline of 100%; (3) exponential growth of the RECIST sum of longest diameters at net rate kg opposed by a sigmoidal shrinkage rate ks driven by the percent pAkt inhibition I = 100 - %pAkt. Imax and the pAkt Hill exponent are fixed at 1, and kout is not estimated but derived as kin/%pAkt(0) = kin/100, so the biomarker pool equilibrates essentially instantaneously and %pAkt tracks the plasma concentration directly. Only the five efficacy parameters (kg, kmax, ki50, gamma2, proportional error) and the four tumor IIV terms were estimated in this integrated fit. Companion preclinical model: modellib(‘Moein_2024_apitolisib_mouse’).
Apitolisib mouse (Moein 2024) Preclinical (mouse; 786-O human renal cell adenocarcinoma xenograft). Integrated PK-PD-efficacy model for orally administered apitolisib (GDC-0980), a dual PI3K/mTOR inhibitor. Three sequentially-fitted layers in one ODE system: (1) one-compartment first-order-absorption PK (CL/F, V/F, ka carried in from the mouse PK model and held fixed); (2) an indirect-response model in which apitolisib inhibits the production rate of phosphorylated Akt (pAkt) measured in TUMOR TISSUE, expressed as %pAkt relative to a drug-free control baseline of 100%; (3) exponential tumor-volume growth at net rate kg opposed by a sigmoidal shrinkage rate ks driven by the percent pAkt inhibition I = 100 - %pAkt. Imax and the pAkt Hill exponent are fixed at 1, and kout is not estimated but derived as kin/%pAkt(0) = kin/100, so the biomarker pool equilibrates essentially instantaneously and %pAkt tracks the plasma concentration directly. Only the five efficacy parameters (kg, kmax, ki50, gamma2, proportional error) and the two tumor IIV terms were estimated in this integrated fit.
Apixaban (Ueshima 2018) One-compartment population pharmacokinetic and pharmacogenomic model for oral apixaban in Japanese adult patients with atrial fibrillation (Ueshima 2018). Apparent oral clearance CL/F is the sum of an apparent renal arm (power on creatinine clearance, CCR/70) and an apparent non-renal arm carrying two recessive-/dominant-style pharmacogenomic factors: CYP3A5 3 carrier (genotype 1/3 or 3/*3) reduces non-renal CL/F by a factor of 0.312, and ABCG2 421A/A (rs2231142 homozygous variant) reduces non-renal CL/F by a factor of 0.341. Apparent volume of distribution Vd/F = 24.7 L (no significant covariates). Absorption rate constant ka was fixed at 0.42 1/h from a prior publication (Frost 2013 Br J Clin Pharmacol, reference 13 in the paper) because the sparse trough-and-2-point-postdose sampling design lacked enough absorption-phase data to identify ka.
Apomine (Bonate 2004) Two-compartment population PK model for oral apomine (a synthetic bisphosphonate-ester anti-cancer agent) in 38 subjects – 19 healthy adult males and 19 male and female patients with advanced solid tumours – pooled from four model-development studies (Bonate 2004 Studies 1, 2, 5, 6) with three validation studies (Studies 3, 4, 7). Apomine is administered orally with or without food in single and multiple-dose regimens over 30 to 2100 mg total daily dose. Disposition is a two-compartment model with linear elimination and a first-order absorption mixture: a dominant Group 1 subpopulation (97 %, paper P1 logit) with an estimable lag time and faster absorption rate, and a minority Group 2 subpopulation (3 %) with no lag time and slower absorption (Table 3). Apparent oral clearance is time-dependent via an empirical sigmoid Emax auto-induction model in elapsed time (CL = CL0 + CLmax * time^n / (cl_hill_t50^n + time^n); Table 3), reaching 50 % of the maximal induction-driven increment in about two days. Relative bioavailability F1 is dose-saturable (F1 = D50 / (Dose + D50)) with an additional fractional food effect (1 + theta_food * FED), where the F1max anchor is structurally fixed at 1 (Table 3). Cancer patients have lower baseline CL/F and lower central volume than healthy males (encoded via the DIS_CANCER indicator with log-additive effects e_cancer_cl and e_cancer_vc back-derived from Table 3); intercompartmental clearance and peripheral volume are common to both populations. Central volume scales proportionally with body weight at a fixed allometric exponent of 1.0 (Table 3). A bimodal high-Vp subpopulation observed in the four healthy-male multiple-dose subjects (Bonate 2004 Study 2) is encoded as a binary indicator MIX_HIGH_VP multiplying the typical peripheral volume by 23.5. Inter-occasion variability (18 % on CL and Vc, Table 3) is omitted from this file because occasion definitions are study-design-specific (per the standard nlmixr2lib practice; see vignette Assumptions and deviations).
Apx3330 (Silva 2024) Population pharmacokinetic model for APX3330 (E3330), a selective APE1/Ref-1 redox inhibitor, measured as total quinone in plasma after oral dosing. Two-compartment disposition with first-order absorption and an absorption lag time, fitted to 1460 plasma concentrations pooled from four Eisai studies in 49 healthy Japanese male volunteers (single dose-escalation 10-600 mg, multiple-dose 120 mg once or twice daily, and a fasted-vs-fed crossover) and one Apexian phase I study in 19 patients with advanced solid tumors dosed 120-360 mg twice daily. All disposition parameters are apparent (oral) values because absolute bioavailability was never measured. Body weight is a power-model covariate on CL/F (exponent 0.659) and V1/F (exponent 0.839) referenced to 70 kg; the oncology cohort carries a log-additive shift on CL/F, and dosing in the fed state carries a log-additive shift on the absorption lag time. Inter-individual variability is estimated on CL/F, V1/F and ka with a CL/F-V1/F correlation of 0.349, and residual error is proportional (16%). The paper’s companion GastroPlus ACAT semi-mechanistic absorption model is a commercial-platform model and is not reproduced here; see the validation vignette.
Arginine (Brussee 2016) Two-compartment population PKPD model for intravenous L-arginine adjunctive therapy in 73 adults with moderately severe falciparum malaria. Exogenous L-arginine PK is two-compartment IV with allometric scaling on CL and V1 and a multiplicative Papuan-ethnicity effect on CL. Endogenous L-arginine concentration follows a second-order polynomial recovery function indexed from approximately two days before presentation (start of symptoms); lognormal between-subject variability multiplies the typical polynomial. Pharmacodynamic output 1 (exhaled NO, ppb) is linear in the exogenous arginine concentration with a per-subject baseline. Pharmacodynamic output 2 (reactive-hyperemia peripheral-arterial-tonometry index, RH-PAT) is linear in the predicted NO minus its baseline, i.e., linear in the exogenous arginine concentration.
Aripiprazole (Kim 2008) Joint one-compartment population PK model for oral aripiprazole and its active metabolite dehydroaripiprazole in 80 Korean psychiatric patients (Kim 2008). First-order absorption (Ka FIXED at 1.06 1/h per a prior popPK analysis; the sparse-sampling design could not identify Ka) into a single aripiprazole central compartment with first-order elimination, and a metabolite central compartment that receives the entire parent elimination flux (fm = 1 assumed for identifiability; metabolite CL and V are apparent values scaled by the unknown fm and reported as CL(m)/fm and V(m)/fm) with first-order elimination of dehydroaripiprazole. Covariate analysis retained CYP2D6 genetic polymorphisms as the only significant covariate on parent CL/F, with four genotype strata fit as independent typical-value clearances (Group I 3.15 L/h, Group II 2.66, Group III 2.27, Group IV 1.83); the paper rejected pooling Groups I+II +III into a single CYP2D6 extensive-metabolizer stratum (uniting them increased OFV by 15.8 points). Age, body weight, gender, and CYP3A5 genetic polymorphisms were screened and not retained. Inter-individual variability is fit on CL/F (shared across strata), V/F, and CL(m)/fm with an estimated covariance between CL/F and CL(m)/fm (value not reported by the paper; see vignette Assumptions and deviations). A proportional residual-error model is used separately for aripiprazole and dehydroaripiprazole.
Aripiprazole (Knights 2015) Two-compartment population PK model for oral aripiprazole in adult psychiatric patients (Knights 2015), with first-order absorption, linear-deviation weight (gated by WT < 115 kg) and age effects on apparent oral clearance, multiplicative CYP2D6 poor-metabolizer effect on CL/F, linear weight (gated by WT < 115 kg) and age effects with multiplicative female-sex effect on the peripheral volume, linear weight (gated by WT < 115 kg) effect with multiplicative female-sex effect on apparent inter-compartmental clearance, correlated inter-individual variability across Vc/F, Q/F, and Vp/F, independent IIV on ka and CL/F, and a proportional residual error.
Aripiprazole (Koue 2007) Two-compartment population PK model for oral aripiprazole in healthy Japanese male volunteers (Koue 2007), with first-order absorption, an absorption lag time, body-weight linear scaling on Vc/F, Vp/F, Q/F, and CL/F, additive linear-deviation CYP2D6 intermediate- and poor-metabolizer effects on CL/F (Group 1 = extensive metabolizer reference), an additive linear-deviation itraconazole-coadministration (CYP3A4 inhibitor) effect on CL/F, independent inter-individual variability on every structural parameter, and a log-normal (exponential) residual error.
Artemether (Hietala 2010) Joint parent-metabolite population PK model for oral artemether (ARM) and its active metabolite dihydroartemisinin (DHA) in 50 Tanzanian children (ages 1-10 years, weights 8-30 kg) with uncomplicated Plasmodium falciparum malaria treated with the standard six-dose weight-based Coartem (artemether 20 mg + lumefantrine 120 mg per tablet) regimen at 0, 8, 24, 36, 48, and 60 hours (Hietala 2010). Absorption is first-order with ka fixed at 1/h. Disposition is two-compartment for ARM with complete in-vivo conversion to DHA (bioavailability of DHA fixed at 1 to render the metabolite model identifiable); DHA disposition is one-compartment. The apparent oral clearance of ARM is time-dependent with a linear increase per dose number occasion (CL/F_ARM = theta1 * (1 + theta2 * (OCC - 1)), OCC = 1..6), reproducing a ~3.4-fold rise over the six-dose regimen attributed to enzyme induction. All PK parameters are reported per kilogram body weight (linear weight normalisation applied inside model()).
Artemether (Hoglund 2015) Joint parent-metabolite population PK model for oral artemether and its active metabolite dihydroartemisinin (DHA) in 89 HIV-infected Ugandan adults receiving artemether-lumefantrine (Coartem) with or without concomitant antiretroviral therapy (efavirenz, nevirapine, or lopinavir/ritonavir) (Hoglund 2015). 3-transit-compartment absorption with ka = ktr feeds a 1-compartment artemether disposition; complete in-vivo conversion of artemether to a 1-compartment DHA disposition with stoichiometric molar conversion. Enzymatic auto-induction of the first-pass demethylation of artemether is modelled as a Hill function of time-since-first-dose with fixed maximum maturation (100 % increase in apparent CL) and fixed maturation half-time (62 h, literature value) and an estimated Hill coefficient (0.445). Relative bioavailability F is anchored at 1 (fixed) with log-normal IIV (58.6 % CV). Three antiretroviral drug-drug interactions are encoded as linear-deviation effects: lopinavir/ritonavir increases AM CL/F by 32.8 % and DHA CL/F by 143 %; efavirenz and nevirapine decrease relative bioavailability by 71.5 % and 66.3 % respectively; nevirapine additionally decreases DHA CL/F by 44.5 %. IIV is retained on AM CL, DHA CL, the mean transit time, and F. NONMEM additive residual error on log-transformed concentrations is encoded as a proportional residual in linear concentration space for both parent and metabolite.
Artemether (Mosha 2014) Joint parent-metabolite population PK model for oral artemether (AM) and dihydroartemisinin (DHA) in 33 pregnant (2nd or 3rd trimester) and 22 non-pregnant women with uncomplicated Plasmodium falciparum malaria in Rufiji, Tanzania after standard fixed-dose artemether-lumefantrine (Mosha 2014). One-compartment AM disposition with first-order absorption and linear metabolism to a one-compartment DHA disposition, including a presystemic AM-to-DHA conversion fraction (1 - F1) with F1 = expit(logit_F1) parameterised on the logit scale. Absorption ka fixed at 0.70 1/h and DHA volume fixed equal to AM Vc per the source paper. IIV is present only on AM CL (99% CV); the remaining structural parameters carry no IIV in the final model. None of the available covariates (pregnancy, body weight, BMI, age, gestational age, diarrhoea) reached statistical significance on AM or DHA PK in Mosha 2014 and so none are encoded. AM residual error is combined (proportional plus additive); DHA residual error is proportional.
Artemether (Tarning 2012) Joint parent-metabolite population PK model for oral artemether and its active metabolite dihydroartemisinin (DHA) in 21 pregnant women (2nd or 3rd trimester) with uncomplicated Plasmodium falciparum malaria in Uganda after the standard fixed-dose oral artemether-lumefantrine regimen (Tarning 2012). Absorption is flexible: zero-order dissolution into a depot of duration DUR feeds a 6-compartment transit chain at rate ktr; the same ktr empties transit6 into central (ka set equal to ktr). Disposition is 1-compartment for both artemether and DHA with complete in-vivo conversion of artemether to DHA. Relative bioavailability F is fixed at 1 with log-normal IIV; no statistically significant covariates were retained in the final model (Methods / Results); a single combined additive residual on log-transformed plasma concentrations is shared by both species.
Artemether parasitemia (Hietala 2010) Joint artemether (ARM) + dihydroartemisinin (DHA) PK model coupled to a semimechanistic Plasmodium falciparum parasite life-cycle PD model in Tanzanian children (ages 1-10 years, weights 8-30 kg) with uncomplicated falciparum malaria (Hietala 2010). The PK is the same Table 1 two-compartment ARM (with time-dependent CL/F_ARM via the OCC dose-occasion covariate) and one-compartment DHA structure as modellib(‘Hietala_2010_artemether’). The PD (Table 3) is a five-stage parasite life-cycle model: parasites mature through tinyrings (PTR), smallrings (PSR), largerings (PLR), and mature trophozoites / schizonts (PMT), and parasites killed or injured by drug accumulate in a spleen compartment (Pspleen) before clearance at a fixed elimination rate k_spleen = 0.26 / h (Gordi et al. 2002, ref 9). Replication is encoded as a multiplication factor REPL_p applied to the PMT -> PTR transit (estimated to 4 in symptomatic patients; fixed to 1 in asymptomatic children, not encoded in this file). Drug killing is modelled on all visible developmental stages as k_ARM = S * log[ARM] and k_DHA = S * log[DHA] with shared slope S_ARMDHA = 0.073 (Table 3). Visible parasitemia is the sum of the ring-stage compartments plus the spleen pool. Lumefantrine effect was tested but not retained in the source paper and is intentionally absent from this model.
Artemisinin (Asimus 2007) Semiphysiological autoinduction popPK model for oral artemisinin, fit to pooled plasma data from six clinical studies (33 healthy male Vietnamese volunteers + 54 male falciparum-malaria patients). The structural model is identical to the original Gordi 2005 saliva-based model except no absorption lag-time is estimated. Three artemisinin compartments (gut depot, liver V_H = 1 L fixed, sampling V_S = 26.1 L) are linked in a circular well-stirred-hepatic-extraction loop with hepatic plasma flow Q_H = 0.63 L/h/kg of body weight. Two enzyme states (precursor1 + enzyme pool) drive autoinduction: hepatic artemisinin amount linearly stimulates precursor formation (slope s_ind); the precursor transitions to enzyme with rate constant kpout = 1/MIT = 1/(2.0 h); the enzyme decays with first-order rate kdeg = ln(2)/94 h. Intrinsic clearance is proportional to enzyme amount and saturates in hepatic concentration via Michaelis-Menten kinetics (CL_int,t = vmax * enzyme / (km + C_H)), giving a pre-induced hepatic extraction E_H = 0.74 increasing to 0.98 after autoinduction (a roughly 13-fold drop in oral bioavailability with only a modest change in systemic clearance).
Artemisinin (Birgersson 2016) One-compartment population PK model for oral artemisinin in 15 healthy male Vietnamese volunteers, with a seven-compartment transit-absorption chain (number of transits fixed at 7). The published final model carries inter-occasion variability on apparent clearance and mean transit-time and inter-individual variability on relative bioavailability; for forward simulation in nlmixr2lib the IOV terms are mapped onto etas (etalcl, etalmtt) so a single-occasion simulation reproduces the population variability. The published full covariate analysis found no clinically significant effect (>20 %) of formulation, dose level (160 vs 500 mg), or concomitant piperaquine on the structural PK parameters, so no covariates are carried in the model.
Artemisinin (Sidhu 1998) One-compartment population PK model with first-order absorption for oral artemisinin in 23 paediatric (2-12 y) and 31 adult (16-45 y) Vietnamese patients with uncomplicated falciparum malaria, fit to sparse capillary plasma samples from a 5-day 10 mg/kg/day field-setting regimen. Separate population estimates for CL/F and V/F are carried for adults (per-subject) and children (per-kg body weight) via a CHILD age-group covariate. Time-dependency in artemisinin disposition is modelled as a 6.9-fold systematic decrease in oral bioavailability between the Day 1 and Day 5 doses, with the published inter-occasion variability on apparent CL/F and V/F retained per occasion. Inter-individual variability on CL/F and V/F is collectively estimated for both age groups; the published etaCL/etaVc correlation ‘near unity’ is encoded at 0.95 for numerical stability.
Artesunate (Hendriksen 2013) Joint parent-metabolite population PK model of intramuscular artesunate (ARS) and its active metabolite dihydroartemisinin (DHA) in 70 African children aged 7 months to 11 years admitted with severe Plasmodium falciparum malaria (Hendriksen 2013). Each species has a one-compartment apparent-volume disposition; ARS is delivered by a zero-order input over a 1-min fixed duration (the IM absorption from the injection site, fixed because too few samples were collected during the absorption phase to identify the rate) and is converted mole-for-mole to DHA with no separate parent elimination. Body weight is the dominant covariate (allometric scaling with fixed exponents 0.75 on apparent clearance and 1.0 on apparent volume for both species; reference 10.9 kg), with hemoglobin additionally lowering DHA clearance by 10.2% per g/dL above the reference 7.1 g/dL.
Artesunate (LohyDas 2018) Joint parent-metabolite population PK model of oral artesunate (ARS) and its active metabolite dihydroartemisinin (DHA) in 50 adult patients with uncomplicated, artemisinin sensitive or resistant Plasmodium falciparum malaria in southern Myanmar (Lohy Das 2018, Malaria Journal). ARS absorption is described by a 3-transit-compartment chain (n = 3 fixed) followed by a one-compartment ARS disposition; complete in-vivo conversion of ARS to DHA is assumed (all ARS clearance is metabolic conversion). DHA disposition is one-compartment. Allometric body-weight scaling is applied to all CL (exponent 0.75) and V (exponent 1.0) parameters, centered on the population-median 50 kg. F is fixed at 1. The packaged model file omits the published time-varying parasite-density covariates on MTT and on F (Eqs. 3 and 4) and the entire PD layer (mixture-Emax parasite-killing model with effect compartment); both depend on the upstream Lohy Das 2017 AAPS J paper (ref [36]) which is not on disk. See the vignette’s Assumptions and deviations section for the rationale.
Artesunate (Morris 2011) Joint parent-metabolite population PK model for single-dose oral artesunate (AS) and its active metabolite dihydroartemisinin (DHA) in 26 pregnant and 25 non-pregnant women with asymptomatic Plasmodium falciparum malaria in the Democratic Republic of Congo (Morris 2011). Each species has a one-compartment apparent-volume disposition, with mixed zero-order plus lagged first-order absorption of AS and complete in-vivo conversion of AS to DHA (no separate AS elimination). Pregnancy increases DHA apparent clearance by 42.3% relative to non-pregnant controls (the only retained covariate); the postpartum sub-cohort could not be characterised by a structural model and is not represented.
Artesunate (Simpson 2013) In vitro (P. falciparum). Sigmoid Emax inhibition model of artesunate effect on hypoxanthine uptake by clinical Plasmodium falciparum isolates from the Thai-Myanmar border (Shoklo Malaria Research Unit, 1993-2005), with pfmdr1 genotype covariate effects on EC50. The ‘subject’ in the NLME framework is a parasite isolate (n=474 isolates with artesunate data). STIM_ARTESUNATE_NM is the per-well drug concentration in the in vitro hypoxanthine-uptake-inhibition assay; the model has no PK and no time evolution. E0 and Emax are fixed per Simpson 2013 Table 3 footnote.
Artesunate (Tan 2009) Joint parent-metabolite population PK model of oral artesunate (AS) and its active metabolite dihydroartemisinin (DHA) in 91 healthy Korean adult volunteers (Tan 2009). AS is described by a one-compartment first-order absorption / first-order elimination disposition; DHA by a two-compartment disposition (central + peripheral). AS is converted mole-for-mole to DHA as the only elimination pathway. Body weight linearly increases DHA apparent clearance (1.9 L/h per kg above the 61.5 kg reference) and a high-fat / high-caloric meal at dosing reduces AS absorption rate Ka by 84%. Subjects pooled across four Phase I studies (single-dose ascending, drug-interaction with pyronaridine, food-effect, and three-day multiple-dose) at 2-5 mg/kg oral AS.
Artesunate hamster pbpk (Kang 2023) Preclinical (golden hamster). PBPK (minimal, joint parent + metabolite; WinNonlin 8.3). Oral artesunate and its active metabolite dihydroartemisinin, each described by a five-compartment minimal PBPK model in which lung and trachea – the antiviral target tissues – are explicit perfusion-rate-limited organs and all nontarget tissues are lumped into a rest-of-body compartment (Kang et al. 2023, Pharmaceutics 15:838). Artesunate is treated as a prodrug that is completely hydrolysed to dihydroartemisinin in the blood compartment, so the parent apparent clearance CL/F is simultaneously the metabolite formation rate; the two subsystems are therefore one jointly fitted model and are kept in a single file. Plasma was the assay matrix and is recovered from blood via the fixed blood-to-plasma ratio 0.75 for both compounds. States carry molar amounts (nmol) so that the 1:1 artesunate to dihydroartemisinin conversion is mass-balanced. Physiological volumes and blood flows (Table 1) are fixed; the biochemical parameters (Table 2) were fitted to 60 artesunate and 127 dihydroartemisinin pooled measurements by naive pooled-data analysis, so the model carries no between-subject variability and is intended for typical-value simulation. Two Table 2 / equation anomalies are reproduced verbatim – see the vignette Errata.
Asc10a (Lv 2025) Population PK model of ASC10-A (also called NHC, beta-D-N4-hydroxycytidine), the active nucleoside metabolite of the oral double prodrug ASC10, in Chinese healthy adult volunteers (Lv 2025). ASC10 is under clinical development for COVID-19 and shares an active moiety with molnupiravir. Data pooled 57 subjects with 1,634 ASC10-A plasma concentrations from a Phase I trial (NCT05523141) with a multiple-ascending-dose part (50-800 mg BID x 5 days + one AM dose on day 6) and a fasted/fed food-effect part. The final model is a two-compartment disposition with first-order elimination and a two-transit-compartment absorption chain (rate constant KTR). Body weight enters as an estimated power on CL/F (exponent 0.903) and a fixed power on Vc/F (exponent 1.0), normalised to the median 61.4 kg. Food status enters KTR only: fed KTR is scaled by theta_KTR-food = 0.474, i.e., ~52.6% reduction relative to fasted, with no effect on relative bioavailability (F is fixed at 1). Residual error is combined additive (2.3 ng/mL, fixed) + proportional (SD = 0.41).
Asenapine (Dogterom 2018) Two-compartment population PK model with first-order sublingual absorption for asenapine in pediatric patients (aged 10-17 years) with schizophrenia, bipolar I disorder, or other psychiatric disorders (Dogterom 2018 Drug Design, Development and Therapy). Central / peripheral volumes and absorption-rate constant were fixed from a Phase I-only fit; no intrinsic covariate (age, BMI, race, sex) was retained in the final model. Residual error switches per observation between intensive Phase I PK sampling (27.8% CV) and sparse Phase III efficacy sampling (56.0% CV), with an additional between-subject scaling of the residual SD (19.2% CV).
ASP8232 (Snelder 2020) Integrated multiple-target-mediated drug disposition (TMDD) population PK-PD model for the small-molecule vascular adhesion protein-1 (VAP-1 / AOC3) inhibitor ASP8232 in adults (Snelder 2020). First-order oral absorption with lag time, three-compartment disposition (central + two peripheral compartments; V3 fixed equal to V2). ASP8232 binds under a quasi-steady-state assumption to soluble VAP-1 (sVAP-1) in the central compartment and to membrane-bound VAP-1 (mVAP-1) in the central and both peripheral compartments; sVAP-1 turnover and drug-target complex elimination are assumed negligible so all VAP-1 pool sizes are held constant per subject. A single dissociation constant KD applies to every binding site. Three observed analytes: (1) total ASP8232 in the central compartment (free + drug-sVAP-1 complex, matching the LC-MS assay), (2) total sVAP-1 in the central compartment (matching the ELISA assay), and (3) VAP-1 plasma activity (a power function of free sVAP-1). Pooled fit across four studies – two phase 1 (first-in-human 8232-CL-0001 and renal-impairment / T2DM-CKD 8232-CL-0002) and two phase 2 (VIDI in diabetic macular oedema, ALBUM in diabetic kidney disease). Between-population PK differences are explained by baseline eGFR (CKD-EPI) effects on clearance (sigmoid Emax, Hill coefficient fixed at 10) and relative bioavailability (power). Female subjects have 12.5% higher VAP-1 concentrations. Full omega block on CL, sVAP-1c, and the activity slope SL. Log-additive residual errors with a multiplicative factor of 1.88 on the phase-2 (VIDI + ALBUM) residuals for ASP8232 PK and VAP-1 activity.
Astegolimab (Kotani 2022) Two-compartment population PK model for astegolimab (anti-ST2 IgG2) in adults with severe asthma (Kotani 2022)
Asundexian (Yassen 2025) Two-compartment population PK model with two transit absorption compartments for asundexian, an oral selective Factor XIa inhibitor, in healthy volunteers and adult patients at risk for thromboembolic / cardiovascular events (Yassen 2025)
AT9283 (Duong 2017) Two-compartment IV population PK model for AT9283 (aurora kinase inhibitor) in adults and children with leukaemia or solid tumours (Duong 2017): allometric body-weight scaling on all four disposition parameters (CL, Vc, Q, Vp) with a power effect of estimated GFR on CL. Population residual error switches between adults (combined additive + proportional) and children (additive only) via the CHILD binary indicator.
Atazanavir (Colombo 2006) One-compartment first-order-absorption population PK model with absorption lag-time for orally administered atazanavir in HIV-1 infected adults; binary low-dose ritonavir (RTV) coadministration reduces apparent oral clearance by 46% (Colombo 2006).
Atazanavir (Dickinson 2009) One-compartment first-order-absorption population PK model with absorption lag-time for oral ritonavir-boosted atazanavir in HIV-infected adults and healthy volunteers; ritonavir AUC0-24 (median 7.52 mg*h/L) enters CL/F via a power function (Dickinson 2009).
Atazanavir (Foissac 2011) One-compartment first-order-absorption population PK model for orally administered atazanavir in 51 HIV-1-infected children and adolescents (3-18 years, 13-79 kg) on therapeutic drug monitoring. Body weight is carried through a fixed-exponent allometric scaling on CL/F (0.75) and V/F (1.0) referenced to 70 kg. Two binary co-medication indicators enter linearly on apparent oral clearance: low-dose ritonavir as a PK booster reduces CL/F (the typical CL/F = 7.1 L/h is the RTV-boosted reference, and absence of ritonavir multiplies CL/F by 1.80) and concomitant 300 mg tenofovir disoproxil fumarate increases CL/F by 25%. Between-subject variability is retained only on CL/F; residual error is proportional (Foissac 2011).
Atazanavir (Hong 2011) C0-delinked one-compartment first-order-absorption population PK model with absorption lag-time for orally administered atazanavir (ATV) in HIV-infected adults and pediatric patients (3 months to 21 years), with covariate effects of age (ka), body weight (CL/F, V/F), sex, study-site region (Africa), ritonavir comedication (CL/F and Frel), and capsule-vs-powder formulation (Frel) (Hong 2011).
Atazanavir (Rekic 2011) Population PK / PD model for atazanavir (boosted with ritonavir 100 mg QD) and its concentration-dependent effect on plasma bilirubin in adult antiretroviral-naive HIV-positive patients from the NORTHIV trial (Rekic 2011). Atazanavir disposition is described by a one-compartment model with first-order absorption and an absorption lag, fitted to log-transformed plasma atazanavir concentrations; ka and the lag time were fixed to the published values from the Colombo 2006 atazanavir popPK report (ref 27) because sparse absorption-phase sampling did not support their re-estimation, and CL/F and V/F were re-estimated under fixed allometric scaling on body weight centred at 70 kg (exponents 0.75 on CL/F and 1 on V/F, both fixed a priori). The bilirubin response is described by an indirect-response (turnover) model with concentration-dependent inhibition of the fractional turnover rate kout: dB/dt = kin - kout * (1 - Imax * Cc / (IC50 + Cc)) * B, with kin re-parameterised at steady state as kin = kout * Baseline. Inter-individual variability is supported only on V/F, CL/F (PK), and bilirubin baseline (PD); the paper notes that the data did not support IIV on the remaining PD parameters. PK residual variability is proportional; bilirubin residual variability is combined additive + proportional (the paper’s ‘slope-intercept’ model).
Atazanavir ritonavir (Molto 2016) Simultaneous one-compartment popPK model for oral atazanavir (ATV, parent / substrate) and ritonavir (RTV, sibling-drug suffix _rtv) in 83 HIV-1-infected Caucasian adults receiving either ATV 400 mg or ATV 300 mg + RTV 100 mg once daily. Both drugs use a Savic transit- compartment absorption chain (ATV: N = 7, MTT = 0.80 h, ka = 2.05 1/h; RTV: N = 11, MTT = 0.522 h, ka = 1.21 1/h) feeding a depot, followed by first-order elimination from a one-compartment central. ATV apparent clearance is exponentially inhibited by RTV plasma concentration: CL/F_ATV(t) = exp(lcl) * exp(-e_crtv_cl * C_RTV(t)) with the unboosted CL/F_ATV = 11.7 L/h and inhibition coefficient 0.296 L/mg. This functional form reproduces the paper’s reported ~18% reduction in ATV CL at the cohort-mean RTV concentration of 0.63 mg/L and explains 17.5% of inter-individual variability in ATV CL. Demographic covariates (weight allometric, gender, age, TDF, HCV, dose-timing, AAG, albumin) were screened by GAM and tested in NONMEM but not retained; an Emax-form and a linear-form inhibition were also tested and rejected (Emax: unrealistic estimates; linear: biased fit). IIV on ka / CL/F / V/F is reported for both drugs with unusually large IIV on absorption (~200% CV) confirmed in the paper Results. ATV residual error is combined (27.0% proportional + 0.07 mg/L additive); RTV residual error is proportional only (28.0%; the additive component of the initial combined error was deleted as negligible) (Molto 2016).
Atazanavir ritonavir (Schipani 2013) Simultaneous one-compartment first-order-absorption popPK model for oral atazanavir (ATV) and ritonavir (RTV) in 30 HIV-infected adults receiving ATV/RTV 300/100 mg once daily, with a direct sigmoidal-Emax inhibition of ATV apparent clearance by RTV plasma concentration (Imax = 0.988, IC50 = 0.221 mg/L). Both drugs share a one-compartment structure with first-order absorption and an absorption lag time; ka values are fixed to the separate-model final estimates (ATV ka = 1.81 1/h, RTV ka = 0.898 1/h) because joint estimation produced numerical instability. Inter-individual variability is carried on V/F for both drugs and on CL/F for RTV (correlated with V/F RTV, rho = 0.75); ATV CL/F is fitted without IIV. Demographic covariates and tenofovir co-administration were tested and none retained (Schipani 2013).
Atenolol qsp (Fu 2022) QSP (preclinical, beagle dog). Novel cardiovascular systems (CVS-CTR) model integrating heart rate (HR), left-ventricular end- diastolic volume (EDV), total peripheral resistance (TPR), and myocardial contractility (CTR) via pressure-volume-loop theory (Fu 2022). Four indirect-response (turnover) ODEs – one per hemodynamic state – are coupled by mean arterial pressure (MAP) feedback and by algebraic PV-loop relations that derive stroke volume (SV), end-systolic volume (ESV), cardiac output (CO), and MAP from the four states plus a fixed literature-derived isovolumic contraction time (TIC = 0.256 s, Templeton 1979). The paper uses atenolol (0.3-30 mg/kg PO) as proof-of-concept beta1-blocker; Emax inhibition on HR-production and CTR-production is driven by the central-compartment atenolol concentration with EC50 fixed at the beta1 KD (58.3 ng/mL, Baker 2005). A positive TPR effect (beta2 mediated) was rejected during model development (Emax_TPR -> 0) and is not included. Atenolol PK is fixed from a prior three-compartment beagle-dog popPK model (Snelder 2013 / Venkatasubramanian 2018 as cited in Fu 2022 references 16 and 18). Circadian rhythm modulates the production rates of HR / TPR / CTR (24 h period on HR and CTR; 8 h effective period on TPR); amplitudes are shared within a study but differ between the two contributing studies (Servier – Study 1, AstraZeneca – Study 2). Four outputs are observed with log-normal residual error (HR, dP/dtmax); CO and MAP are derived algebraically. EDV is included structurally (Fu 2022 Discussion: ‘baseline EDV was fixed to a reported literature value’ 31.13 mL) to enable future extension to compounds with a primary effect on EDV.
Atezolizumab carboplatin nabpaclitaxel qsp (Wang 2024) QSP. Quantitative systems pharmacology model of non-small-cell lung cancer (NSCLC) describing the triple combination of the anti-PD-L1 checkpoint inhibitor atezolizumab with carboplatin and nab-paclitaxel, extended from the published immuno-oncology QSP platform. Four physiological compartments (central, peripheral, tumour, tumour-draining lymph node) plus APC endosomal and surface compartments, two immunological-synapse compartments and a separate three-compartment Michaelis-Menten PK model for nab-paclitaxel. Covers naive and activated CD4+/CD8+ T cell trafficking and proliferation, logistic tumour growth against a dynamic vasculature-driven carrying capacity with a drug-sensitive and a chemo-resistant cancer clone, self-antigen and neo-antigen release, uptake, endosomal processing and MHC-I presentation, TCR-mediated activation, PD-1/PD-L1/PD-L2 checkpoint binding with atezolizumab blockade, CTLA-4/CD28/CD80/CD86 co-signalling, MDSC recruitment with arginase-I and nitric-oxide suppression, Th-to-Treg transdifferentiation, TGF-beta / IFN-gamma / IL-2 / CCL2 dynamics, and the cytotoxic plus anti-angiogenic effects of nab-paclitaxel and carboplatin. 140 ODE states, 55 algebraic (repeated- and initial-assignment) rules and 247 parameters, translated from the SimBiology export in Supplementary Tables S2-S7. Deterministic mechanism model: the authors generated virtual patients by Latin hypercube sampling of the 26 parameter distributions in Supplementary Table S1 rather than by fitting IIV or residual error, so no etas and no error model are encoded.
Atezolizumab cibisatamab qsp (Anbari 2023) QSP. Modular quantitative systems pharmacology platform for immuno- oncology in metastatic colorectal cancer, describing combination therapy with the anti-PD-L1 checkpoint inhibitor atezolizumab and the CEA-directed bispecific T cell engager cibisatamab (CEA-TCB, RO6958688). Four physiological compartments (central, peripheral, tumour, tumour-draining lymph node) plus APC endosomal / surface compartments and four immunological-synapse compartments. Covers naive and activated CD4+/CD8+ T cell trafficking and proliferation, logistic tumour growth with two cancer clones, antigen release, uptake, endosomal processing and MHC-I presentation for a self-antigen and a neo-antigen, TCR-mediated activation, PD-1/PD-L1/PD-L2 checkpoint binding with atezolizumab blockade, CEA-cibisatamab-CD3 trimer formation in the immunological synapse, MDSC recruitment with arginase-I and nitric oxide suppression, Th-to-Treg transdifferentiation, and TGF-beta / IFN-gamma / IL-2 / CCL2 dynamics. 90 ODE states, 38 algebraic (repeated- and initial-assignment) rules and 192 parameters, translated from the SimBiology export in Supplementary Tables S2-S6. Deterministic mechanism model: the authors generated virtual patients by Latin hypercube sampling of the parameter distributions in Supplementary Table S7 rather than by fitting IIV or residual error, so no etas and no error model are encoded.
Atorvastatin (Kakara 2014) PD-only indirect-response Imax model for LDL-cholesterol lowering by atorvastatin (Kakara 2014). One LDL-C compartment with zero-order synthesis Kin inhibited by Imax * DOSE / (ID50 + DOSE), where DOSE is the current daily atorvastatin dose (mg/day) supplied as a time-varying covariate column. An additive 0.109 contribution to the inhibition fraction is applied when ezetimibe is coadministered (CONMED_EZE = 1). The LDL-C synthesis-elimination loop is set up at steady state by enforcing Kin = Baseline * Kout (Kout derived inside model() as Kin / Baseline). Baseline LDL-C is age-scaled as 152 * (AGE/62)^(-0.240). Imax (0.567), Kin (32.8 mg/dL/day), Baseline (152 mg/dL), the age power exponent (-0.240), the ezetimibe INH contribution (0.109), and the IIV magnitudes are shared with Kakara_2014_pitavastatin and Kakara_2014_rosuvastatin (one joint NONMEM 7.2 FOCE-INTER fit across 378 patients). Atorvastatin ID50 = 2.22 mg per Kakara 2014 Table 2.
Avalglucosidase alfa (Tuffal 2023) Concatenated 3-compartment population PK model for avalglucosidase alfa (enzyme replacement therapy) in adolescent and adult patients with late-onset Pompe disease (Tuffal 2023). Pooled analysis of 2042 plasma drug-activity determinations from 75 patients across three trials (phase I/II NCT01898364, its follow-up NCT02032524, and phase III NCT02782741) at 5, 10, and 20 mg/kg IV Q2W. Elimination from the central compartment is the sum of a linear clearance (CL) and a parallel saturable Michaelis-Menten clearance (Vmax, Km). The two peripheral compartments are arranged in SERIES rather than in parallel: central exchanges bidirectionally with peripheral1 (Q2), peripheral1 feeds peripheral2 one way (Q3), and peripheral2 returns drug directly to central one way (Qpc). That cycle is what produces the two chronological kinetic sequences seen in the data - a first exposure-driving sequence contributing about 99% of AUC, then a slow low-concentration rebound peaking near 87 h and remaining above the LLOQ out to the next dose at 336 h. The authors explicitly tested and rejected a bidirectional Q3 (‘Markedly increased OFV’). Q2, V2, Q3, and V3 were fixed after preliminary screening to avoid identifiability problems. No covariates are retained: 12 demographic / laboratory covariates and 3 antidrug-antibody parameterisations were screened and none qualified, so the screened set is documented in covariatesDataExcluded rather than covariateData. Allometric scaling was tested and not retained, so the model carries no body-weight term.
AvalglucosidaseAlfa (Tiraboschi 2023) Cyclic (concatenated) three-compartment population PK model for intravenous avalglucosidase alfa in patients with late-onset and infantile-onset Pompe disease (Tiraboschi 2023), with one-way back-redistribution from the second peripheral compartment to central, parallel linear and Michaelis-Menten elimination from the central compartment, and time-varying body-weight allometric scaling on CL, Vc and Vmax.
Avelumab (Masters 2022) Two-compartment population PK model for avelumab (anti-PD-L1 IgG1) with time-dependent clearance in patients with advanced solid tumors (Masters 2022)
Avibactam qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for avibactam (diazabicyclooctane beta-lactamase inhibitor): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Axatilimab (Yang 2024) Semimechanistic population PK/PD model for axatilimab (anti-CSF-1R IgG4 monoclonal antibody) in healthy adults, patients with advanced solid tumors, and patients with chronic graft-versus-host disease (Yang 2024). Two-compartment IV PK with parallel linear clearance and CSF-1R-mediated saturable elimination via competitive Hill binding with circulating CSF-1; CSF-1, NCMC, AST, and CPK pharmacodynamics integrated as turnover indirect-response biomarkers.
Axitinib (Garrett 2014) Two-compartment population PK model for axitinib in healthy volunteers (Garrett 2014). First-order absorption with fixed lag time, allometric power-form effect of body weight on the central volume of distribution (reference 75 kg), linear-proportional fasting effects on the first-order absorption rate constant ka and on bioavailability F, and a linear-proportional reduction in F for the marketed crystal polymorph Form XLI relative to the earlier Form IV reference. Pooled data from 337 healthy subjects across ten Pfizer Phase I studies.
Axomadol (MangasSanjuan 2016) Semi-physiological population pharmacokinetic and joint pharmacodynamic model of axomadol (a racemic analgesic with opioid agonistic and monoamine-reuptake-inhibitor activity) and its O-demethyl (ODM) metabolite in healthy adult volunteers. The PK structure carries two parallel enantiomer chains (RR-suffix r and SS-suffix s), each consisting of a first-order absorption depot, a liver compartment mimicking first-pass conversion, a parent central compartment, and a metabolite central compartment. Within each enantiomer the parent and metabolite share a single apparent volume of distribution (VP = VM) and share a single typical first-order elimination rate constant (kP0 = kM0), although between-subject variability is estimated separately for the two elimination pathways. The PD layer is shared across the enantiomer chains and is driven by the SS parent in plasma (mydriatic Emax) and by the RR metabolite at a hysteresis effect site (linearly miotic). Pupil diameter is the sum of those two opposing effects; cold-pressor analgesic AUC is a linear function of the parent and metabolite contributions to pupil diameter. Parameter values are from Mangas-Sanjuan et al. 2016 Tables 2, 4, and 5.
AZD6088 rat (Viberg 2012) Preclinical (rat). Direct-effect population PK/PD model for the muscarinic agonist AZD6088 (MW 406.57) in male Sprague-Dawley rats using the spinal-nerve-ligation (SNL) heat-hyperalgesia model of neuropathic pain. One-compartment oral PK (depot -> central) with the first-order absorption rate constant ka fixed at 5 x kel (equivalent to 5 x cl/vc; Viberg 2012 Results PK-model paragraph: ‘Lack of plasma concentration data in the absorption phase made it impossible to estimate rate of absorption and ka was therefore fixed to 5 time clearance’); the 40 umol/kg highest-dose cohort (DOSE_HIGH = 1) gates a distinct typical-value apparent oral clearance of 3.92 L/h/kg vs 10.87 L/h/kg for all other cohorts, per the paper’s OFV-supported categorical CL split (delta OFV = -32). PD is a direct-effect Emax on paw withdrawal latency (baseline 5.947 s, Emax 10.09 s, EC50 0.0433 umol/L; no time delay, no sigmoidicity, no tolerance-development term - all three alternatives increased OFV or degraded goodness-of-fit). Typical-value only: the paper reports ‘exponential models were used to describe inter-individual variability’ on CL and V but never publishes the omega^2 / CV% magnitudes; see the vignette Assumptions and deviations section.
Azithromycin (Merchan 2015) Population PK model for intravenous azithromycin in preterm neonates at risk for Ureaplasma respiratory tract colonization (Merchan 2015). Pooled re-analysis of three studies (single 10 mg/kg, single 20 mg/kg, and 3 daily doses of 20 mg/kg). Two-compartment linear model with all PK parameters allometrically scaled on body weight: fixed exponent 0.75 on CL and Q, fixed exponent 1.0 on V1 and V2, reference body weight 1 kg.
Azithromycin (Sampson 2014) Four-compartment mamillary population PK model for oral azithromycin simultaneously describing concentrations in whole blood, peripheral blood mononuclear cells (PBMCs), and polymorphonuclear cells (PMNs) in healthy adults (Sampson 2014). First-order absorption with lag; unidirectional flow from central to PBMC and to PMN compartments; bidirectional flow between central and a peripheral tissue compartment; elimination from central, PBMC, and PMN compartments. The observed whole-blood concentration is a weighted sum of plasma, PBMC, and PMN concentrations.
Azithromycin (Zheng 2014) Semi-mechanistic tissue distribution population PK model for oral azithromycin in healthy adults (Zheng 2014). Three-compartment plasma PK (depot with absorption lag time and first-order absorption, central, two peripheral compartments) with concentration-dependent fraction unbound in plasma (equation 1). Three tissue distribution compartments (muscle interstitial space fluid, subcutaneous adipose tissue interstitial space fluid, polymorphonuclear-leukocyte (PML) cytosol) each driven by free unbound (or, for PML cytosol, free unionized) plasma drug via first-order rate constants kin and kout, with tissue-specific distribution factors df_muscle, df_adipose, df_pmn that scale the steady-state tissue:plasma free-unbound ratio. Each tissue compartment also exchanges with a deep nonspecific phospholipid-binding compartment via shared kon and koff (Methods equations 1-13).
Azithromycin (Zheng 2018) Pediatric population PK model for intravenous azithromycin in children with community-acquired pneumonia (Zheng 2018). Two-compartment model with linear elimination, allometric scaling on clearance and intercompartmental clearance (exponent 0.75 fixed) and on central and peripheral volumes (exponent 1.0 fixed) with reference body weight 21.5 kg, and a binary alanine aminotransferase covariate that reduces CL by 24 percent when ALT > 40 IU/L.
Aztreonam qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for aztreonam (monobactam): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Bapineuzumab (Hu 2014) Two-compartment population PK model for bapineuzumab in adults with mild-to-moderate Alzheimer’s disease following IV administration (Hu 2014, reduced model)
Baricitinib (Decker 2024) Two-compartment population PK model with zero-order absorption, an absorption lag, and fixed allometric scaling for oral baricitinib in 217 pediatric patients aged 2 to <18 years with polyarticular-course juvenile idiopathic arthritis (JUVE-BASIS, NCT03773978). Apparent total clearance is partitioned semi-mechanistically into an eGFR-dependent apparent renal arm (CLr/F) and an eGFR-independent apparent non-renal arm (CLnr/F). Baseline body weight enters through allometric exponents fixed at 0.75 on all clearance terms (CLr/F, CLnr/F, Q) and at 1 on both volumes (V1/F, V2/F), referenced to 74 kg. The structure was carried unchanged from the adult rheumatoid-arthritis baricitinib population PK model; no further covariate met the stepwise-covariate-modeling inclusion criteria, so the base model is the final model.
BAY81 8973 (Garmann 2017) Two-compartment population PK model for BAY 81-8973 (Kovaltry, full-length unmodified recombinant human factor VIII) in patients with severe haemophilia A aged 1-61 years pooled from the LEOPOLD I, II and Kids trials (Garmann 2017). Final model uses NONMEM M3 likelihood for samples below the chromogenic-assay limit of quantitation (1.5 IU/dL).
BAY94 9027 (Solms 2020) One-compartment population PK model for BAY 94-9027 (damoctocog alfa pegol, Jivi, an extended-half-life site-specifically PEGylated B-domain-deleted recombinant factor VIII) in 198 male patients aged 2-62 years with severe haemophilia A pooled from the BAY 94-9027 phase I (NCT01184820), PROTECT VIII (NCT01580293), and PROTECT VIII Kids (NCT01775618) trials (Solms 2020). Final chromogenic-assay model has lean body weight (LBW) as a power-form covariate on CL and Vc and von Willebrand factor antigen (VWF) as a power-form covariate on CL; between-subject variability is a BLOCK(2) on CL and Vc with correlation 0.449; residual error is combined additive plus proportional. NONMEM M3 likelihood was used for samples below the chromogenic-assay lower limit of quantitation (1.5-3 IU/dL).
Bedaquiline (Svensson 2013) Three-compartment population PK model for bedaquiline (BDQ) with a two-compartment N-desmethyl metabolite M2 and a two-compartment N,N-bis-desmethyl metabolite M3 in healthy adult volunteers following single 400 mg oral doses, with Savic 2007 analytical transit-compartment absorption (non-integer NN feeding a first-order depot at rate ka) and an instantaneous-switch concomitant-efavirenz induction factor of 2.07 on apparent CL_BDQ and CL_M2 and 1.12 on apparent CL_M3, applied from 1 week after the start of 600 mg once-nightly efavirenz co-administration.
Bedaquiline (Svensson 2014) Three-compartment population PK model for bedaquiline (BDQ) and a two-compartment N-desmethyl metabolite M2 in healthy adult volunteers following single 400 mg oral doses, with Savic 2007 analytical transit-compartment absorption (non-integer NN feeding a first-order depot at rate ka), fixed allometric scaling on disposition (0.75 on CL/Q at 70 kg, 1 on Vc/Vp), and multiplicative rifampicin or rifapentine drug-drug-interaction factors of 4.78 and 3.96 on bedaquiline and M2 apparent clearance, applied at full induction from day 3 of rifamycin co-administration.
Bedaquiline (Svensson 2017) Pharmacodynamic exposure-response model for the mycobacterial load (MBL, n bacteria per sample inoculum) in adult patients with drug-resistant pulmonary tuberculosis treated with bedaquiline plus an optimized background regimen. The latent MBL state declines mono-exponentially with a half-life HL that is prolonged by 28.1% in patients with pre-XDR or XDR tuberculosis and shortened by individual bedaquiline weekly-average plasma concentration CAV via an Emax model with the maximum fractional effect on HL fixed at -100% (EC50 1.42 mg/L). The per-subject starting MBL_0 is informed by the baseline mean Time-to-Positivity in MGIT liquid culture (TTP_MGIT_BASE) via a power-form covariate (exponent -3.69 around the cohort median 6.8 days). Inter-individual variability on log HL uses a Box-Cox-transformed eta distribution (Petersson 2009 form, shape 0.66, variance 0.33); inter-occasion variability in sputum sampling on log MBL (variance 3.71) is folded into the residual log-scale error. The Svensson 2017 source’s full 3-component model (longitudinal MBL plus per-sample probability of bacterial presence plus MGIT-tube logistic-growth-driven time-to-event for observed TTP) is reduced here to the MBL component, with the latent MBL state treated directly as the observable; the probability-of-presence and tube-growth-driven TTP machinery are measurement-model artifacts of how MBL was inferred from TTP data and are dropped (see vignette Assumptions and deviations). Bedaquiline CAV is supplied as a time-varying covariate column from any popPK source; the upstream popPK paper (Svensson 2016 CPT PSP, reference 21) is shipped in nlmixr2lib as modellib(‘Svensson_2016_bedaquiline’).
Bedaquiline (Svensson 2018) Three-compartment population PK model for the antimycobacterial bedaquiline (BDQ) and a two-compartment N-desmethyl metabolite M2 in healthy adult volunteers following single 400 mg oral doses, with four-transit-compartment first-order absorption (rate of absorption from the last transit compartment fixed equal to the inter-transit transfer rate, i.e. KA = KTR) and a multiplicative formulation effect adding 23% to the typical mean absorption time when the four 100 mg tablets are suspended in water before swallowing relative to swallowing the tablets whole.
Bedaquiline lpvr (Svensson 2014) Three-compartment population PK model for bedaquiline (BDQ) and a two-compartment N-desmethyl metabolite M2 in healthy adult volunteers following single 400 mg oral doses, with Savic 2007 analytical transit-compartment absorption (non-integer NN feeding a first-order depot at rate ka), fixed allometric scaling on disposition (0.75 on CL/Q at 70 kg, 1 on Vc/Vp), and multiplicative ritonavir-boosted lopinavir (LPV/r) drug-drug-interaction factors of 0.347 on bedaquiline apparent clearance and 0.578 on M2 apparent clearance during LPV/r co-administration (study C110).
Bedaquiline mpbpk (Mehta 2023) mPBPK (minimal physiologically based, translational mouse-to-human). Bedaquiline and its N-desmethyl metabolite M2 in pulmonary tuberculosis patients, with explicit cavitary lung-lesion and uninvolved-lung site-of-action compartments. Parallel first-order absorption from a fast and a slow depot, a well-stirred liver compartment that converts bedaquiline to M2 via a hepatic extraction ratio, a lumped peripheral tissue pool for parent and for M2, and perfusion-limited lesion / uninvolved-lung compartments driven by penetration ratios. Physiological volumes and flows are computed from body weight (Brown 1997); clearance terms are allometrically scaled from the mouse fit with an exponent of 0.75. The lesion and uninvolved-lung states hold concentrations (mg/L), not amounts. Intended for typical-value and population simulation of site-of-action target attainment; the paper did not fit this model to individual human data.
Bedaquiline nvp (Svensson 2014) Three-compartment population PK model for bedaquiline (BDQ) and a two-compartment N-desmethyl metabolite M2 in HIV-1-infected ART-naive adult volunteers following single 400 mg oral doses, with Savic 2007 analytical transit-compartment absorption (non-integer NN feeding a first-order depot at rate ka), fixed allometric scaling on disposition (0.75 on CL/Q at 70 kg, 1 on Vc/Vp), and multiplicative nevirapine (NVP) drug-drug-interaction factors of 0.915 on bedaquiline and 1.05 on M2 apparent clearances during steady-state NVP co-administration (study C117). The factors are fixed-effects only because BSV on the NVP interaction effects was not estimated.
Belantamab (Papathanasiou 2025) Two-compartment population PK model for the antibody-drug conjugate (ADC) belantamab mafodotin in patients with relapsed/refractory multiple myeloma, with sigmoidal time-varying clearance and covariate effects of baseline body weight, BMI, albumin, soluble BCMA, serum IgG, race, and combination therapy (Papathanasiou 2025; ADC moiety only – the cys-mcMMAF payload sub-model is not included; see vignette for rationale)
Belantamab mprotein (Collins 2023) Concentration-driven tumor growth inhibition (TGI) model for serum M-protein in patients with relapsed/refractory multiple myeloma treated with the antibody-drug conjugate belantamab mafodotin, with logistic growth plus a modified weak Allee term, a saturating effect-compartment driven kill term, an exponential resistance decay, and covariate effects of baseline beta-2-microglobulin on the growth rate, baseline M-protein on the kill rate, and extramedullary disease plus baseline soluble BCMA on the effect-compartment rate constant (Collins 2023). The embedded two-compartment time-varying-clearance PK layer is a typical-value ADC driver taken from Collins 2023 Table S1; for the full covariate population PK model of belantamab mafodotin see modellib(‘Papathanasiou_2025_belantamab’). The paper’s ocular-safety discrete time Markov model is not included (see vignette).
Belatacept (Shen 2013) PK/PD model for belatacept (CTLA-4/IgG1 fusion protein, selective T-cell co-stimulation blocker) in adult kidney transplant recipients (Shen 2013). The PK side is a one-compartment IV-infusion model derived from the paper’s noncompartmental analysis (Table 1, 10 mg/kg substudy, n = 10): typical clearance and volume for a 70 kg adult are set so the model reproduces the reported geometric-mean CL, Vss, AUC over a 4-week dosing interval, and ~8-9 day terminal half-life. The PD side is the inhibitory Emax model of Eq. 2 (Section 3.2, n = 62 in the phase II corticosteroid-avoidance substudy IM103034) describing free CD86 receptor expression on peripheral-blood monocytes (MESF) as E0 - Emax * Cc / (EC50 + Cc); CD86 receptor occupancy is derived as 100 * (E0 - freeCD86) / E0. Belatacept exhibited linear PK across 5-10 mg/kg with relatively low between-subject variability; the full population PK with body-weight covariates was published separately by Zhou et al. (2012) and is not refit here.
Belimumab (Marcantonio 2022) QSP. One-compartment monospecific anti-ligand mechanistic PKPD model of belimumab-BAFF binding in adults with systemic lupus erythematosus (SLE) (Marcantonio 2022 Early Feasibility Assessment, Case Study 3). Belimumab binds soluble BAFF (BLyS); model treats BAFF as the ligand and BAFF-R as the cognate receptor (BAFF also binds BCMA and TACI in vivo but this model uses only the BAFF-R affinity 15 nM per Day 2005 / Cachero 2006 / Hymowitz 2005 as reported in Table S5). Belimumab valency = 1 (BAFF trimer binds 3:3 per Shin 2018). Parameters FIXED from paper Table S5.
Belimumab (Struemper 2017) Linear two-compartment subcutaneous population PK model for belimumab in healthy volunteers and adult patients with systemic lupus erythematosus, with first-order absorption + lag time, allometric body-weight scaling on CL/Vc/Q/Vp, and baseline BMI on Vc and baseline albumin and IgG on CL (Struemper 2017)
Belimumab (Zhou 2021) Linear two-compartment IV population PK model for belimumab in Chinese and non-Chinese adult and pediatric patients with systemic lupus erythematosus (Zhou 2021)
Belzutifan (Marathe 2023) Two-compartment population PK model for oral belzutifan (Welireg, a hypoxia-inducible factor 2 alpha inhibitor) in 239 pooled subjects across four phase I studies and one pivotal phase II study: 83 healthy participants, 74 patients with advanced renal cell carcinoma, 21 with other advanced solid tumors, and 61 with von Hippel-Lindau disease-associated RCC. First-order absorption with an absorption lag and linear elimination from the central compartment. Body weight scales clearances (shared exponent on CL/F and Q/F) and volumes (shared exponent on V2/F and V3/F); age scales CL/F and V2/F; fed state and the final market formulation each slow the absorption rate constant; and the polymorphic UGT2B17 and CYP2C19 metabolizer phenotypes shift CL/F, with UGT2B17 poor metabolizers additionally showing 11% higher relative bioavailability. Residual variability is split into healthy-participant and patient strata.
BempedoicAcid (Jadhav 2023) Population PK model for bempedoic acid (ATP-citrate lyase inhibitor) in healthy subjects and patients with dyslipidemia, renal or hepatic impairment, or type 2 diabetes mellitus (Jadhav 2023): two-compartment disposition with a single transit absorption compartment and linear elimination, pooled across 22 phase 1/2/3 studies. Covariate effects on apparent clearance (sex, body weight, Black race, hyperlipidemia, type 2 diabetes, eGFR, ezetimibe), on apparent central volume (sex, age, body weight, simvastatin), on the absorption rate constant (food), and on relative oral bioavailability (atorvastatin).
BempedoicAcid ldlc (Jadhav 2023) Joint population PK + PK/PD model for bempedoic acid and serum low-density lipoprotein cholesterol (LDL-C) in patients with dyslipidemia (Jadhav 2023). The PK layer is the two-compartment / single-transit-absorption popPK model from the companion Jadhav_2023_bempedoicAcid.R file (Table 2). The PD layer (Table 3) is a type 1 indirect-response model in which bempedoic acid inhibits LDL-C production, with covariate effects on the maximum fractional inhibition Imax (sex, body weight, Black race, concomitant statin intensity, concomitant ezetimibe, prior statin therapy) and on baseline LDL-C (concomitant statin intensity, HeFH, type 2 diabetes, prior ezetimibe therapy, prior statin therapy). The authors fit the PD layer sequentially, conditioned on individual post hoc PK parameters from the popPK model.
Benazepril pbpk (Luo 2024) PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). Joint benazepril + benazeprilat disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Benazepril is an inactive ester prodrug hydrolysed by hepatic CES1 to the active diacid benazeprilat, which is eliminated renally. Both analytes are highly bound to albumin (free fractions 0.03 and 0.05), which Section 4 cites as a reason the cirrhosis AUCR was poorly predicted. The semi-PBPK circuit is stomach, three small-intestinal lumen segments (duodenum / jejunum / ileum), the matching three gut-wall segments, portal vein, liver, kidney and a two-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper’s virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded.
Benralizumab (Wang 2017) Two compartment PK model of benralizumab (anti-IL-5Ralpha) in healthy volunteers and patients with asthma (Wang 2017)
Benzathine benzylpenicillin g (Kado 2020) One-compartment population PK model for penicillin released from benzathine benzylpenicillin G (Bicillin L-A) with three parallel absorption pathways (slow and fast via a transit compartment, plus an immediate pathway that bypasses the transit compartment) and route-specific structural parameters for intramuscular (IM) and subcutaneous (SC) administration, developed from 311 dried-blood-spot penicillin concentrations in a randomized crossover of 15 healthy adult male volunteers each receiving 1.2 MIU IM and 1.2 MIU SC into the dorsogluteal region (Kado 2020).
Benznidazole (Soy 2015) One-compartment population PK model with first-order absorption and first-order elimination for oral benznidazole in adult patients with chronic Chagas disease (Soy 2015; CINEBENZ trial, n = 39 index plus n = 10 external validation). Apparent clearance CL/F = 1.73 L/h, apparent volume of distribution V/F = 89.6 L, and absorption rate constant Ka = 1.15 1/h fixed from the literature (Raaflaub & Ziegler 1979). Inter-individual variability is on CL/F (33.4% CV) and V/F (68.8% CV); inter-occasion variability is on CL/F (29.5% CV), folded into the CL/F eta as BSV-equivalent for forward simulation. Residual error is combined additive (0.57 mg/L) plus proportional (19.53% CV). No demographic or biological covariates were retained in the final model.
Benzylpenicillin (Nielsen 2011) In vitro (Streptococcus pyogenes M12 NCTC P1800). Semimechanistic PKPD model of benzylpenicillin time-kill kinetics; two-stage bacterial life-cycle (proliferating drug-sensitive S and non-growing drug-insensitive R) with sigmoidal Emax killing of S via an effect compartment; first-order drug elimination (ke set per in vitro kinetic-system flow rate) plus drug-specific degradation kdeg. Parameter values are from the combined static and dynamic estimation in Table 3.
Benzylpenicillin (Shah 2023) Two-compartment intravenous population PK model for benzylpenicillin in critically unwell adults, with a priori allometric body-weight scaling on all disposition parameters and a serum-creatinine power effect on clearance.
Benzylpenicillin horse (Lallemand 2023) Preclinical (horse). Three-compartment population pharmacokinetic model for benzylpenicillin (BP) in horses, developed as an international meta-analysis to derive PK/PD cutoff values supporting equine clinical breakpoints (VetCAST approach). Seven parallel absorption depots reproduce the five modalities the authors modelled simultaneously: (1) IV sodium or potassium BP dosed directly into the central compartment; (2) IM sodium BP with two SEQUENTIAL first-order rate constants, the rapid Ka being replaced by the slower Ka2 once time-after-dose exceeds a lag time; (3) IM procaine BP from four single-ingredient products (France Depocilline, Sweden Penovet, USA1 Norocillin, Japan Procaine BP G sol ‘KS’), sharing one Ka and one logit-scale bioavailability that are each modified by a ‘source of dataset’ covariate; (4) the Duplocilline fixed combination, whose total BP dose is split by SPC composition into a procaine BP fraction and a benzathine BP fraction absorbed in PARALLEL, each with its own Ka and bioavailability; and (5) IM penethamate hydriodide given at three separate injection sites, each site carrying its own sequential rapid-then-slow Ka pair and its own switch delay. All seven depots and the IV route share one set of disposition parameters (Vc, Vp, Vp2, CL, Q, Q2), which were FROZEN in the full model at the values estimated from the 23 IV profiles alone; plasma clearance carries a ‘source of dataset’ covariate with France as reference. The PK/PD layer reproduces the paper’s Monte Carlo machinery: a fixed unbound fraction of 0.4 converts total to free BP, and the free AUC and the cumulative time above MIC are integrated as states so that fAUC/MIC and fT>MIC can be read directly. Pooled meta-analysis of 40 horses providing 63 rich profiles and 1022 BP plasma concentrations from France, Sweden, Japan and the USA (Lallemand 2023).
Betamethasone (Schoenmakers 2025) Two-compartment population PK model with first-order absorption (no lag time) for intramuscular betamethasone in pregnant women admitted with imminent preterm birth, including early-onset pre-eclampsia (eoPE; diagnosed before 34 weeks gestation). Apparent clearance is multiplied by 0.617 (a 38% reduction, or ~60% of the non-eoPE clearance) when eoPE is present; this is the only retained covariate after backward elimination at P < 0.01. Body weight, BMI, lean body weight, age, gestational age, number of foetuses, white blood cell counts and CRP were screened but did not retain after backward elimination.
Bevacizumab (Han 2015) Two-compartment population PK model for IV bevacizumab in paediatric cancer patients (0.5-21 years) with fixed allometric body-weight scaling on all four disposition parameters, plus additional covariate effects of sex, baseline serum albumin, and primary-CNS-tumour-vs-sarcoma indication on CL and V1 (Han 2015, Table 2 final model)
Bevacizumab (Han 2016) Two-compartment population PK model for IV bevacizumab in adult cancer patients (Han 2016) with allometric body-weight scaling and covariate effects of sex, baseline albumin, baseline alkaline phosphatase, and concomitant interferon alpha on clearance.
Bevacizumab (Panoilia 2015) Quasi-steady-state target-mediated drug-disposition (TMDD QSS) model for IV bevacizumab and free VEGF165 in adults with stage IV colorectal cancer, with fixed allometric body-weight scaling on PK clearances and volumes (Panoilia 2015, Table 3 TMDD model column)
Bevacizumab pk (Papachristos 2020) Two-compartment population PK model for IV bevacizumab in adults with metastatic colorectal cancer, with allometric weight scaling and ICAM-1 / VEGF-A genotype covariates (Papachristos 2020, Table 1)
Bevacizumab pkpd (Papachristos 2020) Two-compartment population PK plus immediate-response Imax PK/PD model for IV bevacizumab and free VEGF-A in adults with metastatic colorectal cancer, with allometric weight scaling and ICAM-1 / VEGF-A genotype covariates (Papachristos 2020, Table 3)
Bevacizumab qss (Papachristos 2020) Quasi-steady-state target-mediated drug-disposition (TMDD QSS) model for IV bevacizumab and free VEGF-A in adults with metastatic colorectal cancer, with allometric weight scaling and ICAM-1 / VEGF-A genotype covariates (Papachristos 2020, Table 2)
BIIB023 (Galluppi 2016) Two-compartment population PK model for intravenous BIIB023, an anti-TWEAK monoclonal antibody, in healthy Chinese, Japanese and Caucasian volunteers and adults with rheumatoid arthritis (Galluppi 2016); parallel first-order linear and Michaelis-Menten elimination from the central compartment with body weight on CL and V, and sex on V.
Biib107 (Toukam 2025) Two-compartment population PK model with parallel linear and Michaelis-Menten elimination, plus direct sigmoidal Emax PK/PD model of alpha-4 integrin receptor saturation, for BIIB107 (humanized aglycosyl anti-alpha-4 integrin IgG4 monoclonal antibody) in healthy adult volunteers (Toukam 2025).
Bivalirudin (Zhang 2012) Population PK and PK-PD model for bivalirudin, a synthetic bivalent direct thrombin inhibitor, in young healthy Chinese volunteers (Zhang 2012). PK: two-compartment intravenous disposition with body-weight-normalised structural parameters (CL = 0.323 L/h/kg, V1 = 0.086 L/kg, Q = 0.0957 L/h/kg, V2 = 0.0554 L/kg); no covariates retained after a 30-covariate screen; log-normal IIV on CL, V1, and V2 with IIV on Q fixed to zero. PD: direct-response sigmoid Emax (Hill coefficient fixed at 1) on activated clotting time (ACT) using the central-compartment concentration as the effect site (E0 = 134 s, Emax = 318 s, EC50 = 2.44 mg/L); one covariate retained – red blood cell count (RBC, 10^12 cells/L) on EC50 via the linear-deviation form EC50_i = theta_EC50 * exp(eta_EC50) * (1 + 1.70 * (RBC - 4.40)) centred at the cohort median 4.40.
Bosentan pbpk (Aoki 2024) PBPK-TMDD (semi-mechanistic, dispersion liver). Bosentan disposition after single intravenous doses of 10-750 mg in healthy adults, re-estimated by Aoki & Sugiyama (2024) with the Cluster Gauss-Newton method (CGNM) on the model structure of Koyama et al. (2021). The liver is resolved as a five-compartment tandem dispersion model: five hepatic extracellular (sinusoidal) sub-compartments in series with the hepatic blood flow, each exchanging with its own hepatocyte sub-compartment. Sinusoid-to-hepatocyte transport is the sum of saturable OATP-mediated uptake (Michaelis-Menten in unbound sinusoidal concentration) and passive diffusion; efflux back to the sinusoid is passive diffusion scaled by the influx/efflux ratio gamma_dif. Drug is eliminated by hepatic metabolism from the hepatocytes and by renal clearance from plasma. Muscle, skin and adipose are perfusion-limited well-stirred tissues. Superimposed on the PBPK backbone is a target-mediated drug disposition (TMDD) layer in plasma: unbound drug binds a finite endothelin-receptor pool of total amount rtot with dissociation constant kd and off-rate koff, which is the source of the low-dose nonlinearity and which permits receptor occupancy to be simulated. The eight parameters below without fixed() are the CGNM estimates of Aoki 2024 Table S2 (all eight identifiable when the 10 mg arm is included); every other value is a fixed physiological or compound constant taken from the published model code. CGNM is a fixed-effects nonlinear-least-squares method, so this model has no between-subject variability and the paper reports no residual-error model – the propSd term is a placeholder. See the vignette Errata.
Brazikumab (Zhang 2023) Two-compartment population PK model with first-order lagged SC absorption and linear clearance for brazikumab (anti-IL-23 p19 mAb) in healthy adults and adults with Crohn’s disease (Zhang 2023)
Brazikumab crp (Zhang 2023) Indirect-response PK/PD model of Crohn’s Disease Activity Index (CDAI) for brazikumab with a constant inhibitory placebo effect and a baseline-C-reactive-protein-dependent sigmoid drug effect on the CDAI input rate, in adults with moderately to severely active Crohn’s disease (Zhang 2023)
Brazikumab il22 (Zhang 2023) Indirect-response PK/PD model of Crohn’s Disease Activity Index (CDAI) for brazikumab with a constant inhibitory placebo effect and a baseline-serum-IL-22-dependent sigmoid drug effect on the CDAI input rate, in adults with moderately to severely active Crohn’s disease (Zhang 2023)
Brentuximab (Li 2017) Semimechanistic coupled population PK model for brentuximab vedotin antibody-drug conjugate (ADC) and its released small-molecule payload monomethyl auristatin E (MMAE) in adults with CD30-expressing hematologic malignancies (Li 2017). ADC is described by a linear 3-compartment model with first-order elimination; MMAE by a linear 2-compartment model with first-order elimination. MMAE formation is driven by (1) proteolytic degradation of the ADC (scaled by a time-decaying drug-antibody ratio DAR(t) and a cycle-dependent fraction Fmc = Cycle^Fm) and (2) a first-order deconjugation flux proportional to the per-ADC MMAE payload above the minimum-detectable DAR. Modeled in molar units (amount nmol, volume L, concentration pmol/mL = nmol/L = nM) following the paper’s convention.
Brentuximab (Suri 2018) Coupled population PK model for brentuximab vedotin antibody-drug conjugate (ADC) and its released payload monomethyl auristatin E (MMAE) in 380 patients with CD30-positive malignancies (Hodgkin lymphoma, systemic anaplastic large-cell lymphoma, mycosis fungoides, primary cutaneous ALCL) pooled from six clinical studies including the phase III ALCANZA study (Suri 2018). ADC is described by a linear 3-compartment model with zero-order input and first-order elimination; MMAE by a 2-compartment model with first-order elimination, fed from ADC by (a) a saturable target-binding flux KdTargetADC (initial Target = 1, irreversibly depleted) and (b) a proteolytic flux FMexp(-ALFMtad)K10ADC whose conversion fraction declines as a function of time after the most recent dose. Both fluxes accumulate in an intermediate Lag compartment that empties to MMAE central with rate Klag (FM is fixed to 1).
Brentuximab (Zhou 2025) Coupled population PK model for brentuximab vedotin antibody-drug conjugate (ADC) and its released payload monomethyl auristatin E (MMAE) in pediatric patients (5-18 years) with relapsed/refractory or newly diagnosed Hodgkin lymphoma or systemic anaplastic large-cell lymphoma (Zhou 2025). ADC is described by a linear 3-compartment model with first-order elimination; MMAE by a 2-compartment model with first-order elimination. ADC -> MMAE flux is the sum of (a) a one-time saturable target-binding flux KdTargetADC (initial Target = 1 unitless, irreversibly depleted) and (b) a proteolytic flux FMexp(-ALFMtad)K10ADC where the conversion fraction declines as a function of time after the most recent dose. Both fluxes accumulate in an intermediate Lag compartment that empties to MMAE central with rate Klag. Final-model parameter values come from Zhou 2025 supplementary Tables S1 (ADC) and S2 (MMAE); equations come from the NONMEM control streams in Zhou 2025 Supplementary Methods.
Brepocitinib (Maleki 2024) Two-compartment population pharmacokinetic model with first-order absorption for oral brepocitinib (PF-06700841, a dual TYK2/JAK1 inhibitor) in healthy participants and patients with alopecia areata, hidradenitis suppurativa, psoriasis, psoriatic arthritis, ulcerative colitis, or vitiligo. Apparent (oral) disposition parameters carry fixed allometric body-weight scaling; clearance additionally depends on baseline aspartate aminotransferase and Asian race, apparent central volume on sex, and the absorption rate constant on ulcerative-colitis status. An absorption lag time applies only to the tablet formulation, and relative bioavailability is 30 percent higher above 100 mg/day. Interindividual variability on CL/F and Vc/F is Box-Cox transformed (Petersson 2009 form) and its magnitude, together with the combined additive-plus-proportional residual error, differs between healthy participants and patients.
Brigatinib (Hanley 2024) Two-compartment oral population pharmacokinetic reduction of the Simcyp minimal-PBPK-with-single-adjusting-compartment (SAC) model for the ALK inhibitor brigatinib in healthy adults (Hanley 2024). The source model was built in the Simcyp Population-based Simulator (versions 15 and 17) and its whole-body mass-balance equations are not published, so the platform model itself cannot be encoded here. What IS fully reported is the brigatinib compound layer, and it is sufficient to reconstruct the disposition as an ordinary compartmental model: first-order absorption into a depot, distribution between a systemic compartment and the SAC (the paper’s k_in / k_out, encoded as the canonical k12 / k21), and first-order elimination from the systemic compartment. Systemic clearance is obtained from the reported unbound hepatic intrinsic clearance through the well-stirred liver model plus the reported renal clearance; bioavailability is the reported f_a times f_G times the f_H implied by that same well-stirred calculation. No parameter is fitted here and none is imported from a Simcyp population file: every value is either a Table 1 / Appendix S1 input or an arithmetic consequence of one. The reduction reproduces the paper’s own predicted Cmax, Tmax and AUC after single 90 mg and 180 mg oral doses to within 6.4%, and Cmax at both doses to within 1% (see the validation vignette). This is a typical-value simulation model: the source reports no inter-individual variance components and no residual-error model, so there are no etas and propSd is fixed at zero. The drug-drug-interaction predictions that are the paper’s main contribution (itraconazole, rifampin, diltiazem, verapamil, efavirenz, and the transporter substrates) depend on proprietary Simcyp perpetrator compound files and are NOT reproducible from this model; see the vignette for the full list of deviations.
Brivaracetam (Schoemaker 2017) One-compartment population PK model for oral brivaracetam in paediatric patients with epilepsy aged 1 month to 16 years (Schoemaker 2017). First-order absorption, single-compartment distribution, and first-order elimination, with allometric scaling of CL/F (exponent 0.750 fixed) and V/F (exponent 1.00 fixed) on lean body weight normalised to a 50 kg adult typical value. Co-administration of phenobarbital (PB; pooled with primidone), carbamazepine (CBZ), or valproate (VPA) modify apparent oral clearance via linear-additive multiplicative factors.
Brodalumab (Timmermann 2019) Two-compartment population PK model for brodalumab in adults with moderate-to-severe plaque psoriasis (Timmermann 2019), with first-order SC absorption, fixed bioavailability, and combined linear plus Michaelis-Menten (target-mediated) elimination from the central compartment.
Busulfan (Choe 2012) One-compartment IV PK model for intravenous busulfan in adult Korean hematopoietic stem cell transplant recipients, with allometric scaling on actual body weight (fixed exponent 0.5) on CL and Vd and a sex effect on Vd (Choe 2012).
Busulfan (Lawson 2022) Two-compartment IV PK model for once-daily busulfan in pediatric hematopoietic stem cell transplant recipients with allometric normal-fat-mass (NFM) scaling, postmenstrual-age maturation on CL, and a time-associated within-treatment-course CL decline (Lawson 2022).
Busulfan (Long-Boyle 2015) One-compartment IV PK model with Michaelis-Menten elimination for busulfan in pediatric and young adult patients (0.1-24 yrs) undergoing hematopoietic cell transplant. Allometric body-weight scaling on intrinsic clearance (CLin, exponent fixed 0.75) and central volume (Vc, exponent fixed 1) with reference weight 22 kg; hockey-stick age effect on CLin (linear increase below the 12-yr breakpoint applied to AGE directly, multiplicative linear decrease above). Correlated IIV on CLin and Vc; combined proportional + additive residual error (Long-Boyle 2015).
Busulfan (Schreib 2024) One-compartment population pharmacokinetic model with intravenous infusion and a time-varying elimination rate constant for busulfan in 124 pediatric patients undergoing hematopoietic stem cell transplantation (HSCT) at the University Children’s Hospital Zurich, October 2010 to February 2020, receiving twice-daily (q12h) 2-hour to 4-hour intravenous busulfan infusions as conditioning. Schreib 2024 fit ln(V) and ln(k) directly rather than CL and V, because ‘building the model on V and CL would result in two parameters with similar covariates … potentially masking the covariates for k’. Two further parameters describe an exponential decline of the elimination rate constant over the course of therapy: kel(t) = kel(0) * (1 + kel_exp_famp * (1 - exp(-kel_exp_kdes * t))), with kel_exp_famp = -0.167 (a 16.7% average fall in k and CL at steady state) and a 13.4 h half-life for the change. The fractional amplitude carries its own inter-individual variability and is doubled to -0.312 (a 31% fall) in the hemophagocytic lymphohistiocytosis / X-linked lymphoproliferative disease group. Volume scales near-proportionally with calculated total body water; the elimination rate constant depends on total body water, a postmenstrual-age maturation function, serum albumin, an acute lymphoblastic leukemia indicator, and the infusion duration. Total body water and the maturation function are derived inside model() from weight, height, age, and sex, so no separate columns are required. IMPORTANT: simulate this model with rxSolve(…, useLinCmt = FALSE). It is a one-compartment linear-elimination model, so rxSolve()’s default ODE-to-linCmt() auto-conversion replaces the ODE with a closed-form solution that holds the elimination rate constant at its t = 0 value and therefore silently discards the time dependence that is the entire point of this paper. The error is large and one-sided: exposure per dosing interval is under-predicted by about 17% at steady state (45% in the HLH/XLP group). See the validation vignette for the demonstration.
Butanediol rat (Fung 2008) Preclinical (rat). Population PK model for 1,4-butanediol (BD) and its bioactivation pathway in adult male Sprague-Dawley rats after intravenous and oral dosing, jointly with the unmeasured semialdehyde intermediate (ALD), the gamma-hydroxybutyric acid metabolite (GHB), and co-administered ethanol (ETOH). Each of the four substances follows two-compartment disposition with Michaelis-Menten (mixed-order) elimination; GHB additionally has a parallel first-order elimination. The metabolic flux is BD -> ALD -> GHB; ETOH has no metabolic connection to the BD/ALD/GHB chain. Mutual competitive inhibition is encoded for BD elimination inhibited by GHB and by ETOH; ALD elimination inhibited by BD; GHB elimination inhibited by BD; ETOH elimination inhibited by BD. Oral BD absorption is first-order with a 7.5 min lag-time; the absorbed fraction (F = 0.93, dose-independent) is split between BD and ALD central compartments to represent pre-systemic conversion of BD to ALD. The fraction entering as BD was dose-dependent in the source paper (30 percent at 1.58 mmol/kg and 55 percent at 6.34 mmol/kg); the model defaults to the low-dose 1.58 mmol/kg fractions (30 percent as BD, 70 percent as ALD); see the vignette Assumptions and deviations for the high-dose alternative. Volume of distribution of ALD is mathematically non-identifiable and was constrained so that Vss of ALD equals Vss of BD (source paper footnote c). Volume of the central compartment for GHB was fixed at the average rat plasma volume of 0.010 L (source paper footnote b). Fit by NONMEM VI ADVAN9 with FOCE-I.
C3G (Jeon 2012) One-compartment first-order absorption population PK model with an absorption lag time for cyanidin-3-glucoside (C3G) following 2-week multiple oral dosing of 1 g black bean (Phaseolus vulgaris, Cheongjakong-3-ho) seed coat extract once daily in 12 healthy adult Korean volunteers (Jeon 2012), with log-normal IIV on CL/F and V/F (with correlation rho = 0.883) and on Ka, and proportional residual error.
Cabamiquine (Courlet 2023) Semi-mechanistic joint blood- and liver-stage population PK/PD model for the Plasmodium elongation factor 2 (eEF2) inhibitor cabamiquine (formerly DDD107498 / M5717) in healthy adult men challenged with Plasmodium falciparum (Courlet 2023, Table 1 and Fig. 2). The PK layer is the three-compartment transit-absorption plus enterohepatic- recycling model of modellib(‘Courlet_2023_cabamiquine_pk’), whose individual parameters were carried forward. The PD layer is a parasite turnover model at two life-cycle stages: a liver compartment seeded with Finc * inoculum infected hepatocytes that grows at a fixed kgr,l = ln(30000)/(624) = 0.072 /h and releases merozoites into blood through a logistic transfer function Tr_lb(t) centred on T50 = 6 days, and a blood compartment seeded at baseline parasitemia P0 that grows at kgr,b. Cabamiquine concentration in the central compartment drives parasite killing at both stages through a sigmoid Emax term with a shared maximum kill rate kki and a shared inverse-exponential onset delay (1 - exp(-kt t)) that lets the killing rate maximise over the first ~48 h after dosing. Blood-stage potency differs between the two challenge populations: EC50,b,SpzCh = EC50,b,IBSM * (1 - 0.83 * STUDY_SPZCH). Liver-stage potency (EC50,l = 0.66 ng/mL) is higher than blood-stage potency, reproducing the paper’s finding of superior efficacy against the liver stage. Set STUDY_SPZCH = 0 to simulate the induced blood stage malaria (IBSM) design (blood inoculation only) and STUDY_SPZCH = 1 for the sporozoite challenge (SpzCh) design (liver inoculation only).
Cabamiquine pk (Courlet 2023) Population PK model for the Plasmodium elongation factor 2 (eEF2) inhibitor cabamiquine (formerly DDD107498 / M5717) in healthy adult men, pooled from a first-in-human single-ascending-dose / induced blood stage malaria (IBSM) trial and a sporozoite-challenge (SpzCh) chemoprophylaxis trial (Courlet 2023 Supplementary Material 1). Three-compartment apparent disposition with linear elimination; a Savic transit-absorption chain feeds a depot that is absorbed into central at rate ka; and an enterohepatic recycling loop moves drug from central into a recycling (gallbladder / bile) compartment at rate k2g and releases it back into the absorption depot at rate kg1 once the depot emptying time MTIME is reached, reproducing the secondary concentration peak observed 24-30 h post-dose. The observed lack of dose proportionality is captured empirically by a power effect of the administered dose on the apparent central volume V2/F (exponent -0.50); allometric body-weight scaling is fixed at 0.75 on the apparent clearances and 1 on the apparent volumes. This is the parsimonious PK model that was carried forward, with individual parameters fixed, into the parasitemia PK/PD model; see modellib(‘Courlet_2023_cabamiquine’).
Cabazitaxel (Ferron 2013) Three-compartment population PK model for intravenous cabazitaxel in patients with advanced solid tumors (Ferron 2013)
Cabazitaxel (Janssen 2017) Two-compartment population PK model for intravenous cabazitaxel in 10 men with metastatic castration-resistant prostate cancer, with individual midazolam clearance as a CYP3A metabolic-phenotype covariate on cabazitaxel clearance (Janssen 2017 Table 2B ‘Metabolic phenotype model’). The covariate enters via a linear deviation from the population midazolam clearance reference of 26 L/h. The individual midazolam clearance is sourced from the companion Janssen 2017 midazolam model (see modellib(‘Janssen_2017_midazolam’)).
Cabozantinib (Lacy 2018) Two-compartment population PK model for oral cabozantinib (tyrosine kinase inhibitor) in healthy volunteers and patients with renal cell carcinoma, castration-resistant prostate cancer, medullary thyroid carcinoma, glioblastoma multiforme, or other advanced malignancies (Lacy 2018, n=1534 across 9 clinical studies). Absorption is described by parallel dual processes: a fraction F1 enters depot1 via first-order absorption rate Ka with absorption lag time ALAG1, and the remaining (1-F1) enters depot2 via zero-order infusion over duration D2. Capsule (vs tablet, reference) formulation reduces both Ka and overall bioavailability; Ka also scales with dose via a power function (DOSE/60 mg)^0.677. Two-compartment disposition (central + peripheral1) with first-order elimination from central. Covariates on CL/F and Vc/F are baseline age (power on median 64 y), body weight (power on median 81 kg), female sex, race (Black/Asian/Other vs White reference), and tumor type (RCC/CRPC/MTC/GB/Other vs HV reference); MTC cancer type drives an approximately 93% higher CL/F.
Cabozantinib dose modification (Lacy 2018) Repeated time-to-event (RTTE) hazard model for the ‘dose modification of any kind’ (DMAK) endpoint in adults with advanced renal cell carcinoma (RCC) treated with oral cabozantinib in the phase III METEOR study (Lacy 2018 exposure-response analysis, n=317 patients with 0-52 events per patient). The instantaneous risk of a dose modification (interruption, reduction, or escalation) depends on whether the patient is currently on an active dose or on a dose interruption. When on active dose (DOSE > 0) the hazard increases log-linearly with the time-varying average cabozantinib plasma concentration CAV. When on a dose interruption (DOSE = 0) the hazard is governed by a separate, larger baseline log-hazard with no cabozantinib effect (cabozantinib effect on hold-state hazard was tested and dropped during base-model development). The drug input Cavg is the individual predicted daily average plasma cabozantinib concentration (ng/mL) derived from the upstream Lacy 2018 popPK model Lacy_2018_cabozantinib. Forward simulation exposes hazard (instantaneous DMAK rate per day) and sur (survival = probability of no DMAK event since t = 0) as derived outputs.
Cabozantinib tumor (Lacy 2018) Longitudinal sum-of-tumor-diameter (SOD) growth-inhibition PD model for oral cabozantinib in adults with advanced renal cell carcinoma (RCC) enrolled in the phase III METEOR study (Lacy 2018 exposure-response analysis, n=319 patients with 1637 evaluable tumor-diameter measurements). The tumor diameter Y follows first-order exponential growth at rate k_grow, with a saturable cabozantinib drug-effect of the form Cavg/(EC50 + Cavg) modulating a time-dependent decay rate decay(t) = k_dmax + k_dmax_tot * exp(-k_tol * t). The k_dmax term is the non-attenuating asymptotic decay rate, k_dmax_tot is the magnitude of the resistance-driven loss of decay rate, and k_tol governs the attenuation kinetics (paper-reported attenuation half-life 25.6 days). The drug input Cavg is the individual predicted daily average plasma cabozantinib concentration (ng/mL) carried as a time-varying CAV data column; the upstream popPK model is Lacy_2018_cabozantinib (Lacy 2018 popPK companion paper). Residual error is additive on Y (mm); IIV is exponential on Y(0), k_grow, k_dmax, and k_dmax_tot, with IIV on EC50 and k_tol fixed at a near-zero variance (paper Supplemental Table 3 footnote b).
Canakinumab (AitOudhia 2012) Integrated population PK/PD model of canakinumab (anti-IL-1beta IgG1/k mAb) in adults with rheumatoid arthritis (Ait-Oudhia 2012). Two-compartment popPK for total canakinumab is coupled to a quasi-equilibrium target-binding model with endogenous IL-1beta (zero-order production ksyn, linear clearance CLL). Predicted free IL-1beta drives downstream PD: (1) a three-compartment CRP transduction chain with a power-law stimulation (beta) on free-IL-1beta ratio and an empirical amplification (gamma) on the input to the third compartment, and (2) a single-compartment ACR latent variable (ACRL) regulated by a sigmoid Emax on the drop in free IL-1beta below baseline plus a first-order placebo build-up; the latent is mapped to ACR20/50/70 response probabilities via a logit transform with a between-subject random effect. Only body weight was a significant covariate (allometric on CL, CLL, CLDL, Vc, Vp with reference 70 kg).
Canakinumab (Chakraborty 2012) Population pharmacokinetic-binding model for canakinumab (anti-IL-1b IgG1/k monoclonal antibody) and its endogenous target IL-1b in adult cryopyrin-associated periodic syndromes (CAPS) patients (Chakraborty 2012). Two physical compartments (central and peripheral) each carry three species: free canakinumab, free IL-1b, and the canakinumab-IL-1b complex. Drug, ligand, and complex share the same volumes of distribution; complex clearance is set equal to free-drug clearance (CLX = CLD). Distribution between compartments uses two permeability-surface-area coefficients: PSD for free drug and complex, PSL for free ligand. Endogenous IL-1b production RLI enters the peripheral compartment. Drug-ligand binding is solved algebraically under a quasi-steady-state assumption with dissociation constant KD (Hayashi 2007 form). Subcutaneous bioavailability F1 was estimated on the logit scale; this file uses the Sp2/0 cell-line value (commercial Ilaris). Body weight modifies CLD, VC, VP via centred power covariates; serum albumin modifies CLD; age modifies SC ka. Two observed analytes: total canakinumab (free + complex) in ug/mL and total IL-1b (free + complex) in pg/mL.
Canrenone (Suyagh 2012) One-compartment population PK model for canrenone, the pharmacologically active metabolite of intravenous potassium canrenoate (K-canrenoate), in 23 paediatric patients (2 days to 10 years; median weight 4 kg, range 2.16-28.0 kg) receiving K-canrenoate in the NICU / PICU for retained fluids or congestive heart failure (Suyagh 2012). The K-canrenoate dose compartment (modelled as ‘depot’ because only canrenone is measured) is converted to canrenone by a first-order metabolic transformation rate kf (paper symbol) = 5.25 1/h. Canrenone disposition is described by apparent clearance CL/F and apparent central volume V/F (the ‘F’ factor absorbs the unknown fraction of K-canrenoate that ultimately reaches the canrenone compartment; the model assumes the total dose is converted to canrenone). Bodyweight scales CL/F and V/F by fixed allometric exponents (0.75 on CL, 1.0 on V) with a reference weight of 70 kg. No other covariate (gestational age, postnatal age, postmenstrual age, serum creatinine, serum albumin, haematocrit, sex) was retained in the final model. Residual error is proportional only.
Capecitabine (Blesch 2003) Population PK model ‘CAP7440’ for oral capecitabine, evaluated by simultaneously fitting plasma concentrations of three sequential metabolites: 5’-DFUR (5’-deoxy-5-fluorouridine), 5-FU (5-fluorouracil), and FBAL (alpha-fluoro-beta-alanine). Parent capecitabine and the first metabolite 5’-DFCR are not modelled as compartments; the dose enters the 5’-DFUR pool directly through a first-order absorption rate constant KA with absorption lag TLAG. Sequential first-order kinetics carry mass through 5’-DFUR -> 5-FU -> FBAL using apparent oral clearances CL/F and volumes V/F (CL1/V1 for 5’-DFUR, CL2/V2 for 5-FU, CL3/V3 for FBAL). 5-FU volume V2 is fixed at 17.8 L from literature (Heggie 1987) because it was not sensitive to the dataset. Three retained covariate effects in the final population PK model: alkaline phosphatase on 5-FU clearance (CL2), creatinine clearance on FBAL clearance (CL3) and volume (V3), and body surface area on FBAL volume (V3) – all multiplicative power forms. Fit to 481 patients with advanced or metastatic colorectal cancer (Phase III studies) plus 24 patients with extensive sampling from a Phase I bioequivalence study.
Capecitabine (Urien 2005) Population PK model for oral capecitabine and its three sequential metabolites 5’-DFCR (5’-deoxy-5-fluorocytidine), 5’-DFUR (5’-deoxy-5-fluorouridine), and 5-FU (5-fluorouracil) in 40 adult patients with metastatic cancer (Urien 2005). Capecitabine PK is a one-compartment apparent V1/F model with first-order absorption (Ka) and a lag time; non-transformation elimination CL10/F runs in parallel with the formation clearance CL12/F to 5’-DFCR. Each metabolite has its own central compartment with apparent volume fixed to 1 L (only output rate constants are identifiable in the source NONMEM ADVAN6 fit), so the chain 5’-DFCR -> 5’-DFUR -> 5-FU -> output is described by first-order rate constants K23, K34, K40 (paper’s notation). Total bilirubin (canonical TBILI; source column BILT, umol/L) is the only retained covariate: power exponent +0.32 on CL10/F and -0.36 on K34, both centred on the median bilirubin 8.8 umol/L. Inter-individual variability is reported on TLAG, V1, CL10, K23, K34, and K40; ISV on CL12 was fixed to 0 and ISV on Ka was deleted in favour of a large inter-occasion variability on Ka that is not represented in this static model file (see vignette Errata).
Carbenicillin qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for carbenicillin (penicillin): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Carboplatin (Ekhart 2008) Two-compartment population PK model for free (ultrafilterable) carboplatin in adult cancer patients (Ekhart 2008)
Carboplatin (Zandvliet 2008) Two-compartment population PK model for free (ultrafilterable) carboplatin in adult cancer patients receiving combination chemotherapy with indisulam (Zandvliet 2008). Clearance is modelled as a renal + non-renal split: a Cockcroft-Gault creatinine-clearance-proportional renal component (theta1 = 0.76) plus a fixed non-renal component (theta2 = 1.5 L/h, fixed at the Calvert 1989 estimate).
Carfentanil iv (Mann 2022) Three-compartment IV carfentanil population PK with a first-order biophase (effect-site) equilibrium compartment, used as the agonist input layer of the Mann 2022 translational opioid-overdose model. Mann 2022 had no full carfentanil clinical PK study to fit – only a single microdose case report (Minkowski 2012) reporting a roughly 45-minute plasma half-life. They therefore took the fentanyl Algera 2021 PK micro-constants and applied a fixed set of rate- constant modifications (k_el and k13 divided by 10; k21 and k31 multiplied by 10; k1 increased to 10/min) to reproduce the longer plasma persistence and faster effect-site equilibration of carfentanil. The resulting macro-constants encoded in ini() below yield exactly those micro-constants when combined with (WT/70)^0.75 / (WT/70) allometric scaling. Inter-subject variability is assumed equal to the fentanyl model (same omega^2 values, no carfentanil-specific IIV estimated). Outputs plasma concentration Cc in ng/mL and effect-site Ce in pM for downstream consumption by Mann_2022_mu_receptor_binding.
Carvedilol (Honda 2005) One-compartment population PK model for orally administered racemic carvedilol in 23 healthy Japanese volunteers, with R- and S-enantiomer whole-blood concentrations measured by chiral HPLC at 2 h and 6 h after a single 5- or 10-mg oral dose (Honda 2005). NONMEM ADVAN1/TRANS2 with very rapid absorption: the racemic dose is split equally between two parallel central compartments (central_r, central_s) with no separate absorption depot. CL/F and V/F scale linearly with body weight; an S/R ratio theta_3 (CL/F) and theta_4 (V/F) parameterise the stereoselective difference. One subject-level eta on CL/F and one on V/F are shared between enantiomers (correlated block IIV, rho ~ 0.90). Power-variance residual error with fixed exponent 1/2 (Honda Eq. 3), shared between R- and S-enantiomer observations. CYP2D610 genotype is not in the structural model; Honda 2005 reports the 10-carrier effect only as a post-hoc stratification of the individual Bayes estimates (Figs. 3-4).
Carvedilol (Hwang 2023) Sequential population PK-PD model for oral carvedilol in 21 healthy Korean male volunteers genotyped for CYP2D6 (Hwang 2023). PK is a two-compartment model (central + peripheral1) with zero-order absorption of duration D1 into the central compartment preceded by an absorption lag time Tlag, and first-order elimination; volumes and clearances are apparent (CL/F, Vc/F, Vp/F, Q/F). The only retained covariate is the CYP2D610/10 (intermediate-metabolizer-2) genotype, which reduces CL/F by 32.8 percent relative to the pooled CYP2D61/1, 1/2, 1/10, 2/10 reference. PD is a direct-effect competitive-antagonism Emax model for the heart-rate response to an isoproterenol sensitivity test: HR = E0 + Emax * D / (ED50 * (1 + C / IC50) + D), where D is the isoproterenol challenge dose in ug (covariate DOSE_ISOPROTERENOL_UG), C is the carvedilol plasma concentration in ng/mL, and the Hill exponent was fixed at 1. Carvedilol competitively shifts the isoproterenol dose-response curve to the right without changing Emax; no covariate, including CYP2D6 phenotype, was retained on any PD parameter.
Carvedilol (Othman 2007) Two-compartment population PK model for S(-)-carvedilol in healthy volunteers after oral administration of the immediate-release (IR) and the new controlled-release (CR) dosage forms of carvedilol (Othman 2007). Three parallel depot compartments encode the dosage-form-specific absorption: depot (CR, 3-stage time-varying KA), depot2 (IR morning, 2-stage), depot3 (IR evening, 2-stage). Diurnal variability in IR absorption is captured by separate morning and evening KAs and a lower IR-PM relative bioavailability.
Casirivimab (Lin 2024) Two-compartment population PK model for casirivimab in pediatric and adult subjects (non-infected, ambulatory or hospitalized SARS-CoV-2-infected, or household contacts) following IV or SC administration (Lin 2024, casirivimab arm of the joint casirivimab + imdevimab popPK model)
Caspofungin (Wurthwein 2013) Linear two-compartment population PK model with proportional residual error for once-daily 2-hour intravenous caspofungin infusions (70, 100, 150, 200 mg QD) in adults with proven or probable invasive aspergillosis (Wurthwein 2013). Clearance and central volume share a single linear body-weight fractional change centred on the cohort median body weight of 76 kg (CL_i = CL_typ * [1 + 0.0102 * (WT - 76)]; V1_i = V1_typ * [1 + 0.0102 * (WT - 76)]). Inter-individual variability is modelled exponentially on CL, V1, and V2 with an estimated CL-V1 covariance (correlation 0.802). Inter-occasion variability (16% CV) is included on CL across five sampling occasions (days 1, 4, 7, 14, 28) via the OCC covariate; downstream users who only need typical-value or IIV-only simulations can pass OCC = 0 (or any value outside 1..5) so the IOV terms zero out. Dose-level, gender, age, baseline serum bilirubin and baseline creatinine clearance were screened but not retained.
CC292 (Li 2017) Two-compartment population PK model for oral CC-292 (spebrutinib, a potent Bruton tyrosine kinase inhibitor) in 145 pooled subjects: 32 healthy adults (AVL-292-004) and 113 patients with relapsed and/or refractory B-cell malignancies including chronic lymphocytic leukemia (AVL-292-003). First-order absorption with a single absorption lag, linear elimination from the central compartment, with linear-deviation female-sex effect on apparent clearance (females have 26% lower CL/F) and a power age effect on apparent central volume (reference age 62 years). Residual variability is split into healthy-volunteer and patient strata.
Cebranopadol (Kleideiter 2017) Two-compartment population PK model for oral cebranopadol with two lagged transition compartments in healthy subjects and chronic-pain patients (Kleideiter 2017; with 2018 correction)
Cebranopadol (Kleideiter 2018) Two-compartment population PK model for cebranopadol, a NOP / opioid receptor agonist, in healthy adults and adult chronic-pain patients (low back pain or osteoarthritis, diabetic polyneuropathy, post-bunionectomy), with two transit absorption compartments before central, first-order elimination, and covariate effects from sex, CYP2C9 phenotype, ALT, CrCl, age, body weight, formulation, and disease status (Kleideiter 2018)
Cediranib (Li 2017) Two-compartment population PK model for oral cediranib (AZD2171) in adult cancer patients (Li 2017), with sequential zero- and first-order absorption (zero-order release into depot followed by first-order absorption to central), bioavailability fixed to 1, allometric power scaling on apparent clearance ((WT/73 kg)^0.517 and (Age/59 y)^-0.409) and on apparent central volume ((WT/73 kg)^0.65), correlated inter-individual variability between CL/F and Vc/F (correlation 0.839), independent IIV on Ka, and proportional residual error (rich-sampling estimate).
Cefadroxil mouse pbpk (Saleh 2023) PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, C57BL/6 Pept2+/+). Nine-compartment CNS PBPK model predicting unbound cefadroxil concentrations in brain extracellular fluid (brain ECF) after a single 36 mg/kg intravenous dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug’s physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.05 (estimated by the authors; Table V). The plasma PK model is the empirical one-compartment model of Table IV (in-house Monolix fit) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published.
Cefathiamidine (Zhi 2018) Two-compartment population PK model for intravenous cefathiamidine (a first-generation cephalosporin) in 54 children (age 2.0-11.8 years; weight 8.0-36.0 kg) with hematologic disease, developed in NONMEM v7.2 (FOCE-I) from 120 sparse plasma samples. Structural model: first-order elimination from a central compartment, with allometric body-weight scaling on CL, Q (exponent 0.75) and V1, V2 (exponent 1), reference weight 17.75 kg (the cohort median current weight). Inter-individual variability (exponential) is estimated for CL and V2 only; residual variability is exponential (lognormal on the linear scale). Bodyweight was the only retained covariate; age and creatinine clearance were not significant in the limited cohort (CrCL range 130-462 mL/min).
Cefepime (Capparelli 2005) One-compartment population PK model for cefepime in preterm and term neonates (Capparelli 2005); additive renal-plus-non-renal CL on serum creatinine, additive Vc step for PCA < 30 weeks.
Cefepime (Jonckheere 2019) Two-compartment population PK model for IV cefepime in critically ill ICU patients (Jonckheere 2019), updated by simultaneously fitting plasma + urine PK from the original Jonckheere 2017 pilot (STDY1) and the Jonckheere 2019 target-controlled-infusion cohort (STDY2). Total clearance is the sum of an estimated-creatinine-clearance-driven renal arm (CL_renal = 2.29 * (eCrCL/60)^0.943 L/h per 70 kg) and a covariate-free non-renal arm (CL_nonren = 0.795 L/h per 70 kg); all PK parameters are scaled allometrically with body weight (reference 70 kg, exponent 3/4 for clearances, 1 for volumes). The structural form encodes the non-dialysis patient (paper Equations 1-4); a separate CL_dialysis = 4.48 L/h applied during intermittent hemodialysis sessions in the source dataset is documented in the vignette but not enabled in this model file.
Cefepime (Shoji 2016) Two-compartment IV population PK model for cefepime in 91 neonates, infants, and children (Shoji 2016); body-weight allometric scaling (fixed exponents 0.75 on CL and Q, 1.0 on Vss), nonlinear postmenstrual-age maturation on CL, a power effect of serum creatinine on CL, and a power effect of gestational age on Vss. Central volume of distribution enters as a fixed fraction of steady-state volume (V/Vss = 0.460).
Cefepime qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for cefepime (cephalosporin): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Cefotaxime (Ahsman 2010) One-compartment population PK model for cefotaxime (CTX) and its active metabolite desacetylcefotaxime (DACT) in critically ill neonates and infants on extracorporeal membrane oxygenation (ECMO). IV bolus parent with first-order elimination; the metabolite is generated 1:1 from parent elimination on a CTX-equivalent mass basis (the source paper converted observed DACT concentrations to CTX equivalents by the molecular weight ratio Mr_CTX / Mr_DACT = 455.5 / 413.4 before fitting, and assumed a conversion fraction FDACT/CTX = 1). Parent CL is scaled by body weight (WT) via a power model centred at 3.5 kg; both parent and metabolite CL include a power covariate effect of time after ECMO decannulation (T_POST_ECMO) centred at 100 h that is removed during ECMO (T_POST_ECMO = 0). Metabolite CL additionally includes a power covariate effect of continuous venovenous hemofiltration flow Q_CVVH centred at 193 mL/min that is removed when CVVH is not running (Q_CVVH = 0). Volumes (Vc, Vc_dact) carry no covariate effects.
Cefotaxime (Leroux 2016) Two-compartment IV population PK model for cefotaxime in neonates and young infants (Leroux 2016). Clearance, central volume, peripheral volume, and inter-compartmental clearance are allometrically scaled to current body weight (fixed exponents 0.75 on CL and Q, 1.0 on V1 and V2; reference weight 1.665 kg). Clearance carries a power-form maturation function on gestational age (reference 30 weeks) and postnatal age (reference 12 days). Only CL has inter-individual variability; residual error is proportional.
Cefoxitin qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for cefoxitin (cephalosporin): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Cefpirome (Bulitta 2011) Three-compartment population PK model for IV cefpirome with simultaneous fit of plasma concentrations and amounts excreted unchanged in urine. Built from a pooled cohort of 24 Caucasian adults: 12 cystic fibrosis (CF) patients and 12 healthy volunteers (HVs) each given a single 10-min IV infusion of 2 g cefpirome. Body size is captured by allometric scaling on lean body mass (LBM) with fixed exponents 0.75 on clearance terms and 1.0 on volumes (reference LBM = 53 kg). Total clearance is split into an estimated renal arm (CL_R, urinary recovery is tracked in the canonical urine compartment) and a non-renal arm (CL_NR). A CF / HV cohort indicator DIS_CF (1 = CF patient, 0 = HV reference) carries three disease- specific scale factors estimated by the paper: FCYF_CLR = 1.07 applied to CL_R, FCYF_CLNR = 1.13 applied to CL_NR, and FCYF_VSS = 0.98 applied uniformly to V1 (central), V2 (shallow peripheral), and V3 (deep peripheral). The inter-compartmental clearances Q12 (central <-> shallow) and Q23 (central <-> deep) are shared across cohorts. Typical-value clearance and volume estimates are anchored to DIS_CF = 0 (HV reference) per the DIS_CF covariate convention registered in inst/references/covariate-columns.md.
Cefpirome (Kang 2020) Two-compartment IV-bolus population PK model for cefpirome in critically ill adults on venoarterial extracorporeal membrane oxygenation (VA-ECMO) (Kang 2020). Final-model covariates: power-form serum creatinine on CL (reference 1.6 mg/dL), and a binary ECMO-active treatment-status indicator on CL (1.41-fold higher when ECMO-ON) and V1 (4.22-fold higher when ECMO-ON).
Cefquinome pbpk (Mi 2023) Veterinary (pig). PBPK (whole-body, six-compartment, Berkeley Madonna 10.1.3) for the cephalosporin cefquinome after intramuscular injection in swine, developed to optimise the dosage regimen against respiratory-tract pathogens and to predict the withdrawal interval for edible tissues (Mi et al. 2023, PLoS Comput Biol 19:e1011331). Plasma (venous and arterial blood pools), liver, kidney, muscle and a lumped rest-of-body compartment are perfusion-limited and well stirred; the lung is permeability-limited and resolves into vascular blood, interstitial fluid and tissue sub-compartments, because lung interstitial fluid is the site of action for respiratory pathogens and was measured directly by microdialysis in this study. Only unbound drug crosses between the lung sub-compartments and only unbound arterial drug perfuses the tissues. Intramuscular absorption is a two-step process: 90% of the injected dose is immediately available at the injection site and is absorbed into venous blood at first-order rate Kim, while 10% is released from a slow depot at first-order rate Kdiss into the same injection-site pool. Elimination is renal (from kidney) plus hepatobiliary (from liver). The 11 parameters that Mi 2023 Table 4 carried through the 1000-animal Monte Carlo pop-PBPK analysis carry between-animal variability here; every other parameter is a fixed physiological or fitted chemical-specific constant. The model was hand-calibrated in Berkeley Madonna against digitised literature data, so no standard errors and no residual-error model are reported – the propSd term is a placeholder. Several Table 3 / supplement-code discrepancies are reproduced as coded; see the vignette Errata.
Ceftaroline (Minichmayr 2024) Two-compartment population PK model with linear elimination for ceftaroline (the active metabolite of the prodrug ceftaroline fosamil) in 12 healthy male volunteers receiving either the standard dosing regimen (ceftaroline fosamil 600 mg every 12 h as a 1 h intravenous infusion) or the approved intensified regimen (600 mg every 8 h as a 2 h intravenous infusion). Doses are entered as milligrams of the administered prodrug ceftaroline fosamil; the model converts to ceftaroline via the fixed ceftaroline-to-ceftaroline-fosamil molar-mass ratio of 0.883 applied as the bioavailability of the central compartment, reflecting the paper’s finding that prodrug-to-drug conversion is complete and effectively instantaneous (a depot / k_trans conversion compartment worsened the fit, and a sensitivity analysis favoured k_trans > 1000/h). Exponential interindividual variability on CL, Vc and Vp; combined additive and proportional residual error. No covariates were retained: creatinine clearance was screened on CL as linear and power functions and via a previously published CL-CLcr relationship, but was not significant in this small, homogeneous cohort with unimpaired renal function. The model underpins the paper’s systematic comparison of the three components of dosing intensification (dosing interval, infusion duration and total daily dose) on the PK/PD index fT>MIC against Staphylococcus aureus.
Ceftaroline (Riccobene 2016) Joint two-compartment ceftaroline fosamil (prodrug) plus three-compartment ceftaroline (active metabolite) population PK model with an algebraic epithelial-lining-fluid (ELF) partition coefficient, developed on plasma and bronchoalveolar-lavage ELF data from 50 healthy adults given 600 mg ceftaroline fosamil as a 1-h IV infusion q12h or q8h to steady state (Riccobene 2016). Ceftaroline fosamil is assumed to be converted completely to ceftaroline, so the whole prodrug elimination clearance CLcf enters the ceftaroline central compartment. ELF concentrations are not a distribution compartment: the parallel decline of plasma and ELF made an ELF compartment unidentifiable, so ELF is modelled as the ceftaroline plasma concentration scaled by a partition coefficient (0.193) carrying inter-individual variability. Body weight enters allometrically on all clearances (0.75) and volumes (1); BSA-normalised creatinine clearance below 80 mL/min/1.73 m2, age, hemodialysis status and healthy-versus-patient status act on ceftaroline clearance, and healthy-versus-patient status acts on ceftaroline central volume. An intramuscular depot (first-order ka, bioavailability fixed to 1) is carried from the upstream adult model; the ELF study itself used IV dosing only.
Ceftazidime (Bulitta 2010) Three-compartment population PK model for ceftazidime after 5-min IV infusion in cystic fibrosis patients and healthy volunteers (Bulitta 2010), with allometric fat-free-mass scaling and a cystic-fibrosis-vs-healthy disease-group factor on total clearance.
Ceftazidime (Conil 2007) Two-compartment IV population PK model for ceftazidime in adult burn-ICU patients, with creatinine clearance on CL and sex / mechanical ventilation / creatinine clearance on the peripheral volume V2 (Conil 2007)
Ceftazidime (Georges 2009) Two-compartment IV population PK model for ceftazidime in critically ill adults (ICU). Total clearance is an additive linear function of MDRD-estimated glomerular filtration rate; central volume V1 is selected by mechanical-ventilation status; peripheral volume V2 is selected by ICU admission etiology (polytrauma, postsurgical, or medical).
Ceftazidime (Shi 2018) One-compartment IV population PK model for ceftazidime in infants 0.1-2.0 years (Shi 2018) with allometric body-weight scaling and a power-form creatinine-clearance effect on clearance.
Ceftazidime (Zieck 2023) One-compartment IV population PK model for ceftazidime in adult general-ward patients spanning adequate to severely impaired renal function (Zieck 2023); clearance is a power function of CKD-EPI estimated glomerular filtration rate and is 1.56-fold higher with concomitant use of another antibiotic.
Ceftazidime avibactam control (Zazo 2024) Semi-mechanistic PK/PD model of ceftazidime-avibactam against Pseudomonas aeruginosa (strain 2154) in the NON-ICU control population. One-compartment linear PK for each drug, with volume proportional to body weight and clearance a linear function of creatinine clearance, driving a two-state bacterial growth model (an actively growing population carried on the log10 scale and a resting population carried on the linear CFU/mL scale) under a shared logistic capacity limit. Ceftazidime kill follows a sigmoidal Emax function whose EC50 is lowered biexponentially by avibactam; exponential delay functions retard growth and initial kill; ceftazidime is degraded by a bacterial-density-dependent Hill function that avibactam inhibits. Sibling model for the critically ill population: Zazo_2024_ceftazidime_avibactam_icu.
Ceftazidime avibactam fosfomycin hfim (Kroemer 2024) In vitro (Escherichia coli, clinical MDR isolate; CTX-M-15 + TEM-4 ESBL and OXA-244 carbapenemase). Semi-mechanistic PK/PD model of the dynamic hollow fiber infection model for ceftazidime/avibactam and fosfomycin: the static time-kill model (susceptible plus joint resistant subpopulation, sigmoidal Emax or power-model kill, Bliss independence, GPDI synergy) extended with two phenotypically less susceptible subpopulations growing at 3x MIC, whose emergence is suppressed by ceftazidime and fosfomycin through subpopulation synergy. Drug concentrations are dosable one-compartment states declining with the nominal hollow fiber half-life. Sibling model: Kroemer_2024_ceftazidime_avibactam_fosfomycin_tkc.
Ceftazidime avibactam fosfomycin tkc (Kroemer 2024) In vitro (Escherichia coli, clinical MDR isolate; CTX-M-15 + TEM-4 ESBL and OXA-244 carbapenemase). Semi-mechanistic PK/PD model of the static time-kill experiments for ceftazidime, avibactam and fosfomycin: a susceptible and a joint resistant bacterial subpopulation under a shared capacity limit, with sigmoidal Emax or power-model mono-drug kill, Bliss independence as the additivity criterion, and general pharmacodynamic interaction (GPDI) terms describing the >99% reduction of the ceftazidime EC50 by avibactam and of the fosfomycin EC50 by ceftazidime. Drug concentrations are static covariates (no dosing events). Sibling model: Kroemer_2024_ceftazidime_avibactam_fosfomycin_hfim.
Ceftazidime avibactam icu (Zazo 2024) Semi-mechanistic PK/PD model of ceftazidime-avibactam against Pseudomonas aeruginosa (strain 2154) in the CRITICALLY ILL (intensive care unit) population. One-compartment linear PK for each drug, with volume proportional to body weight and clearance a linear function of creatinine clearance, driving a two-state bacterial growth model (an actively growing population carried on the log10 scale and a resting population carried on the linear CFU/mL scale) under a shared logistic capacity limit. Ceftazidime kill follows a sigmoidal Emax function whose EC50 is lowered biexponentially by avibactam; exponential delay functions retard growth and initial kill; ceftazidime is degraded by a bacterial-density-dependent Hill function that avibactam inhibits. Sibling model for the non-ICU control population: Zazo_2024_ceftazidime_avibactam_control.
Ceftazidime qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for ceftazidime (cephalosporin): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Ceftobiprole (Cojutti 2023) Three-compartment IV population PK model for ceftobiprole in adults with severe Gram-positive infections (real-life multicentre therapeutic drug monitoring cohort, Italy). Clearance rises exponentially with CKD-EPI estimated glomerular filtration rate; central volume V1 is larger in males. Supports probability-of-target-attainment analysis against free-trough or free-steady-state fCtrough/MIC and fCss/MIC targets.
Ceftriaxone (Garot 2011) Two-compartment IV-infusion population PK model for ceftriaxone in critically ill adult ICU patients with sepsis, severe sepsis, or septic shock (Garot 2011)
Ceftriaxone (Tsai 2023) Two-compartment population PK model for intravenous ceftriaxone in Indigenous Australian adults with end-stage renal disease on three-times-weekly intermittent high-flux hemodialysis, receiving a novel 2 g three-times-weekly post-dialysis regimen. PK is parameterised on unbound drug: the central state carries unbound ceftriaxone and an explicit second-order albumin-binding exchange (k1 on / k2 off) against a capacity bmax derived from serum albumin carries the bound drug, so total and unbound plasma concentrations are both model outputs. Clearance is replaced (not augmented) by a > 10-fold higher dialytic clearance while a session is running, gated by the time-varying RRT_HEMODIAL_ACTIVE covariate; interdialytic clearance falls with serum bilirubin through an inverse-power relationship. Estimated with the Pmetrics non-parametric adaptive grid (NPAG). Tsai 2023, n = 16 subjects, 122 total-and-unbound plasma samples.
Cefuroxime (Alqahtani 2018) Two-compartment IV population PK model for cefuroxime in adults undergoing coronary artery bypass graft (CABG) surgery with cardiopulmonary bypass (Alqahtani 2018), with a power-form creatinine-clearance (Cockcroft-Gault) effect on clearance.
Cefuroxime (Nielsen 2011) In vitro (Streptococcus pyogenes M12 NCTC P1800). Semimechanistic PKPD model of cefuroxime time-kill kinetics; two-stage bacterial life-cycle (proliferating drug-sensitive S and non-growing drug-insensitive R) with sigmoidal Emax killing of S via an effect compartment; first-order drug elimination (ke set per in vitro kinetic-system flow rate) plus drug-specific degradation kdeg. Parameter values are from the combined static and dynamic estimation in Table 3.
Cefuroxime (Viberg 2006) Two-compartment population PK model for intravenous cefuroxime in adult patients with bacterial infections and a wide range of renal function (Viberg 2006); reciprocal serum cystatin C (1/CYSC) and body weight enter as centred-linear covariates on clearance, and body weight enters as a centred-linear covariate on the central volume of distribution.
Cefuroxime axetil (Bulitta 2009) Semiphysiological population PK model for oral cefuroxime axetil (acetoxyethyl-ester prodrug of cefuroxime) in healthy adult male volunteers after a standardized high-fat breakfast. Three drug compartments (stomach -> intestine -> central): a saturable, time-dependent Michaelis-Menten release from the stomach to the intestine followed by first-order absorption from the intestine to the central compartment, with one-compartment linear disposition. The maximum gastric-release rate Vmax is modulated over time-past- meal by a sigmoidal (Hill) function whose maximum fractional change Emax is logit-transformed to range over [-1, 9]. Vmax and Km are estimated as fractions of dose (Bulitta 2009 Results, ‘estimated and are reported as fractions of the cefuroxime dose’), so the absolute Vmax (mg/h) and Km (mg) scale linearly with the stomach-compartment dose amount. Parameter values reproduced here are from the S-ADAPT importance-sampling Monte Carlo EM fit (Bulitta 2009 Table 2, ‘S-ADAPT Population mean’ column), which the authors recommend as the best parametric fit; NONMEM and NPAG estimates are reported alongside in the paper for comparison.
Cemiplimab (Yang 2021) Two-compartment population PK model for cemiplimab (anti-PD-1 IgG4) with time-varying clearance (sigmoid Emax) in adults with advanced solid tumors including cutaneous squamous cell carcinoma (Yang 2021)
Cephalexin rat (Padoin 1998) Preclinical (rat, male Wistar). Two-compartment population PK model for cephalexin after intra-arterial (IA) or oral (gastric-tube) administration in rats, with first-order absorption and a competitive drug-drug interaction from coadministered oral quinapril that lowers cephalexin Ka (paper Table 4: 0.249 to 0.177 1/h; ~29% lower) and CL (paper Table 4: 0.810 to 0.640 L/h/kg; ~21% lower) when both drugs are given by the oral route. The paper parameterizes the disposition as {CL, Vc, CL_D, Vss = Vc + Vp}; this implementation uses the canonical {CL, Vc, Q, Vp} parameterization with the typical-value Vp derived as Vss - Vc = 1.23 - 0.416 = 0.814 L/kg. Intra-arterial quinapril or intra-arterial cephalexin produced no detectable interaction on cephalexin elimination, attributed by the authors to the much higher cephalexin renal concentrations outcompeting quinapril at the renal anionic transport carrier (and to Ka being irrelevant for IA dosing).
Cephalothin (Kuroda 2023) Veterinary (Thoroughbred horse). Three-compartment population PK model for the first-generation cephalosporin cephalothin (CET) in Thoroughbred horses, fitted jointly to single-dose intramuscular (11 mg/kg bwt, 8 horses, this study) and single-dose intravenous (22 mg/kg bwt, 12 horses, Kuroda 2021 Equine Vet J 53:1239-1249) plasma data by nonlinear mixed-effects modelling in Phoenix WinNonlin 8.3. Absorption from the intramuscular site is first order (Kabs) with a bioavailability factor F; disposition is a central compartment (V1) connected to a rapidly equilibrating peripheral compartment (V2, distribution clearance CL2) and a slowly equilibrating peripheral compartment (V3, distribution clearance CL3), with elimination clearance CL from V1. Every structural parameter is reported per kilogram of body weight in Kuroda 2023 Table 1, so the model scales each of them linearly with WT (exponent 1 fixed); doses are therefore given in mg and concentrations come out in ug/mL. The unbound concentration Cu = fu * Cc (fu = 0.8, taken from Ambrose 2007 and used unchanged by Kuroda 2023) is the quantity that drives the paper’s PK/PD target, fT > MIC for 40% of the dosing interval, and its Monte Carlo probability of target attainment. Kuroda 2023 states that between-subject variability was described by an exponential model on the structural parameters but reports NO variance estimates anywhere (the Table 1 CV% column is bootstrap precision of the typical value, not BSV), so every eta is encoded as fixed(0) and the model simulates typical-value profiles unless the user supplies variances. The residual model is combined proportional plus additive; the additive component was estimated separately per route and the packaged value is the intramuscular one (see the addSd comment for the intravenous value).
Certolizumab (Wade 2015) One-compartment population PK model with first-order SC absorption and an additive baseline concentration for certolizumab pegol in adults with Crohn’s disease (Wade 2015)
Cetagliptin (Zhou 2024) Two-compartment population PK model with first-order absorption and saturable Michaelis-Menten elimination for cetagliptin, coupled by a direct-effect sigmoid Emax model to plasma DPP-4 inhibition, in Chinese patients with type 2 diabetes mellitus. Total bilirubin is a power covariate on the peripheral volume of distribution.
Cetuximab (Dirks 2008) Two-compartment population PK model for intravenous cetuximab (anti-EGFR chimeric IgG1) in adults with recurrent and/or metastatic squamous cell carcinoma of the head and neck (SCCHN), with Michaelis-Menten (target-mediated) elimination from the central compartment. Ideal body weight and white blood cell count are linear-deviation covariates on Vmax; total body weight is a linear-deviation covariate on V1 (Dirks 2008 J Clin Pharmacol; Chapter 3 of the Dirks 2010 UTHSC PhD dissertation).
ChlorogenicAcid rat (Chen 2024) Preclinical (rat). Two-compartment intravenous pharmacokinetic model for chlorogenic acid (CGA), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the chlorogenic acid dose is the GHI volume dose times its content in GHI (12.9 ug/mL), i.e. 32.25 ug/kg, 64.5 ug/kg, 129 ug/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Direct-effect sigmoid-Emax models link the chlorogenic acid plasma concentration to the GHI-minus-model-group difference in creatine kinase-MB (CK-MB), ischemia-modified albumin (IMA), cardiac troponin I (cTn I) and alpha-hydroxybutyrate dehydrogenase (alpha-HBDH) (E = Emax * C^gamma / (ED50^gamma + C^gamma); Tables 12, 13, 14, 15). Chen 2024 fitted a PK/PD model only for the analyte/biomarker/dose combinations whose PLSR coefficient was negative, so the effect of an unfitted combination is returned as zero rather than extrapolated. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero.
Chloroquine (Simpson 2013) In vitro (P. falciparum). Sigmoid Emax inhibition model of chloroquine effect on hypoxanthine uptake by clinical Plasmodium falciparum isolates from the Thai-Myanmar border (Shoklo Malaria Research Unit, 1993-2005), with pfmdr1 genotype covariate effects on EC50. The ‘subject’ in the NLME framework is a parasite isolate (n=421 isolates with chloroquine data). STIM_CHLOROQUINE_NM is the per-well drug concentration in the in vitro hypoxanthine-uptake-inhibition assay; the model has no PK and no time evolution. E0 and Emax are fixed per Simpson 2013 Table 3 footnote.
Ciclosporin (Fanta 2007) Three-compartment population PK model with first-order absorption for ciclosporin in paediatric renal transplant candidates (Fanta 2007)
Ciclosporin (Frobel 2013) Parametric time-to-event (TTE) model for the first acute rejection (AR) event after paediatric kidney transplantation in patients receiving oral ciclosporin A (Neoral microemulsion). The baseline hazard is a five-interval step-function exponential with break-points at 5, 8, 25, and 100 days after transplantation. The final model carries no covariates: 15 candidate covariates (including ciclosporin AUC, baseline AUC, demographics, donor characteristics, HLA mismatches, dialysis time, basiliximab induction) were screened by univariate testing, stepwise covariate modelling, cross-validated SCM, and bootstrap-SCM, and none reached statistical significance or clinical relevance. The model output sur is the probability of remaining acute-rejection-free at time t; hazard and cumhaz are exposed as derived outputs.
Ciclosporin (Ni 2013) One-compartment first-order-absorption population PK model for oral ciclosporin in Chinese children with aplastic anemia (Ni 2013)
Ciclosporin (Press 2010) Two-compartment population pharmacokinetic model for oral ciclosporin A (Neoral) in adult kidney transplant recipients (Press 2010). Delayed absorption is described by one transit compartment with the first-order transit rate constant set equal to the absorption rate constant ka (chain: depot -> transit1 -> central at common rate ka; mean absorption time = (n+1)/ka with n = 1 transit compartment). Oral bioavailability is FIXED at 0.5 (Methods ‘Structural model’). Apparent clearance CL and apparent central volume of distribution Vc are allometrically scaled to body weight at a 76 kg median reference with theory-based exponents 0.75 on CL and 1.0 on Vc; the peripheral volume Vp and intercompartmental clearance Q are not weight-scaled. Concomitant high-dose oral prednisolone (PRED_DOSE >= 20 mg/day) is associated with a 55% reduction in the absorption rate constant and a 22% reduction in bioavailability (binary threshold-form covariate). Inter-occasion variability on bioavailability is encoded here as IIV on lfdepot because the source does not specify a per-subject occasion count for downstream simulation (see vignette Assumptions and deviations).
Ciclosporin (Wilhelm 2012) Two-compartment population PK model for ciclosporin (CsA) in adults undergoing haematopoietic allogeneic stem cell transplantation, with first-order oral absorption + lag time and a 3 h intravenous infusion directly into the central compartment (Wilhelm 2012). Twenty subjects on routine fluconazole antimycotic prophylaxis (a CYP3A4 inhibitor) were included; ciclosporin was assayed in whole blood by FPIA (AxSYM, Abbott). Body weight, body surface area, co-medication with CYP3A4 inducers and co-medication with CYP3A4 inhibitors were tested but none reached statistical or clinical significance, so no covariates are retained in the final model. Inter-individual variability was reported on every PK parameter (CL, Vc, Q, Vp, ka, F, tlag); the paper estimated a full omega variance-covariance matrix but did not publish the off-diagonal elements, so the packaged model uses diagonal IIVs only (see vignette Assumptions and deviations).
Ciclosporin (Willemze 2008) Two-compartment population PK model for ciclosporin in children (aged 1.8-16.1 years) after allogeneic haematopoietic stem cell transplantation (Willemze 2008). First-order absorption with lag time and partial bioavailability for oral Neoral microemulsion; intravenous Sandimmune is given as a 2-hour infusion to the central compartment. The ‘alternative parameterization’ (CL, Q, Vp, plus Ka, Vc, Tlag, F) reported in Table 2 is used directly because it is the more physiologically interpretable set. IIV on Vc was fixed to zero; IIVs on Ka, CL, Q, Vp, Tlag, and F are estimated. Residual error is proportional. No covariate (body weight, length, age, or estimated GFR) was retained in the final model; those covariates are documented in covariatesDataExcluded.
Ciclosporin (Woillard 2014) Two-compartment population PK model with Erlang-distributed transit absorption (5 sequential delay compartments) and first-order elimination for oral ciclosporin (CsA) in adult haematopoietic stem cell transplant (HSCT) recipients on graft-versus-host disease prophylaxis (Woillard 2014, NONMEM final model). The apparent peripheral volume of distribution Vp/F is fixed at 500 L; no covariate effects were retained in the final model. Combined additive plus proportional residual error.
Cilazapril pbpk (Luo 2024) PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). Joint cilazapril + cilazaprilat disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Cilazapril is an inactive ester prodrug hydrolysed by hepatic CES1 to the active diacid cilazaprilat, which is eliminated renally. Absorption is described by a directly reported first-order ka rather than by a permeability. The semi-PBPK circuit is stomach, three small-intestinal lumen segments (duodenum / jejunum / ileum), the matching three gut-wall segments, portal vein, liver, kidney and a two-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper’s virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded.
Cilostazol (Yoo 2009) Two-compartment population PK model for oral cilostazol with first-order absorption from the depot and an absorption lag time, estimated in 104 healthy Korean male volunteers receiving a single 50- or 100-mg dose (Yoo 2009). Apparent oral clearance CL/F is modulated by two pharmacogenetic covariates entered in linear-fractional form: a three-level CYP3A5 genotype (CYP3A51/1 = reference, 1/3 = -22.3%, 3/3 = -40.7%) and a three-level CYP2C19 metabolizer phenotype (extensive metabolizer = reference, intermediate = -14.7%, poor = -27.2%). The final NONMEM ADVAN4/TRANS4 model places exponential IIV on CL/F, Vc/F, Q/F and Vp/F with a partial OMEGA BLOCK retaining the (Vp/F, CL/F), (Q/F, Vc/F) and (Vp/F, Q/F) covariances; the remaining off-diagonals are held at zero. Residual error is combined additive plus proportional.
Cipaglucosidase (Hajjar 2018) Two-compartment population PK model for IV cipaglucosidase alfa (ATB200; recombinant human acid alpha-glucosidase / rhGAA) in adult patients with Pompe disease (Hajjar 2018 ACCP poster, phase 1/2 study ATB200-02 / NCT02675465). Disposition has parallel linear clearance (CL 0.569 L/h per 70 kg) and Michaelis-Menten saturable elimination (Vmax 98.6 mg/h per 70 kg, Km 62.4 mg/L) from the central compartment. Clearance and volume parameters are allometrically scaled by total body weight (exponent 0.75 fixed on all clearances, exponent 1 fixed on all volumes; reference 70 kg). Co-administration of the pharmacological chaperone miglustat (AT2221) reduces ATB200 linear CL: 130 mg AT2221 multiplies CL by 0.738 (26.2% reduction), 260 mg AT2221 multiplies CL by 0.595 (40.5% reduction); both effects are estimated as separate categorical covariate multipliers (paper Methods ‘A categorical covariate effect model was implemented’). Residual error is proportional (variance 0.0317, SD 0.178 on the linear-scale concentration).
Cipargamin (Hien 2017) Population PK / PD model of the spiroindolone antimalarial cipargamin (KAE609) in Vietnamese adults with acute uncomplicated Plasmodium falciparum malaria (Hien 2017). PK is a flexible transit-absorption chain (NN = 3 transit compartments fixed) into a one-compartment disposition model with fixed allometric body-weight scaling on CL/F and V/F. Bioavailability is anchored at F = 100% with between-subject variability. PD is a two-population parasite-clearance model: the asexual parasite pool at enrolment is split by a sensitive fraction Fsen and its complement (refractory subpopulation); sensitive parasites grow at a fixed Kgrow (= ln(10)/48 = 0.0479 /h corresponding to 10-fold multiplication per 48-h intraerythrocytic cycle) and are killed via an Emax / EC50 saturable-effect term driven by cipargamin plasma concentration; refractory parasites become active (join the sensitive pool) at first-order rate Kact and are then subject to the same drug kill. The typical Emax is dose-dependent per Hien 2017 Table 3 equation 7: Emax_i = TVEmax * (dose/10)^COVdose_Emax. Structural PK estimates are the Table 3 population values referenced to a 59 kg typical patient receiving 10 mg cipargamin. Reproduces the individually predicted parasite clearance profiles in Hien 2017 Fig 3.
Ciprofloxacin (Schaefer 1996) Two-compartment population PK model for intravenous and oral ciprofloxacin in 10 pediatric cystic fibrosis patients aged 6-16 years (Schaefer 1996). First-order absorption from a depot, two disposition compartments, plus a cumulative urine compartment driven by an independently estimated renal clearance. Total clearance is a linear-with-intercept function of body weight (CL = 8.8 + 0.396 * WT, L/h), and central / peripheral volumes are directly proportional to body weight with slopes 0.698 and 1.3 L/kg respectively. Intercompartmental clearance, absorption rate, renal clearance, and oral bioavailability are weight-independent. IIV is retained on CL, Vc, and Vp only; residual error is proportional. Calibrated to the study weight range (15-42 kg); extrapolation beyond is not appropriate per the source authors.
Ciprofloxacin (Thuo 2011) One-compartment population PK model with first-order absorption and absorption lag for oral ciprofloxacin in Kenyan children with severe malnutrition (Thuo 2011). Apparent CL and apparent Vc are allometrically scaled to body weight (exponents 0.75 and 1) and modified by linear deviations from a serum sodium reference of 136 mmol/L; apparent CL is further reduced by 28.3% in the paper-defined high-mortality-risk stratum.
Ciprofloxacin (Zhao 2014) Two-compartment population PK model with first-order elimination for intravenous ciprofloxacin in neonates and young infants less than three months of age (Zhao 2014). Central and peripheral volumes (V1, V2) scale allometrically with current body weight (fixed exponent 1, reference 1.955 kg); clearance (CL) and inter-compartmental clearance (Q) scale with current body weight at a fixed exponent of 0.75. CL is further multiplied by a renal-maturation factor in gestational age and postnatal age (F_age), a renal-function factor in serum creatinine (RF = exp((CREAT - 42 umol/L) * theta7)), and a fractional reduction (factor 0.708) when inotropic / vasoactive agents are coadministered. IIV is reported on V1, V2, and CL as %CV on an exponential model. Residual error is proportional. Inter-occasion variability on CL (16.4%CV) reported by Zhao 2014 is not encoded structurally here – the source paper does not define an operational occasion mapping for the model-library use case; users who need IOV can add an OCC indicator and per-occasion eta downstream.
Ciprofloxacin colistin invitro (Zhao 2024) In vitro (Escherichia coli; one clinical urinary isolate C47 plus MG1655 wild type LM347 and its isogenic gyrA1/marR mutants LM378 and LM421). Semi-mechanistic PK/PD model of static-concentration time-kill for ciprofloxacin and colistin alone and in combination. Five pre-existing subpopulations (the 2x2 ciprofloxacin/colistin susceptible-resistant grid plus a pre-existing resting subpopulation present only in C47), each cycling between a growing S state, a density-driven resting R state and a transient ciprofloxacin-induced non-colony-forming Nc state. Killing is additive across drugs (subpopulation synergy) with concentration-independent interaction factors shifting the ciprofloxacin and colistin EC50 in combination. Colistin loss to labware is described by an Emax free-fraction model at time zero plus saturable binding kinetics during the experiment; ciprofloxacin is held at its nominal concentration. Sibling models: Zhao_2024_ciprofloxacin_colistin_plasma, Zhao_2024_ciprofloxacin_colistin_kidney.
Ciprofloxacin colistin kidney (Zhao 2024) Semi-mechanistic PK/PD (QSP) model predicting Escherichia coli killing in the kidney interstitium (pyelonephritis) by intravenous ciprofloxacin and colistimethate sodium, alone and in combination. The in vitro time-kill PD model of Zhao 2024 (five pre-existing subpopulations on the 2x2 ciprofloxacin/colistin susceptible-resistant grid plus a pre-existing resting subpopulation, additive killing with EC50-shifting interaction factors) is driven by UNBOUND KIDNEY INTERSTITIAL concentrations, predicted by appending a lumped kidney tissue-plus-tubule compartment per species to two published critically-ill population PK models: a two-compartment model for ciprofloxacin (Khachman 2011) and the CMS1/CMS2 prodrug plus formed-colistin model of Kristoffersson 2020. Kidney concentrations are predicted about 10-fold (ciprofloxacin) and 150-fold (colistin) higher than plasma, so effective killing is reached at much lower doses and much higher MICs than for a bloodstream infection. Colistin EC50 is rescaled to the simulated strain MIC. Sibling models: Zhao_2024_ciprofloxacin_colistin_invitro (the underlying time-kill fit) and Zhao_2024_ciprofloxacin_colistin_plasma (the same model driven by unbound plasma concentrations).
Ciprofloxacin colistin plasma (Zhao 2024) Semi-mechanistic PK/PD (QSP) model predicting Escherichia coli killing in human plasma (bloodstream infection) by intravenous ciprofloxacin and colistimethate sodium, alone and in combination. The in vitro time-kill PD model of Zhao 2024 (five pre-existing subpopulations on the 2x2 ciprofloxacin/colistin susceptible-resistant grid plus a pre-existing resting subpopulation, additive killing with EC50-shifting interaction factors) is driven by unbound plasma concentrations from two published critically-ill population PK models: a two-compartment model for ciprofloxacin (Khachman 2011) and the CMS1/CMS2 prodrug plus formed-colistin model of Kristoffersson 2020. Colistin EC50 is rescaled to the simulated strain MIC. Sibling models: Zhao_2024_ciprofloxacin_colistin_invitro (the underlying time-kill fit) and Zhao_2024_ciprofloxacin_colistin_kidney (the same model driven by kidney interstitial concentrations).
Cisplatin (Boer 2015) Two-compartment population PK model for long-term circulating platinum (Pt) decay after cisplatin-based chemotherapy in adult testicular cancer survivors followed 1-13 years post-treatment (Boer 2015). Dose is the cumulative cisplatin dose expressed as elemental Pt in mg (multiply cumulative cisplatin in mg by 0.6502, the Pt/cisplatin mass ratio 195.08/300.05). An apparent bioavailability F1 (fdepot) accounts for the fraction of the administered Pt remaining in the body after the rapid pre-measurement urinary-excretion phase. Pt is assumed to be cleared solely via urine.
Cisplatin (Royer 2011) Four-compartment population PK model for platinum after cisplatin intraperitoneal perioperative chemotherapy (PIPC), with or without intraperitoneal epinephrine, in adult women with recurrent epithelial ovarian cancer (Royer 2011). Compartments are peritoneum (paper- mechanistic IP dosing / sampling site), central (serum ultrafilterable platinum), peripheral1, and bound (paper-mechanistic protein-bound plasma platinum). Ultrafiltered platinum transfers peritoneum -> central at rate IPCL * (peritoneum / IPV), distributes central <-> peripheral at rate constants k12 = 0.632 /h and k21 = 0.0425 /h (paper notation k23 / k32), and is eliminated from central at rate CL * Cc. Protein-bound platinum is formed from central via a Michaelis-Menten binding term modulated by baseline serum total protein (Vmax * Cc / (KM + Cc) * TPRO, per Supplementary Data S1) and eliminated first-order at rate kB. Epinephrine decreases IPCL by 53.1% and increases central volume V by 80.5% (multiplicative fractional coefficients on the typical values). Ultrafiltered and protein-bound platinum are fitted simultaneously to 316 peritoneum + 577 unbound plasma + 577 bound plasma observations from 55 patients (26 with epinephrine, 29 without).
Cisplatin (Urien 2004) Integrated two-compartment population PK model for ultrafilterable (unbound) plasma platinum coupled to a metabolite compartment representing irreversibly protein-bound plasma platinum (Urien 2004 BJCP). Fitted simultaneously to 396 unbound and 477 total plasma platinum concentration-time observations from 43 adult cancer patients receiving 30-min cisplatin infusions. Unbound clearance depends on body surface area and Cockcroft-Gault creatinine clearance; the unbound central volume depends on BSA. The bound formation parameter fm/Vm depends on dose per m^2 and total serum protein; its BSA and creatinine-clearance exponents are fixed at the negatives of the unbound CL exponents so the formation flux (fm/Vm)CLCc is net BSA- and CLCr-neutral. The apparent metabolite volume Vm is not separately identifiable; the composite parameters fm/Vm (1/L) and CLm0/Vm (1/h) absorb Vm and the bound state is carried in concentration units.
Cladribine (Savic 2017) Population PK model for cladribine (CdA) in patients with relapsing-remitting multiple sclerosis (Savic 2017): three-compartment disposition with first-order oral absorption, separate fasted vs fed (or unknown food-state) absorption parameters, renal clearance proportional to Cockcroft-Gault creatinine clearance, and a multiplicative non-renal-clearance effect of concomitant subcutaneous interferon beta-1a coadministration.
Clarithromycin (Abduljalil 2009) Semimechanistic population pharmacokinetic model for oral clarithromycin and its 14-(R)-hydroxy metabolite during repeated b.i.d. administration (Abduljalil 2009): a single-phase Weibull absorption (kw, lambda) into a one-compartment parent disposition with linear distribution and a parent clearance that is partly inhibited by the parent’s own concentration in a hypothetical effect-style inhibition compartment (Imax form with FCLp = fraction of CLp not subject to inhibition and IC50 = inhibition-compartment concentration giving 50% of maximum inhibition); all parent metabolic clearance feeds a parallel one-compartment metabolite disposition (14-OH-clarithromycin). Body weight enters allometrically with fixed exponents 0.75 on CL and 1.0 on V (parent and metabolite), reference 70 kg.
Clazakizumab (Muthukrishnan 2025) Population PK and PK-PD model for the anti-interleukin-6 monoclonal antibody clazakizumab given as a 3-minute IV bolus to adults with end-stage kidney disease undergoing maintenance dialysis (POSIBIL 6 ESKD phase 2b, NCT05485961; Muthukrishnan 2025). Two-compartment linear PK with allometric weight scaling on CL and V1 and a baseline free IL-6 covariate on CL; sequential-fit indirect-response inhibition of hs-CRP zero-order production (kin) with Imax fixed to 1, an estimated Hill coefficient, and a Manly-transformed IIV on IC50. Baseline hs-CRP enters kin as a power covariate so the pre-dose steady state E0 = kin/kout tracks the observed baseline distribution.
Clesrovimab (Hu 2026) Two-compartment population PK model for clesrovimab in preterm and full-term infants (Hu 2026)
Clindamycin (Bouazza 2012) One-compartment population PK model for clindamycin administered orally (immediate-release tablet) or intravenously (20-min infusion) in 50 adult patients (ages 18-93 y, body weight 23-133 kg) treated for bone and joint infections (Bouazza 2012). First-order absorption for oral dosing with estimated absolute bioavailability F = 0.876; apparent clearance CL/F = 15.2 L/h at 70 kg, with an estimated (non-allometric) body-weight exponent of 0.497 on CL. Apparent volume V/F = 66.2 L and absorption rate Ka = 0.967 1/h, neither carrying retained interindividual variability. IIV is retained only on CL/F (omega = 0.39). Residual variability is proportional (sigma = 0.38). Rifampicin co-administration was screened but not retained in the final model (see covariatesDataExcluded).
Clindamycin (Muller 2010) Three-compartment intravenous popPK model for clindamycin in pregnant women during the peripartum period (Muller 2010). Fit to 175 maternal venous serum concentrations from 7 women receiving either 600 mg over 20 min every 6 h (endocarditis prophylaxis) or 900 mg over 30 min every 8 h (group B streptococcal disease prophylaxis). No covariates were retained in the final model; demographic and laboratory screens (maternal age, gestational age, BMI, weight, edema, temperature, creatinine, ALP, AST, ALT, mode of delivery) are documented in covariatesDataExcluded. Proportional residual error with a per-subject log-normal scaling eta on the residual error magnitude (NONMEM omega-sigma interaction with an extra ETA on epsilon).
Clindamycin (Smith 2017) One-compartment population PK model for intravenous clindamycin in obese and nonobese children, with allometric total body weight on CL and V, sigmoidal Hill maturation on CL by postmenstrual age, and power effects of serum albumin and alpha-1 acid glycoprotein on V (Smith 2017).
Clomethiazole (Zingmark 2003) Two-compartment intravenous population PK model for clomethiazole (Zingmark 2003) in 774 adult acute-stroke patients dosed with a three-phase IV infusion of clomethiazole edisilate over 24 h (6 mg/kg over 0.25 h then 31 mg/kg over 0.25-8 h then 31 mg/kg over 8-24 h, total 68 mg/kg edisilate). The structural model is parameterized in CL/V1/Q/V2 with body weight as a linear covariate on V1 and V2 and a piecewise-linear covariate on CL (linear up to WT50 = 100 kg, constant above) plus a multiplicative effect of concomitant liver-enzyme- inducing drugs (carbamazepine, phenytoin, rifampicin) on CL. IIV uses parameter-specific etas combined with a shared eta common to all four PK parameters (paper text: attributed to clomethiazole adsorption to the infusion tubing) – the joint structure induces a single pairwise correlation among the structural parameters. The paper also reports a proportional-odds sedation-score PD model with a sensitive/non-sensitive mixture component; that PD layer is not encoded here – it requires a NONMEM MIXNUM-style mixture construct that is not naturally expressed in nlmixr2 / rxode2 model files, and the NIH stroke-scale covariate is not yet in the canonical covariate register. See the validation vignette’s Assumptions and deviations section.
Clonazepam pediatric (Yukawa 2002) Steady-state population PK model for clonazepam relative clearance (CL/F) in 137 Japanese pediatric and adult epileptic patients (Yukawa 2002 Table III row 4). CL/F is a body-weight power function with a 3-tier drug-interaction factor for concomitant antiepileptic drugs (monotherapy, +1 AED (CBZ or VPA), +>=2 AEDs).
Clonidine (Kleiber 2017) One-compartment population PK model for intravenous clonidine in critically ill neonates and children on venovenous or venoarterial extracorporeal membrane oxygenation (ECMO) with concomitant continuous venovenous hemofiltration (CVVH). Central compartment only with IV bolus and IV infusion dosing (no absorption). Clearance carries the standard allometric fixed exponent 0.75 on body weight (reference 70 kg), a Hill-type maturation function of postnatal age (steep exponent 3.02 and T50 = 1.13 weeks – ~70% of mature clearance by 10 days of PNA), and a multiplicative diuretic-use effect (CL x 0.659 when any diuretic is active). Volume of distribution carries the standard fixed allometric exponent 1 on body weight and a sigmoidal Emax effect of time-on-ECMO (Emax = +55%, T50 = 51.7 h, Hill exponent 18.5 – effectively a step increase reached by 72 h on ECMO). Residual error is proportional only.
Clopidogrel (Danielak 2017) Joint parent-metabolite population PK model for oral clopidogrel and its active thiol H4 metabolite (the antiplatelet-active diastereomer) in adult Caucasian patients undergoing elective coronarography or percutaneous coronary intervention on chronic clopidogrel 75 mg/day (Danielak 2017). Clopidogrel is described by a one-compartment model with first-order absorption (rate constant ka = source k12) and first-order elimination (CL/F = source CL/F, V/F = source V2/F). The H4 metabolite is described by a one-compartment model with irreversible first-order formation from clopidogrel central at the rate FM * CL/F * (clopidogrel central / Vc) and first-order elimination (CL_h4/F = source Q2/F, V_h4/F = source V3/F). FM was constrained to <= 20% in the source fit because clopidogrel undergoes extensive first-pass metabolism to the inactive carboxylic acid (the competing CES1 pathway accounts for ~85% of the absorbed dose); the final estimate is FM = 4.5%. H4 plasma concentrations were assayed after bromo-3’-methoxyacetophenone derivatisation of the labile thiol and were adjusted to the mass equivalent of clopidogrel, so the parent <-> H4 flux carries 1:1 molar / mass-equivalent stoichiometry. Inter-individual variability is reported on ka, V/F, CL/F, and FM with a covariance between ka and V/F. The only retained covariate is CYP2C19*2 carriage on FM (linear-deviation effect, e_cyp2c19_s2_fm = -0.45); carriers convert 45% less of the absorbed dose to the active H4 metabolite. Bioavailability F was assumed to be unity (typical value 1, not estimated because no IV clopidogrel data exist). Residual error is proportional on the linear-concentration scale for both observed analytes; M3-method handling was used for samples below the quantitation limit (0.25 ng/mL for both clopidogrel and H4).
Clopidogrel (Jung 2024) Joint parent-plus-two-metabolite population PK model for oral clopidogrel with a sequentially-fitted platelet-reactivity (PRU) turnover PD submodel, in healthy Korean male adults stratified by CYP2C19 metabolizer phenotype (Jung 2024). A hepatic first-pass compartment (liver) sits between the depot and the central compartment; all absorbed parent drug is assumed to be metabolized in the liver, so the parent’s only elimination pathway is the hepatic metabolic clearance CLc. Clopidogrel itself is described by a two-compartment model (central + peripheral1) exchanging with the liver at Qc and with the periphery at Qp; the hepatic volume was set equal to the central volume (VH = Vc) because the two were not separately identifiable. The hepatic metabolic flux CLc * (liver / VH) is split three ways by two nested logit-scale fractions: fm1 * fm2 to the active thiol metabolite clopidogrel H4 (one compartment, central_h4), (1 - fm1) * fm2 to the inactive clopidogrel carboxylic acid (two compartments, central_cloca + peripheral1_cloca), and 1 - fm2 to unmeasured other metabolites. Each metabolite formation flux carries a molar-mass ratio scaling factor (1.106 for H4, 0.956 for the carboxylic acid) so that a mass-unit parent flux produces the correct mass-unit metabolite amount. CYP2C19 phenotype is the only retained covariate and acts as an additive shift on the logit scale of both fm1 and fm2, reducing the active-metabolite fraction from 0.120 in extensive metabolizers to 0.071 in intermediate and 0.034 in poor metabolizers. The PD endpoint is the P2Y12 reaction unit (PRU) from the VerifyNow P2Y12 assay, modeled as a turnover pool in which clopidogrel H4 stimulates the fractional turnover rate Kout through a sigmoid Emax function; the drug-free baseline is Kin / Kout = 212.67 PRU. Inter-individual variability is reported on Vc, CLc, Qc (with a correlation between Vc and CLc), fm1, fm2, Vp2, CLm2, Kin, and Emax. Residual error is proportional for all three analytes and additive for PRU. PK and PD were fitted sequentially: all PK parameters were fixed to their final PK-model estimates before the PD parameters were estimated.
Clopidogrel (Pejcic 2024) Joint semi-physiological population PK model for oral clopidogrel (CLO) and its inactive carboxylic acid metabolite (CLO-CA) in healthy Caucasian adults, pooled from two 2-way crossover bioequivalence studies of 150 mg (2 x 75 mg) single oral doses (Pejcic 2024). Absorption is an Erlang-type transit chain (depot -> transit1 -> transit2 -> liver) with a single transit rate constant Ktr = 3 / MTT. The first-pass effect is represented by a hepatic compartment (Vh = 1.5 L/70 kg, liver plasma flow Qh = 50 L/h, both fixed) that receives the whole absorbed dose and also receives the parent returning from the systemic circulation at CLP, so that all parent drug is ultimately metabolised there. Hepatic outflow Qh/Vh is partitioned into three branches whose fractions sum to one: FP to systemic clopidogrel, FiaM to CLO-CA, and FaM to the active thiol CLO-TH (a sink; CLO-TH concentrations were not measured). The fractions use the source’s softmax reparameterisation FiaM = FR1 / (1 + FR1 + FR2), FaM = FR2 / (1 + FR1 + FR2), FP = 1 / (1 + FR1 + FR2) with FaM fixed at 12%, giving FiaM = 87.27% (Study 1) and 86.87% (Study 2). A molecular-weight factor of 0.9565 (307.80 / 321.82) scales the CLO-CA formation flux. Clopidogrel is one-compartment; CLO-CA is two-compartment. Body weight is applied allometrically (exponent 0.75 on clearances, 1 on volumes) to a 70 kg reference. MTT, the generic-product relative bioavailability, FR1, the IIV and IOV magnitudes and the proportional residual error are all study- specific, selected by the STUDY_CLO_BE2 indicator; inter-occasion variability on bioavailability and MTT reflects the crossover design.
Clozapine (Li 2012) One-compartment parent-plus-metabolite population PK model for oral clozapine and its primary active metabolite norclozapine (N-desmethylclozapine) in 162 Chinese adult inpatients (74 male, 88 female; 35.5 +/- 10.6 years) with refractory schizophrenia on maintenance oral clozapine therapy (Li 2012). First-order absorption (Ka fixed at 1.3 1/h from prior rich-data clozapine PK studies) into a single central compartment with first-order elimination; a fixed fraction (KF = 0.66) of the absorbed clozapine dose is converted in the parent central compartment to norclozapine and feeds a separate one-compartment metabolite compartment with its own apparent clearance and apparent volume. Two binary covariates were retained in the final forward-and-backward-selected model: current-smoker status increases apparent clearance of both species (clozapine by 45%, norclozapine by 54.3%), and male sex increases apparent clearance of both species (clozapine by 20.8%, norclozapine by 24.2%); the typical values reported in Table 2 are for the female-nonsmoker reference stratum. A combined additive-plus-proportional residual error model is reported separately for clozapine and norclozapine. The model was internally validated using normalized prediction distribution errors (NPDE).
Colchicine mouse pbpk (Saleh 2023) PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, NMRI). Nine-compartment CNS PBPK model predicting unbound colchicine concentrations in brain extracellular fluid (brain ECF) after a single 1.5 mg/kg intravenous dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug’s physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.14 (taken from ref [20]; Table V). The plasma PK model is the empirical two-compartment model of Table IV (in-house NONMEM fit) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published.
Colistin (Jacobs 2016) Two-state parent-metabolite population PK model for colistimethate sodium (CMS, prodrug) and colistin (active polymyxin) in critically ill ICU patients with acute renal failure requiring intermittent hemodialysis (n=8). One compartment each for CMS and colistin. CMS renal clearance is structurally fixed at 0 (anuric HD population); the estimated CMS clearance is therefore nonrenal (CL_NRCMS). Colistin disposition is parameterised in apparent units (V_col/f_m and CL_col/f_m) because the fraction f_m of nonrenally cleared CMS that becomes colistin is not separately identifiable from plasma data. Hemodialysis clearances of CMS (90 mL/min) and colistin (137 mL/min) are fixed experimental constants from Marchand 2010 (ref 7 of Jacobs 2016) and are gated on/off by the time-varying RRT_HEMODIAL_ACTIVE covariate; PK sampling in the source study was conducted between HD sessions, so RRT_HEMODIAL_ACTIVE = 0 over the model-fit data.
Colistin (Karaiskos 2015) Population PK model for colistimethate sodium (CMS, prodrug) and colistin (active polymyxin formed by in vivo hydrolysis) in critically ill adults after a 9 MU CMS loading dose. CMS distributes through four compartments representing two states of the prodrug (CMS1 = more fully sulfomethylated, CMS2 = partially sulfomethylated derivatives); each state has central and peripheral compartments sharing volumes Vc and Vp but distinct inter-compartmental clearances Q1 and Q2. The same nonrenal clearance CL_NR drives the first-order hydrolysis CMS1 -> CMS2 (in both central and peripheral, with the same rate constant) and CMS2 -> colistin (central only); CMS1 and CMS2 central compartments are additionally cleared by renal clearance proportional to creatinine clearance. Colistin disposition follows a one-compartment model with apparent clearance and volume (CL/fm, V/fm) scaled to the unknown fraction of administered CMS converted to colistin. Measured colistimethate concentration is the sum of CMS1 and CMS2 central concentrations.
Colistin (Lee 2013) One-compartment population PK model of colistin in adult burn-ICU patients receiving colistimethate sodium (CMS) as a 30-minute IV infusion every 12 hours, with first-order CMS-to-colistin conversion (Lee 2013). Apparent CL and Vc of colistin are scaled inversely by the relative fraction of CMS converted to colistin (RFM = 1 - theta4 * (CRCL/128)); the CMS-to-colistin turnover rate constant TR is reduced in patients with clinically-evident peripheral edema (TR = theta3 - theta5 * DIS_EDEMA).
Colistin (Mohamed 2012) Two-compartment population PK model for colistin methanesulfonate (CMS, prodrug) plus a one-compartment apparent model for colistin (formed metabolite) in critically ill patients, with concentration-dependent unbound fraction of colistin A and a semimechanistic Pseudomonas aeruginosa bacterial-kill PKPD (susceptible / resting compartments from Bulitta 2010)
Colistin (Plachouras 2009) Two-compartment population PK model for colistin methanesulfonate (CMS, prodrug) and one-compartment model for the formed colistin (active metabolite) in critically ill adults receiving 3 MU q8h IV CMS for multidrug-resistant Gram-negative infections (Plachouras 2009). Colistin metabolite parameters are apparent values scaled by the unknown fraction (fm) of CMS that forms colistin (CL_col is CL/fm; Vc_col is V/fm).
Colistin (Rodjun 2023) Population PK model for colistimethate sodium (CMS, the inactive prodrug) and formed colistin (the active polymyxin) in critically ill adults, as encoded by Rodjun 2023 for Monte Carlo probability-of-target-attainment simulation against carbapenem-, multidrug- and colistin-resistant Acinetobacter baumannii. CMS follows a two-compartment model with an intravenous 30-min infusion; its total clearance is the sum of a renal component proportional to creatinine clearance and a non-renal component. The entire non-renal CMS clearance is the formation pathway for colistin, which then follows a one-compartment model whose total clearance is likewise the sum of a CrCL-proportional renal component and a non-renal component. Unbound colistin concentration is returned as Ccu_col using the reported unbound fraction of 0.49, because the paper’s PK/PD target is the unbound AUC ratio fAUC/MIC >= 7.4.
Concizumab (Yuan 2019) QSP. Systems PK/PD model for concizumab (humanized anti-TFPI IgG4) describing binding to both membrane-bound TFPI (mTFPI; non-linear clearance via receptor-mediated endocytosis) and soluble TFPI (sTFPI; linear clearance via FcRn-recycled pinocytosis) in a minimal physiologically-based PK framework with two nested endothelial endosome compartments. Parameter values for 70 kg adult humans (Yuan 2019 Tables 1-2); the paper also tabulates monkey and rabbit parameter sets.
ConestatAlfa (Farrell 2013) One-compartment population PK model with Michaelis-Menten elimination for intravenous recombinant human C1 inhibitor (rhC1INH; conestat alfa; Ruconest) in healthy volunteers and adolescent / adult patients with hereditary angioedema (Farrell 2013). Total functional plasma C1INH is modelled as the sum of an estimated endogenous baseline (separate baselines for healthy volunteers and HAE patients) plus exogenously administered rhC1INH, with the endogenous production rate derived from the Michaelis-Menten elimination at baseline so the no-dose steady state is preserved. Allometric power scaling of central volume on body weight (exponent 0.612).
Corifollitropin alfa (Zandvliet 2016) One-compartment subcutaneous population pharmacokinetic model for corifollitropin alfa (a long-acting recombinant gonadotrophin) in women undergoing controlled ovarian stimulation (Zandvliet 2016). Pooled analysis of 2557 evaluable women from five phase II and III trials (single SC doses of 60-180 ug). Corifollitropin alfa is absorbed first-order from a subcutaneous depot into a one- compartment central pool with first-order elimination; body weight is the major covariate, with allometric (WT/60)^exponent power effects on apparent clearance and apparent volume. Apparent bioavailability is modulated by body-mass index, race (Asian and Black indicators vs Caucasian reference), and remains anchored at F = 1. The model jointly describes total FSH immunoreactivity by adding an endogenous follicle stimulating hormone (FSH) compartment whose pre-dose steady-state baseline is FSHbaseline and whose synthesis is set to zero from corifollitropin administration onwards (per the paper’s structural model). Total FSH immunoreactivity (IU/L) equals SCALE * corifollitropin concentration (ng/mL) plus the endogenous FSH compartment value, where SCALE is fixed at 6.11 IU/L per ng/mL from an upstream analysis. Trial- specific multiplicative effects on the FSH immunoreactivity prediction (1.26 for trial 06029 and 1.12 for trial 38825) are exposed via binary study indicators that default to zero for general simulation use.
Cotadutide qsp (Bosch 2024) QSP. 4GI quantitative systems pharmacology model (glucose, insulin, GLP-1, glucagon, GIP) coupled to a one-compartment first-order- absorption cotadutide PK model in adults with type 2 diabetes mellitus (Bosch 2024). Cotadutide is a dual GLP-1/glucagon receptor agonist; in vivo EC50s for cotadutide on each receptor are derived from the in vitro EC50 ratio vs the endogenous ligand (Eq 1). The drug’s free-fraction-corrected central concentration drives four saturable Emax effects on the system: (1) stimulation of glucose- dependent insulin secretion via GLP-1R, (2) inhibition of meal- glucose absorption via GLP-1R, (3) inhibition of glucagon production via GLP-1R, and (4) stimulation of glucose production via GCGR. A fifth Emax inhibits endogenous active GLP-1 production (Eq 3). The placebo arm’s lifestyle-change effect on fasting plasma glucose is modelled as an inverse Bateman attenuation of endogenous glucose production (Eq 2). Cotadutide PK structure and typical values are fixed from the upstream popPK analysis of Guan et al. 2022 (KA=0.343 1/h, CL=1.04 L/h, V=18.7 L). All 4GI system- specific disposition and effect parameters are fixed from the upstream 4GI model of Bosch et al. 2022; meal-effect, baseline, lifestyle and EMAX_5/EC50_5S parameters were re-estimated against the cotadutide MAD/Ph2a dataset (NCT02548585; n=51, T2DM). Five outputs: plasma glucose (mmol/L), insulin (pmol/L), GLP-1 (pmol/L), glucagon (pmol/L) and GIP (pmol/L), each with proportional residual error. Individual fasting plasma glucose enters via the FPG covariate; meal glucose enters as dosing events on the glucose-gut compartment. Defaults are T2DM; healthy-volunteer parameter set from Bosch 2022 is given in source-trace comments. No IIV is encoded (sequential model fit with individual PK / glucose- baseline inputs from Guan 2022 and the observed dataset).
Crisantaspase (Sassen 2017) Two-compartment population PK model for intravenous Erwinia asparaginase (crisantaspase; Erwinase) in pediatric acute lymphoblastic leukemia patients, with allometric scaling on clearance and volumes and a higher first-month clearance (Sassen 2017).
CrisprCas9 human (Desai 2024) QSP. Translational quantitative systems pharmacology platform for in vivo CRISPR-Cas9 gene editing delivered by lipid nanoparticle (LNP), covering the whole-body disposition of all three components of the therapy - the LNP delivery vehicle, the single guide RNA (sgRNA) and the Cas9 messenger RNA (mRNA) - plus intracellular Cas9 translation and ribonucleoprotein (RNP) assembly (Desai 2024). The PK layer is a minimal-PBPK structure with plasma, lymph, a three-region liver (vascular / endosomal / interstitial-cellular), kidney and a lumped remainder tissue; the kidney compartment is present for LNP and sgRNA but absent for mRNA, which is too large to be renally cleared. LNP additionally undergoes opsonin bio-corona formation (kass / kdis), phagocytosis into the mononuclear phagocyte system (kint) and LDL-receptor-mediated endocytosis (kon,LNP / koff,LNP) with receptor recycling. This is the clinical arm of the platform (step 4 of the paper’s four-step workflow), scaled to human physiology and re-estimated against NTLA-2001 LNP plasma PK and serum transthyretin from the first-in-human study (0.1-1 mg/kg total RNA by 2 h IV infusion). The pharmacodynamic layer is an indirect-response model in which the liver ribonucleoprotein concentration inhibits TTR production; unlike the NHP arm the human TTR model carries no Hill coefficient. Sibling extractions: Desai_2024_crisprCas9_mouse and Desai_2024_crisprCas9_nhp.
CrisprCas9 mouse (Desai 2024) Preclinical (mouse). QSP. Translational quantitative systems pharmacology platform for in vivo CRISPR-Cas9 gene editing delivered by lipid nanoparticle (LNP), covering the whole-body disposition of all three components of the therapy - the LNP delivery vehicle, the single guide RNA (sgRNA) and the Cas9 messenger RNA (mRNA) - plus intracellular Cas9 translation and ribonucleoprotein (RNP) assembly (Desai 2024). The PK layer is a minimal-PBPK structure with plasma, lymph, a three-region liver (vascular / endosomal / interstitial-cellular), kidney and a lumped remainder tissue; the kidney compartment is present for LNP and sgRNA but absent for mRNA, which is too large to be renally cleared. LNP additionally undergoes opsonin bio-corona formation (kass / kdis), phagocytosis into the mononuclear phagocyte system (kint) and LDL-receptor-mediated endocytosis (kon,LNP / koff,LNP) with receptor recycling. This is the mouse arm of the platform (step 2 of the paper’s four-step workflow): it was fitted to plasma sgRNA and mRNA concentrations after a single 2 mg/kg total RNA IV bolus, and carries no pharmacodynamic layer because no biomarker was measured in mice. The rate of exocytosis and the mRNA degradation rate were estimated here; the rate of endocytosis was fixed from the NHP estimate and scaled with an allometric exponent of -0.25. Sibling extractions: Desai_2024_crisprCas9_nhp and Desai_2024_crisprCas9_human.
CrisprCas9 nhp (Desai 2024) Preclinical (cynomolgus monkey). QSP. Translational quantitative systems pharmacology platform for in vivo CRISPR-Cas9 gene editing delivered by lipid nanoparticle (LNP), covering the whole-body disposition of all three components of the therapy - the LNP delivery vehicle, the single guide RNA (sgRNA) and the Cas9 messenger RNA (mRNA) - plus intracellular Cas9 translation and ribonucleoprotein (RNP) assembly (Desai 2024). The PK layer is a minimal-PBPK structure with plasma, lymph, a three-region liver (vascular / endosomal / interstitial-cellular), kidney and a lumped remainder tissue; the kidney compartment is present for LNP and sgRNA but absent for mRNA, which is too large to be renally cleared. LNP additionally undergoes opsonin bio-corona formation (kass / kdis), phagocytosis into the mononuclear phagocyte system (kint) and LDL-receptor-mediated endocytosis (kon,LNP / koff,LNP) with receptor recycling. Two pharmacodynamic case studies are attached to the liver RNP concentration: NTLA-2001 (transthyretin amyloidosis) via a sigmoidal indirect-response model inhibiting TTR production, and VERVE-101 (LDL-cholesterol lowering) via a Friberg-type transit / feedback model on serum PCSK9 feeding a precursor-dependent LDL-cholesterol model. Estimated against cynomolgus monkey LNP plasma PK (1-3 mg/kg total RNA), serum TTR (1.5-6 mg/kg) and serum PCSK9 / LDL-cholesterol (0.75-1.5 mg/kg), all by 2 h IV infusion. Sibling extractions: Desai_2024_crisprCas9_mouse and Desai_2024_crisprCas9_human.
Crizotinib mouse (Yamazaki 2008) Preclinical (athymic mouse; GTL16 gastric carcinoma or U87MG glioblastoma xenograft). Integrated PK + cMet phosphorylation (effect-compartment link model) + exponential tumor-growth-inhibition (TGI) model for orally administered crizotinib (PF02341066), an ATP-competitive cMet receptor tyrosine kinase inhibitor. PK is one-compartment first-order absorption with a fixed 0.8 h lag, fitted by naive-pooled analysis (dose-group-specific estimates due to nonlinear kinetics; the encoded set is Study 2 at 50 mg/kg). The cMet phosphorylation response is the Sheiner 1979 link model with E0, Emax, and Hill coefficient all fixed at 1 (Imax 1/(1 + Ce/EC50) form). The tumor-growth model is exponential, with the growth rate inhibited by the plasma concentration via a sigmoidal Imax 1/(1 + Cc/EC50_tumor) function (Emax fixed at 1; the saturable tumor-volume capacity term TG50 was rejected by the authors as TG50 >> Tmax). Default TGI parameters reproduce the GTL16 fit; the U87MG variant (kin_tumor=0.0134, kout_tumor=0.00236, EC50_tumor=94.1 ng/mL) is documented in population$notes and demonstrated in the validation vignette.
CyaaE7 (ParraGuillen 2013) Preclinical (mouse, female C57BL/6 with subcutaneous TC1 tumor expressing HPV E7). Semi-mechanistic K-PD tumor-growth-dynamics model of single-dose CyaA-E7 cancer vaccine: a virtual vaccine compartment feeds a two-compartment transit chain to a vaccine-elicited inhibitory signal SVAC that reduces tumor size via a second-order k3 * SVAC * tumor_size term, inhibited by a Hill-function regulator REG driven by tumor size; a binary mixture covariate MIX_VAC_RELAPSE gates the SVAC degradation rate (k2 = 0 for cure, k2 = k1 for relapse).
Cyclophosphamide mouse (Campagne 2019) Preclinical (mouse). Plasma and brain/tumor extracellular-fluid (ECF) population PK model for cyclophosphamide (CTX) and its sequential metabolites 4-hydroxy-cyclophosphamide (4OH-CTX) and carboxyethylphosphoramide mustard (CEPM) in female CD-1 nude mice (non-tumor-bearing and orthotopic Group 3 medulloblastoma G3MB), following a single 130 mg/kg intraperitoneal dose of cyclophosphamide (Campagne 2019). Three sequential two-compartment plasma sub-models are linked by full (Fm = 1) conversion CTX -> 4OH-CTX -> CEPM (so reported CL and V for the two metabolites are apparent CL/F and V/F); each compound additionally has a one-compartment brain/tumor ECF sub-model linked to its plasma central via influx (CLin) and efflux (CLef) clearances driven by the unbound plasma concentration FU x Cp. ECF volume fixed at 0.001 L/kg (Stewart 2010, ref 26 of source). No covariate effects retained; pooled fit across non-tumor-bearing and G3MB mice.
Cyclophosphamide mouse pbpk (Saleh 2023) PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, CD1 nude). Nine-compartment CNS PBPK model predicting unbound cyclophosphamide concentrations in brain extracellular fluid (brain ECF) after a single 130 mg/kg intraperitoneal dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug’s physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.339 (estimated by the authors; Table V). The plasma PK model is the empirical two-compartment model of Table IV (ref [7]) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published.
Cyclosporin (Debord 2001) Two-compartment population PK model for oral cyclosporin microemulsion (Neoral) in stable renal transplant recipients (Debord 2001), with a gamma-distribution absorption (Savic 2007 analytical transit-compartment form) feeding the central compartment directly, F fixed to 1, and population typical values derived from the means of the 21 individually-fitted patients in Table I of the paper.
Cyclosporine (Philippe 2015) Pediatric PK-PD-time-to-event model for oral cyclosporine in children with severe aplastic anemia (Philippe 2015). PK is a two-compartment model with first-order absorption, lag time, and linear elimination; absorption parameters (F, Tlag, ka) are fixed from the literature, and V1, V2, Cl, Q are allometrically scaled to body weight (reference 34 kg; fixed exponents 0.75 on clearance and 1 on volume). The pharmacodynamic interface model (Eq. 5) describes an effective concentration Ce driven by the predicted trough concentration Ctrough, with production active only when Ctrough lies inside an effective range (lower bound gamma1 = 87 ng/mL, upper bound gamma2 = 120 ng/mL) and first-order elimination at rate alpha. The instantaneous hazard of neutrophil response (Eq. 6) is lambda(t) = lambda0 * (1 + slope * Ce); cumhaz and sur are exposed as derived outputs. In this implementation the predicted Cc (multiplied by 1000 to convert mg/L to ng/mL) is used as the Ctrough input to the interface model; see vignette Assumptions and deviations for the full justification.
Cysteamine (Belldina 2003) Two-compartment population PK model with first-order oral absorption and an absorption lag, sequentially linked to a one-compartment effect-site PD model with fractional inhibitory Emax (Hill = 1) for white-blood-cell cystine content reduction by cysteamine in 11 paediatric and young-adult patients (age 3-15 y, weight 14.3-60.2 kg) with nephropathic cystinosis at steady state on cysteamine bitartrate (Cystagon) approximately every 6 hours. PK and PD parameters in the source paper were estimated as individual NONMEM fits per subject and summarised as arithmetic mean / geometric mean / median / min / max across the 11 patients (Tables 2 and 3); this package encodes the arithmetic means as the typical values, with linear allometric weight scaling fixed at exponent 1.0 to reflect the paper’s per-kg parameterisation of all clearance and volume terms. Dose is in mg cysteamine bitartrate salt (MW 227.24 g/mol); the model converts internally to plasma cysteamine in micromolar (free-base moiety, MW 77.15 g/mol, the measured analyte). PD output cystine is white-blood-cell cystine content in nmol cystine per mg protein.
Dabigatran (Liesenfeld 2013) Two-compartment population PK model for oral dabigatran (after dabigatran etexilate prodrug) in seven end-stage renal disease (ESRD) subjects undergoing intermittent hemodialysis, with first-order absorption, absorption lag, an apparent total body clearance (renal + non-renal), and an apparent dialysis clearance described by the Michaels equation as a function of blood and dialysate flow rates and a hemodialyzer mass transfer-area coefficient (Liesenfeld 2013).
Dabigatran aPTT (Liesenfeld 2006) Pharmacodynamic model for the prolongation of activated partial thromboplastin time (aPTT) by dabigatran in orthopaedic surgery patients receiving oral dabigatran etexilate after total hip replacement (Liesenfeld 2006 BISTRO I PK-PD analysis). The concentration-aPTT relationship combines a linear and an Emax model; the baseline aPTT and the maximum nonlinear effect Emax both decline with time since surgery via a proportional inhibitory Emax form sharing a single ET50. Covariate analysis retained no demographic, comedication, or laboratory variables. The 2006 paper does not develop a PK model; the PK component embedded here is the Liesenfeld 2013 two-compartment dabigatran disposition with all PK thetas fixed so the model is self-contained for simulation. The 2013 PK was fit in end-stage renal-disease subjects and will overestimate dabigatran exposure for the 2006 BISTRO I orthopaedic-surgery population; users targeting BISTRO I-style scenarios should override the PK thetas or supply observed concentrations to the PD layer.
Dabigatran ECT (Liesenfeld 2006) Pharmacodynamic model for the prolongation of ecarin clotting time (ECT) by dabigatran in orthopaedic surgery patients receiving oral dabigatran etexilate after total hip replacement (Liesenfeld 2006 BISTRO I PK-PD analysis). The concentration-ECT relationship is a single linear function whose slope decays exponentially from an initial value SLO0 to a final value SLO_F with rate constant KM; the baseline ECT also declines with time-since-surgery via a proportional inhibitory Emax form. Covariate analysis retained no demographic, comedication, or laboratory variables. The 2006 paper does not develop a PK model; the PK component embedded here is the Liesenfeld 2013 two-compartment dabigatran disposition with all PK thetas fixed so the model is self-contained for simulation. The 2013 PK was fit in end-stage renal-disease subjects and will overestimate dabigatran exposure for the 2006 BISTRO I orthopaedic- surgery population; users targeting BISTRO I-style scenarios should override the PK thetas or supply observed concentrations to the PD layer.
Daclizumab (Othman 2014) Two-compartment population PK model with first-order subcutaneous absorption and lag time for daclizumab high-yield process (HYP) in healthy volunteers (Othman 2014)
Daclizumab cd25 (Diao 2016) Kinetic-binding PK/PD model of CD25 receptor occupancy on peripheral CD4+ T cells following subcutaneous daclizumab high-yield process (HYP) in adults with relapsing-remitting multiple sclerosis (Diao 2016). The receptor-occupancy state occ_cd25 obeys dOcc/dt = konCc(1 - Occ) - koffOcc; the PD output is the percentage of CD4+ T cells staining positive for unoccupied CD25, computed as E0 (1 - occ_cd25). This is a REPARAMETERISATION of the paper’s published sigmoidal Emax (Equation 1) with its two-parameter-set (saturation-phase / desaturation-phase) Hill function – kon and koff are calibrated to reproduce the paper’s phenomenology (rapid saturation within 7 h after first 150 mg SC dose per Figure 1A; return of unoccupied CD25 to baseline in ~24 weeks after last steady-state dose per Figure 1B) rather than transcribed from a paper table. See the companion vignette Assumptions and deviations for the calibration rationale. The PK backbone is inherited verbatim from Othman 2014.
Daclizumab cd56bright (Diao 2016) Indirect-response PK/PD model of CD56 bright natural killer (NK) cell expansion following subcutaneous daclizumab high-yield process (HYP) in adults with relapsing-remitting multiple sclerosis (Diao 2016). Daclizumab HYP serum concentration stimulates the zero-order production rate (Kin) of CD56 bright NK cells (% of all lymphocytes) via a saturable Smax function; first-order elimination rate Kout is fixed by the median baseline. The PK backbone is the two-compartment, first-order SC absorption + lag model from Othman 2014 (file inst/modeldb/specificDrugs/Othman_2014_daclizumab.R), copied verbatim with weight-based allometric scaling.
Daclizumab treg (Diao 2016) Sigmoidal Emax PK/PD model of regulatory T cell (Treg) reduction following subcutaneous daclizumab high-yield process (HYP) in adults with relapsing-remitting multiple sclerosis (Diao 2016). The PD output is the percentage of Treg (CD4+ CD127low/- Foxp3+) among all CD4+ T cells; daclizumab HYP serum concentration drives a maximum 60% reduction via a sigmoidal Emax function. The PK backbone is the two-compartment, first-order SC absorption + lag model from Othman 2014 (file inst/modeldb/specificDrugs/Othman_2014_daclizumab.R), copied verbatim with weight-based allometric scaling.
Dactinomycin (Mondick 2006) Three-compartment intravenous population PK model for actinomycin-D (dactinomycin) in 33 pediatric and young-adult patients (1.58-20.3 years) with Wilms’ tumor or rhabdomyosarcoma. All disposition parameters are allometrically scaled by total body weight, normalized to a reference weight of 70 kg, with theory-based fixed exponents (0.75 on clearances, 1.0 on volumes; not explicitly stated in the abstract). Inter-individual variability was reported only for V1 (54.4% CV) and CL (57.2% CV); residual error and the remaining IIV terms (V2, V3, Q2, Q3) were not reported in the source conference abstract and are encoded as fixed(0). Mondick 2006 (PAGE 15 Abstr 938).
Dalteparin (Schoemaker 1996) One-compartment population PK/PD model for the low molecular weight heparin dalteparine (trade name Fragmin) in healthy volunteers, fitted simultaneously to intravenous and subcutaneous administration data (Schoemaker & Cohen 1996, Example 3 / Table 4). The kinetic sub-model is a one-compartment IV bolus / first-order SC absorption disposition with an estimated constant basal anti-Xa activity (extending the Schoemaker 1996 Example 2 enoxaparine model with a depot compartment and bioavailability). The pharmacodynamic sub-model links anti-Xa activity (Cc) to activated partial thromboplastin time (APTT) through an exponential concentration-effect relationship parameterised by I10 (the anti-Xa activity increment required to produce a 10% increase in APTT). Common kinetic parameters are shared between IV and SC routes within each subject; only F (bioavailability) and ka (absorption rate) differ between routes. Validation of this model and the companion enoxaparine PK model share a single vignette.
Danoprevir (Canini 2015) Combined pharmacokinetic / viral-kinetic (PK/VK) model for the HCV NS3/4A protease inhibitor danoprevir in adults with chronic hepatitis C (Canini 2015). PK is a two-compartment disposition (central = plasma, peripheral1 = tissue including liver) with zero-order absorption of duration Tk0 into central after a post-dose lag Tlag, followed by first-order elimination. The viral-kinetic layer follows the standard Neumann 1998 model with target cells assumed to remain constant (only the productively-infected cell pool infected and the free-virion pool virus are carried as ODE states). Drug effectiveness e(t) = C1 / (EC50 + C1) is a Hill-Emax function with Emax fixed at 1 and Hill coefficient fixed at 1 (the Canini 2015 model-selection winner over h = 2 and h = 3 by AICc/BIC); e(t) blocks virion production so d/dt(virus) = (1 - e) * p * infected - c * virus. Because Canini 2015 does not estimate the virion production rate p or the infection rate beta separately, the packaged model fixes p = 1 as a mathematical scaling (the identifiability constraint p * beta_T = d * c at the pre-treatment steady state fully determines beta_T = d * c and I(0) = c * V0). All 10 estimated typical values from Canini 2015 Table 2 (Tlag, Tk0, V1, ke, k12, k21, V0, c, d, EC50) and all 10 IIVs are carried; cohort-specific V0 offsets (Table 2 cohorts 2-5) and the screened HCV genotype covariate are documented in vignette Errata but not applied as covariates in the packaged model (see notes in population and in the vignette Assumptions section). PK is in per-hour units to match the paper’s Table 2 reporting; the viral-kinetic /day-scale rate constants c and d are converted inline to /h so the integrated system runs on a single hour-scale time axis.
Dapagliflozin (vanderWalt 2013) Semi-mechanistic joint parent-metabolite population PK model for dapagliflozin and its inactive UGT1A9 glucuronide metabolite dapagliflozin 3-O-glucuronide (D3OG, identified as M15 in chromatography) in healthy adults, T2DM subjects with normal or impaired renal function, and patients with hepatic impairment (van der Walt 2013). Parent: 2-compartment disposition with first-order absorption fed by a Savic 2007 transit-compartment chain (continuous N estimated alongside MTT) and a logit bioavailability anchor; three parallel parent elimination pathways are estimated separately as renal excretion of unchanged dapagliflozin (CLP_renal, proportional to baseline creatinine clearance), metabolic formation of D3OG (CLP_M15), and metabolic clearance to unmeasured metabolites (CLP_other, allometrically scaled like CLP_M15). Metabolite: 1-compartment with renal elimination CLM proportional to creatinine clearance. Plasma observations only are emulated here – the source paper also fitted urine dapagliflozin and D3OG concentrations simultaneously with a replicate residual-error structure; see the validation vignette for the urine and replicate-residual deviations. Covariates: creatinine clearance (CRCL; IBW-corrected, mL/min) on CLP_M15, CLP_renal, and CLM; AGE on CLP_other; Child-Pugh Class C (HEPIMP_SEV) on CLP_M15 and V2M; Child-Pugh Class B or C (HEPIMP_MODSEV) on V3P and CLM; female sex (SEXF) on total CLP and on CLM; allometric WT scaling on CLP_M15, CLP_other, V2P, V3P, V2M.
Dapsone (Gatti 1996) One-compartment population PK model with first-order oral absorption and first-order elimination for dapsone 100 mg twice weekly oral Pneumocystis carinii pneumonia prophylaxis in 53 HIV-infected adults (Gatti 1996). Apparent clearance CL/F and apparent central volume V/F are scaled multiplicatively by concomitant rifampin co-administration (shared 69.6% increase on both parameters, reflecting a first-pass / bioavailability effect). Apparent absorption rate constant Ka is scaled multiplicatively by total serum bilirubin (per-mg/dL fractional decrease). IIV on CL/F (35% CV) and Ka (85% CV); V/F inter-individual variability was found non-significant after covariate inclusion and dropped from the final model. Residual-error magnitudes were not reported in the publication; propSd and addSd are FIXED at 0 in this packaged model so users must supply their own residual error to run any stochastic VPC – see the validation vignette’s Errata section.
Dapsone (Hall 2017) One-compartment population PK model with first-order oral absorption for dapsone in healthy US adults across a wide weight range; covariate effects on Ka, CL, and Vc are encoded via the published MARS piecewise-linear basis functions of weight, age, and blood urea nitrogen (Hall 2017).
Daptomycin (Garonzik 2016) In vitro (Staphylococcus aureus USA300, methicillin-resistant CA-MRSA reference strain). Mechanism-based mathematical pharmacodynamic (MBM) model of daptomycin time-kill activity in supplemented Mueller-Hinton broth with 0%, 10%, 30%, 50%, or 70% v/v heat-inactivated human serum. The bacterial population is split into three subpopulations of decreasing daptomycin susceptibility (susceptible, intermediate, resistant), each described by two states (state 1 vegetative, state 2 replicating; six bacterial compartments total). Replication of state 2 cells back into state 1 is gated by a successful-replication probability (REP = 2 x Plateau, with Plateau saturating at a maximum CFU/mL CFUm), and the vegetative-to-replicating transition k12 is modulated by an exponential lag-phase term (Eq 3) and a saturable carrying-capacity term (Eq 7) parameterised by Imax_k12 and IC50_k12. Daptomycin acts on each subpopulation via two mechanisms: a Hill-type stimulation of the probability of death (STI; reduces successful replication via IREP = 1 - STI) and a Hill-type direct killing of bacteria (Kill); the relative balance of the two is the dominant pharmacodynamic feature, with SC50 (0.05 mg/L) much lower than KC50 (4.8 mg/L). The intermediate and resistant subpopulations share the same SC50 and KC50 but have reduced Smax and Kmax (Smax_r and Kmax_r fixed to 0) and the resistant subpopulation has a slower vegetative-to-replicating transition (FR_K12r = 0.0442). Protein binding by human serum is encoded as an ‘active fraction’ factive(HS) multiplying the total static daptomycin concentration to give an effective drug concentration DAP_EF; the active fraction takes five experimental levels (factive = 1 at 0% HS, then 0.346, 0.284, 0.239, 0.252 at 10%, 30%, 50%, 70% HS). The model is in-vitro PD only – there is no human PK component; daptomycin is dosed once at t = 0 into the dap compartment and is chemically stable in the medium for the 24-h experiment. Random effects (eta) are NOT present: the paper reports replicate-level experimental fits with additive plus small-count Poisson residual error on log10 CFU/mL.
Daratumumab (Xu 2020) Two-compartment population PK model for intravenous daratumumab (anti-CD38 IgG1k) in adults with multiple myeloma, with parallel linear and Michaelis-Menten eliminations from the central compartment. The maximum velocity of the saturable (target-mediated) elimination decays mono-exponentially from its baseline value at first-order rate KDES, mimicking depletion of the CD38 target over weekly 16 mg/kg therapy (Xu 2020 MMY1001 D-Kd / D-KRd cohorts).
Daratumumab qsstmdd (Li 2021) Two-compartment semi-mechanistic target-mediated drug-disposition (TMDD) population PK model for IV daratumumab (anti-CD38 IgG1) in adults with multiple myeloma, with parallel non-specific linear clearance and CD38-mediated saturable clearance under the quasi-steady-state (QSS) approximation of Gibiansky 2008. The TMDD/QSS form supersedes an earlier empirical Michaelis-Menten parameterisation with time-dependent Vmax: receptor (CD38) turnover and complex internalisation reproduce mechanistically the observed Vmax time-dependency. PAGE 29 (2021) abstract II-52 by Li, Perez Ruixo, Zhou, Perez Ruixo, and Dosne (Janssen R and D, Beerse). Distinct from Xu 2020 daratumumab, which uses the empirical 2-cmt parallel-linear / time-dependent Vmax form.
Darbepoetin (Takama 2007) Two-compartment intravenous population PK model for darbepoetin alfa in Japanese adult haemodialysis (HD) and peritoneal dialysis (PD) patients with an additive endogenous erythropoietin baseline concentration (Takama 2007). Body weight enters as a linear-deviation effect (centred on 54 kg) on clearance and central volume; peritoneal-dialysis modality adds a +17% multiplicative increment to central volume relative to the HD reference.
Darbepoetin alfa (Agoram 2006) Two-compartment population PK model with first-order subcutaneous absorption for darbepoetin alfa in healthy adult subjects (Agoram 2006). Both IV and SC routes are supported. SC bioavailability is a linear function of the SC dose amount (in ug). Body weight modifies clearance and central volume via a normalized power model (reference 70 kg); subject age modifies the absorption rate constant via a normalized power model (reference 47 years, the development-cohort mean). Total measured serum concentration is the sum of the simulated darbepoetin alfa and an individual-specific endogenous-erythropoietin (eEPO) constant that the ELISA assay cross-detects. Exponential (log-normal) residual error.
Darunavir (Brochot 2015) Two-compartment population PK model for ritonavir-boosted darunavir with alpha-1 acid glycoprotein (AAG)-dependent apparent clearance and allometric weight scaling, in HIV-1-infected pediatric (>=3 to <18 years) and adult patients (Brochot 2015).
Dasabuvir (Mensing 2017) Two-compartment population PK model for oral dasabuvir in HCV genotype-1 infected adults receiving the 3D regimen (Mensing 2017). First-order absorption, linear elimination, combined proportional + additive residual error, IIV on CL/F only. The author’s final model retained cirrhosis, gender, creatinine clearance, and body weight as significant covariates on CL/F (and age, body weight on Vc/F and Vp/F), but the paper does not publish point estimates for these covariate coefficients (only graphical exposure-ratio forest plots in Figure 2); the implemented model is the structural typical-value model with covariate coefficients omitted (documented in covariatesDataExcluded).
Datopotamab (Hong 2025) Coupled population PK model for datopotamab deruxtecan (Dato-DXd, anti-TROP2 antibody-drug conjugate) and its released payload DXd in adults with advanced solid tumors (Hong 2025). Dato-DXd disposition is a two-compartment model with parallel linear (CL_lin) and Michaelis-Menten (Vmax / Km) elimination from the central compartment. DXd is a one-compartment model whose formation rate equals the total Dato-DXd elimination rate (linear + nonlinear) scaled by the molecular-weight ratio (493.5 / 150000) and a time-and-cycle-dependent drug-to-antibody ratio DAR(tad, CYCLE) = 4 * (0.25 + 0.75 * exp(-beta * tad)) * (1 if CYCLE = 1 else Factor1). Body weight is included as a mechanistic covariate with a fixed allometric exponent of 0.75 on Dato-DXd linear clearance and estimated exponents on Dato-DXd volumes (paper Eq. 8-10) and on DXd CL/Vc (Eq. 14-15).
Daunorubicin (Bogason 2011) Two-compartment population PK model for daunorubicin (DNR) in adults with acute myeloid leukaemia, with baseline white blood cell count as a covariate on central volume of distribution (Bogason 2011)
Daunorubicin (Varatharajan 2016) Population PK model for IV daunorubicin (Dnr) and its primary carbonyl-reductase metabolite daunorubicinol (DOL) in adult de novo acute myeloid leukaemia (AML) patients (Varatharajan 2016). Each component (parent and metabolite) is described by an independent two-compartment disposition parameterised on apparent clearance, central volume, and the inter-compartmental rate constants K12 and K21. Daunorubicin is converted to daunorubicinol via parent elimination (the model assumes the fraction metabolised fm = 1, so the published DOL CL and V are ‘apparent’ values that absorb fm). No covariates were retained in the final structural model; demographic / pharmacogenetic associations in the paper are reported on post hoc empirical-Bayes estimates rather than as fixed-effects covariate parameters.
Daunorubicin liposomal (Hempel 2003) One-compartment IV-infusion population PK model for total daunorubicin (free plus liposome-encapsulated) following liposomal daunorubicin (Daunoxome) in paediatric and adolescent oncology patients (Hempel 2003). Clearance and volume of distribution scale linearly with total body weight (CL = theta_CL * WT; V = theta_V * WT, i.e. the source paper’s per-kg parameterisation with allometric exponent fixed to 1 and no reference-weight normalisation). The final model (Table 2 model 15) retains inter-individual variability on CL (51% CV) and V (27% CV), inter-occasion variability on CL (16.7% CV) – documented but NOT encoded structurally here, per the Andrews 2017 / Brooks 2021 nlmixr2lib precedent for IOV without an operational occasion column – and a proportional residual error (22%). Distinct from Varatharajan_2016_daunorubicin (free daunorubicin + daunorubicinol metabolite in adult AML).
Decitabine (Han 2015) Two-compartment IV population pharmacokinetic model coupled with two parallel Friberg-style myelosuppression PD chains (absolute neutrophil count, ANC, and platelet count, PC) for decitabine post-transplant maintenance in adult Korean patients with higher-risk myelodysplastic syndrome or secondary acute myeloid leukemia (Han 2015). The platelet feedback baseline rises asymptotically over cycles per the paper’s IMP / IMK extension (BASE_P_t = BASE_P + IMP * (1 - exp(-IMK * t))); the neutrophil chain uses a time-invariant baseline. PK parameters are body-surface-area-normalized (per m^2): doses must be supplied in mg/m^2 and central-compartment concentrations are returned in mg/L (= ug/mL = 1000 ng/mL). PD outputs ANC and PLT are in 10^9 cells/L.
Deferiprone (Bellanti 2014) One-compartment population PK model for the oral iron chelator deferiprone in healthy adult subjects, with first-order absorption, absorption lag time, and a binary sex effect on the apparent volume of distribution (Bellanti 2014).
Deferiprone (Bellanti 2017) One-compartment population PK model for the oral iron chelator deferiprone in paediatric patients aged <6 years with transfusion-dependent haemoglobinopathies, with first-order absorption and first-order elimination and fixed allometric scaling of clearance and volume on body weight (Bellanti 2017).
Deferoxamine (Bellanti 2015) Indirect-response disease model for serum ferritin in chronic transfusional iron overload (beta-thalassaemia major) with proportional deferoxamine effect on ferritin degradation rate. Two-compartment 8-h SC-infusion deferoxamine PK (literature-derived) enters as a time-varying steady-state concentration covariate; the ferritin compartment captures the baseline turnover (Kin, Kout), the disease-status-modulated transfusion-driven production (CRT), and the chelator effect (1 + DFO) on Kout.
Degarelix (Tornoe 2006) Population PK/PD model of the hypothalamic-pituitary-gonadal (HPG) axis after repeated subcutaneous (s.c.) injections of the GnRH receptor blocker degarelix in prostate-cancer patients. PK is a two-compartment disposition model with two parallel first-order absorption routes from a self-forming s.c. depot: a rapid release (fraction Fr via ka,fast) and a prolonged slow release ((1 - Fr) via ka,slow). The packaged values correspond to the 40 mg/mL dose-concentration arm of Tornoe 2007 Table 3 (Fr_40, F_40 and t_1/2,slow,40); the 20 and 60 mg/mL alternatives are tabulated in the validation vignette. PD is a four-state HPG-axis feedback model (feedback compartment F, LH pool P, LH, testosterone Te) with sigmoidal Imax inhibition of LH pool release by degarelix and a positive interaction (F) from the feedback compartment on LH synthesis and release; testosterone secretion is stimulated by LH via a sigmoidal Emax model. ke_LH, ke_F, lambda, LH_base and Te_base are the degarelix-study-specific values from Table 4.
Desmopressin (Agerso 2004) Three-compartment population PK model for intravenous desmopressin with simultaneous plasma and urinary-amount outputs. Systemic clearance is split into renal and non-renal components, each modulated linearly by creatinine clearance (CRCL), in healthy subjects and patients with varying degrees of renal impairment (Agerso 2004).
Desmopressin (Schutte 2018) Two-compartment apparent population PK model describing the time profile of endogenous factor VIII coagulant activity (FVIII:C) following a desmopressin (DDAVP) administration in nonsevere haemophilia A patients (Schutte 2018; final covariate model with FVIII-recent on baseline FVIII, V1 and CL). Desmopressin is the administered intervention; the apparent PK parameters describe the resulting endogenous FVIII:C release as if it were a unit-dose drug input (the source paper fixed the dose to unity because no FVIII concentrate was infused).
Dexamethasone rat (Li 2012) Preclinical (rat). Mechanism-based PK/PD model for CYP3A1/2 induction by dexamethasone (DEX, single 100 mg/kg ip dose) in male Sprague-Dawley rats. PK is a two-compartment mammillary model with zero-order ip absorption of duration T0 directly into the central compartment (no first-order rate constant; CL/F, Q/F, Vc/F, Vp/F all reported as kg-normalised apparent values). PK BSV is exponential and is retained only on Q/F (all other PK BSVs were not significant). The PD cascade describes CYP3A1 and CYP3A2 induction at three molecular levels: (1) mRNA dynamics use an indirect-response (Dayneka-style) model in which a Hill-type fractional occupancy of CYP3A DNA-responsive elements by the DEX-PXR complex (FO = Cp^gamma / (SC50^gamma + Cp^gamma)) drives a stimulation signal Si,0 = Smax * FO that flows through a per-isoform chain of transit compartments with mean transit time tau (one compartment for CYP3A1, eight compartments for CYP3A2) before stimulating mRNA synthesis as d/dt(mRNAi) = kin,i * (1 + Si,ni) - kout,i * mRNAi. (2) Protein dynamics translate mRNA to CYP3A protein via d/dt(CYP3Ai) = ksyn,i * mRNAi^mi - kdeg,i * CYP3Ai, where the per-isoform amplification factor mi is a paper-mechanistic power exponent on mRNA. (3) Enzyme activity (rate of 6beta-hydroxytestosterone formation) is the algebraic linear combination EA = alpha * CYP3A1 + beta * CYP3A2 with per-isoform turnover-number rates alpha and beta (pmol 6beta-OHT / min / pmol CYP3A). The PK and PD layers were fit sequentially in NONMEM 7.1.2 with FOCE+I, the PK model first then the PD layers simultaneously with PK fixed. Three PD layers were fit by the naive pool approach (each animal contributed one PD observation per time point), so no PD IIVs are present. Numbers of transit compartments (n1 = 1, n2 = 8) are paper-mechanistic fixed structural integers.
Dexmedetomidine (Perez-Guille 2018) Two-compartment IV population PK with sigmoidal Imax PD on heart rate (HR) and mean arterial pressure (MAP) fractional responses for dexmedetomidine in Mexican Mestizo children (2-18 y) undergoing ambulatory surgery, with a priori allometric scaling on CL and Q (exponent 0.75) and V1 and V2 (exponent 1) at a 70 kg reference weight (Perez-Guille et al. 2018, Tables 2 and 3, allometric model)
Dexmedetomidine (Smuszkiewicz 2017) Two-compartment population PK model for intravenous dexmedetomidine continuous infusion in adult ICU patients undergoing analgosedation (Smuszkiewicz 2017). 27 medical and surgical ICU patients (17 male, 10 female; median age 59.5 y, median weight 75 kg) received continuous infusions of 0.1-1.5 ug/kg/h for 23.7-102 h. Age, sex, body weight, infusion duration, pretreatment SOFA score, and inotrope use were screened as covariates but none reached statistical significance, so the final model contains no covariate effects. IIVs on Vc, CL, Vp, and Q are diagonal (no clear correlations). Proportional residual error.
Dexmedetomidine (Su 2016) Two-compartment IV population PK model for dexmedetomidine in neonates (1 day-1 month) and infants (1-24 months) after open heart surgery, with a priori allometric weight scaling on CL, Q (exponent 0.75) and V1, V2 (exponent 1) at a 70 kg reference weight; an Emax-form postnatal-age maturation on CL with TM50 = 0.032 months; a power-form effect of total cardiopulmonary bypass duration on CL centred at 60 min; and a 1.24-fold multiplicative increase in CL for patients with right-to-left intracardiac shunt (Qp:Qs < 1) (Su 2016 Table 4, allometric weight-normalized full covariate model)
Dexmedetomidine (Talke 2018) Three-compartment IV population PK plus effect-compartment sigmoid Emax PD model for dexmedetomidine-induced peripheral vasoconstriction (ADC units from finger photoplethysmography) in healthy adult volunteers, with a priori allometric body-weight scaling on CL, Q2, Q3 (exponent 0.75) and V1, V2, V3 (exponent 1) at a 70 kg reference weight (Talke and Anderson 2018, Tables 3 and 4)
Dexmedetomidine piglet (Ezzati 2014) Preclinical (newborn piglet). One-compartment IV population PK model of dexmedetomidine in a piglet perinatal-asphyxia model with therapeutic hypothermia (Ezzati 2014). Clearance scales allometrically with body weight (Holford exponent 0.75) standardised to 70 kg, decreases with body temperature centred at 37 C (Ftemp), and is multiplied by a paper-specific factor FAED (= 0.558) in the post-hypoxic-ischemic state; volume scales allometrically with weight (exponent 1).
Dextroamphetamine (Castelli 2022) One-compartment population PK model for dextroamphetamine transdermal system (d-ATS) in adults and children with ADHD (Castelli 2022 APNA poster), with sequential zero- and first-order absorption (zero-order release over duration D1 into depot, then first-order Ka into central), power-law body-weight scaling on CL/F (exponent 0.47), V/F (0.53), and Ka (-0.29) at an assumed 70 kg reference, independent IIV on CL/F, V/F, Ka, and D1, bioavailability anchored at F = 1, and residual error not reported in the conference poster (encoded fixed at 0; see vignette Errata).
Dextromethorphan (TerHeine 2014) Semi-physiological eight-compartment population PK model for dextromethorphan and its three phase I metabolites (dextrorphan, 3-methoxymorphinan, 3-hydroxymorphinan) in adult breast-cancer patients receiving chronic oral tamoxifen, used as a dual CYP2D6 / CYP3A phenotypic probe (single 30 mg oral dose). Pre-systemic and systemic metabolism are integrated via a hypothetical hepatic metabolism compartment in rapid (quasi-steady-state) equilibrium with the dextromethorphan central compartment; the algebraic hepatic concentration drives parallel CYP2D6 (dextromethorphan -> dextrorphan) and CYP3A (dextromethorphan -> 3-methoxymorphinan) formation steps. Subsequent CYP3A-mediated conversion of dextrorphan and CYP2D6-mediated conversion of 3-methoxymorphinan both feed the terminal 3-hydroxymorphinan pool, which is eliminated by a single clearance to other species. All metabolite apparent volumes are fixed to 419 L (Abduljalil 2009 literature value) for identifiability. Individual post-hoc CYP2D6 (CL_CYP2D6,1) and CYP3A (CL_CYP3A,1) clearances serve as the phenotypic probe covariates used downstream by the companion tamoxifen / endoxifen model (TerHeine_2014_tamoxifen).
Dha ring early (Cao 2017) In vitro (P. falciparum 3D7 laboratory strain, early-ring-stage parasites). Dynamic stress PD model from Cao 2017 capturing the delayed dihydroartemisinin (DHA) killing effect observed in tightly age-synchronized parasite cultures; one of four stage-specific NLME fits in Table 1 (early-ring stage corresponds to 2 h post-infection per the Klonis 2013 experimental design that supplied the viability data). The killing rate k = kmax(S) * C^hill / (Kc(S)^hill + C^hill) is modulated by a dynamic stress variable S(t) that accumulates while drug concentration C exceeds C* = 0.1 nM (dS/dt = lambda(1 - S)) and resets toward zero otherwise. Stress-dependent kmax(S) = alphaS and Kc(S) = beta1*(1 - S) + beta2 (paper eq 7 and 8). DHA concentration evolves in the central compartment with first-order decay (default kdrug = log(2)/8 /h for in vitro experiments; override for in vivo simulations). Parasite count N(t) is normalized to N(0) = 1, so the deterministic parasites state IS the surviving viability fraction (paper eq 17). See modellib(‘Cao_2017_dha_ring_mid’), modellib(‘Cao_2017_dha_troph_early’), modellib(‘Cao_2017_dha_troph_late’) for the other three stage-specific fits combined in the in vivo PK-PD simulation of Fig 6.
Dha ring mid (Cao 2017) In vitro (P. falciparum 3D7 laboratory strain, mid-ring-stage parasites). Dynamic stress PD model from Cao 2017 capturing the delayed dihydroartemisinin (DHA) killing effect; one of four stage-specific NLME fits in Table 1 (mid-ring stage corresponds to 7.5 h post-infection). Mid-ring is the slowest-accumulating stage (half-life of the unstressed state ~1.86 h) and exhibits the strongest delay in the drug concentration-killing rate curve (Fig 3, Fig 4A). The killing rate k = kmax(S) * C^hill / (Kc(S)^hill + C^hill) is modulated by a dynamic stress variable S(t) that accumulates while drug concentration C exceeds C* = 0.1 nM (dS/dt = lambda(1 - S)) and resets toward zero otherwise. Stress-dependent kmax(S) = alphaS and Kc(S) = beta1*(1 - S) + beta2 (paper eq 7 and 8). DHA concentration evolves in the central compartment with first-order decay (default kdrug = log(2)/8 /h for in vitro experiments; override for in vivo simulations). Parasite count N(t) is normalized to N(0) = 1, so the deterministic parasites state IS the surviving viability fraction (paper eq 17). See modellib(‘Cao_2017_dha_ring_early’), modellib(‘Cao_2017_dha_troph_early’), modellib(‘Cao_2017_dha_troph_late’) for the other three stage-specific fits combined in the in vivo PK-PD simulation of Fig 6.
Dha troph early (Cao 2017) In vitro (P. falciparum 3D7 laboratory strain, early-trophozoite-stage parasites). Dynamic stress PD model from Cao 2017 capturing the delayed dihydroartemisinin (DHA) killing effect; one of four stage-specific NLME fits in Table 1 (early-trophozoite stage corresponds to 24 h post-infection). The killing rate k = kmax(S) * C^hill / (Kc(S)^hill + C^hill) is modulated by a dynamic stress variable S(t) that accumulates while drug concentration C exceeds C* = 0.1 nM (dS/dt = lambda(1 - S)) and resets toward zero otherwise. Stress-dependent kmax(S) = alphaS and Kc(S) = beta1*(1 - S) + beta2 (paper eq 7 and 8). Trophozoite stages have substantially higher maximum killing rate alpha than ring stages (Fig 4B), reflecting greater drug susceptibility once stress has accumulated. DHA concentration evolves in the central compartment with first-order decay (default kdrug = log(2)/8 /h for in vitro experiments; override for in vivo simulations). Parasite count N(t) is normalized to N(0) = 1, so the deterministic parasites state IS the surviving viability fraction (paper eq 17). See modellib(‘Cao_2017_dha_ring_early’), modellib(‘Cao_2017_dha_ring_mid’), modellib(‘Cao_2017_dha_troph_late’) for the other three stage-specific fits combined in the in vivo PK-PD simulation of Fig 6.
Dha troph late (Cao 2017) In vitro (P. falciparum 3D7 laboratory strain, late-trophozoite-stage parasites). Dynamic stress PD model from Cao 2017 capturing the delayed dihydroartemisinin (DHA) killing effect; one of four stage-specific NLME fits in Table 1 (late-trophozoite stage corresponds to 34 h post-infection). The killing rate k = kmax(S) * C^hill / (Kc(S)^hill + C^hill) is modulated by a dynamic stress variable S(t) that accumulates while drug concentration C exceeds C* = 0.1 nM (dS/dt = lambda(1 - S)) and resets toward zero otherwise. Stress-dependent kmax(S) = alphaS and Kc(S) = beta1*(1 - S) + beta2 (paper eq 7 and 8). DHA concentration evolves in the central compartment with first-order decay (default kdrug = log(2)/8 /h for in vitro experiments; override for in vivo simulations). Parasite count N(t) is normalized to N(0) = 1, so the deterministic parasites state IS the surviving viability fraction (paper eq 17). See modellib(‘Cao_2017_dha_ring_early’), modellib(‘Cao_2017_dha_ring_mid’), modellib(‘Cao_2017_dha_troph_early’) for the other three stage-specific fits combined in the in vivo PK-PD simulation of Fig 6.
Diazepam (Ku 2018) Two-compartment population PK model for intravenous diazepam in children aged 3 months to 18 years treated for status epilepticus. Clearance, central volume, inter-compartmental clearance, and peripheral volume scale allometrically with total body weight referenced to a 70 kg adult (fixed exponents 0.75 on CL and Q; 1 on V1 and V2). IIV is estimated on CL and V1 only; IIV on Q and V2 was held fixed at 0 in the final model to avoid >50% shrinkage. Proportional residual error.
Diazepam pbpk kpu human (Yau 2023) PBPK (14-compartment whole-body, four common Kpu clusters). Diazepam disposition in healthy adults after IV administration. The second of two whole-body PBPK simplification strategies investigated by Yau et al. (2023): rather than lumping compartments, the full kinetic structure and every individual tissue blood flow and volume are retained, and only the number of unknown distribution parameters is reduced. Tissues are grouped a priori by similarity of tissue composition using hierarchical clustering (drug-independent), and each cluster is assigned one common unbound tissue:plasma partition coefficient (Kpu). This reduces the distribution parameters from 13 to 4 while leaving the kinetic structure of the whole-body model intact, so individual tissue concentration-time profiles remain available. The lung, arterial and venous blood are treated as a single quasi-steady-state ‘central’ compartment (the only kinetic assumption); the remaining 13 tissues are perfusion-limited well-stirred compartments. Stomach, gut, spleen and pancreas drain into the liver via the portal vein; hepatic elimination uses the well-stirred model and renal elimination is placed on the central compartment. The four clusters are (1) bone, brain, muscle, pancreas and rest of body, (2) lung, gut, stomach, kidney, heart, spleen and liver, (3) skin and (4) adipose. Together with the four-scalar variant this was one of the two models the paper found best described the human diazepam data.
Diazepam pbpk kpu rat (Yau 2023) Preclinical (rat). PBPK (14-compartment whole-body, four common Kpu clusters, k-means clustering). Diazepam disposition in rats after IV administration - the rat counterpart of Yau_2023_diazepam_pbpk_kpu_human. The full kinetic structure and every individual tissue blood flow and volume are retained; only the number of unknown distribution parameters is reduced, by grouping tissues a priori on similarity of tissue composition and assigning one common unbound tissue:plasma partition coefficient (Kpu) per group. The lung, arterial and venous blood are treated as a single quasi-steady-state ‘central’ compartment; the remaining 13 tissues are perfusion-limited well-stirred compartments, with stomach, gut, spleen and pancreas draining into the liver via the portal vein. Clustering here is the k-means solution, which the paper selected as one of its three primary rat candidates: the four clusters are (1) bone, brain, muscle, pancreas and rest of body, (2) kidney, spleen and liver, (3) skin, lung, gut, stomach and heart, and (4) adipose. Note this differs from the hierarchical grouping used for the human model, where lung and the splanchnic organs cluster with kidney and liver. Rat physiology is scaled from body weight using the authors’ own control-stream construction, which reproduces the 250-g reference-rat values of Table S1 exactly. Because the rat hepatic clearance implied by CLb exceeds hepatic blood flow, hepatic clearance is capped at 0.99 * Q_liver as the authors’ control stream and the Figure S4 footnote specify; the balance is carried as renal clearance.
Diazepam pbpk lumped human (Yau 2023) PBPK (kinetically lumped 3-compartment, physiologically parameterised). Diazepam disposition in healthy adults after IV administration. One of two whole-body PBPK simplification strategies investigated by Yau et al. (2023): Nestorov kinetic lumping collapses the 16-state whole-body PBPK model into three compartments on the basis of tissue time constants (paper Table 1). The lung, arterial and venous blood are lumped into a quasi-steady-state ‘central’ compartment; all remaining tissues except adipose form the moderately equilibrating ‘peripheral1’ compartment; adipose alone forms the slowly equilibrating ‘peripheral2’ compartment. Micro-constants are not free parameters - they are computed from the 70-kg reference-man blood flows and volumes (Table S1) and the estimated unbound tissue:plasma partition coefficients via supplement Eq S12-S16, so the model retains physiological interpretability while having the same number of free parameters as an empirical 3-compartment model. Only total blood clearance and three lumped Kpu values are estimated. Kpu1 is the lung Kpu (blood + lung central pool), Kpu2 the common Kpu of the peripheral1 tissues and Kpu3 the adipose Kpu. Elimination is split into hepatic and renal components using the well-stirred model and the reported urinary excretion fraction fe = 0.0005; because only 0.05% of the dose is excreted unchanged the renal term is negligible and clearance is effectively all hepatic.
Diazepam pbpk lumped rat (Yau 2023) Preclinical (rat). PBPK (kinetically lumped 3-compartment, physiologically parameterised). Diazepam disposition in rats after IV administration - the rat counterpart of Yau_2023_diazepam_pbpk_lumped_human. Nestorov kinetic lumping collapses the 16-state whole-body PBPK model into three compartments on the basis of tissue time constants (paper Table 1). The lung, arterial and venous blood are lumped into a quasi-steady-state ‘central’ compartment; all remaining tissues except skin form the moderately equilibrating ‘peripheral1’ compartment; skin alone forms the slowly equilibrating ‘peripheral2’ compartment. Note the lump differs from the human model: skin is the slow compartment in rat (skin is a much larger fraction of body weight in rat) whereas adipose is the slow compartment in human, which is precisely the cross-species scaling limitation the paper discusses. Micro-constants are computed from body-weight-scaled blood flows and volumes via supplement Eq S12-S16, so only total blood clearance and three lumped Kpu values are estimated. Rat physiology is scaled from body weight using the authors’ own control-stream construction (cardiac output = 0.235 * BW^0.75, fractional cardiac outputs and fractional tissue weights), which reproduces the 250-g reference-rat values of Table S1 exactly. Because the rat hepatic clearance implied by CLb exceeds hepatic blood flow, hepatic clearance is capped at 0.99 * Q_liver exactly as the authors’ control stream and Figure S4 footnote specify; the balance is carried as renal clearance.
Diazepam pbpk scalar human (Yau 2023) PBPK (14-compartment whole-body, four common Kpu scalars). Diazepam disposition in healthy adults after IV administration. Variant 2.2 of the Yau et al. (2023) simplification strategies: the full kinetic structure and every individual tissue blood flow and volume are retained, tissue Kpu values are first predicted bottom-up with the Rodgers and Rowland (R&R) equations from tissue composition and diazepam physicochemistry, and then corrected by a small number of estimated multiplicative scaling factors shared within composition-based clusters (Eq 10: Kpu_i = Kpu_predRR,i * SF_i). The scalar therefore quantifies the systematic bias of the bottom-up R&R prediction rather than replacing it, which is why the paper judged this parameterisation the most promising for interspecies translation - the bias is assumed to be similar across species. The R&R weak-base branch is used because diazepam’s basic pKa is 3.4 (< 7), so tissue binding is driven by extracellular albumin rather than acidic phospholipids. The lung, arterial and venous blood are treated as a single quasi-steady-state ‘central’ compartment; the remaining 13 tissues are perfusion-limited well-stirred compartments, with stomach, gut, spleen and pancreas draining into the liver via the portal vein. The four clusters are (1) bone, brain, muscle, pancreas and rest of body, (2) lung, gut, stomach, kidney, heart, spleen and liver, (3) skin and (4) adipose. Together with the four-common-Kpu variant this was one of the two models the paper found best described the human diazepam data.
Diazepam pbpk scalar rat (Yau 2023) Preclinical (rat). PBPK (14-compartment whole-body, four common Kpu scalars, hierarchical clustering). Diazepam disposition in rats after IV administration - the rat counterpart of Yau_2023_diazepam_pbpk_scalar_human. The full kinetic structure and every individual tissue blood flow and volume are retained; tissue Kpu values are first predicted bottom-up with the Rodgers and Rowland (R&R) equations from rat tissue composition and diazepam physicochemistry, and then corrected by a small number of estimated multiplicative scaling factors shared within composition-based clusters (Eq 10: Kpu_i = Kpu_predRR,i * SF_i). The scalar therefore quantifies the systematic bias of the bottom-up R&R prediction rather than replacing it, which is why the paper judged this parameterisation the most promising for interspecies translation. The R&R weak-base branch is used because diazepam’s basic pKa is 3.4 (< 7). The lung, arterial and venous blood are treated as a single quasi-steady-state ‘central’ compartment; the remaining 13 tissues are perfusion-limited well-stirred compartments, with stomach, gut, spleen and pancreas draining into the liver via the portal vein. Clustering here is the hierarchical solution, which the paper selected as one of its three primary rat candidates: the four clusters are (1) bone, brain, muscle, pancreas and rest of body, (2) lung, gut, stomach, kidney, heart, spleen and liver, (3) skin and (4) adipose. Rat physiology is scaled from body weight using the authors’ own control-stream construction, which reproduces the 250-g reference-rat values of Table S1 exactly. Hepatic clearance is capped at 0.99 * Q_liver as the control stream and the Figure S4 footnote specify.
Diclazuril chicken pbpk (Yang 2023) PBPK (whole-body, flow-limited; broiler chicken). Nine-compartment physiologically based pharmacokinetic model for the anticoccidial diclazuril in broiler chickens after continuous oral exposure via medicated feed or drinking water, comprising intestinal contents (absorption site), liver, kidney, lumped skin + fat, muscle, a lumped rest-of-body compartment, lung, arterial plasma and venous plasma; all tissues are perfusion (flow) limited with tissue:plasma partition coefficients, absorption from the gut lumen is first order (Ka) in competition with first-order fecal loss of unabsorbed drug (Kgut), and elimination is hepatic (Clhe) plus fecal excretion. Built to predict edible-tissue residues and withdrawal periods against Chinese and European maximum residue limits (Yang 2023).
Diclofenac (Henin 2012) Semi-mechanistic Gastro-Intestinal Transit Time (GITT) absorption model applied to enteric-coated diclofenac (Henin 2012, The AAPS Journal 14:155-163). The intact tablet moves through the stomach (no drug release due to enteric coating) into the proximal small intestine, distal small intestine, and colon; drug release and absorption are collapsed into a single first-order rate per GI region because the drug release and absorption phenomena were indistinguishable in the observed plasma-only dataset. The effective absorption rate from the intact tablet is a step- function-weighted combination KA_PSI (proximal SI, 1.06 1/h), KA_DSI (distal SI, 8.64 1/h), and KA_Col (colon, 1 1/h FIXED) gated by tablet-position STEP functions with per-subject inflection points IP_APSI (gastric emptying), IP_PSI_DSI, and IP_DSI_C (population distributions from Table II). Total absorbed fraction is limited by the gut-wall bioavailability FA = 0.61 (Table III, IIV 8% CV). Disposition is a three-compartment model with population parameters allometrically scaled to a 70 kg reference (CL = 16.5 L/h/70kg^0.75, V1 = 3.68 L/70kg, Q2 = 1.75 L/h/70kg^0.75, V2 = 7.48 L/70kg, Q3 = 7.21 L/h/70kg^0.75, V3 = 3.79 L/70kg). Combined additive (10 nmol/L) + proportional (12.6%) residual error. This extraction encodes only the ‘no return to fundus’ subpopulation and treats the fundus + antrum transit as a single stomach-to-PSI event, appropriate for an enteric-coated tablet whose contents do not release in the stomach.
Diclofenac (Standing 2008) One-compartment population PK model for oral diclofenac suspension in children and adult volunteers (Standing 2008): two parallel transit-absorption arms (Savic 2007 analytical input form) feeding two depot compartments that each absorb into a single central disposition compartment with linear elimination. Allometric weight scaling on clearance and volume to a 70 kg reference. Captures the double-peak absorption profile common to immediate-release diclofenac. Separate proportional residual error for paediatric and adult cohorts. Source paper additionally fits between-occasion variability (BOV) on CL/F (20%) and Vd/F (93%) – BOV is not implemented in this nlmixr2 model file because it requires an OCC column in the user dataset; users who want BOV can add an etalcl_bov / etalvc_bov occasion-level random effect themselves.
Didanosine (Hirt 2009) One-compartment population PK model for didanosine (ddI) administered once daily as buffered chewable Videx tablets in West African HIV-1-infected children; first-order absorption with ka fixed at 4 1/h, additive residual error, exponential IIV on CL/F and Vc/F with off-diagonal covariance
Digoxin (Jelliffe 2014) Two-compartment population PK/PD model of digoxin in adults with first-order oral absorption, creatinine-clearance-dependent renal elimination, and a peripheral effect compartment normalized per body weight (Jelliffe et al. 2014, Ther Drug Monit; structural parameters carried from Reuning et al. 1973).
Digoxin (Zhou 2010) One-compartment first-order oral absorption population PK model of digoxin in older Chinese patients (Zhou 2010, Acta Pharmacol Sin); concomitant spironolactone, body weight, and serum creatinine modify Cl/F via multiplicative linear-deviation terms.
Dihydroartemisinin (Tarning 2012) One-compartment population PK model for oral dihydroartemisinin (parent drug, dosed as a fixed-dose tablet co-formulated with piperaquine) in 24 pregnant (second / third trimester) and 24 matched non-pregnant women with uncomplicated malaria on the Thai-Myanmar border (Tarning 2012 AAC). Transit-compartment absorption with 7 fixed transit compartments (ktr = (n+1)/MTT with n=7); drug-transit rate is set equal to the absorption rate from the last transit to central (single estimated ktr). Allometric scaling of CL/F (exponent 3/4) and V/F (exponent 1) on body weight centered at the cohort median 48.5 kg. F fixed at 1 with log-normal IIV (CV 30.3%); proportional pregnancy effect on F (-37.5%) and linear effect of log10 admission parasitaemia on F (+27.8% per log10 unit centered at 3.98). IIV on V/F (12.8% CV); between-occasion variability (BOV across 3 dose occasions) on MTT (50.9% CV) multiplexed by the OCC indicator. Additive residual on natural-log concentrations (sigma = 0.580), encoded as proportional residual on the linear-concentration scale per Kloprogge 2018 lumefantrine precedent. Companion file Tarning_2012_piperaquine.R models the co-administered piperaquine arm.
Dilmapimod (Yang 2016) Three-compartment IV population PK model for dilmapimod (SB-681323, a p38 MAPK inhibitor) coupled with an empirical indirect-response model for the inflammatory biomarker C-reactive protein (CRP) in severe-trauma adults at risk for acute respiratory distress syndrome (Yang 2016). BMI is a power covariate on CL and Q2. No statistically significant dilmapimod effect on CRP was retained in the final PD model, so the CRP component is an empirical post-injury production-decline / first-order-loss profile that is decoupled from dilmapimod exposure (Yang 2016 Results section 3.3.1).
Dimethyltryptamine alphaPower (Eckernas 2023) Effect-compartment sigmoidal Imax PK/PD model for the suppression of EEG alpha power by intravenous N,N-dimethyltryptamine (DMT) in healthy adults (Eckernas 2023). Plasma DMT is described by a two-compartment model with first-order elimination from the central compartment whose parameters are fixed from the upstream population PK analysis (Eckernas 2022, reproduced in Table S1). Alpha power is driven by the effect-site (biophase) concentration through a sigmoidal Imax function with Imax fixed to 1, i.e. DMT is able to fully suppress alpha power. Between-subject variability is on baseline alpha power (with a negatively skewed Box-Cox transformation of the eta) and on the effect-site IC50; between-occasion variability is on baseline alpha power. Residual variability on alpha power is proportional.
Dimethyltryptamine betaPower (Eckernas 2023) Effect-compartment sigmoidal Imax PK/PD model for the suppression of EEG beta power by intravenous N,N-dimethyltryptamine (DMT) in healthy adults (Eckernas 2023). Plasma DMT is described by a two-compartment model with first-order elimination from the central compartment whose parameters are fixed from the upstream population PK analysis (Eckernas 2022, reproduced in Table S1). Beta power is driven by the effect-site (biophase) concentration through a sigmoidal Imax function; unlike alpha power, suppression is only partial, with Imax estimated at 0.70. Between-subject variability is on baseline beta power and on the effect-site IC50 as a correlated block; between-occasion variability is on baseline beta power. Residual variability on beta power is proportional.
Dimethyltryptamine LZc (Eckernas 2023) Effect-compartment sigmoidal Emax PK/PD model for the increase in EEG signal diversity, measured as the Lempel-Ziv complexity (LZc) score, produced by intravenous N,N-dimethyltryptamine (DMT) in healthy adults (Eckernas 2023). Plasma DMT is described by a two-compartment model with first-order elimination from the central compartment whose parameters are fixed from the upstream population PK analysis (Eckernas 2022, reproduced in Table S1). The LZc score is driven by the effect-site (biophase) concentration through a sigmoidal Emax function with a maximum relative increase of about 10% over baseline. Between-subject variability is on baseline LZc, on the effect-site EC50 and on Emax; between-occasion variability is on baseline LZc. Residual variability on the LZc score is additive.
Docetaxel (Janssen 2023) Semi-physiological enriched three-compartment population PK model for intravenous docetaxel in pregnant cancer patients, applying gestational changes in AAG binding, glomerular filtration, hepatic plasma flow, CYP3A4 activity and body fluid volumes to the non-pregnant Koolen 2010 base model
Docetaxel (Koolen 2010) Five-compartment population PK model for intravenous and oral docetaxel with concomitant oral ritonavir in 36 adults with advanced cancer. Docetaxel: depot + single transit (Savic-style; ktr = 2/MAT) + three disposition compartments (central, peripheral1, peripheral2) with Bruno-style 3-compartment IV disposition (V_central, V_peripheral1, V_peripheral2; Q1 central-peripheral1, Q2 central-peripheral2). Clearance is parameterised via a well-stirred hepatic-extraction model (Wilkinson 1975) so that elimination is driven by intrinsic clearance CLi modulated by ritonavir plasma concentration via competitive inhibition (Ki = 0.028 ug/mL); CL = Q_hep * CLi / (CLi + Q_hep) with Q_hep fixed at 80 L/h. Hepatic bioavailability F_hep = Q_hep / (CLi + Q_hep) multiplies the depot -> transit transition to encode oral first-pass extraction. Gut bioavailability F_gut switches between F_doc = 0.19 (no ritonavir) and F_RTV = 0.39 (concomitant ritonavir, gated by the binary covariate CONMED_RTV). Polysorbate-80-driven micelle sequestration after IV docetaxel is encoded by a route-dependent central volume (V_central_iv = 9.8 L vs V_central_po = 44 L; gated by the binary per-dose-record covariate ROUTE_IV). Embedded one-compartment first-order-absorption ritonavir PK (depot_rtv + central_rtv) carries fixed typical-value parameters from Kappelhoff 2005 (CL/F = 10.5 L/h, V/F = 96.6 L, ka = 0.871 1/h, Tlag = 0.778 h) so that the ritonavir concentration that drives docetaxel CLi-inhibition is simulated within this single model file (modellib(‘Kappelhoff_2005_ritonavir’) is the upstream source). Inter-individual variability on CLi,0, V_central_iv, V_central_po, MAT, F_depot (shared between F_doc and F_RTV), and Ki; correlated etas for CLi,0 ~ V_central_iv (rho = 0.446). Proportional residual error is encoded at the final-model typical value (32%); the source paper’s separately-estimated higher 63% proportional RUV for the first 4 hours after oral administration is documented in the validation vignette’s Assumptions and deviations section. Inter-occasion variability on CLi,0 (22%), MAT (52%), and F_RTV (44%) reported in the source is not propagated – see vignette Assumptions and deviations.
Docetaxel (Ozawa 2007) Three-compartment IV PK coupled with a modified Friberg-style semimechanistic-physiological PK/PD model for docetaxel-induced neutropenia in Japanese cancer patients (Ozawa 2007). The PD layer extends Friberg 2002 with an additional zero-order input compartment that captures the transient ANC increase attributable to dexamethasone premedication; alpha-1 acid glycoprotein modulates the linear drug-effect slope on the proliferating compartment via a power-law form. Per-subject baseline ANC is supplied as a covariate and is used to initialise the proliferation, transit, and circulating compartments.
Docetaxel (Puisset 2007) Friberg-style semi-mechanistic myelosuppression PD model for docetaxel-induced neutropenia in adult cancer patients (Puisset 2007). PD-only: docetaxel plasma concentration is supplied as the time-varying CP_MGL covariate (mg/L) and drives a linear drug effect E_drug = Slope * CP_MGL on the proliferating compartment. The five-compartment Friberg PD chain (one proliferating pool, three transit compartments, one circulating ANC compartment) and feedback (Circ0 / circ)^gamma reproduce the structure of Friberg LE et al. (2002) J Clin Oncol 20(24):4713-4721. Three covariates retained in the published final covariate model act multiplicatively on Slope: alpha-1 acid glycoprotein (AAG) as a power form (AAG / 1.29)^(-0.72), prior chemotherapy >= 2 lines (PRIOR_CHEMO_LINES_GE2) as a 1.69-fold multiplier, and treatment centre Toulouse vs Paris (STUDY_TOULOUSE) as a 1.82-fold multiplier (the centre effect is acknowledged by the authors to most likely reflect a between-centre HPLC-assay bias on docetaxel concentration rather than a clinical PD covariate). The upstream docetaxel PK was held fixed at Baille 1997 / Bruno 1996 individual posthoc profiles during the published PD fit; users couple this model with their preferred docetaxel popPK (e.g. modellib(‘Ozawa_2007_docetaxel’) or modellib(‘Netterberg_2017_docetaxel’)) to drive CP_MGL.
Docetaxel (Rietveld 2025) Integrated plasma + tumour population PK model for docetaxel (DTX) delivered as CPC634, a core-crosslinked polymeric micelle that covalently entraps DTX via a pH-responsive sulfone-ester linker (Rietveld 2025). Released / conventional DTX is described by a canonical three-compartment IV plasma model (central, peripheral1, peripheral2; CL, Vc, Q1, Vp1, Q2, Vp2). Unreleased DTX (still bound to CPC634) is described by a two-compartment plasma model on the paper-specific compartments entrapped + peripheral_entrapped with linear elimination CLcpc and intercompartmental clearance Qcpc. Release of DTX from CPC634 in plasma is time-dependent: six first-order release rates K122 / K123 / K124 / K125 / K126 / K12 active in the post-dose time windows 0-0.5 / 0.5-1 / 1-2 / 2-6 / 6-168 / 168+ hours (paper Methods 2.6 + Figure 1A). The two tumour-tissue states tumor_entrapped (unreleased DTX in tumour) and tumor_released (released DTX in tumour) are connected to plasma by an influx parameter Kbtn (unreleased DTX, one-way plasma -> tumour) and an in/efflux balance parameter KbtDTX (released DTX), and to each other by the tumour-local release rate KrelT; both share the same tumour distribution volume VcT (paper Methods 2.7). Inter-individual variability on CL (released-DTX clearance), Vcpc (CPC634 central volume), and K122 (first-window release rate). Additive-on-log residual error (equivalent to proportional in linear space) estimated separately for the four therapeutically relevant observation streams (released-DTX plasma, unreleased-DTX plasma, released-DTX tumour, total-DTX tumour). Pragmatic deviations from the published model: (1) the $MIXTURE on Qcpc (subpopulation 1, P = 0.69, Qcpc1 = 0.00122 L/h; subpopulation 2, Qcpc2 = 0.00769 L/h) is encoded as a typical value at the dominant subpopulation 1 (Qcpc = Qcpc1); the minority Qcpc2 = 0.00769 L/h is documented in the vignette Assumptions and deviations section and can be plugged in by overriding lqcpc <- log(0.00769) at simulation time. (2) The 89Zr-CPC634 radiotracer arm of the PICCOLO PET imaging study (compartments 6 and 7 in the supplement; fast-loss rate K002 = 0.336 1/h active in the first 2 h after the 89Zr dose) is omitted -- the radiotracer arm shared the unreleased DTX disposition parameters (Vcpc / VPcpc / Qcpc / CLcpc) so it added no structural information beyond what entrapped already carries. See the vignette Assumptions and deviations section. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chandasana_2024_dolutegravir_lamivudine.html">Dolutegravir (Chandasana 2024)</a> </td> <td style="text-align:left;"> One-compartment oral population PK model with first-order absorption and elimination for dolutegravir 50 mg once daily as the dolutegravir/lamivudine fixed-dose combination in virologically suppressed adults living with HIV-1 (TANGO; Chandasana 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chandasana_2024_abacavir_dolutegravir_lamivudine_pediatric.html">Dolutegravir (Chandasana 2024b)</a> </td> <td style="text-align:left;"> One-compartment oral population PK model with first-order absorption, allometric weight scaling and a postmenstrual-age enzyme-maturation function on clearance for dolutegravir in infants, children and adolescents with HIV-1, applied without re-estimation to the ABC/DTG/3TC fixed-dose combination (dispersible tablet and tablet) in IMPAACT 2019 (Chandasana 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2015_dolutegravir.html">Dolutegravir (Zhang 2015)</a> </td> <td style="text-align:left;"> One-compartment oral population PK model with first-order absorption, absorption lag time, and first-order elimination for dolutegravir (integrase strand transfer inhibitor) in HIV-1-infected treatment-naive adults (Zhang 2015) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wojciechowski_2022_domagrozumab.html">Domagrozumab (Wojciechowski 2022)</a> </td> <td style="text-align:left;"> Quasi-steady-state TMDD population PK/PD model for domagrozumab (anti-myostatin IgG1) in healthy adult volunteers and pediatric patients with Duchenne muscular dystrophy (Wojciechowski 2022): two-compartment IV/SC drug disposition with parallel linear and Michaelis-Menten elimination, a synthesis-degradation total-myostatin compartment with drug-mediated internalization, and a study-population covariate (DIS_DMD) shifting myostatin baseline and turnover. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Khwarg_2024_donepezil.html">Donepezil im (Khwarg 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for long-acting intramuscular donepezil (GB-5001) with three-phase absorption: two lagged parallel first-order depots plus a simultaneous zero-order input, in healthy adult men </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Khwarg_2024_donepezil.html">Donepezil oral (Khwarg 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral donepezil (Aricept 10 mg tablet) with first-order absorption and an absorption lag time, in healthy adult men </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Diep_2026_donidalorsen.html">Donidalorsen (Diep 2026)</a> </td> <td style="text-align:left;"> Two-compartment population PK and indirect-response PD model for the GalNAc3-conjugated antisense oligonucleotide donidalorsen targeting prekallikrein (PKK) mRNA, fit to pooled data from phase 1 to phase 3 studies in healthy volunteers and patients with hereditary angioedema (Diep 2026). First-order SC absorption with categorical covariates on ka (arm vs abdomen/thigh injection site; autoinjector vs vial drug presentation), allometric scaling of CL/F, Vc/F, Q/F, and Vp/F on total body weight with paper-estimated exponents, multiplicative disease-status effects on Vc/F and Q/F, full 5x5 omega block on PK random effects, and an indirect-response model with donidalorsen-driven inhibition of PKK production carrying multiplicative disease-status effects on baseline PKK and IC50. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Singh_2025_donidalorsen.html">Donidalorsen (Singh 2025)</a> </td> <td style="text-align:left;"> Longitudinal exposure-response model relating per-4-week normalized hereditary angioedema (HAE) attack rate to 4-week-averaged plasma prekallikrein (PKK) concentration in adult and adolescent HAE patients receiving subcutaneous donidalorsen (Singh 2025). Sigmoidal Emax response driven by PKK with per-subject covariate effects of baseline HAE attack rate on Emax (power form) and baseline PKK on EC50 (power form). IIV on Emax (log-normal, CV 27.9%) and Hill (log-normal, CV 131%); no IIV on EC50 in the final model (the paper reports the EC50-IIV variant was numerically unstable during stepwise covariate evaluation). Per-4-week attack count is drawn from a Poisson likelihood with mean equal to the sigmoidal attack rate; the paper's generalized-Poisson dispersion parameter is not reported so extraction uses the standard Poisson approximation (documented in vignette Errata). Age, body weight, sex, and treatment-emergent ADA status were screened but not retained in the final model. Upstream PKK trajectories are generated by the paired Diep 2026 popPK/PD model (modellib('Diep_2026_donidalorsen')); this exposure-response model consumes the 4-week-averaged PKK as a time-varying covariate. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/AbdulAziz_2016_doripenem.html">Doripenem (AbdulAziz 2016)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for doripenem in 12 Malaysian critically ill adults with sepsis receiving 500 mg as a 1-hour infusion every 8 hours (Abdul-Aziz 2016). Reported on free (unbound) doripenem; observed total concentrations were corrected by multiplying by 0.90 to account for ~10% protein binding. Body-weight allometric scaling is fixed (0.75 on CL/Q, 1 on V1/V2, reference 70 kg); Cockcroft-Gault creatinine clearance has an exponential effect on CL centred at the cohort mean 82.5 mL/min. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2017_doripenem.html">Doripenem (Lee 2017)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for doripenem in 37 Korean adults with acute infections (pyelonephritis, intra-abdominal infection, neutropenic fever) and CLCR ranging 20-50 or >50 mL/min (Lee 2017). Clearance and central volume scale linearly with body weight (CL/WT = 0.109 L/h/kg, V/WT = 0.280 L/kg at WT=70 kg, CLCR=57 mL/min); CL additionally scales by a power exponent on Cockcroft-Gault creatinine clearance (raw mL/min, reference 57). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LopezArguello_2023_pbp_binding_pseudomonas.html">Doripenem qsp (LopezArguello 2023)</a> </td> <td style="text-align:left;"> QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for doripenem (carbapenem): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2023_dorzagliatin.html">Dorzagliatin (Wang 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral dorzagliatin (HMS5552, a first-in-class dual-acting glucokinase activator approved in China for type 2 diabetes mellitus) in healthy subjects and patients with T2DM (Wang 2023; N = 1062 subjects, 7686 concentrations pooled from six clinical trials). Disposition is a two-compartment model with sequential zero-order then first-order oral absorption (dose enters the depot over a zero-order window of duration D1 = 0.418 h, then is absorbed first-order at ka = 3.29 h^-1) and first-order elimination from the central compartment. Typical values for a 69 kg, 55-year-old male with AST 18 U/L enrolled in the later-phase studies are CL/F = 10.4 L/h, Vc/F = 80.6 L, Q/F = 3.02 L/h and Vp/F = 26.5 L. CL/F carries three power covariate effects plus one study indicator: (WT/69)^0.255, (AST/18)^-0.103, (AGE/55)^-0.135 and a multiplicative exp(0.203) = 1.23 uplift for the three early-phase studies (HMM0102 / HMM0103 / HMM0110), which the paper attributes to a formulation / bioavailability difference that could not be resolved as an F effect. Vc/F carries (WT/69)^0.553 and a multiplicative exp(-0.170) = 0.843 factor in females (15.7% lower than males). D1 carries exp(0.816) = 2.26 when food is consumed at least 1 h after dosing rather than 0.5 h after dosing. Because sex acts only on Vc/F and the meal-timing effect only on D1, neither changes steady-state AUCtau (the paper reports 0.01% and 0% respectively) while both shift Cmax,ss and Cmin,ss. Inter-individual variability is a correlated CL/F-Vc/F block (approximate CV 22.5% and 14.9%; covariance 0.0181) plus independent terms on Vp/F (48.8%) and D1 (82.8%); ka and Q/F carry no IIV. Residual error is combined additive plus proportional (109 ng/mL and 32.9%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kuchimanchi_2024_dostarlimab.html">Dostarlimab (Kuchimanchi 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for dostarlimab (anti-PD-1 IgG4) with sigmoid I_max time-dependent clearance, fitted to GARNET (advanced solid tumours) plus RUBY Part 1 (primary advanced or recurrent endometrial cancer with carboplatin-paclitaxel) data (Kuchimanchi 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Melhem_2022_dostarlimab.html">Dostarlimab (Melhem 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for dostarlimab (anti-PD-1 IgG4) with time-dependent (sigmoid I_max) clearance in adults with advanced solid tumours (Melhem 2022) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Freyer_2000_AVI_regimen.html">Doxorubicin (Freyer 2000)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for doxorubicin in 24 small cell lung cancer patients on the AVI regimen (Freyer 2000). Structural reduction from the paper's three-compartment model: only central CL and V were reported in Table 2, so the peripheral compartments are omitted. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Janssen_2023_pregnancy_cytotoxics.html">Doxorubicin (Janssen 2023)</a> </td> <td style="text-align:left;"> Semi-physiological enriched two-compartment population PK model for intravenous doxorubicin in pregnant cancer patients, applying gestational changes in albumin binding, glomerular filtration, hepatic plasma flow, CYP3A4 activity and body fluid volumes to the non-pregnant Joerger 2007 base model </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kunarajah_2017_doxorubicin.html">Doxorubicin (Kunarajah 2017)</a> </td> <td style="text-align:left;"> Population PK/PD model for IV doxorubicin (3-compartment) with first-order metabolism to doxorubicinol (1-compartment) and a cardiac troponin I (cTnI) turnover sub-model in paediatric oncology patients (Kunarajah 2017). Body surface area enters as a linear factor on every clearance and volume parameter; age enters as an additional power factor on doxorubicin clearance. The cTnI turnover sub-model is driven by a saturable Emax stimulation of cTnI synthesis by the combined doxorubicin + doxorubicinol plasma concentration, with the cTnI baseline shifted linearly by the prior cumulative anthracyclines dose received by the patient before the first dose analysed. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PerezBlanco_2016_doxorubicin.html">Doxorubicin (PerezBlanco 2016)</a> </td> <td style="text-align:left;"> Joint population PK model for IV doxorubicin (DOX, 3-compartment) and its active C-13 alcohol metabolite doxorubicinol (DOXol, 2-compartment) in adult patients with non-Hodgkin's lymphoma receiving R-CHOP chemotherapy (Perez-Blanco 2016). DOX was administered as a 0.5-h IV infusion at the protocol dose of 50 mg/m^2. The fraction Fm = 0.22 of total DOX clearance is routed to DOXol formation; the remaining (1 - Fm) fraction represents non-DOXol elimination pathways. The five volumes of distribution (V1/V2/V3 for DOX and V4/V5 for DOXol) were held fixed during estimation: the DOX volumes to the Kontny 2013 (doi:10.1007/s00280-013-2261-3) adult-reference values, and the DOXol volumes to the values obtained from a sensitivity analysis carried out for that purpose. No covariates were retained in the final model; bilirubin and AST showed an influence on CL and CLm but the OFV decrease was not statistically significant. Residual variability is proportional for both DOX and DOXol. This was the first published two-compartment DOXol popPK model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mody_2023_doxorubicin_dexrazoxane_breast_cancer.html">Doxorubicin dexrazoxane clinical jimt1 (Mody 2023)</a> </td> <td style="text-align:left;"> In vitro to clinical translation (JIMT-1 human breast cancer cell line; HER2-expressing, trastuzumab-resistant breast cancer) of the doxorubicin (DOX) + dexrazoxane (DEX) anticancer pharmacodynamic model (Mody 2023 Methods, 'In vitro-in vivo translation of PD responses'; Figure 6). The in-vitro cellular PD model of Mody_2023_doxorubicin_dexrazoxane_jimt1 is driven here by clinically relevant TUMOUR-SITE concentrations instead of cell-culture-medium concentrations. Clinical plasma PK is reproduced from the companion paper (Mody 2023 Sci Rep 13:3100, doi:10.1038/s41598-023-29964-4): DOX is a 3-compartment mammillary model with linear elimination and DEX a 2-compartment mammillary model parameterised by rate constants, both given as 15-min IV infusions for a typical 1.8 m^2 body-surface-area subject. Tumour-site concentrations are obtained by scaling the simulated plasma concentrations by a total tumour:plasma AUC ratio, kp_tumor_dox = 57.1 (digitised from He 2018 human tumour distribution data) and kp_tumor_dexrazoxane, which is ARBITRARY because no DEX tumour-distribution data exist (the paper simulates 0.1, 1 and 10). Plasma concentrations in mg/L are converted to uM at the PK-PD interface using literature molar masses. The paper's simulated regimen is DOX 50 mg/m^2 Q3W with DEX 500 mg/m^2 (a 10:1 DEX:DOX ratio, previously identified as maximally cardioprotective) over three cycles. For JIMT-1 the paper predicts a HIGHER AUEC of percent cell viability for DOX + DEX than for DOX alone (about 1560-fold), i.e. LESS tumour-cell killing when DEX is added, consistent with the antagonistic in-vitro interaction. An arbitrary 10% CV inter-individual variability is applied to the PD parameters and to both tumour:plasma ratios, exactly as the paper's 500-subject population simulation does; these are simulation conventions, not fitted variances, and are wrapped in fixed(). Sibling models: Mody_2023_doxorubicin_dexrazoxane_clinical_mdamb468, Mody_2023_doxorubicin_dexrazoxane_jimt1 (the in-vitro fit these PD parameters come from) and Mody_2023_doxorubicin_dexrazoxane_static_jimt1. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mody_2023_doxorubicin_dexrazoxane_breast_cancer.html">Doxorubicin dexrazoxane clinical mdamb468 (Mody 2023)</a> </td> <td style="text-align:left;"> In vitro to clinical translation (MDA-MB-468 human breast cancer cell line; triple-negative breast cancer) of the doxorubicin (DOX) + dexrazoxane (DEX) anticancer pharmacodynamic model (Mody 2023 Methods, 'In vitro-in vivo translation of PD responses'; Figure 6). The in-vitro cellular PD model of Mody_2023_doxorubicin_dexrazoxane_mdamb468 is driven here by clinically relevant TUMOUR-SITE concentrations instead of cell-culture-medium concentrations. Clinical plasma PK is reproduced from the companion paper (Mody 2023 Sci Rep 13:3100, doi:10.1038/s41598-023-29964-4): DOX is a 3-compartment mammillary model with linear elimination and DEX a 2-compartment mammillary model parameterised by rate constants, both given as 15-min IV infusions for a typical 1.8 m^2 body-surface-area subject. Tumour-site concentrations are obtained by scaling the simulated plasma concentrations by a total tumour:plasma AUC ratio, kp_tumor_dox = 57.1 (digitised from He 2018 human tumour distribution data) and kp_tumor_dexrazoxane, which is ARBITRARY because no DEX tumour-distribution data exist (the paper simulates 0.1, 1 and 10). Plasma concentrations in mg/L are converted to uM at the PK-PD interface using literature molar masses. The paper's simulated regimen is DOX 50 mg/m^2 Q3W with DEX 500 mg/m^2 (a 10:1 DEX:DOX ratio, previously identified as maximally cardioprotective) over three cycles. For MDA-MB-468 the paper predicts a COMPARABLE AUEC of percent cell viability across the DOX and DOX + DEX groups, consistent with the additive in-vitro interaction. An arbitrary 10% CV inter-individual variability is applied to the PD parameters and to both tumour:plasma ratios, exactly as the paper's 500-subject population simulation does; these are simulation conventions, not fitted variances, and are wrapped in fixed(). Sibling models: Mody_2023_doxorubicin_dexrazoxane_clinical_jimt1, Mody_2023_doxorubicin_dexrazoxane_mdamb468 (the in-vitro fit these PD parameters come from) and Mody_2023_doxorubicin_dexrazoxane_static_mdamb468. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mody_2023_doxorubicin_dexrazoxane_breast_cancer.html">Doxorubicin dexrazoxane jimt1 (Mody 2023)</a> </td> <td style="text-align:left;"> In vitro (JIMT-1 human breast cancer cell line; HER2-expressing, trastuzumab-resistant breast cancer). Cellular-level pharmacodynamic (PD) model for the single-agent and combined anticancer effects of doxorubicin (DOX) and dexrazoxane (DEX) on percent cell viability over 96 h (Mody 2023 Eqs 3, 3a-3e and 4a-4i; Table 2, JIMT-1 column). Both drugs degrade in the cell culture medium by first-order kinetics, so their driving concentrations are dosable one-compartment states conc_dox and conc_dexrazoxane that decay with kdeg (Mody 2023 Sci Rep Eqs 1-2, rate constants restated in this paper's Results). Each drug stimulates cell death through a capacity-limited Hill function whose signal is delayed by a three-compartment signal-transduction transit chain (transit1_<drug> to transit3_<drug>, first-order rate ktr = 1/tau). The two terminal transit signals act ADDITIVELY on the loss term of an exponentially growing viability state (Eq 4i). The drug-drug interaction parameter psi multiplies the DOX half-maximal killing concentration kc50_dox ONLY (Eq 4a); psi > 1 is antagonistic, psi < 1 synergistic and psi = 1 additive. Estimated psi is 2 (antagonistic) in JIMT-1. Setting psi = 1 recovers the paper's single-agent DOX equations (Eq 3a-3e) exactly, which is how the single-agent arms were fitted. No IIV or residual error was estimated (Table 2 reports point estimates with %RSE only); an arbitrary 10% CV IIV is introduced at simulation time per the paper's Methods. Sibling models: Mody_2023_doxorubicin_dexrazoxane_mdamb468 (the other cell line), Mody_2023_doxorubicin_dexrazoxane_static_jimt1 (the 72-h static concentration-response fit) and Mody_2023_doxorubicin_dexrazoxane_clinical_jimt1 (the same PD model driven by clinical tumour-site concentrations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mody_2023_doxorubicin_dexrazoxane_breast_cancer.html">Doxorubicin dexrazoxane mdamb468 (Mody 2023)</a> </td> <td style="text-align:left;"> In vitro (MDA-MB-468 human breast cancer cell line; triple-negative breast cancer). Cellular-level pharmacodynamic (PD) model for the single-agent and combined anticancer effects of doxorubicin (DOX) and dexrazoxane (DEX) on percent cell viability over 96 h (Mody 2023 Eqs 3, 3a-3e and 4a-4i; Table 2, MDA-MB-468 column). Both drugs degrade in the cell culture medium by first-order kinetics, so their driving concentrations are dosable one-compartment states conc_dox and conc_dexrazoxane that decay with kdeg (Mody 2023 Sci Rep Eqs 1-2, rate constants restated in this paper's Results). Each drug stimulates cell death through a capacity-limited Hill function whose signal is delayed by a three-compartment signal-transduction transit chain (transit1_<drug> to transit3_<drug>, first-order rate ktr = 1/tau). The two terminal transit signals act ADDITIVELY on the loss term of an exponentially growing viability state (Eq 4i). The drug-drug interaction parameter psi multiplies the DOX half-maximal killing concentration kc50_dox ONLY (Eq 4a); psi > 1 is antagonistic, psi < 1 synergistic and psi = 1 additive. Estimated psi is 1 (additive) in MDA-MB-468. Setting psi = 1 recovers the paper's single-agent DOX equations (Eq 3a-3e) exactly, which is how the single-agent arms were fitted. No IIV or residual error was estimated (Table 2 reports point estimates with %RSE only); an arbitrary 10% CV IIV is introduced at simulation time per the paper's Methods. Sibling models: Mody_2023_doxorubicin_dexrazoxane_jimt1 (the other cell line), Mody_2023_doxorubicin_dexrazoxane_static_mdamb468 (the 72-h static concentration-response fit) and Mody_2023_doxorubicin_dexrazoxane_clinical_mdamb468 (the same PD model driven by clinical tumour-site concentrations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mody_2023_doxorubicin_dexrazoxane_breast_cancer.html">Doxorubicin dexrazoxane static jimt1 (Mody 2023)</a> </td> <td style="text-align:left;"> In vitro (JIMT-1 human breast cancer cell line; HER2-expressing, trastuzumab-resistant breast cancer). Static 72-h concentration-response model for doxorubicin (DOX) and dexrazoxane (DEX) alone and in combination (Mody 2023 Eqs 1 and 2; Table 1, JIMT-1 columns). This is the ALGEBRAIC endpoint model, not a time-course model: it predicts percent cell viability at 72 h as a function of the two nominal applied concentrations, which enter as the covariates CONC_DOXORUBICIN and CONC_DEXRAZOXANE rather than as ODE states. The combination is described by the Chakraborty-Jusko / Pawaskar competitive interaction model (Eq 2), in which the interaction parameter psi scales the half-maximal inhibitory concentration of BOTH drugs; psi > 1 is antagonistic, psi < 1 synergistic and psi = 1 additive. Estimated psi is 1.11 (modestly antagonistic) in JIMT-1. The single-agent Imax and IC50 values were FIXED from the Eq-1 single-agent concentration-response fits and only psi was estimated (Monolix 2016R1). Setting either concentration to zero and psi to 1 recovers the single-agent inhibitory Hill model of Eq 1. Sibling models: Mody_2023_doxorubicin_dexrazoxane_static_mdamb468 (the other cell line) and Mody_2023_doxorubicin_dexrazoxane_jimt1 (the time-course PD model fitted to the same cells). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mody_2023_doxorubicin_dexrazoxane_breast_cancer.html">Doxorubicin dexrazoxane static mdamb468 (Mody 2023)</a> </td> <td style="text-align:left;"> In vitro (MDA-MB-468 human breast cancer cell line; triple-negative breast cancer). Static 72-h concentration-response model for doxorubicin (DOX) and dexrazoxane (DEX) alone and in combination (Mody 2023 Eqs 1 and 2; Table 1, MDA-MB-468 columns). This is the ALGEBRAIC endpoint model, not a time-course model: it predicts percent cell viability at 72 h as a function of the two nominal applied concentrations, which enter as the covariates CONC_DOXORUBICIN and CONC_DEXRAZOXANE rather than as ODE states. The combination is described by the Chakraborty-Jusko / Pawaskar competitive interaction model (Eq 2), in which the interaction parameter psi scales the half-maximal inhibitory concentration of BOTH drugs; psi > 1 is antagonistic, psi < 1 synergistic and psi = 1 additive. Estimated psi is 0.84 (modestly synergistic) in MDA-MB-468. The single-agent Imax and IC50 values were FIXED from the Eq-1 single-agent concentration-response fits and only psi was estimated (Monolix 2016R1). Setting either concentration to zero and psi to 1 recovers the single-agent inhibitory Hill model of Eq 1. Sibling models: Mody_2023_doxorubicin_dexrazoxane_static_jimt1 (the other cell line) and Mody_2023_doxorubicin_dexrazoxane_mdamb468 (the time-course PD model fitted to the same cells). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hopkins_2017_doxycycline.html">Doxycycline (Hopkins 2017)</a> </td> <td style="text-align:left;"> Two-compartment oral population PK model for doxycycline with two transit absorption compartments, fat-free-mass allometric scaling (CL exponent 0.75, V exponent 1.0, reference 70 kg FFM), and Doryx tablet (reference) / Doryx MPC delayed-release tablet / Doryx capsule formulation effects on relative bioavailability and absorption rate, plus a food (fed-status) effect on relative bioavailability and a formulation-dependent food effect on transit rate, plus a 14.4% increase in CL for female sex. Pooled from eight phase 1 healthy-volunteer trials (n = 178). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Foo_2016_droperidol.html">Droperidol (Foo 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption for intramuscular droperidol in 41 acutely agitated adults presenting to the emergency department (Foo 2016). Absorption rate constant ka and its IIV are fixed (ka = 10 1/h, omega_ka^2 = 1) because the available samples did not characterise absorption. A single shared random effect drives both CL and Vc (Table 2 footnote a: 'The same random effect was used for both Vc and CL'); Q and Vp have no IIV. No covariates were retained -- coingestion of alcohol was screened but not associated with CL or Vc, and patient weight was not available (Methods). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kovalenko_2016_dupilumab.html">Dupilumab (Kovalenko 2016)</a> </td> <td style="text-align:left;"> Dupilumab exploratory population PK model (Kovalenko 2016; 2-cmt with parallel linear + Michaelis-Menten elimination) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kovalenko_2020_dupilumab.html">Dupilumab (Kovalenko 2020)</a> </td> <td style="text-align:left;"> Dupilumab PK model (Kovalenko 2020) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2021_dupilumab.html">Dupilumab (Zhang 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for dupilumab in adult and adolescent patients with asthma (Zhang 2021), with first-order SC absorption and parallel linear plus Michaelis-Menten elimination from the central compartment. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kapitanov_2025_dupilumab.html">Dupilumab 3cmt qsp (Kapitanov 2025)</a> </td> <td style="text-align:left;"> QSP. Three-compartment physiologically-inspired PKRO (piPKRO) model of dupilumab (anti-IL4R IgG4) with full-binding TMDD in central, a lumped-interstitial peripheral, and an inflamed-skin site-of-action (SoA) compartment (Kapitanov 2025 Case Study 3 Approach 2). Extends the 2-compartment Case Study 2 model by adding a peripheral2 compartment representing 50 percent of the total interstitial skin volume (0.563 L), with dupilumab distribution parameterised by a fixed skin partition coefficient (P_dist,13 = 0.3) and a fixed central-to-skin distribution half-time (t_dist,13 = 30 h). IL4R concentration in the inflamed-skin SoA compartment is set to 2.02 nM (Table S4) reflecting higher immune-cell infiltrate and higher per-cell IL4R expression in atopic-dermatitis skin (Table S5). All other parameters (t_half, P_dist,12, C_R,1, C_R,2, k_deg, k_on, k_off) are carried unchanged from the Case Study 2 fit. The lumped peripheral volume V_2 is kept at 12.4 L per Kapitanov 2025 Table S4 Approach 2 (note: Section 3.3 narrates that V_2 was decreased by subtracting the SoA volume, but Table S4 Approach 2 reports V_2 = 12.4 L identical to Case 2 -- this file follows Table S4). The central-peripheral distribution half-time t_dist,12 = 55.9 h is recomputed for the 3-cpt structure per Eq 22. Case Study 3 illustrates that a specific SoA compartment with high local target burden and slow distribution can lead to substantial local drug depletion and lower receptor occupancy at the SoA than in the central and lumped-peripheral compartments (Kapitanov 2025 Figure 5). Deterministic single-subject typical-value simulation; no IIV / residual error is reported by the paper. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kovalenko_2020_dupilumab.html">Dupilumab base (Kovalenko 2020)</a> </td> <td style="text-align:left;"> Dupilumab primary base population PK model from Kovalenko 2020 (Model 3): 2-compartment with parallel linear + Michaelis-Menten elimination and a 3-transit-compartment SC absorption chain; fit to Phase 3 atopic-dermatitis data with only body weight as a covariate of central volume. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kovalenko_2020_dupilumab.html">Dupilumab covariate (Kovalenko 2020)</a> </td> <td style="text-align:left;"> Dupilumab primary covariate population PK model from Kovalenko 2020 (Model 4): 2-compartment with parallel linear + Michaelis-Menten elimination and a 3-transit-compartment SC absorption chain; fit to Phase 3 atopic-dermatitis data with body weight + albumin on Vc and BMI + EASI + race (White) on the linear elimination rate. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kapitanov_2025_dupilumab.html">Dupilumab qsp (Kapitanov 2025)</a> </td> <td style="text-align:left;"> QSP. Two-compartment physiologically-inspired PKRO (piPKRO) model of dupilumab (anti-IL4R IgG4) with full-binding TMDD in both the central and lumped-interstitial peripheral compartment (Kapitanov 2025 Case Study 2, calibrated to digitised first-in-human single-dose IV PK in healthy volunteers from Li 2015 at 1, 3, 8, and 12 mg/kg). The piPK framework replaces the classical (CL, Q, V1, V2) parameterisation with the physiological (V1, V2 = 13 L interstitial, t_half, t_dist, P_dist) macroparameter set of Kapitanov 2025 Eqs 3-7: drug elimination occurs at the same first-order rate in every compartment (k_el = log(2)/t_half), the peripheral volume is fixed at the total- body interstitial volume, and the peripheral partition coefficient P_dist is a first-class parameter. IL4R is expressed in both compartments with a bottom-up-derived concentration (Table S5; central 0.00605 nM, peripheral 0.127 nM); free drug + free receptor -> drug-receptor complex is described by mass-action forward binding with k_on = 1e-3 /(nM*s) and k_off = 3.3e-5 /s (K_D = 33 pM per Kuo 2010; Kapitanov 2025 ref 41). Complex degrades at the same rate as the free receptor (Section 3.1). Deterministic single-subject typical-value simulation; no IIV / residual error is reported by the paper for Case Study 2 and none is encoded here (fixed(0) alternative was not used because the paper reports no baseline IIV structure). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ogasawara_2020_durvalumab.html">Durvalumab (Ogasawara 2020)</a> </td> <td style="text-align:left;"> Two compartment PK model of durvalumab (anti-PD-L1) in patients with hematologic malignancies (Ogasawara 2020) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gisleskog_1999_dutasteride.html">Dutasteride (Gisleskog 1999)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for dutasteride (GI198745, a dual type-1/type-2 5-alpha-reductase inhibitor) in healthy male volunteers after single oral doses, with first-order absorption, an absorption lag-time, and parallel linear (CL_l) plus Michaelis-Menten (Vmax / Km) elimination from the central compartment (Gisleskog 1999). All volumes and clearances are apparent (oral, no IV reference); bioavailability is assumed dose-independent. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Keizer_2011_E7820.html">E7820 human (Keizer 2011)</a> </td> <td style="text-align:left;"> Population PK/PD model for the alpha2-integrin inhibitor E7820 in patients with advanced solid tumors or lymphoma (Keizer 2011 clinical column). One-compartment oral PK with first-order absorption (PK structure and parameter values inherited from an earlier phase I popPK analysis of the same study and reproduced in Keizer 2011 Table II; the absorption model was simplified from the original turnover-absorption form to a first-order form to ease multi-dose simulations). PD is an indirect-response (turnover) model for alpha2-integrin expression on platelets, with an Emax inhibition function (Emax fixed at 1, Hill exponent gamma fixed at 1) acting on the input rate kin. BSV is reported on baseline integrin expression and on drug sensitivity (IC50). No tumor-growth submodel is included in the clinical analysis (Keizer 2011 Figure 3 caption: 'The clinical model had the same structure, but did not incorporate a sub-model for tumor size'). Parameter values from Keizer 2011 Tables II (clinical PK) and III (clinical integrin PD). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Keizer_2011_E7820.html">E7820 mouse (Keizer 2011)</a> </td> <td style="text-align:left;"> Preclinical (mouse, female nude with subcutaneous KP-1 pancreatic-carcinoma xenograft). Sequential PK/PD/tumor-growth model for the alpha2-integrin inhibitor E7820 (Keizer 2011). Stage 1: one-compartment oral PK with first-order absorption, per-kg parameterisation. Stage 2: indirect-response (turnover) model for alpha2-integrin expression on platelets, with an Emax inhibition function (Emax fixed at 1, Hill exponent gamma fixed at 1) acting on the input rate kin. Stage 3: exponential tumor growth on diameter with an initial-slow-growth term (1 - exp(-beta*t)) gating the growth rate, and a sigmoidal Emax inhibition driven by relative alpha2-integrin inhibition ((I_base - integrin)/I_base) with Hill coefficient fixed at 5. Parameters from Keizer 2011 Tables II (preclinical PK), III (preclinical integrin PD), and IV (tumor growth). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2024_eculizumab.html">Eculizumab (Lee 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of eculizumab (SB12 proposed biosimilar and reference Soliris) with a direct inhibitory sigmoid Imax model for terminal complement activity and a direct sigmoid Emax model linking terminal complement activity to serum LDH, in healthy subjects and patients with paroxysmal nocturnal haemoglobinuria (Lee 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jonsson_2015_edoxaban.html">Edoxaban (Jonsson 2015)</a> </td> <td style="text-align:left;"> Joint parent plus metabolite population PK model for edoxaban and its main metabolite M4 in adults with normal kidney function through severe renal impairment, from a dedicated renal impairment study. Edoxaban disposition is two-compartment with Savic 2007 analytical transit-compartment absorption (non-integer NN, with the absorption rate constant constrained equal to the transit rate constant ktr) and absolute oral bioavailability estimated on the logit scale; edoxaban clearance is split into a renal arm that is linear in creatinine clearance and drives a urinary-excretion compartment, and a non-renal arm that is assumed to form all of the M4 metabolite. M4 is described by a one-compartment model with first-order formation. Fixed allometric body-weight scaling (0.75 on all clearance terms that are not a function of kidney function, 1 on all volumes) is applied around a 70 kg reference subject. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Niebecker_2015_edoxaban.html">Edoxaban (Niebecker 2015)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and a lag time for edoxaban in adults; pooled phase 1 healthy volunteers (13 studies) and Hokusai-VTE phase 3 patients with deep-vein thrombosis or pulmonary embolism (Niebecker 2015). Apparent clearance is split into a non-renal component and a piecewise-linear renal component driven by creatinine clearance, with a phase-3 patient effect on the upper-CLcr slope and on Q/F. Asian race increases Vc/F; concomitant P-glycoprotein inhibitors increase phase-1 CL/F and F. The fed-state study 6 has a slower ka and higher non-renal CL/F (FED covariate). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Stein_2018_mAb_nonlinear_PK.html">Efalizumab (Stein 2018)</a> </td> <td style="text-align:left;"> Two-compartment QSS TMDD typical-value fit for efalizumab (anti-CD11a mAb) used to illustrate the critical concentration (Ccrit) for nonlinear PK (Stein and Peletier 2018 Table 1) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gastonguay_2005_efaproxiral.html">Efaproxiral (Gastonguay 2005)</a> </td> <td style="text-align:left;"> Linear 2-compartment IV popPK model for efaproxiral (RSR13) with an algebraic linear RBC:plasma proportionality (Crbc = SLPRBC * Cc) and a linear PD model relating partial pressure of oxygen at 50% hemoglobin saturation (p50, mmHg) to RBC efaproxiral concentration (p50 = INTp50 + SLPp50 * Crbc). Population of 451 cancer patients receiving radiation therapy pooled across six phase I-III trials. Full covariate model: BSA and AGE on CL; BSA, AGE and baseline albumin (BALB) on V1 and V2; BSA on Q; max administered dose (MDOS), AGE and BALB on SLPRBC; primary cancer type indicators (breast, glioma, other; reference = lung) on SLPp50. Three-way correlated IIV block on CL, V1, V2 plus diagonal IIV on Q, SLPRBC, INTp50 and SLPp50. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bienczak_2016_efavirenz.html">Efavirenz (Bienczak 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral efavirenz in African children (Bienczak 2016), with Savic 2007 transit-compartment absorption (NN = 25 fixed transit compartments and a separate first-order absorption step ka from the depot to the central compartment), oral bioavailability fixed to 1 (no intravenous data), Anderson-Holford allometric scaling of all clearance and volume parameters to a 15.4 kg reference child (exponents 0.75 on CL and Q, 1.0 on Vc and Vp), and a composite CYP2B6 516G>T (rs3745274) | 983T>C (rs28399499) SNP-vector effect on apparent oral clearance that distinguishes six metabolic subgroups (516GG|983TT extensive metabolizer reference, 516GG|983TC and 516GT|983TT intermediate, 516TT|983TT and 516GT|983TC slow, 516GG|983CC ultra-slow). Encoded as log-ratio multiplicative shifts on the 516GG|983TT EM reference so the single etalcl IIV applies uniformly on the log-CL scale across all six SNP-vector subgroups. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dhoro_2015_efavirenz.html">Efavirenz (Dhoro 2015)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral efavirenz in HIV-positive and HIV/TB co-infected adults in Zimbabwe (Dhoro 2015), with apparent clearance CL/F stratified by CYP2B6 983T>C (CYP2B6*18, rs28399499) genotype and multiplicative fractional covariate effects of CYP2B6 516G>T (CYP2B6*6, rs3745274) genotype, body weight, and sex on CL/F. Absorption rate constant ka and apparent volume V/F are fixed from the upstream Nyakutira 2008 Zimbabwean cohort. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Heathman_2024_efavirenz.html">Efavirenz (Heathman 2024)</a> </td> <td style="text-align:left;"> Population pharmacokinetic model for efavirenz (EFV) and its 8-hydroxy- and 7-hydroxy-metabolites in 135 healthy volunteers receiving a single 600 mg dose followed by 17 days of 600 mg/day (4594 plasma concentration samples). Each of EFV, 8-OH EFV, and 7-OH EFV is a 2-compartment model with the metabolite central volume fixed equal to that of EFV (lack of identifiability). EFV absorption is sequential zero- (D1 = 1.74 h) plus first-order (KA = 0.165/h). Two independent enzyme-turnover models drive CYP2B6 and CYP2A6 autoinduction: R(t) = kout * (1 + Emax * Cc_EFV / (EC50 + Cc_EFV)); dE/dt = R(t) - kout * E; E(0) = 1. CYP2B6 modulates the EFV-to-8-OH formation arm (CL-EFV,2B6 = 3.64 L/h; Emax-2B6 = 15.5; EC50-2B6 = 32000 nM = 10.10 mg/L) and the 8-OH-onward CYP2B6 arm (CL-8OH,2B6 = 0.758 L/h); CYP2A6 modulates the EFV-to-7-OH formation arm (CL-EFV,2A6 = 0.0947 L/h; Emax-2A6 = 4.22; EC50-2A6 = 12500 nM = 3.95 mg/L). UGT2B7 elimination arms (CL-EFV,UGT = 0.0504 L/h; CL-8OH,UGT = 5.44 L/h) and the 7-OH total clearance (CL-7OH = 3.39 L/h) are not autoinduced. CYP2B6 phenotype reduces the EFV-to-8-OH formation arm by 9.72% (IM) / 9.06% (PM, encoded as canonical CYP2B6_SM) and reduces the CYP2B6 Emax by 53.5% (IM) / 93.2% (PM); NM (extensive metabolizer) is the reference. PM subjects show essentially no autoinduction (effective Emax-2B6 ~ 1.05) and accumulate over 2-3 weeks. IIV is exponential on PK parameters (estimated for most; 15.1% CV fixed for several); residual variability is proportional for all three analytes (25.8% / 28.0% / 29.9% CV). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hirt_2009_efavirenz.html">Efavirenz (Hirt 2009)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption and elimination for once-daily oral efavirenz (EFV) in treatment-naive HIV-1-infected West African children (Hirt 2009). CL/F and V/F scale linearly with body weight (shared allometric exponent fixed at 1) and CL/F additionally varies with postnatal age via a power covariate centred at the cohort median 6.35 years (signed exponent -0.535, so apparent clearance decreases with age); the inter-individual variability of V/F is forced to perfect correlation with the eta of CL/F and is constructed as vc_eta_scale * etalcl (the K parameter in Hirt 2009 Table 2); multiplicative residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Luo_2016_efavirenz.html">Efavirenz (Luo 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and elimination for oral efavirenz in pediatric HIV-1-infected patients (Luo 2016). Capsule / capsule-sprinkle formulation; body weight is a power covariate on CL/F, Vc/F, and Ka with reference 20 kg. The adult cohort (n = 24 healthy adults) and oral-solution formulation (study-specific Frel) reported in the same paper are documented in the validation vignette but not encoded as separate sub-models. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mukonzo_2009_efavirenz.html">Efavirenz (Mukonzo 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for single-dose oral efavirenz in 121 healthy Ugandan adults, with sequential zero-order followed by first-order absorption to the central compartment. Apparent oral clearance CL/F is reduced by 21% in homozygous CYP2B6*6 (rs3745274 T/T) and by 20% in homozygous CYP2B6*11 (rs35303484 G/G) carriers (multiplicative fractional effects). Relative bioavailability Frel is increased by 26% in ABCB1 rs3842 mutant carriers (heterozygote or homozygote). Apparent peripheral volume Vp/F is 2.08-fold higher in women than in men. Concentrations are reported in mg/L (1 mg/L efavirenz = 3.168 micromol/L). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Olagunju_2018_efavirenz.html">Efavirenz (Olagunju 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral efavirenz in HIV-positive pregnant women (Olagunju 2018), with composite CYP2B6 516G>T (rs3745274) and 983T>C (rs28399499) metaboliser status (slow / intermediate / fast) as a categorical covariate on CL/F and fixed-exponent allometric body-weight scaling on CL/F and V/F. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Robarge_2017_efavirenz.html">Efavirenz (Robarge 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for a single 600 mg oral dose of efavirenz in 73 HIV-seronegative adult volunteers (Robarge 2017), with parallel zero- and first-order absorption (independent lag times t_lag1 and t_lag2, zero-order duration D2), allometric fat-free-mass scaling on CL/F (exponent 3/4, fixed), allometric fat-mass scaling on Vp/F (exponent 1, fixed), and CYP2B6 metaboliser status (normal / intermediate / slow) reducing CL/F by 0%, 25% and 51% respectively. Bioavailability is fixed to F = 1 (no IV reference formulation); the first-order absorption fraction F1 = 0.414 was estimated and the zero-order fraction F2 = 1 - F1 = 0.586 was assigned by mass balance. All absorption-related typical values (F1, t_lag1, K_a, t_lag2, D2) were estimated in an interim model and then fixed prior to covariate evaluation; the IIVs on those absorption parameters were re-estimated in the final model. Block-structured between-subject variability is estimated on (CL/F, Q/F, V_p/F) with correlations rho(CL/F, V_p/F) = 0.196 and rho(Q/F, V_p/F) = 0.849; rho(CL/F, Q/F) was fixed at 0. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Salem_2014_efavirenz.html">Efavirenz (Salem 2014)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral efavirenz in HIV-1-infected children (Salem 2014). Allometric body-weight scaling on apparent clearance (fixed exponent 0.75) and apparent volume of distribution (fixed exponent 1.0) referenced to 70 kg; sigmoid Emax maturation of CL/F with postnatal age (TM50 = 4.6 months, Hill = 3.4); 51% reduction in CL/F for CYP2B6-516 T/T homozygotes; Emax maturation of relative bioavailability for the oral liquid (suspension or solution) formulations vs the capsule reference (mature F = 0.79; TM50 = 10.6 months; Hill fixed at 1). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sanchez_2011_efavirenz.html">Efavirenz (Sanchez 2011)</a> </td> <td style="text-align:left;"> One-compartment population PK/pharmacogenetic model for oral efavirenz in Caucasian HIV-infected adults (Sanchez 2011), with GGT, CYP2B6*6 genotype (linked 516G>T + 785A>G), and ABCC4 (MRP4) 1497C>T carrier covariate effects on apparent oral clearance CL/F. Absorption rate ka fixed at 0.3 h^-1 (sparse TDM data could not estimate it); no covariate effect on V/F. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jansson_2008_eflornithine.html">Eflornithine rat (Jansson 2008)</a> </td> <td style="text-align:left;"> Preclinical (rat, Sprague-Dawley). Stereoselective two-enantiomer population PK model of racemic eflornithine after single oral or IV doses in male Sprague-Dawley rats (Jansson 2008). Each enantiomer (L = active, D) carries its own 2-compartment disposition (CL, Vc) with shared Q and Vp from the IV fit; oral absorption is modeled with a shared Savic 2007 transit-compartment chain (continuous number of compartments via Stirling approximation) feeding per-enantiomer depots that drain to central via saturable Michaelis-Menten kinetics (Tmax, Kt). Bioavailability differs between enantiomers and shifts upward at the highest oral dose level (3000 mg/kg) via a categorical indicator. Racemic plasma concentration is the algebraic sum Cc_rac = Cc_l + Cc_d. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/DoldanMartelli_2013_EGF_IFN_chimera.html">EGF IFN chimera (DoldanMartelli 2013)</a> </td> <td style="text-align:left;"> In vitro (Daudi human Burkitt lymphoma cell line). Mechanistic kinetic model of an EGF-IFNalpha-2a chimeric ligand binding to EGFR and IFN receptor on the cell membrane: sequential two-subunit engagement, receptor lateral diffusion, and internalization (Doldan-Martelli 2013). Default parameters are wild-type IFN chimera in Daudi-EGFR cells (overexpressing EGFR ~300x parental); k2on / k2off can be overridden for K133A and R144A IFN mutants, and R1_0 / R2_0 for parental Daudi cells (see vignette). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ide_2020_elotuzumab.html">Elotuzumab (Ide 2020)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for elotuzumab (anti-SLAMF7 humanized IgG1) in Japanese and non-Japanese patients with multiple myeloma (Ide 2020); parallel linear and Michaelis-Menten elimination from the central compartment plus second-order target-mediated elimination from the peripheral compartment driven by a non-renewable target pool, with time-varying serum M protein on Vmax. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Farrell_2014_eltrombopag.html">Eltrombopag (Farrell 2014)</a> </td> <td style="text-align:left;"> Population PK/PD model for eltrombopag in healthy male volunteers (single dose) and adult patients with chronic liver disease (CLD; multiple daily doses) (Farrell 2014). Two-compartment apparent disposition with dual sequential first-order absorption: Ka1 acts on the depot from the end of the absorption lag time (ALAG1) until time MTIME after the dose, and Ka2 acts thereafter. CL/F is reduced in females, in East Asian subjects, and in CLD patients with a linear-in-Child-Pugh-score gradient (HEPIMP_CP_SCORE >= 5). Vc/F is approximately three-fold higher in South/Central Asian subjects. Platelet dynamics use a four-compartment lifespan model (three maturing precursor pools feeding the circulating-platelet pool) with linear stimulation of precursor production by plasma eltrombopag; the slope SLOP is 34% lower in East Asian CLD patients. PD parameters are CLD-specific (median baseline platelet 41 Gi/L). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Goggin_2004_emfilermin.html">Emfilermin (Goggin 2004)</a> </td> <td style="text-align:left;"> One-compartment population PK model for subcutaneous emfilermin (recombinant human leukaemia inhibitory factor, r-hLIF) in healthy postmenopausal women and in infertile women undergoing in vitro fertilization and embryo transfer (IVF-ET) (Goggin 2004). Absorption is zero-order (D1 = 0.84 h, invariant, no IIV) directly into the central compartment, followed by first-order elimination. Apparent clearance CL/F is decreased by 35% in IVF-ET patients (typical 37 L/h) relative to healthy postmenopausal women (typical 57 L/h). Apparent volume V/F is linear in body weight on the natural scale: V/F = 235 L at the median 62 kg, increasing or decreasing by 6.7 L/kg (~29% per 10 kg) -- an absolute-linear covariate form, not log-multiplicative. Inter-individual variability is log-normal on CL/F (17% CV) and V/F (28% CV); inter-occasion variability is log-normal on V/F (23% CV) across three protocol-defined occasions (first dosing day = 1, intermediate dosing days = 2, last dosing day = 3). Residual error is proportional (20% CV). Studied weight range was 48-83 kg; the linear V/F-WT term is extrapolation-unsafe below ~27 kg where the typical V/F would become negative. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yoneyama_2017_emicizumab.html">Emicizumab (Yoneyama 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order subcutaneous absorption and elimination for emicizumab (ACE910), a bispecific anti-FIXa/FX humanized monoclonal antibody mimicking the cofactor function of activated factor VIII, in healthy male adult volunteers (Japanese and Caucasian) and Japanese male adult/adolescent patients with severe hemophilia A with or without factor VIII inhibitors (Yoneyama 2017). Body-weight allometric exponents are fixed (0.75 on CL/F, 1 on Vd/F) per Yoneyama 2017 Methods. Anti-emicizumab neutralizing antibody (ADA_POS) increases CL/F by a factor of exp(2.01) and the effect onsets 33.4 days post the first SC dose (NONMEM MTIME parameterisation). The companion repeated time-to-event (RTTE) bleeding-hazard model from Yoneyama 2017 Section 2.4 is not included here; nlmixr2lib does not currently support TTE models. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Baron_2016_empagliflozin.html">Empagliflozin (Baron 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK with lagged first-order absorption for empagliflozin in patients with type 2 diabetes (T2DM), coupled with two indirect-response PK/PD models for fasting plasma glucose (FPG) and glycated hemoglobin (HbA1c). The drug effect on FPG elimination is driven by steady-state AUC (AUCss = DOSE_EMPA_MGD * 1e6 / MW / CL) via an Emax function (Gmax, AUC50); FPG in turn drives HbA1c production with a boundary-condition baseline (HbA1climit). Pooled popPK/PD analysis of 4065 T2DM patients (PK n = 2761 active) from two phase I, four phase II, and four phase III studies (Baron 2016 Diabetes Therapy). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jiang_2024_empagliflozin.html">Empagliflozin (Jiang 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with Savic transit-compartment absorption for oral empagliflozin in healthy Korean adults (Jiang 2024). Pools two randomized, open-label, two-period crossover phase 1 studies that compared a novel empagliflozin L-proline cocrystal formulation (CKD-370) against the conventional empagliflozin formulation at 25 mg, and as a 5 mg/1000 mg empagliflozin/metformin fixed-dose combination. An analytical transit chain (Ktr = 9.19 1/h, Mtt = 0.63 h, derived N = 4.79) feeds first-order absorption (Ka) into a two-compartment disposition model. Log-transformed body weight scales apparent clearance (exponent 0.64) and the apparent peripheral volume (exponent 0.57). Interindividual variability on Ka, CL, and the peripheral volume; interoccasion variability across the two crossover periods on Ktr, Mtt, Ka, and CL. Formulation (cocrystal vs conventional) was screened as a covariate and was NOT retained: the paper's central finding is that the L-proline cocrystal does not alter the absorption phase or any PK parameter, so this single model describes both formulations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Johnston_2019_empagliflozin.html">Empagliflozin (Johnston 2019)</a> </td> <td style="text-align:left;"> Exposure-response (PD-only) model for the effect of empagliflozin on HbA1c in patients with type 1 diabetes mellitus (T1DM) on background insulin therapy (M-EASE-2; Johnston 2019). A direct-response Emax function of individual steady-state empagliflozin AUC (AUC_EMPA, supplied as a per-subject covariate column from an upstream popPK analysis -- Mondick 2018 plus EASE-2 / EASE-3 data-on-file) reduces the model-predicted baseline HbA1c, with an additional linear placebo drift over time. Full covariate model on baseline HbA1c, Emax, and placebo (sex, insulin delivery type, body weight, eGFR, baseline insulin daily dose, and -- on Emax only -- baseline HbA1c). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Johnston_2021_empagliflozin.html">Empagliflozin MEASE1 (Johnston 2021)</a> </td> <td style="text-align:left;"> Semi-mechanistic direct-effect exposure-response (PD-only) model for empagliflozin on total daily insulin dose (TDID), mean daily glucose (MDG), and glycated hemoglobin (HbA1c) in adults with type 1 diabetes (T1D) (M-EASE-1; Johnston 2021). Individual steady-state empagliflozin AUC (AUC_EMPA, supplied as a per-subject covariate column from the upstream Johnston 2021 popPK; see Johnston_2021_empagliflozin_popPK) drives (i) a proportional-Emax reduction in TDID, (ii) a direct-Emax reduction in MDG offset by a linear time-dependent placebo drift and coupled to the TDID reduction via a power-form insulin effect, and (iii) HbA1c as a power-form function of MDG relative to baseline. Covariates: sex, body weight, eGFR, insulin delivery type (CSII vs MDI), and baseline HbA1c. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Johnston_2021_empagliflozin.html">Empagliflozin popPK (Johnston 2021)</a> </td> <td style="text-align:left;"> Population pharmacokinetic (PopPK) model for empagliflozin in adults with type 1 diabetes (T1D) enrolled in the EASE clinical program (pooled EASE-1 phase II, EASE-2 phase III, and EASE-3 phase III). Two-compartment model with first-order elimination and a sequential zero-order (duration D1) plus first-order (ka) oral absorption with a depot lag time (ALAG). Full covariate power / multiplicative model on CL/F (age via TPRO/AP/eGFR/TDID/WT + sex + smoking), V2/F, V3/F, Q/F, and ka (Johnston 2021 Table S2). The PopPK was adapted from a prior T2D empagliflozin analysis (Baron 2016) and re-fit on the EASE-1 + EASE-3 dataset with EASE-2 as external evaluation, then re-estimated on the combined EASE-1/-2/-3 data to generate the individual AUCss inputs to the M-EASE-1 and M-EASE-2 exposure-response models. This file encodes the final full-covariate model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valade_2014_emtricitabine.html">Emtricitabine (Valade 2014)</a> </td> <td style="text-align:left;"> Two-compartment oral population PK model for emtricitabine (FTC) in HIV-infected pregnant and non-pregnant women, with first-order absorption and elimination. Creatinine clearance (Cockcroft-Gault, raw mL/min) on apparent oral clearance via the power model CL/F = 22.3 * (CRCL/135)^0.33 captures the 18% CL/F increase observed during pregnancy as a manifestation of the pregnancy-associated 50% rise in estimated glomerular filtration rate; pregnancy itself, gestational age, age, weight, serum creatinine and co-medication were screened but not retained after CLcr inclusion (Valade 2014, BJCP). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valade_2015_emtricitabine.html">Emtricitabine (Valade 2015)</a> </td> <td style="text-align:left;"> Two-compartment oral population PK model for emtricitabine (FTC) in HIV-1-infected men on combined antiretroviral therapy, with an asymmetric effect compartment of negligible volume describing seminal plasma distribution via distinct blood-plasma-to-seminal-plasma transfer rate (k1e) and seminal-plasma elimination rate (ke1) constants (Valade 2015, EVARIST ANRS-EP 49 study) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Steichert_2025_enalapril_enalaprilat_pediatric.html">Enalapril enalaprilat pediatric (Steichert 2025)</a> </td> <td style="text-align:left;"> Simultaneous parent + active-metabolite population PK model for oral enalapril (ODMT) and enalaprilat in ACEi-naive children with heart failure (Steichert 2025, LENA studies). Combined one-compartment model for enalapril (first-order absorption with a lag) coupled with a one-compartment model for enalaprilat via a fixed fraction metabolised fm = 0.7. Allometric scaling (fixed exponents 0.75 on CL, 1 on V) referenced to 5 kg body weight. Covariate effects retained in the final model: age and serum creatinine on the apparent clearance of enalaprilat, and modified Ross score on the apparent volume of distribution of enalaprilat. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Luo_2024_CES1_cirrhosis.html">Enalapril pbpk (Luo 2024)</a> </td> <td style="text-align:left;"> PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). Joint enalapril + enalaprilat disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Enalapril is an inactive ester prodrug hydrolysed by hepatic carboxylesterase 1 (CES1) to the active diacid enalaprilat, which is cleared renally. The semi-PBPK circuit is stomach, three small-intestinal lumen segments (duodenum / jejunum / ileum), the matching three gut-wall segments, portal vein, liver, kidney and a one-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper's virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Barras_2009_enoxaparin.html">Enoxaparin (Barras 2009)</a> </td> <td style="text-align:left;"> Two-compartment first-order absorption population PK model of anti-factor Xa activity in 118 adults (PE / DVT / ACS / atrial fibrillation) receiving subcutaneous enoxaparin treatment doses (1 mg/kg BID by total or lean body weight, 1.5 mg/kg BID for LBW-based obese dosing) under conventional vs lean-body-weight-and-renal-function individualised dosing (Barras 2009 randomised controlled trial). CL is a composite renal + non-renal model with LBW substituted into the Cockcroft-Gault CrCl equation; central volume scales linearly with LBW. The paper additionally reports a three-category proportional-odds bleeding / bruising adverse-event PD model with logit(P[S<=1]) = 2.83 - 2.75*(Age/61) - 0.536*(cAUC/23) and logit(P[S<=2]) = logit(P[S<=1]) + 2.05, driven by patient Age and cumulative AUC (cAUC) of anti-Xa activity from first dose to event. The proportional-odds PD layer is NOT encoded in this model file -- it requires canonical parameter names for cumulative-logit / proportional-odds PD models that are not yet registered in references/parameter-names.md. The PD equation is reproduced in the validation vignette, where it is applied deterministically to cAUC values derived from the simulated PK profile (see vignette Source trace and Assumptions and deviations sections). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Berges_2007_enoxaparin.html">Enoxaparin (Berges 2007)</a> </td> <td style="text-align:left;"> Two-compartment first-order absorption population PK model of anti-factor Xa activity in elderly patients (>75 years) receiving prophylactic subcutaneous enoxaparin 4000 IU once daily (Berges 2007 PROPHRE.75 study) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Feng_2006_enoxaparin.html">Enoxaparin (Feng 2006)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for enoxaparin in adult inpatients receiving continuous intravenous infusion (CII) or subcutaneous (SC) dosing (Feng 2006) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Green_2003_enoxaparin.html">Enoxaparin (Green 2003)</a> </td> <td style="text-align:left;"> Two-compartment first-order-input population PK model for subcutaneous enoxaparin in adults treated at the Royal Brisbane Hospital for acute coronary syndrome, deep vein thrombosis, pulmonary embolism, or DVT prophylaxis (Green & Duffull 2003). Anti-Xa activity is the observation; lean body weight (LBW; James 1976 formula) is the size descriptor on clearance and total body weight is the size descriptor on the central volume. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Green_2005_enoxaparin.html">Enoxaparin (Green 2005)</a> </td> <td style="text-align:left;"> Two-compartment, first-order absorption population PK model for subcutaneously administered enoxaparin (anti-Xa activity) in 38 adults with acute coronary syndromes and a wide range of renal function (Green 2005). Total clearance is the sum of a renal arm scaled linearly to estimated creatinine clearance (CRCL, Cockcroft-Gault with ideal body weight; reference 80 mL/min) and a covariate-free non-renal arm: CL = 0.681 * (CRCL / 80) + 0.229 L/h. Central volume of distribution scales linearly with total body weight (reference 80 kg): Vc = 5.22 * (WT / 80) L. A constant basal anti-Xa activity (49.9 IU/L) is added to the model prediction to represent endogenous and assay-baseline anti-Xa activity, per the Schoemaker parameterisation referenced in the paper. Inter-individual variability is log-normal on total CL, Vc, Q, and basal anti-Xa activity (paper Table 2 Covariate Model). Residual error is combined additive (52.4 IU/L) plus proportional (20.0 percent CV) on observed anti-Xa concentrations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Oualha_2018_enoxaparin.html">Enoxaparin (Oualha 2018)</a> </td> <td style="text-align:left;"> Population PK model for subcutaneous enoxaparin in 22 children during the first post-operative week after paediatric liver transplantation (Oualha 2018). One-compartment open model with first-order absorption (ka fixed at 1/h) and first-order elimination, measured as anti-Xa activity (target 0.2-0.4 IU/mL). Apparent clearance CL/F is allometrically scaled by pre-operative bodyweight BWPREOP (fixed exponent 0.75); apparent central volume V/F is allometrically scaled (fixed exponent 1) by a time-varying post-operative bodyweight BW(t) that captures peri-operative fluid resuscitation followed by post-operative diuresis: BW(t) = (BWPREOP + PFA/1000) * (1 - (1 - fbw) * t^hill_bw / (tbw50^hill_bw + t^hill_bw)). Bodyweight-evolution parameters fbw / hill_bw / tbw50 are jointly estimated with the enoxaparin PK and carry their own between-subject variability. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/SanchezPena_2005_enoxaparin.html">Enoxaparin (SanchezPena 2005)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model of anti-factor Xa activity after intravenous enoxaparin in 546 adults undergoing elective percutaneous coronary intervention (Sanchez-Pena 2005). The IV bolus is modelled as a brief zero-order input phase of duration T0 with linear elimination. Body weight is the only retained covariate, applied as estimated allometric exponents on clearance (0.9) and volume (0.7) with reference 75 kg. A fixed basal anti-Xa activity (0.0725 IU/mL) is added to the dose-driven concentration to account for the endogenous pre-dose background measured by the chromogenic anti-Xa assay. Doses must be entered in IU (1 mg enoxaparin = 100 IU anti-Xa); the typical 0.5 mg/kg clinical dose corresponds to 3830 IU for a 76 kg patient. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schoemaker_1996_low_molecular_weight_heparin_modeling.html">Enoxaparin (Schoemaker 1996)</a> </td> <td style="text-align:left;"> One-compartment population PK model with intravenous bolus input and an estimated constant basal anti-Xa activity for the low molecular weight heparin enoxaparine (trade name Clexane) in healthy volunteers (Schoemaker & Cohen 1996, Example 2 / Table 3, Solution 2). Enoxaparin amount in the central compartment plus an additive endogenous baseline reproduces the lingering low post-dose anti-Xa activity that would otherwise force a second compartment if pre-value subtraction were applied; the authors recommend the basal-activity formulation over the competing two-compartment model (Solution 1, Table 2) because it matches the dose / AUC clearance estimate from the upstream Stiekema 1993 paper. Anti-Xa activity is the surrogate concentration measure; doses are in anti-Xa IU and concentration is in IU/mL. Validation of this model and the companion dalteparin PK/PD model share a single vignette. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Foster_2023_enrofloxacin_cat.html">Enrofloxacin ciprofloxacin cat (Foster 2023)</a> </td> <td style="text-align:left;"> Veterinary (domestic cat). Joint parent + metabolite population PK model for enrofloxacin and its active metabolite ciprofloxacin after a single 5 mg/kg intravenous enrofloxacin dose infused over 30 minutes in 34 client-owned cats hospitalised for clinical illness and spanning normal to markedly reduced kidney function (Foster 2023 Figure S2 and Table 3). Each analyte gets one plasma compartment: enrofloxacin is eliminated by a renal clearance into urine plus a metabolic formation clearance that generates ciprofloxacin, and ciprofloxacin is eliminated by its own clearance into urine. Retained covariates are body weight on the enrofloxacin volume and blood urea nitrogen on the metabolic formation clearance -- the paper's finding that more azotemic cats form ciprofloxacin faster, which the authors themselves argue is unlikely to be causal because serum creatinine and SDMA were not significant. This is a SECOND, independent fit reported alongside the paper's primary total-fluoroquinolone model (Foster_2023_enrofloxacin_totalFluoroquinolone_cat), not a sub-model of it. Two caveats carried from the source: the volumes and clearances are ABSOLUTE (mL, mL/h) despite Table 3's 'mL/kg' labels, and the published clearance pair reproduces a ciprofloxacin fraction of total fluoroquinolone of only about 2% against the ~18% the same paper reports -- see the lvc_cipro / lcl_met comments and the vignette Errata. Between-subject variances and the residual-error magnitude are not reported, so every eta and residual SD is fixed(0) and the model simulates typical-value profiles unless the user supplies variances. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Foster_2023_enrofloxacin_cat.html">Enrofloxacin totalFluoroquinolone cat (Foster 2023)</a> </td> <td style="text-align:left;"> Veterinary (domestic cat). Two-compartment population PK model for TOTAL FLUOROQUINOLONE (the arithmetic sum of plasma enrofloxacin and its active metabolite ciprofloxacin, which Foster 2023 treats as one analyte because the two compounds' antibacterial effects are additive) after a single 5 mg/kg intravenous enrofloxacin dose infused over 30 minutes in 34 client-owned cats hospitalised for clinical illness and spanning normal to markedly reduced kidney function. Sparse sampling (3 samples per cat, 98 samples total) with nonlinear mixed-effects modelling in Phoenix NLME. Body weight, centred at the population value of 3.8 kg, is the only retained covariate and acts as a power effect on both the central volume and the elimination clearance; serum creatinine, symmetric dimethylarginine, blood urea nitrogen, age and sex were screened and none was significant, which is the paper's central finding (no dose adjustment is indicated for azotemic cats). Foster 2023 Table 1 labels the volumes 'mL/kg' and the clearances 'mL/kg/h', but those parameters are ABSOLUTE (mL and mL/h), not weight-normalised: only the absolute reading reproduces the paper's own Table 1 AUC0-24 of 46.0 ug*h/mL and the concentration range of Figure 1 -- see the lvc comment and the vignette Errata for the arithmetic. Between-subject variances and the residual-error magnitude are not reported anywhere in the paper or its supplement, so every eta and propSd is fixed(0) and the model simulates typical-value profiles unless the user supplies variances. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Csajka_2005_ephedrine_caffeine.html">Ephedrine caffeine (Csajka 2005)</a> </td> <td style="text-align:left;"> Mechanistic simultaneous population PK model for co-administered ephedrine, its N-demethylation metabolite norephedrine, and caffeine in healthy adults after single oral doses (Csajka 2005). Caffeine is described by a 1-compartment first-order-absorption model with a fractional decrease in apparent clearance during oral contraceptive therapy. Ephedrine uses a 1-compartment depot + central + cumulative-urine model with an absorption lag time, renal clearance, and saturable Michaelis-Menten conversion to norephedrine; norephedrine is carried as a pseudo-concentration state because its volume of distribution V_NE is unidentifiable, so the reported parameter is the compound Vmax/V_NE and the norephedrine elimination is first order. The interaction term reproduces the paper's indirect-action absorption model (equation 10b/10e final form): the caffeine amount in the absorption compartment depresses ephedrine ka by an asymptotic fraction d, with caffeine acting as the f(C) inhibitor on its own absorption-compartment amount. Parameter values are the pharmaceutical-formulation defaults from Table 3; herbal-formulation alternatives (bioavailability F_E,herbal = 0.78 instead of F_E,pharm = 0.59, plus a 22.2-min caffeine absorption lag) are documented in inline comments and can be applied by overriding lfdepot and ltlag_caf at simulation time. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sarashina_2005_epinastine.html">Epinastine (Sarashina 2005)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption for oral epinastine in healthy adults and paediatric atopic dermatitis patients (Sarashina 2005), with linear-in-WT CL/F and V1/F plus food-status and formulation covariate effects </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Abboud_2009_epinephrine.html">Epinephrine (Abboud 2009)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous epinephrine (adrenaline) infusion in adults with septic shock, with a constant endogenous epinephrine production rate (R0) feeding the central compartment and body weight and SAPS II severity score as power covariates on clearance (Abboud 2009). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Oualha_2014_epinephrine.html">Epinephrine (Oualha 2014)</a> </td> <td style="text-align:left;"> Population PK/PD model for continuous IV epinephrine in critically ill children following cardiopulmonary bypass for repair of congenital heart defects (Oualha 2014). One-compartment open PK with first-order elimination plus an endogenous zero-order production rate q0 and circulating-volume-anchored Vc = 0.08*WT; allometric scaling of CL and q0 on body weight (exponents fixed to 3/4). Hemodynamic Emax sub-models for heart rate (HR) and the stroke-volume * systemic-vascular-resistance product (SV*SVR) with age power effects on basal HR and SV*SVR and a RACHS-1 categorical effect on SV*SVR_max. Glucose/lactate turnover sub-model: epinephrine stimulates the zero-order plasma glucose production rate via an Emax function; plasma lactate is produced at the rate of glucose elimination and itself follows first-order elimination. kGLY and kLAC are derived at steady state (Eq. 12-13). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Janssen_2023_pregnancy_cytotoxics.html">Epirubicin (Janssen 2023)</a> </td> <td style="text-align:left;"> Semi-physiological enriched three-compartment population PK model for intravenous epirubicin in pregnant cancer patients, applying gestational changes in albumin binding, glomerular filtration, hepatic plasma flow, CYP3A4 activity and body fluid volumes to the non-pregnant Sandstrom 2006 base model </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Diep_2022_eplontersen.html">Eplontersen (Diep 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK and indirect-response PD model for the GalNAc3-conjugated antisense oligonucleotide eplontersen targeting transthyretin (TTR) mRNA, fit to pooled data from two phase 1 studies in healthy volunteers (Diep 2022). First-order SC absorption with site-specific typical ka (arm vs abdomen), allometric scaling on CL by lean body mass, on Vc/Q/Vp by total body weight, and an indirect-response model with eplontersen-driven inhibition of TTR production. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PerezRuixo_2008_epoetinAlfa.html">EpoetinAlfa (PerezRuixo 2008)</a> </td> <td style="text-align:left;"> Population PK/PD model for subcutaneous recombinant human erythropoietin (rHuEPO / epoetin alfa) in healthy adult male volunteers (Perez-Ruixo 2008). PK is the Olsson-Gisleskog 2007 prior (two-compartment with linear + Michaelis-Menten elimination and dual subcutaneous absorption: a fast sequential zero-order infusion into the depot of duration D1 feeding first-order absorption ka into central, plus a slower zero-order direct infusion into central of duration D2 after lag time tlag2; dose-dependent absolute bioavailability F = F0 + Emax(F)*Dose/(ED50(F)+Dose)). Endogenous EPO is maintained at the baseline BSL by a constant input rate kEPO derived from the steady-state balance against linear + MM elimination (equation 4). The PD layer is the maturation-structured cytokinetic model D: rHuEPO stimulates the progenitor production rate kin*C/(SC50+C) into a 10-stage bone-marrow precursor age chain (transfer rate Np/Tp), which feeds a 10-stage circulating reticulocyte age chain whose transfer rate (NR/TR)*(S0/SM) is inhibited by a 5-stage signal transduction (transit time tau) driven by C/(EC50+C). Output RET = sum of reticulocyte compartments reproduces the percentage of reticulocytes in % units. No demographic covariate effects were retained in either layer. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hayashi_1998_epoetinBeta.html">EpoetinBeta (Hayashi 1998)</a> </td> <td style="text-align:left;"> One-compartment population PK model for subcutaneous recombinant human erythropoietin (epoetin beta) in healthy adult male Japanese volunteers with a constant endogenous EPO production rate carrying a fixed circadian sinusoid (acrophase near midnight) feeding the central compartment, and body weight as a power covariate on apparent absorption rate ka and apparent central volume V/F, plus serum creatinine and age as power covariates on the elimination rate constant k_e (reparameterised here onto canonical CL/F so the k_e covariates ride on CL/F together with the V/F weight exponent); apparent V/F and E/F throughout because bioavailability was not separately estimable from this SC-only study (Hayashi 1998). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kawamura_2018_eribulin.html">Eribulin (Kawamura 2018)</a> </td> <td style="text-align:left;"> Three-compartment IV PK driver coupled with a Friberg-style semi-mechanistic PD model for eribulin-induced neutropenia in Japanese patients with recurrent or metastatic breast cancer (Kawamura 2018). Plasma eribulin concentrations are produced by a 3-compartment model with linear elimination from the central compartment whose parameters are FIXED from the Majid 2014 popPK analysis (reproduced verbatim in Kawamura 2018 section 2.3): CL depends on body weight (allometric 0.75), serum albumin, alkaline phosphatase, and total bilirubin; V1, V2, V3 scale linearly with body weight; Q2 and Q3 scale allometrically with body weight. The PD layer (proliferation + three transit compartments + circulating neutrophils + feedback) is estimated on 401 patients / 5199 ANC measurements (Table 2): MTT = 104.5 h, Kprol = 0.0377 /h, Kout = 0.0295 /h, Gamma = 0.203, Slope = 0.0413 mL/ng (linear drug effect). Serum albumin influences Kprol (negative exponent), MTT (positive exponent), and Kout (positive exponent); a binary low-baseline-ANC indicator (BNEU3 = 1 when baseline ANC < 3000/uL) multiplies Kprol. IIV is reported on Kprol, Kout, and Slope (no IIV on MTT or Gamma). Additive residual error on circulating ANC (sigma = 1.15 cells/nL = 1150 cells/uL). Eribulin doses must be supplied in milligrams of eribulin-FREE-BASE equivalent (1.4 mg/m^2 mesilate = 1.23 mg/m^2 free base, conversion factor 1.23/1.4). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vanHasselt_2013_eribulin.html">Eribulin (vanHasselt 2013)</a> </td> <td style="text-align:left;"> Three-compartment IV PK driver coupled with a Friberg-style semi-physiological PD model for eribulin-induced neutropenia in adult patients with late-stage metastatic breast cancer (van Hasselt 2013). The PK layer parameters are FIXED from the Majid 2014 popPK analysis (J Clin Pharmacol 54:1134); van Hasselt 2013 cites the same Eisai popPK analysis as unpublished data on file and does not report PK parameter values in the paper. CL depends on body weight (allometric 0.75), serum albumin, alkaline phosphatase, and total bilirubin; V1, V2, V3 scale linearly with body weight; Q2 and Q3 scale allometrically with body weight. The PD layer (proliferation + three transit compartments + circulating neutrophils + feedback) is estimated on 1579 patients / 23 427 ANC measurements from a pool of 12 phase I / II / III studies (Table 3 final covariate model): ANC0 = 4.03e9 cells/L, MTT = 109 h, Gamma = 0.216, SLOPE = 0.0451 L/ug (linear drug effect). Albumin, bilirubin, alkaline phosphatase, LDH, and G-CSF-receival covariates enter MTT; albumin, AST, and G-CSF-receival covariates enter SLOPE. IIV on ANC0, MTT, Gamma, and SLOPE (diagonal; off-diagonal covariances were not computationally feasible per the paper). Proportional residual error on circulating ANC. Eribulin doses must be supplied in milligrams of eribulin FREE BASE (conversion factor 1.23/1.4 from the mesilate dose). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vanHasselt_2015_eribulin.html">Eribulin (vanHasselt 2015)</a> </td> <td style="text-align:left;"> Disease-progression (DP) model for prostate-specific antigen (PSA) dynamics in metastatic castration-resistant prostate cancer (CRPC) patients treated with eribulin mesilate (van Hasselt 2015). K-PD framework: the per-dose predicted eribulin AUC enters a single transient drug-effect compartment depot_kpd that decays with rate KP (fixed to 6000 /day so the effect is nearly instantaneous after each dose); PSA evolves under a first-order growth rate KG counteracted by an inhibition rate KD0 multiplied by the K-PD state depot_kpd and an exponentially decaying resistance factor exp(-k_res*t). PSA0, KD0, KG, k_res have correlated lognormal IIV; proportional residual error on PSA (log-transform-both-sides). Prior taxane treatment (binary PRIOR_TAXANE) multiplies PSA0; cumulative number of days of prior taxane treatment (continuous PRIOR_TAXANE_DAYS) enters KD0 as (1 + NTRT/720)^theta. The companion parametric Weibull survival sub-model fit in R survreg is documented in the vignette but not encoded here (not an ODE / nlmixr2 structure). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Saleh_2023_leicns_pk30_mouse_brain.html">Erlotinib mouse pbpk (Saleh 2023)</a> </td> <td style="text-align:left;"> PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, FVB). Nine-compartment CNS PBPK model predicting unbound erlotinib concentrations in brain extracellular fluid (brain ECF) after a single 50 mg/kg oral dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug's physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.628 (estimated by the authors; Table V). The plasma PK model is the empirical one-compartment model of Table IV (in-house NONMEM fit) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Eyler_2014_ertapenem.html">Ertapenem (Eyler 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous ertapenem in critically ill adults with acute kidney injury receiving continuous venovenous hemodialysis (CVVHD) or hemodiafiltration (CVVHDF). PK is parameterised on unbound drug; total serum concentrations are reconstructed via a single-site saturable albumin-binding equation Cb = Bmax * Cu / (KD + Cu). Systemic (body) clearance and a separate dialytic clearance arm are estimated as primary parameters; the dialytic arm is added to body clearance only while the CRRT circuit is running, gated by the time-varying RRT_HEMODIAL_ACTIVE covariate. Eyler 2014, n = 8 subjects, single 1 g IV dose over 30 min. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lakota_2018_ertapenem.html">Ertapenem (Lakota 2018)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for ertapenem in adults across a wide range of body sizes (Lakota 2018) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LopezArguello_2023_pbp_binding_pseudomonas.html">Ertapenem qsp (LopezArguello 2023)</a> </td> <td style="text-align:left;"> QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for ertapenem (carbapenem): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nielsen_2011_antibacterial_efficacy.html">Erythromycin (Nielsen 2011)</a> </td> <td style="text-align:left;"> In vitro (Streptococcus pyogenes M12 NCTC P1800). Semimechanistic PKPD model of erythromycin time-kill kinetics; two-stage bacterial life-cycle (proliferating drug-sensitive S and non-growing drug-insensitive R) with sigmoidal Emax killing of S via an effect compartment; first-order drug elimination (ke set per in vitro kinetic-system flow rate); drug-specific degradation kdeg fixed at zero. Parameter values are from the combined static and dynamic estimation in Table 3. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Areberg_2006_escitalopram.html">Escitalopram (Areberg 2006)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and lag time for escitalopram in healthy and hepatic-impaired adults (Areberg 2006) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PerezRuixo_2020_esketamine.html">Esketamine (PerezRuixo 2020)</a> </td> <td style="text-align:left;"> Joint three-compartment esketamine + two-compartment apparent noresketamine population PK model with a hepato-portal first-pass compartment (well-stirred model) and three parallel absorption routes (intranasal direct, intranasal-swallowed via PO depot, and PO solution via PO depot) developed from 9784/9397 esketamine/noresketamine plasma observations in 820 healthy volunteers and patients with treatment-resistant depression receiving intranasal, intravenous, and oral esketamine (Perez-Ruixo 2020). Asian race decreases esketamine kel (x0.36) and noresketamine apparent CLn/F (x0.81); Japanese race increases the nasal-cavity-absorbed fraction FRn (x1.34); and hepatic blood flow Qh declines linearly by 2.19 L/h per year of age above 60. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fang_2010_etanercept.html">Etanercept (Fang 2010)</a> </td> <td style="text-align:left;"> One-compartment population PK model for rhTNFR-Fc (recombinant human TNF receptor-Fc fusion protein; etanercept-class molecule from Celgen Bio-Pharmaceutical) with first-order subcutaneous absorption, absorption lag time, and linear elimination in healthy Chinese volunteers (single SC doses 12.5-50 mg) and Chinese male patients with ankylosing spondylitis (multiple SC doses 25 mg BIW or 50 mg QW) (Fang 2010). Female sex is the typical-value reference: males have 0.655x lower CL/F. Single-dose administration is the typical-value reference: multi-dose administration in AS patients has 0.674x lower apparent bioavailability F. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Horita_2018_ethambutol.html">Ethambutol (Horita 2018)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model with zero-order absorption (lag time + zero-order duration) and first-order elimination for oral ethambutol in Ghanaian children with active tuberculosis (Horita 2018); allometric weight scaling on CL/F, Q/F, V1/F, V2/F with non-canonical estimated exponents (0.382, 0.474, 0.228, 0.858) normalised to the cohort median 14.3 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jonsson_2011_ethambutol.html">Ethambutol (Jonsson 2011)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral ethambutol in adult South African pulmonary tuberculosis patients (Jonsson 2011), with one transit compartment preceding first-order absorption, allometric scaling on clearance (3/4) and volume (1) terms relative to a 50 kg reference, an HIV-status effect on bioavailability (15.4% reduction), and 4-occasion inter-occasion variability on apparent oral clearance. Parameter values are taken from the publication's Table 2 (NONMEM final estimates column); see inst/modeldb/ddmore/Jonsson_2011_ethambutol_ddmore.R for the DDMoRE-bundle replicate of the same fit. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ramachandran_2023_tuberculosis_eptb_pbpk.html">Ethambutol pbpk (Ramachandran 2023)</a> </td> <td style="text-align:left;"> PBPK (whole-body, 18 ODEs, MATLAB ode15s). Oral ethambutol disposition at extrapulmonary tuberculosis (EPTB) sites in a 70-kg reference adult male (Ramachandran and Gadgil 2023, CPT Pharmacometrics Syst Pharmacol). Seventeen perfusion-limited well-stirred tissue compartments connected by blood and lymph flow, plus a gut-lumen state for enterohepatic recycling. The model's two novel states are a pleural-fluid compartment (a filtrate of the lung that drains via lymphatics) and a consolidated lymph-node compartment that collects afferent lymph from every organ except bone and spleen and returns it to venous blood - the two most common EPTB sites. Physiological volumes, blood flows, and lymph flows are fixed literature fractions of body weight, cardiac output (5200 mL/min), and afferent lymph flow (8 L/day) respectively (Appendix S1 Tables S2 and S3). Tissue:plasma partition coefficients were computed by the Rodgers and Rowland method from the drug physicochemistry in Table S4 and are tabulated in Table S5. Only the first-order oral absorption rate ka and total systemic clearance CL were estimated, by weighted least squares against reported plasma concentrations after a 400 mg oral dose (Appendix S1 Table S7). Clearance is split into a renal component fR * CL driven by arterial concentration and a hepatic component (1 - fR) * CL driven by the hepatic inlet concentration; the hepatic output enters the gut lumen, which for ethambutol is a terminal sink cleared only by faecal transit at kF = 0.252 /h (the paper assigns a non-zero gut reabsorption rate only to rifampicin). The model is a deterministic typical-value simulation: the paper reports no interindividual variability and no residual error model, so propSd is fixed at 0. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nemoto_2017_ethanol.html">Ethanol (Nemoto 2017)</a> </td> <td style="text-align:left;"> Bayesian population PK model for orally ingested ethanol (alcohol) in 34 healthy Japanese adults (Nemoto 2017). One-compartment model with first-order absorption and Michaelis-Menten elimination; covariates: sex, age, body weight, ALDH2 and ADH1B genotypes. Final model fit by a fully conditional MCMC Bayesian analysis with informative priors derived from Seng et al. 2014 (Chinese + Indian cohort). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ye_2017_ethaselen.html">Ethaselen (Ye 2017)</a> </td> <td style="text-align:left;"> Preclinical (mouse, BALB/c nude with A549 NSCLC xenograft). Integrated dose-biomarker-response PD model for the thioredoxin reductase (TrxR) inhibitor ethaselen (Ye et al. 2017). The TrxR biomarker is described by an indirect-response (IDR) turnover in which the zero-order production Kin is linearly amplified by the instantaneous natural tumor growth rate (linear correction factor gamma1) and the first-order degradation Kout is increased by a sigmoidal Emax function of the current administered ethaselen dose (Smax, SC50, Hill = gamma2). Tumor volume follows a smooth exponential-to-linear growth law (paper Eq 5: dX/dt = 2*lambda0*lambda1*X / (lambda1 + 2*lambda0*X)) tempered by a zero-order Emax killing rate driven by the TrxR-inhibition ratio P = 1 - TrxR_treatment / TrxR_control (paper Eq 7). The control TrxR trajectory is carried internally as a shadow state (trxr_ctrl) so P is defined per-subject without requiring an external control-arm simulation. No pharmacokinetic compartment is included; the paper acknowledges ethaselen plasma concentrations were not measured. The current daily dose enters the model through the time-varying covariate DOSE (mg/kg/day), which the published study toggles between 0 (vehicle / off-treatment) and one of {36, 72, 108} mg/kg/day for days 0-9 (oral gavage QD x10 d). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Freyer_2000_AVI_regimen.html">Etoposide (Freyer 2000)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for etoposide in 24 small cell lung cancer patients on the AVI regimen (Freyer 2000), with a linear serum creatinine effect on clearance. Structural reduction from the paper's two-compartment model: only central CL and V were reported in Table 2, so the peripheral compartment is omitted. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Toffoli_2001_etoposide.html">Etoposide (Toffoli 2001)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral and IV etoposide in adult patients with solid tumours (Toffoli 2001). Additive-linear creatinine-clearance covariate on CL from Equation 2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Moein_2022_etrolizumab.html">Etrolizumab (Moein 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for etrolizumab with first-order SC absorption and time-decreasing clearance in adults with moderately-to-severely active ulcerative colitis (Moein 2022) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/deWit_2016_everolimus.html">Everolimus (deWit 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order oral absorption for everolimus 10 mg once-daily in 40 adult patients with advanced thyroid carcinoma (de Wit 2016). Bioavailability F is structurally fixed at 1 (absolute F unknown), so reported CL, V1, Q, and V2 are apparent (oral / F). Allometric scaling on apparent clearance (exponent 0.75) and apparent central volume (exponent 1.0) using a 70 kg reference weight per the Anderson and Holford theory cited by the paper. Apparent peripheral volume V2/F was held fixed at 400 L in the final model. Bioavailability is multiplied by 0.792 in subjects who carry at least one ABCB1 TTT haplotype (CYP3A / P-gp efflux marker). Inter-occasion variability on F captures the day-1-vs-day-15 sampling occasion contrast (CV 19.2%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/TerHeine_2018_everolimus.html">Everolimus (TerHeine 2018)</a> </td> <td style="text-align:left;"> Semi-mechanistic two-compartment population PK model for everolimus in pooled adult oncology (metastatic thyroid or breast cancer) and renal transplant patients (ter Heine 2018). Oral absorption is modelled with a chain of four transit compartments parameterised by the mean absorption time MAT and the Savic 2007 convention ktr = (n + 1) / MAT (n = 4 transit compartments). Hepatic disposition uses a well-stirred liver model: hepatic plasma flow QHP = QH * (1 - HCT); hepatic extraction EH = fu * CLint / (QHP + fu * CLint) with FIXED unbound fraction fu = 0.27; oral bioavailability F = 1 - EH and systemic plasma clearance CLH = QHP * EH. Volume parameters (VC, VP) and flow parameters (QH = 90 L/h FIXED, Q) are allometrically scaled to fat-free mass FFM at a 57.2 kg reference (equivalent to a 70 kg, 1.80 m adult male) with theory-based exponents 0.75 on flows and 1.0 on volumes (Anderson and Holford). Concomitant high-dose oral prednisolone (PRED_DOSE >= 20 mg/day, a CYP3A4 inducer) increases apparent CLint by 31%. Modelled plasma concentrations were derived externally from observed whole-blood concentrations and HCT via a Langmuir-plus-linear erythrocyte binding model (Bmax = 0.964 mg/L, Kd = 0.0920 mg/L, Kns = 0.153); the vignette uses the same back-calculation to compare simulated plasma concentrations against the paper's whole-blood trough targets. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pu_2021_evinacumab.html">Evinacumab (Pu 2021)</a> </td> <td style="text-align:left;"> Population PK/PD model for evinacumab in healthy volunteers and adults / pediatric patients with homozygous familial hypercholesterolemia (Pu 2021): two-compartment PK with first-order SC absorption (with lag time) and parallel linear plus Michaelis-Menten elimination from the central compartment, linked to a Type 1 indirect-response model for low-density lipoprotein cholesterol (LDL-C) where evinacumab inhibits LDL-C production. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kuchimanchi_2018_evolocumab.html">Evolocumab (Kuchimanchi 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model for evolocumab with first-order SC absorption and parallel linear plus Michaelis-Menten (target-mediated) elimination from the central compartment, in healthy adults and patients with hypercholesterolemia (Kuchimanchi 2018) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kuchimanchi_2018_evolocumab.html">Evolocumab ldlc (Kuchimanchi 2018)</a> </td> <td style="text-align:left;"> Joint population PK + static Emax-on-AUC exposure-response model for evolocumab LDL-C lowering in adults with hypercholesterolemia (Kuchimanchi 2018). The PK layer (Table 3) is the one-compartment model with parallel linear and Michaelis-Menten elimination and SC bioavailability from the companion Kuchimanchi_2018_evolocumab.R file. The PD layer (Table 4) is an algebraic Emax model linking AUC over weeks 8-12 of dosing to the mean week-10-and-12 LDL-C reduction, with statin / ezetimibe / HeFH covariate effects on baseline LDL-C, a statin covariate effect on Emax, and a regimen-effect multiplier on EC50 distinguishing once-monthly (QM) from once-every-2-weeks (Q2W) dosing. AUC of evolocumab is integrated inside an extra rxode2 state over the 56-84-day window; the LDLC observable is meaningful only at t >= 84 (vignette documents the time-window discipline). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2024_evorpacept_qsp.html">Evorpacept qsp (Wang 2024)</a> </td> <td style="text-align:left;"> QSP. Quantitative systems pharmacology platform for immuno-oncology in advanced non-small cell lung cancer (NSCLC), extended with a pharmacokinetic and pharmacodynamic module for the anti-CD47 antibody evorpacept (ALX-148) in combination with the anti-PD-L1 antibody durvalumab. Four physiological compartments (central, peripheral, tumour, tumour-draining lymph node) plus an endothelial endosomal compartment for FcRn recycling, APC endosomal / surface compartments and three immunological-synapse compartments (T-cell:cancer-cell, T-cell:APC and macrophage:cancer-cell). The evorpacept module adds target-mediated drug disposition through CD47 on red blood cells with cross-arm (bivalent) binding, FcRn-mediated recycling, and CD47-SIRP-alpha plus PD-1/PD-L1 control of macrophage phagocytosis. The surrounding platform covers naive and activated CD4+/CD8+ T cell trafficking and proliferation, logistic tumour growth, antigen release, uptake, endosomal processing and MHC-I presentation for a self-antigen and a neo-antigen, TCR kinetic proofreading, PD-1/PD-L1/PD-L2/CTLA-4/CD28/CD80/CD86 checkpoint binding, M1/M2 macrophage recruitment and polarisation, MDSC recruitment with arginase-I and nitric-oxide suppression, Th-to-Treg transdifferentiation, and TGF-beta / IFN-gamma / IL-2 / IL-10 / IL-12 / CCL2 dynamics. 153 ODE states, 6 algebraic species, 40 repeated-assignment rules and 276 parameters, translated from the SimBiology export in Supplementary Tables S2-S6 and cross-checked against the authors' MATLAB source (Zenodo doi:10.5281/zenodo.11093921). Deterministic mechanism model: the authors generated virtual patients by sampling the parameter distributions in Supplementary Table S1 rather than by fitting IIV or residual error, so no etas and no error model are encoded. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valle_2005_exemestane.html">Exemestane (Valle 2005)</a> </td> <td style="text-align:left;"> Three-compartment population PK with first-order absorption + lag time, coupled to an indirect-response PD model on plasma estrone sulphate (E1S), for oral exemestane (25 mg single dose) in healthy postmenopausal women. Crossover study comparing a sugar-coated tablet (SCT) under fasting versus an extemporaneous tablet-suspended-in-water suspension under fasting versus a SCT taken after a standard high-fat breakfast. Disposition is independent of formulation and food; absorption rate ka and apparent bioavailability F depend on formulation (suspension: ka 7.6 vs SCT 2.35 1/h, F 1.2x) and on the high-fat meal (ka 1.13 1/h, F 1.6x). Exemestane inhibits E1S synthesis via a sigmoid Imax function with IC50 22.1 pg/mL and Hill coefficient 1.73. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cirincione_2017_exenatide.html">Exenatide (Cirincione 2017)</a> </td> <td style="text-align:left;"> Population PK model for exenatide immediate-release (Cirincione 2017): two-compartment, parallel linear and Michaelis-Menten elimination, sequential zero-order then saturable first-order absorption after SC dosing. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cirincione_2017_exenatide_er.html">Exenatide er (Cirincione 2017)</a> </td> <td style="text-align:left;"> Population PK model for extended-release (ER) microsphere SC exenatide in patients with type 2 diabetes (Cirincione 2017 AAPS J): two-compartment disposition with three parallel SC-absorption processes (initial first-order release plus two Savic 2007 analytical transit-compartment chains for the second- and third-phase microsphere release) and parallel linear plus saturable Michaelis-Menten elimination. Disposition parameters (CL, Q, Vc, Vp, Vmax, Km) and the eGFR-on-CL and WT-on-Vc covariate effects are fixed from the IR companion model (Cirincione 2017 BJCP). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ng_2018_exendin939.html">Exendin939 (Ng 2018)</a> </td> <td style="text-align:left;"> Two-compartment intravenous-infusion population PK model for exendin-(9-39) in patients with congenital hyperinsulinism (Ng 2018). Pooled paediatric (neonates and children) and adult cohort with allometric scaling fixed at 0.75 on CL and Q and 1.0 on Vc and Vp (reference WT 70 kg); inter-individual variability retained only on CL. Residual variability follows the NONMEM Poisson error model (Var(Y|F) = F * sigma^2), encoded as a power-error with fixed exponent 0.5. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hazendonk_2016_factor_viii.html">Factor viii (Hazendonk 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for factor VIII (FVIII) concentrates in severe and moderate hemophilia A patients (adults and children, FVIII plasma concentration < 0.05 IU/mL) undergoing elective / minor / major surgery (Hazendonk 2016). PK parameters are allometrically scaled to a 68 kg reference body weight with fixed exponents of 0.75 on clearances and 1.0 on volumes; typical CL, V1, Q, V2 at the reference body weight are 150 mL/h, 2810 mL, 160 mL/h, 1900 mL. Clearance carries three covariate effects (age with power exponent -0.17 centered at 40 years; +26% for blood group O; -7% for a major surgical procedure) and central volume carries an age effect (power exponent -0.09 centered at 40 years). B-domain-deleted recombinant FVIII products (Refacto AF) are under-detected by the one-stage clotting assay by a fixed 34%, encoded as a multiplicative correction on the predicted concentration. IIV on CL and V1 is 37% and 27% (exponential); IIV on Q and V2 was not estimable. Residual error is combined (proportional 18% + additive 0.15 IU/mL) using the values reported for the majority center cluster (centers 1, 2, 3). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Noe_1996_factor_viii.html">Factor viii (Noe 1996)</a> </td> <td style="text-align:left;"> Deterministic mechanistic model of coagulation factor VIII kinetics with von Willebrand factor (vWF) binding equilibrium in adult humans; captures endogenous synthesis, reversible binding to vWF, and differential elimination of the free and vWF-bound forms of factor VIII </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Brekkan_2016_factorIX.html">FactorIX (Brekkan 2016)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for plasma-derived factor IX (FIX) activity in patients with moderate or severe haemophilia B, developed by Brekkan et al. 2016 to support pharmacokinetic dose individualisation. Disposition is described by linear three-compartment kinetics with intravenous input and first-order elimination from the central compartment; allometric body-weight scaling on CL/Q (0.75) and V1/V2/V3 (1.0) is fixed with a reference weight of 70 kg, and an endogenous baseline FIX activity is estimated as a structural parameter. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Koopman_2023_factorix.html">Factorix (Koopman 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for recombinant factor IX-Fc fusion concentrate (rFIX-Fc, eftrenonacog alfa) in haemophilia B patients aged 2-71 years (Koopman 2023) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviia dog (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (beagle dog). Two-compartment population PK model for activated recombinant factor VII (rFVIIa) in male beagle dogs following single IV bolus administration, from the Larsen 2018 preclinical scaling programme (n=10). Body-weight-normalised allometric scaling (exponents 1 for V/V2, 0.75 for CL/Q; scaling principle I) is applied within-species around the median weight of 12.8 kg. IIV is estimated on V and CL. Endogenous rFVIIa was not detectable in dogs. Parameter values from Larsen 2018 Table 2 (Dog column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviia monkey (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (cynomolgus monkey). Two-compartment population PK model for activated recombinant factor VII (rFVIIa) in cynomolgus monkeys following single IV bolus administration, from the Larsen 2018 preclinical scaling programme (n=27; 16 males and 11 females). Endogenous rFVIIa was detectable in monkeys and is carried as an additive baseline (0.287 IU/mL) on the observation. Body-weight- normalised allometric scaling (exponents 1 for V/V2, 0.75 for CL/Q; scaling principle I) is applied within-species around the median weight of 2.78 kg. IIV is estimated on V, CL, and the endogenous baseline. Parameter values from Larsen 2018 Table 2 (Monkey column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviia mouse (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (mouse). Two-compartment population PK model for activated recombinant factor VII (rFVIIa) in male C57BI/6 and NMRI mice following single IV bolus administration, developed as part of the Larsen 2018 preclinical scaling programme (n=51 mice pooled across two strains). Strain enters as a fractional multiplier on both V and CL (C57BI/6 = 0.588 x V, 0.87 x CL relative to the NMRI reference), and body-weight-normalised allometric scaling (exponents 1 for V/V2, 0.75 for CL/Q; scaling principle I) is applied within-species around the median weight of 0.026 kg. IIV is estimated only on CL. Endogenous rFVIIa was not detectable in mice. Parameter values from Larsen 2018 Table 2 (Mouse column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviia rat (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment population PK model for activated recombinant factor VII (rFVIIa) in male Sprague-Dawley rats following single IV bolus administration, from the Larsen 2018 preclinical scaling programme (n=37 rats). Body-weight-normalised allometric scaling (exponents 1 for V/V2, 0.75 for CL/Q; scaling principle I) is applied within-species around the median weight of 0.24 kg. IIV is estimated on V and CL. Endogenous rFVIIa was not detectable in rats. Parameter values from Larsen 2018 Table 2 (Rat column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nestorov_2014_factorviii.html">Factorviii (Nestorov 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for recombinant factor VIII Fc fusion protein (rFVIIIFc, efmoroctocog alfa) in previously treated patients with severe hemophilia A (Nestorov 2014; final covariate model with VWF on CL and WT and HCT on V1) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valke_2024_factorviii_thrombin_plasmin.html">Factorviii (Valke 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for factor VIII (FVIII) activity (IU/dL) after a single intravenous bolus of plasma-derived von Willebrand factor / factor VIII concentrate (pdVWF/FVIII, Humate-P) in 29 adults with severe hemophilia A (Valke 2024). The model was newly estimated after the earlier Bukkems 2022 popPK model over-predicted FVIII activity in this cohort (MPE 60.9%). CL, V1, Q and V2 are allometrically scaled to a 70 kg reference with exponents fixed at 0.75 on clearances and 1 on volumes; typical values are CL 3.07 dL/h, V1 39.1 dL, Q 1.09 dL/h, V2 9.16 dL, giving a terminal half-life of 12.7 h. Clearance is 1.53x higher in patients with a Nijmegen-modified Bethesda assay inhibitor. FVIII activity was measured by both the one-stage clotting assay (OSA, the reference scale of this model) and the chromogenic substrate assay (CSA); CSA samples read 0.939x the OSA value and carry their own residual-error magnitudes. IIV on CL (57.3%) and V1 (38.2%) is correlated at 77.0%. Residual error is combined proportional plus additive, with assay-specific magnitudes. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviii dog (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (haemophilia A dog). Two-compartment population PK model for B-domain truncated recombinant factor VIII (rFVIII) in haemophilia A dogs (1 male, 2 female) following single IV bolus administration, from the Larsen 2018 preclinical scaling programme (n=3). Body- weight-normalised allometric scaling (exponents 1 for V/V2, 0.75 for CL/Q; scaling principle I) is applied within-species around the median weight of 19.55 kg. No IIV was estimated (only 3 subjects). Endogenous FVIII is not present (haemophilia A background). Parameter values from Larsen 2018 Table 3 (Dog column). Note: Q is estimated with high RSE (83%), reflecting the small sample size. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chelle_2019_factorviii_fanhdi.html">Factorviii fanhdi (Chelle 2019)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for Fanhdi/Alphanate (plasma-derived factor VIII concentrate, Grifols) in hemophilia A patients pooled from 12 hemophilia centers in the WAPPS-Hemo platform (Chelle 2019). Final model has fat-free mass (FFM) as a power-form covariate on CL, V1, and V2, and a piecewise-linear age effect on CL above the median age of 25 years; between-subject variability is a BLOCK(2) on CL and V1 with correlation 0.797; residual error is proportional only. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviii monkey (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (cynomolgus monkey). One-compartment population PK model for B-domain truncated recombinant factor VIII (rFVIII) in male cynomolgus monkeys with concentration-dependent (nonlinear) clearance, from the Larsen 2018 preclinical scaling programme (n=35). Clearance rises exponentially with total FVIII activity (Larsen 2018 Eq. 1: CL(C) = CL_0 * exp(beta_cl * C)) to capture saturation of the von Willebrand factor buffer at supraphysiological doses. Body-weight-normalised allometric scaling (exponents 1 for V, 0.75 for CL; scaling principle I) is applied within- species around the median weight of 2.96 kg. IIV is estimated on CL and the endogenous baseline. Endogenous FVIII (1.75 IU/mL) is carried as an additive baseline on the total observed activity. Parameter values from Larsen 2018 Table 3 (Monkey column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviii mouse (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (mouse). One-compartment population PK model for B-domain truncated recombinant factor VIII (rFVIII) in FVIII-knockout (FVIII KO) mice following single IV bolus administration, from the Larsen 2018 preclinical scaling programme (n=132; 71 males, 61 females). Sex enters as a fractional multiplier on CL (females clear rFVIII 24% faster than males), and body- weight-normalised allometric scaling (exponents 1 for V, 0.75 for CL; scaling principle I) is applied within-species around the median weight of 0.021 kg. IIV is estimated on CL. Endogenous FVIII is not present (KO). Parameter values from Larsen 2018 Table 3 (Mouse column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valke_2024_factorviii_thrombin_plasmin.html">Factorviii plasminPeak (Valke 2024)</a> </td> <td style="text-align:left;"> Sequential PK-PD model relating factor VIII (FVIII) activity to normalized plasmin peak height (% of normal pooled plasma) measured with the Nijmegen hemostasis assay, after a single intravenous bolus of plasma-derived von Willebrand factor / factor VIII concentrate (pdVWF/FVIII, Humate-P) in 29 adults with severe hemophilia A (Valke 2024). The PK layer is the two-compartment allometric model of Valke 2024 Table 2 (see modellib('Valke_2024_factorviii')); the PD layer is an inhibitory proportional-baseline Imax model, E = Ebase * (1 - Imax * C^n / (EC50^n + C^n)), with baseline 81.2% of normal pooled plasma and EC50 256 IU/dL, and Imax and the Hill coefficient both fixed at 1. FVIII replacement therefore suppresses the raised fibrinolytic activity of untreated severe hemophilia A. Inter-individual variability is carried on the baseline (26.4%); residual error on the PD readout is proportional (16.3%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Larsen_2018_haemophilia_animal_popPK.html">Factorviii rat (Larsen 2018)</a> </td> <td style="text-align:left;"> Preclinical (rat). One-compartment population PK model for B-domain truncated recombinant factor VIII (rFVIII) in male and female Sprague-Dawley rats with concentration-dependent (nonlinear) clearance, from the Larsen 2018 preclinical scaling programme (n=60; 30 males, 30 females). Clearance rises exponentially with total FVIII activity (Larsen 2018 Eq. 1: CL(C) = CL_0 * exp(beta_cl * C)) to capture saturation of the von Willebrand factor buffer at supraphysiological doses. Sex enters as a fractional multiplier on CL (females clear rFVIII 39% faster than males), and body-weight-normalised allometric scaling (exponents 1 for V, 0.75 for CL; scaling principle I) is applied within-species around the median weight of 0.32 kg. Endogenous FVIII (1.12 IU/mL) is carried as an additive baseline on the total observed activity. Parameter values from Larsen 2018 Table 3 (Rat column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valke_2024_factorviii_thrombin_plasmin.html">Factorviii thrombinPeak (Valke 2024)</a> </td> <td style="text-align:left;"> Sequential PK-PD model relating factor VIII (FVIII) activity to normalized thrombin peak height (% of normal pooled plasma) measured with the Nijmegen hemostasis assay, after a single intravenous bolus of plasma-derived von Willebrand factor / factor VIII concentrate (pdVWF/FVIII, Humate-P) in 29 adults with severe hemophilia A (Valke 2024). The PK layer is the two-compartment allometric model of Valke 2024 Table 2 (see modellib('Valke_2024_factorviii')); the PD layer is a proportional-baseline Emax model, E = Ebase * (1 + Emax * C^n / (EC50^n + C^n)), with baseline 11.2% of normal pooled plasma, Emax 8.06 as a factor of baseline, EC50 51.6 IU/dL and the Hill coefficient fixed at 1. Inter-individual variability is carried on the baseline (34.7%); residual error on the PD readout is additive (17.4% of normal pooled plasma). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valke_2024_factorviii_thrombin_plasmin.html">Factorviii thrombinPotential (Valke 2024)</a> </td> <td style="text-align:left;"> Sequential PK-PD model relating factor VIII (FVIII) activity to normalized thrombin potential (% of normal pooled plasma) measured with the Nijmegen hemostasis assay, after a single intravenous bolus of plasma-derived von Willebrand factor / factor VIII concentrate (pdVWF/FVIII, Humate-P) in 29 adults with severe hemophilia A (Valke 2024). The PK layer is the two-compartment allometric model of Valke 2024 Table 2 (see modellib('Valke_2024_factorviii')); the PD layer is an additive-baseline Emax model, E = Ebase + Emax * C^n / (EC50^n + C^n), with baseline 21.9% of normal pooled plasma, Emax 65.3% of normal pooled plasma, EC50 1.93 IU/dL and the Hill coefficient fixed at 1. The very low EC50 means thrombin potential stays near normal for 24 h after a bolus even as FVIII activity falls. Inter-individual variability is carried on Emax (33.1%); residual error on the PD readout is additive (16.2% of normal pooled plasma). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Farrell_2012_farletuzumab.html">Farletuzumab (Farrell 2012)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for farletuzumab (humanized IgG1 anti-folate-receptor-alpha monoclonal antibody) with first-order linear elimination after IV infusion in women with advanced epithelial ovarian cancer (Farrell 2012). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Naik_2013_fasiglifam.html">Fasiglifam (Naik 2013)</a> </td> <td style="text-align:left;"> Joint population PK / FPG / HbA1c model of fasiglifam (TAK-875, a novel GPR40 / free-fatty-acid-receptor-1 agonist) in adults with type 2 diabetes mellitus (T2DM) inadequately controlled on metformin (Naik 2013 CPT PSP). Structural PK is a two-compartment model with first-order absorption and linear elimination (Q/F, V2/F, ka FIXED to values estimated in an earlier multiple-rising-dose study because the phase-2 sparse-sampling design could not identify them). Sex is an additive covariate on CL/F (males ~41 pct higher). Drug effect on fasting plasma glucose (FPG) is a semi-mechanistic indirect-response model with Emax stimulation of Kout; Emax depends exponentially on baseline FPG (BFPG) and on baseline aspartate aminotransferase (AST = SGOT). HbA1c is driven by FPG via a first-order production / elimination pair, with a placebo factor LIFE(t) = 1 - MPL * (1 - exp(-ln(2) * t / HL_pl)) that scales HbA1c production down over the 12-week trial (MPL half-life FIXED to 720 h from graphical analysis). Baseline HbA1c depends linearly on disease duration (T_DIAG_DIAB) and exponentially on BFPG. MPL has an additive sex effect (males have larger placebo response). Data: 1211 PK samples from 286 T2DM patients and 2710 FPG + 1381 HbA1c samples from 346 T2DM patients on 6.25-200 mg oral once-daily fasiglifam for 12 weeks. Fasiglifam was subsequently withdrawn from phase-3 development in December 2013 for hepatotoxicity signals; this popPK/PD model characterises the phase-2 dose-selection analysis that preceded that finding. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yoshida_2024_fazpilodemab.html">Fazpilodemab (Yoshida 2024)</a> </td> <td style="text-align:left;"> Population PK + longitudinal gastrointestinal-adverse-event (GIAE) discrete-time Markov + treatment-discontinuation logistic model for the anti-FGFR/KLB bispecific antibody fazpilodemab (BFKB8488A) in adults with type 2 diabetes mellitus or non-alcoholic fatty liver disease (Yoshida 2024). The PK structure is a 2-compartment disposition with parallel direct (ka) plus transit-mediated (ka2 via the depot2 absorption compartment, transit rate ktr) subcutaneous absorption; linear plus Michaelis-Menten elimination acting on free drug; target-mediated quasi-steady-state binding to a constant target concentration rmax; and a sigmoidal time-onset ADA-mediated clearance arm cl_ada that activates when ADA_POS = 1. The DTMM transition probabilities (grade 0 / grade 1 / grade 2-3) and the treatment-discontinuation probability are emitted as algebraic outputs conditional on the time-varying PREV_AE_SCORE covariate (the previous-day GIAE grade); the actual Markov-chain stochastic simulation is performed downstream in R using these probabilities (matching the mrgsolve plus Rcpp framework of the source paper supplement S2.3). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_febuxostat.html">Febuxostat (Chen 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for febuxostat in 128 healthy Chinese adult volunteers (Chen 2024), with a dual-depot parallel first-order absorption model built to reproduce the multiple-peak plasma profiles of this BCS class II compound. A fraction F1 of the dose is absorbed from a first depot (rate ka1, lag time Tlag1) and the remaining 1 - F1 from a second depot (rate ka2, lag time Tlag2). A standardised high-fat high-calorie meal slows both absorption rate constants and lengthens Tlag1; body weight scales the apparent central and peripheral volumes. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hirai_2016_febuxostat.html">Febuxostat (Hirai 2016)</a> </td> <td style="text-align:left;"> Semi-mechanistic PK-PD model for the hypouricemic effect of febuxostat (non-purine xanthine oxidase inhibitor) in hyperuricemic patients with or without chronic kidney disease. Combines a 2-compartment first-order absorption PK model for febuxostat (literature-digitized, WinNonlin-fit) with (i) a reversible mixed-type inhibition of xanthine oxidase (parameters transferred from Takano 2005 in vitro bovine XO study), (ii) a 2-compartment endogenous uric acid disposition model (adapted from Scott 1969), and (iii) a 1-compartment endogenous xanthine model. Renal uric acid clearance is a power function of creatinine clearance (Tykarski 1991). Non-renal uric acid clearance is assumed constant. All parameters are literature-derived and fixed (deterministic mechanistic model; no population-PK fit). Reproduces Figure 2 of Hirai 2016 (10 mg febuxostat QD in a normouricemic 60 kg patient with CrCl 100 mL/min). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/HillMcManus_2017_febuxostat_lesinurad.html">Febuxostat lesinurad (HillMcManus 2017)</a> </td> <td style="text-align:left;"> Semi-mechanistic dual-drug PKPD model for the impact of non-adherence to febuxostat (xanthine oxidase inhibitor) plus lesinurad (URAT1 uricosuric) urate-lowering therapy in gout; combines published 2-compartment first-order absorption PK for each drug with a 4-compartment xanthine / uric-acid PD system (Hill-McManus 2017) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Henin_2012_gastrointestinal_transit_time_absorption.html">Felodipine (Henin 2012)</a> </td> <td style="text-align:left;"> Semi-mechanistic Gastro-Intestinal Transit Time (GITT) absorption model applied to extended-release felodipine (Henin 2012; the AAPS Journal 14:155-163). The tablet moves through five GI regions (fundus -> antrum -> proximal small intestine -> distal small intestine -> colon) governed by sigmoid step functions with per-subject inflection points IP_FA, IP_APSI, IP_PSI_DSI, IP_DSI_C supplied as covariates (population distributions from Table II). Zero-order drug release from the tablet into the current GI region at rates D1 (fundus 0.68 mg/h), D2/D3 (antrum + PSI 1.91 mg/h), D4/D5 (DSI + colon 1.16 mg/h); dissolved drug flows downstream via K23 (fundus -> antrum, 0.43 1/h) and K34 (antrum -> PSI, 3.48 1/h fasted or 0.81 1/h fed) with both K23 and K34 increased by 5 1/h once the tablet reaches the small intestine (Table I footnote a). Dissolved drug in PSI, DSI, and colon is absorbed to a semi-physiological liver compartment (Vliver 0.0143 L/kg, QH 3.5 L/h/kg^0.75) at rates K47 = K57 = 2.87 1/h and K67 = 1.15 1/h with gut-wall bioavailability FA (0.23 fasted or 0.39 fed). Hepatic extraction ratio EH = 0.50 and blood flow QH set the first-pass loss; systemic distribution is a 3-compartment model (Vcentral, Qperiph1/Vperiph1, Qperiph2/Vperiph2). This extraction encodes only the 'no return to fundus' subpopulation (7/12 felodipine subjects in the paper); the 3-component antrum-fundus-return mixture is documented in the vignette Assumptions and deviations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Back_2018_fenofibrate.html">Fenofibrate (Back 2018)</a> </td> <td style="text-align:left;"> Mechanism-based oral absorption / disposition model for fenofibrate (parent) and fenofibric acid (active form, measured analyte) in healthy Korean adults under fasted, standard-meal, and high-fat-meal conditions. Three drug compartments (stomach -> duodenum -> central) coupled to a 2-compartment calorie sub-model (stomach -> duodenum) via a bile-acid-driven coupling: the combined fenofibrate-metabolism / fenofibric-acid-absorption rate constant km&a is multiplied by (1 + Ebile * calories_in_duodenum), and a time-varying gastric emptying rate constant kg is multiplied by (1 + Efood) for the first 6.94 h after a meal. Meal-type-specific shifts on Vc/F encode the additional bioavailability change between fasted, standard, and high-fat meals. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bista_2015_fentanyl.html">Fentanyl (Bista 2015)</a> </td> <td style="text-align:left;"> One-compartment population PK model for transdermal fentanyl (Durogesic patch) in adult cancer patients with first-order absorption from the patch and allometric body-weight scaling on CL/F and V/F (Bista 2015) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Oosten_2016_fentanyl.html">Fentanyl (Oosten 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for fentanyl administered by continuous subcutaneous infusion and transdermal matrix patch in adult cancer patients, with separate first-order absorption for each route, transdermal lag time, allometric body-weight scaling on CL/F and V/F (V/F fixed at 280 L), IIV on Ka (sc and td), F (td), and CL/F, IOV on transdermal Ka multiplexed by occasion, and proportional residual error (Oosten 2016). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Laffont_2025_opioid_overdose_reversal_simulation.html">Fentanyl iv (Mann 2022)</a> </td> <td style="text-align:left;"> Three-compartment IV fentanyl population PK with a first-order biophase (effect-site) equilibrium compartment, used as the agonist input layer of the Mann 2022 translational opioid-overdose model. Parameter values are the Algera 2021 popPK fit re-tabulated in Mann 2022 Supplement 1 Table S1 (intravenous fentanyl, healthy opioid- naive and chronic opioid-user volunteers pooled, n = 30). Allometric scaling: CL and inter-compartmental clearances on (WT/70)^0.75, volumes on (WT/70). Outputs plasma concentration Cc in ng/mL and effect-site Ce in both ng/mL and pM for downstream consumption by the Mann 2022 mu-opioid receptor binding model. Intended for use as the IV-fentanyl agonist input in a simulated overdose-rescue chain; no residual error is reported in the source supplement. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Plock_2014_ferumoxytol.html">Ferumoxytol (Plock 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with Michaelis-Menten elimination for IV ferumoxytol in healthy adults and adults with chronic kidney disease (Plock 2014). Encodes the typical non-dialysing-patient form; the haemodialysis-driven time-varying central volume (VSLOPE) and the within-session weight-loss effect on V1 (WLO) are described in the vignette but not enabled in this model file. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Oishi_2014_fesoterodine.html">Fesoterodine (Oishi 2014)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption and a lag time for the active metabolite 5-hydroxymethyl tolterodine (5-HMT) following oral administration of the prodrug fesoterodine as a sustained-release tablet (Oishi 2014). Pooled analysis of 10 pharmacokinetic and 3 efficacy/safety studies in Western and East Asian healthy volunteers and adult patients with overactive bladder. Dose entered as mg of fesoterodine fumarate; CL/F and V/F are reported by the paper as molecular-weight-adjusted values for 5-HMT so that dose (mg fumarate) can be used directly with the reported parameters to yield 5-HMT concentrations. CL/F varies with Cockcroft-Gault creatinine clearance (power exponent 0.303 centred at 80 mL/min), moderate hepatic impairment (Child-Pugh Class B multiplier 0.422), CYP2D6 poor-metaboliser phenotype (multiplier 0.626), concomitant CYP3A inhibitor coadministration (multiplier 0.504, source population coadministered ketoconazole), concomitant CYP3A inducer coadministration (multiplier 3.90, source population coadministered rifampicin), female sex (multiplier 0.903), and Japanese ethnicity (multiplier 1.12). Inter-individual variability is a full block on CL/F, V/F, and ka; Tlag IIV was fixed to zero by the source authors. Residual variability is a combined additive + proportional error model on 5-HMT plasma concentrations (LOQ 0.02 ng/mL for most studies). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sano_2023_fesoterodine.html">Fesoterodine (Sano 2023)</a> </td> <td style="text-align:left;"> Population pharmacokinetic model for 5-hydroxymethyl tolterodine (5-HMT), the active metabolite of fesoterodine, in pediatric patients aged 6-17 years with overactive bladder or neurogenic detrusor overactivity. One-compartment disposition with first-order absorption, a first-order absorption lag and first-order elimination, fitted to 428 5-HMT concentrations from 142 patients pooled across the phase II study 1066 (NCT00857896) and the phase III study 1047 (NCT01557244). Body weight is allometrically scaled onto CL/F (exponent 0.75 FIXED) and Vd/F (exponent 1 FIXED) referenced to 35 kg; CYP2D6 poor metabolizers carry 0.546-fold CL/F, female patients 0.862-fold CL/F and 0.634-fold Vd/F, and the beads-in-capsule (BIC) formulation carries 0.648-fold bioavailability relative to the tablet reference. Absorption is flip-flop (ka 0.0897 1/h is far slower than kel = CL/Vd = 1.05 1/h), so the reported 7.73 h terminal half-life is absorption-rate-limited. Inter-individual variability on CL/F, Vd/F and ka is a full 3x3 covariance block and residual error is additive on the log-transformed concentration scale (i.e. log-normal). Doses are supplied in micrograms so that amount/volume yields ng/mL, matching the source NM-TRAN dataset. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sano_2023_fesoterodine.html">Fesoterodine mcc (Sano 2023)</a> </td> <td style="text-align:left;"> Population pharmacokinetic/pharmacodynamic exposure-response model relating 5-hydroxymethyl tolterodine (5-HMT, the active metabolite of fesoterodine) average steady-state plasma concentration to maximum cystometric capacity (MCC) in pediatric patients aged 6-16 years with neurogenic detrusor overactivity. An Emax model in which the maximum attainable MCC is the age-based expected bladder capacity (EBC) rather than an estimated parameter: MCC = BASE + (Emax - BASE) * Cavg,ss / (EC50 + Cavg,ss), with Emax = 30 * (AGE + 1) mL up to age 12 and a 390 mL plateau thereafter (FIXED, not estimated), and typical baseline MCC scaling with age by the same (AGE + 1)/13 factor from a 190 mL plateau. Fitted to 242 MCC observations (baseline and week 12) from 121 patients in the phase III study 1047 (NCT01557244); EC50 is 6.22 ng/mL. This is the pharmacodynamic half of a sequential PK-then-PD analysis: Cavg,ss is supplied as the CAV covariate from individual empirical-Bayes estimates of the companion population PK model, available in this library as modellib('Sano_2023_fesoterodine'). This is a PD-only model with no dose events and no ODE states; correlated inter-individual variability is carried on baseline MCC and Emax, and residual error is combined proportional plus additive. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Harrold_2020_filgrastim_ars.html">Filgrastim (Harrold 2020)</a> </td> <td style="text-align:left;"> Semi-mechanistic population PK / absolute-neutrophil-count / overall-survival model for subcutaneous filgrastim treatment of hematopoietic syndrome of acute radiation syndrome (HS-ARS) in adult and pediatric humans. PK is one-compartment (subcutaneous depot -> central drug amount) with target-mediated disposition through quadratic-equilibrium free / bound filgrastim partitioning against the time-varying G-CSF receptor pool. PD is a 5-stage granulopoiesis cascade (progenitor stem -> mitotic stem -> two precursor stages -> circulating neutrophils); bound drug stimulates receptor production (ST1) and transit between bone-marrow stages (ST2). Acute radiation effect is a kinetic-pharmacodynamic depot (depot_kpd) seeded by the radiation dose in Gy that decays first-order at rate kpde and kills the mitotic-stem stage at rate kpdkill * kpd ^ gamma; gamma depends on the radiation dose rate via a Hill-type function gamma = tgamma * DR / (DR + dr50). Overall survival is integrated as a Cox cumulative hazard (cumhaz_os) on a Box-Cox transformation of an effect-compartment ANC. All structural and IIV parameters fixed at the values from Harrold 2020 Table 2 (granulopoiesis values from Melhem 2018 popPK / ANC in healthy adults and chemotherapy-induced neutropenia; radiation / OS values scaled from rhesus-macaque NHP study with kpde and kpdkill multiplied by 0.72 to match the human LD50 and the kpdkill IIV omega halved per Harrold 2020 Methods 1.3 to address NHP-data sparsity). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kim_2017_fimasartan.html">Fimasartan (Kim 2017)</a> </td> <td style="text-align:left;"> Population PK-PD model for fimasartan (an angiotensin II receptor blocker) in healthy adult Korean men and men with mild or moderate hepatic impairment (Kim 2017). Plasma fimasartan is described by a 2-compartment model with parallel mixed-input absorption: a first-order arm with rate Ka and absorption lag time LAG (fraction F1 = (1 - alpha) * F of the dose) running in parallel with a zero-order arm of virtual duration D2 (fraction F2 = alpha * F of the dose), where the total relative bioavailability F is fixed at 0.18 in healthy subjects (Kim 2008) and incremented to 0.18 + IL1 in mild and 0.18 + IL2 in moderate hepatic impairment to capture the markedly higher Cmax observed in cirrhotic patients via reduced first-pass extraction and intrahepatic shunting. The PD model describes systolic and diastolic blood pressures as indirect-response (turnover) compartments with zero-order synthesis Kin inhibited by fimasartan via a sigmoid-Imax function E(C) = 1 - Emax * Cc / (EC50 + Cc) and first-order loss Kout = Kin / Base; the steady-state baseline rides a fixed cosinor circadian rhythm Bsl(t) = MESOR * (1 + Amp1% * cos(2*pi*(t - AC1)/24) + Amp2% * cos(2*pi*(t - AC2)/12)) with amplitudes and phases inherited from Park 2014 (healthy Korean reference). EC50 is stratified by hepatic-impairment severity: for SBP, healthy versus any-impairment pooled (mild + moderate); for DBP, healthy + mild versus moderate alone, reflecting the contrasting impact of hepatic dysfunction on the two pressure outputs. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Taubert_2018_finafloxacin.html">Finafloxacin (Taubert 2018)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for finafloxacin (a novel fluoroquinolone with enhanced antibacterial activity at acidic pH) with linear elimination, parallel first-order plus zero-order oral absorption (each with its own absorption lag time), an additive renal + non-renal clearance decomposition, and a cumulative-urinary excretion compartment. Built from pooled data of 266 subjects across three trials: 127 healthy volunteers (Trial I oral 25-1,000 mg/day; Trial II IV 200-1,000 mg/day) and 139 patients with complicated urinary tract infections (Trial III IV 800 mg/day, 60-min infusions). Covariates: body surface area on the central volume of distribution (power form, exponent 1.50, reference 1.829 m^2) and healthy / patient cohort status (DIS_HEALTHY) on both the renal and non-renal clearance arms. The paper-reported total apparent clearance (20.9 L/h healthy; -29% in patients) and population-specific fraction renally excreted (FER1 = 0.40 healthy, FER2 = 0.21 patient) are re-parameterised into the canonical lcl_renal + lcl_nonren additive decomposition; the typical values are anchored to DIS_HEALTHY = 0 (patient reference) per the inst/references/covariate-columns.md DIS_HEALTHY convention. The IIV translation between the paper and the re-parameterised forms is documented in the validation vignette Assumptions and deviations (Taubert 2018). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Eissing_2024_finerenone.html">Finerenone (Eissing 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with a 4-transit-compartment delayed first-order absorption for finerenone in adults with chronic kidney disease and type 2 diabetes (FIGARO-DKD final PK model; Eissing 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Goulooze_2022_finerenone.html">Finerenone (Goulooze 2022)</a> </td> <td style="text-align:left;"> Population PKPD turnover model for serum potassium response to finerenone in patients with chronic kidney disease and type 2 diabetes (FIDELIO-DKD Phase III). Indirect-response model with an Emax effect of finerenone steady-state AUC on the potassium dissipation rate Kout, with a linear annual disease-progression slope on serum K (different typical value for active-treatment vs placebo arms). Finerenone PK is upstream (van den Berg 2022) and reduced here to AUCss = DOSE / CL with typical apparent clearance 28.0 L/h. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vandenBerg_2021_finerenone.html">Finerenone (vandenBerg 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral finerenone (Bayer BAY 94-8862, a non-steroidal selective mineralocorticoid receptor antagonist) in adults with chronic kidney disease and type 2 diabetes, developed on n=2284 subjects / 5057 sparse PK observations from the FIDELIO-DKD Phase III trial (NCT02540993). Absorption is modelled via a chain of four sequential first-order steps (depot + three transit buffers, all at common rate Ka = 22.5 1/h, mean transit time MTT = n_steps / Ka = 0.178 h with the depot counted as the first compartment in the chain) preceded by a fixed 0.215 h absorption lag time; the central-peripheral disposition is two-compartment with the peripheral volume fixed equal to the central volume Vp/F = Vc/F (ratio fixed at 1). Covariates retained in the final model are body weight and Korean ethnicity on Vc/F; time-varying eGFR-CKD-EPI, body height, serum creatinine, smoking status (current or former vs never), long-term (>=50% of treatment period) SGLT2 inhibitor use, gamma glutamyl-transferase, and a two-tier CYP3A4-inhibitor coadministration categorisation (strong/moderate/weak inhibitor >=50% of treatment period vs any other inhibitor exposure) on CL/F; with each of the CL/F covariates (except GGT) ALSO applied inversely to the relative bioavailability F1 in the paper's NONMEM control stream (so the covariate appears on both CL/F and F simultaneously, the net effect on steady-state AUC scales as 1 / covariate-factor^2 and the net effect on Cmax scales as 1 / covariate-factor). Inter-individual variability is a 2x2 block on CL/F and Vc/F (omega^2 0.0961 / 0.104, covariance 0.0442, correlation ~0.44); no IIV on Ka or absorption. Residual error is proportional (sigma^2 = 0.313, propSd = sqrt(0.313) ~= 0.5595). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Goulooze_2022_finerenone_uacr_egfr.html">Finerenone egfr (Goulooze 2022)</a> </td> <td style="text-align:left;"> Population PKPD disease-progression model for the estimated glomerular filtration rate (eGFR) response to finerenone in patients with chronic kidney disease and type 2 diabetes (FIDELIO-DKD Phase III). eGFR declines at a constant chronic slope that is flattened by an exponential stabilisation function as eGFR approaches 16.1 mL/min/1.73 m^2, and the chronic slope is driven by the model-predicted UACR time course from the companion UACR model rather than by baseline UACR. Finerenone acts twice: an acute fully reversible eGFR decline through a power function of steady-state daily AUC acting via its own effect compartment, and, via its UACR reduction, a sustained flattening of the chronic slope. Finerenone PK is upstream (van den Berg 2022) and reduced here to AUCss = DOSE / CL with typical apparent clearance 29.9 L/h. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Goulooze_2022_finerenone_uacr_egfr.html">Finerenone uacr (Goulooze 2022)</a> </td> <td style="text-align:left;"> Population PKPD disease-progression model for the urine albumin-to-creatinine ratio (UACR) response to finerenone in patients with chronic kidney disease and type 2 diabetes (FIDELIO-DKD Phase III). UACR is integrated as a state whose fractional progression rate is corrected by the model-predicted UACR and by UACR over baseline; the finerenone effect is a power function of steady-state daily AUC acting through an effect compartment, and concomitant SGLT2 inhibitor use gives a direct proportional shift. Finerenone PK is upstream (van den Berg 2022) and reduced here to AUCss = DOSE / CL with typical apparent clearance 29.9 L/h. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2007_flucloxacillin.html">Flucloxacillin (Landersdorfer 2007)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for IV flucloxacillin in healthy adult volunteers (Landersdorfer 2007) with linear renal and non-renal elimination. The structural model splits total clearance into a renal arm (CL_R = 5.37 L/h) and a non-renal arm (CL_NR = 2.73 L/h); their sum reproduces the derived total clearance CL_T = 8.10 L/h reported in Table 2. The renal arm also drives a cumulative urinary excretion compartment that the paper fits jointly with plasma. Distribution uses a shallow peripheral (V_2 = 2.61 L, CLic_shallow = 15.3 L/h) and a deep peripheral (V_3 = 2.17 L, CLic_deep = 1.23 L/h); central volume V_1 = 4.79 L. Between-subject variability is reported as a full 5x5 variance-covariance matrix (Table 3, natural-log scale) on CL_R, CL_NR, V_1, V_2, V_3; no BSV is included on the inter-compartmental clearances. Residual error is combined additive + proportional on both plasma concentrations (9.4% CV, 0.155 mg/L) and cumulative urinary amounts (20.9% CV, 1.04 mg). The 5-min infusion duration used in the study is supplied via dose records (DUR / RATE) rather than as a model parameter. No structural covariates were retained: the cohort was 10 healthy Caucasian adults (5 M / 5 F, weight 52-83 kg, age 23-34 years) and demographics are not used inside the model. Monte Carlo dose-attainment simulations in the paper (continuous, 4-h, 0.5-h infusions) reuse these PK parameters together with 96% protein binding. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Han_2013_fluconazole.html">Fluconazole (Han 2013)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for fluconazole in adult burn-ICU patients with suspected or confirmed Candida infection, with a piecewise CL covariate model that switches between a fixed CRRT-cohort CL and a Cockcroft-Gault-CrCl / postburn-recency / sepsis-shifted non-CRRT CL plus an additive WT / edema / postburn-recency model on volume (Han 2013) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Leroux_2018_fluconazole_micafungin.html">Fluconazole (Leroux 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model of intravenous fluconazole in preterm and term neonates with suspected or proven systemic candidiasis (Leroux 2018), with linear current-weight scaling of CL and V. Typical-value structural model only: the source paper and Data S1 supplement (goodness-of-fit plots only) do not report inter-individual variability magnitudes, residual error structure, or a maturation covariate functional form, so IIV and RUV are encoded as fixed(0) and no postmenstrual / corrected gestational age effect is encoded. See vignette Errata. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/McLachlan_1996_fluconazole.html">Fluconazole (McLachlan 1996)</a> </td> <td style="text-align:left;"> One-compartment population PK model for fluconazole in adults with HIV/AIDS, fit to 770 plasma concentrations from 113 male subjects pooled across an intensive-sampling sub-study (Study 1, n=13, 12-17 samples per dose) and a sparse routine-care sub-study (Study 2, n=100, single sample per subject). Oral capsules (Diflucan, 50-800 mg) and 50 mg per 15 min IV infusions are described by a single linear central compartment with first-order absorption from a depot and zero-order input during IV infusion. The final NONMEM clearance model is an additive intercept-plus-slopes regression on Cockcroft-Gault creatinine clearance (raw, not BSA-normalized) and absolute CD4+ T-lymphocyte count: CL (L/h) = 0.25 + 0.0057 * CLcr (mL/min) + 0.00068 * CD4 (cells/mm^3); volume of distribution, absorption rate, and bioavailability are not modulated by covariates. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Momper_2016_fluconazole.html">Fluconazole (Momper 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for fluconazole with first-order oral absorption and IV administration in extremely premature infants with birth weights < 750 g (Momper 2016) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Patel_2011_fluconazole.html">Fluconazole (Patel 2011)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous fluconazole in 10 critically ill anuric adults receiving continuous venovenous hemodiafiltration (CVVHDF) (Patel 2011). Total fluconazole clearance from the central compartment is partitioned into a CVVHDF-route arm (CL_CVVHDF, 1.66 L/h typical, encoded as lcl_renal) and a non-CVVHDF arm (CL_NCVVHDF, 1.01 L/h typical, encoded as lcl_nonren) that are fitted simultaneously to the plasma concentration-time profile and the cumulative amount of fluconazole in the CVVHDF effluent. Dialysis-filter membranes in use for more than 48 hours reduce CVVHDF efficiency to 36.8 percent of the fresh-filter baseline (FILT_AGE_HI indicator on the CL_CVVHDF arm, multiplicative effect 0.368, bootstrap 95 percent confidence interval 0.326 to 0.426; informed by 1 of 10 patients with a > 48 h filter). Input into the central compartment is zero-order over an estimated infusion duration D1 (typical 0.689 h, near the 60-min nominal infusion). No subject-level covariates (age, total body weight, sex, APACHE II score) reached the OFV-3.84 retention threshold on CL, Vc, Q, or Vp, so no demographic covariates are encoded in this file. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wade_2008_fluconazole.html">Fluconazole (Wade 2008)</a> </td> <td style="text-align:left;"> One-compartment intravenous population PK model for fluconazole in preterm and term infants (gestational age 23-40 weeks, postnatal age <120 days) with allometric body weight on CL and V (fixed exponents 0.75 and 1.0, reference 1 kg), power effects of gestational age at birth (reference 26 weeks) and postnatal age (reference 2 weeks) on CL, and an on/off power effect of serum creatinine on CL gated when SCR > 1 mg/dL (Wade 2008). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Watt_2015_fluconazole.html">Fluconazole (Watt 2015)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous fluconazole in critically ill children (1 day to 17 years; n=40) supported with extracorporeal membrane oxygenation (ECMO) or matched non-ECMO controls (Watt 2015). Clearance scales linearly with body weight and is modulated by serum creatinine via a power function (CREAT/0.4)^-0.29 centered at the cohort median initial SCR of 0.4 mg/dL; allometric 3/4- power scaling on CL was tested and rejected (delta-OFV +9.7) so a linear weight scaling was retained. Central volume scales linearly with body weight and is increased 1.39-fold in subjects on ECMO support via a multiplicative power factor 1.39^ECMO_STATUS. Proportional residual error (15.3% CV); diagonal Omega with IIV on CL (33.2% CV) and V (22.2% CV) only -- the paper retained the proportional-only error model after showing the proportional-plus-additive form could not precisely estimate the additive component. Used by the authors to derive dosing recommendations for invasive candidiasis prevention (12 mg/kg loading then 6 mg/kg daily) and treatment (35 mg/kg loading then 12 mg/kg daily) in children on ECMO. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Langenhorst_2019_fludarabine.html">Fludarabine (Langenhorst 2019)</a> </td> <td style="text-align:left;"> Three-compartment IV population PK model for the circulating fludarabine metabolite F-ara-A in children and adults (n = 258, age 0.3-74 years) receiving fludarabine phosphate as part of myeloablative conditioning prior to allogeneic hematopoietic cell transplantation (Langenhorst 2019). Total clearance is decomposed into a non-renal component (3.2 L/h at 70 kg) and an eGFR-driven renal component with slope 0.78 (unitless), both allometrically scaled to actual body weight at a fixed 0.75 exponent referenced to 70 kg: CL = (3.2 + eGFR_L_per_h * 0.78) * (BW/70)^0.75 L/h, where eGFR is expressed in L/h (converted from Cockcroft-Gault / Schwartz eGFR in mL/min/1.73 m^2 by * 60/1000). Central volume V1 (39 L), first peripheral V2 (20 L), and second peripheral V3 (50 L) all scale allometrically at a fixed 1.0 exponent to 70 kg. Inter-compartmental clearances Q2 (8.6 L/h, V1-V2) and Q3 (3.8 L/h, V1-V3) scale at a fixed 0.75 exponent to 70 kg. Because the random effects on the volume triplet (V1, V2, V3) and on the clearance triplet (CL, Q2, Q3) were highly correlated in the original NONMEM fit, the paper encodes a SINGLE shared eta on {V1, V2, V3} (48% CV) and a SINGLE shared eta on {CL, Q2, Q3} (23% CV) -- implemented here as identical eta terms across the three parameters in each group (perfect correlation, matching the paper's coding). Inter-occasion variability (12% CV on CL/Q2/Q3, 31% CV on V1/V2/V3, one occasion per dose) reported by Langenhorst 2019 Table 2 is NOT encoded structurally: the source paper defines each dose as an occasion but nlmixr2lib omits IOV when no operational occasion column is defined for downstream use (Brooks 2021 / Andrews 2017 precedent); downstream users can add an OCC indicator and per-occasion etas in rxode2. Residual error is proportional (6.3% CV) on the linear concentration scale. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Polito_2016_fludrocortisone.html">Fludrocortisone (Polito 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral fludrocortisone with first-order absorption, an absorption lag time, and first-order elimination, estimated in 14 adults with septic shock (out of 21 enrolled; 7 had undetectable plasma concentrations) receiving a single 50 ug oral dose of fludrocortisone acetate via naso-gastric tube (Polito 2016). The Simplified Acute Physiology Score II (SAPS II) is retained as a positive power covariate on both apparent oral clearance CL/F (exponent 0.019) and absorption lag time Tlag (exponent 0.036), normalised to the cohort median SAPS II = 53. Inter-individual variability is exponential on every PK parameter (ka, V/F, CL/F, Tlag) with a diagonal OMEGA matrix; residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Luo_2024_CES1_cirrhosis.html">Flumazenil pbpk (Luo 2024)</a> </td> <td style="text-align:left;"> PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). flumazenil disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Flumazenil is a direct-acting benzodiazepine-receptor antagonist that CES1 INACTIVATES to flumazenil acid, so no active metabolite is tracked. Renal clearance is negligible (CLint,K 1.67 mL/min). Section 4 notes enzymes other than CES1 also metabolise flumazenil, which the paper cites as a reason its cirrhosis AUCR was underpredicted. The semi-PBPK circuit is stomach, three small-intestinal lumen segments (duodenum / jejunum / ileum), the matching three gut-wall segments, portal vein, liver, kidney and a two-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper's virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kumpulainen_2010_flurbiprofen.html">Flurbiprofen (Kumpulainen 2010)</a> </td> <td style="text-align:left;"> Three-compartment population PK model with a separate cerebrospinal-fluid (CSF) compartment for flurbiprofen in 64 healthy children aged 3 months to 13 years (Kumpulainen 2010). Two parallel absorption routes: oral syrup via an absorption compartment with lag time (K12) and a single first-order ka, and IV flurbiprofen axetil prodrug via a separate dosing compartment that converts to flurbiprofen with a first-order rate constant (K42). Plasma kinetics scaled allometrically by weight (exponents fixed at 0.75 for CL and 1 for all volumes, including the CSF volume held fixed at 0.15 L/70 kg per literature). The paper's QCSF + UPTK parameterisation is encoded as canonical influx / efflux clearances clin = QCSF * UPTK and clef = QCSF, with fraction unbound (fu) gating only the central-to-CSF flux. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2018_flurbiprofen.html">Flurbiprofen (Zhang 2018)</a> </td> <td style="text-align:left;"> One-compartment IV population PK plus Holford-Sheiner effect-compartment for cerebrospinal fluid (CSF) disposition of flurbiprofen, the active metabolite of flurbiprofen axetil, in Chinese adults with postoperative pain receiving 1 mg/kg IV flurbiprofen axetil (Zhang 2018, Tables 1-2, Eq. 3 covariate form). Final-model typical values CL = 1.55 L/h, Vd = 7.91 L, plasma-CSF equilibration rate Ke = 0.0015/h; linear-multiplicative covariate effects of weight and height on Ke centered on the population medians (68.5 kg, 165 cm). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Weber_2015_fluticasone_inhaled.html">Fluticasone inhaled (Weber 2015)</a> </td> <td style="text-align:left;"> Semi-mechanistic. Population PK model for inhaled fluticasone propionate (FP) in healthy adult volunteers (Weber 2015), used for Monte-Carlo simulation of PK-based bioequivalence trials. Separate central (LC1 -> LC2) and peripheral (LP1 -> LP2) lung deposition compartments hold undissolved drug particles (LC1, LP1) and dissolved drug (LC2, LP2); mucociliary clearance kmuc removes undissolved particles from central lung regions only; dissolved drug is absorbed into a two-compartment systemic disposition with central-to-peripheral rate constants k12 and k21. Each administration splits across LC1 (bioavailability flung * fc) and LP1 (bioavailability flung * (1 - fc)); the remaining (1 - flung) fraction is assumed to have negligible oral bioavailability. F_Lung and F_C are logit-transformed; all other parameters are log-transformed. Structural parameters and BSV were taken from the validated FP inhalation model of Weber and Hochhaus 2013 (reference 13 of Weber 2015); BOV on F_Lung, F_C, and kmuc described as a paper-specific extension for crossover-trial simulation is NOT encoded in this model file (see vignette Assumptions and deviations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Geldof_2008_fluvoxamine_rat.html">Fluvoxamine rat (Geldof 2008)</a> </td> <td style="text-align:left;"> Preclinical (rat, male Wistar). Non-linear pharmacokinetic brain distribution model for fluvoxamine in plasma, brain extracellular fluid (ECF) and total brain tissue, fit by Geldof et al. (2008, Pharm Res 25(4):792-804) using simultaneous analysis of microdialysate ECF (n = 26 rats, frontal-cortex CMA/12 probe) and total brain tissue (n = 35 rats, destructive brain sampling) after a single 30 min IV infusion of 1, 3.7 or 7.3 mg/kg fluvoxamine. The structural model is a three-compartment plasma disposition (central + 2 peripherals, with PK parameters fixed at the mean post-hoc estimates from the upstream Geldof 2007 rat population PK model, Table I 'Microdialysis + brain sampling' row) coupled to a single-state lumped brain compartment (brain_total) whose dynamics follow dCT/dt = kin*Cp - kout*CSP (paper Eq 10), with the shallow perfusion-limited CSP and deep brain CDB (= ECF) concentrations recovered algebraically at every time step from CT via the rapid-equilibrium saturable-efflux quadratic (paper Appendix Eq 47). The single lumped efflux parameter N***max (NstarMax in this file) and C50 govern the saturation of the active (Pgp / MRP-mediated) removal flux from the deep brain back to the shallow brain. Inter-individual variability is on kin and kout only, with the correlation reported in Table II. The proportional residual error sigma^2 = 0.042 is shared between the ECF (Cecf) and total- brain (Cbrain) observations per Table II. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Rose_2016_follitropin_delta.html">Follitropin delta (Rose 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for FE 999049 (recombinant human FSH; INN follitropin delta) with first-order subcutaneous absorption through a single transit compartment and first-order elimination, in 27 healthy pituitary-suppressed female subjects after a single subcutaneous dose of 37.5-450 IU (2.2-26.3 ug). Body weight enters as an allometric covariate on apparent clearance (exponent 0.75) and apparent volume of distribution (exponent 1) with reference weight 65 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zufferey_2018_fondaparinux.html">Fondaparinux (Zufferey 2018)</a> </td> <td style="text-align:left;"> Parametric time-to-event model for major bleeding after major orthopaedic surgery under fondaparinux thromboprophylaxis (POP-A-RIX 2.5 mg once daily and PROPICE 1.5 mg once daily pooled cohorts; n = 1393, 64 adjudicated bleeding events). The hazard is hz(t) = h0(t) * exp(beta1*SEX + beta2*AUCinf/8.5 + beta3*LBM/44), with gamma-shaped baseline h0(t) = theta1*theta2*(t-theta3)*exp(-theta2*(t-theta3)) for t > theta3 and 0 otherwise (lag time theta3 ~= 17.6 h, peak ~4 days post-surgery). AUCinf is derived inside the model from daily dose and clearance using the paper's PK equation CL = 0.34 * (CRCL/60)^0.485 * exp(eta) (lean-body-weight Cockcroft-Gault CrCl). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fisher_2008_fosamprenavir.html">Fosamprenavir (Fisher 2008)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption for orally administered fosamprenavir (FPV), measured as the active amprenavir (APV) metabolite, in HIV-1-infected pediatric patients aged 4 weeks to 18 years (Fisher 2008). Allometric scaling on apparent clearance (CL/F, Q) at a fixed exponent of 0.75 and on apparent volumes (V2/F, V3) at a fixed exponent of 1.0 (reference 70 kg). Apparent CL/F is reduced ~60% by concomitant ritonavir (RTV) co-administration (maximal CYP3A4 inhibition assumed at the RTV doses used), and is further modified by a piecewise age-maturation factor (linearly declining additive offset for AGE <= 2*AG50, zero above), by sex (lower in females), by race (separate multipliers for Black and for the non-Black non-White composite vs the White reference), and by a power effect of serum alpha-1-acid glycoprotein (AAG, centred at 0.77 g/L). Apparent V2/F also carries a power effect of AAG. Bioavailability is anchored on suspension-under-fed conditions (F=1), with a separate relative bioavailability for the tablet formulation (F_tab) and a separate relative bioavailability for the suspension administered fasted (F_food,sus). Inter-occasion variability on CL/F (~34% CV) reported by the source poster is NOT structurally encoded here (no operational occasion column is defined for the model-library use case); downstream users who want to simulate IOV can add an OCC indicator and a per-occasion eta in rxode2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Weatherley_2018_fosdagrocorat.html">Fosdagrocorat (Weatherley 2018)</a> </td> <td style="text-align:left;"> Simultaneous parent-metabolite population PK model for oral fosdagrocorat (PF-04171327, a phosphate-ester prodrug of the dissociated glucocorticoid receptor agonist PF-00251802) in adult rheumatoid-arthritis patients receiving stable background methotrexate (Weatherley 2018). The prodrug is fully cleaved by alkaline phosphatase in the gut wall before absorption; only the active Metabolite-1 (PF-00251802) and its circulating N-oxide Metabolite-2 (PF-04015475) are modelled. Metabolite-1 is described by a two-compartment disposition (apparent CL, V2, Q, V4 fixed at 209 L) with first-order absorption (K12) and bioavailability F1 fixed at 1 (apparent F absorbed via the prodrug-to-Metabolite-1 conversion). Metabolite-2 is described by a one-compartment disposition (apparent Vm, CLm) with Fm fixed at 1 (assumed 100 percent molar conversion of Metabolite-1 to Metabolite-2). Standard allometric weight scaling is fixed on Metabolite-1 disposition (exponent 0.75 on CL and Q; exponent 1.00 on V2 and V4); body-weight scaling on Metabolite-2 CLm is estimated as a power-form covariate (exponent 0.450) and no weight effect is applied to Vm (rejected in stepwise covariate testing). Retained covariates on Metabolite-1 CL are female-vs-male (-26.8 percent) and a small linear age effect (-0.00633 L/h per year above 40). Retained covariates on Metabolite-2 CLm are female-vs-male (-34.1 percent) and body weight. Inter-individual variability is reported on Metabolite-1 CL (33 percent CV) and absorption rate K12 (249 percent CV), and on Metabolite-2 Vm (44 percent CV) and CLm (26 percent CV). The publication's interoccasion variability on Metabolite-1 F1 (23.8 percent CV across dosing occasions) is encoded here as IIV on F1 because nlmixr2lib simulation does not carry an occasion column; this approximation is documented in the vignette Assumptions section. Residual error is combined additive plus proportional on the linear-concentration scale separately for each analyte (Metabolite-1 proportional 19.9 percent + additive 0.305 ng/mL; Metabolite-2 proportional 7.8 percent + additive 0.10 ng/mL). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shoji_2017_fosdagrocorat_oc.html">Fosdagrocorat oc (Shoji 2017)</a> </td> <td style="text-align:left;"> Kinetic-pharmacodynamic (K-PD) model for serum osteocalcin (OSTCALC) bone-formation biomarker following once-daily oral fosdagrocorat (PF-04171327, a dissociated agonist of the glucocorticoid receptor) or oral prednisone comparator in adults with rheumatoid arthritis on background methotrexate (Shoji 2017). Sister model to Shoji_2017_fosdagrocorat_p1np: identical K-PD structure (virtual K-PD depot with zero-order Input mg/week and first-order KDE; sigmoid Emax inhibition of biomarker synthesis with Hill coefficient fixed to 1; empirical dose-and-time-dependent rebound multiplier; additive placebo-period slope). For the OSTCALC fit Shoji 2017 fixed KDE to the P1NP-derived estimates and fixed Imax to 1 for both drugs, and used independent (not block) IIV on KDE, EDK50, and BL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shoji_2017_fosdagrocorat_p1np.html">Fosdagrocorat p1np (Shoji 2017)</a> </td> <td style="text-align:left;"> Kinetic-pharmacodynamic (K-PD) model for serum amino-terminal propeptide of type I collagen (P1NP) bone-formation biomarker following once-daily oral fosdagrocorat (PF-04171327, a dissociated agonist of the glucocorticoid receptor) or oral prednisone comparator in adults with rheumatoid arthritis on background methotrexate (Shoji 2017). A virtual K-PD depot for the drug (zero-order Input mg/week, first-order elimination KDE) feeds a sigmoid Emax inhibition of biomarker synthesis (Hill coefficient fixed to 1); the synthesis rate carries an empirical dose-and-time-dependent rebound multiplier and an additive linear placebo-period slope captures the methotrexate-only time trend. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Huppe_2023_fosfomycin.html">Fosfomycin (Huppe 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for multiple-dose intravenous fosfomycin in critically ill adults with renal insufficiency during continuous venovenous hemodialysis (CVVHD). Total clearance is the sum of a renal arm driven linearly by measured urinary creatinine clearance (gated off in anuric patients) and a CVVHD arm given by the Michaels hemodialyzer equation as a function of blood flow rate, dialysate flow rate, and a mass transfer-area coefficient. Central volume increases linearly with time since first dose (Huppe 2023). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gibiansky_2005_fospropofol.html">Fospropofol (Gibiansky 2005)</a> </td> <td style="text-align:left;"> Joint two-compartment fospropofol (GPI 15715, AQUAVAN) prodrug + intermediate delay compartment + two-compartment propofol active-metabolite population PK model in adults receiving IV bolus AQUAVAN for procedural sedation (Gibiansky 2005, ASCPT poster, colonoscopy sedation Phase II study). The model assumes complete metabolism of GPI 15715 to propofol via systemic alkaline-phosphatase hydrolysis; the intermediate compartment captures the appearance delay between GPI 15715 elimination from plasma and the corresponding rise in propofol concentration. Lean body mass (LBM, reference 55 kg) was retained as a linear-fractional covariate on GPI 15715 central volume Vc_GPI, GPI 15715 metabolic clearance CL_GPI, and propofol central volume Vc_PR; fentanyl premedication exposure, age, sex, and other demographics/laboratory covariates were tested but not retained. Propofol Vc_PR (6.91 L) was fixed as the data were insufficient for joint estimation with CL_PR (the model is identifiable on CL_PR = K10_PR * Vc_PR = 4.53 L/min). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fiedler-Kelly_2019_fremanezumab.html">Fremanezumab (Fiedler-Kelly 2019)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for fremanezumab (anti-CGRP IgG2 delta-a/kappa mAb) with first-order SC absorption, absorption lag time, and route-specific central volume / residual error supporting both IV and SC administration in healthy adults and adults with chronic or episodic migraine (Fiedler-Kelly 2019). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/FiedlerKelly_2020_fremanezumab_cm.html">Fremanezumab cm (FiedlerKelly 2020)</a> </td> <td style="text-align:left;"> Population PD exposure-response model relating fremanezumab average plasma concentration (Cav) to monthly moderate-to-severe headache days in adults with chronic migraine. Placebo time-course is a Hill (sigmoid) function in months and the drug effect is a power function of Cav centered on the population median Cav. Fitted to 5312 monthly observations from 1361 chronic-migraine patients pooled across the LBR-101-021 phase 2b and TV48125-CNS-30049 phase 3 studies (Fiedler-Kelly 2020). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/FiedlerKelly_2020_fremanezumab_em.html">Fremanezumab em (FiedlerKelly 2020)</a> </td> <td style="text-align:left;"> Population PD exposure-response model relating fremanezumab average plasma concentration (Cav) to monthly migraine days in adults with episodic migraine. Placebo time-course is an exponential growth in months (predicted reduction = exp(exponent * t)) and the drug effect is an Emax/EC50 of Cav scaled by individual baseline migraine days. Fitted to 4444 monthly observations from 1142 episodic-migraine patients pooled across the LBR-101-022 phase 2b and TV48125-CNS-30050 phase 3 studies (Fiedler-Kelly 2020). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zou_2022_furmonertinib.html">Furmonertinib (Zou 2022)</a> </td> <td style="text-align:left;"> Semi-mechanistic joint parent-metabolite population PK model for oral furmonertinib (AST2818, a third-generation irreversible EGFR TKI) and its active metabolite AST5902 (Zou 2022). The parent has two-transit-compartment absorption (rate constant ka shared for depot and both transits) feeding a two-compartment parent disposition (CL/F, Vc/F, Q/F, Vp/F). The parent is eliminated through the central compartment; a fraction Fm of the parent elimination becomes AST5902, which is described by a two-compartment metabolite (Clm/(F*Fm), Vcm/(F*Fm), and inter-compartment rate constants k67 and k76). Because no intravenous data were available, the absolute parent bioavailability F and the fraction Fm are non-identifiable and are absorbed into the apparent parameters. Autoinduction of furmonertinib metabolism (mediated by CYP3A4) is modelled as an indirect-response (IDR III) enzyme pool with unity baseline: d(A_ENZ)/dt = kENZ * (1 + S * Cc) - kENZ * A_ENZ, and the apparent parent clearance is CLbase/F * A_ENZ. Covariates: alkaline phosphatase (ALP, U/L) via power effect on CLbase/F and on Clm/(F*Fm) (median 77.2 U/L); body weight (WT, kg) via power effect on Clm/(F*Fm) (median 65 kg); and a categorical food-with-a-high-fat-meal effect on parent oral bioavailability (+22.4%) and on the fraction converted to AST5902 (-33.5%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yoshioka_2018_FXa_inhibitors_mbma.html">FXa inhibitors mbma (Yoshioka 2018)</a> </td> <td style="text-align:left;"> MBMA. PT-ratio-driven logistic event-rate model for direct oral factor Xa inhibitors (rivaroxaban, apixaban, edoxaban) in non-valvular atrial fibrillation. Inputs a population-mean prothrombin-time ratio (PTR) supplied per observation time; outputs per-arm probability of ischemic stroke/SE (p_isse) and of major bleeding (p_mb), plus a derived per-arm mortality probability. Fit by NONMEM 7.3 to per-arm event counts from 5 large RCTs (Yoshioka 2018; 57,655 patients). Suitable for simulating per-arm summary outcomes only; the upstream popPK -> PT-ratio layer for each FXa inhibitor is out of scope and PTR must be supplied externally. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Willmann_2021_fxi_aso.html">FxiLica (Willmann 2021)</a> </td> <td style="text-align:left;"> Population PK/PD model for the GalNAc-conjugated FXI antisense oligonucleotide FXI-LICA (BAY2976217, formerly ION-957943) after subcutaneous administration in healthy volunteers (study LICA-CS1), pooled with IONIS-FXIRX data to fit the FXI-activity indirect- response arm (Willmann 2021). Pharmacokinetics: two-compartment model with parallel first-order (fraction F1, rate ka) and zero-order (fraction 1 - F1, duration D2) subcutaneous absorption and first-order elimination from the central compartment. The final PK model retained an OMEGA block for the correlations between IIV on CL, V2, and Q (rho ~ 0.6-0.7; Table S3). Pharmacodynamics: indirect-response model on FXI activity with sigmoid-Imax inhibition (Imax fixed to 1) of the zero-order synthesis rate Kin = baseline * kout, driven by an effect-site concentration linked to the plasma central compartment via first- order equilibration ke0. FXI-LICA has an approximately 20-fold higher hepatocyte potency than IONIS-FXIRX (IC50 = 2.59 vs 167 ng/mL) owing to GalNAc-mediated ASGPR uptake into hepatocytes. The main-analysis FXI-LICA PK model was fitted to healthy-volunteer data only; the authors' ESRD simulations for FXI-LICA (Figures 3a, 4) borrow the IONIS-FXIRX ESRD covariate effects on CL, V3, and the effect-site driving concentration (keoPAT) under the assumption that the receptor-mediated hepatic uptake pathway is not altered in ESRD. These borrowed ESRD factors are exposed as fixed parameters in this model so that the FXI-LICA-in-ESRD scenario can be reproduced. The FXI-activity arm was fitted simultaneously with pooled ASO-CS1 + ASO-CS4 + LICA-CS1 data (shared kout, baseline, ke0, and Hill). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Melhem_2018_g_csf.html">G csf (Melhem 2018)</a> </td> <td style="text-align:left;"> Semi-mechanistic population PK/PD model for filgrastim and pegfilgrastim in healthy adult volunteers and adult cancer patients undergoing chemotherapy-induced neutropenia (CIN). Subcutaneous drug enters a depot compartment, absorbed at first-order rate KSC into a total-drug compartment (DT). Target-mediated disposition uses a quasi-equilibrium quadratic between total drug and total G-CSFR pool (circ), giving free drug (FDC) and bound complex (RDC). Free drug is cleared linearly (CLD/VD) and bound drug is internalised (KINT). The granulopoiesis PD cascade is a five-state receptor chain (SM stem -> MT mitotic -> PM1 -> PM2 -> RB blood receptors) with baseline set by KP/KTR (bone-marrow pools) and KP/KC (blood pool); ANC = RB / SR. Drug binding stimulates receptor production (ST1 = 1 + STM1*driver) and bone-marrow transit (ST2 = 1 + STM2*driver). Chemotherapy is modelled as a KPD virtual compartment with lag LAG9; its output (KCHM*CHM) scaled by CHMSL adds to the mitotic-cell elimination rate. Corticosteroid is a second KPD compartment with lag LAG8; its output drives an Emax stimulation modulating ST1 and ST2 through CRTM = KCRT*CRT/(CRT50+KCRT*CRT). Filgrastim and pegfilgrastim differ on FSC, KSC, VD, CLD and KD (categorical covariate FORM_GCSF_PEG); pegfilgrastim at 300 ug/kg has an additional lower CLD (categorical DOSE_PEG_300UGKG); patient vs healthy-volunteer status shifts KINT and STM2 (categorical DIS_HEALTHY); body weight allometrically scales CLD (exponent 0.641) and VD (exponent 0.943) with reference 70 kg. Filgrastim (2.6 h) and pegfilgrastim (10.1 h) apparent half-lives. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kielbasa_2020_galcanezumab.html">Galcanezumab (Kielbasa 2020)</a> </td> <td style="text-align:left;"> One-compartment population PK model for galcanezumab (humanized IgG anti-CGRP mAb) with first-order SC absorption, linear elimination, and allometric body weight scaling on CL/F (Kielbasa 2020) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Acosta 2007)</a> </td> <td style="text-align:left;"> One-compartment population PK model for ganciclovir after IV ganciclovir and an oral liquid valganciclovir formulation in neonates with symptomatic congenital CMV disease (Acosta 2007), with first-order absorption, estimated bioavailability, and a steep un-normalized body-weight power effect on clearance. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Acosta 2007 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Caldes_2009_ganciclovir.html">Ganciclovir (Caldes 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ganciclovir after IV ganciclovir and oral valganciclovir administration in solid organ transplant patients infected with cytomegalovirus, with first-order absorption, lag time, logit-transformed bioavailability, and creatinine-clearance scaling on CL (Caldes 2009) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2021_ganciclovir.html">Ganciclovir (Chen 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral ganciclovir (the active metabolite of valganciclovir) in adult Chinese renal allograft recipients (Chen 2021), with first-order absorption after a lag time and a linear creatinine-clearance effect on apparent oral clearance (CL/F). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Facchin 2019)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ganciclovir after oral valganciclovir in children with a renal transplant (Facchin 2019), parameterized in apparent oral clearances and volumes, with first-order absorption after a lag time and serum-creatinine, body-surface-area and sex effects on CL/F and Vc/F. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Facchin 2019 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Franck 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ganciclovir after IV ganciclovir and oral valganciclovir in pediatric solid-organ and stem-cell transplant recipients (Franck 2021), with first-order absorption after a lag time, estimated bioavailability, allometric body-weight scaling, a creatinine-clearance power effect on clearance, and a purely additive residual error. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Franck 2021 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Horvatits 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for IV ganciclovir in critically ill patients receiving continuous venovenous hemodiafiltration (Horvatits 2014). No covariates were retained because of the small cohort (n = 9), so every disposition parameter is a covariate-free typical value with between-subject variability. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Horvatits 2014 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Koloskoff_2025_ganciclovir.html">Ganciclovir (Koloskoff 2025)</a> </td> <td style="text-align:left;"> Indirect-response viral turnover PD model for cytomegalovirus (CMV) viral load decline in pediatric solid-organ and hematopoietic-stem-cell transplant recipients receiving (val)ganciclovir (Koloskoff 2025). The model treats the q12h-interval ganciclovir AUC (AUC_0-12) as a time-varying covariate input AUC_GCV that stimulates first-order viral degradation through an Emax-EC50 relationship. The upstream popPK that produces AUC_0-12 (Franck 2021 Bayesian estimator) is NOT included here; AUC_GCV must be supplied per record by the user, either from the Franck 2021 model or any other AUC source. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Krens 2020)</a> </td> <td style="text-align:left;"> One-compartment population PK model for IV ganciclovir in critically ill adults (Krens 2020), with a CKD-EPI eGFR power effect on clearance and a fixed central volume. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Krens 2020 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Lalagkas 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ganciclovir after IV ganciclovir and oral valganciclovir in Caucasian solid-organ transplant recipients with established CMV infection (Lalagkas 2023), with first-order absorption after a lag time, estimated bioavailability, allometric body-weight scaling, and a CKD-EPI eGFR power effect on clearance. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Lalagkas 2023 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Li 2021)</a> </td> <td style="text-align:left;"> One-compartment population PK model for IV ganciclovir in critically ill children (Li 2021), with a kidney-function power effect and near-linear body-weight scaling on clearance. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Li 2021 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Nguyen 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for IV ganciclovir and oral valganciclovir in a pediatric population (Nguyen 2021), with allometric body-weight scaling on all disposition parameters, a power effect of eGFR on clearance, and a multiplicative critical-illness factor on clearance. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Nguyen 2021 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Perrottet_2009_ganciclovir.html">Ganciclovir (Perrottet 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ganciclovir (administered as oral valganciclovir prodrug) in adult solid-organ transplant recipients (Perrottet 2009) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Yuen 1995)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for IV ganciclovir in patients with CMV retinitis, asymptomatic CMV urine shedding, or a solid-organ transplant with renal dysfunction (Yuen 1995), with an additive-intercept clearance whose renal-elimination term scales with body weight and creatinine clearance and is reduced in transplant recipients and in CMV-retinitis patients. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Yuen 1995 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Zhao 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ganciclovir after oral valganciclovir in pediatric renal transplant patients (Zhao 2009), parameterized in apparent oral clearances and volumes, with first-order absorption after a lag time and a two-component additive apparent clearance built from a steep creatinine-clearance power term plus a weight-proportional term. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Zhao 2009 when the primary is obtained -- the transcribed creatinine-clearance exponent of 2.93 is unusually steep and is flagged in the validation vignette. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2023_ganciclovir_model_repository.html">Ganciclovir (Zhou 1996)</a> </td> <td style="text-align:left;"> One-compartment population PK model for IV ganciclovir in newborns with acute symptomatic congenital CMV disease (Zhou 1996), with additive-linear renal-function scaling on clearance, additive-linear body-weight scaling on the central volume, and correlated between-subject variability on CL and Vc. Parameters transcribed from the Yang 2023 ganciclovir / valganciclovir population-PK model repository (Table 3), not from the primary publication; re-verify against Zhou 1996 when the primary is obtained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Grimm_2023.html">Gantenerumab (Grimm 2023)</a> </td> <td style="text-align:left;"> Gantenerumab PK model in cynomolgus monkeys (Grimm 2023): two-compartment plasma PK with brain extracellular distribution across six brain regions (brain_cerebellum, brain_hippocampus, brain_striatum, brain_cortex, choroid plexus, CSF). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/VanWart_2004_garenoxacin.html">Garenoxacin (VanWart 2004)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with first-order absorption and first-order elimination for oral garenoxacin (a des-F(6) quinolone) in adults with community-acquired respiratory tract infections (Van Wart 2004); CL/F covariates are creatinine clearance, ideal body weight, age, obesity (WT > 130% IBW), and concomitant pseudoephedrine; V/F covariates are body weight and male sex. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Smythe_2013_gatifloxacin.html">Gatifloxacin (Smythe 2013)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral gatifloxacin in adult African pulmonary tuberculosis patients co-administered rifampin, isoniazid, and pyrazinamide (Smythe 2013). Savic transit-compartment absorption (analytical form, N = 12.6, MTT = 0.65 h) feeds first-order absorption into a one-compartment disposition model. Apparent oral clearance is split into a GFR-mediated component scaled linearly with Cockcroft-Gault creatinine clearance and a non-GFR (other) component scaled allometrically with fat-free mass (FFM, Janmahasatian formula); apparent volume is scaled linearly with FFM. Age, sex, and HIV status modify the absorption rate constant. Relative bioavailability is fixed at 1 on the first dose and 11.7% lower at steady state. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chawla_2023_gefapixant.html">Gefapixant (Chawla 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and an absorption lag time for the P2X3-receptor antagonist gefapixant in healthy volunteers and adults with refractory or unexplained chronic cough (Chawla 2023) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jiang_2008_gemcitabine.html">Gemcitabine (Jiang 2008)</a> </td> <td style="text-align:left;"> Population PK model for intravenous gemcitabine and its primary inactive metabolite 2',2'-difluorodeoxyuridine (dFdU) in 94 adult patients with cancer pooled from three clinical studies (Jiang 2008 Br J Clin Pharmacol 65(3):326-333). Both gemcitabine and dFdU are described by a two-compartment model with first-order elimination, joined by a first-order formation step. The fraction of gemcitabine converted to dFdU (F) is not identifiable, so dFdU parameters are apparent (CL/F, Q/F, Vc/F, Vp/F); the NONMEM ADVAN6 parent-metabolite encoding treats the total gemcitabine clearance as the apparent formation flux into the dFdU central compartment. Retained covariates after forward inclusion / backward elimination (Table 4): estimated creatinine clearance on apparent dFdU clearance via a linear-additive scaling CL_dFdU/F = 0.04 * (1 + 0.48 * CRCL/70); body surface area (power exponent 0.93, reference 1.73 m^2), oxaliplatin co-administration order (multiplicative factors 0.65 when gemcitabine is given first and 0.54 when oxaliplatin is given first), and a non-small-cell lung cancer indicator (multiplicative factor 1.24) on apparent dFdU central volume. Residual error is proportional for both analytes. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2009_gemifloxacin.html">Gemifloxacin (Landersdorfer 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for gemifloxacin in healthy adults with first-order absorption + lag time, additive renal (filtration + saturable Michaelis-Menten tubular secretion with competitive probenecid inhibition) and non-renal clearance, and treatment-arm-static probenecid effects on absorption rate, absorption lag, and non-renal clearance (Landersdorfer 2009 gemifloxacin / probenecid DDI study). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Abbasi_2023_gentamicin.html">Gentamicin (Abbasi 2023)</a> </td> <td style="text-align:left;"> One-compartment intravenous pooled meta-analytic population PK model for gentamicin in critically ill adult ICU patients (n=1215 pooled from 21 published studies; Abbasi 2023). Vd (0.33 +/- 0.20 L/kg), CL (4.70 +/- 2.89 L/h) and total body weight (70.8 +/- 19.9 kg) were pooled as mean +/- SD across the 21 studies and used as normal-distribution inputs to a Monte Carlo Simulation (10,000 virtual patients, Crystal Ball, Oracle) for probability-of-target-attainment (PTA) analysis of gentamicin once-daily doses 5-10 mg/kg (0.5-h IV infusion, 24-h dosing interval) against AUC24h/MIC and Cmax/MIC efficacy targets. The reported variability reflects combined between-study and between-patient effects rather than a NONMEM-fitted OMEGA. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bijleveld_2016_gentamicin.html">Gentamicin (Bijleveld 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of intravenous gentamicin in term neonates with hypoxic-ischaemic encephalopathy undergoing controlled hypothermia (Bijleveld 2016), with fixed allometric body-weight scaling (exponents 0.75 on CL and Q, 1 on Vc and Vp), an estimated gestational-age power effect on CL, and a categorical post-rewarming (study-day-5, > 96 h PNA) multiplicative increase in CL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bijleveld_2017_gentamicin.html">Gentamicin (Bijleveld 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of intravenous gentamicin in (pre)term neonates with suspected or proven Gram-negative sepsis (Bijleveld 2017), with fixed allometric body-weight scaling (exponents 0.75 on CL and Q, 1 on Vc and Vp) and an estimated postmenstrual-age power effect on CL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Frymoyer_2013_gentamicin.html">Gentamicin (Frymoyer 2013)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for gentamicin in 29 term neonates with hypoxic ischemic encephalopathy (HIE) receiving therapeutic hypothermia (Frymoyer 2013), with fixed allometric birth-weight scaling (exponent 0.75 on CL, 1 on Vc, reference 3.3 kg) and a power effect of serum creatinine on PNA day 1 on CL (reference 0.9 mg/dL). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fuchs_2014_gentamicin.html">Gentamicin (Fuchs 2014)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for gentamicin in 1449 preterm and term neonates (Fuchs 2014) with fixed allometric body-weight scaling (0.75 on CL/Q, 1 on Vc/Vp), linear centred-on-median effects of gestational age on CL and Vc, postnatal age on CL, and dopamine co-administration on CL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hodiamont_2017_gentamicin.html">Gentamicin (Hodiamont 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of intravenous gentamicin in critically ill adult ICU patients (Hodiamont 2017) estimated without retained covariates, with correlated between-subject variability on CL and central volume V1, combined additive plus proportional residual error, and substantial inter-occasion variability on CL and V1 reported in the source (documented in the vignette assumptions, not encoded structurally). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Llanos_2017_gentamicin.html">Gentamicin (Llanos 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of gentamicin in pediatric oncology patients with febrile neutropenia (Llanos-Paez 2017) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Llanos-Paez_2017_gentamicin.html">Gentamicin (Llanos-Paez 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for gentamicin in pediatric oncology patients (Llanos-Paez 2017 AAC) extended with a renal-cortex accumulation compartment and an Emax model of relative renal-function reduction (Llanos-Paez 2017 AAPS J). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Llanos-Paez_2020_gentamicin.html">Gentamicin (Llanos-Paez 2020)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for gentamicin in pediatric oncology and nononcology patients (Llanos-Paez 2020); body composition is described by normal fat mass (NFM = FFM + Ffat * (TBW - FFM)) with separate Ffat estimates for CL (0.48) and V1 (0.10) and Ffat fixed to 0 for Q and V2; CL is driven by Holford 2017 GFR-maturation (PMA-based Hill function) and a power ratio of age/sex-matched physiological mean serum creatinine (Ceriotti 2008) over individual SCR; oncology cohort has 15.4% lower V1 and 32.1% lower Q than nononcology. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/MedellinGaribay_2015_gentamicin.html">Gentamicin (MedellinGaribay 2015)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for gentamicin in infants 1-24 months (Medellin-Garibay 2015) with linear body-weight scaling on CL and central volume Vc and an additive (CLCR/75)-driven term on CL; intercompartmental clearance Q and peripheral volume Vp are not weight-scaled in the published parameterisation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mohamed_2012_gentamicin.html">Gentamicin (Mohamed 2012)</a> </td> <td style="text-align:left;"> In vitro (Escherichia coli ATCC 25922). Semi-mechanistic PKPD model of gentamicin bactericidal activity with adaptive resistance: drug-susceptible growing bacteria (bact_growing) plus insusceptible resting bacteria (bact_resting), with a binding model (ar_off / ar_on) by which gentamicin reduces its own Emax. Fit jointly to static and dynamic in-vitro time-kill curves. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sampson_2014_gentamicin.html">Gentamicin (Sampson 2014)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model of gentamicin in term neonates with hypoxic-ischaemic encephalopathy undergoing whole-body hypothermia, as reported (model originally developed by Frymoyer 2013; this model file reproduces the parameter values stated by Sampson 2014 during the model's external predictive-performance evaluation). Allometric body-weight scaling on CL (fixed exponent 0.75) and linear body-weight scaling on V (exponent 1) referenced to a 3.3 kg neonate, with a power effect of serum creatinine on CL (exponent -0.566); inter-individual variability on CL only and proportional residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Singu_2024_gentamicin.html">Gentamicin (Singu 2024)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for gentamicin in 50 PK-evaluable Namibian neonates (82.7% preterm) with suspected or confirmed sepsis receiving 5 mg/kg every 24 h, developed from a truncated (7.5 h) two-sample informative-block randomised sampling design (Singu 2024). Clearance carries an allometric birth-weight effect (exponent 1.30, reference 1.57 kg), a power effect of log-transformed white blood cell count (exponent -0.560, reference log-WBC 2.39), and a logistic postnatal-age maturation function FMAT = PNA^GAMMA / (PNA^GAMMA + PNA50^GAMMA) with GAMMA and PNA50 fixed at 0.551 and 0.0332 years; volume carries an allometric birth-weight effect (exponent 1.76, reference 1.57 kg). Residual variability is a log-scale combination (additive + proportional) model with two concentration-dependent magnitudes switched at a predicted concentration of 2264.25 ng/mL (log(concentration) = 7.725), an approach the authors take from Hussein 2005; see modellib('Hussein_2005_levocetirizine'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Staatz_2005_gentamicin_vancomycin.html">Gentamicin (Staatz 2005)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for gentamicin in adult cardiothoracic-surgery patients with unstable renal function (Staatz 2005). Clearance scales linearly with raw Cockcroft-Gault creatinine clearance centred at the population baseline median (63 mL/min); central and peripheral volumes scale linearly with body weight; intercompartmental clearance is a population constant. Operator-resolved sidecar (request-001) replaced the paper's Wahlby 2004 baseline-CrCl + change-from-baseline (BCOV+DCOV) decomposition with the simpler CrCl-only covariate form to avoid adding a new canonical baseline-CrCl column; for stable-CrCl subjects the published final-model parameters reproduce the paper's CL exactly because DCOV is zero (see vignette Errata). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Thomson_2003_gentamicin.html">Gentamicin (Thomson 2003)</a> </td> <td style="text-align:left;"> One-compartment population PK model of intramuscular gentamicin in African infants with suspected severe sepsis (Thomson 2003). The 8 mg/kg i.m. dose is modelled as an IV bolus into the central compartment because first-order absorption could not be characterised from the sparse 1 h / next-morning sampling (the paper documents that ka estimates were poorly identified and absorption appeared complete by 1 h). Apparent clearance scales linearly with body weight and as a power function of (postnatal age + 1 day) normalised to the cohort median + 1 day; apparent volume of distribution scales linearly with body weight relative to the cohort median 3 kg. Reported CL and V are apparent values (CL/F, V/F) because all doses were administered by intramuscular injection and bioavailability could not be estimated. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Veinstein_2013_gentamicin.html">Gentamicin (Veinstein 2013)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous gentamicin in critically ill adult ICU patients with acute kidney injury undergoing 4-hour intermittent hemodialysis (n=10, all male; 6 mg/kg infused over 30 min, with hemodialysis starting 30 min after the end of the infusion; Veinstein 2013). Disposition is parameterised in terms of non-hemodialysis (interdialytic body) clearance, an additive hemodialysis-arm clearance, and volume of distribution. The dialysis arm is gated on/off by the time-varying RRT_HEMODIAL_ACTIVE covariate. Body weight enters the model as a linear (exponent = 1) structural scaler on all three parameters because the published Table 4 estimates are reported per kg; weight was tested as an explicit covariate on V and not retained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wahlby_2004_time_varying_covariates.html">Gentamicin (Wahlby 2004)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous gentamicin in 210 cancer patients, demonstrating Wahlby 2004's extended covariate-model formulation in which the within- and between-subject components of a time-varying covariate are entered as separate model terms. Final-model clearance depends on baseline creatinine clearance (CRCL_BASE) and the time-varying delta-from-baseline (CRCL - CRCL_BASE); central volume depends on baseline body surface area (BSA_BASE) and the time-varying albumin (ALB) ratio (ALB/34)^-0.41. Underlying structural PK (compartment count, IIV/residual-error structure) follows Rosario et al. 1998 (Br J Clin Pharmacol 46:229-236). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kelman_1984_gentamicin.html">Gentamicin glasgow (Kelman 1984)</a> </td> <td style="text-align:left;"> One-compartment population PK model for gentamicin in neonates and very young infants (Kelman 1984, Glasgow I cohort; n=43, postnatal age 2-120 days, body weight 0.8-3.7 kg). Clearance is a linear-additive function of body weight, postnatal age, and serum creatinine (paper Equation 4, Model 1: CL = theta1*WT + theta2*AGE + theta3*CREAT); volume of distribution is proportional to body weight (Equation 5: V = theta4*WT). Encoded from Kelman 1984 Table 2 Model 1 (FULL model, paper's best by NONMEM objective function). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kelman_1984_gentamicin.html">Gentamicin manchester1 (Kelman 1984)</a> </td> <td style="text-align:left;"> One-compartment population PK model for gentamicin in neonates and young infants (Kelman 1984, Manchester I cohort; n=32, postnatal age 1-153 days, body weight 1.6-9.1 kg). Clearance is a linear-additive function of body weight, postnatal age, and serum creatinine (paper Equation 4, Model 1: CL = theta1*WT + theta2*AGE + theta3*CREAT); volume of distribution is proportional to body weight (Equation 5: V = theta4*WT). Encoded from Kelman 1984 Table 3 Model 1 (FULL model, paper's best by NONMEM objective function for this cohort). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kelman_1984_gentamicin.html">Gentamicin manchester2 (Kelman 1984)</a> </td> <td style="text-align:left;"> One-compartment population PK model for gentamicin in older children and adolescents (Kelman 1984, Manchester II cohort; n=68, age 6 months-15 years, body weight 5.7-62 kg). Clearance is a linear-additive function of body weight, age (in years), and serum creatinine (paper Equation 4, Model 1: CL = theta1*WT + theta2*AGE + theta3*CREAT); volume of distribution is proportional to body weight (Equation 5: V = theta4*WT). Encoded from Kelman 1984 Table 4 Model 1 (FULL model, paper's best by NONMEM objective function for this cohort). For Manchester II the paper's theta2 coefficient is in L/h per year (sign negative), implying age units are years rather than the days used in the Glasgow I and Manchester I cohorts; encoded here with the canonical AGE-in-years covariate. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cao_2013_gevokizumab.html">Gevokizumab (Cao 2013)</a> </td> <td style="text-align:left;"> Second-generation minimal physiologically-based PK (mPBPK) model for gevokizumab in adults (Cao 2013 Model A; clearance from plasma) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Karakitsios_2024_gevokizumab.html">Gevokizumab (Karakitsios 2024)</a> </td> <td style="text-align:left;"> mPBPK. Second-generation minimal physiologically-based population PK model for gevokizumab (anti-interleukin-1beta humanized IgG2) in adults with type 2 diabetes, with between-subject variability estimated from AGGREGATE (mean and SD versus time) plasma concentration data rather than individual records. Karakitsios and Dokoumetzidis re-estimated the three drug-specific parameters of the Cao 2013 second-generation mPBPK structure (the tight- and leaky-tissue vascular reflection coefficients and plasma clearance) plus two IIV terms, using a Bayesian reconstruct-the-likelihood-from-aggregate-data method implemented in RStan: at each MCMC iteration a Latin-hypercube virtual population is solved, its mean and SD versus time are computed, and those summaries are fitted to the digitised published mean and SD profiles. This file holds the SINGLE dosage-group fit (7 mg IV, n = 10; Table 3). A single lognormal body-size random effect (etalvc) is shared by all four physiologic volumes, per the authors' assumption that a larger patient has proportionally larger plasma, tight-tissue ISF, leaky-tissue ISF and lymph volumes. The five-dosage-group meta-analysis from the same paper is a separate model; see modellib('Karakitsios_2024_gevokizumab_mbma'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Karakitsios_2024_gevokizumab.html">Gevokizumab mbma (Karakitsios 2024)</a> </td> <td style="text-align:left;"> mPBPK / MBMA. Five-dosage-group meta-analysis of gevokizumab (anti-interleukin-1beta humanized IgG2) in adults with type 2 diabetes, fitted to AGGREGATE (mean and SD versus time) plasma concentration data from five dose arms simultaneously. The structural model is the Cao 2013 second-generation minimal PBPK model; Karakitsios and Dokoumetzidis estimated its three drug-specific parameters (tight- and leaky-tissue vascular reflection coefficients and plasma clearance) hierarchically, with an exponential INTER-GROUP variability (IGV) term on each, following a log-Student's t distribution with 5 degrees of freedom to down-weight outlying dose arms. The random effects in this file are therefore DOSAGE-GROUP-level (eta_study_*), not between-subject: the model simulates group-mean concentration-time profiles and is NOT suitable for individual-subject simulation. Between-subject variability was estimated separately for each of the five arms (a semi-hierarchical design with no distribution assumed across arms), so no single population IIV exists; the five per-arm pairs are tabulated in population$notes. For an individual-level version of the same structure, see the single dosage-group fit modellib(‘Karakitsios_2024_gevokizumab’).
Givosiran (Ayyar 2024) Mechanistic translational PK model for the GalNAc-siRNA givosiran (Ayyar & Song 2024) parameterized for human (70 kg adult). 22-ODE system covering SC depot, central plasma (parent + AS(N-1)3’ active metabolite), competitive ASGPR receptor binding (free target, parent-target complex, metabolite-target complex), receptor-mediated hepatocyte internalization, endolysosomal sequestration / degradation / endosomal escape, free cytoplasmic siRNA, RISC-loaded siRNA (combined parent + metabolite), kidney vascular and tissue distribution with a deep bound pool and GFR elimination - for parent and metabolite. Pharmacodynamic ALAS1 mRNA silencing (rat-only in the paper) is not included in the human parameterization.
Glasdegib QTcF (Fostvedt 2021) Population PD model for glasdegib concentration-driven prolongation of the QT interval corrected for heart rate using Fridericia’s formula (QTcF) in 70 adult patients with advanced cancer pooled from two phase 1 dose-escalation trials (B1371001 in hematologic malignancies; B1371002 in solid tumors). The exposure-response form is a linear mixed-effects model: QTcF = theta1 + theta2 * (CP_GLASDEGIB_NGML / 1000) + eta1 + W * eps, with an additive random effect on the intercept (eta1 ~ N(0, omega2_1)) and a ‘thetarized’ additive residual error (W is a fitted scalar; eps ~ N(0, 1)). The covariate analysis (age, sex, study) retained no covariates; a random effect on the slope was considered and dropped because shrinkage exceeded 20%. PD-only model: plasma glasdegib concentration is supplied as a time-varying covariate CP_GLASDEGIB_NGML (ng/mL). The slope is reported in the source publication on the microgram-per-mL scale (4.3 msec per microgram per mL) which is equivalent to 0.0043 msec per ng/mL; this file keeps the slope in the paper’s microgram-per-mL form and applies the unit conversion CP_GLASDEGIB_NGML / 1000 inside model(). The source publication does not fit a population PK model; users wishing to drive the PD model from a simulated PK source must supply their own concentration trajectory (no glasdegib popPK model exists in the nlmixr2lib registry).
Glasdegib QTcS (Fostvedt 2021) Population PD model for glasdegib concentration-driven prolongation of the QT interval corrected for heart rate using a population- specific Fridericia-style correction factor (QTcS; beta estimated at 0.312 in this cohort versus the fixed beta = 1/3 of QTcF) in 70 adult patients with advanced cancer pooled from two phase 1 dose- escalation trials (B1371001 in hematologic malignancies; B1371002 in solid tumors). The exposure-response form is the same linear mixed-effects model as the companion QTcF extraction: QTcS = theta1 + theta2 * (CP_GLASDEGIB_NGML / 1000) + eta1 + W * eps, with additive random intercept (eta1 ~ N(0, omega2_1)) and a ‘thetarized’ additive residual error (W is a fitted scalar; eps ~ N(0, 1)). Covariate analysis (age, sex, study) retained no covariates; a random effect on the slope was considered and dropped because shrinkage exceeded 20%. PD-only model: plasma glasdegib concentration is supplied as a time-varying covariate CP_GLASDEGIB_NGML (ng/mL). The slope is reported in the source publication on the microgram-per-mL scale (4.31 msec per microgram per mL); this file keeps that scaling and applies the unit conversion CP_GLASDEGIB_NGML / 1000 inside model(). The source publication does not fit a population PK model; users wishing to drive the PD model from a simulated PK source must supply their own concentration trajectory (no glasdegib popPK model exists in the nlmixr2lib registry).
Glibenclamide (Rambiritch 2016) Two-compartment population PK model with first-order oral absorption for glibenclamide in poorly controlled South African adults with type 2 diabetes (Rambiritch 2016). All disposition parameters are apparent (CL/F, Vc/F, Vp/F, Q/F); F is not estimated. Concentration data were log-transformed prior to NONMEM fitting (LTBS), giving an effectively proportional residual error in linear space. No covariate effects were retained in the final model.
Glp1ra bodyweight (Bosch 2024) QSP. GLP-1R agonist body composition model (Bosch 2024) extending the Hall 2009 three-compartment energy-balance model with (1) an inverse-Bateman lifestyle-change effect on energy intake, (2) a body-weight-dependent activity effect on physical activity energy expenditure for studies that included weight management and intensive behavioural treatment, and (3) a GLP-1R agonist drug effect driven by the in-vitro EC50- normalised free drug concentration, with a time-dependent tolerance term that shifts the in-vivo EC50 upward. Liraglutide and semaglutide PK are encoded inline as fixed one-compartment first-order absorption models (parameters from the Bosch 2024 supplement S10 reproducing FDA clinical pharmacology review (liraglutide, 17 Dec 2018) and Carlsson Petri et al. 2018 (semaglutide); both PK paths share the body composition machinery, and the total normalised free concentration drives the GLP-1R effect so a user simulating a single drug doses to that drug’s depot only. Body weight (kg) and percent change from baseline are the primary observation outputs. Initial conditions are derived from baseline body weight, BMI, age, sex and height via the Jackson body-fat regression and the Mifflin resting-metabolic-rate equation; baseline energy intake is set to maintain steady state at PAL = 1.6 (low-active-to-active). 11 active ODEs (3 macronutrient stores, 2 extracellular-water states, lipolysis diet target, adaptive thermogenesis, plus 2 first-order PK chains for each drug).
Glp1ra hba1c (Bosch 2025) QSP. Integrated glucose-red blood cell-HbA1c (IGRH) sub-model from the Bosch 2025 4GI-HbA1c systems framework, used to predict long-term HbA1c response from a time-varying plasma glucose driver in adults with type 2 diabetes mellitus receiving GLP-1R / GLP-1R + glucagon receptor agonists (cotadutide, liraglutide). The model is a 24-state transit chain (12 unglycated red blood cell age cohorts + 12 glycated cohorts; NC = 12 transit compartments per Bosch 2025 supplement model code S2) with a glucose-concentration-dependent shortening of the RBC life span; the HbA1c output is the percentage glycated fraction of the total RBC pool. All structural parameters are fixed from the Lledo-Garcia 2013 / Kjellsson 2015 IGRH publications and held constant during the Bosch 2025 calibration; only the residual error and the IIV on the RBC life span were estimated on the cotadutide Ph2a HbA1c dataset (Bosch 2025 Table 2, third-block ‘IGRH model’). Plasma glucose drive is supplied as the time-varying regressor GLU in mmol/L (linearly interpolated by rxode2 between dataset rows) and the per-subject baseline glucose anchor FPG is in mmol/L; both are converted to mg/dL inside model() to match the published IGRH parameterisation (KG in dL/mg/day, reference glucose 149 mg/dL = 8.27 mmol/L).
Glucarpidase (Kimura 2023) Modified Michaelis-Menten PK/PD simulation model for glucarpidase (CPG2) rescue after high-dose methotrexate (Kimura 2023). MTX disposition is 2-compartment IV with renal-only first-order elimination (Kr fixed at ~10% of literature total MTX CL from Fukahara 2008); the remaining elimination is captured by a saturable hydrolysis term coupled to a 1-compartment IV CPG2 disposition. All structural parameters are literature-sourced point values (no estimation in the source paper).
Glutamine (Sadaf 2024) One-compartment population PK model with first-order absorption for oral L-glutamine (Endari) in children and adults with sickle cell disease and in healthy adult volunteers (Sadaf 2024). The model is fitted to baseline-subtracted plasma L-glutamine, i.e. the exogenous increment above each participant’s own endogenous pre-dose concentration, so the predicted Cc is zero before the first dose. Apparent oral clearance and apparent central volume are allometrically scaled to body weight with the theoretical exponents 0.75 and 1 (70 kg reference). Apparent clearance falls as the pre-dose endogenous L-glutamine baseline rises (power -0.96, reference 683 umol/L), interpreted as competition between endogenous and exogenous glutamine for elimination. Apparent volume rises with the administered dose per kg (power 0.27, reference 0.1 g/kg), which is how the paper captures the capacity-limited, less-than-dose-proportional rise in exposure seen at 0.3 and 0.6 g/kg after a Michaelis-Menten base model proved unstable. Inter-occasion variability is carried on clearance across the four study visits; inter-individual variability on clearance and on the absorption rate constant, and the additive residual error, were all held at zero in the published final model.
GNbAC1 (Cao 2013) Second-generation minimal physiologically-based PK (mPBPK) model for GNbAC1 in adults (Cao 2013 Model A; clearance from plasma)
Granisetron (Li 2023) One-compartment population PK model for the granisetron transdermal delivery system (GTDS, Sancuso 34.3 mg/52 cm2 patch) in healthy Caucasian volunteers, with first-order absorption from the patch (no lag time) and first-order elimination; no covariates retained (Li 2023)
Guselkumab (Chen 2022) One-compartment population PK model with first-order SC absorption for guselkumab (anti-IL-23 human IgG1 lambda mAb) in patients with active psoriatic arthritis (DISCOVER-1 and DISCOVER-2 phase 3 trials; Chen 2022)
Guselkumab (Yao 2018) One-compartment population PK model with first-order SC absorption and first-order elimination for guselkumab (anti-IL-23 p19 human IgG1-lambda mAb) in adults with moderate-to-severe plaque psoriasis (pooled phase 2 X-PLORE and phase 3 VOYAGE 1 / VOYAGE 2 trials; Yao 2018)
Haloperidol (Franken 2017) One-compartment population PK model for haloperidol in 28 terminally ill adult palliative-care patients (Franken 2017). Two parallel first-order absorption routes (oral and subcutaneous) with route-specific absorption rate constants fixed from literature (Ka oral = 0.236 1/h, Ka SC = 20 1/h derived from intramuscular Tmax = 20 min). Oral bioavailability F = 0.861 is estimated; SC F is assumed to be 1. IIV is included on F, CL, and Vd; the IIV on F and CL was 99% correlated and is encoded with correlation fixed to unity (BLOCK pattern). Residual variability is additive on log-transformed concentrations (LTBS). Covariate analysis (body weight, age, sex, primary diagnosis, plasma creatinine, urea, bilirubin, GGT, ALP, ALT, AST, CRP, albumin, concomitant CYP2D6 / CYP3A inducers and inhibitors, time-to-death) did not retain any covariate in the final model.
Haloperidol panss subscales (PillaReddy 2013) Population pharmacokinetic / pharmacodynamic (PK/PD) model for haloperidol against the three PANSS subscales (positive, negative, general) in adults with schizophrenia from Pilla Reddy 2013 Part II (pooled dataset of 1338 placebo-treated patients with subscale data available for n=741 of them; 12 industry-sponsored Phase II/III trials between 1989 and 2009 plus one open-label study). The PK sub-model uses the haloperidol structural model inherited from Part I (PMID 23473810): a simplified one-compartment representation parameterised with the typical apparent oral clearance CL/F = 88 L/h, apparent central volume Vc/F = 669 L, and first-order absorption rate ka = 0.23 1/h reported in Part I Table 2 (Pilla Reddy 2012a originally described the haloperidol PK as two-compartment with Q/F = 233 L/h and Vp/F = 2500 L; the one-compartment simplification preserves the steady-state average concentration Css = Dose / (CL * tau) used as the PD driver and is documented as a deliberate Part II simplification in the vignette Assumptions and deviations). The PD sub-model has three outputs that share the Weibull placebo time-course form Pplacebo = Pmax * (1 - exp(-(t/TD)^POW)) but each subscale carries its own placebo baseline, Pmax, TD, and POW (Pilla Reddy 2013 Part II Table 1) and haloperidol’s own Emax / EC50 / KT triplet per subscale (Part II Table 2). The combined PANSS subscale prediction is BASL * (1 - Pplacebo) * (1 - Drug), where Drug = Emax * Cc / (EC50 + Cc) * (1 - exp(-KT * t)) with the KT delay capturing the onset time to the maximum drug effect. Concentration drives the PD with a steady-state assumption: the paper uses Css from the PK model and feeds it into the Emax equation; in the rxode2 implementation Cc is the time-varying plasma concentration derived from the one-compartment ODE, which is approximately constant at steady state for the once- and twice-daily regimens studied here. An exponential time-to-event dropout sub-model with subscale- and drug-specific baseline hazards (Part II Table 4) is reported in the paper but is not encoded in this nlmixr2lib model body; the dropout parameters are documented in population$dropout_model and discussed in the vignette. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Feng_2012_higenamine.html">Higenamine (Feng 2012)</a> </td> <td style="text-align:left;"> Population PK/PD model for intravenous higenamine in 10 healthy Chinese subjects (Feng 2012). Two-compartment disposition with Michaelis-Menten (saturable) elimination from the central compartment, plus a direct-effect Emax sub-model for the cardiovascular-stress heart-rate response (E = E0 + Emax * Cc / (EC50 + Cc)). No demographic covariates were retained in the final model (sex, height, weight, BMI, and age were graphically screened but did not influence PK or PD). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ngo_2020_HL2351.html">HL2351 (Ngo 2020)</a> </td> <td style="text-align:left;"> Population PK model for HL2351 (hIL-1Ra-hyFc, ~97 kDa) in healthy adult Korean men: a quasi-steady-state target-mediated drug disposition (QSS-TMDD) model coupled with FcRn-mediated recycling. The injection-site depot feeds a separate distribution space where free drug equilibrates with FcRn (QSS dissociation constant AKSS1, total FcRn AFcRn_t); free drug moves to the central compartment either directly (Ka2) or by FcRn-mediated recycling of the FcRn-drug complex (Krec). In the central compartment free drug equilibrates with IL1R (QSS dissociation constant KSS2, total IL1R CIL1R_t), is taken up back to the distribution space (Kup), exchanged with one peripheral compartment (Q/F), and eliminated linearly (CL/F). The IL1R-drug complex degrades at Kdeg2. All drug amounts and concentrations are in nmol / nmol/L; convert mg dosing using molecular weight 97 kDa (1 mg HL2351 = approximately 10306 nmol). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Royer_2010_HuHMFG1.html">HuHMFG1 (Royer 2010)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with linear elimination for HuHMFG1 (AS1402), a humanised anti-MUC1 monoclonal antibody, in patients with metastatic breast cancer; serum AST enters the typical clearance equation additively (Royer 2010) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Carmichael_2003_hydroxychloroquine.html">Hydroxychloroquine (Carmichael 2003)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model with an absorption lag time for oral hydroxychloroquine (HCQ) whole-blood concentrations in 123 adult rheumatoid arthritis patients (74 on HCQ alone plus 49 on HCQ + methotrexate) pooled from four Australian studies, with bioavailability fixed at the value 0.746 estimated from a nine-patient IV/oral crossover sub-study and a linear additive shift in central volume of distribution for concomitant methotrexate coadministration (V_MTX = 1070 L added to the base V of 605 L when MTX is present) (Carmichael 2003). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sharma_2016_hydroxyprogesteroneCaproate.html">HydroxyprogesteroneCaproate (Sharma 2016)</a> </td> <td style="text-align:left;"> Population PK model for 17alpha-hydroxyprogesterone caproate (17-OHPC) in pregnant women with singleton gestation receiving 250 mg weekly IM injections for prevention of recurrent preterm birth (Sharma 2016). Structural model is a maternal central compartment with first-order IM absorption (ka fixed to 3 /day) and first-order elimination, linked by reversible first-order rate constants kMF / kFM to a fetal compartment whose amount is tracked as a dynamic state. Allometric (power) scaling on apparent CL/F (exponent 0.80) and apparent Vmaternal/F (exponent 0.84) around the cohort median weight of 68 kg. Inter-individual variability is encoded as a shared random effect on log(CL) with an estimated scale factor (Theta6 = 1.90) applied to the same eta on log(Vmaternal), preserving the > 0.9 ETA correlation reported in the source. Inter-occasion variability (PK1 = 20-24 weeks gestation, PK2 = 31-34 weeks gestation) on CL and Vmaternal is encoded as paired per-occasion etas keyed to OCC. Residual error is combined proportional + additive on linear-scale ng/mL concentrations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_guhongInjection.html">HydroxysafflorYellowA rat (Chen 2024)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment intravenous pharmacokinetic model for hydroxysafflor yellow A (HSYA), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the hydroxysafflor yellow A dose is the GHI volume dose times its content in GHI (1 mg/mL), i.e. 2.5 mg/kg, 5 mg/kg, 10 mg/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Direct-effect sigmoid-Emax models link the hydroxysafflor yellow A plasma concentration to the GHI-minus-model-group difference in creatine kinase-MB (CK-MB), ischemia-modified albumin (IMA), cardiac troponin I (cTn I) and alpha-hydroxybutyrate dehydrogenase (alpha-HBDH) (E = Emax * C^gamma / (ED50^gamma + C^gamma); Tables 12, 13, 14, 15). Chen 2024 fitted a PK/PD model only for the analyte/biomarker/dose combinations whose PLSR coefficient was negative, so the effect of an unfitted combination is returned as zero rather than extrapolated. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dong_2016_hydroxyurea.html">Hydroxyurea (Dong 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral hydroxyurea in pediatric patients with sickle cell anaemia (Dong 2016, HUSTLE trial NCT00305175; n = 96 children aged 1.2-16.6 years on a 20 mg/kg starting dose, all African-American). Saturable Michaelis-Menten elimination from the central compartment (Vmax 490 mg/h per 70 kg, Km fixed at 25 mg/L based on a prior report) with allometric scaling of Vmax (exponent 0.75 fixed) and apparent central volume V/F (exponent 1.0 fixed) on total body weight (reference 70 kg). Cystatin C is a power-model covariate on Vmax (exponent -0.509, reference 0.74 mg/L), and was the only covariate retained over serum creatinine, eGFR, and direct 99mTc-DTPA-measured GFR. Oral absorption is described by a Savic 2007 transit-compartment chain (NN = 12.4 transit compartments fitted, MTT = 0.158 h) feeding a depot with first-order absorption Ka = 8.19 /h. Residual error is combined additive (0.117 mg/L) and proportional (39.7% CV). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Paule_2011_hydroxyurea.html">Hydroxyurea (Paule 2011)</a> </td> <td style="text-align:left;"> Two-compartment population PK + indirect-response PD models for hydroxyurea (HU) in adults with sickle cell anemia (Paule 2011): bicompartmental oral PK with first-order absorption and elimination, allometric scaling on CL/F and Vc/F; turnover models for HbF percentage and mean corpuscular volume (MCV) where HU inhibits the elimination rate of each PD response. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_guhongInjection.html">Hyperoside rat (Chen 2024)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment intravenous pharmacokinetic model for hyperoside (HYP), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the hyperoside dose is the GHI volume dose times its content in GHI (3.4 ug/mL), i.e. 8.5 ug/kg, 17 ug/kg, 34 ug/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Chen 2024 fitted no concentration-effect model for hyperoside: its weighted PLSR regression coefficients against all four myocardial biomarkers were positive (Table 10), so the paper's PK/PD tables (Tables 12-15) contain no entry for it and this file carries the PK layer only. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pillai_2004_ibandronate.html">Ibandronate (Pillai 2004)</a> </td> <td style="text-align:left;"> Kinetic-pharmacodynamic (K-PD) model for ibandronate (a nitrogen-containing bisphosphonate) suppression of urinary C-telopeptide of type-I collagen (uCTX) in postmenopausal women with osteoporosis. A virtual K-PD effect compartment receives the administered dose and decays at rate KDE, producing a dose-driving rate (DODR = KDE * central * F) that inhibits the uCTX synthesis rate KS via a sigmoid Emax (inhibition-fraction form). uCTX follows an indirect-response synthesis-degradation turnover (KS, KD). A multiplicative placebo disease-progression drift (1 + SLOPE * t) and a calcium + vitamin D supplementation suppression term (1 - VIT * [1 - exp(-KVIT * t)]) modify the observed uCTX. The supplementation indicator CONMED_CAVITD also switches KDE between the no-supplement (0.112 /day) and with-supplement (0.014 /day) typical values. The model handles intravenous and oral routes via the bioavailability factor F (default F=1 for IV; oral users override lfdepot to log(0.008) for 2.5 mg oral or log(0.007) for 20 mg oral, both fixed from a separate absolute-bioavailability study). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ibrahim_2023_ibrutinib.html">Ibrutinib competing risk (Ibrahim 2023)</a> </td> <td style="text-align:left;"> Competing-risk multistate model for study dropout and death in patients with chronic lymphocytic leukemia treated with ibrutinib (phase Ib/II PCYC-1102). Three ODE states track the marginal probability of being alive and on study (s_alive, the only transient state), of having dropped out (s_dropout), and of having died (s_death); both terminal states are absorbing. The transition rate to dropout falls as the model-predicted leukocyte count rises (hazard ratio 4.92 for every 10-unit decrease in leukocyte count -- patients whose predicted count fell below normal, potentially reflecting neutropenia, were the most likely to leave the study), while the transition rate to death rises with the model-predicted lymph-node burden (hazard ratio 1.35 for every 10-unit increase in SPD) and with the deletion(17p) chromosomal abnormality (hazard ratio 4.16). The two time-varying predictors are supplied as covariates and are intended to be the model-predicted leukocyte count and SPD from the sister efficacy model, held constant between visits as in the source analysis. There are no drug-dosing events in this model and ibrutinib exposure does not enter it directly. Sister model files from the same paper: modellib('Ibrahim_2023_ibrutinib_leukocyte_spd'), modellib('Ibrahim_2023_ibrutinib_sbp'), modellib('Ibrahim_2023_ibrutinib_dbp'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ibrahim_2023_ibrutinib.html">Ibrutinib dbp (Ibrahim 2023)</a> </td> <td style="text-align:left;"> Ibrutinib-induced diastolic blood pressure (dBP) turnover model in patients with chronic lymphocytic leukemia, from the phase Ib/II PCYC-1102 study. A zero-order production / first-order loss turnover model preceded by a single transit compartment reproduces the slow onset of ibrutinib-associated hypertension; ibrutinib stimulates the zero-order dBP production rate through an Emax function of the daily AUC(0-24). The transit rate constant and the turnover rate constant are equal, both set to (n + 1) / MTT with n = 1 transit compartment. Baseline age lowers the dBP baseline. Structurally identical to the sister systolic model but fitted independently, with a smaller Emax (0.0694 vs 0.113), a lower AUC50 (63.1 vs 91.7 h*ng/mL), a roughly 2-fold slower onset (MTT 161 vs 79.9 days), no random effect on MTT, and the age effect placed on the baseline rather than on MTT. Ibrutinib enters only through the time-varying covariate AUC_IBRU; there are no drug-dosing events in this model. Sister model files from the same paper: modellib('Ibrahim_2023_ibrutinib_sbp'), modellib('Ibrahim_2023_ibrutinib_leukocyte_spd'), modellib('Ibrahim_2023_ibrutinib_competing_risk'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ibrahim_2023_ibrutinib.html">Ibrutinib leukocyte spd (Ibrahim 2023)</a> </td> <td style="text-align:left;"> QSP / semi-mechanistic PK-PD. Integrated ibrutinib exposure-response model for leukocyte count and lymph-node burden (sum of the products of perpendicular diameters, SPD) in chronic lymphocytic leukemia (CLL), from the phase Ib/II PCYC-1102 study. Five ODEs: one relative-quantity turnover compartment for phosphorylated Bruton tyrosine kinase (pBtk) whose zero-order production is inhibited by an Imax function of the daily ibrutinib AUC(0-24), plus four CLL cell subpopulations that reproduce the Calissano CLL life cycle -- two proliferating stroma-attached colonies in lymphoid tissue with different detachment sensitivities (cll_subpop1, cll_subpop2), a released/activated tissue pool that exits to blood (cll_subpop3), and a resting peripheral-blood pool (cll_bld). Free (unphosphorylated) Btk drives four simultaneous drug effects: inhibition of CLL proliferation, ibrutinib-induced apoptosis of cll_subpop3, enhanced detachment of cll_subpop1/cll_subpop2 from the stroma, and blockade of homing from blood back to tissue -- the last of which produces the characteristic treatment-related lymphocytosis. Resistance is an exponential decay of the proliferation and apoptosis effects with time (t1/2 = 761 days). Two observed outputs: leukocyte count (10^9 cells/L) and total SPD (cm^2). Ibrutinib enters only through the time-varying covariate AUC_IBRU; there are no drug-dosing events in this model. Sister model files from the same paper: modellib('Ibrahim_2023_ibrutinib_sbp'), modellib('Ibrahim_2023_ibrutinib_dbp'), modellib('Ibrahim_2023_ibrutinib_competing_risk'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ibrahim_2023_ibrutinib.html">Ibrutinib sbp (Ibrahim 2023)</a> </td> <td style="text-align:left;"> Ibrutinib-induced systolic blood pressure (sBP) turnover model in patients with chronic lymphocytic leukemia, from the phase Ib/II PCYC-1102 study. A zero-order production / first-order loss turnover model preceded by a single transit compartment reproduces the slow onset of ibrutinib-associated hypertension; ibrutinib stimulates the zero-order sBP production rate through an Emax function of the daily AUC(0-24). The transit rate constant and the turnover rate constant are equal, both set to (n + 1) / MTT with n = 1 transit compartment. Baseline age lowers the mean transit time (older patients reach the new steady state faster). Ibrutinib enters only through the time-varying covariate AUC_IBRU; there are no drug-dosing events in this model. Sister model files from the same paper: modellib('Ibrahim_2023_ibrutinib_dbp'), modellib('Ibrahim_2023_ibrutinib_leukocyte_spd'), modellib('Ibrahim_2023_ibrutinib_competing_risk'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hirt_2008_ibuprofen.html">Ibuprofen (Hirt 2008)</a> </td> <td style="text-align:left;"> One-compartment population PK model with linear elimination for intravenous ibuprofen-lysine (15-min infusion) administered for closure of patent ductus arteriosus in preterm neonates (Hirt 2008). Total-body clearance increases with postnatal age via a power function (CL = 9.49 mL/h x (PNA / 96.3 h)^1.49) anchored at the cohort median PNA of 96.3 h; the apparent volume of distribution is not influenced by postnatal age, gestational age, body weight, Apgar score, or baseline serum sodium / creatinine / albumin / urine output. Exponential inter-individual variability on CL and V; proportional residual error. The PK-PD link reported by the authors (AUC1D > 600 mg L^-1 h or AUC3D > 900 mg L^-1 h associated with >= 91% PDA closure) is illustrated in the validation vignette rather than carried in this model file. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lodise_2018_iclaprim.html">Iclaprim (Lodise 2018)</a> </td> <td style="text-align:left;"> Two-compartment IV-infusion population PK model for iclaprim, a bacterial dihydrofolate reductase inhibitor, in adult patients with complicated skin and skin-structure infections from the pooled ASSIST-1 and ASSIST-2 phase 3 trials (Lodise 2018). Structural typical-value equations are additive-linear (NONMEM theta-sum form rather than power form): central volume V1 carries a body-weight slope; clearance CL carries age + sex (male shift) + sampling-occasion (day 1-2 vs day 4 +/- 1) shifts; peripheral volume V2 has no covariates; inter-compartmental clearance Q carries a severe-cSSSI-infection shift. Block-correlated log-normal IIV on V1, CL, V2 was retained in the source paper but only diagonal CV% values are tabulated -- off-diagonal covariances are not reported and are implemented here as diagonal-only (documented in the vignette Assumptions and deviations section). Combined proportional + additive residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Brain_2008_ifosfamide.html">Ifosfamide (Brain 2008)</a> </td> <td style="text-align:left;"> Joint population PK / PD model for ifosfamide in adults with advanced solid tumours (Brain 2008, n=17, single-agent ifosfamide 9 g/m^2 per cycle by either 3 h x 3 daily or 72 h continuous infusion, n=1 randomised crossover, NONMEM VI FOCE INTERACTION). One-compartment ifosfamide PK with Kerbusch 2000-style autoinduction of clearance via a relative enzyme-pool state (drug inhibits enzyme degradation), three coupled apparent-volume metabolite states (4-hydroxy-ifosfamide, 3-dechloroethyl-ifosfamide, 2-dechloroethyl-ifosfamide), an indirect-response model for urinary beta-2-microglobulin (BMG, renal tubular toxicity) with linear stimulation of production by parent ifosfamide concentration, and a five-compartment Friberg-style myelosuppression chain for absolute neutrophil count (ANC) with linear inhibition of proliferation by parent ifosfamide concentration and (CIRC0 / circ)^gamma feedback. No covariates were retained in the final model (one outlier patient on carbamazepine was excluded prior to the final analysis). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Freyer_2000_AVI_regimen.html">Ifosfamide (Freyer 2000)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for ifosfamide in 24 small cell lung cancer patients on the AVI regimen (Freyer 2000), with day-1/day-2 clearance collapsed to the mean per operator instruction. Structural reduction from the paper's two-compartment model: only central CL and V were reported in Table 2, so the peripheral compartment is omitted; the paper's day-1 to day-2 autoinduction of clearance is also not represented in this extraction. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kerbusch_2000_ifosfamide.html">Ifosfamide (Kerbusch 2000)</a> </td> <td style="text-align:left;"> One-compartment population PK model for ifosfamide with autoinduction of CYP3A4-mediated metabolism implemented as ifosfamide-driven inhibition of enzyme-pool degradation (no lag time); estimated in 15 adults with soft tissue sarcoma receiving 9 or 12 g/m^2 as a 72-h continuous IV infusion. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/ParraGuillen_2013_tumor_immunotherapy.html">Il12 (ParraGuillen 2013)</a> </td> <td style="text-align:left;"> Preclinical (mouse, female C57BL/6 with subcutaneous MC38 tumor). Applicability re-fit of the Parra-Guillen 2013 semi-mechanistic K-PD tumor-growth-dynamics model to a single dose of murine IL-12 delivered by hydrodynamic plasmid injection. Structural equations are identical to the CyaA-E7 build (ParraGuillen_2013_cyaaE7); cell-line and immunotherapeutic-kinetics parameters (Ts0, lambda, k1, REG50) are re-estimated against the Medina-Echeverz 2014 dataset, while vaccine- efficacy and regulator-shape parameters (k3, k4, gamma) and the mixture probability P(1) = 0.844 are carried fixed from the CyaA-E7 fit. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yu_2024_ilaprazole.html">Ilaprazole (Yu 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order elimination for ilaprazole, a proton-pump inhibitor, after 0.75 h intravenous infusion in Chinese healthy subjects and patients with duodenal ulcer (Yu 2024). Pooled analysis of 1,560 plasma concentrations from 58 subjects across four phase I studies (healthy) and one phase IIa study (duodenal ulcer), fit in Phoenix NLME 8.3 by FOCE-ELS. Female sex lowers clearance (exp(-0.213)) and duodenal-ulcer disease status raises both clearance (exp(0.290)) and peripheral volume (exp(0.356)); peripheral volume additionally scales with body weight by a power of 1.545 around a 60.6 kg median. Typical values are for a healthy male at 60.6 kg. Inter-individual variability on the inter-compartmental clearance (CLp) was fixed in the final model because of 84% eta-shrinkage and no numeric variance was reported, so it is encoded as fixed(0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pei_2016_iloperidone.html">Iloperidone (Pei 2016)</a> </td> <td style="text-align:left;"> Population PK model for iloperidone and its two major plasma metabolites P-88 (M1, contributes to the therapeutic profile via D2 / 5-HT2A binding affinity comparable to the parent) and P-95 (M2, CYP2D6-mediated hydroxylation metabolite, pharmacologically less active) in 70 Chinese patients with schizophrenia or schizoaffective disorder receiving oral iloperidone 12-24 mg/day twice daily (Pei 2016). One-compartment first-order absorption (Ka FIXED at 2.26 1/h, estimated in a separate forward analysis of healthy-volunteer concentration-time data digitised from Pei 2016 ref [23] and fixed for the patient model to stabilize absorption identification under sparse sampling) and first-order parallel-pathway elimination of iloperidone via three rate constants: K20 (other elimination pathways), K23 (formation of M1), and K24 (formation of M2). Each metabolite then occupies its own one-compartment model with FIXED apparent volume V3 = V4 = 10 L (the fractions of iloperidone converted to each metabolite are not identifiable from the cohort because no co-administered tracer was available, so K23, K30 and K24, K40 are estimated against the FIXED apparent metabolite volume per Methods) and first-order elimination (K30 for M1, K40 for M2). Inter-occasion variability was retained on K20 in the published final model but is NOT carried as a separate eta in this nlmixr2lib extraction (see Errata in the validation vignette). Mass units (mg) rather than molar units are used per the source's convention because the molecular weights of iloperidone (427.3 g/mol), M1 (429.4 g/mol), and M2 (429.2 g/mol) are within 0.5%. CYP2D6*10 (rs1065852) polymorphism affects both metabolite formation rate constants: T/T homozygotes have K23 1.34-fold the C/C + C/T pooled reference; C/T heterozygotes and T/T homozygotes have K24 reduced to 0.693 and 0.492 of the C/C wild-type reference respectively. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chien_2022_imatinib.html">Imatinib (Chien 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral imatinib in healthy adult volunteers (Chien 2022); first-order absorption preceded by a Savic 2007-style analytical transit-compartment chain (mean transit time and number of transit compartments estimated), first-order elimination, and an OMEGA BLOCK between the IIV on CL and V1 motivated by their estimated correlation r > 0.9. No covariates were retained in the final model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jiang_2023_imatinib.html">Imatinib (Jiang 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption and first-order elimination for oral adjuvant imatinib in postoperative Chinese adults with gastrointestinal stromal tumors (Jiang 2023). The absorption rate constant is fixed at 1.22 1/h. Apparent oral clearance CL/F carries two covariate effects: a power-form red blood cell count effect ((RBC/3.7)^0.49) and a three-level ABCG2 rs2231142 genotype effect encoded with paired binary indicators (GG wild-type reference = 1, GT heterozygote = 0.879, TT homozygous variant = 0.976). Inter-individual variability is estimated on CL/F only; the apparent volume of distribution V/F is a typical value with no IIV because only trough concentrations were collected. Residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schindler_2017_imatinib.html">Imatinib (Schindler 2017)</a> </td> <td style="text-align:left;"> Joint tumor-dynamics PD model for imatinib-treated GIST liver metastases (Schindler 2017). Three size metrics (maximum transaxial diameter MTD in mm, software-segmented actual volume Vactual in mL, calculated ellipsoidal volume Vellipsoid in mL) follow a logistic tumor-growth model with a linear DOSE-dependent shrinkage term and a mono-exponential drug-effect washout (resistance development). Tumor density (Hounsfield units) follows an indirect-response model in which imatinib linearly stimulates the loss rate. Each subject can carry up to two liver lesions (lesion 1 has the larger baseline by convention); the binary covariate MIX_LARGE_BASE selects between a mixture subpopulation with larger lesion baselines (MIX_LARGE_BASE = 1, P = 0.348) and a smaller-baseline subpopulation (MIX_LARGE_BASE = 0). Drug exposure enters via the daily dose normalized to the median 400 mg, so DOSE is supplied as a per-record time-varying covariate (in mg/day). The OS and PFS time-to-event arms of the source publication are not encoded as ODE compartments here (see vignette Assumptions and deviations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/GonzalezSales_2024_imetelstat.html">Imetelstat (GonzalezSales 2024)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for imetelstat (GRN163L), a 13-mer N3'-P5' thio-phosphoramidate oligonucleotide telomerase inhibitor, fit to 4375 plasma concentrations from 424 adults with hematologic malignancies (lower-risk MDS, myelofibrosis, multiple myeloma, ET/PV, CLD) or solid tumors who received IV imetelstat 0.4-11.7 mg/kg weekly to every-4-weeks (Gonzalez-Sales 2024). Imetelstat is described by a two-compartment nonlinear disposition model with saturable binding/distribution to a peripheral binding (BIND) compartment (Snoeck 1999 / Peletier 2017 parameterisation): free drug binds reversibly to a target pool with capacity Bmax (Kon, Koff); bound drug is internalised to a deep peripheral tissue (Kint) and returns to central as free drug (Kback); free drug also undergoes linear elimination from central (CL). Theory-based allometric exponents for body weight (1 on Vc, 0.75 on CL, -0.25 on Kback) are fixed. Final covariates: sex, dose, time, and MF / MM malignancy on CL; sex and MM malignancy on Vc; MF malignancy and baseline spleen volume on Bmax. The time effect encodes a hyperbolic decay of baseline CL: CL(t) = CL * cl_hill_t50 / (t + cl_hill_t50). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Admiraal_2023_iminobiotin.html">Iminobiotin (Admiraal 2023)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for 2-iminobiotin (2-IB, a selective neuronal nitric oxide synthase inhibitor) in adults after out-of-hospital cardiac arrest, with a power-model eGFR-on-clearance covariate effect. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vos_2025_iminobiotin.html">Iminobiotin (Vos 2025)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model for the selective neuronal and inducible nitric oxide synthase inhibitor 2-iminobiotin (2-IB) in adults with acute large-vessel-occlusion (LVO) ischemic stroke treated with endovascular thrombectomy (Vos 2025). Central, peripheral, and inter-compartmental clearance are fixed at the upstream TIBOHCA-trial values (Vc = 10.2 L, Q = 15.0 L/h, Vp = 10.4 L); clearance is the only estimated structural parameter. Two typical clearance values are reported, one for patients who did not receive concomitant intravenous thrombolysis (alteplase) (9.29 L/h) and one for patients who did (15.3 L/h, a +65% increase). Baseline estimated glomerular filtration rate enters as an allometric power-form covariate on CL with exponent 0.817 and reference 90 mL/min/1.73 m^2. IIV is estimated on CL only (omega^2 = 0.046, ~22% CV); Vc had no IIV because the volume was fixed. Residual error is a single proportional component with SD 12.3%. The 24-hour continuous infusion uses an eGFR-stratified pump-speed table (Supplemental Table S2) so that all subjects regardless of renal function target the same average exposure (AUC_avg_4h ~ 365 ng*h/mL). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Couffignal_2014_imipenem.html">Imipenem (Couffignal 2014)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for imipenem in 51 critically ill adult ICU patients with suspected ventilator-associated pneumonia due to Gram-negative bacilli (Couffignal 2014). All patients received imipenem as a 0.5 h IV infusion every 8 hours; the protocol dose (500, 750 or 1000 mg) was chosen by Cockcroft-Gault creatinine clearance per the European Medicine Agency renal-adjustment table. Central clearance scales as a power of measured 4-hour creatinine clearance (reference 86.4 mL/min, the cohort median); central volume scales jointly with total bodyweight (reference 77 kg) and serum albumin (reference 18 g/L). The model was fitted in Monolix 4.1.2 using the SAEM algorithm with M3-equivalent BQL handling. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lamoth_2009_imipenem.html">Imipenem (Lamoth 2009)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for imipenem in adult febrile neutropenic patients with hematological malignancies (Lamoth 2009). Total clearance is the additive sum of a non-renal arm and a renal arm linear in Cockcroft-Gault GFR; the central volume of distribution scales linearly with total body weight referenced to 70 kg. A single log-normal inter-individual variability term is applied multiplicatively to the total clearance (TVCL = CL_nonren + CL_renal * GFR / 100), and residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yadav_2017_imipenem_aminoglycoside_pseudomonas.html">Imipenem amikacin PA001 (Yadav 2017)</a> </td> <td style="text-align:left;"> In vitro (static-concentration time-kill). Mechanism-based PK/PD (Bulitta life-cycle growth) model of bacterial killing and resistance for imipenem combined with amikacin against carbapenem- and amikacin-resistant clinical Pseudomonas aeruginosa isolate FADDI-PA001 (MIC_IPM = 32 mg/L, MIC_AMK = 32 mg/L). Three pre-existing bacterial subpopulations with signal-molecule growth inhibition and aminoglycoside-mediated outer-membrane permeabilisation (mechanistic synergy) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yadav_2017_imipenem_aminoglycoside_pseudomonas.html">Imipenem amikacin PA088 (Yadav 2017)</a> </td> <td style="text-align:left;"> In vitro (static-concentration time-kill). Mechanism-based PK/PD (Bulitta life-cycle growth) model of bacterial killing and resistance for imipenem combined with amikacin against carbapenem- and tobramycin-resistant clinical Pseudomonas aeruginosa isolate FADDI-PA088 (MIC_IPM = 16 mg/L, MIC_AMK = 4 mg/L). Three pre-existing bacterial subpopulations with signal-molecule growth inhibition and aminoglycoside-mediated outer-membrane permeabilisation (mechanistic synergy) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LopezArguello_2023_pbp_binding_pseudomonas.html">Imipenem qsp (LopezArguello 2023)</a> </td> <td style="text-align:left;"> QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for imipenem (carbapenem): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yadav_2017_imipenem_aminoglycoside_pseudomonas.html">Imipenem tobramycin PA001 (Yadav 2017)</a> </td> <td style="text-align:left;"> In vitro (static-concentration time-kill). Mechanism-based PK/PD (Bulitta life-cycle growth) model of bacterial killing and resistance for imipenem combined with tobramycin against carbapenem- and amikacin-resistant clinical Pseudomonas aeruginosa isolate FADDI-PA001 (MIC_IPM = 32 mg/L, MIC_TOB = 4 mg/L). Three pre-existing bacterial subpopulations with signal-molecule growth inhibition and aminoglycoside-mediated outer-membrane permeabilisation (mechanistic synergy) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yadav_2017_imipenem_aminoglycoside_pseudomonas.html">Imipenem tobramycin PA022 (Yadav 2017)</a> </td> <td style="text-align:left;"> In vitro (static-concentration time-kill). Mechanism-based PK/PD (Bulitta life-cycle growth) model of bacterial killing and resistance for imipenem combined with tobramycin against carbapenem-resistant and aminoglycoside-resistant clinical Pseudomonas aeruginosa isolate FADDI-PA022 (MIC_IPM = 16 mg/L, MIC_TOB = 8 mg/L). Three pre-existing bacterial subpopulations with signal-molecule growth inhibition and aminoglycoside-mediated outer-membrane permeabilisation (mechanistic synergy) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yadav_2017_imipenem_aminoglycoside_pseudomonas.html">Imipenem tobramycin PA088 (Yadav 2017)</a> </td> <td style="text-align:left;"> In vitro (static-concentration time-kill). Mechanism-based PK/PD (Bulitta life-cycle growth) model of bacterial killing and resistance for imipenem combined with tobramycin against carbapenem- and aminoglycoside-resistant clinical Pseudomonas aeruginosa isolate FADDI-PA088 (MIC_IPM = 16 mg/L, MIC_TOB = 32 mg/L). Three pre-existing bacterial subpopulations with signal-molecule growth inhibition and aminoglycoside-mediated outer-membrane permeabilisation (mechanistic synergy) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pardridge_2023_brain_plasma_protein_binding.html">Imipramine pbpk (Pardridge 2023)</a> </td> <td style="text-align:left;"> PBPK (brain, semi-mechanistic). Pardridge 2023 Pharmaceutical Research partly-flow / partly-compartmental model of imipramine delivery to brain that resolves plasma-protein binding as an explicit kinetic process inside the brain capillary. Seven brain states carry albumin-bound, alpha-1-acid-glycoprotein (AGP)-bound and free drug in the brain capillary, free AGP in the capillary, free drug in brain, and free / drug-occupied brain cytoplasmic binding protein. Drug crosses the blood-brain barrier by bi-directional first-order permeation and may be metabolised in brain. Imipramine shows a high degree of plasma-protein-mediated uptake from BOTH the AGP-bound and the albumin-bound pools (KG in vivo 90 uM vs KG in vitro 1.2 uM; KA in vivo >1,000 uM vs KA in vitro 42 uM), so free imipramine in brain is 18- to 31-fold higher than equilibrium dialysis of plasma predicts. Arterial input is a constant total plasma concentration (steady-state IV-infusion model, c_plasma_ss, Table II LT0 = 100 nM); the paper did not run a non-steady-state oral model for imipramine because detailed oral pharmacokinetic parameters were not available, so this model has no dosing compartment. All parameters are FIXED at the paper's literature-sourced values; the paper reports no IIV and no residual error because it is a deterministic simulation study rather than a fit to individual data. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cheng_2026_immunoglobulin.html">Immunoglobulin (Cheng 2026)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous immunoglobulin (IVIG) replacement therapy in pediatric primary-immunodeficiency and secondary-antibody-deficiency patients (Cheng 2026) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Byrne_2022_imr687.html">Imr687 (Byrne 2022)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for IMR-687 (a selective PDE9 inhibitor) in healthy subjects and patients with sickle cell disease (SCD), coupled with a repeated time-to-event (RTTE) exposure-response model for vaso-occlusive crisis (VOC) events. The PD hazard uses a saturable (Michaelis-Menten) drug effect on a constant baseline hazard. The model supports forward simulation of typical-value PK and cumulative-VOC hazard at any once-daily dose; the published covariate effects (body weight on CL/F and V/F; capsule formulation, capsule daily dose, and high-fat meal on absorption) are NOT encoded because the source conference poster reports the covariate point estimates without the functional forms or reference values needed to apply them. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Renard_2011_indacaterol.html">Indacaterol (Renard 2011)</a> </td> <td style="text-align:left;"> MBMA. Study-level Bayesian Emax meta-analysis of trough FEV1 dose-response to once-daily inhaled indacaterol in adults with moderate-to-severe chronic obstructive pulmonary disease (COPD), pooled from 11 placebo-controlled trials (7,476 patients; indacaterol doses 18.75 to 600 ug once daily). Algebraic Emax dose-response on placebo-corrected steady-state trough FEV1 (mL); the model is constrained to a null response at dose = 0 because the source data are contrasts to placebo. The original Bayesian analysis included between-study (delta_i) and between-arm-within-study (gamma_ij) random effects on Emax with unif(0, 0.25) priors; the paper reports only the posterior means of the structural Emax and ED50, not the random-effect posterior summaries, comparator mean effects (formoterol, salmeterol, tiotropium), or a per-observation residual sigma. The model file therefore encodes the indacaterol-only structural Emax curve with between-study and between-arm variances fixed to zero following the Vargo 2014 MBMA precedent. Suitable for simulating typical-trajectory study-arm-mean trough FEV1 improvement vs placebo at steady state (Week 2 to Month 6); not suitable for individual-subject simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Csajka_2004_indinavir.html">Indinavir (Csajka 2004)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model for oral indinavir 800 mg three-times-daily (alone) or 800 mg twice-daily with low-dose ritonavir in HIV-infected adults; concomitant ritonavir, sex, and body weight enter apparent oral clearance as multiplicative covariate effects (Csajka 2004). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kappelhoff_2005_indinavir.html">Indinavir (Kappelhoff 2005)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption popPK model for oral indinavir in HIV-1-infected adults, with multiplicative covariate effects of concomitant ritonavir (CL/F x 0.354) and concomitant NNRTI (efavirenz/nevirapine; CL/F x 1.41) on apparent clearance and of female sex on apparent bioavailability (F x 1.48). A 0.485 h absorption lag-time is applied only when ritonavir is co-administered (Kappelhoff 2005). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Berends_2019_infliximab.html">Infliximab (Berends 2019)</a> </td> <td style="text-align:left;"> Two-compartment TMDD-QSS population PK/target-dynamics model of infliximab and free TNF in adults with moderate-to-severe ulcerative colitis (Berends 2019) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Faelens_2021_infliximab.html">Infliximab (Faelens 2021)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model of infliximab in adults with moderate-to-severe ulcerative colitis (Faelens 2021 adapted model; baseline-covariate-only re-fit of Dreesen 2019) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fasanmade_2009_infliximab.html">Infliximab (Fasanmade 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of infliximab (anti-TNF-alpha) in patients with ulcerative colitis (Fasanmade 2009) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Frymoyer_2017_infliximab.html">Infliximab (Frymoyer 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of intravenous infliximab in children and adults with Crohn's disease (Frymoyer 2017; structural model and parameter values from the Fasanmade et al. REACH + ACCENT I analysis reproduced in Frymoyer 2017 Methods) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hanzel_2021_infliximab.html">Infliximab (Hanzel 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of subcutaneous and intravenous infliximab CT-P13 (biosimilar) in adults with Crohn's disease and ulcerative colitis (Hanzel 2021) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marcantonio_2022_efa.html">Infliximab (Marcantonio 2022)</a> </td> <td style="text-align:left;"> QSP. One-compartment monospecific anti-ligand mechanistic PKPD model of infliximab-TNF-alpha binding in adults with rheumatoid arthritis (Marcantonio 2022 Early Feasibility Assessment, Case Study 1). Structure identical to the Marcantonio 2022 adalimumab model; differs only in the drug-specific parameters (KD, half-life, molecular weight, route). Infliximab is administered IV (dose directly into Ab_00), binds soluble TNF-alpha monovalently (effective valency = 1 per Tran 2017 / Lim 2018), and blocks the reversible TNF:TNFR interaction. All parameters are FIXED from paper Table 2 (no fitting, no IIV, no residual error reported). Run in the Applied BioMath Assess Monospecific Anti-Ligand model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wu_2024_inotuzumab.html">Inotuzumab (Wu 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for inotuzumab ozogamicin in pediatric and adult patients with relapsed/refractory B-cell precursor acute lymphoblastic leukemia (BCP-ALL) and adult patients with B-cell non-Hodgkin's lymphoma (NHL); linear plus time-dependent (target-mediated) clearance with covariate effects on CL_SS, Vc, CL_TIME, and cl_exp_kdes (Wu 2024, ITCC-059 pediatric trial pooled with 11 adult studies). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chatelut_1999_interferon_alfa_2b.html">Interferon alfa 2b (Chatelut 1999)</a> </td> <td style="text-align:left;"> One-compartment population PK model for subcutaneous alpha-2b interferon (Intron A) in adults with chronic hepatitis C virus infection (Chatelut 1999), with sequential zero-order then first-order absorption (a fraction Fz of the bioavailable dose is absorbed at zero-order over duration tk0, the remaining (1 - Fz) is absorbed at first-order rate ka after tk0) and first-order elimination. Apparent oral clearance CL/F is reduced by 63.8% in chronic-haemodialysis patients relative to patients with normal renal function (RRT_HEMODIAL_STATUS = 1 vs 0); apparent central volume of distribution V/F scales linearly with body surface area (BSA). Proportional residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jeon_2013_interferonAlfa2a.html">InterferonAlfa2a (Jeon 2013)</a> </td> <td style="text-align:left;"> Joint PK-PD model for a sustained-release subcutaneous formulation of interferon alfa-2a (SR-IFN-alpha) and the serum neopterin response in healthy adult male volunteers (Jeon 2013). Pharmacokinetics: one-compartment with first-order elimination and a parallel mixture of zero- and first-order absorption. A fraction Fz = exp(RF)/(1 + exp(RF)) of the dose is absorbed by a zero-order process with duration D2 entering the central compartment directly; the remaining 1 - Fz is absorbed by a first-order process (rate Ka) from a depot compartment with lag time ALAG, accounting for the second concentration peak observed around 100 h post-injection. Pharmacodynamics: indirect-response (turnover) model for serum neopterin (baseline BASE = Kin/Kout) with a single transit compartment placed between the stimulus and the observed neopterin, delaying the neopterin response through mean transit time MTT. The drug stimulates the zero-order production rate of neopterin through a sigmoid Emax function E(C) = EMAX * C^GA / (EC50(t)^GA + C^GA), where EC50 is time-dependent and increases monotonically over time as EC50(t) = ECB * (1 + CA * (1 - exp(-CB * t))) -- an empirical saturation device that captures the observed loss of the neopterin dose-response between groups (9, 18, 27, 36 MIU) over the 0-264 h observation window. No covariate effects were retained in the final model. Doses are entered in MIU (10^6 IU); the published apparent clearance (CL/F = 12.2 L/h) and apparent volume of distribution (V/F = 691 L) match values previously reported for IFN-alpha in healthy subjects (Reference [19] of Jeon 2013). The model uses an explicit specific-activity conversion (1 MIU = 4 ug = 4e6 pg, from the WHO IFN-alpha-2a International Standard at 2.5e8 IU/mg) so the doses in user data can be entered in MIU and the simulated Cc is returned in pg/mL. The specific-activity conversion is not stated in the paper itself; it is documented in the validation vignette's Assumptions and deviations section. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Willmann_2021_fxi_aso.html">IonisFxirx (Willmann 2021)</a> </td> <td style="text-align:left;"> Population PK/PD model for the unconjugated FXI antisense oligonucleotide IONIS-FXIRX (BAY2306001) after subcutaneous administration in healthy volunteers (study ASO-CS1) and patients with end-stage renal disease requiring hemodialysis (study ASO-CS4), pooled with FXI-LICA data to fit the FXI-activity indirect-response arm (Willmann 2021). Pharmacokinetics: two-compartment model with parallel first-order (fraction F1, rate ka) and zero-order (fraction 1 - F1, duration D2) subcutaneous absorption and first-order elimination from the central compartment. Covariates on PK: end-stage renal disease reduces CL by 53% and peripheral volume V3 by 38% (proportional NONMEM form); body weight enters V2 as a power (V2 propto (WT/70)^0.967). Pharmacodynamics: an indirect-response model on FXI activity with sigmoid-Imax inhibition (Imax fixed to 1) of the zero-order synthesis rate Kin = baseline * kout, driven by an effect-site concentration linked to the plasma central compartment via first- order equilibration ke0. An additional multiplicative factor keoPAT modulates the effect-site concentration in ESRD patients (Ce_ESRD = keoPAT * Cc at steady state; keoPAT = 0.329 reduces the effective driving concentration to ~one third of the healthy-volunteer value). The PK arm was fitted separately per compound; the FXI-activity arm was fitted simultaneously to pooled ASO-CS1 + ASO-CS4 + LICA-CS1 data with shared kout, baseline, ke0, and Hill exponent (Table S4). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yoshida_2021_ipatasertib.html">Ipatasertib (Yoshida 2021)</a> </td> <td style="text-align:left;"> Joint parent + metabolite population pharmacokinetic model for oral ipatasertib (AKT kinase inhibitor under development for breast and prostate cancer) and its primary active metabolite M1 (G-037720) in 342 adult patients with cancer from five Phase 1 and 2 studies (Yoshida 2021). Each analyte is described by a 3-compartment disposition model with sequential zero-order then first-order absorption from its own depot. The two depots receive the oral parent dose simultaneously (the user supplies one event per depot with the same amount and time); both bioavailability anchors are fixed at F = 1 because absolute parent F and the fraction of parent metabolised to M1 are not separately identifiable from oral data alone, and the apparent M1 absorption parameters (kf, Dur, F) subsume formation, first-pass survival, and metabolite bioavailability per the source. Retained parent covariates: power effect of age on apparent CL/F, linear-additive effect of abiraterone coadministration on apparent CL/F, power effect of body weight on apparent F, and a +20.1% multiple-dose increment in apparent F representing CYP3A auto-inhibition by ipatasertib. Retained metabolite covariates: power effects of body weight on apparent V3 and Q3 of M1, a +33.1% multiple-dose increment in apparent F_M1, and an additional +61.5% abiraterone-by-multiple- dose effect on apparent F_M1. The paper fitted parent and metabolite in TWO SEPARATE NONMEM runs (Yoshida 2021 Discussion); this file collapses them into one rxode2 model with NO mechanistic fractional-conversion linkage, mirroring the paper's simulation strategy. See vignette Assumptions and deviations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Feng_2014_ipilimumab.html">Ipilimumab (Feng 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous ipilimumab (anti-CTLA-4 IgG1) in patients with unresectable stage III or IV melanoma (Feng 2014) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sanghavi_2020_ipilimumab.html">Ipilimumab (Sanghavi 2020)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous ipilimumab (anti-CTLA-4 IgG1) with time-varying clearance via a sigmoid cl_hill_max function in patients with advanced solid tumors receiving ipilimumab alone or in combination with nivolumab (Sanghavi 2020) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhu_2023_oxaliplatin_irinotecan_colorectal_cancer.html">Irinotecan (Zhu 2023)</a> </td> <td style="text-align:left;"> Minimal-PBPK. Joint irinotecan and SN-38 population PK model in adults with metastatic colorectal cancer: two-compartment plasma disposition for each of irinotecan and its active metabolite SN-38, plus an explicit tumour compartment. Irinotecan enters the tumour interstitial space by tumour plasma flow and permeates into tumour cells at a permeability-surface-area product; SN-38 enters a single lumped tumour-tissue space governed by a tumour/plasma partition coefficient. Zhu 2023 uses the tumour SN-38 concentration as the driver of an in vitro-derived tumour-killing model; that in vivo tumour-growth layer is NOT included here because its growth equation and limiting parameter are not reported anywhere in the paper - see the vignette Errata. The tumour-distribution parameters are transferred from the tumour-bearing-mouse fit, see modellib('Zhu_2023_irinotecan_mouse'); the in vitro drug-effect half is modellib('Zhu_2023_sn38_organoid'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhu_2023_oxaliplatin_irinotecan_colorectal_cancer.html">Irinotecan mouse (Zhu 2023)</a> </td> <td style="text-align:left;"> Preclinical (tumour-bearing mouse). QSP / minimal-PBPK. Joint irinotecan and SN-38 minimal-PBPK model: two-compartment plasma disposition for each of irinotecan and its active metabolite SN-38, plus an explicit tumour compartment. Irinotecan enters the tumour interstitial space by tumour plasma flow and permeates into tumour cells at a permeability-surface-area product; SN-38 enters a single lumped tumour-tissue space governed by a tumour/plasma partition coefficient. This is the preclinical fit that supplied the tumour-distribution parameters (KP_IRI, KP_SN, PS_IRI) for the human model, see modellib('Zhu_2023_irinotecan'); it carries no PD component because Zhu 2023 took the drug effect from patient-derived tumour organoids rather than from mouse tumour growth. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Brillac_2025_isatuximab.html">Isatuximab (Brillac 2025)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with linear elimination for isatuximab in pediatric and adult patients with relapsed/refractory acute leukemias (Brillac 2025) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fau_2020_isatuximab.html">Isatuximab (Fau 2020)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous isatuximab (anti-CD38 IgG1) in adults with relapsed/refractory multiple myeloma, with parallel time-varying linear and Michaelis-Menten eliminations from the central compartment (Fau 2020). The linear clearance follows a sigmoidal Emax decay from baseline to steady state; the magnitude of the decay differs by multiple-myeloma immunoglobulin type. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Desai_2016_isavuconazole.html">Isavuconazole (Desai 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for isavuconazole (administered as the prodrug isavuconazonium sulfate) in healthy adults and adults with mild (Child-Pugh A) or moderate (Child-Pugh B) hepatic impairment, following single 100 mg oral or 2-h intravenous doses (Desai 2016). Weibull absorption for the oral route; hepatic-impairment-group-specific typical CL and Q; linear BMI effect on peripheral volume. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Horita_2018_isoniazid.html">Isoniazid (Horita 2018)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model with first-order absorption and linear elimination for oral isoniazid in Ghanaian children with active tuberculosis (Horita 2018); NAT2 slow-vs-nonslow acetylator phenotype on apparent oral clearance with separate typical-value clearances and separate IIV omegas; allometric weight scaling on CL/F and Q/F (fixed 0.75) and V1/F and V2/F (fixed 1.0) normalised to the cohort median 14.3 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Seng_2015_isoniazid.html">Isoniazid (Seng 2015)</a> </td> <td style="text-align:left;"> Parent + two-metabolite population pharmacokinetic model for oral isoniazid (INH), acetylisoniazid (AcINH), and isonicotinic acid (INA) in 33 healthy Asian adults (Seng 2015; Singapore single-dose 300 mg oral INH study with crossover rifampin / efavirenz arms). Two-compartment INH disposition with first-order absorption, linked to a two-compartment AcINH disposition and a one-compartment INA disposition; metabolite formation splits via the fraction-of- clearance parameters F_AcINH (INH -> AcINH) and F_INA (AcINH -> INA), with the complementary (1 - F_AcINH) routing INH directly to INA. The NAT2-derived acetylator phenotype (rapid / intermediate / slow) selects between three typical-value INH clearances (65.2 / 32.6 / 6.52 L/h at 63 kg). Creatinine clearance enters as a power- law covariate on AcINH clearance with exponent 0.4 referenced to the cohort median 113 mL/min. All clearance and volume terms are allometrically scaled by total body weight (0.75 exponent on clearance, 1.0 on volume) with reference weight 63 kg. AcINH and INA central volumes are fixed at 17 L (apparent central volume of AcINH from Boxenbaum & Riegelman 1976) to keep the metabolite model identifiable in the absence of intravenous data. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wilkins_2011_isoniazid.html">Isoniazid (Wilkins 2011)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model for oral isoniazid in South African pulmonary tuberculosis patients (Wilkins 2011; 235 patients, 2352 plasma concentrations). First-order absorption with an absorption lag time, first-order elimination, and allometric scaling on all clearance and volume terms (WT exponent 0.75 on CL and Q, exponent 1 on Vc and Vp, reference weight 70 kg). A two-class mixture model on apparent clearance characterises the bimodal isoniazid elimination phenotype that arises from N-acetyltransferase-2 (NAT2) polymorphism: typical CL/F is 21.6 L/h in fast eliminators (13.2 % of subjects) and 9.70 L/h in slow eliminators (86.8 %). Two covariate effects were retained: female sex reduces Vc/F by 10.3 % and HIV-positive comorbidity reduces CL/F by 17.4 %. Inter-individual variability is reported on CL/F, Vc/F, Q/F, relative bioavailability F, and lag time; inter-occasion variability on ka (90.1 %) and F (8.4 %) is not propagated -- see the validation vignette Assumptions and deviations section for the single-occasion approximation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ramachandran_2023_tuberculosis_eptb_pbpk.html">Isoniazid pbpk (Ramachandran 2023)</a> </td> <td style="text-align:left;"> PBPK (whole-body, 18 ODEs, MATLAB ode15s). Oral isoniazid disposition at extrapulmonary tuberculosis (EPTB) sites in a 70-kg reference adult male (Ramachandran and Gadgil 2023, CPT Pharmacometrics Syst Pharmacol). Seventeen perfusion-limited well-stirred tissue compartments connected by blood and lymph flow, plus a gut-lumen state for enterohepatic recycling. The model's two novel states are a pleural-fluid compartment (a filtrate of the lung that drains via lymphatics) and a consolidated lymph-node compartment that collects afferent lymph from every organ except bone and spleen and returns it to venous blood - the two most common EPTB sites. Physiological volumes, blood flows, and lymph flows are fixed literature fractions of body weight, cardiac output (5200 mL/min), and afferent lymph flow (8 L/day) respectively (Appendix S1 Tables S2 and S3). Tissue:plasma partition coefficients were computed by the Rodgers and Rowland method from the drug physicochemistry in Table S4 and are tabulated in Table S5. Only the first-order oral absorption rate ka and total systemic clearance CL were estimated, by weighted least squares against reported plasma concentrations after a 300 mg oral dose (Appendix S1 Table S7). Slow and fast acetylators are modelled as two discrete cases selected by the NAT2_SLOW covariate, each with its own ka, CL, and fractional renal clearance fR; the partition coefficients are identical for the two groups. Clearance is split into a renal component fR * CL driven by arterial concentration and a hepatic component (1 - fR) * CL driven by the hepatic inlet concentration; the hepatic output enters the gut lumen, which for isoniazid is a terminal sink cleared only by faecal transit at kF = 0.252 /h (the paper assigns a non-zero gut reabsorption rate only to rifampicin). The model is a deterministic typical-value simulation: the paper reports no interindividual variability and no residual error model, so propSd is fixed at 0. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Abuhelwa_2015_itraconazole.html">Itraconazole (Abuhelwa 2015)</a> </td> <td style="text-align:left;"> Population PK model for oral itraconazole and its active metabolite hydroxy-itraconazole in healthy adults (Abuhelwa 2015). Two-compartment parent with 4-transit-compartment Savic-style absorption and a one-compartment hydroxy-itraconazole metabolite eliminated by mixed linear and Michaelis-Menten kinetics. Encodes the SUBA-itraconazole vs Sporanox formulation effect on relative bioavailability (with formulation-dependent scaling of the F variability) and the fed-vs-fasted effect on both relative bioavailability and the transit-absorption rate constant; the metabolic conversion ratio fm is assumed = 1 so all parent clearance becomes metabolite, and the metabolite CL/V are apparent values scaled by the unknown fm. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hennig_2006_itraconazole.html">Itraconazole (Hennig 2006)</a> </td> <td style="text-align:left;"> Population PK model for oral itraconazole and its active metabolite hydroxy-itraconazole in paediatric cystic-fibrosis and bone-marrow-transplant patients (Hennig 2006). One-compartment parent + one-compartment metabolite with first-order absorption, first-order metabolic conversion (fm fixed to 1), allometric weight scaling on parent CL/F (0.75) and Vd/F (1.0), and formulation-specific ka and relative bioavailability for capsule vs oral solution. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hennig_2007_itraconazole.html">Itraconazole (Hennig 2007)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral itraconazole and its one-compartment hydroxy-itraconazole metabolite in adult cystic fibrosis patients (Hennig 2007), with first-order absorption from a depot, formulation-specific absorption rate constants and bioavailability for capsule vs. oral solution selected by the binary FORM_CAPSULE covariate, and a single absorption lag-time shared across both formulations. The fraction of itraconazole metabolised to hydroxy-itraconazole is fixed to 1; metabolite parameters are reported as CL_m/(F*f_m) and V_m/(F*f_m). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Alshehri_2023_ivermectin.html">Ivermectin (Alshehri 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ivermectin (IVM) in 56 adults (32 Wuchereria bancrofti-infected, 24 uninfected) in Cote d'Ivoire given a single 200 ug/kg oral dose as part of ivermectin + diethylcarbamazine + albendazole (IDA) triple-drug therapy for lymphatic filariasis, taken after a high-fat breakfast (724 plasma samples). Absorption is a zero-order input into the depot compartment (TK0 = 3.74 h) preceded by a lag time (Tlag = 0.757 h), followed by first-order transfer into the central compartment (Ka = 0.718 1/h) and linear elimination. Body weight enters as allometric scaling with exponents fixed at 1 on Vc/F and Vp/F and 0.75 on CL/F and Q/F, centered on the study population's body weight. Sex is the only covariate retained by the stepwise selection: Vp/F is 53% lower in men than in women (424.3 L in women, 201.8 L in men). Lymphatic filariasis infection status was screened and had no effect on any PK parameter. Residual variability is combined additive (0.461 ng/mL) plus proportional (22.8%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gupta_2015_ixazomib.html">Ixazomib (Gupta 2015)</a> </td> <td style="text-align:left;"> Three-compartment population pharmacokinetic model for the oral proteasome inhibitor ixazomib (MLN9708) developed from pooled data of 226 adult patients with advanced multiple myeloma, lymphoma, or solid tumours across four phase I dose-escalation studies (Gupta 2015). Combined intravenous and oral data are described by a three-compartment model with first-order absorption and linear elimination; IV and oral data share the same disposition kinetics. Inter-individual variability is estimated on clearance, central volume V2, the second peripheral volume V4, absorption rate constant Ka, and bioavailability F; IIV on Q3, V3, and Q4 was fixed to zero. Body surface area on V4 (reference 1.90 m^2, exponent 2.3) is the only retained covariate; weight, age, gender, race, creatinine clearance, ALT, AST, albumin, and bilirubin had no clinically relevant effect on ixazomib pharmacokinetics. Residual error is additive on log-transformed concentration (NONMEM Y = LOG(F) + EPS(1)) which maps to a proportional error in linear concentration space. This analysis supported the switch from BSA-based to fixed (4 mg) dosing in subsequent ixazomib clinical studies. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gupta_2017_ixazomib.html">Ixazomib (Gupta 2017)</a> </td> <td style="text-align:left;"> Three-compartment population pharmacokinetic model for the oral proteasome inhibitor ixazomib (Ninlaro) in 755 adult patients with multiple myeloma, lymphoma, solid tumours, or light-chain amyloidosis pooled across ten phase I, I/II, and III trials including TOURMALINE-MM1 (Gupta 2017). First-order linear absorption with a 13 min lag time describes oral dosing; intravenous and oral data share the same disposition kinetics. Inter-individual variability is estimated on clearance, bioavailability F, and the second peripheral volume V4, with a strong (82%) correlation between log CL and log F. Body surface area on V4 (reference 1.87 m^2, exponent 2.06) is the only retained covariate; sex, age, race, mild/moderate renal impairment, mild hepatic impairment, smoking status, and CYP-modulatory concomitant medications had no clinically relevant effect on systemic exposure. Residual error is additive on log-transformed concentration with a time-after-dose-varying standard deviation declining exponentially from SD1 = 1.90 to SD0 = 0.46 with rate KSD = 0.84/h (Karlsson 1995 model 3). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jackson_2022_ixekizumab.html">Ixekizumab (Jackson 2022)</a> </td> <td style="text-align:left;"> Two-compartment linear population PK model for subcutaneous ixekizumab in paediatric patients with moderate-to-severe plaque psoriasis (IXORA-PEDS; Jackson 2022) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_guhongInjection.html">Kaempferol3OGlucoside rat (Chen 2024)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment intravenous pharmacokinetic model for kaempferol-3-O-glucoside (K-3-G), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the kaempferol-3-O-glucoside dose is the GHI volume dose times its content in GHI (8.7 ug/mL), i.e. 21.75 ug/kg, 43.5 ug/kg, 87 ug/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Direct-effect sigmoid-Emax models link the kaempferol-3-O-glucoside plasma concentration to the GHI-minus-model-group difference in cardiac troponin I (cTn I) (E = Emax * C^gamma / (ED50^gamma + C^gamma); Tables 14). Chen 2024 fitted a PK/PD model only for the analyte/biomarker/dose combinations whose PLSR coefficient was negative, so the effect of an unfitted combination is returned as zero rather than extrapolated. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_guhongInjection.html">Kaempferol3ORutinoside rat (Chen 2024)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment intravenous pharmacokinetic model for kaempferol-3-O-rutinoside (K-3-R), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the kaempferol-3-O-rutinoside dose is the GHI volume dose times its content in GHI (65.7 ug/mL), i.e. 164.25 ug/kg, 328.5 ug/kg, 657 ug/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Direct-effect sigmoid-Emax models link the kaempferol-3-O-rutinoside plasma concentration to the GHI-minus-model-group difference in creatine kinase-MB (CK-MB) (E = Emax * C^gamma / (ED50^gamma + C^gamma); Tables 12). Chen 2024 fitted a PK/PD model only for the analyte/biomarker/dose combinations whose PLSR coefficient was negative, so the effect of an unfitted combination is returned as zero rather than extrapolated. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Park_2001_ketoprofen.html">Ketoprofen (Park 2001)</a> </td> <td style="text-align:left;"> One-compartment oral PK plus Holford-Sheiner effect-compartment for synovial fluid disposition of ketoprofen in adults with arthritis at steady state on 100 mg oral twice-daily dosing (Park 2001 Tables 2-3, Eq. 1; effect-compartment elimination rate keo = 0.16 1/h, peak synovial:plasma ratio 0.77 with 3.1 h time lag). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valitalo_2017_ketorolac.html">Ketorolac (Valitalo 2017)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for IV ketorolac in adults, jointly fit to R-ketorolac and S-ketorolac plasma concentrations after racemic IV dosing in women at delivery, postpartum women, nonpregnant women, and men (Valitalo 2017 BJCP). Body-weight allometric scaling on clearance and volumes (reference 71 kg) plus proportional pregnancy-at-delivery and male-sex effects on clearance (and pregnancy-at-delivery on volumes), shared between enantiomers. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Granda_2024_kidney_pbpk.html">Kynurenicacid pbpk (Granda 2024)</a> </td> <td style="text-align:left;"> PBPK (mechanistic kidney, 35 states). Individualised prediction of kynurenic acid renal clearance from per-subject biomarker measurements of kidney blood flow and OAT1/3-mediated tubular secretory capacity, on top of tracer-measured GFR (Granda 2024, Clin Transl Sci). The kidney is resolved into 11 longitudinal subsegments (proximal tubule S1-S3, descending and ascending loop of Henle, distal tubule, and five collecting-duct subsegments), each with a tubular lumen, a tubular epithelial cell and a peritubular blood subspace (33 states), plus a systemic blood compartment and a bladder. Mechanisms: unbound glomerular filtration, OAT1/3-mediated active secretion in the proximal tubule only, pH-dependent bidirectional passive diffusion along the whole nephron, and CKD tubular-flow adaptation. The structural equations, segment volumes, surface areas, tubular pH profile and adaptive flow factors are the Huang & Isoherranen 2018 / 2020 framework; Granda 2024 replaces the framework population defaults for kidney blood flow and secretory clearance with per-subject measured / fitted values. The biomarker leg of the workflow: for each subject the unbound intrinsic secretory clearance is fitted so that the model reproduces that subject's MEASURED kynurenic acid renal clearance (Table 3). The fitted value is then carried into the tenofovir and oseltamivir carboxylate models via a drug-specific scalar. Deterministic typical-value model: the paper reports no IIV and no residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lindauer_2017_lacosamide.html">Lacosamide dropout (Lindauer 2017)</a> </td> <td style="text-align:left;"> Repeated time-to-event dropout model (dropout not because of a lack of efficacy) for adult patients newly diagnosed with focal or generalized tonic-clonic epilepsy on lacosamide (LCM) or carbamazepine controlled-release (CBZ-CR) monotherapy, from Lindauer 2017 (SP0993 trial; NCT01243177). The base hazard is a smoothed 4-breakpoint step function h0(t) = k1 + (k2-k1)*S1 + (k3-k2)*S2 + (k4-k3)*S3 + (k5-k4)*S4 where S_i is a logistic sigmoid centred at breakpoint BP_i with steepness GAM=50 (paper Frobel et al. parameterisation). Covariate effects are multiplicative on the hazard: hazard_drop = h0(t) * exp(e_sexf_drop*SEXF + e_conmed_lcm_drop*CONMED_LCM). Sister time-to-seizure model for the same trial: modellib('Lindauer_2017_lacosamide_seizure'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lindauer_2017_lacosamide.html">Lacosamide seizure (Lindauer 2017)</a> </td> <td style="text-align:left;"> Repeated time-to-seizure model for adult patients newly diagnosed with focal or generalized tonic-clonic epilepsy on lacosamide (LCM) or carbamazepine controlled-release (CBZ-CR) monotherapy, from Lindauer 2017 (SP0993 trial; NCT01243177). Two Weibull sub-model hazards are exposed as separate outputs: hazard_1st for the time to the first seizure and hazard_2nd for the time to the second and subsequent seizures. Base Weibull hazard h(t) = lam * p * (lam * t)^(p - 1) with distinct scale (lam) and shape (p) for each sub-model. Baseline disease severity (NSP3M: number of seizures in the previous 3 months, categorised <2 / 2-6 reference / 7-50 / >50), daily AUC of the assigned drug (LCM or CBZ-CR; centred at 104 mg*h/L for LCM and 132 mg*h/L for CBZ), and age (LCM-only, on the first-seizure hazard) enter as log-linear covariate effects on lam. IIV on ln(lam_2nd) with SD 2.03 captures the substantial between-subject variability in the subsequent-seizure hazard. Sister dropout model for the same trial: modellib('Lindauer_2017_lacosamide_dropout'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Archary_2019_lamivudine.html">Lamivudine (Archary 2019)</a> </td> <td style="text-align:left;"> One-compartment population PK model for lamivudine in severely malnourished HIV-infected children (Archary 2019); CL/F matures with age via a sigmoid Emax function, Vc/F decreases linearly with serum triglyceride, and ka steps up between day 1 and day 14 of antiretroviral treatment </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bouazza_2010_lamivudine.html">Lamivudine (Bouazza 2010)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for once-daily oral lamivudine in HIV-infected West African children (Bouazza 2010); allometric weight scaling on CL/F, Q/F, Vc/F, and Vp/F with reference body weight 16.8 kg, and absorption rate constant Ka structurally fixed to the disposition distribution-phase eigenvalue (Ka = alpha = 0.71 1/h) from the literature </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bouazza_2011_lamivudine.html">Lamivudine (Bouazza 2011)</a> </td> <td style="text-align:left;"> Two-compartment oral popPK model for lamivudine in HIV-infected children from neonates to adolescents (Bouazza 2011) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chandasana_2024_dolutegravir_lamivudine.html">Lamivudine (Chandasana 2024)</a> </td> <td style="text-align:left;"> Two-compartment oral population PK model with first-order absorption and elimination for lamivudine 300 mg once daily as the dolutegravir/lamivudine fixed-dose combination in virologically suppressed adults living with HIV-1 (TANGO; Chandasana 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chandasana_2024_abacavir_dolutegravir_lamivudine_pediatric.html">Lamivudine (Chandasana 2024b)</a> </td> <td style="text-align:left;"> One-compartment oral population PK model with first-order absorption, an absorption lag time and formulation-dependent absolute bioavailability for lamivudine in children with HIV-1 weighing 6 to 40 kg, applied without re-estimation to the ABC/DTG/3TC fixed-dose combination (dispersible tablet and tablet) in IMPAACT 2019 (Chandasana 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Moore_1999_lamivudine.html">Lamivudine (Moore 1999)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral lamivudine in HIV-1-infected adults pooled from the NUCA3001 and NUCA3002 phase III trials (Moore 1999); CL/F scales with a Cockcroft-Gault-style renal function index ((140 - AGE)/(CREAT * 100), * 0.85 if female) raised to an estimated power and with linear body weight, V/F and ka carry no covariates </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/He_2012_lamotrigine.html">Lamotrigine (He 2012)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral lamotrigine in Chinese paediatric patients with epilepsy aged 0.5-17 years (He 2012). First-order absorption with Ka fixed at 1.0 1/h and bioavailability fixed at 1 (lamotrigine steady-state trough therapeutic-drug-monitoring data, which do not identify Ka or F), and first-order elimination from a single central compartment. Apparent oral clearance is scaled by an estimated power of total body weight (exponent 0.635) and modified exponentially by concomitant antiepileptic comedication: valproate (CONMED_VPA) reduces CL, while the enzyme-inducers carbamazepine (CONMED_CBZ) and phenobarbital (CONMED_PB) increase CL. Apparent central volume is fixed at 16.7 L at the 27.87 kg reference weight, scaled linearly with total body weight (allometric exponent fixed at 1.0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hussein_1997_lamotrigine.html">Lamotrigine (Hussein 1997)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral lamotrigine monotherapy in adults and adolescents newly diagnosed with epilepsy; apparent oral clearance carries a first-order auto-induction term that decays toward a steady-state value over treatment duration and a multiplicative race effect for Asians vs Caucasians; apparent volume of distribution and absorption rate constant are time-invariant with no covariate effects retained in the final model (Hussein 1997). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Milosheska_2016_lamotrigine.html">Lamotrigine (Milosheska 2016)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption parent + one-compartment metabolite population pharmacokinetic model for oral lamotrigine (LTG) and its N-2-glucuronide (LTG-glu) in 100 adult Slovenian epilepsy patients on stable mono- or adjunctive therapy (Milosheska 2016). Complete conversion of parent to metabolite is assumed (LTG-N-5-glucuronide is a minor route, < 10% urinary excretion of unchanged drug per the paper's Discussion citing Ref [5]). Parent apparent oral clearance (CL/F) carries a power effect of total body weight and additive-in-fraction effects of smoking, concomitant enzyme-inducing antiepileptic drugs (carbamazepine, phenobarbital, or phenytoin, pooled as CONMED_EIAED), concomitant UGT2B7 inhibitors (valproic acid or sertraline, pooled as CONMED_UGT_INH), Cockcroft-Gault estimated creatinine clearance (deviation from 110 mL/min), and two UGT2B7 SNP genotype categoricals (-161C>T rs7668258 and 372A>G rs28365063). Parent apparent volume (V/F) carries a linear deviation-from-reference weight effect. Metabolite apparent clearance (CL_LTG-glu / F_metab) carries a power weight and linear Cockcroft-Gault CLcr effect; metabolite apparent volume (V_LTG-glu / F_metab) is estimated as a typical value only (no IIV supported by the sparse metabolite data). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Le_2015_lampalizumab.html">Lampalizumab (Le 2015)</a> </td> <td style="text-align:left;"> Combined ocular-serum target-mediated drug-disposition (TMDD) model with quasi-steady-state binding approximation for intravitreally administered lampalizumab (anti-complement factor D Fab) and total complement factor D (CFD) in adults with geographic atrophy secondary to age-related macular degeneration. Vitreous humor is the dosing compartment (depot) and the site of drug-target binding; aqueous humor lampalizumab and aqueous humor total CFD observations are derived from vitreous via constant partition coefficients; serum lampalizumab is the central elimination compartment with linear first-order clearance. Age and female sex modify ocular and systemic elimination rates respectively (Le 2015 Table 1, Eq. 1-7). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Le_2015_lampalizumab_cyno.html">Lampalizumab cyno (Le 2015)</a> </td> <td style="text-align:left;"> Preclinical (cynomolgus monkey). Semi-mechanistic and integrated ocular-systemic target-mediated drug-disposition (TMDD) PK/PD model of lampalizumab (anti-complement factor D antigen-binding fragment) and complement factor D (CFD) after intravenous or bilateral intravitreal (ITV) administration. Uses the quasi-steady-state (QSS) approximation of Gibiansky 2008. States: vitreous total lampalizumab amount (depot), serum total lampalizumab amount (central), a lampalizumab peripheral distribution compartment (peripheral1), and total CFD concentrations in vitreous (total_target), serum (total_target_central) and a target peripheral distribution compartment (total_target_peripheral1). Total lampalizumab concentrations in aqueous humor and retina are algebraic partitions of vitreous total drug. The model represents one eye; vitreous-to-serum flows in the serum ODEs carry a factor of 2 to reflect bilateral ITV dosing. Naive-pool fit (no IIV); residual error was reported qualitatively (proportional) but no numeric variance is given in the paper, so residual SDs are encoded as fixed(0) so deterministic simulation reproduces the paper exactly (see vignette Errata). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kunisawa_2015_landiolol.html">Landiolol (Kunisawa 2015)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model with lag time for landiolol hydrochloride (an ultra-short-acting cardioselective beta1-adrenergic receptor blocker) in adult patients with peripheral arterial disease undergoing peripheral arterial surgery, with linear body-weight normalization on CL, Vc, Q and Vp (Kunisawa 2015) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Buil-Bruna_2015_lanreotide.html">Lanreotide (Buil-Bruna 2015)</a> </td> <td style="text-align:left;"> One-compartment population PK model with parallel first- and zero-order subcutaneous absorption for lanreotide Autogel/Depot in patients with gastroenteropancreatic neuroendocrine tumors (Buil-Bruna 2015). A linear effect of body weight on apparent clearance and a small categorical effect of sex on the first-order absorbed fraction are retained; absolute bioavailability F is not identifiable and is structurally anchored at 1, so apparent CL/F and Vd/F are reported. Concentrations are predicted in ng/mL; residual error is additive on the log-transformed observations (LTBS), mapped to proportional in linear space. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sakurai_2007_lansoprazole.html">Lansoprazole (Sakurai 2007)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenously administered lansoprazole in 56 healthy Japanese adult males (Sakurai 2007). Volumes (V1, V2) and clearances (CL, Q) scale linearly with body weight via per-kg reference values; systemic clearance is stratified by CYP2C19 metabolizer phenotype using two binary indicators (homoEM reference; heteroEM and PM groups carry multiplicative factors of 0.612 and 0.212 respectively). Inter-individual variability is log-normal on V1, CL, V2 (no IIV on Q); residual error is combined proportional plus additive. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhu_2017_lebrikizumab.html">Lebrikizumab (Zhu 2017)</a> </td> <td style="text-align:left;"> Lebrikizumab population PK model (Zhu 2017): two-compartment model with first-order absorption after SC dosing in adults with moderate-to-severe asthma. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hopkins_2015_leflunomide.html">Leflunomide (Hopkins 2015)</a> </td> <td style="text-align:left;"> Parametric time-to-event (TTE) model for cessation of daily oral leflunomide due to toxicity in adult rheumatoid-arthritis patients. Two piecewise-exponential hazards describe (i) the toxicity-cessation hazard h0(t) with four intervals (< 50, 50-112, 112-204, > 204 days after leflunomide initiation) and (ii) a parallel random-censoring hazard h0ran(t) with five intervals (< 147, 147-210, 210-314, 314-350, > 350 days). CYP1A2 rs762551 C-allele carrier status is the single retained covariate: it multiplies the cessation hazard by (1 + 1.29 * SNP_CYP1A2_RS762551_C_CARRIER), a 2.29-fold increase for C-carriers vs AA homozygotes. Teriflunomide steady-state trough concentrations (total and free), leflunomide dose, CYP2C19 phenotype, and 20 other covariates were screened but not retained in the final model. Outputs sur / hazard / cumhaz (cessation) and sur_cens / hazard_cens / cumhaz_cens (random censoring) are exposed for forward simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gupta_2016_lenvatinib.html">Lenvatinib (Gupta 2016)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for lenvatinib in healthy subjects and patients with cancer (Gupta 2016). Simultaneous first-order plus zero-order oral absorption into the central compartment, linear elimination, and covariate effects of body weight (allometric on CL/F and Q/F with exponent 0.75 and linear on V/F), CYP3A4 inducers (+30 percent on CL/F), CYP3A4 inhibitors (-7.8 percent on CL/F), serum albumin < 30 g/L (-16.3 percent on CL/F), alkaline phosphatase > ULN (-11.7 percent on CL/F), healthy-subject cohort (+15 percent on CL/F vs cancer patients), and capsule vs tablet formulation (relative bioavailability 0.896). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Majid_2024_lenvatinib.html">Lenvatinib (Majid 2024)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for lenvatinib pooled across healthy subjects and patients with differentiated thyroid cancer (DTC), renal cell carcinoma (RCC), hepatocellular carcinoma (HCC), and other solid tumors (Majid 2024). Zero-order input into the absorption depot over a duration D1 followed by first-order absorption (ka) into the central compartment, linear elimination, and covariate effects of body weight (allometric exponent 0.75 on CL/F and Q/F, 1.0 on the volumes), concomitant CYP3A4 inhibitors (-10.4 percent on CL/F), serum albumin < 30 g/L (-10.0 percent), alkaline phosphatase > ULN (-9.0 percent), healthy-subject cohort (+19 percent), and tumor type (DTC -4.9 percent, HCC -13.8 percent, RCC -14.9 percent versus the other-solid-tumor reference), plus a capsule-versus-tablet relative bioavailability of 0.882. This is the updated successor to the Gupta 2016 lenvatinib popPK model; see modellib('Gupta_2016_lenvatinib'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Majid_2024_lenvatinib.html">Lenvatinib biomarkers (Majid 2024)</a> </td> <td style="text-align:left;"> Simultaneous four-biomarker population PK/PD indirect-response model for serum VEGF, Tie-2, Ang-2 and FGF-23 during lenvatinib treatment of radioiodine-refractory differentiated thyroid cancer (Majid 2024). Each biomarker is a turnover pool with its own baseline, mean residence time and Hill coefficient; a single common Emax, a single common EC50 and a single common linear disease-progression slope are shared across all four. Lenvatinib exposure enters as the steady-state daily AUC covariate AUC_LEN and acts through a sigmoid Emax function that inhibits Kout for VEGF and FGF-23 (levels rise) and inhibits Kin for Tie-2 and Ang-2 (levels fall). There is no PK ODE: AUC_LEN is supplied as a data column, as in the source sequential fit. Companion models: modellib('Majid_2024_lenvatinib') for the PK layer and modellib('Majid_2024_lenvatinib_tumor') for the tumor-growth-inhibition layer. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Majid_2024_lenvatinib.html">Lenvatinib tumor (Majid 2024)</a> </td> <td style="text-align:left;"> Integrated tumor-growth-inhibition and serum-biomarker PK/PD model for lenvatinib in radioiodine-refractory differentiated thyroid cancer (Majid 2024). First-order tumor growth (rate KG) is opposed by three shrinkage terms acting on the RECIST 1.1 sum of longest diameters: an Emax function of lenvatinib average daily AUC whose potency decays exponentially with an acquired-resistance rate lambda, and two linear terms proportional to the model-predicted relative change from baseline of Tie-2 and Ang-2. The Tie-2 and Ang-2 time courses are re-integrated inside this model from the companion four-biomarker turnover model with their parameters fixed. Lenvatinib exposure enters as the AUC_LEN covariate; there is no PK ODE. Time is in weeks. Companion models: modellib('Majid_2024_lenvatinib') and modellib('Majid_2024_lenvatinib_biomarkers'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Majid_2024_lenvatinib.html">Lenvatinib tumor asdeposited (Majid 2024)</a> </td> <td style="text-align:left;"> AS-DEPOSITED variant of the Majid 2024 integrated tumor-growth-inhibition and serum-biomarker PK/PD model for lenvatinib in radioiodine-refractory differentiated thyroid cancer: a literal transcription of the deposited NONMEM control stream (Supplementary Text S3), including the copy-paste slip in its Tie-2 drug-effect line. Text S3 defines Tie-2-specific effect variables (IMAX1T, HILLT, IC50T) and then never uses them; the EFFT line picks up the VEGF-indexed IMAX1, HILL and IC50 instead, so the Tie-2 arm is driven by the VEGF Hill coefficient of 1 and by an EC50 left on the ng*h/mL scale while the stream's AUC has been rescaled to ug*h/mL. The resulting 1000-fold mismatch makes the Tie-2 pool essentially drug-insensitive, so its contribution to tumor dynamics is a pure disease-progression term. This is the code that produced the published Table 3 estimates and Figure 4; modellib('Majid_2024_lenvatinib_tumor') is the companion variant that instead follows the paper's printed Equations 2 and 4. Lenvatinib exposure enters as the AUC_LEN covariate; there is no PK ODE. Time is in weeks. Companion models: modellib('Majid_2024_lenvatinib') and modellib('Majid_2024_lenvatinib_biomarkers'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jaworowicz_2006_levalbuterol.html">Levalbuterol (Jaworowicz 2006)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for (R)-albuterol following inhaled levalbuterol (90 ug) or racemic albuterol (180 ug) via a hydrofluoroalkane metered-dose inhaler in pediatric (4-11 years) and adult (12-81 years) asthma patients. First-order absorption, linear elimination, body-weight effects on apparent clearance (linear-additive) and central volume (power), and a pediatric-vs-adult split on absorption rate. The reference parameters are the Adult / Study 051-353 / single-dose levalbuterol-visit values (bioavailability anchor F1 = 1). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/KreeftmeijerVegter_2015_levamisole.html">Levamisole (KreeftmeijerVegter 2015)</a> </td> <td style="text-align:left;"> One-compartment oral PK model for levamisole in 38 children with steroid-sensitive nephrotic syndrome (Kreeftmeijer-Vegter 2015, EudraCT 2005-005745-18). First-order absorption, first-order elimination, allometric scaling of CL/F (exponent 0.75) and V/F (exponent 1) to 70 kg, and a linear proportional age effect on CL/F centred on the population median age of 6.28 years (-10.1% per additional year). The typical ka (1.2 1/h) was fixed in the final model with IIV retained. IIV on V/F was modelled as perfectly correlated with IIV on CL/F (single eta scaled to V/F), encoded here as a full omega block with covariance equal to sqrt(var_CL * var_V). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shin_2017_levetiracetam.html">Levetiracetam (Shin 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for levetiracetam in Korean neonates with seizures (Shin 2017). Structural parameters (V, CL) reported on a per-kg-body-weight basis (linear scaling by body weight). Drug absorption was not modelled because trough-style sampling between 6 and 23 hours after dose did not capture the absorption phase; intravenous and oral doses are therefore modelled as bolus inputs directly into the central compartment with bioavailability fixed at 1. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2012_levetiracetam.html">Levetiracetam (Wang 2012)</a> </td> <td style="text-align:left;"> One-compartment population PK model for levetiracetam (LEV) in Chinese pediatric epilepsy patients (Wang 2012). First-order oral absorption and linear elimination (NONMEM ADVAN2 TRANS2). Body weight is the only retained covariate; it enters CL/F as a power-style allometric term with reference weight 25 kg (cohort median). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hussein_2005_levocetirizine.html">Levocetirizine (Hussein 2005)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption and first-order elimination for orally administered levocetirizine in atopic young children (12-48 months, 8-20 kg) receiving 0.125 mg/kg twice-daily levocetirizine (administered as 0.25 mg/kg twice-daily racemic cetirizine) for 18 months in the ETAC study (Hussein 2005). CL/F and V/F are linear functions of body weight (CL/F = 0.244 + 0.0442 * WT L/h; V/F = 0.639 * WT L). The absorption rate constant ka is parameterised as ka = theta_ka + CL/V to guard against flip-flop kinetics, with theta_ka = 1.140 1/h and CL/V contributing on average less than 5% to ka. Residual variability is additive with two concentration-dependent magnitudes: 53.5 ng/mL for Cc <= 400 ng/mL and 316 ng/mL for Cc > 400 ng/mL (the 400 ng/mL threshold was selected by sensitivity analysis and has no clinical or therapeutic implication). Bioavailability is anchored at F = 1 here; the paper additionally estimated F_noncomp = 0.281 applied to 12% of records flagged as suspected noncompliance and recorded in the vignette Assumptions and deviations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Denti_2018_levofloxacin.html">Levofloxacin (Denti 2018)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral levofloxacin in South African children with multidrug-resistant tuberculosis (MDR-TB) disease or exposure (Denti 2018; n = 109; median age 2.1 yr; median weight 12.4 kg). First-order absorption with an absorption lag time, allometric scaling fixed to 0.75 on CL / Q and 1 on Vc / Vp with the population-median 12 kg as the reference weight, and a Hill-type maturation function on CL driven by postmenstrual age (PMAGE_50 = 10.6 mo, gamma = 3.39; PMAGE = postnatal age + 9 mo assuming term gestation). Covariate effects: HIV-positive children have 15.9% lower CL; nasogastric-tube (NGT) administration shortens the absorption lag time by 85.6% relative to the oral reference. F is fixed at 1; the additive residual error is fixed at 20% of the LLOQ (0.0160 mg/L). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hurtado_2014_levofloxacin_rat.html">Levofloxacin rat (Hurtado 2014)</a> </td> <td style="text-align:left;"> Preclinical (rat). Three-compartment population PK model for unbound levofloxacin in plasma and prostate interstitial fluid in male Wistar rats after a single 7 mg/kg IV bolus, with simultaneous fitting of total plasma concentrations (central, Vc) and free prostate ISF concentrations measured by microdialysis (effect compartment, apparent volume V3* = V_prostate / fu_prostate). Prostate kinetics are asymmetric: uptake from central is first-order (k13), efflux back to central combines a linear first-order term (k31) with a saturable Michaelis-Menten efflux (Vmax, kM) consistent with active transporter involvement. The standard central <-> peripheral1 disposition uses macro-constants CL, Q, Vc, Vp (Hurtado 2014). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Adeojo_2024_levonorgestrel.html">Levonorgestrel (Adeojo 2024)</a> </td> <td style="text-align:left;"> MBMA. Two-compartment population PK model for levonorgestrel fit to pooled MEAN plasma concentration-time profiles digitised from four published trials (Adeojo 2024). The four trials span single 0.25 mg oral, single 0.25 mg intravenous, single 0.75 mg oral, two-dose 0.75 mg oral, and 150 mg subdermal implant administration in women, with two of the trials contributing arms co-administered with efavirenz 600 mg orally once daily. Absorption is first-order and applies to oral dosing only (ka fixed at 4.15 /h); the intravenous and subdermal-implant routes enter the central compartment directly, so the estimated bioavailability F is an oral bioavailability. Concomitant efavirenz enters as a single binary covariate with two effects: clearance rises from 5.86 to 10.1 L/h (+72.4%) through CYP3A4 induction and oral bioavailability falls from 0.837 to 0.533 (-36.3%) through gut-wall CYP3A4 induction. The model carries BETWEEN-STUDY variability on oral bioavailability only (21.2% CV) and no between-subject variability, because it was fit to study-level mean profiles rather than individual concentrations; it therefore simulates study-level mean concentration-time curves, NOT individual patient concentrations. This is the reference mixed-effects model of the source paper, used there to anchor a Simcyp physiologically-based model by retrograde determination; the Simcyp PBPK model itself is not reproducible from the published sources and is not packaged here (see the validation vignette Assumptions and deviations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bienczak_2025_ligelizumab.html">Ligelizumab (Bienczak 2025)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ligelizumab in adolescent and adult patients with chronic spontaneous urticaria and healthy adult volunteers (Bienczak 2025) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Retlich_2015_linagliptin.html">Linagliptin (Retlich 2015)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with concentration-dependent (saturable) binding of linagliptin to dipeptidyl peptidase-4 in both central and peripheral compartments, coupled with a population sigmoid Emax PK/PD model relating total linagliptin plasma concentration to plasma DPP-4 activity, in adults with type 2 diabetes mellitus (Retlich 2015 Tables 4 and 5). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tadayasu_2014_linagliptin.html">Linagliptin (Tadayasu 2014)</a> </td> <td style="text-align:left;"> Two-compartment target-mediated drug disposition population PK model for linagliptin with quasi-equilibrium concentration-dependent binding to DPP-4 in both the central and peripheral compartments, coupled with an occupancy-based DPP-4-inhibition pharmacodynamic model (DPP-4 inhibition = Emax * Cbound/BMAX in the central compartment), in Japanese patients with type 2 diabetes mellitus (Tadayasu 2014 Table 3). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Abdelgawad_2024_linezolid.html">Linezolid (Abdelgawad 2024)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral linezolid in plasma and lumbar cerebrospinal fluid (CSF) in adults with HIV-associated tuberculous meningitis receiving high-dose rifampicin (35 mg/kg) co-treatment (LASER-TBM PK substudy, Abdelgawad 2024). Absorption is a Savic transit chain (mean transit time 0.211 h, estimated chain length 5.68) feeding a first-order depot (ka 1.21 1/h); elimination from the central compartment is saturable Michaelis-Menten with maximal clearance 7.25 L/h and Michaelis-Menten constant 27.2 mg/L, so Vmax = CLmax * km = 197 mg/h. CLmax and central volume (40.8 L) are allometrically scaled on fat-free mass with a 45 kg reference and fixed 0.75 / 1 exponents (FFM beat total body weight by dOFV -30 vs -7.7). CSF is a Sheiner-style effect compartment holding a concentration, equilibrating with plasma at 0.198 1/h (equilibration half-life 3.5 h) toward a pseudo-partition coefficient PPC; PPC rises linearly with CSF total protein and plateaus at PPCmax = 0.365 once CSF protein reaches an estimated 1.18 g/L breakpoint (a broken-stick whose intercept and amplitude were fixed to 0 and 1, so the slope 0.847 = 1/1.18 is fully determined by the breakpoint). Random effects are between-subject variability on CLmax (9.60%), between-visit variability on CLmax across the day-3 and day-28 PK visits (20.3%), and five-occasion between-occasion variability on ka (87.9%) and mean transit time (110%); all reported percentages are the omega standard deviation on the log scale. Residual error is combined proportional plus additive, separately for plasma (21.5%, 0.173 mg/L) and CSF (91.5%, 0.02 mg/L). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Boak_2014_linezolid.html">Linezolid (Boak 2014)</a> </td> <td style="text-align:left;"> Population PK/toxicodynamic model for linezolid in 41 critically ill adult patients (Australia + USA, 42 treatment courses of 600 mg linezolid q12h IV and/or oral for 5-54 days). PK is a one-compartment model with three serial absorption-lag compartments (lat1 -> lat2 -> lat3) followed by a gut compartment (depot) before the central compartment; oral bioavailability is fixed at 100%. Renal function is described by a weight-normalised Cockcroft-Gault GFR expressed as the ratio RF = GFR / 120, driving an additive renal-plus-nonrenal total clearance CL = F_Size_CL * (CL_NR + CL_R * RF) with fixed allometric exponents F_Size_CL = (WT/65)^0.75 and F_Size_V = (WT/65)^1. The toxicodynamic model represents platelet turnover with a 15-compartment bone-marrow precursor chain (precursor1..precursor15, rate ktr = 15/MTT_Pre) feeding a 15-compartment circulating platelet chain (transit1..transit15, rate kout = 15/MTT_PL, Friberg-Bulitta life-span distribution). The observed platelet count is the average of the 15 circulating compartments. Linezolid inhibits synthesis of platelet precursor cells (Imax fixed to 1, so the inhibitory factor simplifies to IC50 / (IC50 + Cc)) with IC50 = 8.06 mg/L, and a homeostatic feedback (Base_PL/PL)^gamma with gamma = 1.02 stimulates precursor synthesis when circulating platelets fall below baseline. Initial-condition scaling factors F_Ini_Pre and F_Ini_PL (typical values fixed to 1, BSV estimated) allow the model to describe subjects whose platelet counts were not at steady state at the start of therapy. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Milakovic_2024_linezolid.html">Linezolid (Milakovic 2024)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order elimination for intravenous linezolid in nine critically ill adults with COVID-19-associated acute respiratory distress syndrome (CARDS) supported by veno-venous extracorporeal membrane oxygenation (vv ECMO), who received a higher-than-standard 600 mg dose as a 30-min infusion every 8 h. Between-subject variability is exponential on both clearance and volume of distribution, estimated as a correlated 2x2 block with a strong negative CL-Vd covariance; residual variability is proportional. No covariate was retained: the automated covariate search found none significant, which the authors attribute to the small, deliberately homogeneous sample. The model was used for Monte Carlo probability of target attainment (PTA) and cumulative fraction of response (CFR) analyses comparing 600 mg every 8 h against the standard 600 mg every 12 h. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schmidt_2009_oxazolidinones.html">Linezolid (Schmidt 2009)</a> </td> <td style="text-align:left;"> In vitro (Staphylococcus aureus MRSA strain OC2878). Mechanism-based PD model of bacterial-killing time-kill curves for linezolid, the first-in-class FDA-approved oxazolidinone (Schmidt 2009). Susceptibility-based two-subpopulation structure: an active self-replicating susceptible pool with logistic carrying-capacity limit and a dormant persister pool that is insusceptible to killing; first-order S->P conversion (P->S held fixed at 0), natural-death loss from both pools, exponential turn-on of growth and of drug-induced killing, and Emax killing of the susceptible subpopulation by the antibiotic. Linezolid was experimentally stable over 24 h so the published model carries kdeg = 0; for dynamic syringe-replacement experiments the user supplies the dilution-equivalent rate (~log(2)/5 1/h for a t1/2 ~ 5 h linezolid regimen) via rxSolve(..., params = c(kdeg = ...)). The same joint fit is shared with Schmidt_2009_rwj416457 (only EC50 and kdeg differ). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Taubert_2016_linezolid.html">Linezolid (Taubert 2016)</a> </td> <td style="text-align:left;"> Population PK model for linezolid in critically ill ICU patients (Taubert 2016). Two-compartment model with first-order absorption from a depot, first-order elimination, complete oral bioavailability (F = 1), and a combined proportional + additive residual error model. Covariate effects: body weight (power on Vc, exponent 1.31) and peritonitis (multiplier on Vc, factor 1.53); fibrinogen concentration (power on CL, exponent 0.04), serum lactate concentration (power on CL, exponent -0.21), and ARDS (multiplier on CL, factor 1.82). All continuous covariates are normalised to the patient-group-1 median (WT = 76 kg, FIB = 13.0 umol/L, LACT = 1.91 mmol/L). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tsuji_2017_linezolid.html">Linezolid (Tsuji 2017)</a> </td> <td style="text-align:left;"> Population PK/PD model for linezolid in hospitalized adult and pediatric patients with MRSA or gram-positive cocci infections (Tsuji 2017). PK is a two-compartment model with first-order oral absorption and an additive renal-plus-non-renal clearance structure (CL = CL_nonren + CL_renal * RF, where RF = CrCl / 100 mL/min/70 kg standardized to 70 kg by allometry); plasma total and unbound concentrations are modelled simultaneously with an estimated fraction-unbound (FU = 0.823) linking the two. PD is a Friberg-style semi-mechanistic platelet turnover model (one proliferating compartment, three transit compartments, one circulating compartment) with an empirical (PLTZERO/PLT)^gamma feedback term and a published mixture model of two thrombocytopenia mechanisms: linear inhibition of platelet synthesis (PDI, 97% of patients, SLOPE on RFORM) and saturable stimulation of platelet elimination (PDS, 3% of patients, Emax on Kcirc), selected per subject by the binary covariate MIX_PDI. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Litjens_2023_linezolid_tuberculous_meningitis.html">Linezolid cns pbpk (Litjens 2023)</a> </td> <td style="text-align:left;"> PBPK (permeability-limited 4-compartment CNS model, adapted from the Simcyp Simulator V19R1 brain module). Linezolid disposition in plasma, brain blood, brain mass, cranial cerebrospinal fluid (CSF) and spinal CSF in adults and children, developed to predict cranial CSF exposure and AUC0-24:MIC target attainment in tuberculous meningitis (Litjens et al. 2023, Antibiotics). The four CNS ODEs, the CNS physiological volumes and the CSF/brain fluid flows are those of the upstream framework paper Verscheijden et al. 2019 (PLoS Comput Biol; Eqs 2-5, S1 Table), which Litjens 2023 adapted with permission; the linezolid drug-specific inputs (Vss, plasma clearance, ka, fa, B:P, fu, fu-brain, PSB, PSC, PSE, and the BCRP / P-gp efflux clearances measured in-house in MDCKII monolayers) are from Litjens 2023 Supplementary Table S1 and S2. Adult and paediatric brain physiology branch on AGE (< 18 y uses the paediatric equations of Verscheijden 2019 S1 Table). IMPORTANT DEVIATION: the systemic side of the published model is the proprietary Simcyp full-PBPK whole-body distribution model (Method 2, Rodgers and Rowland) whose tissue:plasma partition coefficients are not reported anywhere in the paper or its supplement; here it is replaced by a single well-stirred plasma compartment parameterised with the exact aggregate values the paper does report - Vss (L/kg) and the per-study total plasma clearance carried in the Simcyp 'additional clearance' slot. Plasma AUC (= Dose * F / CL, the quantity every PK-PD conclusion of the paper rests on) is preserved exactly; the distribution phase of the plasma curve is approximate. See the vignette Assumptions and deviations section. Typical-value forward-simulation model: the paper reports no IIV or residual-error estimates. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/CarlssonPetri_2021_liraglutide.html">Liraglutide (CarlssonPetri 2021)</a> </td> <td style="text-align:left;"> Liraglutide PK model in adolescents (Carlsson Petri 2021) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Overgaard_2016_liraglutide.html">Liraglutide (Overgaard 2016)</a> </td> <td style="text-align:left;"> Liraglutide 3.0 mg population PK model in overweight and obese adults with and without type 2 diabetes (Overgaard 2016 SCALE Obesity/Prediabetes + SCALE Diabetes pooled analysis) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Watson_2010_liraglutide.html">Liraglutide (Watson 2010)</a> </td> <td style="text-align:left;"> One-compartment population PK model for subcutaneous liraglutide (once-daily human GLP-1 analog) with sequential zero-order-then- first-order (sequential dual) absorption, pooled across four Phase 1 studies (Studies A-D) in healthy volunteers and adults with type 2 diabetes (Watson 2010). The combined-data final model estimates apparent clearance and volume (CL/F, V/F) on a per-kg body-weight basis. Absolute bioavailability F=0.51 was identifiable only in the Study A submodel (with IV data) and is documented in the vignette. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bosch_2022_liraglutide_qsp.html">Liraglutide qsp (Bosch 2022)</a> </td> <td style="text-align:left;"> QSP. Integrated four-hormone-plus-glucose (4GI) model of human glucose regulation coupling glucose, insulin, GLP-1, glucagon, and glucose-dependent insulinotropic peptide (GIP) dynamics after meal and drug perturbation, with liraglutide (GLP-1RA) PK+PD baked in as the training drug. Glucose disposition is a two-compartment model with insulin-dependent and -independent elimination fed by an oral-meal dose compartment plus a buffer and three serial gut transit compartments; insulin, GLP-1, glucagon, and GIP each have one-compartment (GIP two-compartment) turnover with steady-state baseline production. Feedbacks: glucose stimulates insulin production (amplified by GLP-1 and GIP); insulin drives insulin-dependent glucose clearance through an effect compartment; glucose inhibits glucagon production (a below-baseline glucose-on-glucagon effect is estimable in healthy volunteers only); glucagon and GIP each stimulate glucagon and glucose production; GLP-1 inhibits gastric emptying and glucagon production, and stimulates insulin secretion. Liraglutide PK is the Watson 2010 body-weight-scaled one-compartment model (KAdrug 0.154 /h, CL/F 0.013 L/h/kg, V/F 0.16 L/kg, fu 0.005) and drives the three GLP-1 receptor pathways via unbound-liraglutide EC50s derived from the in vitro EC50 ratio between endogenous GLP-1 (1.919 pM) and liraglutide (6 pM). Population-mean fits only (no IIV per Bosch 2022 Methods; five proportional residual errors, one per biomarker output). Healthy volunteer (HV) vs type 2 diabetes mellitus (T2DM) subjects differ in glucose clearance, insulin-dependent glucose clearance, the below-baseline glucose-on-glucagon exponent, and the GIP-on-insulin-secretion exponent, switched by the DIS_DIAB indicator (1 = T2DM, 0 = HV). 18 ODE states; 5 outputs. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Thomson_1989_lisinopril.html">Lisinopril (Thomson 1989)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral lisinopril (an ACE inhibitor) at steady state in elderly and renal-disease hypertensive adults (Thomson 1989). First-order absorption with apparent clearance CL/F driven by body weight, serum creatinine, age, and a binary compensated-cardiac-failure indicator; apparent volume V/F and absorption rate ka are population means without retained covariate effects. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jensen_2023_lngIus52mg.html">LngIus52mg (Jensen 2023)</a> </td> <td style="text-align:left;"> Eight-year population PK / SHBG-turnover model for levonorgestrel (LNG) released from the LNG-releasing intrauterine system (LNG-IUS) 52 mg (Mirena). In-vivo release from the reservoir is described as a mixture of a first-order term (rate coefficient c12), a constant zero-order term (c13), and a time-dependent first-order term with time-decay constant t1, giving DADT(depot) = -c12 * depot - c13 * (1 + depot / (t1 + t)). Two- compartment LNG disposition treats only unbound drug as distributable / eliminable via K20 * fuLNG * central and K23 * fuLNG * central, where fuLNG is the closed-form free-fraction solution to reversible LNG binding to albumin (constant, KDA = 18209 nmol/L, ALB = 700000 nmol/L) and to SHBG (KDS = 1.82 nmol/L). SHBG serum concentration is modelled with an indirect-response turnover (zero-order kin, first-order kout) whose synthesis is linearly inhibited by a delay-compartment-smoothed unbound LNG signal (delay time-constant tau). Body weight enters as an allometric-style power on apparent (free) LNG clearance and on the SHBG baseline. Bioavailability of the loaded LNG reservoir is fixed via a logit anchor (F1 ~ 0.971 of the 52 mg reservoir). All rate constants are expressed per hour. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Archary_2018_lopinavir.html">Lopinavir (Archary 2018)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model for oral lopinavir/ritonavir in severely malnourished HIV-infected children, with FFM allometric scaling and a linear total-cholesterol effect on apparent clearance (Archary 2018). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Crommentuyn_2005_lopinavir.html">Lopinavir (Crommentuyn 2005)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model for oral lopinavir co-administered with ritonavir in 122 HIV-1-infected adults on BID lopinavir/ritonavir 400-666/100-166 mg. Apparent oral clearance CL/F follows an inverse-saturable function of per-subject ritonavir AUC over the 12 h dosing interval (CONMED_RTV_AUC_12h, mg*h/L, computed from the upstream Kappelhoff 2005 ritonavir popPK model) plus a pooled +39% NNRTI co-medication factor (efavirenz or nevirapine, encoded as the CONMED_NNRTI class indicator). IIV is estimated on ka, CL/F, and V/F as a full 3x3 correlated block; residual error is combined additive plus proportional. The reported IOV on relative bioavailability F (17.5% CV) is NOT encoded structurally (Brooks 2021 precedent); downstream users who want IOV can add an OCC covariate and a per-occasion eta in rxode2 (Crommentuyn 2005). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jullien_2006_lopinavir.html">Lopinavir (Jullien 2006)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral lopinavir (boosted with ritonavir) in HIV-infected children from birth to 18 years, with the absorption and elimination rate constants constrained to a single shared rate constant k = CL/F divided by V/F (Jullien 2006, simplified parameterisation per Wahlby 2002). Body weight is allometrically scaled on CL/F and V/F (reference 27 kg), nevirapine coadministration increases CL/F by 34%, and male sex increases CL/F by 39% in children older than 12 years. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schipani_2012_lopinavir.html">Lopinavir (Schipani 2012)</a> </td> <td style="text-align:left;"> Population PK model for boosted lopinavir (lopinavir/ritonavir 400/100 mg) in HIV-infected adults from the Liverpool Therapeutic Drug Monitoring Registry. One-compartment with first-order absorption; apparent clearance is modified additively by body weight (deviation from median 72 kg) and by SLCO1B1 521T>C (rs4149056) genotype, encoded via the paired SLCO1B1_HAP15_HET / SLCO1B1_HAP15_HOM indicators (the source paper genotyped only 521T>C so *5- and *15-haplotype carriers are pooled, per the canonical's documented pooling rule). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Thoueille_2023_lopinavir.html">Lopinavir (Thoueille 2023)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model for oral ritonavir-boosted lopinavir (LPV/r) in HIV-negative individuals receiving 5-day COVID-19 post-exposure prophylaxis (COPEP study) pooled with people living with HIV followed by routine therapeutic drug monitoring; body weight enters apparent oral clearance as a linear deviation from a 70 kg reference (Thoueille 2023). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fauchet_2015_lopinavir.html">Lopinavir placental (Fauchet 2015)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model for total lopinavir in HIV-infected pregnant and nonpregnant women with a maternal-to-fetal effect-compartment placental-transfer chain and a downstream fetal-to-amniotic-fluid distribution-and-elimination chain; a 39% pregnancy effect is applied multiplicatively to apparent maternal CL (Fauchet 2015 MFLA submodel). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2012_lopinavir_ritonavir.html">Lopinavir ritonavir (Zhang 2012)</a> </td> <td style="text-align:left;"> Simultaneous integrated population pharmacokinetic model of oral lopinavir (LPV, parent) and ritonavir (RTV, sibling-drug suffix _rtv) in 21 HIV-infected South African adults with and without concomitant antitubercular rifampicin (Zhang 2012). Structure: LPV one-compartment with first-order absorption (ka 0.991 1/h) and LPV CL/F dynamically inhibited by RTV plasma concentration via a sigmoid Imax (Imax = 0.953, IC50 = 0.0351 mg/L); RTV two-compartment with a Savic transit- compartment absorption chain (NN = 2.03, MTT = 1.44 h) feeding RTV depot at rate ktr = (NN+1)/MTT and absorbed to RTV central at ka_rtv = 3.28 1/h. Allometric scaling fixed at the Holford / Anderson literature values: fat-free mass (Janmahasatian) drives CL/F (exponent 0.75) and total body weight drives Vc/F and Vp/F (exponent 1.0). Rifampicin (CONMED_RIF) increases LPV CL/F by 71.0% and RTV CL/F by 36.0%, reduces LPV F by 20.0% and RTV F by 45.0% (at the 100 mg reference RTV dose), and the RTV F when on rifampicin scales upward with RTV dose at 8.1% per 10 mg above the 100 mg reference (saturation of first-pass metabolism / P-gp self-inhibition; identifiable only within the RIF-coadministered arm of the source study). Diurnal variation is encoded via the simulation convention t = clock-hours- from-midnight: doses given during the overnight window (clock 20:00 to 08:00) carry +42.0% (LPV) and +45.0% (RTV) relative bioavailability vs morning doses, and oral CL/F of both drugs is reduced by 32.7% overnight. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2013_lopinavir_ritonavir.html">Lopinavir ritonavir (Zhang 2013)</a> </td> <td style="text-align:left;"> Integrated population PK model for lopinavir (1-compartment, first-order absorption) and ritonavir (2-compartment, transit-chain absorption with N=2 transit compartments) co-administered to HIV-infected adults (n=21) and children (n=74; 35 of whom received rifampicin-based antitubercular treatment). Ritonavir plasma concentration inhibits lopinavir apparent clearance via a sigmoidal Emax DDI (Emax=0.82, EC50=0.098 mg/L, Hill=2.8). Rifampicin coadministration increases apparent clearance and reduces relative bioavailability of both drugs, with separate magnitudes for adults vs children. Ritonavir dose (mg/kg) drives a linear increase in relative bioavailability of both drugs; lopinavir-on-adults is the only dose-effect cell not supported by the data. Diurnal variation is encoded as a step function with overnight reduction in apparent clearance (adults 51%, children 27%) and increased bioavailability at the evening lopinavir dose for adults (+19%). Allometric scaling on apparent CL/Q (exponent 0.75) and apparent V/Vp (exponent 1) with reference body weight 65 kg (Zhang 2013). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2012_lopinavir_ritonavir_pedi.html">Lopinavir ritonavir pedi (Zhang 2012)</a> </td> <td style="text-align:left;"> Integrated one-compartment popPK model for oral lopinavir (LPV) and ritonavir (RTV) in 74 HIV-infected children (6 months to 4.5 years) treated with LPV/r oral solution with or without concomitant rifampicin-based antitubercular treatment (Zhang 2012). LPV uses a one-compartment model with first-order absorption; RTV uses a one- compartment model with a Savic-style 10-transit-compartment absorption chain followed by a separate first-order absorption step from the last transit to central. Apparent CL/F and V/F are allometrically scaled to the cohort median 10 kg with fixed exponents 0.75 / 1. The dynamic LPV-RTV interaction is encoded as direct sigmoid-Emax inhibition of LPV apparent clearance by RTV plasma concentration (Emax = 0.9 fixed, EC50 = 0.0519 mg/L). Lopinavir bioavailability is modulated by concomitant rifampicin-based antitubercular treatment (-83.2% at the no-extra-ritonavir reference) and by the concomitant ritonavir dose in mg/kg (+2.1% per mg/kg above the 3 mg/kg reference). Ritonavir apparent clearance is +50% in subjects on rifampicin-based antitubercular treatment. Both drugs share random effects modelled as log-normal between- subject variability with selected inter-occasion variabilities folded in as BSV-equivalent (see vignette Assumptions and deviations). Residual error is proportional on the linear scale (implemented via NONMEM exponential error on log-transformed data). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fauchet_2015_lopinavir.html">Lopinavir unbound (Fauchet 2015)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model for lopinavir in HIV-infected pregnant and nonpregnant women parameterised on the unbound fraction, with total LPV reconstructed from a linear HSA binding term plus a saturable single-site AAG binding term (Fauchet 2015 unbound submodel). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Swart_2004_lorazepam_midazolam.html">Lorazepam (Swart 2004)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for lorazepam by continuous infusion in mechanically ventilated critically ill adult ICU patients. Clearance is selected by chronic alcohol-abuse status (a flat 0.74 L/h for alcohol-abuse subjects; a PEEP-adjusted 4.13 - (PEEP - 5) * 0.417 L/h otherwise). Steady-state volume of distribution decreases linearly with age above 58 years. Fitted by NONMEM V in the Swart 2004 learning cohort (n = 28). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ahmed_2016_lorenzosOil.html">LorenzosOil (Ahmed 2016)</a> </td> <td style="text-align:left;"> Population pharmacodynamic model of Lorenzo's oil effect on plasma C26:0 in asymptomatic boys with X-linked adrenoleukodystrophy: inhibitory fractional Emax model relating observed plasma erucic acid concentration to plasma C26:0. The paper does not develop a PK model for erucic acid; observed erucic acid plasma concentration is supplied as a time-varying covariate. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2021_lorlatinib.html">Lorlatinib (Chen 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral lorlatinib (a third- generation ALK/ROS1 tyrosine kinase inhibitor) in adult patients with advanced ALK-positive or ROS1-positive non-small cell lung cancer and healthy participants (Chen 2021; N = 425 across seven studies). Disposition is a two-compartment model with sequential zero-order and first-order oral absorption (dose enters the depot via a zero-order window of duration D1 = 1.15 h followed by first-order absorption at ka = 3.11 h^-1) and time-varying metabolic auto-induction of clearance: CL(t) = CLI + (CLMX - CLI) * (1 - exp(-cl_exp_kdes * t)), rising from a single-dose CLI = 9.04 L/h to a steady-state CLMX = 14.5 L/h with induction rate constant cl_exp_kdes = 0.020 h^-1 (~7.25 d to functional steady state; Chen 2021 abstract, Table 4). CLI and CLMX share a fixed allometric exponent 0.75 on body weight (reference 70 kg) and both are modulated by a shared multiplicative covariate block: 1 + e_alb_cl * (ALB - 40 g/L) with e_alb_cl = 0.00670 per g/L, 1 + e_dose_lor_cl * (DOSE_LOR_MGD - 100 mg/day) with e_dose_lor_cl = 0.00100 per mg/day, and a power effect (CRCL / 100)^e_crcl_cl with e_crcl_cl = 0.235. V2 = 121 L carries a fixed allometric exponent 1.0 on body weight; V3 = 155 L and Q = 22.0 L/h have no allometric scaling. ka is modulated by proton-pump inhibitor co-administration: ka x (1 - 0.675 * CONMED_PPI), i.e. a 67.5% ka reduction on PPI (which reduces Cmax by ~30% with no effect on AUCinf per the Chen 2021 Discussion). Bioavailability F1 = 0.759. Inter-individual variability is a correlated CL/F block (log-scale variances 0.030 and 0.022; covariance -0.006), a correlated V2/V3 block (0.086, -0.017, 0.101), and an independent ka block (2.33). Residual error is a route-specific log-scale additive term (approximately proportional in linear space): propSd 11.5% CV for IV data (Study B7461007 absolute bioavailability arm) and 43.8% CV for oral data. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2013_losmapimod.html">Losmapimod (Yang 2013)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order elimination and time-dependent first-order absorption (change-point MTIME between two absorption rate constants XKA1 and XKA2) plus an absorption lag time ALAG1, for oral losmapimod (a p38 alpha/beta mitogen-activated protein kinase inhibitor). Data pooled from 30 healthy volunteers, 23 patients with rheumatoid arthritis, and 24 patients with chronic obstructive pulmonary disease across four GSK studies (Yang 2013). Retained covariates on inter-individual variability are sex on CL/F (males reference; females get a lower CL/F), age on V1 (power exponent centred at median age 54 years), and body weight on both absorption rate constants (shared power exponent centred at median 74 kg). Residual error is proportional and larger in COPD patients than in the pooled healthy + RA reference. Doses were 5, 7, 7.5, 10, and 20 mg oral single- or repeat-dose; 60 mg data were excluded from the analysis due to lack of dose-proportionality (possible saturable absorption). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Siebinga_2023_lu177psma617.html">Lu177psma617 (Siebinga 2023)</a> </td> <td style="text-align:left;"> Six-compartment population PK dosimetry model for the PSMA-targeted radioligand [177Lu]Lu-PSMA-617 in men with low-volume metastatic prostate cancer, built from SPECT/CT-derived organ and tumor activity plus blood samples over two treatment cycles (~3 and ~6 GBq). States are blood (1), salivary gland (2), kidney (3), liver (4), tumor (5) and a lumped other-tissue compartment (6); every state carries a decay-corrected radioactivity amount (MBq) and only the blood compartment has an estimated volume (V1 = 10.3 L), so blood is observed as a concentration (MBq/L) and the tissues as amounts. Salivary-gland uptake is saturable in the bound amount (Bmax 40.4 MBq, IIV 59.3% CV); all other tissue exchange is first order. The tumor uptake rate constant carries a power effect of total tumor volume (exponent 0.705), interindividual variability (58.8% CV) and interoccasion variability across the two cycles (43.5% CV), and the blood excretion rate constant is scaled a priori by Cockcroft-Gault creatinine clearance under an assumption of complete renal elimination. The blood compartment has two observation channels with separate residual error: blood samples, and SPECT-derived blood activity linearly recalibrated to the blood-sample scale. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Siebinga_2024_lu177psmaIT.html">Lu177psmaIT (Siebinga 2024)</a> </td> <td style="text-align:left;"> Five-compartment population PK model with a sequentially fitted PSA pharmacodynamic layer for the PSMA-targeted radioligand [177Lu]Lu-PSMA-I&T in men with metastatic castration-resistant prostate cancer treated with ~7.4 GBq per cycle. PK states are central (1), salivary glands (2), kidneys (3), tumor lesions (4) and a lumped remaining-tissue compartment (5); the tissue states carry radioactivity amounts (MBq) and the central state is observed as a concentration (MBq/L) through a fixed volume V1 = 10.3 L. Renal excretion leaves the central compartment (k10 = 0.253 1/h, i.e. 2.61 L/h). Salivary-gland uptake is saturable (Bmax 134 MBq); all other exchange is first order. Tumor uptake carries a power effect of segmented tumor volume (exponent 1.08), a structural decline over treatment cycles (73%, 50% and 44% of the cycle-1 rate in cycles 2, 3 and 4-7), interindividual variability (63.4% CV) and interoccasion variability across cycles (37.8% CV), and salivary-gland uptake carries a weak tumor-volume power effect. All PK parameters are allometrically scaled to the 79 kg median body weight. The PD layer is a single PSA state growing exponentially (kG 0.000408 1/h) and eliminated by a direct effect linear in the physical tumor activity concentration plus a delayed effect linear in an effect-compartment concentration (ke0 0.00128 1/h). Baseline PSA is fixed at 140 ug/L with a linear tumor-volume effect. Every state carries a decay-corrected radioactivity amount, matching the source SPECT/CT data and the predecessor [177Lu]Lu-PSMA-617 model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hietala_2010_artemether_lumefantrine_malaria.html">Lumefantrine (Hietala 2010)</a> </td> <td style="text-align:left;"> Population PK model for oral lumefantrine (LUM) in 50 Tanzanian children (ages 1-10 years, weights 8-30 kg) with uncomplicated Plasmodium falciparum malaria treated with the standard six-dose weight-based Coartem (artemether 20 mg + lumefantrine 120 mg per tablet) regimen at 0, 8, 24, 36, 48, and 60 hours (Hietala 2010). One-compartment disposition with first-order absorption preceded by an absorption lag time. The paper tested co-administration with full-fat (3.4%) cow's milk as a categorical covariate on the PK parameters of LUM; the effect did not improve the model fit and is not encoded here (Discussion: 'the resulting number of doses actually administered with an adequate amount of milk may have been too small to allow the detection of a difference'). All PK parameters are reported per kg body weight (linear weight normalisation applied inside model()). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hoglund_2015_artemether_lumefantrine.html">Lumefantrine (Hoglund 2015)</a> </td> <td style="text-align:left;"> Joint parent-metabolite population PK model for oral lumefantrine and its major oxidative metabolite desbutyl-lumefantrine in 89 HIV-infected Ugandan adults receiving artemether-lumefantrine (Coartem) with or without concomitant antiretroviral therapy (efavirenz, nevirapine, or lopinavir/ritonavir) (Hoglund 2015). 1-transit-compartment absorption with ka = ktr feeds a 2-compartment lumefantrine disposition; complete in-vivo conversion of lumefantrine to a 1-compartment desbutyl-lumefantrine disposition with stoichiometric molar conversion. Relative bioavailability F is anchored at 1 (fixed) with log-normal IIV (47.4 % CV). Three antiretroviral drug-drug interactions are encoded as linear-deviation effects on parent clearance and on bioavailability: efavirenz increases LF CL/F by 72.6 %, lopinavir/ritonavir decreases LF CL/F by 62.1 % and increases desbutyl-lumefantrine CL/F by 392 %, nevirapine decreases relative bioavailability by 24.8 %. IIV is retained on LF CL, the mean transit time, and the relative bioavailability F. NONMEM additive residual error on log-transformed concentrations is encoded as a proportional residual in linear concentration space for both parent and metabolite. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kay_2020_lumefantrine.html">Lumefantrine (Kay 2020)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral lumefantrine in 277 HIV- infected and HIV-uninfected Ugandan children (3 months to ~10 years) with uncomplicated malaria receiving artemether-lumefantrine alone or with concomitant ART (efavirenz, lopinavir/ritonavir, or nevirapine) (Kay 2020, ASTMH poster 2167). First-order absorption (depot -> central) feeds a 2-compartment lumefantrine disposition (central + peripheral1). Body-weight allometric scaling enters on all clearance and volume terms with a fixed volume exponent of 1 and a piecewise age-dependent clearance exponent (0.75 for age >60 mo, 0.9 for >24-60 mo, 1.0 for >3-24 mo, 1.2 for <=3 mo). Age also enters as a covariate on relative bioavailability F (younger children have reduced F). Three ART drug-drug interactions are encoded as linear- deviation effects on apparent oral clearance CL/F and on first-order absorption KA: efavirenz, lopinavir/ritonavir, and nevirapine. Diagonal IIV is retained on CL/F, V2/F, Q/F, V3/F, and KA. The NONMEM additive-on-log-scale residual is encoded as a proportional residual in linear concentration space (consistent with the related Hoglund 2015 Ugandan-adult lumefantrine model). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kay_2022_lumefantrine.html">Lumefantrine (Kay 2022)</a> </td> <td style="text-align:left;"> Population PK model for oral lumefantrine in 277 Ugandan children (186 HIV-uninfected, 178 HIV-infected on efavirenz-, nevirapine-, or lopinavir/ritonavir-based antiretroviral therapy plus daily trimethoprim-sulfamethoxazole prophylaxis) ages ~2 months to 8.6 years treated with six-dose weight-based Coartem Dispersible (20 mg artemether + 120 mg lumefantrine per tablet) for uncomplicated Plasmodium falciparum malaria (Kay 2022). Two-compartment disposition with first-order absorption. Body-weight fixed effects scale all clearance and volume terms with a reference weight of 15 kg; volumes use an allometric exponent of 1, clearances use an age-dependent exponent (1.2 for age <= 3 months, 1.0 for >3 to 24 months, 0.9 for >24 to 60 months, 0.75 for >60 months) from Anderson & Holford 2009 (paper ref 34). A power-form age effect on relative bioavailability captures reduced lumefantrine bioavailability in young children (F = (age_months / 50)^0.204). Concomitant antiretroviral therapy enters as mutually-exclusive linear-deviation effects on apparent oral CL/F and ka: efavirenz increases CL/F by 98.2% and ka by 48.4%, lopinavir/ritonavir decreases CL/F by 51.4% and ka by 21.2%, nevirapine has no statistically significant effect (both CIs cross zero). Q/F IIV is fixed at 15.9% CV; the remaining four structural parameters carry estimated log-normal IIV (104%, 112%, 127%, and 16.9% CV for CL/F, V2/F, V3/F, and ka respectively). NONMEM proportional residual error on linear concentration (sigma^2 = 0.200, 44.7% CV). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kloprogge_2013_lumefantrine.html">Lumefantrine (Kloprogge 2013)</a> </td> <td style="text-align:left;"> Population PK model for oral lumefantrine in pregnant and non-pregnant women with uncomplicated Plasmodium falciparum malaria in Uganda after the standard fixed-dose oral artemether-lumefantrine treatment (Kloprogge 2013). Flexible five-compartment transit absorption chain into a two-compartment disposition model with relative bioavailability F1 fixed at 1, log-normal IIV on CL / Vp / MTT / F, and covariate effects of pregnancy on intercompartmental clearance (-36.5%, categorical) and body temperature on mean absorption transit time (+16.5% per degC over 36.0-39.8 degC, linear-deviation centered at the cohort median 36.9 degC). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kloprogge_2015_lumefantrine.html">Lumefantrine (Kloprogge 2015)</a> </td> <td style="text-align:left;"> Simultaneous parent + active-metabolite (desbutyl-lumefantrine, DLF) population PK model for oral lumefantrine in 116 pregnant women (second or third trimester) with uncomplicated Plasmodium falciparum malaria on the Thailand-Myanmar border treated with the standard fixed-dose artemether-lumefantrine regimen (Kloprogge 2015). First-order absorption with lag time into a two-compartment LF disposition with relative bioavailability F fixed at 1 and Box-Cox-transformed IIV on F (Box-Cox shape -0.394 not encoded -- see Errata); DLF is formed mole-for-mole from LF central elimination (linear drug-metabolite chain, fraction metabolised assumed = 1) and disposes through its own two- compartment chain with apparent CL/F = 197 L/h and Vc/F = 6,490 L. Retained covariates: estimated gestational age (power on LF ka, linear-deviation on LF Q/F, both centered on the cohort median 22.8 weeks) and admission parasitaemia (log10-exponential on DLF CL/F, centered on cohort median log10(3,260) = 3.513). Venous-only residual error encoded; capillary-residual variance components and capillary conversion factors (LF 0.878, DLF 0.464) NOT encoded -- see Errata. Time-to-event PD layer (Gompertz hazard with E_max LF effect on recrudescent malaria, Table 4) NOT encoded -- see Errata. Parameter values from Kloprogge 2015 Table 2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kloprogge_2018_lumefantrine.html">Lumefantrine (Kloprogge 2018)</a> </td> <td style="text-align:left;"> Population PK model for oral lumefantrine in 1,347 patients (children, non-pregnant adults, and second-/third-trimester pregnant women) from 26 studies in 12 African, Oceanian, and Southeast Asian countries with uncomplicated Plasmodium falciparum malaria treated with the standard fixed-dose artemether-lumefantrine regimen (Kloprogge 2018 PLOS Medicine). Two-compartment disposition with first-order absorption; F fixed at 1 with log-normal IIV (Box-Cox shape -0.343 on the F IIV departure from log-normal not reproduced here -- see Errata); allometric scaling of CL/F and Q/F (power 3/4) and of Vc/F and Vp/F (power 1) on body weight centered at the model-building median 42 kg; dose-saturable absorption on F with Dose50 = 3.86 mg/kg; exponential effect of log10 admission parasitaemia on F centered at log10(15,800/uL) = 4.2 (coefficient -0.643 per log10 unit); proportional pregnancy effect on ka (+35.2% in second and third trimester). IIV on Vc/F (CV 144%) and F (CV 70.3%); additive log-scale residual SD 0.323. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mosha_2014_artemether_lumefantrine.html">Lumefantrine (Mosha 2014)</a> </td> <td style="text-align:left;"> Population PK model for oral lumefantrine in 33 pregnant (2nd or 3rd trimester) and 22 non-pregnant women with uncomplicated Plasmodium falciparum malaria in Rufiji, Tanzania after standard fixed-dose artemether-lumefantrine (Mosha 2014). One-compartment disposition with first-order absorption and ka fixed at 0.54 1/h. Relative bioavailability F1 is fixed at 1 (structural anchor) with a categorical pregnancy effect of -33% on F1 (linear-deviation form) and log-normal IIV around the typical F1 (65% CV). The published model used a logit transformation on individual F1 to constrain individuals to (0, 1); this encoding uses log-normal IIV on F (matching the established Kloprogge 2013 / 2018 lumefantrine precedents in nlmixr2lib). Structural CL and Vc do not carry IIV in the final model; the F1 IIV absorbs the joint CL/Vc variability via the AUC = D x F / CL relationship. The desbutyl-lumefantrine (DLF) metabolite arm of the published joint model is not encoded; see the validation vignette for the rationale. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Simpson_2013_lumefantrine.html">Lumefantrine (Simpson 2013)</a> </td> <td style="text-align:left;"> In vitro (P. falciparum). Sigmoid Emax inhibition model of lumefantrine effect on hypoxanthine uptake by clinical Plasmodium falciparum isolates from the Thai-Myanmar border (Shoklo Malaria Research Unit, 1993-2005), with pfmdr1 genotype covariate effects on EC50. The 'subject' in the NLME framework is a parasite isolate (n=324 isolates with lumefantrine data). STIM_LUMEFANTRINE_NM is the per-well drug concentration in the in vitro hypoxanthine-uptake-inhibition assay; the model has no PK and no time evolution. E0 and Emax are fixed per Simpson 2013 Table 3 footnote. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/VasquezBahena_2009_lumiracoxib_rat.html">Lumiracoxib rat (VasquezBahena 2009)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment population PK plus indirect-response PK/PD model for the antinociceptive effect of oral lumiracoxib in carrageenan-induced thermal hyperalgesia in female Wistar rats (Vasquez-Bahena 2009). PK: first-order absorption with lag time and dose-dependent relative bioavailability. PD: time-variant (gamma function) carrageenan-induced COX-2 synthesis with first-order COX-2 degradation; lumiracoxib reversibly inactivates COX-2 via a competitive binding model (COX-2_act = KD * COX-2 / (KD + Cp)). The level of inflammatory mediators (MED) equals the active COX-2 amount and drives the paw withdrawal latency response LT = LT0 / (1 + MED). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/VelezdeMendizabal_2012_lumiracoxib_rat.html">Lumiracoxib rat (VelezdeMendizabal 2012)</a> </td> <td style="text-align:left;"> Preclinical (rat). Semi-mechanistic PD model of the formalin-induced antinociceptive response to lumiracoxib in adult female Wistar rats (Velez de Mendizabal 2012). No PK measurements were made: lumiracoxib was tracked through two virtual compartments -- intraplantar local (lumxLocal) and intrathecal central (lumxCns) -- each decaying monoexponentially at first-order rates K_D_Local and K_D_CNS from a bolus equal to the administered dose (10, 30, 100, or 300 ug per route). The biphasic formalin-induced nociceptive response (flinch count per 1-min window) is modeled as the sum of an early phase PN1, a monoexponential decay from an initial pain load PN1_0 with rate K_PN1 (insensitive to lumiracoxib), and a delayed phase PN2 built from upregulated COX-2 in the local and CNS compartments. Both COX-2 species are taken proportional to a pain-mediator signal MED whose time course is the analytical Erlang-transit kernel of Savic 2007 (MED0 = 1; chain length NC = 6.5; transit rate K_TR = 0.233 min^-1), and the proportionality constants theta_COX2_L / theta_COX2_CNS scale MED to flinch units in the local and CNS arms respectively. Lumiracoxib inhibits upregulated COX-2 in each arm via E = 1 / (1 + LUMX) with an implicit IC50 of one dose unit (an IC50 parameter was tested and found not significant). Model is the second-pass selection (Table I of Velez de Mendizabal 2012); the IC50, delayed-COX-2, and Emax variants were rejected during model development. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2020_luspatercept.html">Luspatercept (Chen 2020)</a> </td> <td style="text-align:left;"> One-compartment population PK model for luspatercept (activin receptor type IIB / IgG1 Fc-fusion) in adults with anemia due to myelodysplastic syndromes (Chen 2020), with first-order subcutaneous absorption, first-order linear elimination parameterised in CL/F and V1/F, body weight + age + baseline albumin power covariates on CL/F, and body weight + baseline albumin power covariates on V1/F. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Xie_2000_m3g_rat.html">M3g rat (Xie 2000)</a> </td> <td style="text-align:left;"> Preclinical (rat, male Sprague-Dawley). Blood-brain barrier (BBB) distributional model for morphine-3-glucuronide (M3G) in rat as published by Xie et al. (2000, Br J Pharmacol): a one-compartment plasma PK driven by an unbound systemic clearance CL_u = 3.8 mL/min from the paper's Model A, coupled to a two-compartment brain model (brain 1 = sampled brain extracellular fluid via striatal microdialysis, brain 2 = deeper redistribution compartment) with asymmetric BBB exchange (separate unbound influx CL_u,in and efflux CL_u,out across the BBB) and a symmetric intercompartmental clearance Q_br between the two brain compartments. The model captures a probenecid-sensitive organic-anion transport contribution to BBB influx: CL_u,in is 1.55-fold higher under co-administered probenecid (CONMED_PROBENECID = 1) while CL_u,out, Q_br, and the two brain volumes are unchanged. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lignet_2023_m8891.html">M8891 human (Lignet 2023)</a> </td> <td style="text-align:left;"> Predicted-human translational PK/PD model for M8891, a selective and reversible methionine aminopeptidase 2 (MetAP2) inhibitor. One-compartment oral PK whose disposition (CL, Vss) is the mean of three preclinical-to-human scaling methods and whose absorption rate constant ka comes from a GastroPlus PBPK model, coupled to the effect-compartment plus Met-EF1a turnover PD model estimated in Caki-1 xenograft-bearing mice. No human subjects were dosed in this analysis; the model is the forward projection that was used to select the M8891 dose for the Phase Ia study NCT03138538. PD parameters are carried over unchanged from the mouse model on the hypothesis that the same Met-EF1a modulation level is associated with efficacy in humans. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lignet_2023_m8891.html">M8891 mouse (Lignet 2023)</a> </td> <td style="text-align:left;"> Preclinical (mouse, Caki-1 renal-carcinoma xenograft). One-compartment oral PK of M8891, a selective and reversible methionine aminopeptidase 2 (MetAP2) inhibitor, with first-order absorption and elimination, linked through an effect compartment to a turnover model for the tumour target-engagement biomarker Met-EF1a (uncleaved methionine-elongation-factor-1-alpha). M8891 inhibits the first-order degradation of Met-EF1a, so the biomarker accumulates above its kin/kout baseline. Naive-pooled fit (Phoenix WinNonlin 6.4) to single-dose and 4-day repeated-dose PK/PD data at 10, 25, and 100 mg/kg p.o.; the authors report no inter-individual variability because a nonlinear mixed-effects fit did not converge on this dataset. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Betts_2018_mAb_linear_PK.html">MAb combined (Betts 2018)</a> </td> <td style="text-align:left;"> Preclinical + human. Class-level typical two-compartment population PK model with body-weight allometric scaling for monoclonal antibodies with linear clearance, jointly fitted across human, cynomolgus monkey and hFcRn Tg32 mouse (Betts 2018 combined all-species dataset, n=27 mAbs). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Betts_2018_mAb_linear_PK.html">MAb cyno (Betts 2018)</a> </td> <td style="text-align:left;"> Preclinical (cynomolgus monkey). Class-level typical two-compartment population PK model for monoclonal antibodies with linear clearance in cynomolgus monkeys (Betts 2018, n=23 mAbs). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Betts_2018_mAb_linear_PK.html">MAb human (Betts 2018)</a> </td> <td style="text-align:left;"> Class-level typical two-compartment population PK model for monoclonal antibodies with linear clearance in adult humans (Betts 2018, n=18 mAbs). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Muliaditan_2025_mab_mpbpk.html">Mab mpbpk human (Muliaditan 2025)</a> </td> <td style="text-align:left;"> Human-scaled. Translational minimal physiologically based pharmacokinetic (mPBPK) model for transferrin-receptor (TfR) mediated brain delivery of monoclonal antibodies, projected forward from the cynomolgus monkey fit Muliaditan_2025_mab_mpbpk_nhp by replacing the Bloomingdale 2017 NHP physiology with the human physiology (Muliaditan 2025 Supplementary Table S1, human column), allometrically scaling the bsAb-TfR internalization rate kint by (70/6.2)^(-0.25) = 0.546 (paper Methods: standard rate-constant exponent -0.25), and recalibrating the luminal BCSFB unbound TfR baseline uTFR0_BCSFB to be 3-fold higher than the NHP estimate (0.256 -> 0.768 nM) per the paper Results. The other TfR- related parameters (TfRpt, uTFR0_BBB, FACQ_BECF, TfRtotn, ktrans, kdeg_uTfR_BBB, kdeg_uTfR_BCSFB, FACBR) are assumed identical to the NHP estimates per paper Methods. Per-compound TfR binding parameters (kon_T, koff_T) MUST be set per simulated antibody from biophysical measurements (paper Table S2); the default ini() encodes kon_T = 0 (non-TfR control IgG). For trontinemab in human, the paper reports KD,TfR = 131 nM (versus 249 nM in NHP) with kon_T = 1.0548 nM^-1 h^-1 and koff_T = 138.24 h^-1 (Table S2, Grimm 2023 column for human). Clinical validation in the paper was against single ascending doses 0.1-7.2 mg/kg IV trontinemab in healthy human subjects (NCT04023994; Grimm 2023). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Muliaditan_2025_mab_mpbpk.html">Mab mpbpk nhp (Muliaditan 2025)</a> </td> <td style="text-align:left;"> Preclinical (cynomolgus monkey). Translational minimal physiologically based pharmacokinetic (mPBPK) model for transferrin-receptor (TfR) mediated brain delivery of monoclonal antibodies in non-human primates. 26-compartment NONMEM ADVAN8 structure combining the Bloomingdale 2017 mAb mPBPK framework (plasma, tissue vascular / endosomal / interstitial / FcRn, brain vascular, BBB endosomal (unbound + FcRn-bound), brain ISF, BCSFB endosomal (unbound + FcRn-bound), CSF, lymph, with FcRn recycling) and the Chang 2022 whole-body-plasma + brain-vascular + ISF + neuronal TfR binding with empirical TMDD-style elimination of the bsAb-TfR complex (kint). Two TfR binding sites on the brain barriers (luminal BBB, luminal BCSFB) transcytose bound complex into the abluminal side (brain ISF and CSF respectively), where it may dissociate or be degraded. Parameters were fit to 395 plasma, 81 CSF, and 102 brain mean concentrations digitised from eight literature studies in cynomolgus monkey (7 non-TfR mAbs + 10 anti-TfR bsAbs with KD,TfR 36-1900 nM). The kint mixture (POP1 fast 0.0329 h^-1 fraction 0.437; POP2 slow 0.0125 h^-1) is selected per subject via the MIX_FAST_ELIM covariate (1 = POP1, 0 = POP2). Per-compound TfR binding parameters (kon_T, koff_T) are NOT estimated population values - they are set per simulated antibody from biophysical measurements (paper Table S2); the default ini() encodes kon_T = 0 (non-TfR control IgG). Plasma observation is TOTAL drug (free + TfR-bound complex), CSF observation is unbound CBCSF, brain observation is whole-brain homogenate (simplified volume-weighted average across BBB + ISF + BCSFB endosomal spaces, scaled by the estimated FACBR correction factor 0.05). No inter-individual variability was estimated (dataset was mean digitised profiles). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Betts_2018_mAb_linear_PK.html">MAb tg32mouse (Betts 2018)</a> </td> <td style="text-align:left;"> Preclinical (hFcRn Tg32 transgenic mouse). Class-level typical two-compartment population PK model for monoclonal antibodies with linear clearance in mice expressing human neonatal Fc receptor (Betts 2018, n=11 mAbs). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb1 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment population PK model for the non-disclosed humanized IgG antibody mAb1 (Zheng 2012; pI 6.1) with first-order SC absorption and linear central clearance in Gottingen minipigs (Table 1 population mean parameter estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb2 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment population PK model for the non-disclosed humanized IgG antibody mAb2 (Zheng 2012; pI 8.7; unusually fast systemic clearance in minipig) with first-order SC absorption and linear central clearance in Gottingen minipigs (Table 1 population mean parameter estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb3 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment population PK model for the non-disclosed humanized IgG antibody mAb3 (Zheng 2012; pI 9.1) with first-order SC absorption and linear central clearance in Gottingen minipigs (Table 1 population mean parameter estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb4 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment population PK model for the non-disclosed humanized IgG antibody mAb4 (Zheng 2012; pI 9.3) with first-order SC absorption and linear central clearance in Gottingen minipigs (Table 1 population mean parameter estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb5 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment population PK model for the non-disclosed humanized IgG antibody mAb5 (Zheng 2012; pI 9.4) with first-order SC absorption and linear central clearance in Gottingen minipigs (Table 1 population mean parameter estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb6 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment population PK model for the non-disclosed humanized IgG antibody mAb6 (Zheng 2012; pI 9.2) with first-order SC absorption and linear central clearance in Gottingen minipigs (Table 1 population mean parameter estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb7 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment population PK model with first-order SC absorption plus parallel linear and saturable (Michaelis-Menten) central elimination for the non-disclosed humanized IgG antibody mAb7 (Zheng 2012; pI 8.9) in Gottingen minipigs (Table 1 population mean parameter estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zheng_2012_minipig_mab.html">MAb8 (Zheng 2012)</a> </td> <td style="text-align:left;"> Preclinical (Gottingen minipig). Two-compartment IV-only population PK model for the non-disclosed humanized IgG antibody mAb8 (Zheng 2012; pI 8.7) with linear central clearance in Gottingen minipigs (Table 1 population mean parameter estimates). No SC arm was run for mAb8, so no Ka or F is defined. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Easterling_2018_magnesium_sulfate.html">Magnesium sulfate (Easterling 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model of magnesium sulfate (MgSO4-7H2O) with intravenous administration and an endogenous baseline magnesium term added to the administered drug, in pregnant women with severe preeclampsia comparing continuous IV infusion vs serial IV bolus dosing (Easterling 2018). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Salinger_2013_magnesiumSulfate.html">MagnesiumSulfate (Salinger 2013)</a> </td> <td style="text-align:left;"> One-compartment population PK model of magnesium sulphate (MgSO4-7H2O) with first-order intramuscular absorption, IV dosing into the central compartment, and an endogenous baseline magnesium term added to the administered drug, in pregnant women with pre-eclampsia (Salinger 2013). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chan_2008_maraviroc.html">Maraviroc (Chan 2008)</a> </td> <td style="text-align:left;"> Two-compartment population PK meta-analysis model for oral maraviroc (CCR5 antagonist) in healthy volunteers and asymptomatic HIV-infected adults, with hepatic-extraction-ratio parameterisation of clearance, dose-dependent absorption (sigmoid-Emax F_ABS and power-function ka), food effect on both, Asian-race covariates on hepatic extraction / peripheral volume / inter-compartmental clearance, an age effect on Q, and a TAD-dependent residual error (Chan 2008) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Davis_2008_maraviroc.html">Maraviroc (Davis 2008)</a> </td> <td style="text-align:left;"> Concentration-QT mixed-effects regression model relating single-dose oral maraviroc plasma concentrations to individual heart-rate-corrected QT intervals in healthy adult male and female volunteers (Davis 2008). No structural pharmacokinetic component is fit: the model is the one-stage NONMEM mixed-effects regression of observed QT on observed RR interval and observed maraviroc plasma concentration (Cp), with a fractional female-sex multiplier on the population QT intercept, a population QT/RR correction-factor exponent (Fridericia-style), and a linear concentration-QT slope. The single-dose population slope estimate (0.970 us mL/ng, 95% CI -0.571 to 2.48) was not significantly different from zero across the studied concentration range up to 2363 ng/mL. For simulation, supply observed or simulated maraviroc plasma concentration as the time-varying covariate CP_MVC_NGML and the RR interval as RR. Interoccasion variability on the QT intercept reported by the paper (13.4 ms^2) is not encoded (Hong_2015_moxifloxacin precedent: nlmixr2lib has no idiomatic IOV encoding for distributed models). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Rosario_2008_maraviroc.html">Maraviroc (Rosario 2008)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and lag time for maraviroc (CCR5 antagonist) coupled with a direct Emax CCR5 receptor occupancy model in healthy adults and HIV-1-positive patients (Rosario 2008). PK is parameterised with dose-dependent relative bioavailability F1 and dose-dependent elimination rate constant K across six dose groups (3, 10, 25, 100 (reference), 300 mg b.i.d. and 600 mg q.d.); receptor occupancy on CD4 T cells is modelled as Occ = E0 + Emax * Cp / (KD + Cp) with a background binding baseline. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Weatherley_2009_maraviroc_iv.html">Maraviroc iv (Weatherley 2009)</a> </td> <td style="text-align:left;"> Four-compartment IV maraviroc population PK in 20 healthy young adult males receiving 3, 10, or 30 mg as a 1-hour IV infusion (Weatherley & McFadyen 2009 Br J Clin Pharmacol). NONMEM ADVAN7 + FOCE-I fit to log- transformed plasma concentrations from study A4001009 only. Exponential inter-subject variability on CL, V1, V2, Q3 and V3; proportional residual error (additive on the log scale). No dose effect on clearance over the 3-30 mg range. This file extracts only the IV disposition analysis (paper Analysis 1, Table 3); the paper's two companion analyses (sigmoid Emax NAUC meta-regression of oral phase 1 data, and the S-PLUS closed-form mass balance model) are not ODE PK models and are described in the vignette but not implemented here. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sun_2023_maribavir.html">Maribavir (Sun 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral maribavir with first-order absorption, an absorption lag time, and dose-dependent absorption rate in adult transplant recipients with cytomegalovirus infection/disease (Sun 2023) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kuester_2008_matuzumab.html">Matuzumab (Kuester 2008)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for matuzumab (humanised anti-EGFR IgG1 monoclonal antibody) in adults with advanced carcinoma (Kuester 2008), with parallel first-order linear and Michaelis-Menten elimination from the central compartment; body weight on linear CL and central volume. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Stein_2018_mAb_nonlinear_PK.html">Mavrilimumab (Stein 2018)</a> </td> <td style="text-align:left;"> Two-compartment QSS TMDD typical-value fit for mavrilimumab (anti-GM-CSF receptor mAb) used to illustrate the critical concentration (Ccrit) for nonlinear PK (Stein and Peletier 2018 Table 1) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Xu_2023_MBG453.html">MBG453 (Xu 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for sabatolimab (MBG453, anti-TIM-3 IgG4) with parallel linear and Michaelis-Menten elimination from the central compartment, fit to pooled adult patients with advanced solid tumors and hematologic malignancies (Xu 2023). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Koumaki_2023_mecillinam.html">Mecillinam (Koumaki 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous mecillinam (amdinocillin) in healthy adults, used to determine systemic MIC breakpoints against Enterobacterales for intermittent, extended, and continuous infusion regimens (Koumaki 2023). CL, Q, Vc and Vp are parameterised per kilogram of body weight (linear exponent 1, not allometric). The unbound concentration Cu = fu * Cc (fu = 0.9, i.e. 10 percent protein binding) is the driver of the PK/PD target fT>MIC >= 40 percent of the dosing interval used for probability of target attainment. Parameters were obtained by MCMC (Stan) reanalysis of published AGGREGATE concentration mean and SD data, not individual-level data; no residual error was estimated. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cao_2013_MEDI528.html">MEDI528 (Cao 2013)</a> </td> <td style="text-align:left;"> Second-generation minimal physiologically-based PK (mPBPK) model for MEDI-528 in adults (Cao 2013 Model A; clearance from plasma) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hood_2021_medi7836.html">Medi7836 (Hood 2021)</a> </td> <td style="text-align:left;"> Population PK-PD binding model for MEDI7836 (anti-IL13 IgG1 lambda-YTE mAb) in healthy adult males (Hood 2021): two-compartment SC PK with first-order absorption, ADA-on-CL covariate, plus IL13 turnover, fixed Kon/Koff binding to MEDI7836:IL13 complex, complex distribution sharing CL/Q/V3 with parent drug, and a serum PD observation modelled as the molar sum of free IL13 and a small fraction of complex. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hoglund_2018_mefloquine.html">Mefloquine (Hoglund 2018)</a> </td> <td style="text-align:left;"> Population PK model for oral mefloquine in Burmese adults with uncomplicated Plasmodium falciparum malaria treated with the standard 3-day artesunate-mefloquine combination (Hoglund 2018). One-transit-compartment absorption with ka = ktr feeds a two-compartment disposition model. No covariates were retained in the final model: body-weight allometric scaling (fixed exponents 0.75 / 1.0), sex, admission parasitaemia, and validated molecular markers of mefloquine and artemisinin resistance (pfmdr1, pfcrt, atp6, pfk13) were tested in a step-wise covariate search but did not significantly improve the model. Relative bioavailability F is implicitly 1 (the paper tested adding an estimated F with IIV and excluded it from the final model). NONMEM additive residual error on the log-transformed observation is encoded here as a proportional residual in the linear concentration space. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ramharter_2019_mefloquine.html">Mefloquine (Ramharter 2019)</a> </td> <td style="text-align:left;"> Population PK model for the two mefloquine enantiomers and their carboxy-metabolite carboxymefloquine (CMQ) in pregnant African women receiving intermittent preventive treatment for malaria (Ramharter 2019, MIPPAD trial, Gabon). Each parent enantiomer ((+)-mefloquine = '_r' = (11R, 2'S); (-)-mefloquine = '_s' = (11S, 2'R)) follows a two-compartment disposition with first-order oral absorption. Carboxymefloquine ('_cmq') is formed molar 1:1 from both parents via the apparent parent clearance and follows a two-compartment disposition; its first-order clearance is autoinduced by CMQ plasma concentration via a two-stage RNA + enzyme-pool turnover model (precursor1 = enzymatic-RNA precursor pool, precursor2 = metabolizing enzyme pool; both kdeg-driven, both at unit steady state in the absence of CMQ; CMQ clearance scales linearly with precursor2). A shared body-weight allometric exponent acts on the central volume of each parent enantiomer (reference 55 kg). The split-dose IPTp regimen (REGIMEN_SPLIT = 1: 7.5 mg/kg on two consecutive days) carries a small (+5%) bioavailability increment relative to the single-dose regimen (REGIMEN_SPLIT = 0: 15 mg/kg on day 1). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Simpson_2013_mefloquine.html">Mefloquine (Simpson 2013)</a> </td> <td style="text-align:left;"> In vitro (P. falciparum). Sigmoid Emax inhibition model of mefloquine effect on hypoxanthine uptake by clinical Plasmodium falciparum isolates from the Thai-Myanmar border (Shoklo Malaria Research Unit, 1993-2005), with pfmdr1 genotype covariate effects on EC50. The 'subject' in the NLME framework is a parasite isolate (n=460 isolates with mefloquine data). STIM_MEFLOQUINE_NM is the per-well drug concentration in the in vitro hypoxanthine-uptake-inhibition assay; the model has no PK and no time evolution. E0 and Emax are fixed per Simpson 2013 Table 3 footnote. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nath_2007_melphalan.html">Melphalan (Nath 2007)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for melphalan in paediatric blood or marrow transplant recipients (Nath 2007). Structural CL is a linear additive function of body weight, prior-carboplatin therapy, and 99mTc-DTPA-tracer-measured GFR; central volume Vc is a linear additive function of body weight; intercompartmental rate constants k12 and k21 are estimated directly (not as Q/Vc and Q/Vp). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nath_2010_melphalan.html">Melphalan total (Nath 2010)</a> </td> <td style="text-align:left;"> Two-compartment IV-infusion population PK model for total plasma melphalan in adults with multiple myeloma undergoing high-dose therapy and autologous stem-cell transplant (Nath 2010); additive non-renal + renal CL with hematocrit, fat-free mass, and creatinine-clearance covariates. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nath_2010_melphalan.html">Melphalan unbound (Nath 2010)</a> </td> <td style="text-align:left;"> Two-compartment IV-infusion population PK model for unbound (ultrafiltrate) plasma melphalan in adults with multiple myeloma undergoing high-dose therapy and autologous stem-cell transplant (Nath 2010); additive non-renal + renal CL with hematocrit, fat-free mass, and creatinine-clearance covariates. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cao_2013_mepolizumab.html">Mepolizumab (Cao 2013)</a> </td> <td style="text-align:left;"> Second-generation minimal physiologically-based PK (mPBPK) model for mepolizumab in adults (Cao 2013 Model A; clearance from plasma) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gebhard_2023_leukemia_maintenance_therapy.html">Mercaptopurine (Gebhard 2023)</a> </td> <td style="text-align:left;"> Oral 6-mercaptopurine (6MP) population PK during maintenance therapy for childhood acute lymphoblastic leukemia, predicting the erythrocyte 6-thioguanine-nucleotide concentration E-TGN (Gebhard 2023, model PK^6MP_{mm,pop}). The plasma layer is a FIXED one-compartment model with first-order absorption and a fixed bioavailability of 0.12, all taken from the literature (Supplementary Table S2). A third compartment carries 6-thioguanine nucleotides -- the active metabolites of 6MP -- inside red blood cells, filled by SATURABLE Michaelis-Menten influx driven by the plasma 6MP concentration and drained by a first-order efflux rate constant. V_mm and K_eff were estimated with IIV on 452 children with 4624 E-TGN observations (Table 3); K_mm was estimated as a population-only parameter with no IIV, which is what the 'pop' subscript of the model name denotes. Because every patient's record begins mid-maintenance-therapy, the red-cell compartment is initialised at the patient's first observed E-TGN value (covariate BL_TGN_RBC) rather than at zero. Volumes are per m^2 body-surface area and doses are supplied in umol/m^2, so BSA cancels and no BSA covariate is needed. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hawwa_2008_mercaptopurine.html">Mercaptopurine (Hawwa 2008)</a> </td> <td style="text-align:left;"> Population PK / pharmacogenetic model for oral 6-mercaptopurine (6-MP) and its two active intracellular metabolites 6-thioguanine nucleotides (6-TGNs) and 6-methylmercaptopurine nucleotides (6-mMPNs) measured in erythrocytes (RBCs) of 19 paediatric patients (n = 75 samples; 150 concentrations) with acute lymphoblastic leukaemia receiving maintenance chemotherapy at a target oral dose of 75 mg/m^2/day. The structural model is a one-compartment first-order absorption + first- order elimination model for 6-MP whose plasma concentration is NOT observed; transformation of bioavailable 6-MP into the two RBC metabolites occurs at a common metabolic rate constant kme = 0.78 * k20 (78% of total 6-MP elimination), with the fractional split between metabolites governed by FM3 (the fractional metabolic transformation of 6-MP into 6-TGNs; the complementary fraction 1 - FM3 goes to 6-mMPNs). The fixed structural anchors ka = 1.3 1/h, F = 0.22, k20 = 0.53 1/h, and the kme : k20 ratio of 0.78 are taken from the prior 6-MP literature (Hawwa 2008 cites Zimm 1983 and Lennard 1990; values listed in Methods page 4 of the British Journal of Clinical Pharmacology article). The apparent distribution volume of 6-MP central and of each metabolite compartment is not identifiable from the RBC sampling design and is fixed to 1 L by the ADVAN6 implementation convention used by Hawwa 2008 (analogous to Urien 2005 capecitabine; see vignette Errata for the dimensional-analysis discussion). The only estimated structural parameters are FM3, CL_6TGNs, and CL_6mMPNs; the only retained covariates are TPMT genotype (any TPMT*3* mutation, pooled binary) on FM3 (theta = 2.56) and body surface area (m^2) on the apparent clearance of 6-TGNs as a power-law (theta = 1.16, anchored to BSA = 1 m^2). IIV is estimated only on the two metabolite clearances; no IIV on FM3 was estimable. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gebhard_2023_leukemia_maintenance_therapy.html">Mercaptopurine anc (Gebhard 2023)</a> </td> <td style="text-align:left;"> Joint PK/PD model of oral 6-mercaptopurine (6MP) myelosuppression during maintenance therapy for childhood acute lymphoblastic leukemia (Gebhard 2023, final model PKPD^6MP_{lin,mm}). This is the paper's headline model. A FIXED one-compartment plasma layer with first-order absorption feeds a red-cell compartment holding 6-thioguanine nucleotides (E-TGN) via SATURABLE Michaelis-Menten influx; the red-cell concentration then drives a linear effect function Edrug = slope * E-TGN that suppresses the proliferation rate of a Friberg-style myelosuppression chain (proliferating pool, three transit compartments, circulating neutrophils) with power-law rebound feedback (base / ANC)^gamma. Because every patient's record begins mid-maintenance-therapy rather than at drug-free baseline, the red-cell compartment is initialised at the patient's first observed E-TGN (covariate BL_TGN_RBC) and the Friberg chain is initialised at a treatment steady state scaled by an estimated fraction-of-baseline parameter inieff. Estimated on 452 children with 4624 E-TGN and 9808 ANC observations (Table 5, whole-data-set column). Two outputs: E-TGN (umol/L) and ANC (G/L). Volumes are per m^2 body-surface area and doses are supplied in umol/m^2, so BSA cancels and no BSA covariate is needed. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bergen_2017_meropenem.html">Meropenem (Bergen 2017)</a> </td> <td style="text-align:left;"> In vitro (hollow-fiber infection model). Mechanism-based PK/PD (life-cycle growth) model of meropenem bacterial killing and resistance against Pseudomonas aeruginosa 1280 (meropenem MIC 0.25 mg/L) across simulated critically ill patient renal-function profiles (augmented renal clearance, normal, and impaired). The bacterial population is split into three pre-existing subpopulations of decreasing meropenem susceptibility (susceptible, intermediate, resistant), each described by two states (state 1 preparing for replication, state 2 immediately before replication; six bacterial compartments total). Meropenem acts via inhibition of successful bacterial replication (a Hill-type Inh_Rep function per subpopulation; no direct killing term). The intermediate and resistant subpopulations have higher IC50_Rep and steeper or shallower Hill coefficients than the susceptible subpopulation; the susceptible subpopulation has Imax_Rep and Hill fixed to 1. Meropenem disposition in the HFIM is a fixed-half-life first-order decline parameterised from the upstream popPK model (Mattioli 2016, reference 20 in the source paper); the default half-life is 1.1 h (normal renal function); 0.6 h (augmented renal clearance) and 4.0 h (impaired renal function) are obtained by overriding thalf_mem at simulation time. No patient covariates and no random effects: this is the typical-value MBM fit (Bergen 2017 Table 3) to the simultaneous P. aeruginosa 1280 HFIM data across the three renal-function scenarios and four dosing regimens (2, 1, or 0.5 g q8h plus 1 g q12h for impaired). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fukumoto_2023_meropenem.html">Meropenem (Fukumoto 2023)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for meropenem in 31 adults with sepsis in a Japanese emergency center / ICU (Fukumoto 2023). Clearance scales as a power function of the BSA-normalized MEASURED creatinine clearance obtained from an 8-hour timed urine collection, CL = 13.6 * (CRCL / 87.6)^0.66 L/h, centered on the cohort-median 87.6 mL/min/1.73 m^2. The measured CCr was the single retained covariate and outperformed Cockcroft-Gault CCr and eGFR, which underestimate renal function in septic patients with normal or augmented renal clearance. Interindividual variability was estimated on CL and V1 only; the random effects on Q and V2 were fixed at zero. Residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Germovsek_2018_meropenem.html">Meropenem (Germovsek 2018)</a> </td> <td style="text-align:left;"> One-compartment plasma + CSF (two-state) IV population PK model for meropenem in neonates and young infants (<=90 days) with late-onset sepsis and/or meningitis (Germovsek 2018; NeoMero-1 and NeoMero-2 studies). Plasma CL and Vc are allometrically scaled to body weight (fixed exponent 0.632 on CL, 1.0 on Vc) with a fixed Rhodin-style postmenstrual-age maturation Hill function on CL and a power covariate of (CREAT_REF / CREAT) on CL; an additional CSF compartment with fixed Vcsf = 0.15 L/70 kg and estimated inter-compartmental clearance CL_CSF carries a logit-scale CSF penetration fraction (typical 8.4 %) modulated by CSF total protein concentration. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hanberg_2018_meropenem.html">Meropenem (Hanberg 2018)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for meropenem in critically ill adults receiving venovenous or venoarterial extracorporeal membrane oxygenation (ECMO) treatment, with simultaneous fitting of plasma concentrations (central compartment Ac/Vc) and free subcutaneous adipose-tissue (SCT) concentrations sampled by microdialysis (peripheral compartment Ap/Vp scaled by an estimated fraction unbound in tissue f_u,tissue = 0.79). Elimination clearance is a direct linear function of the patient's estimated creatinine clearance (eCLCr, Cockcroft-Gault, raw mL/min) via CL_i = CLfrac * eCLCr_i with CLfrac = 0.0460 L/h per (mL/min); 9 of 10 patients were also on continuous renal replacement therapy so eCLCr partly reflects the CRRT contribution (Hanberg 2018). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kim_2024_meropenem.html">Meropenem (Kim 2024)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model for meropenem in healthy Korean adults with normal renal function (Kim 2024; n = 12, 84 plasma samples after a single 500 mg 30-min IV infusion). Total clearance depends on serum creatinine through a power model centred at the cohort median 0.86 mg/dL: CL = 12.4 * (CREAT/0.86)^-0.392 L/h, so CL falls from 14.3 L/h at CREAT = 0.6 mg/dL to 11.7 L/h at CREAT = 1.0 mg/dL. Because the estimated correlation between the random effects on CL and Vc was 0.99, the paper codes a SINGLE eta shared by both parameters, with the Vc random effect constructed as 1.53 * eta_CL (Table 2 'V1' row under interindividual variability). Interindividual variability on Q and Vp is fixed, not estimated. Residual error is proportional. The unbound concentration Cu = fu * Cc (fu = 0.98) is the driver of the fT>MIC PK/PD targets (40% fT>MIC, 40% fT>4MIC, 100% fT>MIC, 100% fT>4MIC) used in the paper's Monte Carlo probability-of-target-attainment simulations for intermittent, 3-h extended, and continuous infusion regimens. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Padari_2012_meropenem.html">Meropenem (Padari 2012)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for meropenem in very-low-birth-weight neonates (gestational age <=32 weeks, birth weight <1,500 g; n=19; Padari 2012). Vss scales linearly with current body weight; CL follows the Rhodin (2009) fixed renal-maturation function (allometric exponent 0.75 on CL, Hill-type postmenstrual-age maturation with TM50 = 47.7 weeks and Hill = 3.4); serum creatinine, postnatal age, and gestational age were screened and did not improve fit and are not retained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shekar_2014_meropenem.html">Meropenem (Shekar 2014)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for meropenem in critically ill adult patients on extracorporeal membrane oxygenation (ECMO) and historical critically ill control patients with sepsis, with a piecewise covariate on clearance that switches between a fixed RRT-cohort CL and a Cockcroft-Gault-CrCL-driven non-RRT CL (Shekar 2014) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ulldemolins_2015_meropenem.html">Meropenem (Ulldemolins 2015)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for meropenem in 30 critically ill adults with septic shock and continuous renal replacement therapy (Ulldemolins 2015). Clearance is the sum of a constant CRRT-mediated baseline (3.68 L/h at zero residual diuresis) and an additive linear contribution from 24-hour residual diuresis (0.22 L/h per 100 mL/24h); central volume scales with body weight by power exponent 2.07 around the population-median 73 kg. CRRT intensity, blood flow, filter type, and serum albumin were tested but not retained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wittau_2015_meropenem.html">Meropenem (Wittau 2015)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model for meropenem in morbidly obese adults (Wittau 2015). Allometric scaling on fat-free mass with a reference FFM of 53 kg. Unbound meropenem concentrations in subcutaneous tissue and peritoneal fluid are described as the plasma concentration multiplied by the site-to-plasma AUC ratios FSC and FPF (the final model assumed very rapid equilibration with plasma, so SC and PF are not carried as separate ODE states). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Rees_2018_meropenem_ciprofloxacin.html">Meropenem ciprofloxacin (Rees 2018)</a> </td> <td style="text-align:left;"> In vitro (hollow-fiber infection model). Mechanism-based PK/PD (life-cycle growth) model of bacterial killing and resistance for meropenem plus ciprofloxacin against hypermutable Pseudomonas aeruginosa CW44, with three pre-existing subpopulations and subpopulation plus mechanistic synergy </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2024_meropenem_pbpk.html">Meropenem pbpk (Yang 2024)</a> </td> <td style="text-align:left;"> PBPK-derived reduced one-compartment model for meropenem in critically ill (severe pneumonia with or without sepsis) adult Asian ICU patients. The source paper built an 18-tissue whole-body PBPK model in PK-Sim v11.2 for healthy adults and extrapolated it to critically ill patients by scaling albumin, alpha-1-acid glycoprotein, hematocrit and GFR; that whole-body structure is a PK-Sim platform model whose ODEs, organ volumes, blood flows and partition coefficients are NOT written out in the publication and are therefore not reproduced here. What IS fully reported, and is what this file encodes, is the reduced disposition model the authors carried out of the PBPK and into their Monte Carlo target-attainment analysis: Vd 23.21 L, CL 12.07 L/h and unbound fraction 0.98 for Asian patients with severe infection (Results, Monte Carlo simulations). Intravenous infusion dosing; linear elimination; no covariates. The PK/PD index is the fraction of the dosing interval during which the free concentration exceeds the MIC (f%T>MIC), with 40%fT>MIC as the primary target and 100%fT>MIC as a stricter alternative; f%T>MIC is computed from the simulated profile rather than integrated as a model state, following the Minichmayr_2024_ceftaroline precedent. Interindividual variability was assumed lognormal on Vd and CL in the Monte Carlo but its magnitude is not reported anywhere in the paper or its supplement, so both etas are encoded as fixed(0); no residual error model was reported. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LopezArguello_2023_pbp_binding_pseudomonas.html">Meropenem qsp (LopezArguello 2023)</a> </td> <td style="text-align:left;"> QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for meropenem (carbapenem): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2018_meropenem_tobramycin.html">Meropenem tobramycin PAO1 (Landersdorfer 2018)</a> </td> <td style="text-align:left;"> In vitro (Pseudomonas aeruginosa PAO1 wild-type). Mechanism-based PK/PD (life-cycle growth) model of bacterial killing and resistance for meropenem plus tobramycin, with three pre-existing subpopulations (susceptible, MEM-resistant/TOB-intermediate, MEM-intermediate/TOB-resistant) and mechanistic synergy via tobramycin-induced outer-membrane permeabilisation of meropenem </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2018_meropenem_tobramycin.html">Meropenem tobramycin PAOmutS (Landersdorfer 2018)</a> </td> <td style="text-align:left;"> In vitro (Pseudomonas aeruginosa PAOdelta-mutS hypermutable strain). Mechanism-based PK/PD (life-cycle growth) model of bacterial killing and resistance for meropenem plus tobramycin, with three pre-existing subpopulations (susceptible, MEM-resistant/TOB-intermediate, MEM-intermediate/TOB-resistant) and mechanistic synergy via tobramycin-induced outer-membrane permeabilisation of meropenem </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chae_2012_metformin.html">Metformin (Chae 2012)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for oral metformin in healthy Korean adults, coupled to a three-transit Sun-Jusko signal-transduction PD model for the antihyperglycaemic effect (Chae 2012). Plasma drug concentration in the central compartment drives a Hill-type stimulation function DR = Emax * Cp^r / (EC50^r + Cp^r) that initiates a cascade of three secondary-messenger transit compartments (M1 -> M2 -> M3) with shared mean transit time tau. The third messenger M3 is the measured percent change in plasma glucose from baseline relative to a sugar-bolus control arm. Creatinine clearance enters CL/F as a power covariate with reference 106.5 mL/min and exponent 0.782. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Choi_2018_metformin.html">Metformin (Choi 2018)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral metformin in 36 healthy adult Korean men from a phase I single-dose 2-way crossover bioequivalence study comparing a single-agent metformin tablet against a metformin-containing fixed-dose combination (FDC) tablet (Choi 2018). The absorption process is parallel mixed-input: fraction F1 of the dose is absorbed first-order from the depot compartment (rate Ka), and fraction (1-F1) is absorbed zero-order directly into the central compartment over duration D2 with lag time ALAG2. Formulation enters as a binary covariate (FORM_FDC) with multiplicative power-style effects on Ka (Ka_FDC = 0.83 * Ka_single-agent) and on relative bioavailability F (F_FDC = 0.94 * F_single-agent = 0.94). IIV on CL/F, Vc/F (correlated, rho 0.225), and Ka; proportional residual error only. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vanRongen_2018_metformin.html">Metformin (vanRongen 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral metformin in 22 overweight and obese Caucasian adolescents (van Rongen 2018). First-order absorption into a single central compartment with apparent oral clearance (CL/F) and apparent oral volume of distribution (V/F). Total body weight (TBW) enters linearly on CL/F with reference 75.8 kg (study median): CL/F = 1.17 * (1 + 0.0138 * (TBW - 75.8)) L/min. Proportional residual error; IIV on CL/F, V/F, and ka. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yoon_2013_metformin.html">Metformin (Yoon 2013)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with first-order oral absorption and an absorption-lag time for a single 500 mg oral dose of metformin in healthy Korean male adults (Yoon 2013). Body weight enters V/F as a linear-in-deviation covariate; two organic-cation-transporter (OCT) genetic polymorphisms enter CL/F as multiplicative fractional shifts, OCT2 c.808G>T (SLC22A2, A270S, widely known as rs316019) and OCTN1 c.917C>T (SLC22A4, T306I). Variant carriers are pooled (heterozygotes and homozygous variants together vs the homozygous wild-type reference) per the source paper's dominant-model encoding. Slow first-order absorption (ka = 0.248 1/h) relative to elimination (kel = cl/vc = 1.21 1/h) produces the apparent flip-flop kinetics reported for metformin. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gebhard_2023_leukemia_maintenance_therapy.html">Methotrexate (Gebhard 2023)</a> </td> <td style="text-align:left;"> Low-dose oral methotrexate (MTX) population PK during maintenance therapy for childhood acute lymphoblastic leukemia, predicting the erythrocyte methotrexate concentration E-MTX (Gebhard 2023, model PK^MTX_{fix,bio,lin,peri}). The plasma layer is a FIXED two-compartment model with first-order absorption whose parameters were taken from the literature (Supplementary Table S1) and a saturable dose-dependent bioavailability F = 1 - 0.77 * D / (15.01 + D) with D the MTX dose in mg/m^2. A fourth compartment carries methotrexate (predominantly methotrexate polyglutamates) inside red blood cells, filled by LINEAR influx driven by the PERIPHERAL plasma concentration and drained by a first-order efflux rate constant; both red-cell rate constants were estimated on 452 children with 4192 E-MTX observations (Table 2) and carry IIV. Because every patient's record begins mid-maintenance-therapy, the red-cell compartment is initialised at the patient's first observed E-MTX value (covariate BL_MTX_RBC) rather than at zero. Volumes are per m^2 body-surface area and doses are supplied in umol/m^2, so BSA cancels and no BSA covariate is needed. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Joerger_2006_methotrexate.html">Methotrexate (Joerger 2006)</a> </td> <td style="text-align:left;"> Population PK model for methotrexate (MTX) and its principal circulating metabolite 7-hydroxy-methotrexate (7-OH-MTX) in adult cancer patients receiving high-dose intravenous MTX therapy (Joerger 2006). Joint parent + metabolite model: linear 3-compartment MTX (central + two peripheral compartments) with first-order elimination from the central compartment, feeding a linear 2-compartment 7-OH-MTX disposition through a fixed metabolic fraction of 10 percent. Additive-linear covariate effects of baseline creatinine clearance (Cockcroft-Gault, raw mL/min, truncated at 140), concurrent benzimidazole-class proton-pump-inhibitor comedication, and prior NSAID administration on both MTX and 7-OH-MTX total clearance. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Olivo_2024_methotrexate.html">Methotrexate (Olivo 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for high-dose intravenous methotrexate (12 g/m^2 over a 4 h infusion) in Brazilian paediatric patients with osteosarcoma treated on the Brazilian Osteosarcoma Treatment Group (BOTG) protocol (Olivo 2024; n = 32 patients, 216 cycles, 563 therapeutic-drug-monitoring concentrations). Linear first-order elimination from the central compartment. Serum creatinine scales clearance as a power function normalized to the cohort median 0.58 mg/dL (exponent -0.192), and body surface area scales the central volume as a power function normalized to the cohort median 1.45 m^2 (exponent 0.301). Correlated exponential between-subject variability on CL and Vc (correlation 94%), exponential between-occasion variability on CL shared across MTX cycles, and a proportional residual error. Built to anticipate leucovorin (folinic acid) rescue dose adjustments before the first monitored MTX concentration. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ruhs_2012_methotrexate.html">Methotrexate (Ruhs 2012)</a> </td> <td style="text-align:left;"> PK/PD model of methotrexate (MTX) and homocysteine (HCY) after high-dose MTX treatment in children with acute lymphoblastic leukemia (Ruhs 2012). Two-compartment IV PK for MTX with linear BSA scaling on CL, V1, Q, V2 (theta values reported per m^2 BSA in the paper) and a power effect of the age- and gender-adjusted serum creatinine ratio (CREAT_REF / CREAT) on CL (Eq. 1); coupled to a single-compartment indirect response model for HCY where MTX inhibits the HCY elimination rate kout via an inverse Emax function (Emax fixed to 1) and the typical HCY baseline depends linearly on age. IOV reported in the paper on CL (17.15% CV) and HCYBL (23.83% CV) across the window and four consolidation HDMTX administrations is not encoded in the model file (only between-subject IIV is carried). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Taylor_2020_methotrexate.html">Methotrexate (Taylor 2020)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for intravenous high-dose methotrexate (5 or 8 g/m^2 over 24 h IV infusion) in pediatric NOPHO ALL2000 / ALL2008 patients with acute lymphoblastic leukemia; BSA-normalized PK parameters (reference 1.73 m^2) and a time-varying serum creatinine power effect on clearance (reference 29 umol/L) implemented as the default population PK model behind the MTXPK.org clinical decision support tool (Taylor 2020) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Saleh_2023_leicns_pk30_mouse_brain.html">Methotrexate mouse pbpk (Saleh 2023)</a> </td> <td style="text-align:left;"> PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, CD1 nude). Nine-compartment CNS PBPK model predicting unbound methotrexate concentrations in brain extracellular fluid (brain ECF) after a single 1000 mg/kg intravenous dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug's physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.195 (estimated by the authors; Table V). The plasma PK model is the empirical two-compartment model of Table IV (in-house NONMEM fit) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Billard_1995_methoxsalen.html">Methoxsalen blood (Billard 1995)</a> </td> <td style="text-align:left;"> Three-compartment intravenous population PK model for 8-methoxypsoralen (8-MOP, methoxsalen) whole blood concentrations in healthy adult volunteers receiving 5/10/15 mg over 60 min (Billard 1995) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Billard_1995_methoxsalen.html">Methoxsalen plasma (Billard 1995)</a> </td> <td style="text-align:left;"> Three-compartment intravenous population PK model for 8-methoxypsoralen (8-MOP, methoxsalen) plasma concentrations in healthy adult volunteers receiving 5/10/15 mg over 60 min (Billard 1995) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Teuscher_2015_methylphenidate.html">Methylphenidate (Teuscher 2015)</a> </td> <td style="text-align:left;"> Pediatric population PK model for methylphenidate hydrochloride extended-release multilayer beads (MPH-MLR, Aptensio XR) after a single oral dose, parameterized as a two-input, one-compartment, first-order-elimination structure: a fast-release (IR) depot delivers a fraction F1 of dose with first-order absorption rate Ka1, a slow-release (ER) depot delivers the remaining 1 - F1 with first-order rate Ka2 after an absorption lag tlag, and the central compartment eliminates linearly via clearance CL and apparent volume V. Body weight enters CL via a power covariate CL = CL_TV * WT^theta (Eq 4). Between-individual variability is retained on CL and V; IIV on Ka1, Ka2, F1, and tlag was not in the final pediatric fit (Table 1). The companion exposure-response analysis maps simulated Cmax to change-from-baseline ADHD-RS-IV total score via the Emax model E = Emax * Cmax / (EC50 + Cmax) with Emax = -34.96 and EC50 = 5.77 ng/mL (Table 2); the PD step lives outside the ODE system because the published mapping uses a per-period Cmax, not the instantaneous central concentration. The vignette reproduces the full PK simulation, NCA, and Cmax-to-ADHD-RS-IV exposure-response. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Eugene_2016_metoprolol.html">Metoprolol (Eugene 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral metoprolol tartrate with first-order absorption and lag time in elderly inpatients with multiple comorbidities; sex as the only covariate on apparent clearance (Eugene 2016). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/CohenWolkowiez_2012_metronidazole.html">Metronidazole (CohenWolkowiez 2012)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for metronidazole in preterm infants (Cohen-Wolkowiez 2012). Clearance scales linearly with body weight (reference 1.5 kg) and as a power function of postmenstrual age (reference 32 weeks); central volume scales linearly with body weight. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Suyagh_2011_metronidazole.html">Metronidazole (Suyagh 2011)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous metronidazole in 32 preterm neonates receiving treatment of or prophylaxis against necrotising enterocolitis, with dried-blood-spot HPLC sampling (Suyagh 2011). Clearance is described by an allometric 3/4-power scaling on body weight (reference 1.0 kg) and a linear postmenstrual- age maturation term centred at 30 weeks; volume of distribution is proportional to body weight. The publication is open access only at the abstract level, so inter-individual variability on CL and V and the residual-error magnitude are FIXED at 0 here; users running stochastic VPCs must supply their own variability terms (see the validation vignette's Errata section). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Clinckers_2008_MHD_rat.html">MHD rat (Clinckers 2008)</a> </td> <td style="text-align:left;"> Preclinical (rat). Population PK model for 10,11-dihydro-10-hydroxy- carbamazepine (MHD), the active metabolite of oxcarbazepine, in male Wistar rat plasma and hippocampal extracellular fluid (Clinckers 2008). One-compartment central disposition (V2) with combined zero-order (fraction F1 of dose over duration D2) and lagged first-order (1 - F1, ka with lag ALAG1) absorption after intraperitoneal bolus, coupled to a biophase / effect compartment (V3) reached via inter-compartmental rate constants k23 and k32. Acute focal pilocarpine-induced seizure activity and local intrahippocampal verapamil (efflux-transporter blockade) each shrink the biophase volume (V3a -> V3b under seizure; V3a -> V3c under verapamil); plasma kinetics are unaffected. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zou_2012_MI_219.html">MI 219 (Zou 2012)</a> </td> <td style="text-align:left;"> Predicted-human two-compartment IV PK model for MI-219 (a small-molecule HDM2/p53 inhibitor) in adults, with parameters projected from NONMEM-based interspecies allometric scaling of single-dose IV plasma profiles in rats (5 mg/kg), beagle dogs (2 mg/kg), and cynomolgus monkeys (10 mg/kg). Linear elimination from the central compartment; mouse data were excluded from the joint NONMEM fit because the mouse profile was not superimposable on the other species under Wajima / Dedrick normalisation. The model file encodes the predicted human typical values at a 70 kg reference body weight (Zou 2012 Table 5, NONMEM column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Leroux_2018_fluconazole_micafungin.html">Micafungin (Leroux 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model of intravenous micafungin in preterm and term neonates with suspected or proven systemic candidiasis (Leroux 2018), with linear current-weight scaling of CL and V. Typical-value structural model only: the source paper and Data S1 supplement (goodness-of-fit plots only) do not report inter-individual variability magnitudes, residual error structure, or the functional form / coefficient of the corrected-gestational-age effect on CL that the paper mentions, so IIV and RUV are encoded as fixed(0) and the CGA covariate is omitted. See vignette Errata. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Martial_2017_micafungin.html">Micafungin (Martial 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for IV micafungin in adult intensive-care-unit patients with suspected or proven fungal infection (Martial 2017). Body-weight allometric scaling (fixed exponents 0.75 on CL and Q, 1 on V1 and V2; 70 kg reference), log-normal IIV on CL and V1 (encoded as diagonal; the source reports a qualitative non-zero correlation but no numerical covariance), and a proportional residual error. No covariates were retained in the final model (only weight via the a-priori allometric structure). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Beredaki_2023_micafungin.html">Micafungin clsi (Beredaki 2023)</a> </td> <td style="text-align:left;"> In vitro (Candida albicans clinical isolates in RPMI-1640 with 10% pooled human serum; two-compartment closed dialysis/diffusion PK/PD model). Micafungin in vitro PK/PD model with the exposure-response relationship indexed on CLSI M27 MICs. Beredaki 2023 simulated q24h micafungin exposures in the internal compartment of a dialysis/diffusion model and described the concentration decay with the one-compartment mono-exponential Ct = C0 * exp(-k * t) (Methods, 'In vitro pharmacokinetics'), with an average half-life of 14 h (range 9-15 h) in the presence of 10% pooled human serum. The 72 h change in log10 CFU/mL relative to the starting inoculum was then related to the PK/PD index fAUC0-24/MIC with the sigmoidal variable-slope Emax model E = Emax * EI^n / (EI^n + EI50^n) (Methods, 'PK/PD analysis'), where E is the REDUCTION in log10 CFU/mL relative to the drug-free control. Beredaki 2023 printed the equation but not its coefficients, so e0, lemax, lec50 and lhill were recovered by digitising the fitted curve in Figure 5(a) (CLSI panel, R^2 = 0.92); the recovered coefficients reproduce the paper's own Table 2 CLSI stasis target (2.8 fAUC0-24/MIC) to +3% and its 1-log-kill target (9.2 fAUC0-24/MIC) to -13%, equivalent to 0.10 log10 CFU/mL on the effect axis. The fungal density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * ((e0 - kill72) / 72) * bact so that log10(bact) changes by exactly (e0 - kill72) across the paper's 72 h observation window, reproducing the endpoint model exactly at the time the paper actually fitted it. The companion EUCAST parameterisation is Beredaki_2023_micafungin_eucast. See the vignette for the reported no-serum targets, which are NOT packaged because Beredaki 2023 published no exposure-response curve for that condition. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Beredaki_2023_micafungin.html">Micafungin eucast (Beredaki 2023)</a> </td> <td style="text-align:left;"> In vitro (Candida albicans clinical isolates in RPMI-1640 with 10% pooled human serum; two-compartment closed dialysis/diffusion PK/PD model). Micafungin in vitro PK/PD model with the exposure-response relationship indexed on EUCAST E.Def 7.3 MICs. Beredaki 2023 simulated q24h micafungin exposures in the internal compartment of a dialysis/diffusion model and described the concentration decay with the one-compartment mono-exponential Ct = C0 * exp(-k * t) (Methods, 'In vitro pharmacokinetics'), with an average half-life of 14 h (range 9-15 h) in the presence of 10% pooled human serum. The 72 h change in log10 CFU/mL relative to the starting inoculum was then related to the PK/PD index fAUC0-24/MIC with the sigmoidal variable-slope Emax model E = Emax * EI^n / (EI^n + EI50^n) (Methods, 'PK/PD analysis'), where E is the REDUCTION in log10 CFU/mL relative to the drug-free control. Beredaki 2023 printed the equation but not its coefficients, so e0, lemax, lec50 and lhill were recovered by digitising the fitted curve in Figure 5(b) (EUCAST panel, R^2 = 0.87); the recovered coefficients reproduce the paper's own Table 2 EUCAST stasis target (2.8 fAUC0-24/MIC) to -1% and its 1-log-kill target (7.9 fAUC0-24/MIC) to -6%, equivalent to 0.05 log10 CFU/mL on the effect axis. The fungal density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * ((e0 - kill72) / 72) * bact so that log10(bact) changes by exactly (e0 - kill72) across the paper's 72 h observation window, reproducing the endpoint model exactly at the time the paper actually fitted it. The companion CLSI parameterisation is Beredaki_2023_micafungin_clsi. See the vignette for the reported no-serum targets, which are NOT packaged because Beredaki 2023 published no exposure-response curve for that condition. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Brill_2014_midazolam.html">Midazolam (Brill 2014)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for midazolam with two equalized peripheral volumes and a three-transit-compartment first-order oral absorption chain (Ka = Ktr), supporting oral and intravenous dosing, in 20 morbidly obese patients (mean total body weight 144 kg, range 112-186; mean BMI 47, range 40-68) and 12 non-obese healthy volunteers (mean total body weight 76 kg, mean BMI 22). Total body weight enters as a linear covariate on central volume (reference 127 kg) and a power covariate on peripheral volume (reference 127 kg); morbid-obesity status (BMI > 40) shifts oral bioavailability up and the transit absorption rate down. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Franken_2017_midazolam.html">Midazolam (Franken 2017)</a> </td> <td style="text-align:left;"> Joint parent-metabolite population PK model for midazolam, its primary active metabolite 1-OH-midazolam (1-OH-M), and the secondary metabolite 1-OH-midazolam-glucuronide (1-OH-MG) in 45 terminally ill adult palliative-care patients (Franken 2017). Midazolam: one-compartment disposition with two parallel first-order absorption routes (oral and subcutaneous bolus) using route-specific absorption rate constants fixed from literature (Ka oral = 5.5 1/h, Ka SC = 10 1/h); oral bioavailability F is estimated and SC F assumed = 1. 1-OH-M: one compartment, central volume fixed equal to midazolam V, clearance estimated. 1-OH-MG: one compartment, clearance and volume estimated. All inter-compartment fluxes carry the parent / metabolite signal in midazolam-equivalent mass units (concentrations were adjusted to midazolam equivalents via molecular weight per Methods). Midazolam clearance depends on serum albumin (power form, reference 25 g/L) and 1-OH-MG clearance depends on eGFR (standard four-variable MDRD, power form, reference 104 mL/min/1.73 m^2). IIV on midazolam CL was correlated with oral F (rho fixed to unity per Results); other IIVs are independent. Residual variability is additive on log-transformed concentrations (LTBS) for all three analytes; a cross-output residual correlation noted in Methods is not encoded in this nlmixr2 port (see vignette Assumptions and deviations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Janssen_2017_cabazitaxel_phenotyping.html">Midazolam (Janssen 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous midazolam used as a CYP3A metabolic-phenotyping probe in 10 men with metastatic castration-resistant prostate cancer (Janssen 2017 Table 2A). Each patient received a single 2.5 mg IV midazolam bolus 1-7 days before their scheduled cabazitaxel infusion; the empirical Bayes estimates of individual midazolam clearance from this fit are then used as covariate input to the companion Janssen 2017 cabazitaxel model (see modellib('Janssen_2017_cabazitaxel')). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Swart_2004_lorazepam_midazolam.html">Midazolam (Swart 2004)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for midazolam by continuous infusion in mechanically ventilated critically ill adult ICU patients. Clearance is selected by chronic alcohol-abuse status and decreases linearly with age above 57 years; intercompartmental clearance decreases linearly with APACHE II score above 26. Fitted by NONMEM V in the Swart 2004 learning cohort (n = 21). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vanRongen_2015_midazolam.html">Midazolam (vanRongen 2015)</a> </td> <td style="text-align:left;"> Joint parent-and-sequential-metabolites population PK model for intravenous midazolam, its primary CYP3A oxidative metabolite 1'-hydroxymidazolam (1-OH-midazolam), and the downstream phase-II 1'-hydroxymidazolam glucuronide (1-OH-midazolam glucuronide) in 19 overweight and obese adolescents (12.5-18.9 years, body weight 62-149.8 kg) undergoing surgery (van Rongen 2015). Two-compartment disposition for midazolam (central + peripheral) routes the entire elimination clearance CL1 to 1-OH-midazolam formation. 1-OH-midazolam is described by a one-compartment model with apparent volume of distribution fixed at 0.9 times the midazolam central volume (Mandema 1992); the entire 1-OH-midazolam clearance CL3 is routed to 1-OH-midazolam glucuronide formation. 1-OH-midazolam glucuronide is described by a two-compartment model with renal elimination clearance CL4. Total body weight (TBW) enters the peripheral volume of distribution of midazolam as a power function with reference 104.7 kg (cohort median) and estimated exponent X = 1.68; no other covariate effect was retained. Concentrations are modeled in umol/L throughout (paper Methods), so dosing is in umol; dose_umol = dose_mg * 1000 / MW_midazolam where MW_midazolam = 325.77 g/mol. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vanRongen_2017_midazolam.html">Midazolam (vanRongen 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for midazolam in 19 obese adolescents (12-18.9 years, total body weight 62-149.8 kg, BMI 24.8-55 kg/m^2) and 20 morbidly obese adults (26-57 years, total body weight 112.3-186.3 kg, BMI 39.9-67.6 kg/m^2), with a five-transit-compartment first-order oral absorption chain (Ka = Ktr) supporting oral and intravenous dosing (van Rongen 2017 Final model, Table 2). Study population (adolescent vs morbidly obese adult) separates clearance into two cohort-specific values (CL_104.7 kg in adolescents with an estimated TBW power on top, CL fixed-across-WT in morbidly obese adults). The same V_141.8 kg central value of the peripheral compartment is shared between cohorts but TBW power scaling applies only to morbidly obese adults. Central volume, inter-compartmental clearance Q, transit-absorption rate Ka = Ktr, and oral bioavailability F are shared across both cohorts. Oral data were collected only in morbidly obese adults; adolescents received only IV bolus doses. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cella_2012_midazolam_paediatric_scaling.html">Midazolam children adolescents (Cella 2012)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for midazolam in children and adolescents (Cella 2012 Model 2), IV bolus only, with per-kg linear scaling of the central volume and a linear normalisation of the peripheral volume by age (months) at a 74-month reference. Cohort of 18 paediatric oncology patients (ages 3.2 to 16.2 years, body weights 12.6 to 60.1 kg) dosed at 0.12 mg/kg IV before invasive procedures. Fitted with informative priors from De Wildt 2002 via the NONMEM PRIOR / Wishart subroutine. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cella_2012_midazolam_paediatric_scaling.html">Midazolam infants adults (Cella 2012)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for midazolam in infants, toddlers, and adults (Cella 2012 Model 1), with first-order absorption supporting intravenous and oral dosing, body-weight allometric scaling of clearance (exponent fixed to 0.75 at a 70 kg reference), per-kg linear scaling of the central volume, and a constant peripheral volume. Pooled cohort of 23 infants and toddlers in a paediatric surgical ICU and 34 healthy adult volunteers. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Brussee_2018_midazolam_pbpk.html">Midazolam pbpk (Brussee 2018)</a> </td> <td style="text-align:left;"> PBPK (semi-physiological; well-stirred liver + Qgut gut wall) population PK model for midazolam and its primary metabolite 1-OH-midazolam in 37 preterm neonates (gestational age 26-34 weeks, body weight 0.770-2.030 kg at the time of dosing). Distinguishes first-pass CYP3A-mediated metabolism in the gut wall (Qgut model) and liver (well-stirred model) from systemic hepatic elimination of the metabolite. Tissue volumes (V_h, V_pv, V_gw) and hepatic blood flow Q_h are allometrically scaled from a term-neonate reference (Bjorkman 2005) by body weight with fixed exponents (1 for volumes, 0.75 for flow); intestinal length scales as 2.736 * WT[g]^0.512 cm (Struijs 2009) so the Qgut hybrid flow varies with body size. Supports oral administration (depot, full first-pass through gut wall and liver) and IV (dose directly to central; no first-pass). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hajjar_2018_pompe_disease.html">Miglustat (Hajjar 2018)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral miglustat (AT2221; N-butyl-1-deoxynojirimycin) administered as a pharmacological chaperone for cipaglucosidase alfa in adult patients with Pompe disease (Hajjar 2018 ACCP poster, phase 1/2 study ATB200-02 / NCT02675465). Absorption is described by a sequential zero-order release into the depot (duration D1 = 0.459 h) followed by first-order absorption (Ka = 0.485 /h) into the central compartment. Apparent disposition parameters (CL/F = 8.55 L/h, Vc/F = 36.3 L, Q/F = 3.16 L/h, Vp/F = 45.6 L) are reported at the 70 kg reference body weight and allometrically scaled with exponents 0.75 fixed on clearances and 1 fixed on volumes. Bioavailability F is not estimable from oral-only data and is anchored at 1 (apparent CL/F and V/F parameterisation). Residual error is proportional (variance 0.0408, SD 0.202 on the linear-scale concentration). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dorlo_2008_miltefosine.html">Miltefosine (Dorlo 2008)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order oral absorption and linear elimination for miltefosine in 31 Dutch military personnel with Old World (Leishmania major) cutaneous leishmaniasis contracted in Afghanistan (Dorlo 2008), treated with oral miltefosine 50 mg three times daily (150 mg/day, median 1.76 mg/kg/day) for 28 days with post-treatment follow-up to a maximum of 202 days. CL/F, Vc/F, Q/F, and Vp/F are estimated apparent parameters; relative bioavailability F is unidentifiable from oral-only data and is structurally fixed at 1. Inter-individual variability is log-normal on ka, CL/F, and Vc/F (diagonal in this implementation; see Assumptions in the vignette for the unreported CL/Vc correlation noted by the authors). IIV on Q/F and Vp/F was not estimable from the data. Residual error is proportional (31.5% CV). No covariate effects were retained in the final model. This is the structural model later re-used as the base PK structure in Dorlo 2017 visceral-leishmaniasis miltefosine work. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dorlo_2017_miltefosine.html">Miltefosine (Dorlo 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order oral absorption for miltefosine in 95 Eastern African adults and children (>=7 years) with visceral leishmaniasis (Dorlo 2017), enrolled across three treatment centres in Kenya and Sudan and randomised to either a 28-day monotherapy regimen of oral miltefosine 2.5 mg/kg/day or a 10-day oral miltefosine 2.5 mg/kg/day arm combined with a single 10 mg/kg liposomal amphotericin B IV dose on day 1. CL/F, Q/F, Vc/F, and Vp/F are allometrically scaled on fat-free mass (exponents 0.75 and 1.0; reference FFM 53 kg). Relative bioavailability is structurally fixed at 100% from the end of the initial reduced absorption window onwards, and reduced by a typical 74.3% during the window itself (0 < t <= 7 days for monotherapy, 0 < t <= 1 day for the combination arm); the duration is regimen-dependent via the MIL_REGIMEN indicator. The combined-error residual model is proportional (31.0%) with additive component fixed at 0.001 ug/mL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sahota_2015_miridesap.html">Miridesap (Sahota 2015)</a> </td> <td style="text-align:left;"> Target-mediated drug disposition (TMDD) PK/PD model for CPHPC (miridesap, GSK2315698, Ro 63-8695) and serum amyloid P (SAP) in healthy volunteers (study CPH113776) and patients with systemic amyloidosis (study CPH114527). Two-compartment PK for CPHPC (IV plus first-order subcutaneous depot); two-compartment turnover model for SAP with first-order endogenous production and elimination; bimolecular CPHPC + free SAP -> complex binding treated as effectively irreversible (KOFF set to zero because the complex internalisation rate is much faster than the dissociation rate). Final-model covariates (Sahota 2015 Eq. 1 and Eq. 2): creatinine clearance modifies CPHPC clearance below an 80 mL/min threshold; hepatic amyloid involvement multiplies SAP intercompartmental clearance Q4; whole-body amyloid load (categorical 0-3) multiplies SAP peripheral volume V4 in two cumulative steps; biological sex multiplies baseline plasma SAP. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chua_2025_mirikizumab.html">Mirikizumab (Chua 2025)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for mirikizumab (anti-IL-23p19 IgG4 mAb) in patients with moderately-to-severely active Crohn's disease (Chua 2025 VIVID-1 phase 3) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vorontsova_2022_misoprostol.html">Misoprostol (Vorontsova 2022)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model of misoprostol acid (MPA, the active metabolite) following buccal or vaginal misoprostol tablet administration for full-term labor induction, with mixed linear plus Michaelis-Menten clearance. The absorption rate constant is estimated separately by route (buccal vs vaginal) and dose level (25 vs 50 microgram); a relative bioavailability multiplier captures the higher vaginal exposure, with inter-occasion variability across the first two dose events. IIV is estimated on apparent clearance, apparent central volume of distribution, and the absorption rate constant. Development cohort: 47 women at term gestation (>=37 weeks) undergoing labor induction in the IMPROVE trial (NCT02408315). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/An_2012_mitoxantrone_pbpk.html">Mitoxantrone human pbpk (An 2012)</a> </td> <td style="text-align:left;"> Human-scaled simulation. Semi-mechanistic whole-body PBPK model 3 for mitoxantrone (Novantrone) in adult cancer patients after a single 12 mg/m^2 IV bolus, projected forward from the mouse fit in An_2012_mitoxantrone_mouse_pbpk (An and Morris 2012, AAPS J). Same topology as the mouse model: seven physiological tissue compartments (central plasma plus six perfusion-limited well-stirred organs - lung, heart, spleen, liver, kidney, brain) and a permeability-limited remainder compartment lumping muscle, fat, bone, and skin and resolving into an interstitial (is_remainder) and an intracellular (int_remainder) subspace coupled by a permeability-surface area product. Plasma unbound fraction fu = 0.2, DNA dissociation constant K_DNA = 0.0013 uM, protein-binding dissociation constant K_macro = 1.44 uM, and per-organ T_macro are carried over from the mouse fit as cross-mammalian constants (paper Methods). Per-organ T_DNA values are replaced with the human DNA-content literature values reported in Table III: lung and spleen DNA use the literature rapidly- perfused-organ value (15 uM); brain T_DNA is the mouse-derived value (0.10 uM); the remainder T_DNA uses the literature slowly-perfused- organ value (4.5 uM). The remainder ISF/intracellular split is assumed to follow the mouse proportion 33/67 (Table I) applied to the human remainder volume V_other = 62 L. Hepatic and renal intrinsic clearances are derived from clinical CL_H = 19 L/h/m^2 and CL_R = 2.7 L/h/m^2 (Ehninger 1990 ref 6) via the well-stirred rearrangement Clint = Q_H * CL / (fu * Q_H - fu * CL) yielding Clint_H = 250 L/h and Clint_R = 27 L/h. PS_remainder is allometrically scaled from the mouse value PS = 1.44 mL/min via PS = A * M^0.75 (Kawai 1998 ref 28) giving PS = 31.1 L/h. The model is a typical-value forward simulation; there is no IIV and no residual error from the paper. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/An_2012_mitoxantrone_pbpk.html">Mitoxantrone mouse pbpk (An 2012)</a> </td> <td style="text-align:left;"> Preclinical (mouse). Semi-mechanistic whole-body PBPK model 3 for mitoxantrone (Novantrone) in male ND4 Swiss Webster mice (24-32 g) after a single 5 mg/kg intravenous bolus (penile vein) (An and Morris 2012, AAPS J). Seven physiological tissue compartments: central plasma plus six perfusion-limited well-stirred organs (lung, heart, spleen, liver, kidney, brain), with a permeability-limited remainder compartment that lumps muscle, fat, bone, intestine, and skin and resolves into an interstitial (is_remainder) and an intracellular (int_remainder) subspace coupled by a permeability- surface area product PS. Hepatic and renal elimination act on the unbound cellular concentration of liver and kidney via well-stirred intrinsic clearances Clint_H and Clint_R. Saturable tissue binding to DNA (capacity T_DNA, affinity K_DNA) and macromolecular protein (capacity T_macro, affinity K_macro) is encoded as a Cp-dependent effective tissue:plasma partition coefficient Kp_eff(Cp) that varies instantaneously with plasma concentration (Eqs. 9-11 of the paper). Plasma unbound fraction is fixed to 0.2; tissue binding affinities are shared across all organs. The model is intended for typical- value simulation of mouse plasma and tissue concentration-time profiles. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Honda_2006_mizoribine.html">Mizoribine (Honda 2006)</a> </td> <td style="text-align:left;"> One-compartment oral PK model for mizoribine in healthy Caucasian male volunteers (Honda 2006); first-order absorption with a fixed absorption-lag time, apparent volume of distribution V/F linear in body weight, apparent oral clearance CL/F linear in Cockcroft-Gault creatinine clearance (CLcr), and additive residual error on the serum-concentration scale. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Peng_2014_MK3577.html">MK3577 (Peng 2014)</a> </td> <td style="text-align:left;"> Semi-mechanistic PD model for the glucagon receptor antagonist MK-3577 (Merck) in healthy male subjects undergoing a glucagon challenge (Peng 2014). Glucose / glucagon / insulin homeostasis is described with constant endogenous baselines Gss, Iss, GNss; glucose has central + peripheral distribution plus an effect compartment for delayed regulation of glucose production. MK-3577 drives an inhibitory Imax model on the glucagon-stimulation of glucose production (Imax,MK = 0.961, IC50,MK = 13.9 nM) and a stimulatory Emax model on glucagon secretion (Emax,MK = 0.788 FIX, EC50,MK = 575 nM FIX) for the prechallenge compensatory feedback. Sandostatin (octreotide) PK is modeled with literature CL / V (0.121 L/kg/h, 0.194 L/kg) and inhibits endogenous insulin (IC50,S2 = 0.921 ng/mL) and glucagon (IC50,S1 = 5.50 ng/mL) secretion at a fixed Imax of 1. The MK-3577 PK layer is NOT modeled here because the absorption rate ka, apparent volume V/F, and molecular weight for the mg-to-nM conversion are not reported in the on-disk paper or its tables; users supply the MK-3577 plasma concentration as a time-varying covariate column CP_MK3577_NM (nM) per the standing operator decision (extract PD layer only; PK supplied externally). See vignette Assumptions and deviations for the gap. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Peng_2014_MK3577.html">MK3577 t2dm (Peng 2014)</a> </td> <td style="text-align:left;"> T2DM-patient adaptation of the Peng 2014 semi-mechanistic glucose / glucagon / insulin model with MK-3577 as the glucagon receptor antagonist (Peng 2014 Fig. 1b and 'CTS Method for T2DM Patients'). Three structural changes vs. the healthy model: (1) GPRG1 = 0 (glucose self-regulation of GPROD is fully compromised in T2DM, matching Silber 2007); (2) CL_GI is scaled to 11% of the healthy value (lead- compound finding); (3) baseline glucose is elevated by a fold factor theta with typical value fixed at 1 (i.e., baseline FPG is 2 x healthy G_SS) and IIV fixed at 51% CV based on lead-compound data. The new T2DM baselines for insulin (I_SSP) and glucagon (GN_SSP) are derived from theta via Peng 2014 Eqs. 7-10 (insulin: I_SSP = I_SS * (1+theta)^IPRG; glucagon: closed-form rearrangement of Eqs. 9 and 10 conditioned on the T2DM-scaled CL_GI). The effect compartment for glucose negative feedback is omitted because GPRG1 = 0 makes its contribution vanish; the Sandostatin compartment is also omitted because the T2DM phase IIa CTS used in Peng 2014 had no glucagon challenge -- only endogenous homeostasis under multi-day MK-3577 dosing. The MK-3577 PK layer is NOT modeled here for the same reason as the healthy model (ka, V/F, MW not in the on-disk PDF); users supply the time-varying MK-3577 plasma concentration via CP_MK3577_NM (nM). See vignette Assumptions and deviations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wu_2012_modafinil.html">Modafinil (Wu 2012)</a> </td> <td style="text-align:left;"> Joint parent + metabolite population pharmacokinetic model for oral modafinil and its principal carboxylic-acid metabolite modafinil acid (2-[(diphenylmethyl)sulfonyl]acetic acid) in 49 healthy volunteers from five major ethnic groups of China (Han, Mongolian, Korean, Uygur, Hui) under a single 200 mg oral dose (Wu 2012). Four-compartment NONMEM ADVAN6 / L2 structure: GI depot with first-order absorption (ka), two-compartment modafinil disposition (apparent CL/F, Vc/F, Q/F, Vp/F), and a one-compartment modafinil acid disposition (apparent CL3/F1F2, V3/F1F2). All modafinil elimination is treated as forming modafinil acid at the apparent-parameter level because F2 (the absolute modafinil-to-acid metabolic-conversion fraction) is not identifiable from oral plasma data alone; F2 is absorbed into the apparent acid parameters. Sex acts on CL/F, Q/F, and Vp/F of modafinil; ethnicity acts on Vc/F of modafinil (Korean and Hui share a single composite multiplier; Mongolian and Uygur each have their own) and on CL3/F1F2 of modafinil acid (Han and Mongolian share the reference; Korean has its own multiplier; Uygur and Hui share a composite multiplier). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mukai_2019_mogamulizumab.html">Mogamulizumab (Mukai 2019)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for mogamulizumab in adults with cutaneous T-cell lymphoma or adult T-cell lymphoma (Mukai 2019) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hwang_2023_monalizumab.html">Monalizumab (Hwang 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for monalizumab (anti-CD94/NKG2A IgG4) in patients with advanced solid tumors or squamous cell carcinoma of the head and neck (Hwang 2023) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Abrantes_2017_moroctocog.html">Moroctocog (Abrantes 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for factor VIII activity (IU/dL) following intravenous administration of moroctocog alfa (B-domain-deleted recombinant FVIII, marketed as ReFacto, ReFacto AF and Xyntha) in patients with moderate to severe hemophilia A; pooled analysis of 754 patients across 13 clinical trials over 20 years (Abrantes 2017). The exogenous-drug component is added to a constant endogenous-baseline FVIII activity (severe-subpopulation typical value, 0.474 IU/dL; the paper's full model is a two-class mixture, see vignette deviations). Clearance and inter-compartmental clearance scale allometrically with body weight at theory-based exponent 0.75; central and peripheral volumes share an estimated allometric exponent 0.812. Clearance has a piecewise-linear age effect (increasing from birth to 1 year of age, then decreasing into adulthood; centered at 20 years), a +166% inhibitor (ADA_POS) effect, and a -34.7% study B1831090 effect. The peripheral volume is +88.4% larger in Black subjects. Bioavailability F carries multiplicative covariate effects for product (1.38x for Xyntha vs ReFacto), assay (-39.0% for OSA central, -14.6% for OSA local laboratory) following Abrantes 2017 Table 2 footnote g. Proportional residual error is 19.2% (CSA reference) and switches to 26.9% (+40.3%) for OSA-assayed samples. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/deHoogd_2017_morphine.html">Morphine (deHoogd 2017)</a> </td> <td style="text-align:left;"> Joint parent-metabolite population PK model for morphine and its two glucuronide metabolites (M3G, M6G) in 20 morbidly obese adults (post-gastric-bypass) and 20 healthy adult volunteers (de Hoogd 2017). Morphine: three-compartment IV model with total body weight (TBW) covariate on the second peripheral volume V5M. Non-glucuronide morphine clearance is structurally fixed at 35% of total morphine CL in a 70-kg healthy adult. M3G and M6G are each one-compartment models fed by formation-delay transit chains (n = 5 for M3G, n = 2 for M6G); VM3G = VM6G is a structural equality. TBW covariates apply to CLF M6G, the M3G transit rate Ktr, M3G elimination CL, and M6G elimination CL, all power-form normalised to a reference of 98.5 kg (population median). Proportional residual error is reported separately for the healthy- volunteer cohort and the morbidly obese cohort, selected via the binary indicator DIS_OBESE_MORBID. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Elkomy_2015_morphine.html">Morphine (Elkomy 2015)</a> </td> <td style="text-align:left;"> Joint parent-metabolite population PK model for morphine and its two glucuronide metabolites M3G and M6G in 20 infants and young children (3 days - 5.4 years; 3.1 - 18.5 kg) after congenital heart surgery (Elkomy 2015 AAPS J). Morphine: linear two- compartment IV disposition with allometric body-weight scaling (CL and CLD with exponent 0.75 FIXED; VC and VP with exponent 1.0 FIXED) normalised to a reference of 6 kg (study median). Each metabolite is modeled as a morphine-driven intermediate effect compartment (rate constant Kint chasing morphine plasma concentration via dCint/dt = Kint * (Cc - Cint)) feeding an empirical Emax transduction where metabolite concentration = Mmax * Cint / (Cint50 + Cint). Estimated glomerular filtration rate (Schwartz formula) is a covariate: Kint scales linearly with GFR/70 and Mmax scales as 70/GFR (both exponents FIXED at +/-1 per the paper's covariate analysis). Between-subject random effects on Kint, Mmax, and Cint50 are SHARED across the M3G and M6G channels (one eta per group; Table II). Doses are administered as nmol of morphine equivalents (one nmol of clinical morphine sulfate yields two nmol of free morphine); concentrations are output in nmol/L (nM) for morphine, M3G, and M6G. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Franken_2015_morphine.html">Morphine (Franken 2015)</a> </td> <td style="text-align:left;"> Joint parent-metabolite population PK model for morphine and its two glucuronide metabolites (M3G, M6G) in 47 terminally ill adult palliative care patients (Franken 2015). Morphine: two-compartment disposition with three parallel first-order absorption routes (subcutaneous bolus, immediate-release oral liquid, controlled-release oral tablet) using route-specific fixed absorption rate constants; oral bioavailability F is estimated (SC F assumed 1). M3G and M6G are each one-compartment models fed by fixed-fraction transformation of morphine clearance (Fm1 = 0.55 for M3G, Fm2 = 0.10 for M6G, both fixed from literature). Morphine clearance decreases exponentially as time-to-death (TTD, days) approaches zero. Metabolite clearance depends on estimated glomerular filtration rate (eGFR, MDRD four-variable formula) and serum albumin via shared power-form covariate exponents. Residual variability was reported as additive error on the log-transformed observation (LTBS). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pierre_2017_morphine.html">Morphine (Pierre 2017)</a> </td> <td style="text-align:left;"> Joint parent-metabolite population PK model for IV morphine and its primary glucuronide metabolite morphine-3-glucuronide (M3G) in 14 healthy adults and 7 patients with biopsy-confirmed nonalcoholic steatohepatitis (NASH) following a single 5 mg morphine sulfate IV infusion (Pierre 2017). Morphine is described by a three-compartment disposition (central + two peripherals) with parallel renal (CL_M_R) and non-renal (CL_M_NR) clearances; the entire non-renal clearance is assumed to lead to M3G formation via a single liver transit compartment with first-order rate constant k_trans. M3G is described by a one-compartment model with a single total clearance (CL_M3G). Cumulative urinary morphine and M3G amounts are tracked as elimination-amount compartments. Total body weight enters all CL/Q and V parameters a priori with fixed allometric exponents (0.75 and 1, respectively) referenced to 70 kg. The NASH severity score (NASF; combined NAFLD activity score and fibrosis staging, 0-12) is the only additional covariate retained in the final model; it acts on M3G clearance through a linear effect on the natural logarithm of (NASF / 4) for NASF >= 4 and is identically zero for NASF < 4 so that healthy and benign-NAFLD subjects (NASF < 5) recover the typical CL_M3G. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Colin_2014_moxifloxacin.html">Moxifloxacin (Colin 2014)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for moxifloxacin in post-bariatric (roux-en-y gastric bypass) volunteers (Colin 2014): linear first-order absorption (no lag, no transit) into a central compartment with two peripheral compartments, allometric scaling on lean body mass (exponent 0.75 on all CL terms and 1 on all volumes, reference LBM 60 kg), and inter-individual variability on ka, central volume, and clearance. Single 400 mg oral and 400 mg 1-h IV infusion doses are fit simultaneously with an implicit bioavailability of 1. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hong_2015_moxifloxacin.html">Moxifloxacin (Hong 2015)</a> </td> <td style="text-align:left;"> Sequential population PK + PD (QT-interval) model for single-dose oral moxifloxacin (400 mg or 800 mg, Avelox tablets) in healthy adult Korean male volunteers (Hong 2015): a two-compartment first-order absorption PK model with a lag time and a dose-dependent absorption rate constant (different Ka for 400 mg vs 800 mg), followed by an individually corrected QT-interval PD model that adds two mixed-effect cosine circadian components (24 h and 6 h), a first-order-decaying placebo (water-intake) effect, and an Emax drug effect on QT prolongation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2009_moxifloxacin.html">Moxifloxacin (Landersdorfer 2009)</a> </td> <td style="text-align:left;"> Population PK model for oral moxifloxacin bone penetration (Landersdorfer 2009): two-compartment plasma disposition with first-order absorption from a gut depot, plus two paper-mechanistic bone matrix compartments (cortical and cancellous bone) connected to the central compartment by fixed transfer rate constants. The bone tissue:serum equilibrium concentration ratio is captured by the multiplicative scale terms fcortical and fcancellous on the cortical and cancellous bone observations. Disposition parameters were MAP-Bayesian estimated against Simon 1997 priors; bone-penetration scale terms used noninformative priors. Single 400 mg oral dose in 24 adults undergoing total hip replacement; serum and femoral bone samples (cortical + cancellous, head + neck) collected 2 to 7 hours post-dose. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Morganroth_2015_moxifloxacin.html">Moxifloxacin (Morganroth 2015)</a> </td> <td style="text-align:left;"> Population pharmacodynamic linear concentration-effect model for moxifloxacin-induced placebo- and baseline-corrected QTcF interval prolongation (DDQTcF) in 40 healthy adult male Japanese and 40 healthy adult male Caucasian volunteers following a single 400 mg oral dose in a thorough QT study (Morganroth 2015). The linear mixed-effects PD model of Table 4 has form DDQTcF = (alpha + rho * RACE_WHITE) + (beta + gamma * RACE_WHITE) * Cc, with alpha = 1.71 ms, beta = 2.58 ms per ug/mL (Japanese reference), rho = 2.58 ms (additive intercept shift in Caucasian; USA site), gamma = -0.24 ms per ug/mL (concentration-by-country interaction; Caucasian slope = 2.34 ms per ug/mL). The source publication does not fit a popPK model; the PK driver in this file is a typical-value 1-compartment oral approximation with CL/F = 8.47 L/h, V/F = 132.6 L, and ka = 1.7 /h derived from the pooled NCA summary statistics in Morganroth 2015 Table 1 (see vignette Errata). Per operator sidecar-001 (option C) the paper's BVN(0, Sigma) subject random effects on the PD intercept and slope are OMITTED because Sigma is not numerically reported in the paper, and a small placeholder additive residual SD of 1 ms is used to satisfy rxode2's residual-error requirement; downstream users who need VPC-style simulation must add their own IIV. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nielsen_2011_antibacterial_efficacy.html">Moxifloxacin (Nielsen 2011)</a> </td> <td style="text-align:left;"> In vitro (Streptococcus pyogenes M12 NCTC P1800). Semimechanistic PKPD model of moxifloxacin time-kill kinetics; two-stage bacterial life-cycle (proliferating drug-sensitive S and non-growing drug-insensitive R) with sigmoidal Emax killing of S via an effect compartment; first-order drug elimination (ke set per in vitro kinetic-system flow rate); drug-specific degradation kdeg fixed at zero. Parameter values are from the combined static and dynamic estimation in Table 3. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kaushal_2024_mRNA0184.html">MRNA0184 cyno (Kaushal 2024)</a> </td> <td style="text-align:left;"> Preclinical (cynomolgus monkey). Translational semi-mechanistic PK/PD model for mRNA-0184, a lipid-nanoparticle-encapsulated mRNA encoding human relaxin-2 fused to a variable light chain kappa domain (Rel2-vlk), for heart failure. PK: 3-compartment plasma1-tissue-plasma2 redistribution of Rel2-vlk mRNA (both plasma compartments share V1; elimination is from the tissue compartment) reproducing the delayed second concentration peak characteristic of LNP modalities; observed mRNA concentration is the sum of the two plasma compartments. PD: Rel2-vlk protein production linear in the concentration of a hypothetical effect compartment equilibrating with plasma mRNA, distributed in a 2-compartment (central/peripheral) system with a single symmetric intercompartmental clearance and first-order elimination from the central compartment. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kaushal_2024_mRNA0184.html">MRNA0184 human (Kaushal 2024)</a> </td> <td style="text-align:left;"> Human-scaled allometric projection (NO human data were fitted). Translational semi-mechanistic PK/PD model for mRNA-0184, a lipid-nanoparticle-encapsulated mRNA encoding human relaxin-2 fused to a variable light chain kappa domain (Rel2-vlk), for heart failure with reduced ejection fraction. This is the cynomolgus-monkey model of Kaushal 2024 forward-projected to a 70 kg adult by allometric scaling (exponent 1 on volumes, 0.75 on Rel2-vlk mRNA clearances, 0.85 on Rel2-vlk protein clearances). PK: 3-compartment plasma1-tissue-plasma2 redistribution of Rel2-vlk mRNA (both plasma compartments share V1; elimination from the tissue compartment). PD: Rel2-vlk protein production linear in a hypothetical effect-compartment mRNA concentration, distributed in a 2-compartment system with a single symmetric intercompartmental clearance and first-order central elimination. Used to support the 0.025 mg/kg every-2-weeks first-in-human starting dose for NCT05659264. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Attarwala_2023_mRNA3927.html">MRNA3927 (Attarwala 2023)</a> </td> <td style="text-align:left;"> Preclinical (mouse, rat, cynomolgus monkey; allometrically scalable to humans). Translational semi-mechanistic PK and PK/PD model for mRNA-3927, an LNP-encapsulated dual mRNA encoding propionyl-CoA carboxylase (PCC) subunits PCCA and PCCB. PK: 3-compartment plasma1-tissue-plasma2 redistribution (V shared between the two plasma compartments; V and V2 fixed at the mouse reference and scaled allometrically) with body-weight allometric scaling of clearances (mouse reference 0.025 kg; estimated exponents cla on CL12/CL32 and clb on CL23/CL20). PD: liver PCC protein 2-compartment indirect-response model driven by an effect compartment linked to plasma mRNA, with synthesis linear in effect-compartment mRNA concentration and first-order degradation. Three downstream biomarkers (2-methylcitrate, 3-hydroxypropionate, C3/C2 carnitine ratio) follow direct sigmoidal Imax suppression by liver PCC protein with Imax fixed at 0.999. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Niloy_2026_MTMSATrp_mouse.html">MTMSATrp mouse (Niloy 2026)</a> </td> <td style="text-align:left;"> Preclinical (mouse). One-compartment population PK model for MTMSA-Trp, a novel mithramycin analogue investigated for Ewing sarcoma, in female athymic nu/nu mice following single IV bolus doses of 0.3, 1, 3, 5, or 10 mg/kg. First-order elimination from the central compartment with an empirical power-function effect of dose on apparent clearance (CL decreases with increasing dose; reference dose 3 mg/kg, exponent beta = -0.30). Parameters are expressed in per-kg body-weight units (mL/h/kg for CL, mL/kg for V) so the dose record carries the per-kg dose directly (mg/kg) without an explicit body-weight covariate. Parameter values from Niloy 2026 Table 1 (final model). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Laffont_2025_opioid_overdose_reversal_simulation.html">Mu receptor binding (Mann 2022)</a> </td> <td style="text-align:left;"> QSP. Competitive mu-opioid receptor binding kinetics layer of the Mann 2022 translational opioid-overdose model. Tracks the fraction of receptors bound by an opioid agonist (RL_op) and by an opioid antagonist (RL_antag) under simultaneous exposure to both ligands. All 12 ligands characterised by Mann 2022 Supplement 1 Table S2 are carried inline as fixed parameters (Kon in pM^-n s^-1, Koff in s^-1, slope n unitless); the OPIOID_ID and ANTAGONIST_ID integer covariates select which ligand occupies each binding slot at simulation time, so the same compiled model can simulate any agonist-antagonist pair from the Table-S2 panel without re- instantiation. Ligand effect-site concentrations enter the model as the time-varying covariate columns L_OPIOID_pM and L_ANTAGONIST_pM, typically piped from the PK layer in a composed chain (e.g., Mann_2022_fentanyl_iv or Mann_2022_carfentanil_iv for the opioid slot; Laffont_2024_naloxone or Laffont_2024_nalmefene for the antagonist slot). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Barau_2012_mycophenolic_acid.html">Mycophenolic acid (Barau 2012)</a> </td> <td style="text-align:left;"> One-compartment population PK model for mycophenolic acid (MPA, active moiety of mycophenolate mofetil MMF) after oral MMF dosing in paediatric liver transplant recipients (Barau 2012). First-order absorption and first-order elimination, with diagonal (uncorrelated) inter-individual variability on ka, CL/F, and V/F and proportional residual error. Two covariates are retained in the final model: a linear-with-age effect on ka of the form ka_TV = 3.9 - 2.2 * (AGE / 8.65 years), so ka declines from 3.9 1/h at AGE = 0 to 1.7 1/h at the cohort median age of 8.65 years; and a power-on-binary effect on V/F of the form V/F = 64.7 L * 2.3^POSTTX_EARLY, where POSTTX_EARLY = 1 within the first 6 months post-transplant (POD <= 180 days) and 0 thereafter, so V/F is 64.7 L in the stable post-transplant period and 148.8 L in the immediate post-transplant period (paper attributes the volume increase to the higher unbound MPA fraction associated with low serum albumin in the immediate post-transplant period). Apparent clearance CL/F = 12.7 L/h carries no retained covariate effect in the final model. Enterohepatic recirculation, the MPAG metabolite compartment, and protein binding are not modelled here -- the paper attributes the absence of secondary peaks to surgical removal of the gallbladder in the liver-transplant recipients. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/deWinter_2009_mycophenolic_acid.html">Mycophenolic acid (deWinter 2009)</a> </td> <td style="text-align:left;"> Semi-mechanistic competitive-protein-binding population PK model for mycophenolic acid (MPA, the active moiety of mycophenolate mofetil MMF) and its glucuronide metabolite MPAG in adult renal transplant recipients (de Winter 2009). Free MPA (fMPA) follows a two-compartment disposition with first-order oral absorption (lag-time TLAG; fixed ka = 4.00 1/h); free MPAG (fMPAG) follows a one-compartment disposition. Both species bind competitively to a saturable plasma protein binding pool with capacity BMAX and species-specific association / dissociation rate constants k24 / k42 (MPA) and k56 / k65 (MPAG). The fMPAG-to-gallbladder transport rate constant k57 drives enterohepatic recirculation: fMPAG accumulates in a gallbladder compartment and empties into the fMPA central compartment during a fixed window (TGB to TGB+DGB post-dose) at rate constant k72, completing the EHC loop. Three covariates: a power effect of creatinine clearance (CRCL) on CL fMPAG (exponent 1.36; CRCL reference 45 mL/min); a power effect of plasma albumin (ALB) on BMAX (exponent 1.39; ALB reference 0.5 mmol/L); and a multiplicative power-form effect of cyclosporine cotreatment (CONMED_CSA) on k57 (multiplier 0.002, reducing EHC by ~99.8% under cyclosporine vs the tacrolimus reference). Total MPA (tMPA) and total MPAG (tMPAG) plasma concentrations are reported as the sum of the unbound and bound concentrations of each species. Dosing is BID by default (tau = 12 h hardcoded in model() for the gallbladder-emptying window). Concentrations are in molar units (umol/L) per the source paper's choice to analyse MMF / MPA / MPAG on a molar basis (MMF MW 433.5; MPA MW 320.3; MPAG MW 496.5). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dong_2014_mycophenolic_acid.html">Mycophenolic acid (Dong 2014)</a> </td> <td style="text-align:left;"> Population PK-PD model for oral mycophenolic acid (MPA, the active moiety of mycophenolate mofetil MMF) in paediatric renal transplant recipients in the early post-transplant period (Dong 2014). Two-compartment disposition with a Savic 2007-style 8-transit-compartment absorption chain followed by a first-order absorption step from depot to central; dose-dependent relative bioavailability described by a power function of dose per body surface area (DBSA) with reference 450 mg/m^2; estimated body-weight exponent of 0.31 on CL/F (not the canonical allometric 0.75). The PD layer links MPA plasma concentration to inosine monophosphate dehydrogenase (IMPDH) activity in peripheral blood mononuclear cells via a simplified inhibitory Emax model with Emax fixed at 0 (i.e., complete inhibition achievable in the limit of high MPA concentration). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Frymoyer_2013_mycophenolic_acid.html">Mycophenolic acid (Frymoyer 2013)</a> </td> <td style="text-align:left;"> Population PK model for unbound mycophenolic acid (MPA, the active moiety of mycophenolate mofetil MMF) in adult allogeneic haematopoietic cell transplantation (alloHCT) recipients (Frymoyer 2013). Two-compartment disposition with first-order absorption and linear elimination; oral bioavailability F = 0.560. Absorption lag time follows a two-class mixture: Group 1 (no delay, ALAG = 0, 91 % of subjects) and Group 2 (delayed absorption, ALAG = 1.96 h, 9 % of subjects), gated by the latent MIX_LAGGED_ABS class indicator. Creatinine clearance (Cockcroft-Gault with ideal body weight, NOT BSA-normalized) is the only retained covariate, entering CL as a power scaling (CRCL / 86 mL/min)^0.207. Inter-individual variability is log-normal on all structural parameters and on F; residual error is proportional. Doses are MPA-equivalent (mg) -- MMF mass must be converted externally via F_MW = 0.739 (oral) or 0.682 (IV) per Frymoyer 2013 Methods. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jiao_2008_mycophenolic_acid.html">Mycophenolic acid (Jiao 2008)</a> </td> <td style="text-align:left;"> Population PK model with enterohepatic circulation (EHC) for mycophenolic acid (MPA) and its 7-O-glucuronide metabolite (MPAG) in healthy Chinese male volunteers after a single 500 mg oral dose of mycophenolate mofetil (MMF, Cellcept). Five-compartment chain model (Figure 2 of Jiao 2008): a gastrointestinal depot, a two-compartment MPA disposition (central + peripheral), a one-compartment MPAG disposition (central_mpag), and a gallbladder accumulation compartment (gallbladder_mpag). First-order absorption with an absorption-lag time. Complete (fm = 1, fixed) one-pass conversion of MPA to MPAG by glucuronidation; MPAG is renally cleared in parallel with biliary excretion into the gallbladder. EHC is encoded as time-gated bolus releases of the gallbladder pool back into the GI depot at two postprandial meal times (4 and 10 h post-dose, study-1 design), with rate constant k51 acting over a 0.01 h window; the recycled MPAG is reabsorbed via the same first-order ka as the oral dose. The fraction of MPAG biliary-routed at the branch is encoded as EHCP = k45 / (k40 + k45). Body-weight scaling: paper Eq 5 (linear-proportional 'slope without intercept') with reference 65.5 kg applied to CL_MPA/F, Q/F, and V_3/F via fixed allometric exponent 1. Cross-parameter IIV linkage: eta(CL_MPAG/F) = psi_q_cl_mpag * eta(Q/F) reproduces the paper's joint eta structure where psi_q_cl_mpag is the paper's 'q' parameter. UGT1A9 polymorphisms were screened but not retained in the final model (no significant effect). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Park_2023_mycophenolic_acid.html">Mycophenolic acid (Park 2023)</a> </td> <td style="text-align:left;"> Population PK model for total mycophenolic acid (MPA, the active moiety of mycophenolate mofetil MMF) in pediatric allogeneic haematopoietic stem cell transplantation (HSCT) recipients receiving oral MMF for acute graft-versus-host disease prophylaxis or treatment (Park 2023). One-compartment disposition with first-order absorption and no lag time (NONMEM ADVAN2 TRANS2). Body surface area is the only retained covariate, entering the apparent volume of distribution as a centered-linear term Vd/F = 89.83 * (1 + 0.854 * (BSA - 1.11)). Inter-individual variability is log-normal on CL/F and Vd/F; residual error is combined proportional plus additive. Doses are MMF mass (mg) with no MMF-to-MPA molecular-weight conversion -- CL/F and Vd/F are apparent parameters that absorb both the molecular-weight ratio and oral bioavailability. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sherwin_2012_mycophenolic_acid.html">Mycophenolic acid (Sherwin 2012)</a> </td> <td style="text-align:left;"> Pediatric / adolescent enterohepatic-recirculation population PK model for mycophenolic acid (MPA) and its main inactive metabolite 7-O-MPA-glucuronide (MPAG) in patients with childhood-onset systemic lupus erythematosus (cSLE) on oral mycophenolate mofetil (MMF) (Sherwin 2012; n = 19, age 10-28 years). Absorption uses a Savic 2007 transit-compartment chain (number of compartments NN = 8.2 estimated, mean transit time MTT = 1.1 h estimated) feeding the gut compartment, which delivers MPA into the central compartment at a fixed first-order absorption rate constant Ka = 1.5 1/h. MPA follows two-compartment disposition (CL1/F, V3/F central; CL2/F, V4/F peripheral). The fraction FM of total MPA elimination is converted to MPAG (FM fixed at 0.85; the remaining 0.15 is metabolism to AcMPAG, not contained in the model). MPAG follows one-compartment disposition with V_MPAG fixed equal to V3 MPA and apparent renal clearance CLM/F. Total MPAG elimination is partitioned into renal ((1 - FMPAG) fraction) and biliary (FMPAG = 0.65 fraction, fixed). Biliary MPAG enters a gallbladder compartment that empties to the gut during fixed meal-time windows (1-2 h and 4-6 h post-dose); only a fraction EHC = 0.35 (fixed) of the emptied gallbladder content is reabsorbed (the rest is excreted in feces). Upon return to the gut compartment MPAG is reconverted to MPA and re-enters the absorption pathway via Ka, generating the characteristic secondary peak. The model() block hardcodes the BID dose interval (tau = 12 h) so the meal-time windows recur each interval; CL1/F, V3/F, CL2/F, V4/F, and CLM/F carry exponential IIV (paper-reported CV%). Inter-individual variability and residual error follow the source Table 3. No covariates entered the final model: bodyweight, age, sex, race, ethnicity, and disease duration were screened and rejected (see covariatesDataExcluded). Dose unit is MPA-mass-equivalent mg (MMF mg x 0.739 by molecular-weight ratio MPA/MMF = 320.3 / 433.5). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yu_2017_mycophenolic_acid.html">Mycophenolic acid (Yu 2017)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model with first-order oral absorption (no lag) for mycophenolic acid (MPA, the active component of mycophenolate mofetil, MMF) in Chinese adult renal transplant recipients (Yu 2017). Apparent clearance CL/F follows a linear-additive covariate model in body weight and serum creatinine (CL/F = 0.0916 * BW + 0.0417 * Scr + 7.98 L/h); apparent central volume V1/F follows a linear-additive covariate model in the UGT2B7 211G>T (rs7438135) genotype (V1/F = 14.7 + 7.72 * UGT2B7 L) where the paper's ordinal genotype code maps 211GT to 1, 211GG to 2, and 211TT to 3. Residual error is combined proportional plus additive on plasma MPA. Interoccasion variability (13.7% CV on CL/F and V1/F) reported by the paper is documented in the vignette but not encoded in this typical-value model because Karlsson-Sheiner IOV requires per-occasion etas and an OCC data column; the IIV-only encoding remains usable for typical-value and IIV-only simulations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zeng_2010_mycophenolic_acid.html">Mycophenolic acid (Zeng 2010)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption for mycophenolic acid (MPA, the active moiety of mycophenolate mofetil MMF) in children and young people undergoing blood or marrow, kidney, and liver transplantation (Zeng 2010, Br J Clin Pharmacol). Both intravenous (2 h infusion to the central compartment) and oral (depot with first-order absorption rate ka and bioavailability F) routes were modelled jointly. Apparent clearance CL/F (combined IV/oral typical value) varies linearly with body weight via (1 + theta_WT * WT/27.9) where 27.9 kg is the cohort median, and additively with concomitant calcineurin-inhibitor type via (1 + theta_CYTA * CYTA): CYTA = 0 on ciclosporin (the reference; n = 23) and CYTA = 1 on tacrolimus (n = 15) (Zeng 2010 Table 2 model 4). Ciclosporin inhibits MRP2-mediated biliary efflux of MPAG and thus suppresses enterohepatic recirculation of MPA, so paediatric MPA CL/F is approximately 2.5x higher under ciclosporin (CYTA = 0; multiplier 1.0) than under tacrolimus (CYTA = 1; multiplier 1 + (-0.60) = 0.40). Inter-individual variability is diagonal on CL, ka, and F; residual error is exponential (log-normal) with SD 0.48 on the log scale. Inter-occasion variability of 5.8% CV on CL/F reported in Zeng 2010 Table 3 (with occasion defined as 7 days in patients on daily MMF) is NOT encoded structurally here because the model-library use case does not define an operational occasion column; downstream users who want to simulate IOV can add an OCC indicator and a per-occasion eta in rxode2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhao_2010_mycophenolic_acid.html">Mycophenolic acid (Zhao 2010)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for mycophenolic acid (MPA, the active moiety delivered as oral mycophenolate mofetil MMF) in children with idiopathic nephrotic syndrome (Zhao 2010). First-order absorption (ka = 5.16 1/h) with absorption lag time (tlag = 0.215 h) into a central compartment. Apparent oral clearance CL/F (typical value 9.7 L/h at the cohort medians WT = 23.5 kg, ALB = 38.6 g/L) is modeled with two covariates: a power effect of body weight on CL/F with exponent 0.753 referenced to 23.5 kg (close to allometric but estimated, not fixed), and an unusual linear-in-ratio effect of serum albumin in the form CL/F = q1 * (WT/23.5)^q2 * [1 - q3 * (ALB/38.6)] with q1 = 22.5 L/h, q2 = 0.753, q3 = 0.570 (higher serum albumin reduces apparent CL/F, consistent with stronger MPA-albumin binding in nephrotic patients with restored albumin). Apparent central V1/F = 22.3 L; apparent peripheral V2/F was fixed at 250 L (estimation between 100 and 600 L was non-identifiable; the fixed value lies in the range reported for adult transplant cohorts). Apparent inter-compartment clearance Q/F = 18.8 L/h. Exponential inter-individual variability is estimated on lag time, V1/F, Q/F, and CL/F (no IIV on ka or V2/F). A proportional residual error (44.6%) on MPA plasma concentration completes the model. Dosing in this packaged form is in mg of MMF; the MMF-to-MPA hydrolysis is implicit in the apparent bioavailability F. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2014_nab_paclitaxel.html">Nab paclitaxel (Chen 2014)</a> </td> <td style="text-align:left;"> Three-compartment population PK with saturable (Michaelis-Menten) distribution between the central and first peripheral compartments and saturable elimination from the central compartment, coupled with a Friberg-style 5-compartment semi-mechanistic PD model for paclitaxel-induced neutropenia, fit to 150 adult patients with advanced solid tumors who received nab-paclitaxel (Abraxane) 80-375 mg/m^2 as a 30-minute IV infusion (Chen 2014). The first peripheral compartment exchanges with central via the saturable Vmtr / Kmtr process; the second peripheral compartment exchanges via linear intercompartmental clearance Q2. Baseline albumin lowers the maximal elimination rate VMEL via a power-form covariate; advanced age (>= 65 years) potentiates the linear Slope of paclitaxel-driven inhibition of the proliferating neutrophil precursor pool, and baseline albumin also modifies the baseline ANC via a power-form covariate. PK observation is plasma paclitaxel concentration (ug/L = ng/mL); PD observation is absolute neutrophil count (10^9 cells/L). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_guhongInjection.html">NAcetylglutamine rat (Chen 2024)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment intravenous pharmacokinetic model for N-acetyl-L-glutamine (NAG), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the N-acetyl-L-glutamine dose is the GHI volume dose times its content in GHI (30 mg/mL), i.e. 75 mg/kg, 150 mg/kg, 300 mg/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Direct-effect sigmoid-Emax models link the N-acetyl-L-glutamine plasma concentration to the GHI-minus-model-group difference in creatine kinase-MB (CK-MB), ischemia-modified albumin (IMA), cardiac troponin I (cTn I) and alpha-hydroxybutyrate dehydrogenase (alpha-HBDH) (E = Emax * C^gamma / (ED50^gamma + C^gamma); Tables 12, 13, 14, 15). Chen 2024 fitted a PK/PD model only for the analyte/biomarker/dose combinations whose PLSR coefficient was negative, so the effect of an unfitted combination is returned as zero rather than extrapolated. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_nadroparin.html">Nadroparin (Chen 2024)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order subcutaneous absorption and first-order elimination for nadroparin (a low-molecular-weight heparin) in 40 preterm and term neonates and infants under 8 months of age treated for arterial or venous thromboembolic disease (Chen 2024). The model is fitted to plasma anti-Xa activity (IU/mL) rather than to a drug concentration, so the disposition parameters are apparent anti-Xa clearance (CL/F) and apparent anti-Xa volume of distribution (Vd/F). The single retained covariate is Schwartz-formula creatinine clearance on CL/F, entered as a median-normalised power model CL/F = 0.211 * (CRCL / 51.1)^0.238, consistent with nadroparin being cleared predominantly by renal excretion. Inter-individual variability was retained on CL/F only; the variances on Vd/F and ka approached zero during model building and were fixed to zero. Body-weight allometric scaling with a fixed 0.75 exponent and a sigmoid Emax postmenstrual-age maturation function were both tested but were not retained in the final model once creatinine clearance was included. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kyhl_2016_nalmefene.html">Nalmefene (Kyhl 2016)</a> </td> <td style="text-align:left;"> Population PK model for nalmefene in healthy volunteers (Kyhl 2016): two-compartment model with first-order absorption after oral dosing, separate absorption rates for tablet and solution formulations, and a link to mu-opioid receptor occupancy. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Laffont_2024_nalmefene.html">Nalmefene (Laffont 2024)</a> </td> <td style="text-align:left;"> Population PK model for intranasal (IN) nalmefene HCl in healthy adult volunteers (Laffont 2024): two-compartment model with linear elimination, parallel zero-order plus lagged first-order absorption, and allometric body-weight scaling on apparent clearance. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dowling_2008_naloxone.html">Naloxone (Dowling 2008)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for naloxone in six healthy male volunteers receiving 0.8 mg intravenous, 0.8 mg intramuscular, 0.8 mg intranasal, 2 mg intravenous, and 2 mg intranasal doses in an open-label crossover design (Dowling 2008). Intramuscular and intranasal absorption are modeled as first-order via separate depot compartments (Ka_im 0.65 1/h, F_im 0.36; Ka_in 1.52 1/h, F_in 0.038); intravenous doses go directly to the central compartment (F 1, structural anchor). Fat-free mass (Janmahasatian 2005 formula, called LBW2005 in the paper) is allometrically scaled on clearance with fixed exponent 0.75, and body weight is linearly scaled on central volume (exponent 1); both effects use a 70 kg reference. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Laffont_2024_naloxone.html">Naloxone (Laffont 2024)</a> </td> <td style="text-align:left;"> Population PK model for intranasal (IN) naloxone HCl in healthy adult volunteers (Laffont 2024): two-compartment model with linear elimination and parallel zero-order plus lagged first-order absorption; Q/F and Vp/F fixed to literature values from Yassen 2007. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sharma_2018_naltrexone_bupropion.html">Naltrexone bupropion (Sharma 2018)</a> </td> <td style="text-align:left;"> Dose- and time-dependent population pharmacodynamic (DTPD) body-weight model for the naltrexone/bupropion fixed-dose combination (Contrave) in obese and overweight adults under lifestyle intervention, based on 4591 subjects pooled from six Contrave clinical trials (placebo and active-treatment arms). Indirect-response body-weight model with linear NHANES-derived disease progression, inverse-Bateman lifestyle-intervention stimulation of body-weight loss (kout), and a combined Emax dose- and time-dependent inhibitory drug effect; diabetes (T2DM) and race covariates on key parameters. Does not include the linked Markov dropout layer (Tr10, Tr01, Tr12, Tr02) of Table 3, and does not include the PPPD concentration-driven variant (whose underlying naltrexone/bupropion PopPK model is an unpublished internal Takeda report). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bjornsson_2011_naproxcinod.html">Naproxcinod (Bjornsson 2011)</a> </td> <td style="text-align:left;"> Joint population PK / pain intensity (PI) / informative-dropout model for naproxen following oral administration of naproxcinod (a naproxen nitrate ester prodrug), naproxen, or placebo after wisdom-tooth extraction (Bjornsson 2011, 242 patients with moderate-to-severe post-surgical dental pain). PK: one-compartment disposition of unbound naproxen with parallel Savic transit-compartment absorption chains for naproxcinod (MTT 1.77 h, NN 3.58) and naproxen (MTT 0.500 h, NN 4.23) feeding a shared central compartment. Total naproxen is computed from unbound via a saturable albumin-binding equation Ctot = Cu + Bmax * Cu / (Km + Cu) (Bmax = 643 umol/L, Km = 0.549 umol/L). Relative bioavailability of naproxen via naproxcinod vs naproxen is 59.7%. PD: pain intensity on a 100-mm visual analogue scale modeled as PI(t) = PI_baseline * (1 - placebo(t)) * (1 - drug(t)), where placebo(t) = Pmax * (1 - exp(-kpl * t)) (Pmax 20.2%, kpl 0.237 /h; additive IIV on Pmax allows individual PI to either decrease or increase from baseline) and drug(t) is a sigmoid Emax function of unbound naproxen with Emax fixed at 1, EC50 0.135 umol/L, and Hill exponent 1.61. TTE: rescue-medication request modeled as a Weibull hazard (lambda 0.00999, alpha 0.729) with log-linear covariate effects of PI(t) and (PI_baseline - 55) on the slope of PI(t); the hazard is set to zero for t < 1.5 h to reflect the protocol's rescue-medication abstention window. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Boucher_2018_naproxen_mbma.html">Naproxen mbma (Boucher 2018)</a> </td> <td style="text-align:left;"> MBMA. Model-based meta-analysis longitudinal time-course Emax model for the Western Ontario and McMaster Universities (WOMAC) pain score (0-10 scale) in adults with osteoarthritis, fitted to study-arm-mean data from 18 randomized double-blind placebo-controlled trials of naproxen vs placebo (12 flare designs, 6 non-flare). The WOMAC pain response over time follows a three-parameter Emax model in time: pain = E0 + Emax * time / (ET50 + time), where ET50 is the time to half-maximal effect. Flare design shifts both baseline E0 and Emax; naproxen treatment shifts Emax and shortens ET50 (faster onset: ET50 0.21 week vs placebo 0.69 week). Between-study variability is carried as study-arm-level random effects on E0 (SD 0.62) and Emax (SD 0.74); the residual describes study-arm-mean variability weighted by each arm's observed standard error (sigma fixed to 1). Suitable simulation scope is study-arm-mean WOMAC pain time-course, NOT individual-patient pain scores. Parameter values are the NONMEM column of Table 2 (the same model was fit in NONMEM, BUGS, and R with closely agreeing estimates). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Long_2017_necitumumab.html">Necitumumab (Long 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for necitumumab in cancer patients (Long 2017), with IV infusion input and parallel linear plus Michaelis-Menten (target-mediated) elimination from the central compartment and allometric weight scaling on CL, Q, V1, and V2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Djerada_2014_nefopam.html">Nefopam (Djerada 2014)</a> </td> <td style="text-align:left;"> Joint parent + metabolite population PK model for nefopam (2-compartment; IV zero-order infusion into central) and its N-demethyl metabolite desmethyl-nefopam (1-compartment; fed by an apparent metabolic clearance K13 from the nefopam central compartment) in 48 elderly patients (65-99 years) with normal, moderate, or severe renal impairment after a single 20 mg 30 min IV nefopam infusion for postoperative analgesia (hip-fracture repair). Fit in Monolix v4.1 with SAEM. No covariates were retained in the final model despite an extensive screen (age, sex, weight, height, BMI, LBW, FFM, IBW, BSA, iohexol-clearance GFR, MDRD GFR, and Cockroft-Gault GFR); severe renal impairment (GFR <= 30 mL/min) correlated post-hoc with reduced nefopam clearance and higher AUC0-infinity by Spearman test but did not meet the model-building inclusion criteria. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hirt_2006_nelfinavir.html">Nelfinavir (Hirt 2006)</a> </td> <td style="text-align:left;"> Population PK model for oral nelfinavir and its active metabolite M8 (hydroxy-tert-butylamide) in 182 pediatric HIV-1 infected children aged 3 days to 17 years (Hirt 2006). One-compartment model for nelfinavir (depot + central) with first-order absorption (Ka) and linear elimination via apparent total clearance CL_T/F; the active metabolite M8 is described by a single compartment (central_m8) with apparent volume FIXED to 1 L (not identifiable). Only the fraction F_MT (~2.5%) of nelfinavir's total clearance enters M8; the remainder is lost to non-M8 elimination pathways. Body-weight scaling is linear (per-kg parameterisation of V/F and CL/F); both V/F and CL/F decrease with age via a shared power exponent of -0.29 relative to the median age 8.2 years. M8 elimination rate KM0 is increased ~1.9-fold by concomitant administration of an enzyme-inducing NNRTI (efavirenz or nevirapine, pooled under the indicator CONMED_NNRTI_IND consistent with the paper's finding that the two drugs' inducer effects on KM0 were not significantly different and the two were never administered simultaneously). Inter-individual variability on V/F, CL/F, and KM0; correlation 0.45 between IIVs on CL/F and KM0; additive residual error in mg/L for each output. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hirt_2007_nelfinavir.html">Nelfinavir (Hirt 2007)</a> </td> <td style="text-align:left;"> Six-compartment population pharmacokinetic model for nelfinavir and its M8 metabolite describing placental transfer from maternal plasma into umbilical (cord) plasma and amniotic fluid (Hirt 2007). Oral nelfinavir is absorbed first-order with a lag time into the maternal central compartment. Nelfinavir is then (i) eliminated, (ii) converted to M8 in a maternal M8 compartment, and (iii) transferred to a cord nelfinavir compartment. M8 is eliminated from the mother and transferred to a cord M8 compartment. Both nelfinavir and M8 transfer from cord to amniotic fluid and are eliminated from amniotic fluid by first-order rate constants. The distribution volume of M8 in the mother and the volumes of all cord and amniotic-fluid compartments were not estimable and are fixed at 1 L per the paper. Covariate effects: day-of-delivery indicator increases maternal nelfinavir CL and V each by 92 percent and gates a body-weight effect on CL within the delivery cohort only (reference 73 kg, exponent 2.81); pregnancy increases M8 elimination by 67 percent; body weight scales M8 elimination on the full database (reference 63 kg, exponent 1.41); concomitant NNRTI use increases M8 elimination by 148 percent. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2023_nemonoxacin.html">Nemonoxacin (Chen 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral nemonoxacin capsules in Chinese adults, pooled across phase I to III trials (Chen 2023; n = 161 subjects / 195 cases; 2007 plasma concentrations). First-order absorption with a lag time and estimated relative bioavailability. Clearance is an additive linear function of Cockcroft-Gault creatinine clearance (10.3 L/h intercept + 0.026 L/h per mL/min), and allometric body-weight scaling is fixed at 0.75 on CL / Q and 1 on Vc / Vp with a 70 kg reference. Covariate effects: female sex lowers Vc by 11%; community-acquired pneumonia raises Vp by 23%; food slows absorption (ka x 0.44), lengthens the lag time (T_lag x 1.6), and lowers bioavailability (F x 0.88). Inter-occasion variability on CL distinguishes the single-dose occasion from steady state (72 h after the first of multiple doses). The paper's companion PK/PD analysis is an exposure-response logistic regression on AUC0-24/MIC, Cmax/MIC and %T > MIC; its regression coefficients are not reported, so only the population PK layer is encoded here (the published PK/PD targets are reproduced in the validation vignette). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kemal_2026_nemtabrutinib.html">Nemtabrutinib (Kemal 2026)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for nemtabrutinib (oral BTK inhibitor) in adults with hematologic malignancies including CLL/SLL (Kemal 2026, full covariate model) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bienczak_2016_nevirapine.html">Nevirapine (Bienczak 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral nevirapine in African children (Bienczak 2016) with three-transit-compartment absorption, semi-mechanistic well-stirred hepatic extraction (Gordi-style) splitting oral bioavailability into a pre-hepatic component FpreH (age-driven exponential maturation toward an older-child reference fixed to 1) and a hepatic component FH derived algebraically from intrinsic clearance CLint via FH = QH / (QH + fu * CLint), allometric scaling of CLint and Vc to median weight 14.5 kg and of hepatic plasma flow QH to a 70 kg reference, CYP2B6 516G>T | 983T>C metabolizer-status (EM / IM / SM / USM) effects on CLint, and a 29% diurnal-variation cosine on CLint with zenith at noon. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schipani_2011_nevirapine.html">Nevirapine (Schipani 2011)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral nevirapine in HIV-infected adults (Schipani 2011), with CYP2B6 516G>T (rs3745274) and 983T>C (rs28399499) genotype and body-weight covariate effects on CL/F. Covariate effects are additive on linear-scale CL/F per the published equation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Svensson_2012_nevirapine.html">Nevirapine (Svensson 2012)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral nevirapine in HIV-infected South African adults (Svensson 2012, the multi-source 'mega-model' integration paper) with first-order absorption through two transit compartments (Savic 2007 parameterisation, ktr = (NTRANS+1)/MTT shared across all transit-rate steps), a two-population mixture on apparent oral clearance CL/F (fast eliminators 3.12 L/h at 82.7% probability vs slow eliminators 1.45 L/h at 17.3% probability, the slow class associated by the paper Discussion with CYP2B6 516TT homozygotes), Anderson-Holford allometric scaling (fat-free mass at exponent 0.75 for CL/F with reference FFM 42 kg corresponding to a 70 kg / 1.6 m woman, body weight at exponent 1 for V/F with reference WT 70 kg, both exponents fixed), a fed/fasted binary covariate on absorption mean transit time MTT (2.46 h fed vs 0.596 h fasted, a 4.1-fold slowing of absorption with food) and a concomitant tuberculosis-treatment (rifampicin + isoniazid +/- ethambutol) effect on bioavailability F (39% decrease; F = 0.613 when on TB treatment vs F = 1.0 reference, with an additional 34.1% between-subject variability specific to the TB-treatment effect). Residual error is proportional (8.41% CV). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Iida_2008_nicorandil.html">Nicorandil (Iida 2008)</a> </td> <td style="text-align:left;"> Two-compartment IV PK plus inhibitory Emax PD plus asymptotic- exponential disease-progression population PKPD model for nicorandil in acute heart failure (AHF) patients with pulmonary artery wedge pressure (PAWP, mmHg) as the haemodynamic biomarker (Iida 2008 model 10). Five clinical studies pooled: 11 healthy volunteer subjects (concentration only) and 94 AHF patients (concentration plus PAWP), 618 nicorandil and 559 PAWP observations. Allometric size scaling with a 70 kg reference (exponent 0.75 on CL and Q per paper Equation 5; canonical exponent 1.0 on V1 and V2 per Holford 1996). AHF-vs-healthy disease cohort modifies all four PK parameters multiplicatively (FCL = 1.94, FV1 = 1.39, FQ = 0.519, FV2 = 4.06; paper Table 2). PD layer: inhibitory Emax model on plasma concentration (no effect compartment, paper Results paragraph 'Pharmacodynamic analysis'), summed with an asymptotic-exponential disease- progression term that decreases PAWP from a baseline S0 to a steady-state Sss with half-life Tprog. Inhibitory maximum Emax = -11.7 mmHg, EC50 = 423 ng/mL, S0 = 25.6 mmHg, Sss = 19.5 mmHg, Tprog = 5.83 h. Bootstrap median final estimates. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ahlstrom_2010_nicotinicAcid_rat.html">NicotinicAcid rat (Ahlstrom 2010)</a> </td> <td style="text-align:left;"> Preclinical (rat). PK/PD feedback model for nicotinic acid (NiAc) and non-esterified fatty acids (NEFA) in male Sprague-Dawley rats following IV infusions. NiAc disposition is a two-compartment model with two parallel capacity-limited (Michaelis-Menten) elimination processes (likely glycine conjugation and amidation) plus endogenous synthesis. NEFA turnover is described by an inhibitory drug-mechanism function (Hill-Imax, with Imax fixed at 1) acting on the formation of NEFA, coupled to a moderator feedback chain of 8 transit compartments (precursor1..precursor8): the first compartment inhibits NEFA formation amplified by exponent p and the last compartment stimulates NEFA loss. A NiAc-independent capillary release rate kcap sets the lower physiological limit of NEFA in plasma. All structural parameters are body-weight-normalized (per kg). Parameter values from Ahlstrom 2010 Tables 1 (NiAc PK) and 2 (NEFA PD). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Castro-Surez_2020_nimotuzumab.html">Nimotuzumab (Castro-Surez 2020)</a> </td> <td style="text-align:left;"> Semi-mechanistic two-compartment QSS TMDD population PK model for nimotuzumab (anti-EGFR humanized IgG1) in adults with autosomal dominant polycystic kidney disease (Castro-Suarez 2020); EGFR binding represented in both central (Rtot) and peripheral (Rtotp) compartments under quasi-steady-state, plus a turnover mediator that stimulates non-specific clearance via a sigmoid Emax of free central nimotuzumab. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schmid_2017_nintedanib.html">Nintedanib (Schmid 2017)</a> </td> <td style="text-align:left;"> Population pharmacokinetic model of nintedanib and its main hydrolytic metabolite BIBF 1202 (Schmid 2017): a one-compartment first-order absorption + lag parent (nintedanib) jointly fit with a one-compartment first-order absorption + lag metabolite (BIBF 1202) coupled to the parent via a fixed fractional formation-during-elimination term (kmet = CL/V2 * ffM) and a fixed V3/V2 volume ratio inherited from rat IV data. The 1191-patient pooled data set spans four trials in NSCLC (Reck 2011 NSCLC phase II, LUME-Lung 1, LUME-Lung 2) and IPF (TOMORROW). Covariates include allometric body weight on CL, linear age on F1, smoking-status on F1, ethnic-origin composite (Indian/Chinese/Taiwanese vs Korean vs reference) on F1, study-group effects on F1 and ka, and on the metabolite side body weight on F2 with ethnic-origin (Indian alone, non-Indian Asian) on F2, ECOG status, LDH (hockey-stick), study-group effect on ka2, and NSCLC histology on ka2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Valenzuela_2025_nipocalimab.html">Nipocalimab (Valenzuela 2025)</a> </td> <td style="text-align:left;"> Integrated PK/RO/IgG/MG-ADL QSS TMDD model for nipocalimab in healthy adults and generalized myasthenia gravis (Valenzuela 2025) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chan_2023_nirmatrelvir.html">Nirmatrelvir (Chan 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption for nirmatrelvir coadministered with ritonavir 100 mg (PAXLOVID) in adults with and without COVID-19 (Chan 2023; pooled analysis of 8 phase I and phase II/III studies, N = 1237). Allometric baseline body weight on clearances (fixed 0.75) and volumes (fixed 1) referenced to 70 kg; a breakpoint power model for BSA-normalized creatinine clearance on CL (CL scales with nCLCR below the estimated 70.1 mL/min/1.73 m2 breakpoint and is independent of nCLCR at or above it); fractional carbamazepine, itraconazole and COVID-19 effects on CL; a power effect of age on central volume referenced to 45 years; and a relative-bioavailability model combining a dose power function (referenced to 300 mg) with a fractional 150-mg-tablet formulation effect. Combined additive plus proportional residual error with separate proportional magnitudes for the phase I and phase II/III data. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Clegg_2024_nirsevimab.html">Nirsevimab (Clegg 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for nirsevimab in preterm and term infants (Clegg 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2013_nisin_amikacin_linezolid.html">Nisin amikacin (Landersdorfer 2013)</a> </td> <td style="text-align:left;"> In vitro (methicillin-resistant Staphylococcus aureus USA300). Mechanism-based pharmacodynamic model for nisin plus amikacin in a 48-h static-concentration time-kill assay (S-ADAPT and NONMEM analyses; subpopulation synergy concept). Six pre-existing bacterial populations crossing nisin (susceptible Nis-S, intermediate Nis-I, resistant Nis-R) with amikacin (susceptible Ami-S, resistant Ami-R) susceptibility, each following a Bulitta two-state life-cycle growth model (state 1 -> state 2 -> 2*state 1 with replication rate k21 fixed). Nisin kills with a second-order function (k2*Cnis) and amikacin with a saturating Hill function; a saturating carrying-capacity replication factor (REP = 2*CFUmax/(CFUmax + CFUall)) caps the population. Nisin- and amikacin-cross-resistant populations carry reduced biofitness (multiplicative growth-rate factor fk12). Nisin and amikacin concentrations are external time-varying inputs (covariates Cnis and Cami); the model contains no human PK component. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2013_nisin_amikacin_linezolid.html">Nisin linezolid (Landersdorfer 2013)</a> </td> <td style="text-align:left;"> In vitro (methicillin-resistant Staphylococcus aureus USA300). Mechanism-based pharmacodynamic model for nisin plus linezolid in a 48-h static-concentration time-kill assay (S-ADAPT and NONMEM analyses; subpopulation synergy concept). Three pre-existing bacterial populations (Nis-S/Lin-S, Nis-I/Lin-S, Nis-R/Lin-I), each following a Bulitta two-state life-cycle growth model. Nisin kills with a second-order function (k2*Cnis); linezolid inhibits protein synthesis (turnover of a protein pool P), which (i) raises Inh_Rep = 1 - P and therefore reduces successful replication for the Lin-S populations and (ii) inhibits the slow state 1 -> state 2 growth-rate transition via a steep Hill function (Inh_k12) in all three populations. The Nis-R/Lin-I population has no Inh_Rep effect (only Inh_k12). A saturating carrying-capacity replication factor (REP = 2*CFUmax/(CFUmax + CFUall)) caps the population. Nisin and linezolid concentrations are external time-varying inputs (covariates Cnis and Clin); the model contains no human PK component. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sharma_2023_nitrofurantoin.html">Nitrofurantoin human pbpk (Sharma 2023)</a> </td> <td style="text-align:left;"> PBPK (whole-body, GNU MCSim 6.1, age-dependent physiology). Nitrofurantoin (NFT) disposition in human adults -- Sharma 2023 'Model V5a' extrapolated across species WITHOUT further calibration, the paper's final human model. Same thirteen mass-balance amount states and same structure as the rabbit and rat siblings (perfusion-rate-limited distribution, glomerular filtration + saturable active tubular secretion - first-order tubular reabsorption, Michaelis-Menten hepatic metabolism, saturable hepatobiliary efflux with enterohepatic recirculation). Three things are human-specific. (1) Physiology is not a fixed reference adult: height, body weight, cardiac output, liver / kidney / fat / plasma volumes and liver / kidney / fat blood flows are all AGE- and SEX-dependent polynomials, so AGE and SEXF are genuine covariates that reshape the whole body. (2) Renal and metabolic parameters are allometrically scaled from the RABBIT fit and the enterohepatic-recirculation and gut-absorption parameters from the RAT fit, each as parameter * (BW_species / BW_human)^0.25. (3) Renal function is an explicit input: CRCL carries the subject's absolute glomerular filtration rate in mL/min, which lets the model reproduce the paper's renal-insufficiency analysis (Figure 7) showing that a fall from normal to severely compromised GFR raises plasma Cmax ~1.3-fold, trough ~2-fold and AUC ~1.3-fold while substantially cutting urinary NFT delivery -- simultaneously raising hepatic exposure (a DILI risk) and undercutting efficacy at the bladder. Between-subject variability is a lognormal geometric SD of 1.17 (~16% CV) applied to all 18 drug-specific parameters, as in the paper's Monte-Carlo simulations; the model has no human-fitted parameters at all, so every human prediction is a genuine extrapolation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sharma_2023_nitrofurantoin.html">Nitrofurantoin rabbit pbpk (Sharma 2023)</a> </td> <td style="text-align:left;"> Preclinical (rabbit). PBPK (whole-body, GNU MCSim 6.1, Bayesian MCMC). Nitrofurantoin (NFT) disposition in rabbits -- Sharma 2023 'Model V5a', the paper's final structure. Thirteen mass-balance amount states: gut lumen, gut tissue, liver, cumulative hepatic metabolism, bile, cumulative faeces, kidney, renal tubular filtrate, pre-void urine storage, fat, lumped rest of body, plasma, cumulative urine. Perfusion-rate-limited tissue distribution with a single combined plasma pool (no arterial/venous split); only unbound drug (fu/Kbp, divided by each tissue:plasma partition coefficient) distributes. Renal handling is deliberately non-linear: passive glomerular filtration plus SATURABLE active tubular secretion out of the kidney into the tubules, minus first-order tubular reabsorption from the tubules back into the kidney -- the combination the paper shows is required to reproduce the observed fall in fractional urinary recovery from ~60% to ~10% over a 30-fold IV dose increase. Hepatic elimination is Michaelis-Menten metabolism plus a second saturable hepatobiliary efflux into bile, which returns to the gut lumen at a first-order rate (enterohepatic recirculation, EHR). Renal parameters were calibrated on rabbit IV plasma and urine data (model V4); the EHR parameters are the rabbit arm of a hierarchical rat+rabbit fit (model V5a), and are small in rabbits (biliary excretion < 1% of dose), so EHR mainly adds late-phase biphasic plasma behaviour rather than changing urinary recovery. Deterministic typical-value simulator: the MCMC posterior describes PARAMETER UNCERTAINTY, not between-subject variability, so no IIV is encoded (see the vignette Errata). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sharma_2023_nitrofurantoin.html">Nitrofurantoin rat pbpk (Sharma 2023)</a> </td> <td style="text-align:left;"> Preclinical (rat). PBPK (whole-body, GNU MCSim 6.1, Bayesian MCMC). Nitrofurantoin (NFT) disposition in rats -- Sharma 2023 'Model V5a', the paper's final structure. Structurally IDENTICAL to the rabbit sibling (see modellib('Sharma_2023_nitrofurantoin_rabbit_pbpk')): thirteen mass-balance amount states with perfusion-rate-limited distribution, non-linear renal handling (glomerular filtration + saturable active tubular secretion - first-order tubular reabsorption), Michaelis-Menten hepatic metabolism, and saturable hepatobiliary efflux feeding an enterohepatic recirculation loop. What differs is (a) rat physiology and (b) the rat arm of the hierarchical enterohepatic-recirculation fit. The renal and metabolic parameters were NOT refitted to rat data -- they are the rabbit model V4 estimates carried over unchanged and rescaled by rat body weight, which is what makes the rat plasma and urine predictions a genuine cross-species test rather than a fit (Results 3.3: 'the fact that the model was extrapolated from rabbits by adapting to known rat physiology and that biochemical parameters were scaled just based on rat body weight provides strong confidence'). The rat is the species that supplied the biliary-excretion data, so its hepatobiliary efflux capacity is ~24-fold higher than the rabbit's and enterohepatic recirculation is a real feature of the rat plasma profile (visible late-phase biphasic kinetics). The rat EHR estimates are also the ones the human model inherits by allometric scaling. Deterministic typical-value simulator: the MCMC posterior describes PARAMETER UNCERTAINTY, not between-subject variability, so no IIV is encoded (see the vignette Errata). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bajaj_2017_nivolumab.html">Nivolumab (Bajaj 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for nivolumab (anti-PD-1 IgG4) with time-varying clearance (sigmoid Emax) in patients with advanced solid tumors (Bajaj 2017) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2019_nivolumab.html">Nivolumab (Zhang 2019)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with time-varying clearance for intravenous nivolumab (anti-PD-1 IgG4) in adults with advanced solid tumors, alone or in combination with ipilimumab or chemotherapy (Zhang 2019) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Oualha_2014_norepinephrine.html">Norepinephrine (Oualha 2014)</a> </td> <td style="text-align:left;"> Population PK/PD model for continuous IV norepinephrine in hypotensive critically ill children (Oualha 2014). One-compartment open PK with first-order elimination plus an endogenous zero-order production rate q0 and circulating-volume-anchored Vc = 0.08 * WT; allometric scaling of CL and q0 on body weight (exponents fixed to 3/4). Emax PD sub-model on mean arterial pressure (MAP) with a power-of-postmenstrual-age effect on basal MAP0 and a categorical organ-dysfunction-count effect on the maximal drug-induced MAP increase dMAP (32 mmHg for <=3 dysfunctions vs 12 mmHg for >=4). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_noscapine.html">Noscapine (Chen 2024)</a> </td> <td style="text-align:left;"> Semi-mechanistic population PK model for oral noscapine in healthy adults (Chen 2024) with zero-order release into a four-compartment transit chain, an explicit liver compartment that carries the profound hepatic first-pass extraction (apparent clearance CL/F leaves the liver, not the plasma), and three-compartment systemic disposition. Liver plasma flow Qh (55 L/h) and liver volume Vh (1.5 L) are held fixed at physiologic values. Apparent clearance carries a CYP2C9 genotype-predicted-phenotype effect on three levels keyed to the CPIC activity score (extensive metabolizer reference 958 L/h; intermediate metabolizer with activity score 1.5, 531 L/h; poor and intermediate metabolizers with activity score 1.0, 343 L/h) plus a total-body-weight power term (exponent 1.34, 77.3 kg reference); inter-compartmental clearance to the first peripheral compartment carries an age power term (exponent 0.348, 29 year reference). Relative bioavailability of the reformulated test suspension is 82.8% of the reference suspension. Correlated inter-individual variability on absorption duration, CL/F, Vc/F and F1; inter-occasion variability on the transit rate and on Qp1/F across the two crossover periods; exponential residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Biliouris_2018_nusinersen.html">Nusinersen (Biliouris 2018)</a> </td> <td style="text-align:left;"> Semimechanistic nine-compartment population PK model of nusinersen (antisense oligonucleotide, Biogen ISIS 396443) intended for extrapolation of nonhuman-primate (cynomolgus monkey) fits to paediatric patients with spinal muscular atrophy, following intrathecal lumbar-puncture bolus administration. Structure (Biliouris 2018): CSF is the sampled dosing site, coupled bidirectionally to three anatomically distinct spinal-cord segments (cervical, thoracic, lumbar), to the brain (with a deeper brain-tissue redistribution compartment), and to the pons; CSF also drains one-way into the plasma / central compartment, which has its own peripheral distribution and first-order elimination. Volumes scale linearly with body weight relative to a 2.8 kg reference; rate constants scale as (WT/2.8)^(-0.08); the CSF physiological volume V_CSF is overridden by a stepwise paediatric-age function (120, 130, 135, 140 mL for age classes <0.25, 0.25-0.5, 0.5-1, 1-2 years) rather than by WT. Residual error is proportional per observed matrix; twelve exponential-IIV terms are carried on the elimination and transfer rate constants and on the CSF / plasma volumes. All THETA / OMEGA / SIGMA values are held FIXED from the paper's final NONMEM simulation control stream on file. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Luu_2017_nusinersen.html">Nusinersen (Luu 2017)</a> </td> <td style="text-align:left;"> Four-compartment population PK model for nusinersen (antisense oligonucleotide) following intrathecal administration in pediatric patients with spinal muscular atrophy (Luu 2017): a CSF + CNS-tissue subsystem (intrathecal bolus enters the CSF) coupled by a unidirectional CSF-to-plasma transport to a plasma + systemic-tissue subsystem, with baseline body weight as a power covariate on CL_p and V_CSF and a linear covariate on V_p. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2011_nutlin3a.html">Nutlin3a (Zhang 2011)</a> </td> <td style="text-align:left;"> Preclinical (mouse). Whole-body PBPK model for nutlin-3a (MDM2 inhibitor) in adult C57BL/6 mice after intravenous and oral administration (Zhang et al. 2011, DMD). Thirteen physiological tissue compartments (adipose, adrenal gland, bone marrow, brain, intestine + lumen, liver, lung, muscle, retina, spleen, vitreous, residual diffusion-limited tissue + residual vascular space) with arterial and venous blood pools (75/25 split of total blood volume). Perfusion-limited tissues use a partition coefficient K_i; the eye is modelled as retina + vitreous coupled by a permeability-surface-area product PA_VIT; the residual compartment is diffusion-limited (5% vascular space, 95% tissue, coupled by PA_RES). Elimination is combined linear (hepatic, k_e) and saturable Michaelis-Menten (arterial, V_max/K_m). Oral absorption is first-order from an intestinal lumen depot. Plasma protein binding is reported (B_max = 286 uM, K_A = 0.085 1/uM, Langmuir form) but is only used for an unbound-concentration derivation that the paper applies to tissue exposure / IC50 comparisons, not to the elimination ODEs; the ODE system operates on total concentrations. The model is intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nagy_2017_obiltoxaximab.html">Obiltoxaximab (Nagy 2017)</a> </td> <td style="text-align:left;"> Preclinical and clinical (NZW rabbit, cynomolgus macaque, human). Joint two-compartment population PK model for obiltoxaximab, the anti-protective-antigen chimeric IgG1(kappa) monoclonal antibody approved for inhalational anthrax under the US FDA Animal Rule. One file carries the three species-specific parameter sets used for animal-to-human dose translation (Nagy 2017 Supplementary Table S1), selected by the SPECIES_RABBIT / SPECIES_MACAQUE indicators with human as the reference species. Animals additionally have first-order intramuscular absorption (Ka, F1), and anthrax-infected subjects carry a parallel Michaelis-Menten elimination arm that approximates protective-antigen target-mediated drug disposition. For infected humans the macaque nonlinear-clearance component is carried over and allometrically scaled to human body size, exactly as the paper does to project infected-human exposures. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nagy_2017_obiltoxaximab.html">Obiltoxaximab survival (Nagy 2017)</a> </td> <td style="text-align:left;"> Preclinical (NZW rabbit + cynomolgus macaque). Weibull cure-rate (mixture) survival model for obiltoxaximab treatment of inhalational anthrax, fit simultaneously to infected rabbit and macaque survival data and used to justify the 16 mg/kg human dose under the US FDA Animal Rule. The cure fraction psurv (probability of surviving to the Day-28 end of study) follows an Emax dose-response on the logit scale with an exponentiated log10 prior-to-treatment (PTT) bacteremia term; the Weibull death-rate lambda for the non-cured fraction is log-linear in log10 PTT bacteremia. Species was investigated as a covariate but is not retained in the final model. The model is algebraic and deterministic (no ODE state, no IIV, no residual error): it returns the survivor function directly. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gibiansky_2014_obinutuzumab.html">Obinutuzumab (Gibiansky 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of obinutuzumab (GA101, glycoengineered type II anti-CD20 mAb) in adults with chronic lymphocytic leukemia (CLL) or non-Hodgkin lymphoma (NHL); clearance is the sum of a time-independent component CL_inf and a mono-exponentially decaying time-dependent component CL_T*exp(-cl_exp_kdes*time), with histology (CLL / BCL / DLBCL / MCL), baseline tumor size, body weight, and sex as covariates (Gibiansky 2014). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yu_2022_ofatumumab.html">Ofatumumab (Yu 2022)</a> </td> <td style="text-align:left;"> Population PK / B-cell-count model for subcutaneous ofatumumab in adults with relapsing multiple sclerosis (Yu 2022) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chigutsa_2012_ofloxacin.html">Ofloxacin (Chigutsa 2012)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral ofloxacin in South African adults with multidrug-resistant tuberculosis (MDR-TB) (Chigutsa 2012; n = 65; pooled Cape Town and Durban cohorts). Savic 2007 transit-compartment absorption chain (number of transit compartments NN = 6 estimated). Total apparent oral clearance is an additive sum of two routes: a glomerular-filtration component scaling linearly with creatinine clearance (CrCl computed by a lean-body-weight modification of the Cockcroft-Gault equation; reference 68 mL/min), and an extraglomerular component (active tubular secretion + minor biliary excretion) allometrically scaled to total body weight (exponent 0.75 fixed, reference 70 kg). Central volume is allometrically scaled to lean body mass (exponent 1 fixed, reference 46 kg LBM); peripheral volume to total body weight (exponent 1, reference 70 kg); intercompartmental clearance to total body weight (exponent 0.75, reference 70 kg). Mean transit time is 2.4-fold longer when ofloxacin is administered after a meal (Cape Town cohort, FED = 1) than fasted (Durban cohort, FED = 0). F is fixed at 1; residual error is combined additive (0.6 mg/L) and proportional (9.6%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PillaReddy_2013_panss_subscales.html">Olanzapine panss subscales (PillaReddy 2013)</a> </td> <td style="text-align:left;"> Population PK/PD model for olanzapine against the three PANSS subscales (positive, negative, general) in adults with schizophrenia from Pilla Reddy 2013 Part II. The PK sub-model is the one-compartment olanzapine structural model from Part I (PMID 23473810) Table 2: first-order absorption ka = 0.30 1/h, apparent oral clearance CL/F = 21.8 L/h, apparent central volume of distribution Vc/F = 700 L. The PD sub-model has three outputs that share the Weibull placebo time- course form Pplacebo = Pmax * (1 - exp(-(t/TD)^POW)) but each subscale carries its own placebo Pmax, TD, POW (Part II Table 1) and olanzapine's own Emax / EC50 / KT triplet per subscale (Part II Table 2). The KT for olanzapine PANSS positive and general (0.048 and 0.035 1/day) is the common-across-atypical-antipsychotic value (Part II Methods); the KT for the negative subscale (0.028 1/day) was estimated separately per drug. Olanzapine was numerically superior to the other SGAs for the negative subscale (Emax = 0.33 vs 0.14-0.17 for the other SGAs; Part II Results), making this a clinically meaningful subscale- specific comparison. The exponential time-to-event dropout sub-model from Part II Table 4 is documented in population$dropout_model but is not encoded in this model body.
Olanzapine rat (Johnson 2011) Preclinical (rat). Mechanism-based hybrid physiology-based population PK-PD model for olanzapine and striatal dopamine D2 receptor occupancy (D2RO) in rats (Johnson 2011). Plasma PK is a 2-compartment model fitted across IP, SC, and IV routes in Wistar / Sprague-Dawley rats (single dose 0.01-40 mg/kg, pooled from 12 studies, n = 283); the absorption rate constant was not estimable, so all routes deposit drug directly into the central compartment, and the intraperitoneal bioavailability FIP is estimated (about 64%) with an 87% CV log-normal IIV. SC and IV bioavailability are fixed at 1. The brain submodel adds a brain-vascular compartment (Vbv, fed by cerebral blood flow CLbv from systemic central) and a brain- extravascular compartment (Vbev, fed across the BBB by an estimated clearance CLbev applied to the unbound concentration on each side via fixed fu_plasma and fu_brain). D2 receptor occupancy in striatum is the reduced model published by the authors (Bmax dropped per their sensitivity analysis): dRO/dt = kon * Cfree_bev * (1 - RO) - koff * RO, with kon = koff/Kd and Cfree_bev = fu_brain * (Cbev in nM), so the binding kinetics are driven by the free brain-extravascular concentration converted to nM via the olanzapine molecular weight (312.43 g/mol). All structural parameters are body-weight-normalised (per kg).
Olaratumab (Mo 2018) Two-compartment population PK model with linear clearance for olaratumab in patients with advanced or metastatic cancer (Mo 2018)
Olodaterol (Borghardt 2016) Population PK model for inhaled and intravenous olodaterol (long-acting beta-2-adrenergic receptor agonist) in 148 healthy adult volunteers from three Phase I trials (Borghardt 2016). Four-compartment systemic disposition (central + 3 peripheral) fitted to IV plasma + urine data, with two parallel first-order elimination processes from the central compartment: renal (cl_renal) and nonrenal (cl_nonren). For inhaled administration via the Respimat inhaler, three parallel first-order absorption depots (slow, intermediate, fast) feed the central compartment, with absorption half-lives of 21.8 h, 2.00 h, and 0.268 h respectively. The pulmonary bioavailable fraction (49.4% of the nominal ex-mouthpiece dose) is split across the three depots by two logit-transformed proportionality parameters. Smoking is a covariate on the slow and fast absorption rate constants (active smokers vs ex-smokers and never-smokers pooled). Systemic disposition parameters were estimated from IV data and fixed when fitting the inhalation data.
Olokizumab (Kretsos 2014) Two-compartment population PK with linear elimination and SC first-order absorption (depot, central, peripheral1) plus effect-compartment fractional sigmoid Imax PD model for C-reactive protein (CRP) suppression in mild-to-moderate rheumatoid arthritis patients receiving single-dose IV or SC olokizumab (anti-IL-6 monoclonal antibody, IgG4, CDP6038). Final-analysis estimates from Kretsos et al. 2014 Table 1 (Final column), pooling first-in-human (healthy volunteers, Hickling 2011) and first-in-patient (Cohorts 1+2, n=27 active-treatment subjects) data. The PK observation model adds a per-subject endogenous anti-IL-6 baseline (‘endo’) as an additive offset on the observed OKZ concentration. Body weight was reported as a significant covariate on CL and central volume (paper Discussion) but its functional form / exponents were not reported in main text or supplement; the body-weight covariate effect is omitted here – see vignette Assumptions.
Olprinone (Kunisawa 2014) Two-compartment intravenous population PK model for olprinone (a phosphodiesterase III inhibitor) in healthy adult Japanese male volunteers with body-weight normalization on CL, Vc, Q and Vp (Kunisawa 2014)
Omalizumab (Hayashi 2007) Mechanism-based binding population PK/PD model for omalizumab and IgE in Japanese atopic-asthma patients (Hayashi 2007). Three serum entities (free omalizumab, free IgE, and the omalizumab-IgE complex) each carry their own clearance and volume of distribution and are coupled through instantaneous-equilibrium binding (law of mass action) with a concentration-dependent dissociation constant. Body weight modifies omalizumab CL and Vd; baseline IgE modifies IgE CL and IgE production rate. Subcutaneous absorption is first-order. Disposition parameters are reported as apparent (divided by SC bioavailability f). Three observed quantities: total omalizumab (ug/mL), total IgE (ng/mL), and free IgE (ng/mL).
Omalizumab (Lowe 2009) Mechanism-based binding population PK/PD model for omalizumab and free / total IgE in 1928 patients (1781 with severe persistent allergic asthma across four Phase III studies plus 152 healthy atopic volunteers in a single-dose bioequivalence study; Lowe 2009). Three serum entities (free omalizumab, free IgE, and the omalizumab-IgE complex) each carry their own clearance and apparent volume of distribution and are coupled through instantaneous-equilibrium binding (law of mass action) with a baseline-IgE-dependent dissociation constant that further varies with the instantaneous total-omalizumab-to-total-IgE molar ratio. Body weight modifies all clearances, all volumes, and the IgE production rate via allometric power covariates centred at 70 kg; baseline IgE modifies CL of free IgE, IgE production rate, and Kd via power covariates centred at 365 ng/mL. Subcutaneous absorption is first-order. Disposition parameters are reported as apparent (divided by SC bioavailability f). Extends Hayashi 2007 (modellib(‘Hayashi_2007_omalizumab’)) with (i) IIV on Kd, (ii) baseline IgE as a covariate on Kd, and (iii) bodyweight covariates on the IgE production and clearance parameters. Three observed quantities: total omalizumab, total IgE, and free IgE (all in ng/mL).
Omalizumab (Marcantonio 2022) QSP. One-compartment monospecific anti-ligand mechanistic PKPD model of omalizumab-IgE binding in adults with moderate-to-severe allergic asthma (Marcantonio 2022 Early Feasibility Assessment, Case Study 3). Omalizumab binds soluble IgE; model treats IgE as the ligand and FcepsilonRI (high-affinity IgE receptor on basophils and mast cells) as the cognate receptor. Bivalent binding (crystallography suggests 2:1 stoichiometry with some 2:2 and 3:3 populations per Davies 2017). Parameters FIXED from paper Table S6; the receptor half-life used is the Assess default of 15 min rather than the Table S6 alternative of 2 hr (paper explicitly tests both).
Omalizumab pediatric (Zhu 2023) Population pharmacodynamic indirect-response (IDR Type IV) model for forced expiratory volume in 1 second (FEV1, percent predicted on a 0-1 fractional scale) driven by serum free IgE in pediatric patients (6-11 years) with moderate to severe persistent inadequately controlled allergic asthma treated with omalizumab (Zhu 2023). PD-only – free IgE concentration enters as the exogenous time-varying covariate IGE_FREE (per-FEV1-observation interpolated value, ng/mL). Adapted from the adult/adolescent IgE-FEV1 model (Lowe et al. 2009) with simplifications for the sparser pediatric data: observed baseline FEV1 per subject, common IIV magnitude on Imax and FEV1max, and Hill coefficient gamma fixed at 9. The estimated IC50 in pediatrics (39.4 ng/mL) is higher than in adults/adolescents (19.8 ng/mL); the model still supports the 25 ng/mL free IgE target underpinning the Xolair dosing table.
Ombitasvir (Mensing 2017) One-compartment population PK model for oral ombitasvir in HCV genotype-1 infected adults receiving the 3D (paritaprevir/ritonavir + ombitasvir + dasabuvir) +/- ribavirin regimen (Mensing 2017). First-order absorption, linear elimination, combined proportional + additive residual error, IIV on CL/F only. The author’s final model retained cirrhosis, gender, age, and body weight as significant covariates on CL/F (and age, body weight on Vc/F), but the paper does not publish point estimates for these covariate coefficients (only graphical exposure-ratio forest plots in Figure 2); the implemented model is the structural typical-value model with covariate coefficients omitted (documented in covariatesDataExcluded).
Omega3PUFA (Wang 2024) One-compartment population PK model with first-order oral absorption for omega-3 polyunsaturated fatty acids (PUFA) in community-dwelling adults with type 2 diabetes, with an HDL-cholesterol power covariate on Ka, V and CL, coupled to a direct-effect inhibitory Imax model for glycosylated hemoglobin (HbA1c) (Wang 2024)
Omeprazole (Solana 2014) Two-compartment intravenous-infusion population PK model for omeprazole in 40 critically ill children (Solana 2014), with fixed Anderson-Holford allometric body-weight scaling on all four disposition parameters (exponents 0.75 on CL and Q, 1.00 on Vc and Vp; reference 70 kg). Between-patient variability was retained on CL only; residual error is proportional.
Omeprazole (Zhao 2018) Population PK-pharmacogenetic model for oral omeprazole and its two metabolites 5-hydroxy-omeprazole and omeprazole sulfone in Caucasian neonates and young infants (Zhao 2018). One-compartment parent disposition with first-order absorption (Ka modulated by ABCB1 C3435T genotype) is followed by parallel formation into two one-compartment metabolites with apparent volume V_M/F fixed to 1 L; the omeprazole-to-5-hydroxy-omeprazole formation clearance (CLOMZ-M1) is modulated by CYP2C19 metabolizer phenotype (poor / intermediate / extensive-or-ultrarapid) and a postnatal-age power function, while the omeprazole-to-omeprazole-sulfone formation clearance (CLOMZ-M2) and the metabolite apparent eliminations carry no covariates. Linear omeprazole elimination was estimated as negligible (< 0.0001 L/h) and is therefore not included in the final structural model.
Ondansetron (deAlwis 1998) Two-compartment population PK model with zero-order intravenous-infusion input for ondansetron in pooled paediatric, young-adult, elderly, and aged subjects (de Alwis 1998). The paper uses an empirical additive linear-regression covariate model in the 1990s NONMEM tradition (Maitre 1991 three-step approach): clearance CL and inter-compartmental clearance CLd are sex-stratified with separate male and female intercepts and slopes; the central volume V1 has a body-weight slope only; the steady-state volume Vss has body-weight and age slopes; the peripheral volume Vp is derived as Vss - V1. Inter-individual variability is diagonal log-normal on CL, V1, Vss, and CLd. Proportional residual error is stratified across five paper-defined study sub-populations (young healthy volunteers 18-41 y, elderly healthy volunteers 61-75 y, aged healthy volunteers >= 75 y, paediatric cancer patients receiving chemotherapy, paediatric patients receiving general anaesthesia), switched at runtime via the canonical AGE / DIS_HEALTHY / DIS_CANCER_PED covariates.
Ontamalimab (Wang 2020) Two-compartment population PK model for ontamalimab (SHP647), a fully human IgG2 anti-MAdCAM-1 monoclonal antibody, in adults with moderate-to-severe ulcerative colitis or Crohn’s disease (Wang 2020), with first-order SC absorption, absorption lag time, parallel linear and Michaelis-Menten elimination from the central compartment, and allometric weight scaling on CL, Vc, Q, Vp, and Vmax.
Orin1001 (Li 2024) Two-compartment population PK model with first-order absorption and a lag time for oral ORIN1001 in Chinese patients with advanced solid tumors
Ormutivimab (Zhang 2022) Time-dependent population pharmacodynamic emax model for rabies virus neutralizing antibody (RVNA) activity after rabies vaccination in healthy Chinese adults, with a categorical drug-product covariate that contrasts Ormutivimab (rHRIG, a recombinant human anti-rabies IgG1 monoclonal antibody) against plasma-derived human rabies immunoglobulin (HRIG) (Zhang 2022). Output Cc is neutralizing antibody activity in IU/mL measured by the rapid fluorescent focus inhibition test (RFFIT). The published Y1 two-compartment PK overlay for the passive-antibody component of the combined drug+vaccine groups (E = Y1 + Y2) is NOT included here because the seven structural PK constants (Ka, V1, V2, K10, K12, K21, C0) are not reported anywhere on disk; see the vignette’s Assumptions and deviations section for the omitted-component audit trail.
Ornidazole (Li 2023) One-compartment intravenous population PK of ornidazole in breastfeeding women after caesarean section, with breast-milk concentration linked to plasma by an estimated milk-to-plasma concentration ratio that rises as a power of time postpartum. Apparent clearance decreases with total bilirubin.
Oseltamivir (Chairat 2016) Joint population pharmacokinetic model for oral oseltamivir (parent) and its active antiviral metabolite oseltamivir carboxylate in 12 obese (BMI >= 30 kg/m^2) and 12 non-obese (BMI < 30 kg/m^2) healthy Thai adult volunteers (Chairat 2016 BJCP). First-order absorption (ka) into a one-compartment parent (OS) disposition, an intermediate metabolism compartment delaying carboxylate appearance (rate km), and a one-compartment oseltamivir carboxylate (OC) disposition. Relative oral bioavailability F is fixed to unity with interindividual variability on F absorbing absorption differences. Creatinine clearance computed using Janmahasatian fat-free mass (CLCR(FFM); raw Cockcroft-Gault with FFM substituted for total body weight) is a linear covariate on CL/FOC, centred at the population median CLCR(FFM) of 73 mL/min (3.84% increase per 10 mL/min increase). Obesity itself was not a retained covariate in the formal model. Residual error is additive on log-transformed concentrations of OS and OC (encoded here as a log-normal residual on Cc and Cc_oselcarb).
Oseltamivir (Kamal 2013) Joint parent-metabolite population PK model for oral oseltamivir (prodrug, OP) and its active metabolite oseltamivir carboxylate (OC) in 390 subjects aged 1 to 78 years pooled from 13 clinical trials (healthy adults, influenza-inoculated and naturally infected adults, healthy geriatric subjects, renally impaired adults, and healthy and infected pediatric subjects 1 to 18 years). Oseltamivir is described by a two-compartment model with first-order absorption and first-order conversion to OC (CLp/F treated as the OP-to-OC conversion clearance under the assumption of complete metabolism; <5% of prodrug is excreted unchanged renally). OC is described by a one-compartment model with first-order elimination. All clearance and volume terms are apparent (conditioned on oral bioavailability F; OC terms additionally on the fraction metabolized fm, assumed 1). Covariates: body weight as a power function on OP CLp/F, OC CLm/F, and OC Vcm/F (allometric-style exponents estimated, not fixed); creatinine clearance (BSA-normalized to 1.73 m^2) as a power function on OC CLm/F; and age as a linear (additive) term on OC Vcm/F. Inter-individual variability is exponential on all seven structural parameters, with two off-diagonal covariances (CLp/F with CLm/F, and Vp/F with Vcm/F). Residual error is proportional only for oseltamivir (40.5% CV reduced CCV model) and combined additive plus proportional for OC (14.0% CV proportional + 17.9 ng/mL additive SD).
Oseltamivir (Kamal 2015) Mechanistic drug-disease (viral-dynamics) model of influenza-virus progression and oseltamivir antiviral effect in adults with experimental and naturally-acquired influenza A (H1N1) virus infection (Kamal 2015). Builds on the Baccam et al. (2006) target-cell-limited viral-dynamics framework: uninfected target respiratory epithelial cells (target_cells) are infected by free virus (virus) at second-order rate beta_inf; infected cells (infected_cells) produce virus at rate p_prod per cell per day and die at rate delta_clr; free virus is cleared at rate c_clr. Oseltamivir inhibits viral production through an inhibitory Hill function acting on log10(p) (Equation 4 of Kamal 2015), parameterised so Emax is the maximum log10-fold reduction of p and ED50 is the dose producing a 2-fold (50%) reduction of p on the linear scale. Dose enters via the per-record DOSE covariate (mg per administered oseltamivir dose; 0 during placebo or outside the treatment window); no oseltamivir pharmacokinetics are modelled. Initial conditions are fixed per Baccam et al. (2006): target_cells(0) = 4e8 epithelial cells (from a 160 cm^2 upper-respiratory-tract surface area and 2e-11 to 4e-11 m^2 per epithelial cell), infected_cells(0) = 0, and virus(0) = 10^0.25 TCID50/mL (the viral-titer lower limit of quantification, used as the inoculation viral titer). The viral load viralLoad (TCID50/mL of nasal wash, canonical PD-output name) is the single observed output with proportional residual error, equivalent to the paper’s log10-transformed additive-error model. The three viral-dynamics compartments are declared paper-specific (see paper_specific_compartments).
Oseltamivir (Standing 2012) Population PK model for oral oseltamivir and its active metabolite oseltamivir carboxylate in preterm and term neonates and infants (Standing 2012). One-compartment parent + one-compartment metabolite with first-order absorption, an empirical transit compartment delaying first-pass metabolite appearance, well-stirred-model hepatic first-pass conversion (FM derived from CLI / liver-blood-flow FQ), and physiologically scaled clearances combining (WT/70)^0.75 allometry with a Rhodin 2009 renal-maturation Hill sigmoid on CLU/CLM and a fitted HCE1 Hill sigmoid (PM50 86.1 wk, Hill 3.17) on intrinsic clearance CLI. Volumes (VD, VDM) and liver blood flow (FQ) fixed from external references.
Oseltamivir pbpk (Luo 2024) PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). Joint oseltamivir + oseltamivir carboxylate disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Oseltamivir is an inactive ester prodrug hydrolysed by hepatic CES1 to the antivirally active oseltamivir carboxylate, which is eliminated renally by combined glomerular filtration and tubular secretion (both CLint,K values exceed GFR, Section 3.1.6). The semi-PBPK circuit is stomach, three small-intestinal lumen segments (duodenum / jejunum / ileum), the matching three gut-wall segments, portal vein, liver, kidney and a one-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper’s virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded.
Oseltamivircarboxylate pbpk (Granda 2024) PBPK (mechanistic kidney, 35 states). Individualised prediction of oseltamivir carboxylate renal clearance from per-subject biomarker measurements of kidney blood flow and OAT1/3-mediated tubular secretory capacity, on top of tracer-measured GFR (Granda 2024, Clin Transl Sci). The kidney is resolved into 11 longitudinal subsegments (proximal tubule S1-S3, descending and ascending loop of Henle, distal tubule, and five collecting-duct subsegments), each with a tubular lumen, a tubular epithelial cell and a peritubular blood subspace (33 states), plus a systemic blood compartment and a bladder. Mechanisms: unbound glomerular filtration, OAT1/3-mediated active secretion in the proximal tubule only, pH-dependent bidirectional passive diffusion along the whole nephron, and CKD tubular-flow adaptation. The structural equations, segment volumes, surface areas, tubular pH profile and adaptive flow factors are the Huang & Isoherranen 2018 / 2020 framework; Granda 2024 replaces the framework population defaults for kidney blood flow and secretory clearance with per-subject measured / fitted values. Predicts each subject’s oseltamivir carboxylate renal clearance from their measured GFR and kidney blood flow plus the kynurenic-acid-derived secretory capacity, scaled by 0.038. Reproduces the Predicted OC CLr column of Table 4. Deterministic typical-value model: the paper reports no IIV and no residual error.
Osimertinib (Brown 2017) Joint two-compartment population PK model for osimertinib (AZD9291) and its active metabolite AZ5104 in advanced non-small cell lung cancer (NSCLC) patients pooled with healthy volunteers (Brown 2017). First-order oral absorption into a parent (osimertinib) compartment is followed by a second compartment (AZ5104) in series; the fraction of parent eliminated as AZ5104 is fixed at 0.25 per the publication. Body weight (allometric on parent CL/F and Vc/F and on AZ5104 CL/F), serum albumin (power on parent Vc/F), healthy-volunteer disease state (linear factor on both parent and AZ5104 CL/F), and ethnicity (Chinese, Japanese, Asian-other, and non-Asian non-Caucasian linear factors on AZ5104 CL/F) were retained as significant covariates.
Osimertinib qsp (Liang 2024) QSP. EGFRm+ target-engagement (receptor-occupancy) model for osimertinib (OSI, AZD9291) in NSCLC, describing irreversible covalent binding of OSI to the two EGFR mutants studied by Liang 2024 – T790M and L858R – with target turnover. Two independent free-target / drug-target-complex pairs share a single osimertinib concentration driver, reproducing the pair of mutant inhibition curves in Liang 2024 Figure 1. The osimertinib concentration is NOT fitted here: Liang 2024 generated it with a whole-body PK-Sim 10.0 PBPK model that is a platform port (no ODEs, no organ volumes and no blood flows are published, and no .pksim5 project was deposited), so that layer is not reproducible from the on-disk sources and is deliberately NOT extracted. The total osimertinib concentration is instead supplied per record as the canonical time-varying covariate CEFFECT and multiplied by the fraction unbound fu inside model(), exactly as in the authors’ own supplementary code listing. Deterministic mechanism model: Liang 2024 simulated ten virtual subjects per scenario from PK-Sim population physiology rather than fitting IIV or residual error, so no etas and no error model are encoded. IMPORTANT DEVIATION – the complex ODE carries an inferred ‘- kdeg * complex’ elimination term that is absent from the paper’s printed Eq 5 and from the supplementary listing. As printed, the complex has no elimination at all and (with koff = 0) occupancy rises monotonically to 100% under any sustained exposure, which provably contradicts the flat sawtooth plateau of the paper’s own Figure 1. Restoring the term conserves total target at rbase and yields the steady-state occupancy konCfree / (konCfree + kdeg); it uses only published values (kdeg = 0.025 /h is declared in the supplement and used on the free-target line), but it is an inferred correction rather than a transcription. Extraction performed under operator sidecar decision oare_PMC10946252 request-001 = option B (answered 2026-08-05). See the vignette ‘Assumptions and deviations’ section for this and for the residual ~400-fold kon / concentration scale discrepancy against Figure 1’s plasma band.
Oxaliplatin (Valenzuela 2011) Population PK/PD model for hyperthermic intraperitoneal oxaliplatin (HIO) and induced neutropenia in 30 adults with peritoneal carcinomatosis after cytoreductive surgery (Valenzuela 2011). PK: peritoneum-as-depot first-order absorption (parameterized in the paper as peritoneum-to-plasma clearance Qa and peritoneum volume Va = vd, with ka = Qa/Va as a secondary parameter) feeding an open two-compartment plasma disposition; bioavailability F was fixed to 1 so Cl/F, Vc/F, Q2/F, Vp/F are apparent. PD: Friberg semi-mechanistic myelosuppression chain (one proliferating compartment plus three transit compartments feeding circulating ANC) with a linear drug effect Edrug = alpha * Cc on the proliferation rate and a (Circ0/Circ)^gamma feedback amplification; MTT was fixed at 118 h and the circulating-cell elimination rate constant kCirc was fixed at 0.07 per h (both from Friberg 2002). No subject covariates were retained in the final model; ten demographic and biochemistry covariates were screened graphically and showed no correlation with PK/PD parameters.
Oxaliplatin (Zhu 2023) Two-compartment population PK model for ultrafilterable platinum in plasma after intravenous oxaliplatin in adults with metastatic colorectal cancer, estimated from human concentration-time profiles digitised from the literature. Zhu 2023 uses the central-compartment concentration as the driver of an in vitro-derived tumour-killing model; that in vivo tumour-growth layer is NOT included here because its growth equation and limiting parameter are not reported anywhere in the paper - see the vignette Errata. The in vitro drug-effect half is available as modellib(‘Zhu_2023_oxaliplatin_organoid’).
Oxaliplatin organoid (Zhu 2023) In vitro (patient-derived colorectal-cancer tumour organoids, PDTOs). Exponential organoid-growth model with a sigmoidal Emax (Hill) killing term describing the 96 h concentration-viability relationship of oxaliplatin. A vehicle-control organoid and a treated organoid grow in parallel; the reported readout is cell viability, the ratio of treated to control organoid volume. This is the in vitro half of the Zhu 2023 in vitro-to-in vivo translation: the same Emax / EC50 / Hill parameters are carried into the human oxaliplatin PK/PD model, see modellib(‘Zhu_2023_oxaliplatin’).
Oxcarbazepine (Rodrigues 2017) Parent-metabolite population PK model for oral oxcarbazepine (OXC) and its active monohydroxy derivative (MHD) in epileptic children aged 2-12 years (Rodrigues 2017). Two-compartment OXC + one-compartment MHD with first-order absorption, complete metabolic conversion (Fm fixed to 1), reversible MHD-to-OXC back-transformation (KBT), empirical allometric weight scaling on CL_OXC/F, Vc_OXC/F, CL_MHD/F, and Vc_MHD/F (no scaling on Q_OXC/F or Vp_OXC/F), and a 29.3% increase in MHD clearance under concomitant enzyme-inducing antiepileptic drugs.
Oxyfluorfen human (Decrane 2023) QSP (PBPK + thyroid-hormone kinetics). Human cross-species extrapolation of the rat oxyfluorfen model of Decrane et al. 2023 (Curr Res Toxicol 5:100138), used by the authors to forecast the drop in serum T4 and T3 from long-term oral exposure to the herbicide oxyfluorfen in drinking water. Structurally identical to the rat companion model (13 flow-limited chemical compartments including a diffusion-limited thyroid split into thyroid blood and thyroid tissue, plus five thyroid-hormone states carrying the HPT feedback loop), but with human physiology (Table 2), an age-dependent cardiac output, a hepatic clearance derived from human in-vitro hepatocyte data rather than fitted, human basal thyroid-hormone levels, non-allometric human hormone production and transfer rates, and a DIFFERENT empirical sodium-iodide-symporter (NIS) inhibition function fitted to human hNIS-HEK293T-EPA in-vitro data. There is no human in-vivo calibration data for oxyfluorfen, so the authors caution that the predictions ‘should be considered with caution’. Deterministic typical-value (no IIV, no residual error). Companion model: Decrane_2023_oxyfluorfen_rat.
Oxyfluorfen rat (Decrane 2023) QSP (PBPK + thyroid-hormone kinetics). Preclinical (rat, Sprague-Dawley). Whole-body PBPK model for the diphenyl-ether herbicide oxyfluorfen coupled to a hypothalamic-pituitary-thyroid (HPT) axis thyroid-hormone kinetics sub-model (Decrane et al. 2023, Curr Res Toxicol 5:100138). Oxyfluorfen is distributed across 13 flow-limited compartments (gut lumen, GI tissue, liver, kidney, muscle, skin, fat, brain, slowly perfused, rapidly perfused, a diffusion-limited thyroid split into thyroid blood and thyroid tissue, and a single well-mixed blood pool) with hepatic metabolic clearance and GFR-driven renal excretion. The predicted oxyfluorfen concentration in thyroid blood drives an empirical in-vitro-derived sodium-iodide-symporter (NIS) inhibition function that scales thyroidal T4 and T3 synthesis (the paper’s CBthy; see the model file and vignette Errata for the evidence that CBthy is the thyroid blood rather than the thyroid tissue concentration, a reading that reproduces both species’ published points of departure); five thyroid-hormone states (T4 and T3 in thyroid tissue, T4, T3 and TSH in serum) carry the HPT feedback loop in which TSH stimulates T4 synthesis (stim) and serum T4 regulates TSH production and turnover (feed1, feed2). The model is deterministic typical-value (no IIV, no residual error) – the paper reports point estimates only, with no parameter uncertainty, no between-animal variability and no measurement-error model. Companion model: Decrane_2023_oxyfluorfen_human.
Oxypurinol (Stocker 2012) One-compartment population PK model for oxypurinol (the active metabolite of allopurinol) in adults with gout (Stocker 2012). First-order formation from allopurinol (Kfm taken as the apparent first-order absorption rate into the central compartment), one-compartment distribution, and first-order elimination. Apparent clearance (CL/Fm) is modified by raw Cockcroft-Gault creatinine clearance based on lean body weight (CRCL), concomitant any-class diuretic use (CONMED_DIURETIC; thiazide / furosemide / spironolactone pooled), and concomitant probenecid use (CONMED_PROBENECID), each via a linear-deviation multiplicative factor. Apparent volume (V/Fm) is allometrically scaled on lean body weight (LBW) with the volume exponent held fixed at the theoretical value of 1.0. The dose entered into the model is the oxypurinol-equivalent dose, taken as 0.9 x allopurinol dose per the paper’s prior published assumption.
Ozanezumab (Berges 2015) Two-compartment IV population PK plus effect-compartment sigmoid Emax PKPD model for the proportion of skeletal-muscle membrane Nogo-A co-localized with ozanezumab in adults with amyotrophic lateral sclerosis (ALS), based on the GlaxoSmithKline first-in-human study NCT00875446 (Berges 2015, Table 2)
Ozoralizumab (Takeuchi 2023) One-compartment population PK model with first-order absorption for subcutaneous ozoralizumab (anti-TNF VHH NANOBODY) in Japanese patients with rheumatoid arthritis (Takeuchi 2023)
Paclitaxel (deJonge 2005) Semi-mechanistic population pharmacokinetic model for orally administered paclitaxel formulated in Cremophor EL (CrEL) and coadministered with cyclosporin A in adult cancer patients. Free paclitaxel in the gastrointestinal tract (depot) absorbs first-order (kabs) into a two-compartment plasma disposition (central + peripheral1; linear elimination CL/F, volume V/F, intercompartmental clearance Q derived from the paper’s k23 = Q/Vc and k32 = Q/Vp). A second GI-tract paclitaxel pool (bound) holds drug encapsulated in CrEL micelles; the depot <-> bound equilibrium is governed by a single rate constant keq whose forward binding rate scales with the GI-tract CrEL amount (cremophor), which itself decays first-order with rate kcrem. Bioavailability F1 is fixed at 1 with log-normal between-subject variability; the paper found no dose-dependence in F. Inter-occasion variability on CL collapses to between-subject variability in this packaged form because the source dataset’s occasion column is not encoded – see vignette Assumptions and deviations.
Paclitaxel (Janssen 2023) Semi-physiological enriched three-compartment population PK model for intravenous paclitaxel with saturable elimination and saturable distribution in pregnant cancer patients, applying gestational changes in albumin binding, glomerular filtration, hepatic plasma flow, CYP3A4 activity and body fluid volumes to the non-pregnant Crombag 2019 base model
Paclitaxel myelosuppression (Wahlby 2004) Semi-mechanistic Friberg-Karlsson myelosuppression PD model for paclitaxel-induced neutropenia in 45 cancer patients, demonstrating Wahlby 2004’s extended covariate-model formulation. Final-model adds a time-varying bilirubin (TBILI) effect on mean transit time and a per-occasion delta-from-baseline-bilirubin effect on the linear drug-effect Slope, with inter-individual variability in the delta-bilirubin-Slope coefficient (Wahlby 2004 Eq 3 demonstrated). Paclitaxel PK is supplied via per-subject empirical-Bayes columns (CL_INDIV, VC_INDIV, VP_INDIV) following the Friberg 2002 paclitaxel convention; users can also pair this PD model with the Friberg_2002_paclitaxel PK structure directly via the modellib registry.
Pacmilimab qsp (Ippolito 2024) QSP. Quantitative systems pharmacology model of immuno-oncology (QSP-IO) describing the antitumour immune response to pacmilimab, a conditionally activated (Probody) anti-PD-L1 antibody, in triple-negative breast cancer. Four physiological compartments (central, peripheral, tumour, tumour-draining lymph node) plus APC endosomal and surface compartments, three immunological-synapse compartments (T cell-cancer cell, T cell-APC, macrophage-cancer cell) and four out-of-synapse compartments introduced by this paper (cancer cell, T cell, macrophage, APC). Covers Gompertzian tumour growth with explicit tumour vasculature, naive and activated CD4+/CD8+ T cell trafficking and proliferation, self- and neo-antigen release, uptake, endosomal processing and MHC-I presentation, TCR-mediated activation, PD-1/PD-L1/PD-L2 and CTLA-4/CD28/CD80/CD86 checkpoint binding, MDSC recruitment with arginase-I and nitric-oxide suppression, M1/M2 tumour-associated macrophages with CD47/SIRP-alpha phagocytosis control, and IL-2 / IL-10 / IL-12 / IFN-gamma / TGF-beta / CCL2 dynamics. The Probody module carries six free states per compartment (masked, open and cleaved combinations of the two active sites) with reversible unmasking and irreversible protease cleavage, plus nine mono- and bivalently bound states at each of nine PD-L1 bearing sites. Dosing the fully masked state P_mm gives pacmilimab; dosing the fully cleaved state P_cc gives the unmasked anti-PD-L1 antibody (equivalent to atezolizumab monotherapy), which is how the paper simulates its reference arm. Deterministic mechanism model: the authors generated virtual patients by Latin hypercube sampling of the distributions in Supplementary Table S1 rather than by fitting IIV or residual error, so no etas and no error model are encoded.
Paliperidone panss subscales (PillaReddy 2013) Population PK/PD model for paliperidone extended release against the three PANSS subscales (positive, negative, general) in adults with schizophrenia from Pilla Reddy 2013 Part II. The PK sub-model is the one-compartment paliperidone structural model from Part I (PMID 23473810) Table 2 with the sequential zero-order plus first-order absorption simplified to first-order only (the ER absorption profile is approximately steady at steady state because the OROS-extended- release tablet is dosed once daily): first-order absorption ka = 0.57 1/h, apparent oral clearance CL/F = 14.1 L/h, apparent central volume of distribution Vc/F = 475 L. The PD sub-model has three outputs that share the Weibull placebo time-course form Pplacebo = Pmax * (1 - exp(-(t/TD)^POW)) but each subscale carries its own placebo Pmax, TD, POW (Part II Table 1) and paliperidone’s own Emax / EC50 / KT triplet per subscale (Part II Table 2). The KT for paliperidone PANSS positive and general (0.048 and 0.035 1/day) is the common-across-atypical- antipsychotic value; the KT for the negative subscale (0.13 1/day) was estimated separately per drug. The lag time ALAG1 = 0.67 h and the zero-order absorption duration DUR = 23.6 h reported in Part I Table 2 for paliperidone ER are not encoded here because the Css used by the PD model is approximately invariant to within-dose absorption details at steady state. The exponential time-to-event dropout sub-model from Part II Table 4 is documented in population$dropout_model but not encoded in the model body. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Robbie_2012_palivizumab.html">Palivizumab (Robbie 2012)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for palivizumab (anti-RSV humanized IgG1 kappa mAb) with first-order IM absorption in adults and children (Robbie 2012) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cao_2013_PAmAb.html">PAmAb (Cao 2013)</a> </td> <td style="text-align:left;"> Second-generation minimal physiologically-based PK (mPBPK) model for PAmAb in adults (Cao 2013 Model A; clearance from plasma) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marcantonio_2022_efa.html">Panitumumab (Marcantonio 2022)</a> </td> <td style="text-align:left;"> QSP. Two-compartment monospecific anti-receptor mechanistic PKPD model of panitumumab-EGFR binding in adults with metastatic colorectal cancer (Marcantonio 2022 Early Feasibility Assessment, Case Study 2). Drug administered IV as a bivalent (valency = 2) antibody that binds membrane EGFR (in central and peripheral compartments). Soluble EGFR is not modelled (JSON shed_css = 0). All species eliminate first-order; drug bound to membrane receptor eliminates at the receptor's rate. Structure identical to the Marcantonio 2022 trastuzumab anti-receptor model; differs only in the drug-specific parameters. Parameters FIXED from paper Tables 3 (target) and 4 (drug) and the Assess run file JSON. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pettersen_2009_pantoprazole.html">Pantoprazole (Pettersen 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intravenous pantoprazole in 20 paediatric intensive-care patients aged 10 days to 16.4 years (Pettersen 2009). Pantoprazole is given as a zero-order infusion (15-30 min) into the central compartment with first-order elimination. Body-weight allometric scaling is fixed (0.75 on CL/Q, 1 on Vc/V2, reference 20 kg). Clearance is further modified by age (power on AGE/5 years), and three binary clinical covariates retained at the forward-selection / backward-elimination step: systemic inflammatory response syndrome (DIS_SIRS), concomitant CYP2C19-inhibitor coadministration (CONMED_CYP2C19_INH, pooling fluconazole, voriconazole, and isoniazid), and clinically defined hepatic dysfunction (HEPIMP, paediatric criterion TBILI >= 4 mg/dL OR ALT > 2x ULN for age). Each of the three indicators reduces pantoprazole CL by 62.3%, 65.8%, and 50.5% respectively when present alone. The reference subject is a 20 kg / 5-year-old paediatric ICU patient without SIRS, hepatic dysfunction, or CYP2C19 inhibitor coadministration. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Anderson_1998_paracetamol.html">Paracetamol (Anderson 1998)</a> </td> <td style="text-align:left;"> One-compartment oral PK model for paracetamol (acetaminophen) with an explicit cerebrospinal-fluid (CSF) equilibration compartment in nine ventilator-dependent children (5 months to 12 years) with indwelling ventricular drains for raised intracranial pressure (Anderson 1998 NONMEM fit, Table 3). First-order absorption, single nasogastric dose of 40 mg/kg paracetamol elixir, plasma + CSF sampled hourly for 4 h and 2-hourly through 10 h. CSF concentration follows the plasma concentration with first-order equilibration rate keq = ln(2)/teq and steady-state ratio PC = Ccsf/Cc. Parameters are standardized to a 70 kg adult using fixed allometric exponents (0.75 on CL, 1 on V, 0.25 on the equilibration half-time teq; keq therefore scales with exponent -0.25). The published equation 2 for residual error var = SF^2 * (C^PWR + V) is unconventional and the NONMEM PWR and V terms are not reported; placeholder additive residual SDs are used so the model simulates plausibly (see vignette Errata). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Krekels_2015_paracetamol.html">Paracetamol (Krekels 2015)</a> </td> <td style="text-align:left;"> Parent-and-metabolites population PK model for intravenous paracetamol (administered as the prodrug propacetamol; doses expressed as paracetamol equivalents) and its glucuronide and sulphate phase-II conjugates in 54 preterm and term neonates and infants (Krekels 2015). One-compartment plasma disposition for paracetamol with three parallel elimination pathways from the central compartment: glucuronide formation (CL_gluc), sulphate formation (CL_sulf), and unchanged renal excretion (CL_renal). Each metabolite distributes into a one-compartment plasma space whose volume is fixed at 18% of the parent volume (Vc_gluc = Vc_sulf = 0.18 * Vc, based on the previously reported adult paracetamol model in Allegaert et al. and adult literature). The two metabolites share a common urinary excretion rate constant kE_met = mf * kE_renal, where kE_renal = CL_renal / Vc is the parent unchanged-renal rate constant and mf (multiplication factor) is estimated to be 11.3. Cumulative urinary amounts of parent paracetamol and the two metabolites are tracked as elimination-amount compartments and exposed as additive-error observations. Bodyweight enters linearly on Vc (and so on the metabolite volumes by inheritance), on the glucuronide formation clearance (CL_gluc), and on the unchanged renal clearance (CL_renal); the sulphate formation clearance (CL_sulf) scales with bodyweight as a power with an estimated exponent of 1.40. No postnatal age, postmenstrual age, sex, term-vs-preterm, or study-protocol covariate was retained in the final model, and no time-varying (up-regulation) component was detected on the glucuronidation pathway. Parameter values reported throughout (mL/min/kg, L/kg) are per-kg quantities; individual structural parameters are obtained in model() by multiplying by body weight in kg (linear) or body weight in kg raised to n (power). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wattanakul_2016_paracetamol.html">Paracetamol (Wattanakul 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for paracetamol (acetaminophen) administered as a single 600 mg dose by either intramuscular injection (zero-order absorption over DUR_IM) or oral syrup (first-order absorption with rate constant ka) in 21 adult Thai patients with uncomplicated Plasmodium falciparum malaria and fever > 38 C (Wattanakul 2016). Intramuscular bioavailability is fixed to F_IM = 1; the relative oral bioavailability is F_PO = 0.844 (95% CI 0.682-0.951). The depot compartment carries oral doses (f(depot) = F_PO) while intramuscular doses target central with rate = -2 to invoke the modeled dur(central) = DUR_IM. No covariates were retained: allometric scaling on body weight did not improve the fit and a stepwise covariate search (age, AST, ALT, bilirubin, BUN, creatinine, sex, hemoglobin, parasitaemia, systolic BP, temperature) found no significant effect at p < 0.05. Inter-individual variability for V_C and DUR_IM was estimated below 1% CV and fixed to zero in the source paper without changing the OFV; this model omits the corresponding etas accordingly. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wu_2025_paracetamol.html">Paracetamol (Wu 2025)</a> </td> <td style="text-align:left;"> Parent-and-three-metabolites population PK model for intravenous and rectal paracetamol (PCM) and its glucuronide (PCM-GLU), sulfate (PCM-SULF), and combined oxidative metabolites (PCM-cysteine + PCM-mercapturate, PCM-OXI, denoted with the canonical cysmer suffix) from preterm and term neonates through infants, children, and adults (Wu 2025). Two-compartment plasma disposition for parent PCM with three parallel formation clearances and parallel renal elimination of unchanged parent; one-compartment plasma disposition for each metabolite with renal elimination expressed as a fraction of glomerular filtration rate (GFR). The preterm-and-term-neonate-to-adult (PTNA) maturation equation (Wu 2024) is applied to each formation clearance and to a separate PCM-SULF renal-secretion clearance; an additional adult-only correction factor scales the renal clearance of PCM-GLU in subjects >= 18 years. Rectal absorption parameters (Ka, Tlag, F) are fixed from Wang 2014. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mensing_2017_3D_HCV_regimen.html">Paritaprevir (Mensing 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral paritaprevir (co-dosed with ritonavir) in HCV genotype-1 infected adults receiving the 3D regimen (Mensing 2017). First-order absorption with a fixed absorption lag time, linear elimination, additive residual error on log-transformed concentrations (encoded as log-normal Cc ~ lnorm(expSd)), IIV on CL/F only. The author's final model retained cirrhosis, gender, age, opioid use, and antidiabetic-agent use as significant covariates on CL/F (and age, body weight on Vc/F), but the paper does not publish point estimates for these covariate coefficients (only graphical exposure-ratio forest plots in Figure 2); the implemented model is the structural typical-value model with covariate coefficients omitted (documented in covariatesDataExcluded). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kim_2015_paroxetine.html">Paroxetine (Kim 2015)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for paroxetine (SSRI antidepressant) in Korean adults with major depressive disorder or anxiety disorder receiving therapeutic drug monitoring (Kim 2015). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2023_patritumab.html">Patritumab (Lee 2023)</a> </td> <td style="text-align:left;"> Integrated population PK model for the conjugated (anti-HER3-ac-DXd) and unconjugated payload (DXd) of patritumab deruxtecan (HER3-DXd, U3-1402, anti-HER3 antibody-drug conjugate) in adult cancer patients (Lee 2023 ACoP poster). anti-HER3-ac-DXd disposition is a two-compartment model with three parallel elimination pathways from the central compartment: a transient time-decaying linear clearance CL_t(time) = CL_T * exp(-Kdes * time), a non-specific time-dependent linear clearance CL_ns(time) that declines sigmoidally from CL_ss * (1 + Emax) at time = 0 to CL_ss at infinity via a Hill function CL_ns(time) = CL_ss * (1 + Emax * T50^hill / (T50^hill + time^hill)), and a Michaelis-Menten saturable clearance CL_mm = Vmax / (Km + Cc). DXd is a one-compartment model with parallel linear and Michaelis-Menten elimination; its formation rate equals the sum of the three anti-HER3-ac-DXd elimination rates each scaled by a dimensionless fractional-conversion factor Frac_ns / Frac_t / Frac_mm (Frac_ns fixed at 1 as the identifiability anchor). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lu_2022_patritumab.html">Patritumab (Lu 2022)</a> </td> <td style="text-align:left;"> Joint two-analyte population PK model for patritumab deruxtecan (HER3-DXd, an anti-HER3 antibody-drug conjugate) in adults with HER3-expressing solid tumors (Lu 2022). DXd-conjugated antibody (intact ADC) is described by a 2-compartment model with parallel linear and Michaelis-Menten clearance. Released unconjugated DXd (MAAA-1181a, exatecan-derivative payload) is described by a 1-compartment model with linear clearance and a first-order, time-dependent release rate driven by the level of DXd-conjugated antibody in the central compartment, scaled by the molecular-weight ratio MW_DXd/MW_DXdAb and a payload-to-intact-drug ratio PIR modulated by a cycle-1-vs-later (factor1) and a within-cycle exponential (factor2) modifier. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ouerdani_2015_pazopanib.html">Pazopanib (Ouerdani 2015)</a> </td> <td style="text-align:left;"> Semi-mechanistic tumour growth and angiogenesis-inhibition (TGI) model for pazopanib in renal-cell carcinoma patients (Ouerdani 2015 clinical fit): logistic tumour growth (state tumor_size) limited by a separately tracked vasculature-determined carrying capacity (state carrying_capacity), with antiangiogenic and cytotoxic drug effects parameterised as power functions of per-period mean AUC_PAZO and an exponentially declining resistance on the cytotoxic effect. The empirical exponent on capacity growth (n) is fixed at 0.5 for the clinical fit (vs 1 in the paired mouse model) to better describe the tumour-regrowth and long-term-antiangiogenic phases observed in patients. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ouerdani_2015_pazopanib_mouse.html">Pazopanib mouse (Ouerdani 2015)</a> </td> <td style="text-align:left;"> Preclinical (mouse, CB-17 SCID with CAKI-2 renal-cell carcinoma xenografts). Semi-mechanistic tumour growth and angiogenesis-inhibition (TGI) model for pazopanib (Ouerdani 2015): logistic tumour growth (state tumor_size) limited by a separately tracked vasculature-determined carrying capacity (state carrying_capacity), with an antiangiogenic drug effect on carrying-capacity loss (power form in AUC_PAZO) and a putative cytotoxic drug effect on tumour decay (exponentially declining resistance, no AUC effect after the cytotoxic exponent was fixed to 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_guhongInjection.html">PCoumaricAcid rat (Chen 2024)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment intravenous pharmacokinetic model for p-coumaric acid (pCA), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the p-coumaric acid dose is the GHI volume dose times its content in GHI (78.3 ug/mL), i.e. 195.75 ug/kg, 391.5 ug/kg, 783 ug/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Direct-effect sigmoid-Emax models link the p-coumaric acid plasma concentration to the GHI-minus-model-group difference in creatine kinase-MB (CK-MB), ischemia-modified albumin (IMA), cardiac troponin I (cTn I) and alpha-hydroxybutyrate dehydrogenase (alpha-HBDH) (E = Emax * C^gamma / (ED50^gamma + C^gamma); Tables 12, 13, 14, 15). Chen 2024 fitted a PK/PD model only for the analyte/biomarker/dose combinations whose PLSR coefficient was negative, so the effect of an unfitted combination is returned as zero rather than extrapolated. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wahlby_2004_time_varying_covariates.html">Pefloxacin (Wahlby 2004)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous 1-hour pefloxacin infusions in 74 critically ill adults, demonstrating Wahlby 2004's extended covariate-model formulation. Final-model clearance carries time-varying CRCL (with inter-individual variability in the CRCL effect coefficient, Eq 3), per-subject baseline total bilirubin BIL_BASE (replaces BIL), age, centre indicator, and per-subject baseline weight WT_BASE (replaces WT, with a saturating 'up to median weight' qualifier per Methods). Central volume retains the upstream Karlsson 1993 (ref [10]) WT/CRCL/BIL effects unchanged. Underlying structural PK comes from Karlsson MO, Sheiner LB. The importance of modeling interoccasion variability in population pharmacokinetic analyses. J Pharmacokin Biopharm 1993;21(6):735-750 (not on disk in this worktree). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kloos_2021_pegasparaginase.html">Pegasparaginase (Kloos 2021)</a> </td> <td style="text-align:left;"> One-compartment population PK model with time-dependent (split-point) clearance for intravenous PEGasparaginase in pediatric acute lymphoblastic leukemia patients treated on the Dutch Childhood Oncology Group ALL-11 protocol (Kloos 2021). Clearance and volume of distribution are normalized to body surface area; clearance is constant for the first 12.7 days after a dose and then increases linearly with time after dose as the polyethylene glycol moiety is hydrolyzed. Clearance is 38 percent higher during an active infection and 11-19 percent lower outside induction, encoded as multiplicative treatment-phase factors with protocol 1A as the reference. Inter-individual variability on clearance is shared with the volume of distribution through a scaled eta; inter-occasion variability on clearance uses one occasion per administered dose. Combined proportional and additive residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Brekkan_2018_pegfilgrastim.html">Pegfilgrastim (Brekkan 2018)</a> </td> <td style="text-align:left;"> Bidirectional population PK/PD model for pegfilgrastim (PG) in healthy volunteers after single 6 mg subcutaneous doses. PK is one-compartment with sequential zero- and first-order absorption (zero-order input rate R1 into depot followed by first-order Ka into central), and parallel elimination via a linear ANC-dependent pathway (cl_anc * ANC) and a saturable non-specific Michaelis-Menten pathway (Vmax / Km). PD is a Friberg/Quartino-style maturation cascade (4 transit compartments + circulating compartment) with the production rate set by baseline ANC and a fixed 7-hour circulating neutrophil half-life, plus three Emax drug effects: proliferation (scaling production), maturation (scaling transit rate ktr), and a margination effect on circulating-pool clearance kcirc that is parameterised by scaled Emax,prol and EC50 (Emax,Scale = 0.0622, EC50,Scale = 0.477). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Naik_2013_peginesatide.html">Peginesatide (Naik 2013)</a> </td> <td style="text-align:left;"> Two-compartment population PK/PD model for peginesatide in adult chronic kidney disease (CKD) patients (Naik 2013). PK: first-order subcutaneous absorption with saturable Michaelis-Menten elimination and fixed inter-compartmental clearance. PD: modified precursor-dependent lifespan indirect-response (LIDR) model of hemoglobin (1 progenitor compartment + 7 red-blood-cell aging compartments) with a peginesatide Emax stimulation on progenitor production and an empirical exponential downward-drift factor on the progenitor-to-RBC transit. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bi_2017_peginterferon_alfa_2a.html">Peginterferon alfa 2a (Bi 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for peginterferon alfa-2a in adult patients with chronic hepatitis B (Bi 2017). Creatinine clearance (Cockcroft-Gault, mL/min, not BSA-normalized) modifies clearance via a power form, and body mass index modifies central volume via a power form. Exponential IIV on CL, V, and Ka; combined proportional + additive residual error on plasma concentration. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gupta_2006_peginterferon_alfa_2b.html">Peginterferon alfa 2b (Gupta 2006)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order subcutaneous absorption for peginterferon alfa-2b (PEG-Intron) in adult patients with chronic myelogenous leukaemia (Gupta 2006). Apparent clearance declines over treatment time via an Emax-type function CL(t) = CL0 / (1 + (t / T50)^beta) with beta fixed to 1 in the final model, so CL(t) = CL0 / (1 + t / T50). Cockcroft-Gault creatinine clearance modifies baseline clearance via a power form. Exponential IIV on CL0, T50, and V; proportional residual error on plasma concentration. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hu_2017_peginterferon_beta_1a.html">Peginterferon beta 1a (Hu 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for peginterferon beta-1a in adults with relapsing multiple sclerosis (Hu 2017). First-order SC absorption with the absorption rate constrained above the elimination rate to avoid flip-flop kinetics. BMI is a covariate on both clearance and volume of distribution. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ahamadi_2017_pembrolizumab.html">Pembrolizumab (Ahamadi 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for pembrolizumab (humanized anti-PD-1 IgG4 monoclonal antibody) with allometric scaling and covariate effects of sex, albumin, tumor type, ECOG performance status, prior ipilimumab status, eGFR, and baseline tumor burden, in adults with advanced solid tumors (Ahamadi 2017, KEYNOTE-001/-002/-006) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Elassaiss-Schaap_2017_pembrolizumab.html">Pembrolizumab (Elassaiss-Schaap 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with parallel linear and Michaelis-Menten clearance plus a direct-response Imax PK/PD model on the ex vivo IL-2 stimulation ratio (PD-1 target engagement) for IV pembrolizumab (anti-PD-1 IgG4 mAb) in adults with advanced solid tumors (Elassaiss-Schaap 2017, KEYNOTE-001 parts A, A1, A2). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lindauer_2017_pembrolizumab.html">Pembrolizumab (Lindauer 2017)</a> </td> <td style="text-align:left;"> QSP / mini-PBPK. Translational semi-mechanistic PK/PD/TGI model for the anti-PD-1 monoclonal antibody pembrolizumab in advanced melanoma. Couples a two-compartment plasma PK (parallel linear + Michaelis-Menten clearance, human PK substituted from Elassaiss-Schaap 2017 KEYNOTE-001) to a Shah-Betts (2012) physiologic tumor tissue compartment (vascular, endosomal, interstitial sub-spaces with FcRn recycling), mechanistic pembrolizumab-PD-1 binding in both blood and tumor, an indirect-response positive feedback that upregulates tumor PD-1 expression when the complex forms, and a Simeoni-type tumor-growth model in which the antitumor effect is a power function of the tumor receptor occupancy. Mouse-derived parameter estimates plus three human melanoma growth-rate scenarios (slow/medium/fast) and two kill-rate scaling options (allometric / growth-proportional) are tabulated in Lindauer 2017 Table 1 and Table S3; the default human parameterisation here is medium growth with allometric kill-rate scaling (the central reference scenario). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gong_2023_pemigatinib.html">Pemigatinib (Gong 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral pemigatinib (a selective fibroblast growth factor receptor [FGFR] 1-3 inhibitor) in healthy participants and patients with advanced solid tumors including cholangiocarcinoma (CCA) and myeloid/lymphoid neoplasms with FGFR1 rearrangement (Gong 2023; N = 467 across seven FIGHT studies, 4552 concentration records). Disposition is a two-compartment model with sequential zero-order and first-order oral absorption (dose enters the depot over a zero-order window of duration D1 = 0.810 h, then is absorbed first-order at ka = 3.67 h^-1) and linear elimination (CL/F = 10.7 L/h, Vc/F = 118 L, Vp/F = 95.0 L, Q/F = 25.2 L/h). Covariate effects retained after stepwise forward addition and backward elimination: concomitant phosphate-binding agent use reduces CL/F by 15.5%; male sex increases CL/F by 26.2% and reduces ka by 58.3% (female is the reference); concomitant proton-pump inhibitor use reduces ka by 62.0%; concomitant histamine-2 receptor antagonist use reduces D1 by 33.4%; and baseline body weight scales Vc/F and Vp/F as power functions of (WT / 73.9 kg) with exponents 0.842 and 1.13 respectively. The paper reports all of these covariate effects as statistically significant but not clinically significant on steady-state exposure. Inter-individual variability is a correlated CL/F-Vc/F block (log-scale variances 0.201 and 0.191, covariance 0.129, correlation 0.659) plus independent terms on ka (1.69) and D1 (0.371); IIV on Q/F and Vp/F was fixed to zero in the final model. Residual error is additive on the log scale (approximately proportional in linear space) at 32.3%. Concentrations are expressed in nM using the pemigatinib molecular weight of 487.5 g/mol, matching the h*nM exposure units used throughout the paper's exposure-response analyses. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gong_2023_pemigatinib.html">Pemigatinib creatinine (Gong 2023)</a> </td> <td style="text-align:left;"> Exposure-response Emax model relating steady-state pemigatinib exposure to the percentage change from baseline in serum creatinine concentration, in 300 patients receiving pemigatinib monotherapy across FIGHT-101, FIGHT-102, and FIGHT-202 (Gong 2023). Pemigatinib inhibits the renal transporters OCT2 (IC50 0.075 uM) and MATE1 (IC50 1.1 uM), which mediate active tubular secretion of creatinine; the resulting rise in serum creatinine is exposure-dependent and reversible rather than a marker of renal injury. The model is a direct-effect Emax function of the per-subject steady-state AUC over the 24 h dosing interval: pct change = Emax * AUC_PEMI / (EC50 + AUC_PEMI), with Emax = 43.4% and EC50 = 2595 h*nM. It carries no ODE state -- the exposure metric AUC_PEMI is supplied as a static covariate derived from the companion population PK model (see modellib('Gong_2023_pemigatinib')), mirroring the authors' workflow of simulating a dense steady-state post hoc profile in NONMEM, deriving AUCss noncompartmentally, and regressing the biomarker on it in SAS. Evaluating the model at the paper's own mean steady-state AUC of 2850 h*nM for 13.5 mg once daily, and dose-scaling that value, reproduces the four published predictions of 14.2%, 18.3%, 22.7%, and 26.9% at 6, 9, 13.5, and 20 mg to three significant figures. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gong_2023_pemigatinib.html">Pemigatinib phosphate (Gong 2023)</a> </td> <td style="text-align:left;"> Exposure-response Emax model relating steady-state pemigatinib exposure to the mean change from baseline in serum phosphate concentration at cycle 1 day 8 and cycle 1 day 15, in 300 patients receiving pemigatinib monotherapy across FIGHT-101, FIGHT-102, and FIGHT-202 (Gong 2023). Pemigatinib inhibits FGFR1-3; FGFR1 blockade attenuates FGF23 signalling in the renal proximal tubule, reducing urinary phosphate excretion, so the exposure-dependent rise in serum phosphate is an on-target class effect that the authors use as a pharmacodynamic surrogate for pemigatinib exposure. The model is the paper's three-parameter direct-effect Emax function of the per-subject steady-state AUC over the 24 h dosing interval, with baseline serum phosphate entering as an additive covariate: change = Emax * AUC_PEMI / (EC50 + AUC_PEMI) + coef * baseline, with Emax = 5.51 mg/dL, EC50 = 1665 h*nM, and coef = -0.185 per mg/dL (carried as -0.5730 per mmol/L against the SI-canonical PHOS column). It carries no ODE state -- the exposure metric AUC_PEMI is supplied as a static covariate derived from the companion population PK model (see modellib('Gong_2023_pemigatinib')), mirroring the authors' workflow of simulating a dense steady-state post hoc profile in NONMEM, deriving AUCss noncompartmentally, and regressing the biomarker on it in SAS. The endpoint is reported in the paper's mg/dL. The fitted EC50 of 1665 h*nM approximates the mean AUC0-24h at 8 mg once daily, which is the basis of the paper's claim that serum phosphate can serve as an exposure surrogate. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ogungbenro_2009_penciclovir.html">Penciclovir (Ogungbenro 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and lag time for penciclovir in pooled adults and children (Ogungbenro 2009). Famciclovir is the oral prodrug of penciclovir; both oral famciclovir and intravenous penciclovir doses are described jointly (six clinical studies, 69 subjects of whom 23 are children, 160 occasions, 1676 plasma penciclovir observations). Allometric body-weight scaling with reference 70 kg (exponent 0.75 shared on CL and Q, exponent 1.0 shared on V1 and V2), an empirical piecewise age effect on CL with separate K parameters for AGE < 40 years (rising-with-youth limb) and AGE >= 40 years (declining-with-elderly limb), and a power function of creatinine clearance on CL with reference 100 mL/min (Cockcroft-Gault, raw mL/min). Inter-individual variability on ka, CL, V1 (fixed at omega^2 = 0.003), V2, and Q; combined proportional plus additive residual error (additive variance fixed at 0.01 mg^2/L^2). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Muller_2007_penicillin_G.html">Penicillin G (Muller 2007)</a> </td> <td style="text-align:left;"> Two-compartment IV bolus population PK model for penicillin G (benzylpenicillin) in 20 very preterm neonates with gestational age less than 32 weeks studied on day 3 of life (Muller 2007). Clearance is linearly scaled to current body weight with reference 1.195 kg (cohort mean); central volume, peripheral volume, and intercompartmental clearance are not weight-scaled in the final model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Padari_2018_penicillin_G.html">Penicillin G (Padari 2018)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for penicillin G (benzylpenicillin) in preterm and term neonates (Padari 2018; pooled with Metsvaht 2007 GA <=28 wk cohort). CL and Q are allometrically scaled to body weight (fixed exponent 0.75) with a fixed Rhodin-style postmenstrual-age (PMA) sigmoidal renal-maturation function on CL; Vc and Vp are allometrically scaled (fixed exponent 1.0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Li_2014_penicillinG.html">PenicillinG cattle (Li 2014)</a> </td> <td style="text-align:left;"> Preclinical (cattle). Three-compartment population pharmacokinetic model for penicillin G in cattle, with four parallel first-order absorption depots covering intramuscular penicillin sodium, intramuscular procaine penicillin, subcutaneous procaine penicillin, and oral procaine penicillin (the oral depot feeds the liver compartment directly), plus separate liver and kidney tissue compartments connected to the central compartment by inter-compartmental clearance; pooled meta-analysis of 100 cattle from 30 published studies and FARAD records (Li 2014). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Li_2014_penicillinG.html">PenicillinG swine (Li 2014)</a> </td> <td style="text-align:left;"> Preclinical (swine). Three-compartment population pharmacokinetic model for penicillin G in swine, with two parallel first-order absorption depots covering intramuscular penicillin potassium and intramuscular procaine penicillin, plus separate kidney and muscle tissue compartments connected to the central compartment by inter-compartmental clearance; pooled meta-analysis of 89 pigs from 13 published studies and one unpublished FDA dataset (Li 2014). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Silva_2023_perampanel.html">Perampanel (Silva 2023)</a> </td> <td style="text-align:left;"> One-compartment popPK model of perampanel in Portuguese adults with refractory epilepsy on chronic therapeutic drug monitoring (Silva 2023). NONMEM FOCE-I fit of 72 steady-state plasma concentrations from 44 patients; concentrations were obtained during the elimination phase only (9.7-24 h post-dose) so first-order absorption could not be estimated, and dosing was modelled as an IV bolus into the central compartment (perampanel oral bioavailability is ~100% per the summary of product characteristics). Concomitant enzyme-inducing antiepileptic drugs (carbamazepine, oxcarbazepine, phenobarbital, or phenytoin) increase apparent clearance 2.76-fold; body-mass index enters the volume of distribution as a power function centred at the sample median BMI of 25.1 kg/m^2. Inter-patient variability is exponential on clearance only; residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Luo_2024_CES1_cirrhosis.html">Perindopril pbpk (Luo 2024)</a> </td> <td style="text-align:left;"> PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). Joint perindopril + perindoprilat disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Perindopril is an inactive ester prodrug hydrolysed by hepatic CES1 to the active diacid perindoprilat, which is eliminated renally. Perindopril has a second, parallel hepatic pathway (UGT glucuronidation) that does NOT form perindoprilat; because the responsible UGT isoenzyme is unidentified, its cirrhosis scaling uses the metoprolol AUC0-inf change rate of 0.62 (Section 3.1.4). The semi-PBPK circuit is stomach, three small-intestinal lumen segments (duodenum / jejunum / ileum), the matching three gut-wall segments, portal vein, liver, kidney and a two-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper's virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gaohua_2023_permeabilityLimited_pbpk.html">PermeabilityLimited pbpk (Gaohua 2023)</a> </td> <td style="text-align:left;"> PBPK (whole-body, permeability-limited, 55 ODEs; bespoke MATLAB/SimBiology 2022a model). Theoretical what-if investigation of how passive permeability, metabolism, active transporters and dosing route determine the dynamics of the tissue/plasma partition coefficient Kp(t) and the volume of distribution Vd(t). Twelve tissues (adipose, bone, brain, heart, kidney, muscle, skin, liver, pancreas, spleen, gut, lung) each carry four subcompartments - residual blood cells (_bc), residual plasma (_plasma), extracellular water (_ew) and intracellular water (_iw) - and the three blood compartments (venous, arterial, portal vein) carry the blood-cell and plasma subcompartments only, giving 12*4 + 3*2 = 54 disposition states plus a gut-lumen depot. Passive permeation runs between adjacent subcompartments (Eqs 8-10), active uptake and efflux transporters sit on the cell membrane between _ew and _iw (Eqs 11-12), and metabolism may occur in any subcompartment (Eqs 13-16). There is no drug: the compound is a generic small molecule with fu = fi = 1 everywhere, so a perfusion-limited model would give Kp = 1 in every tissue and Vdss = 1 L/kg exactly; any departure from those values measures the permeability-limited model's divergence from the perfusion-limited one. Every permeability, metabolic clearance and transporter clearance is parameterised as a fold-multiple of that tissue's plasma flow (the fold_* parameters), which is exactly how the paper drives its what-if scenarios - each scenario in Tables 1-4 is a one-parameter change. Deterministic typical-value simulation model: the paper reports no IIV and no residual-error model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Garg_2014_pertuzumab.html">Pertuzumab (Garg 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order linear elimination from the central compartment for intravenous pertuzumab (PERJETA) in patients with a variety of HER2-targeted solid tumors (Garg 2014) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2021_pertuzumab.html">Pertuzumab (Wang 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order subcutaneous absorption and bioavailability for pertuzumab (Perjeta) administered either intravenously or as the fixed-dose combination subcutaneous formulation with trastuzumab (PH FDC SC) in patients with HER2-positive early breast cancer in the FeDeriCa study (Wang 2021) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yin_2020_pexidartinib.html">Pexidartinib (Yin 2020)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral pexidartinib (CSF1R/KIT/FLT3 inhibitor) in healthy subjects and adult patients with tenosynovial giant cell tumour (TGCT) or other advanced solid tumours (Yin 2020). Absorption is sequential zero-order deposition into a depot (duration D1, lag time ALAG1) followed by first-order absorption (KA) into the central compartment, with linear elimination from central. Apparent clearance CL/F scales allometrically on (WT/80)^0.75 and is additionally modified by piecewise power effects of CRCL (active only when CRCL < 90 mL/min), AST (active only when AST > 80 U/L), and total bilirubin (active only when TBILI > 20.5 umol/L), plus multiplicative effects for Asian race (1.27x), healthy-participant cohort (1.26x; the Phase 1 healthy-subject studies), and female sex (0.869x). Apparent central and peripheral volumes Vc/F and Vp/F scale on (WT/80)^1; apparent inter-compartmental clearance Q/F scales on (WT/80)^0.75. Relative bioavailability of the Phase 1 formulation is fixed at 0.855 vs the Phase 3 / commercial reference formulation. Inter-individual variability is a 3x3 block on log(CL,Vc,Vp), independent diagonals on log(KA) and log(Q), and a Phase-1-formulation-specific IIV on the F1 bioavailability anchor. The published inter-occasion variability (5 occasions on KA, 10 occasions on F1) is not encoded structurally here (following the Andrews 2017 / Brooks 2021 tacrolimus precedent for the model-library use case where no operational occasion column is defined). Residual error is proportional with separate magnitudes for patient samples (29.7% CV) and healthy-subject samples (19.6% CV), switched per-subject by the DIS_HEALTHY indicator. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yin_2021_pexidartinib.html">Pexidartinib (Yin 2021)</a> </td> <td style="text-align:left;"> Semi-mechanistic longitudinal tumor-size (RECIST) exposure-response PD model for pexidartinib in adult patients with tenosynovial giant cell tumor (TGCT), driven by pre-computed running average pexidartinib concentration Cavg (mg/L). Tumor size Y(t) declines from an individual baseline Y0 through a saturable drug effect gated by an onset first-order process: Y(t) = Y0 * (1 - Emax * (1 - exp(-kdrug * Cavg)) * (1 - exp(-konset * TAFD))) + growth * time, with Emax fixed at 0.999 and natural tumor growth fixed at 0 (Yin 2021 Results: the placebo-cohort growth rate estimate 0.227 cm/yr had a 95% CI that included zero and was fixed). Baseline Y0, drug-effect rate constant kdrug, and onset rate constant konset each carry three multiplicative covariate effects: joint extremity (upper vs lower reference), joint size (small vs large reference), and age centered at 44 years. Individual variability is a 3x3 log-normal block on baseline, kdrug, and konset. Residual error is a power-of-prediction form SD(Y) = 0.365 * Ypred^0.550. The pexidartinib PK backbone (Cavg input) is a separately extracted model, `modellib('Yin_2020_pexidartinib')`, corresponding to the same authors' Yin 2020 J Clin Pharmacol popPK publication (reference 5 in Yin 2021). This PD model does NOT carry the ORR proportional-odds logistic regression (Table S3) or the piecewise-exponential TTE hepatic-AE models (Table S5) that appear as parallel endpoints in Yin 2021; those are non-ODE statistical / survival regressions and are documented in the paired vignette narrative but not encoded here as separate model files. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Betts_2019_pf_06671008.html">Pf 06671008 qsp (Betts 2019)</a> </td> <td style="text-align:left;"> QSP. Preclinical (mouse). Translational quantitative systems pharmacology model for the CD3 x P-cadherin LP DART bispecific PF-06671008, fit to HCT-116 xenografts in NSG mice with human PBMC engraftment. Couples bispecific PK (2-cpt linear), permeability-diffusion tumor drug disposition, trimer (drug-CD3-Pcad) formation in the TME, and tumor growth via a 4-compartment transduction (Simeoni-based) model (Betts 2019 AAPS J). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Guo_2022_PF_06939999.html">PF 06939999 (Guo 2022)</a> </td> <td style="text-align:left;"> Population PK/PD model for PF-06939999 (a small-molecule PRMT5 inhibitor) in 28 adults with advanced solid tumors enrolled in the dose-escalation part of NCT03854227. PK is a two-compartment model with first-order absorption (CL/F, V1/F, Q/F, V2/F, Ka). Plasma SDMA (the PD biomarker for PRMT5 inhibition) is modelled by an indirect-response model with saturable Imax inhibition on zero-order SDMA production (Kin/Kout), the log-transformed SDMA observation taking an additive (log-normal) residual error. Platelet count is described by the Friberg semi-mechanistic myelosuppression model (proliferating cells plus three transit compartments feeding a circulating compartment) with a linear drug effect Slope*Cc on the proliferation rate and feedback (Circ0/circ)^gamma. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Langdon_2010_PF00821385.html">PF00821385 dog (Langdon 2010)</a> </td> <td style="text-align:left;"> Preclinical (beagle dog). Translational popPK-PD model for PF-00821385, a Pfizer HIV-1 gp120 cell-fusion inhibitor candidate (molecular weight 440.49 g/mol) studied in conscious freely-moving Beagle dogs. PK is a one-compartment disposition model with first-order oral absorption; PD describes heart rate as the sum of (a) a typical-value baseline HR with log-normal inter-subject variability, (b) a 24-h cosine circadian rhythm with typical-value amplitude and a log-normal inter-subject variable peak time, and (c) a linear drug effect on free plasma concentration with no IIV. The PD-slope SLOPE = 1.76 bpm per micromolar free drug is from Langdon 2010 Table 1; plasma unbound fraction fu = 0.64 is FIXED via back-calculation from the published unbound vs total Cmax ratio at 20 mg/kg oral (paper Introduction); see vignette Errata. PK and PD were fit sequentially in NONMEM VI using FOCE INTER with individual Bayesian post hoc PK estimates serving as input to the PD model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Langdon_2010_PF00821385.html">PF00821385 human (Langdon 2010)</a> </td> <td style="text-align:left;"> First-in-human popPK-PD model for PF-00821385, a Pfizer HIV-1 gp120 cell-fusion inhibitor candidate (molecular weight 440.49 g/mol), developed in 24 healthy male volunteers from a single-ascending-dose study (Langdon 2010 Tables 2 and Figure 3). PK is a two-compartment model with first-order oral absorption and an additive residual error on the log-transformed plasma concentrations (i.e., a log-normal residual). PD describes supine pulse rate as the sum of (a) a typical- value baseline rate with log-normal inter-subject variability, (b) a 24-h cosine circadian rhythm with typical-value amplitude and log- normal inter-subject variable peak time, and (c) a linear drug effect on free plasma concentration with no IIV. The PD-slope SLOPE = 0.76 bpm per micromolar free drug is from Langdon 2010 Table 2; plasma unbound fraction fu = 0.64 is FIXED via back-calculation from the canine toxicology Cmax data (see vignette Errata). The PD layer was fit sequentially to individual Bayesian post hoc PK estimates from the population PK fit (NONMEM VI with FOCE INTER; 500-iteration nonparametric bootstrap for SE / CI). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Li_2018_PF04236921.html">PF04236921 (Li 2018)</a> </td> <td style="text-align:left;"> Integrated population PK and indirect-response PK/PD model for the anti-interleukin-6 monoclonal antibody PF-04236921 in healthy volunteers and adults with rheumatoid arthritis, Crohn's disease, or systemic lupus erythematosus (Li 2018). Two-compartment IV/SC PK with first-order absorption and linear elimination from the central compartment; disease-stratified linear clearance and PD parameters; PF-04236921 inhibits the zero-order CRP synthesis rate of an indirect-response model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chang_2011_PF04455242_prolactin.html">PF04455242 human (Chang 2011)</a> </td> <td style="text-align:left;"> Two-compartment population PK and reduced direct-response PD model for PF-04455242 (kappa opioid receptor antagonist) in healthy adult volunteers (Chang 2011). PK is fit with zero-order oral absorption (duration D1) and lag time (ALAG1) into the central compartment; residual error uses the log-transform-both-sides (lognormal) form. PD is the reduced antagonism model (Eq. 11/12) that replaces the spiradoline PK with a deterministic Weibull-scaled placebo prolactin profile and predicts the time-matched prolactin response under PF-04455242 antagonism. Simulation time t = 0 must be aligned with the IM spiradoline challenge dose; PF-04455242 is dosed earlier (typically t = -1 h in the proof-of-mechanism study). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chang_2011_PF04455242_prolactin.html">PF04455242 rat (Chang 2011)</a> </td> <td style="text-align:left;"> Preclinical (Sprague-Dawley rat). Competitive antagonism PK-PD model of PF-04455242 (kappa opioid receptor antagonist) on spiradoline-induced plasma prolactin elevation. One-compartment first-order absorption PK for both spiradoline (KOR agonist challenge) and PF-04455242, with a dose-dependent absorption rate constant for PF-04455242 (1.64 /h at 3.2 mg/kg SC, 0.385 /h at 10 mg/kg SC). Direct-response sigmoid Emax PD: prolactin = baseline + Emax * Csp^gamma / ((EC50 * (1 + Cpf/Ki))^gamma + Csp^gamma) with competitive antagonism of the spiradoline-induced rise by PF-04455242. Spiradoline plasma compartments are declared via paper_specific_compartments rather than registering a new sibling-drug suffix; see Errata in the vignette for the rationale. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Benson_2014_PF04457845_qsp.html">PF04457845 qsp (Benson 2014)</a> </td> <td style="text-align:left;"> QSP. Whole-body four-physiological-compartment (brain, plasma, rest-of-body ROB, blood-brain-barrier microvascular endothelial cells MEC) systems-pharmacology model for the irreversible FAAH-1 inhibitor PF-04457845 and five fatty acid ethanolamide substrates (anandamide AEA, N-oleoyl-ethanolamide OEA, N-palmitoyl-ethanolamide PEA, N-linoleoyl-ethanolamide LEA, N-stearoyl-ethanolamide SEA). 39 ODE states coupling: PF-04457845 pharmacokinetics (first-order absorption, saturable Michaelis-Menten plus linear elimination, plasma-ROB tissue distribution); irreversible PF-04457845 inhibition of FAAH with FAAH protein turnover; NAPE-phospholipase-D-mediated ethanolamide synthesis from the N-acyl phosphatidylethanolamine precursors with competitive substrate inhibition among the five ethanolamides; FAAH-independent clearance via N-acylethanolamine-hydrolyzing acid amidase (NAAA); bidirectional AEA-transporter and passive diffusion transport of ethanolamides between ROB, MEC and plasma; blood-brain-barrier flux of ethanolamides between brain and MEC; and Emax cannabinoid receptor CB1 occupancy in brain by AEA. All 110+ structural, physiologic, enzyme-kinetic, receptor-binding, and PK parameters are fixed from the paper's supplement Table S1 and Model S1 SBML export: Peters and Hultin 2008 PBPK physiology for the average 70-kg-adult tissue volumes; literature in vitro enzyme-kinetic measurements for FAAH kcat/Km, NAPE-PLD forward/reverse rates and substrate Ki, and NAAA affinities; McPartland 2007 CB1 in vitro K_d; and Pfizer Phase I data (Li 2012) for FAAH turnover, drug PK, and the fitted NAT relative activities. No IIV, no residual variability -- the model is deterministic and describes mean biomarker profiles. 2-arachidonoyl-glycerol (2-AG) is not included (fixed at zero) per the paper's stated scope. Model was originally implemented in DBSolve Optimum and Matlab/SimBiology. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jones_2011_PF04878691_HCV.html">PF04878691 (Jones 2011)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model with first-order oral absorption and time-varying clearance for the toll-like-receptor-7 (TLR7) agonist PF-04878691 in healthy male and female adult volunteers (Jones 2011 BJCP, Phase 1 multiple-dose escalation study, twice-weekly oral doses of 3, 6, or 9 mg over 2 weeks). Observed plasma exposure increased over the dosing period inconsistently with the 12-16 h terminal half-life; a standard linear time-invariant two-compartment model over-estimated Cmax on day 1 and under-estimated exposure on day 11. The clearance was therefore parameterised with an exponentially decaying time-dependent component superimposed on a steady-state arm: CL(t) = CL_SS + CL_TIME * exp(-cl_exp_kdes * TAFD), where TAFD is the time after first dose. Reparameterised from the paper's CLF (final = CL_SS) and CL0 (initial = CL_SS + CL_TIME). The hypothesised mechanism for the time-varying clearance is IFN-mediated CYP1A2 inhibition by the TLR7-induced interferon response (Discussion). All disposition parameters are reported per kilogram body weight (paper: doses were body-weight-normalised so estimated PK parameters carry per-kg units); WT is therefore a required covariate. No other covariates retained. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jones_2011_PF04878691_HCV.html">PF04878691 lymphocyte (Jones 2011)</a> </td> <td style="text-align:left;"> Coupled PK + indirect-response pharmacodynamic model for absolute lymphocyte count during oral PF-04878691 (TLR7 agonist) administration in healthy adult volunteers (Jones 2011 BJCP). PK is the two-compartment time-varying clearance model from the companion file Jones_2011_PF04878691.R (Table 1; all PK structural parameters and IIVs fixed at the published Table 1 values so the PK forcing function is the published popPK profile). Drug stimulates the re-distribution (loss) of lymphocytes through a power function on kout: dLYMPH/dt = kin - kout * (1 + slope * Cc^gamma) * LYMPH, with baseline lymphocyte count rbase = kin / kout so that kin = rbase * kout (Jones 2011 Methods / Table 3 lymphocyte model). The typical Emax indirect-response model could not adequately identify the parameters given the limited number of dose levels studied, so the Emax * Cc^gamma / (EC50^gamma + Cc^gamma) drug effect was replaced with the power function slope * Cc^gamma (Methods 'Population PK-OAS and PK-lymphocyte models'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jones_2011_PF04878691_HCV.html">PF04878691 oas (Jones 2011)</a> </td> <td style="text-align:left;"> Coupled PK + indirect-response pharmacodynamic model for the 2',5'-oligoadenylate synthetase (OAS) gene-expression fold change during oral PF-04878691 (TLR7 agonist) administration in healthy adult volunteers (Jones 2011 BJCP). PK is the two-compartment time-varying clearance model inherited from the companion file Jones_2011_PF04878691.R (Table 1; all PK structural parameters and IIVs fixed at the published Table 1 values so the PK forcing function is the published popPK profile). Drug stimulates the production of OAS through a power function: dOAS/dt = kin * (1 + slope * Cc^gamma) - kout * OAS, with baseline OAS fold change rbase = kin / kout so that kin = rbase * kout (Jones 2011 Methods / Tables 2 OAS model). The typical Emax indirect-response model could not adequately identify the parameters given the limited number of dose levels studied, so the Emax * Cc^gamma / (EC50^gamma + Cc^gamma) drug effect was replaced with the power function slope * Cc^gamma (Methods 'Population PK-OAS and PK-lymphocyte models'). The OAS observation is unitless (fold change from baseline). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jones_2011_PF04878691_HCV.html">PF04878691 viralLoad (Jones 2011)</a> </td> <td style="text-align:left;"> Combined PK + OAS + HCV viral RNA pharmacodynamic chain for oral PF-04878691 (TLR7 agonist) used to predict the antiviral efficacy of PF-04878691 in chronic hepatitis C (HCV) patients (Jones 2011 BJCP Figure 10 simulation). PK is the two-compartment time-varying clearance model from Jones_2011_PF04878691.R (Table 1; all PK structural parameters fixed at the published Table 1 values plus the IIVs on CL_SS and ka). The PF-04878691 OAS indirect-response sub-model is the same as Jones_2011_PF04878691_oas.R (Table 2; OAS in fold-change units, baseline rbase_oas = 0.96, drug stimulates production through slope * Cc^gamma). The OAS-viral-load relationship was fit on TLR9-agonist (CPG-10101) data in HCV patients (Jones 2011 Table 4) and is assumed transferable to PF-04878691 under the paper's explicit translation assumption that 'both TLR7 and TLR9 work through the same pathway'. The viral-load model is an inhibitory sigmoid Imax driven by the change in OAS from baseline expressed as a fold change oas_fc_above = oas(t) / rbase_oas - 1, so at the OAS baseline the viral-load deviation from BASE is zero. The viral RNA observation (vload) is in log10 copies/mL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Grasela_1985_phenobarbital.html">Phenobarbital (Grasela 1985)</a> </td> <td style="text-align:left;"> One-compartment population PK model for phenobarbital in preterm neonates (Grasela & Donn 1985), derived from routine clinical data via NONMEM. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Saleh_2023_leicns_pk30_mouse_brain.html">Phenobarbital mouse pbpk (Saleh 2023)</a> </td> <td style="text-align:left;"> PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, ICR). Nine-compartment CNS PBPK model predicting unbound phenobarbital concentrations in brain extracellular fluid (brain ECF) after a single 10 mg/kg intraperitoneal dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug's physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.0121 (estimated by the authors; Table V). The plasma PK model is the empirical one-compartment model of Table IV (in-house NONMEM fit) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hennig_2015_phenytoin.html">Phenytoin (Hennig 2015)</a> </td> <td style="text-align:left;"> One-compartment population PK model for phenytoin in critically ill children with a linear partition coefficient describing protein binding to albumin (Hennig 2015). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tanaka_2012_phenytoin.html">Phenytoin (Tanaka 2012)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for phenytoin after IV fosphenytoin sodium administration in Japanese healthy volunteers and adult / pediatric patients (Tanaka 2012). The fosphenytoin compartment converts first-order (K12) to the phenytoin central compartment; phenytoin is cleared from central and exchanges with a peripheral compartment via Q. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yukawa_1990_phenytoin.html">Phenytoin (Yukawa 1990)</a> </td> <td style="text-align:left;"> Steady-state Michaelis-Menten population PK model for phenytoin in 334 Japanese epilepsy outpatients on chronic oral phenytoin (Yukawa 1990 Model 2). Covariate effects on Vmax (allometric body weight, co-anticonvulsants) and Km (age <15 yr, co-anticonvulsants); dose-dependent powder bioavailability. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Boer-Perez_2026_piperacillin.html">Piperacillin (Boer-Perez 2026)</a> </td> <td style="text-align:left;"> One-compartment population PK model for piperacillin in preterm and term neonates with severe infections (Boer-Perez 2026); body-weight allometric scaling, sigmoidal postmenstrual-age maturation on CL fixed from Rhodin 2009, and a power effect of serum creatinine on CL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bulitta_2007_piperacillin.html">Piperacillin (Bulitta 2007)</a> </td> <td style="text-align:left;"> Two-compartment first-order IV population PK model for piperacillin in 8 adult cystic-fibrosis patients and 26 adult healthy volunteers receiving 4 g piperacillin as a 5-min intravenous infusion (Bulitta 2007). Lean body mass (LBM) is the size descriptor with allometric scaling (exponents 0.75 on CL and Q, 1.0 on V1 and V2; LBM_STD = 53 kg). A cystic-fibrosis disease-state indicator multiplicatively scales V1 and V2 via fcyf_vss^DIS_CF (fcyf_vss = 0.926), with fcyf_cl^DIS_CF retained on CL at its boundary estimate of 1.00 for model-form traceability. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bulitta_2010_piperacillin.html">Piperacillin (Bulitta 2010)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for piperacillin in healthy adult volunteers after a single intravenous infusion, with first-order non-renal clearance and parallel first-order plus mixed-order (Michaelis-Menten) renal elimination, allometrically scaled to 70 kg; a urine compartment accumulates the renally excreted amount (Bulitta 2010 Model 3, final model, NONMEM estimates) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/CohenWolkowiez_2014_piperacillin_tazobactam.html">Piperacillin (CohenWolkowiez 2014)</a> </td> <td style="text-align:left;"> One-compartment population PK model for piperacillin in premature and term infants under 61 days postnatal age (Cohen-Wolkowiez 2014); linear body-weight scaling on CL and V (fixed exponent = 1) and a power effect of postmenstrual age on CL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jeon_2014_piperacillin.html">Piperacillin (Jeon 2014)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for piperacillin in 50 Korean adult burn-ICU patients receiving piperacillin-tazobactam 4.5 g (4 g piperacillin + 0.5 g tazobactam) every 8 h as a 30-min infusion (Jeon 2014) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2012_piperacillin.html">Piperacillin (Landersdorfer 2012)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for piperacillin in healthy adult volunteers after intravenous infusion, with parallel first-order plus mixed-order (Michaelis-Menten) renal clearance and first-order non-renal clearance; a urine compartment accumulates the renally excreted amount (Landersdorfer 2012 Model 3, the final model) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nichols_2016_piperacillin_tazobactam.html">Piperacillin (Nichols 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK model for piperacillin in critically ill children (1-9 years) receiving extended-infusion piperacillin-tazobactam (Nichols 2016); IV zero-order input, first-order elimination, and a linear-additive effect of body weight on CL centered at the cohort median 18 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/ObrinkHansen_2015_piperacillin.html">Piperacillin (ObrinkHansen 2015)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for piperacillin in critically ill adults with septic shock (Obrink-Hansen 2015); linear first-order elimination with an additive linear effect of plasma creatinine on clearance, IIV on CL and central volume, and a proportional residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LopezArguello_2023_pbp_binding_pseudomonas.html">Piperacillin qsp (LopezArguello 2023)</a> </td> <td style="text-align:left;"> QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for piperacillin (penicillin): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wenker_2024_piperacillin_tazobactam.html">Piperacillin tazobactam (Wenker 2024)</a> </td> <td style="text-align:left;"> In vitro (hollow-fibre infection model, HFIM; six Gram-negative strains). Sigmoidal Emax PK/PD-index model for piperacillin/tazobactam in its clinical 8:1 ratio. Wenker 2024 ran a 21-experiment dose-fractionation study (7 growth controls plus 14 treated arms) in an HFIM and fit, by non-linear least squares on the 24 h timepoint, the change in bacterial density over 24 h as a function of each candidate PK/PD index. fT>MIC (the percentage of the dosing interval during which the free piperacillin/tazobactam concentration exceeds the isolate MIC) was the best-fitting index (AIC 94.1, R^2 0.691) versus fCmax/MIC (AIC 102, R^2 0.526) and fAUC/MIC (AIC 96.9, R^2 0.62). The packaged model is the fT>MIC model of Wenker 2024 Figure 1: effect = e0 - emax * FTMIC_TZP^hill / (ec50^hill + FTMIC_TZP^hill), where effect is the change in log10(CFU/mL) accrued over the 24 h experiment (positive = net growth, negative = net kill), e0 = 4.155 (FIXED) is the change in the untreated growth controls, emax = 8.851 (FIXED) is the maximum achievable reduction of that change, ec50 = 51.01 %fT>MIC is the index value giving half-maximal effect (the paper calls it EI50), and hill = 2.11 is the sigmoidicity coefficient (the paper calls it gam). There is NO pharmacokinetic component: exposure enters solely through the externally supplied per-experiment PK/PD index FTMIC_TZP, and the model has no ODE states because Wenker 2024 fit only the 24 h endpoint and never reported the starting inoculum numerically. The corresponding PK/PD targets recovered from these coefficients are 48% fT>MIC for bacteriostasis, 60% fT>MIC for 1 log10 kill and 75% fT>MIC for 2 log10 kill; the paper reports 48%, 60% and 77% respectively (see the vignette Errata for the 2 log10 discrepancy). Neither between-experiment variability nor a residual error magnitude was reported, so no eta parameters are present and addSd is FIXED at 0; the model is intended for typical-value and target-derivation simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hoglund_2012_piperaquine.html">Piperaquine (Hoglund 2012)</a> </td> <td style="text-align:left;"> Population PK model for oral piperaquine in pregnant and non-pregnant Sudanese women with uncomplicated Plasmodium falciparum malaria (Hoglund 2012). Three-transit-compartment absorption (ka = ktr) into a three-compartment disposition model. Body weight is the only retained covariate, applied as an allometric function on all clearances (fixed exponent 0.75) and volumes (fixed exponent 1.0). Relative bioavailability F is fixed at 1. The final model retains BSV on CL and F, treats MTT between-occasion variability as forward-simulation IIV, and uses an additive residual on the log-transformed observation (proportional in linear concentration space). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hoglund_2017_piperaquine.html">Piperaquine (Hoglund 2017)</a> </td> <td style="text-align:left;"> Population PK model for oral piperaquine in adults, children, and healthy volunteers across 11 pooled clinical studies (Hoglund 2017; individual-participant-data meta-analysis, n = 728). Two-transit- compartment absorption with kA = kTR feeding a three-compartment disposition model. Allometric body weight scaling on all clearances (fixed exponent 0.75) and volumes (fixed exponent 1.0) with reference weight 54 kg. Enzyme maturation function on elimination clearance (Hill-type sigmoid with MF50 = 0.575 y, Hill = 5.51). Dose-occasion effect adds 23.7% to relative bioavailability per consecutive dose. Bioavailability anchored at 1 with IIV. Predictions are venous plasma piperaquine base concentrations (ng/mL); a separately estimated capillary-to-venous scale of 106% is documented but not applied because only venous output is simulated. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tarning_2008_piperaquine.html">Piperaquine (Tarning 2008)</a> </td> <td style="text-align:left;"> Population PK model for oral piperaquine in Burmese and Karen adults and children with uncomplicated Plasmodium falciparum malaria (Tarning 2008). Two-compartment disposition with first-order absorption (no lag) and elimination from the central compartment. Body weight is the only retained covariate: a linear (1 + theta * (WT - 48)) effect on apparent oral clearance CL/F and on apparent central volume of distribution Vc/F, centred on the cohort median of 48 kg. The combined three-dose and four-dose Artekin regimens were pooled; no treatment-regimen effect was retained. Exponential IIV on all five disposition / absorption parameters. Residual error is proportional in linear concentration space (the source paper fit an additive error on natural-log-transformed concentrations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tarning_2012_dihydroartemisinin_piperaquine.html">Piperaquine (Tarning 2012)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for oral piperaquine in 24 pregnant (second / third trimester) and 24 matched non-pregnant women with uncomplicated malaria treated with the fixed-dose oral dihydroartemisinin-piperaquine combination once daily for 3 days (Tarning 2012 AAC). Transit-compartment absorption with 5 fixed transit compartments (ktr = (n+1)/MTT with n=5); the drug-transit rate is set equal to the absorption rate from the last transit to central (single estimated ktr). F fixed at 1; CL/F and F carry proportional pregnancy effects (+45.0% on CL/F and +46.8% on F). IIV on CL/F (21.5% CV) and Vc/F (39.5% CV); between-occasion variability (BOV across 3 dose occasions) on MTT (45.8% CV) and F (56.3% CV) multiplexed by the OCC indicator. Additive residual on natural-log concentrations (sigma = 0.285), encoded as proportional residual on the linear-concentration scale per Kloprogge 2018 lumefantrine precedent. Companion file Tarning_2012_dihydroartemisinin.R models the co-administered dihydroartemisinin arm. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tarning_2014_piperaquine.html">Piperaquine (Tarning 2014)</a> </td> <td style="text-align:left;"> Population PK model for oral piperaquine in adults with uncomplicated Plasmodium falciparum malaria in Thailand (Tarning 2014; n = 30, fed vs fasting parallel design). Three-transit-compartment absorption (ka = ktr) feeding a three-compartment disposition model. Allometric body-weight scaling on all clearances (fixed exponent 0.75) and volumes (fixed exponent 1.0); 70 kg reference. Linear dose-occasion effect on relative bioavailability (+25.3% per consecutive dose, OCC = 1, 2, 3). Linear age effect on the first peripheral volume of distribution (+4.10% per year of age). Relative bioavailability anchored at 1 with between-dose-occasion variability (no BSV in the final model). Concomitant low-fat food was tested as a covariate but was not retained in the final model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kakara_2014_statins_LDLC.html">Pitavastatin (Kakara 2014)</a> </td> <td style="text-align:left;"> PD-only indirect-response Imax model for LDL-cholesterol lowering by pitavastatin (Kakara 2014). One LDL-C compartment with zero-order synthesis Kin inhibited by Imax * DOSE / (ID50 + DOSE), where DOSE is the current daily pitavastatin dose (mg/day) supplied as a time-varying covariate column. An additive 0.109 contribution to the inhibition fraction is applied when ezetimibe is coadministered (CONMED_EZE = 1). The LDL-C synthesis-elimination loop is set up at steady state by enforcing Kin = Baseline * Kout (Kout derived inside model() as Kin / Baseline). Baseline LDL-C is age-scaled as 152 * (AGE/62)^(-0.240). Imax (0.567), Kin (32.8 mg/dL/day), Baseline (152 mg/dL), the age power exponent (-0.240), the ezetimibe INH contribution (0.109), and the IIV magnitudes are shared with Kakara_2014_atorvastatin and Kakara_2014_rosuvastatin (one joint NONMEM 7.2 FOCE-INTER fit across 378 patients). Pitavastatin ID50 = 0.860 mg per Kakara 2014 Table 2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lu_2019_polatuzumab.html">Polatuzumab (Lu 2019)</a> </td> <td style="text-align:left;"> Integrated two-analyte population PK model of polatuzumab vedotin (anti-CD79b vc-MMAE antibody-drug conjugate) in adults with non-Hodgkin lymphoma (Lu 2019). The antibody-conjugated MMAE (acMMAE) is described by a two-compartment model with three parallel elimination pathways from the central compartment: a slowly-time-decaying nonspecific linear clearance (CL_NS, sigmoidal Hill decline with cycle), a rapidly-decaying linear clearance (CL_t, mono-exponential decline), and a saturable Michaelis-Menten clearance (CL_MM). All three acMMAE pathways feed unconjugated MMAE formation in the central MMAE compartment with relative conversion fractions FRAC_NS, FRAC_NS x FRAC_CLT, and FRAC_NS x FRAC_MM, modulated by a time-dependent multiplier (1 + FRAC_T x exp(-alpha x t)) on FRAC_NS that captures the cycle-over-cycle decline in MMAE formation. Unconjugated MMAE is described by an apparent two-compartment model with parallel linear (CL_MMAE) and Michaelis-Menten (Vmax_MMAE / KSS) elimination from its central compartment. Modeled in MMAE-equivalent micrograms (pola dose in ug/kg x weight in kg x 3.65 x 718 / 145001 -> MMAE-equivalent ug administered to the acMMAE central compartment), with concentrations in ng/mL = ug/L. The Asian-race indicator on acMMAE Vc (e_asian_vc = 0.929, i.e., 7.1% lower V1 in Asian patients) is retained from the Lu 2019 final model and was subsequently re-quoted and assessed as not clinically meaningful in the Shi 2020 ethnicity-sensitivity analysis (PMID 32770353) of the same upstream popPK model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lu_2017_polatuzumab_neuropathy.html">Polatuzumab neuropathy (Lu 2017)</a> </td> <td style="text-align:left;"> Time-to-event hazard model for the onset of grade >= 2 peripheral neuropathy (PN) during polatuzumab vedotin treatment in adults with relapsed/refractory B-cell non-Hodgkin lymphoma (Lu 2017). The PN hazard is driven by a hypothetical effect compartment receiving plasma antibody-conjugated MMAE (acMMAE) with first-order distribution k1e in and ke0 = k1e out, modulated by a Weibull time function on the drug-effect potency (alpha drug-effect, beta shape) and by twelve baseline-covariate proportional-hazard terms (age, body weight, sex, active grade 1 PN at baseline, prior radiotherapy, prior vinca alkaloid, prior platinum-based chemotherapy, rituximab combination, tumor histology DLBCL vs other-non-FL, baseline tumor sum of products of perpendicular diameters, baseline serum albumin). The acMMAE plasma driver is inlined from the published Lu 2019 integrated two-analyte popPK (acMMAE side only; see Lu_2019_polatuzumab.R) per the standing policy of reusing a published same-drug PK when the originally-used PK source (Lu 2015 ASCPT poster, unpublished) is not on disk. Both the instantaneous hazard and the cumulative hazard / survival outputs are exposed for direct VPC simulation of the Kaplan-Meier curve. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cheah_2016_polymyxin_Abaumannii_dynamics.html">Polymyxin AB3070294 (Cheah 2016)</a> </td> <td style="text-align:left;"> In vitro (Acinetobacter baumannii AB307-0294; clinical heteroresistant isolate). Mechanism-based PK/PD model for polymyxin B and colistin against A. baumannii in a dynamic one-compartment in vitro infection model (IVM). Identical structural model to Cheah_2016_polymyxin_ATCC19606 (three bacterial subpopulations bact_s/bact_r/bact_d with logistic carrying capacity, Bulitta 2010 lipid-A receptor-occupancy submodel, Hill-function killing, single-compartment adaptive-resistance turnover, IVM one-compartment PK); strain-specific parameter values from Cheah 2016 Table 1 column AB307-0294. Fitness cost G_inhib_max was not estimated for this strain (Table 1 'NE') so f_cost is held at 0. SC50 is inherited as a FIXED proxy from Bulitta 2015 (Cheah 2016 does not report it); see vignette Errata for the full inheritance list. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cheah_2016_polymyxin_Abaumannii_dynamics.html">Polymyxin ATCC19606 (Cheah 2016)</a> </td> <td style="text-align:left;"> In vitro (Acinetobacter baumannii ATCC 19606; heteroresistant reference strain). Mechanism-based PK/PD model for polymyxin B and colistin against A. baumannii in a dynamic one-compartment in vitro infection model (IVM). The bacterial system is partitioned into three subpopulations -- polymyxin-susceptible (bact_s, CFU_S in the paper), constitutively polymyxin-resistant (bact_r, CFU_R; killing rate fixed at zero), and dormant or extremely slowly replicating cells (bact_d, Pop_D; nonobservable on viable-count plates) -- with a logistic carrying capacity CFU_max constraining the total observable population and a first-order bidirectional transition between susceptible and dormant states (k_SD, k_DS). Polymyxin in the IVM reservoir (central compartment) follows one-compartment first-order kinetics (CL_IVM, V_IVM) with simulated elimination half-life 11.6 h. Polymyxin target-site binding follows the Bulitta 2010 lipid-A LPS receptor-occupancy model: competitive displacement of bound divalent cations (Ca2+ and Mg2+) by polymyxin gives F_bound_cations (Eq 4), and a Hill function of the unoccupied fraction (Hill_binding, EC50) gives F_polymyxin_eff (Eq 5). The effective polymyxin concentration C_polymyxin_eff is the F_polymyxin_eff-weighted reservoir concentration divided by (1 + R_adaptive) to encode adaptive resistance attenuation (Eq 6). Bacterial killing is a Hill function of C_polymyxin_eff (Eq 7; Kill_max fixed at 100/h, Hill_killing, KillC50). Adaptive resistance R_adaptive is a single-compartment turnover whose driver Stim is a Hill-1 of raw reservoir polymyxin concentration (Eq 8; S_max fixed at 300, SC50 inherited from Bulitta 2015 since Cheah 2016 does not report it), with rate constant k_adapt (Eq 9). For ATCC 19606 the fitness cost G_inhib_max was not estimated by the authors (Table 1 'NE') so f_cost is held at 0. Observation is the log10 of the drug-free agar viable count CFU_S + CFU_R (Eq 11). The model has no inter-experiment IIV (typical-value fit per strain) and the residual error is set to a tiny fixed value because Cheah 2016 does not report it -- see vignette Errata for the full inheritance / approximation list. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cheah_2016_polymyxin_Abaumannii_dynamics.html">Polymyxin FADDIAB008 (Cheah 2016)</a> </td> <td style="text-align:left;"> In vitro (Acinetobacter baumannii FADDI-AB008; clinical heteroresistant isolate; loss-of-LPS resistance mechanism). Mechanism-based PK/PD model for polymyxin B and colistin against A. baumannii in a dynamic one-compartment in vitro infection model (IVM). Identical structural model to Cheah_2016_polymyxin_ATCC19606 (three bacterial subpopulations bact_s/bact_r/bact_d with logistic carrying capacity, Bulitta 2010 lipid-A receptor-occupancy submodel, Hill-function killing, single-compartment adaptive-resistance turnover, IVM one-compartment PK); strain-specific parameter values from Cheah 2016 Table 1 column FADDI-AB008. For this strain the experimental data supported inclusion of a fitness cost f_cost on susceptible replication (Eq 10): G_inhib_max = 0.994 (Table 1). SC50 is inherited as a FIXED proxy from Bulitta 2015 (Cheah 2016 does not report it); see vignette Errata. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cheah_2016_polymyxin_Abaumannii_dynamics.html">Polymyxin FADDIAB030 (Cheah 2016)</a> </td> <td style="text-align:left;"> In vitro (Acinetobacter baumannii FADDI-AB030; clinical polymyxin-susceptible isolate without heteroresistance). Mechanism-based PK/PD model for polymyxin B and colistin against A. baumannii in a dynamic one-compartment in vitro infection model (IVM). Identical structural model to Cheah_2016_polymyxin_ATCC19606 (three bacterial subpopulations bact_s/bact_r/bact_d with logistic carrying capacity, Bulitta 2010 lipid-A receptor-occupancy submodel, Hill-function killing, single-compartment adaptive-resistance turnover, IVM one-compartment PK); strain-specific parameter values from Cheah 2016 Table 1 column FADDI-AB030. For this strain the experimental data supported inclusion of a fitness cost f_cost on susceptible replication (Eq 10): G_inhib_max = 0.991 (Table 1). Also notable for this strain: a very steep Hill_killing of 19.5 (Table 1), reflecting near-switchlike polymyxin killing once C_eff approaches KillC50. SC50 is inherited as a FIXED proxy from Bulitta 2015 (Cheah 2016 does not report it); see vignette Errata. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Liang_2023_polymyxinB.html">PolymyxinB (Liang 2023)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model for polymyxin B in critically ill adults not receiving CRRT or ECMO (Liang 2023). Fitted by nonparametric adaptive grid (NPAG) in Pmetrics. Serum albumin is a power covariate on CL and age is a power covariate on Vc, both normalized to the cohort median (ALB 31.45 g/L, age 68 years) with exponents fixed at -0.95 and +0.95 respectively. Combined additive plus proportional residual error derived from the Pmetrics assay-error polynomial scaled by the final gamma. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nicholas_2009_ponezumab.html">Ponezumab (Nicholas 2009)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model for PF-04360365 (ponezumab), a humanized anti-amyloid IgG2 delta-a monoclonal antibody, in adults with mild-to-moderate Alzheimer's disease; allometric body-weight scaling with estimated exponents on every disposition parameter and a full 4x4 inter-individual block on (CL, V1, V2, Q) (Nicholas 2009 preliminary popPK) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dvorackova_2023_posaconazole.html">Posaconazole (Dvorackova 2023)</a> </td> <td style="text-align:left;"> Population PK model for oral posaconazole tablets in adult lung-transplant recipients (Dvorackova 2023). One-compartment disposition with first-order absorption and first-order elimination, parameterised on the apparent (oral) scale as CL/F and Vd/F because bioavailability was not identifiable from the oral-only therapeutic-drug-monitoring data. The absorption rate constant Ka was fixed to 0.8 1/h (back-calculated from the tmax and half-life reported in the posaconazole SmPC) because all concentrations were sampled in the elimination phase; inter-individual variability on Ka was nevertheless estimated and is very large, reflecting that absorption is essentially unidentifiable from these data. Age is the only covariate retained in the final model, entering log-linearly (exponentially) on apparent clearance so that CL/F declines by about 0.9 percent per year of age. Residual variability is proportional. The model was used for Monte Carlo dose optimisation against the EUCAST trough targets of 0.7 mg/L for prophylaxis and 1.25 mg/L for therapy. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kohl_2010_posaconazole.html">Posaconazole (Kohl 2010)</a> </td> <td style="text-align:left;"> One-compartment population PK model for prophylactic oral posaconazole in adult allogeneic stem cell transplant recipients with hematological malignancies (Kohl 2010); ka fixed, age and concurrent diarrhea as covariates. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shu_2024_posaconazole.html">Posaconazole (Shu 2024)</a> </td> <td style="text-align:left;"> Population PK model for posaconazole oral suspension (Noxafil) in Chinese hematopoietic stem cell transplantation (HSCT) recipients (Shu 2024). One-compartment disposition with first-order absorption and first-order elimination, parameterised on the apparent (oral) scale as CL/F and V/F because the therapeutic-drug-monitoring data were oral-only and bioavailability was not identifiable. The absorption rate constant Ka was fixed to 0.4 1/h from earlier posaconazole-suspension popPK studies, because almost every sample was a pre-dose trough and the absorption phase could not support an estimate. Creatinine clearance enters apparent clearance as a power function centred on the cohort median 103.81 mL/min; body weight enters apparent volume as a power function centred on the cohort median 45.85 kg; and concomitant proton-pump-inhibitor use multiplies apparent volume by 3.83. Inter-individual variability is exponential on CL/F and V/F and is very large (omega 1.118 and 0.826 on the log scale). Residual variability is proportional. The model was used for Monte Carlo dose optimisation of weight-banded BID and TID regimens against steady-state trough targets of 0.7 ug/mL for prophylaxis and 1.0 ug/mL for treatment of invasive fungal disease. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vanIersel_2018_posaconazole.html">Posaconazole (vanIersel 2018)</a> </td> <td style="text-align:left;"> Population PK model for the delayed-release solid oral tablet formulation of posaconazole in adult healthy volunteers and patients at high risk for invasive fungal disease (van Iersel 2018). One-compartment disposition with sequential zero-order then first-order absorption: each oral dose loads into the depot compartment as a zero-order infusion of duration D1, after which depot drains to central with first-order rate constant ka and central eliminates with first-order rate constant CL/V. The random effect on D1 is the same as the random effect on ka multiplied by a correlation factor (cor_kad1 = -0.586). Covariates retained in the final model are body weight on relative bioavailability (allometric power exponent), tablet formulation A/B versus C/D on bioavailability, AML/MDS disease state on bioavailability, fed status on absorption rate, and single-dose-versus-multiple-dose record indicator on clearance. Residual variability is log-additive with separate magnitudes for phase 1 versus phase 3 studies. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PerezRuixo_2025_posdinemab.html">Posdinemab (PerezRuixo 2025)</a> </td> <td style="text-align:left;"> Mechanism-based population PK-PD model with full TMDD for the anti-tau monoclonal antibody posdinemab in serum, CSF, and ISF (Perez-Ruixo 2025): two-compartment serum disposition with linear elimination, distribution into a CSF compartment and a downstream ISF compartment, explicit second-order binding of free posdinemab to free p217+tau in CSF and to tau seeds in ISF, internalization of free target and drug-target complex, and Alzheimer's-disease-vs-healthy effect on baseline p217+tau. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lin_2024_pozelimab.html">Pozelimab (Lin 2024)</a> </td> <td style="text-align:left;"> Two-compartment two-binding-site TMDD-QE population PK model of total pozelimab and total C5 in healthy volunteers, adults with paroxysmal nocturnal hemoglobinuria, and pediatric and adult patients with CHAPLE disease (Lin 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ide_2009_pravastatin.html">Pravastatin (Ide 2009)</a> </td> <td style="text-align:left;"> Population PK model for orally administered pravastatin with enterohepatic circulation (Ide 2009) in healthy Japanese male volunteers. Absorption is described by an Erlang chain of 8 transit compartments (N_depot = 8); disposition is one-compartment central with a gallbladder recirculation compartment whose release is gated by the gallbladder-emptying time tg (continuous filling from central via k12 for t < tg, gated release to central via k21 for t >= tg) producing the characteristic second-peak phenomenon. SLCO1B1 *15 haplotype carrier status (paired heterozygote / homozygote indicators) increases relative oral bioavailability Frel multiplicatively (1.50x and 1.95x respectively). Gastric conversion of pravastatin to its inactive 3'alpha-isopravastatin (RMS-416) is highly variable; the source paper corrected for this by using an apparent dose (actual dose x Fa, where Fa = AUCpra / (AUCpra + AUCrms)) as the model input, so the packaged model fixes the depot bioavailability anchor at the population-mean Fa = 0.571 derived from Table II mean AUC values. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shoji_2011_pregabalin.html">Pregabalin (Shoji 2011)</a> </td> <td style="text-align:left;"> One-compartment population PK model for pregabalin in adults (Shoji 2011 BJCP; pooled healthy volunteers, subjects with impaired renal function, and patients with post-herpetic neuralgia or diabetic peripheral neuropathy from 14 clinical trials). CL/F is proportional to Cockcroft-Gault creatinine clearance (capped at an estimated break point) with an additional ideal-body-weight power effect. V/F depends on ideal body weight, body mass index, age, and sex. Absorption rate and lag-time are reduced by a high-fat meal at the time of dosing. Combined proportional + additive residual error is stratified by healthy-vs-patient status. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bender_2009_pregabalin_rat.html">Pregabalin rat binary (Bender 2009)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment population PK model for pregabalin in male Sprague-Dawley rats following a 2 h intravenous infusion (4 or 10 mg/kg/h) in a chronic- constriction-injury (CCI) neuropathic-pain model, with the concomitant administration of sildenafil encoded as a BINARY presence indicator (CONMED_SILDENAFIL). Sildenafil presence reduces pregabalin clearance by a fixed fraction (theta_SLD = 0.302, i.e. 30.2% reduction) per the paper's discrete-covariate parameterisation; the alternative continuous saturable-metabolite parameterisation is encoded in the companion file Bender_2009_pregabalin_rat_smetab.R. Crossover design with two occasions per rat (Day 1 / Day 4 with a washout) carries between-occasion variability on CL and Vc multiplexed by the OCC indicator. Parameter values from Bender 2009 Table IV (Binary Sildenafil Covariate column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bender_2009_pregabalin_rat.html">Pregabalin rat smetab (Bender 2009)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment population PK model for pregabalin in male Sprague-Dawley rats following a 2 h intravenous infusion (4 or 10 mg/kg/h) in a chronic- constriction-injury (CCI) neuropathic-pain model, with the concomitant administration of sildenafil encoded as a CONTINUOUS saturable inhibition driven by the time-varying plasma concentration of sildenafil's active N-methyl metabolite (SLDM). Effective CL = theta_CL * (1 - SLDM / (theta_SLD + SLDM)) with theta_SLD = 1350 ng/mL acting as the IC50 of metabolite-driven inhibition. Statistically the preferred parameterisation in the paper (delta-OFV = -42.6 vs the no-covariate base; the simpler binary form is in the companion file Bender_2009_pregabalin_rat_binary.R with delta-OFV = -8.5). Crossover design with two occasions per rat (Day 1 / Day 4 with a washout) carries between-occasion variability on CL and Vc multiplexed by the OCC indicator. Parameter values from Bender 2009 Table IV (Continuous Sildenafil Metabolite Covariate column). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gaohua_2012_pregnancy_pbpk.html">Pregnancy pbpk caffeine (Gaohua 2012)</a> </td> <td style="text-align:left;"> PBPK (whole-body, 14-compartment pregnancy p-PBPK adapted from the Simcyp Simulator version 11 full-PBPK platform). Caffeine (CYP1A2 substrate) disposition in healthy Caucasian women aged 20-40 years, with gestational-age-dependent maternal physiology and hepatic CYP1A2 activity. The 14 ODE compartments are arterial blood, venous blood, lung, adipose, bone, brain, heart, kidney, gut, liver, muscle, skin, spleen, and a lumped fetoplacental unit (fetus + placenta + amniotic fluid + membranes + umbilical cord) per Gaohua 2012 Figure 1; the uterus and mammary glands are merged into the muscle compartment, so muscle volume and flow are computed as the residual that balances total body weight and cardiac output during pregnancy. Time-varying physiology (cardiac output, body weight, plasma / RBC volumes, hematocrit, serum albumin, skin / adipose / renal / fetoplacental blood flows, and CYP1A2 / CYP2D6 / CYP3A4 enzyme activities) follows the polynomial formula X = X0 * (a0 + a1*GA + a2*GA^2 + a3*GA^3) in Table 2; the fetoplacental volume follows the Gompertz curve in Eq. 1. Drug-specific values for fa, Fg, ka, fu, B:P, basal CL_int,H, the CYP fractional contributions A_1A2 / A_2D6 / A_3A4, and the 12 tissue:plasma partition coefficients (Rodgers and Rowland) are from Tables 3-4. Set covariate GA = 0 to simulate the non-pregnant reference woman; values 0 < GA < 40 simulate any gestational stage. The model is a perfusion-limited typical-value PBPK forward simulation; the paper added no IIV or residual-error model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gaohua_2012_pregnancy_pbpk.html">Pregnancy pbpk metoprolol (Gaohua 2012)</a> </td> <td style="text-align:left;"> PBPK (whole-body, 14-compartment pregnancy p-PBPK adapted from the Simcyp Simulator version 11 full-PBPK platform). Metoprolol (predominantly CYP2D6 substrate; minor CYP3A4 contribution) disposition in healthy Caucasian women aged 20-40 years, with gestational-age-dependent maternal physiology and hepatic CYP2D6 / CYP3A4 activity. The 14 ODE compartments are arterial blood, venous blood, lung, adipose, bone, brain, heart, kidney, gut, liver, muscle, skin, spleen, and a lumped fetoplacental unit (fetus + placenta + amniotic fluid + membranes + umbilical cord) per Gaohua 2012 Figure 1; the uterus and mammary glands are merged into the muscle compartment, so muscle volume and flow are computed as the residual that balances total body weight and cardiac output during pregnancy. Time-varying physiology (cardiac output, body weight, plasma / RBC volumes, hematocrit, serum albumin, skin / adipose / renal / fetoplacental blood flows, and CYP1A2 / CYP2D6 / CYP3A4 enzyme activities) follows the polynomial formula X = X0 * (a0 + a1*GA + a2*GA^2 + a3*GA^3) in Table 2; the fetoplacental volume follows the Gompertz curve in Eq. 1. Drug-specific values for fa, Fg, ka, fu, B:P, basal CL_int,H, the CYP fractional contributions A_1A2 / A_2D6 / A_3A4, and the 12 tissue:plasma partition coefficients (Rodgers and Rowland) are from Tables 3-4. Set covariate GA = 0 to simulate the non-pregnant reference woman; values 0 < GA < 40 simulate any gestational stage. The model is a perfusion-limited typical-value PBPK forward simulation; the paper added no IIV or residual-error model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Gaohua_2012_pregnancy_pbpk.html">Pregnancy pbpk midazolam (Gaohua 2012)</a> </td> <td style="text-align:left;"> PBPK (whole-body, 14-compartment pregnancy p-PBPK adapted from the Simcyp Simulator version 11 full-PBPK platform). Midazolam (CYP3A4 substrate) disposition in healthy Caucasian women aged 20-40 years, with gestational-age-dependent maternal physiology and hepatic CYP3A4 activity. The 14 ODE compartments are arterial blood, venous blood, lung, adipose, bone, brain, heart, kidney, gut, liver, muscle, skin, spleen, and a lumped fetoplacental unit (fetus + placenta + amniotic fluid + membranes + umbilical cord) per Gaohua 2012 Figure 1; the uterus and mammary glands are merged into the muscle compartment, so muscle volume and flow are computed as the residual that balances total body weight and cardiac output during pregnancy. Time-varying physiology (cardiac output, body weight, plasma / RBC volumes, hematocrit, serum albumin, skin / adipose / renal / fetoplacental blood flows, and CYP1A2 / CYP2D6 / CYP3A4 enzyme activities) follows the polynomial formula X = X0 * (a0 + a1*GA + a2*GA^2 + a3*GA^3) in Table 2; the fetoplacental volume follows the Gompertz curve in Eq. 1. Drug-specific values for fa, Fg, ka, fu, B:P, basal CL_int,H, the CYP fractional contributions A_1A2 / A_2D6 / A_3A4, and the 12 tissue:plasma partition coefficients (Rodgers and Rowland) are from Tables 3-4. Set covariate GA = 0 to simulate the non-pregnant reference woman; values 0 < GA < 40 simulate any gestational stage. The model is a perfusion-limited typical-value PBPK forward simulation; the paper added no IIV or residual-error model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mehta_2023_tb_lesion_mpbpk.html">Pretomanid mpbpk (Mehta 2023)</a> </td> <td style="text-align:left;"> mPBPK (minimal physiologically based, translational mouse-to-human). Pretomanid in pulmonary tuberculosis patients, with explicit cavitary lung-lesion and uninvolved-lung site-of-action compartments. First-order oral absorption with saturable, dose-dependent bioavailability F = 1 / (1 + DOSE/ED50), linear clearance from blood, two lumped peripheral tissue pools that split cardiac output and residual body volume by a fixed fraction, and perfusion-limited lesion / uninvolved-lung compartments driven by penetration ratios. Physiological volumes and flows are computed from body weight (Brown 1997); clearance is allometrically scaled from the mouse fit with an exponent of 0.75. The lesion and uninvolved-lung states hold concentrations (mg/L), not amounts. Intended for typical-value and population simulation of site-of-action target attainment. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Saleh_2023_leicns_pk30_mouse_brain.html">Prexasertib mouse pbpk (Saleh 2023)</a> </td> <td style="text-align:left;"> PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, CD1 nude). Nine-compartment CNS PBPK model predicting unbound prexasertib concentrations in brain extracellular fluid (brain ECF) after a single 10 mg/kg subcutaneous dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug's physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.09 (estimated by the authors; Table V). The plasma PK model is the empirical two-compartment model of Table IV (ref [22]) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wattanakul_2024_primaquine_lactation.html">Primaquine (Wattanakul 2024)</a> </td> <td style="text-align:left;"> Mechanistic joint parent-metabolite population PK model for oral primaquine and its carboxyprimaquine metabolite in lactating women with Plasmodium vivax infection, fitted simultaneously to maternal venous plasma, capillary plasma, and breast-milk concentrations. Four-transit absorption (NN = 4, ka set equal to ktr) into a one-compartment primaquine disposition model whose entire systemic clearance is a formation clearance into a one-compartment carboxyprimaquine disposition model, plus first-pass conversion of a fraction FM of the absorbed dose directly into carboxyprimaquine. Each central compartment exchanges with its own breast-milk compartment through a shared apparent inter-compartmental clearance and an analyte-specific milk:plasma partition coefficient; the exchange is gated by a square-wave breastfeeding function (10 feeds/day, 24-minute feeding window in a 2.4-hour cycle). Separate capillary:venous conversion factors are estimated for each analyte. Allometric body-weight scaling on clearance (0.75) and volume (1) referenced at WT = 51 kg; inter-occasion variability on bioavailability and mean transit time across four sampling occasions. All amounts are molar (nmol) because the primaquine to carboxyprimaquine conversion is 1:1 molar. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wattanakul_2024_primaquine_lactation.html">Primaquine motherinfant (Wattanakul 2024)</a> </td> <td style="text-align:left;"> Mechanistic mother-to-infant transfer model for primaquine and carboxyprimaquine, extending the maternal population PK model of Wattanakul 2024 (see modellib('Wattanakul_2024_primaquine')) with a breastfed-infant compartment chain so that infant exposure can be predicted from maternal dosing. The maternal side is unchanged apart from the milk compartments now being drained by breastfeeding. During each feeding window a square-wave gate transfers the entire breast-milk content of each analyte into the infant's dose compartment; between feeds the gate closes and the milk compartments re-equilibrate with maternal plasma. The infant absorbs through a two-transit chain with mean transit time fixed to a paediatric literature value, and has no first-pass metabolism. Infant clearances and volumes are scaled from the mother's individual estimates by infant body weight (allometric exponents 0.75 and 1), with monoamine-oxidase-A maturation applied to primaquine clearance only. No random effects are placed on the infant parameters: all predicted infant variability is inherited from the mother. Simulation model -- no infant pharmacokinetic parameter was estimated, because all but one infant primaquine sample was below the limit of quantification. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cao_2013_PRO95780.html">PRO95780 (Cao 2013)</a> </td> <td style="text-align:left;"> Second-generation minimal physiologically-based PK (mPBPK) model for PRO95780 (drozitumab) in adults (Cao 2013 Model A; clearance from plasma) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mohamed_2013_procainamide.html">Procainamide (Mohamed 2013)</a> </td> <td style="text-align:left;"> Joint parent-metabolite two-compartment popPK model for procainamide and its major active N-acetyl metabolite N-acetylprocainamide (NAPA) in a single 40-year-old male with chronic kidney disease stage 5 receiving continuous renal replacement therapy (CRRT) after aortic valve replacement (Mohamed 2013). Structural model (paper equations 1-3): procainamide central and peripheral compartments with linear distribution clearance (Cld) and two parallel first-order elimination arms from the central compartment - Cl_other combining all non- metabolism procainamide clearance pathways (CRRT, residual renal, and any additional non-NAPA-forming metabolic routes) and Cl_f,napa the formation clearance of NAPA from procainamide; the metabolite NAPA is described as a single central compartment fed by the Cl_f,napa flux out of the procainamide central compartment and eliminated linearly with total clearance Cl_napa (CRRT plus residual renal). The NAPA apparent volume of distribution (Vc)_N was FIXED at 100 L, derived as the patient's body weight (70 kg) times a literature Vd of 1.5 L/kg for NAPA in functionally anephric patients on hemodialysis (paper ref 9), because the fraction of procainamide elimination going through NAPA formation could not be identified independently from the concentration data. All other structural parameters were successfully estimated by first-order conditional estimation in NONMEM VII. RUV was described by a proportional error model (paper equation 4) on both procainamide and NAPA concentrations; the estimated residual- error magnitude is not reported in Mohamed 2013 Table 1, so this library implementation encodes a nominal 15 percent CV proportional SD on each output and documents the omission in the vignette Errata. Because the source is a single-subject case report, no between- subject variability is identifiable and no eta parameters are encoded; the model is a typical-value description tied to this specific patient's disposition on CRRT. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chi_2018_propofol.html">Propofol (Chi 2018)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for propofol target-controlled IV infusion in Chinese adults with hepatic insufficiency undergoing elective liver transplantation, with additive body-weight effect on clearance and power Child-Turcotte-Pugh score effect on peripheral volume (Chi 2018 final regression model). Typical-value-only model: the source paper reports the six final-model THETAs but provides no OMEGA (IIV), no SIGMA (residual error), and no GOF / VPC, so all etas are fixed at zero and no residual error term is included. See vignette Assumptions and deviations for the resulting limitations on VPC-style validation and the recommendation to consult the modellib('Ye_2012_propofol') companion (when extracted) for a fully-reported 3-compartment propofol popPK fit in a larger Chinese-multicenter cohort that shares the Chi 2018 first author. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Diepstraten_2013_propofol.html">Propofol (Diepstraten 2013)</a> </td> <td style="text-align:left;"> Three-compartment intravenous population PK model for propofol in morbidly obese and nonobese adults, adolescents, and children (Diepstraten 2013 meta-analysis of five previously published studies; N = 94 patients, TBW 37-184 kg, age 9-79 years). Final model E in Table 3: total body weight scales clearance allometrically with an estimated exponent and scales the slow inter-compartmental clearance Q3 linearly; age modifies clearance via a bilinear function centered at 41 years with separate slopes below and above the breakpoint. Inter-individual variability on CL, V1, V3, and Q3 (log-normal) and proportional intra-individual error on log-transformed concentrations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Koo_2012_propofol.html">Propofol (Koo 2012)</a> </td> <td style="text-align:left;"> Pharmacodynamic sigmoid Emax model for the probability of recovery of consciousness (ROC) versus propofol effect-site concentration (Ce, ug/mL) during emergence from propofol-remifentanil target-controlled-infusion (TCI) general anesthesia in 94 ASA I-II adult patients undergoing elective minor eye or ENT surgery (Koo 2012). Age modulates both the effect-site concentration at 50% probability of ROC (Ce50) and the Hill exponent lambda via linear-additive age-centred forms Ce50 = 1.15 - 0.0128 * (AGE - 43) and lambda = 9.69 - 0.141 * (AGE - 43). Inter-individual variability is log-normal on Ce50 (CV 26.0%); IIV on lambda was dropped from the final model. Propofol PK is not fit in the source paper -- the per-record effect-site propofol concentration is supplied as the time-varying covariate CEFFECT (driven by the Schnider 1998/1999 TCI controller in the source study; Keo = 0.459 /min). NONMEM Bernoulli LAPLACE likelihood in the source paper; this implementation exposes the typical-value probability of ROC with a placeholder additive residual error (see vignette Assumptions and deviations), following the Shin_2014_sevoflurane.R precedent from the same Yonsei research group. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Przybylowski_2015_propofol.html">Propofol (Przybylowski 2015)</a> </td> <td style="text-align:left;"> Three-compartment IV population PK plus effect-compartment sigmoidal Emax PD model for propofol in adult ASA III cancer patients undergoing major lung surgery under propofol-fentanyl total intravenous anesthesia (Przybylowski 2015; N = 23). The PD response is the AAI (A-line ARX-Index) auditory-evoked-potential depth-of-anesthesia index with the maximum effect fixed to 1 and the pretreatment baseline fixed to 87 from a prior study. Inter-individual variability was estimated on Vc, CL, and the deep-compartment intercompartmental clearance Q2 for PK and on Ce50, gamma (Hill), and ke0 for PD; IIV on Vt1, Q1, Vt2 was fixed to 0 (data uninformative). No demographic, biochemical, or hemodynamic covariates were retained in the final model (Results). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wahlquist_2024_propofol.html">Propofol (Wahlquist 2024)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for propofol with a symbolic-regression (neural-network) derived covariate model on all four inter-compartmental rate constants, the elimination rate constant and the central volume; trained on the pooled Eleveld 30-study dataset spanning neonates to the elderly </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2012_propofol.html">Propofol (Wang 2012)</a> </td> <td style="text-align:left;"> Three-compartment intravenous population PK model for propofol across the human life-span (Wang 2012; 174 subjects pooled across seven previously published studies covering preterm and term neonates, infants, toddlers, children, adolescents, and adults; body weight 0.68-122.7 kg, age 1 day-81 years). Final 'bodyweight-dependent exponent (BDE)' model (Model IV / Final PK model, Table IV): clearance is scaled by total body weight via a power function whose exponent k changes sigmoidally with body weight from k0 = 1.34 at a theoretical 0 kg to k0 - kmax = 0.55 at large body weights, with k50 = 3.78 kg and a Hill coefficient gamma = 5.24 governing the steepness of the decline. The slow inter-compartmental clearance Q3 and the second peripheral volume V3 scale linearly with body weight (BW/70); the first peripheral volume V2 scales as (BW/70)^0.55; the fast inter-compartmental clearance Q2 is independent of body weight. The central volume V1 = 7.58 L is constant for subjects with postnatal age >= 100 days and scales linearly as V1 * (BW/70) for younger subjects. Inter-individual variability (log-normal) was retained on CL, V1, V2, V3, and Q3; no IIV on Q2. Additive residual error on log-transformed concentrations was used, equivalent to a proportional error on the linear concentration scale. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Knibbe_2005_propofol.html">Propofol human (Knibbe 2005)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model for propofol in a 70 kg adult human, projected from male Wistar rat (0.25 kg) parameters via the allometric power model with literature exponents 0.75 for clearances and 1 for volumes. Parameter values are taken from Knibbe 2005 Table 3 (column 'Scaled for humans (70 kg)'); inter- and intra-individual variability are inherited from the rat fit (Table 3, column 'Observed in the rat (250 g)') per the Methods text 'these human scaled pharmacokinetic parameters, together with ... intra- and interindividual variabilities estimated in the rat were used to simulate propofol concentrations'. The companion file Knibbe_2005_propofol_rat.R carries the rat-side parameters used as the scaling anchor. Knibbe 2005 demonstrated that concentrations simulated from this scaled-human model agreed (r^2 = 0.83, P < 0.0001) with concentrations observed in long-term-sedated critically ill patients (Figure 2). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Levitt_2005_propofol_pbpk.html">Propofol pbpk (Levitt 2005)</a> </td> <td style="text-align:left;"> PBPK (whole-body). Human physiologically based pharmacokinetic model for propofol (Levitt and Schnider 2005 BMC Anesthesiology) fitted to the individual arterial plasma data of 24 adult volunteers in three age groups (18-34, 35-65, >65 years) from Schnider et al. 1998 Anesthesiology. Each subject received an initial ~20-second IV bolus (2 mg/kg for subjects <65; 1 mg/kg for subjects >65) followed 60 minutes later by a 60-minute constant infusion at 25, 50, 100, or 200 microg/kg/min. Twelve well-stirred flow-limited tissue compartments (liver, intestine [paper's 'portal'], muscle, kidney, brain, heart, lung, skin, adipose, bone, other [paper's 'other' + 'tendon' connective tissue lumped], plus venous and arterial blood) use the PKQuest standard-human physiology (organ weights, blood flows, and fraction lipid from Table 1 for a 70-kg reference); propofol-specific tissue/blood partition coefficients are hard-coded from the paper text at the standard freepl = 0.022 (adipose 84, brain 1.87, liver 2.12, intestine 1.7, other tissues 1.45). Lean tissue weights and blood flows scale by lean body mass; adipose scales with body fat percent (Gallagher 1996 regression internally derives body fat from WT, HT, AGE, SEXF, RACE_ASIAN). Hepatic elimination is a single well-stirred apparent-clearance approximation (CL_liver = 0.76 * (Q_hepatic artery + Q_portal)) calibrated to the paper's Table 4 age-averaged fractional liver clearance of 0.76; this replaces the paper's Roberts-Rowland dispersion-model liver (dispersion number 0.3) with an ODE-compatible well-mixed compartment. Optional pulmonary sequestration compartment (paper-specific 'lung_seq') receives a bolus fraction frdose = max(0, 0.548 - 0.00891 * AGE) and releases it exponentially with time constant T = 80 min (paper's average parameter set for Figs 9-11); the user allocates the sequestered bolus fraction via a second dose row on cmt = 'lung_seq'. No IIV or formal residual-error variance was reported (individual fits; 15 percent average absolute weighted residual). All parameters are FIXED at the paper's average values. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Knibbe_2005_propofol.html">Propofol rat (Knibbe 2005)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment intravenous population PK model for propofol in male Wistar rats following a single 30 mg/kg bolus delivered over 5 min, as reported in Table 3 (column 'Observed in the rat (250 g)') of Knibbe 2005. The underlying NONMEM fit was performed by Knibbe et al. (reference 11 of the paper) on 19 whole-blood samples from each of 22 chronically instrumented rats; Knibbe 2005 reproduces those rat point estimates and uses them as the species anchor for an allometric scaling to humans (see the companion model file Knibbe_2005_propofol_human.R, which carries the human-projected parameters from Table 3 column 'Scaled for humans (70 kg)'). Log-normal inter-individual variability on CL, V1, Q, V2 and a constant-CV proportional intra-individual residual error model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ngamprasertwong_2016_propofol_sheep.html">Propofol sheep (Ngamprasertwong 2016)</a> </td> <td style="text-align:left;"> Preclinical (sheep). Maternal-fetal population PK model of propofol in mid-gestational pregnant Dorset ewes (Ngamprasertwong 2016; N = 8 ewe-fetus pairs at 110-125 days gestation; term ~147-150 days). Two-compartment maternal disposition (central + peripheral1) linked to a single fetal compartment via a reversible inter-compartmental clearance QM-F; fetal clearance was tested but estimated near zero (<0.001 L/min, RSE >100%) and set to zero in the final model. Maternal clearance scales with heart rate via the normalised power model CL = theta1 * (HR/158)^theta2; no other covariate (gestational age, body weight, blood pressure, uterine blood flow) reached statistical significance. Inter-individual variability was estimated on CL and QM-F; IIV on Vc, Q, Vp, and VFetus was fixed to zero in the source and is omitted here. Residual error is purely proportional, with separate variances for maternal-ewe and fetal observations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/DelFrari_2018_propranolol.html">Propranolol (DelFrari 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral propranolol in infants (aged 50-243 days, 3.6-9.7 kg) with proliferative Infantile Hemangiomas (Del Frari 2018). First-order absorption and first-order elimination; apparent oral clearance CL/F scales with body weight using a fixed allometric exponent of 0.75 and a reference weight of 6.3 kg (the median weight pooled across visits D1-D84). Apparent volume V/F has no covariate effect (the paper tested but did not retain weight on V/F). Between-subject variability is retained on CL/F and Ka only; BSV on V/F was dropped from the final model (large 95% CI including 0 and 62.3% eta-shrinkage). Residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Takechi_2018_propranolol.html">Propranolol (Takechi 2018)</a> </td> <td style="text-align:left;"> One-compartment first-order absorption population PK model for oral propranolol in Japanese infants with infantile hemangioma (35-150 days postnatal age), with fixed allometric body-weight scaling and a power effect of postnatal age on apparent oral clearance; the companion logistic-regression PD model relating exposure (AUC), treatment duration, and gestational age to treatment-success probability is reproduced in the validation vignette. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pardridge_2023_brain_plasma_protein_binding.html">Propranolol pbpk (Pardridge 2023)</a> </td> <td style="text-align:left;"> PBPK (brain, semi-mechanistic). Pardridge 2023 Pharmaceutical Research partly-flow / partly-compartmental model of propranolol delivery to brain that resolves plasma-protein binding as an explicit kinetic process inside the brain capillary. Seven brain states carry albumin-bound, alpha-1-acid-glycoprotein (AGP)-bound and free drug in the brain capillary, free AGP in the capillary, free drug in brain, and free / drug-occupied brain cytoplasmic binding protein. Drug crosses the blood-brain barrier by bi-directional first-order permeation and may be metabolised in brain. The model quantifies plasma-protein-mediated uptake (PMU): because propranolol dissociates from AGP far faster in vivo in the capillary (KG in vivo 19 uM) than in vitro (KG in vitro 3.3 uM), the free drug in brain is roughly twofold higher than equilibrium dialysis of plasma predicts. Arterial input is either a constant total plasma concentration (steady-state IV-infusion model, c_plasma_ss, Table II LT0 = 100 nM) or a one-compartment first-order oral profile (non-steady-state model, Eq 13; dose the depot compartment and set c_plasma_ss to 0). All parameters are FIXED at the paper's literature-sourced values; the paper reports no IIV and no residual error because it is a deterministic simulation study rather than a fit to individual data. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/DeCock_2012_propyleneGlycol.html">PropyleneGlycol (DeCock 2012)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous propylene glycol (PG) excipient exposure in preterm and term neonates receiving paracetamol-PG or phenobarbital-PG (De Cock 2012). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Alsultan_2017_pyrazinamide.html">Pyrazinamide (Alsultan 2017)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with first-order absorption and first-order elimination for oral pyrazinamide in adults with drug-susceptible pulmonary tuberculosis (Alsultan 2017); body weight is an allometric covariate on CL/F and V/F (fixed exponents 0.75 and 1) and biological sex is an exponential covariate on V/F </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chirehwa_2017_pyrazinamide.html">Pyrazinamide (Chirehwa 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model with Savic-style transit-compartment absorption (NN = 28) for oral pyrazinamide in HIV/TB-coinfected adults on the WHO four-drug fixed-dose combination (Chirehwa 2017); fat-free mass (Janmahasatian formula) drives fixed allometric scaling of CL/F (exponent 0.75) and V/F (exponent 1.0) referenced to a 42 kg subject, and CL/F increases linearly by 14.3% from day 1 to day 29 of treatment, attributed to rifampin-mediated enzyme induction. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Horita_2018_pyrazinamide.html">Pyrazinamide (Horita 2018)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with three-compartment transit absorption followed by first-order absorption and first-order elimination for oral pyrazinamide in Ghanaian children with active tuberculosis (Horita 2018); allometric weight scaling on V/F (estimated exponent 0.677) and CL/F (estimated exponent 0.735) normalised to the cohort median 14.3 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kim_2023_pyrazinamide.html">Pyrazinamide (Kim 2023)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with first-order absorption and first-order elimination for oral pyrazinamide in Korean adults with drug-susceptible tuberculosis (Kim 2023); lean body mass is an allometric covariate on CL/F and V/F (fixed exponents 0.75 and 1) and geriatric diabetes mellitus (age >= 70 years with diabetes mellitus) increases CL/F by 32% </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mehta_2023_tb_lesion_mpbpk.html">Pyrazinamide mpbpk (Mehta 2023)</a> </td> <td style="text-align:left;"> mPBPK (minimal physiologically based, translational mouse-to-human). Pyrazinamide in pulmonary tuberculosis patients, with explicit cavitary lung-lesion and uninvolved-lung site-of-action compartments. This is the framework-qualification model of Mehta 2023: first-order oral absorption, first-order elimination from a blood compartment and perfusion-limited lesion / uninvolved-lung compartments driven by penetration ratios. No lumped peripheral tissue compartment -- the paper reports that adding one did not improve the fit. It is the only one of the paper's three models whose lesion and uninvolved-lung predictions could be checked against observed human site-of-action data. Physiological volumes and flows are computed from body weight (Brown 1997). The lesion and uninvolved-lung states hold concentrations (mg/L), not amounts. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ramachandran_2023_tuberculosis_eptb_pbpk.html">Pyrazinamide pbpk (Ramachandran 2023)</a> </td> <td style="text-align:left;"> PBPK (whole-body, 18 ODEs, MATLAB ode15s). Oral pyrazinamide disposition at extrapulmonary tuberculosis (EPTB) sites in a 70-kg reference adult male (Ramachandran and Gadgil 2023, CPT Pharmacometrics Syst Pharmacol). Seventeen perfusion-limited well-stirred tissue compartments connected by blood and lymph flow, plus a gut-lumen state for enterohepatic recycling. The model's two novel states are a pleural-fluid compartment (a filtrate of the lung that drains via lymphatics) and a consolidated lymph-node compartment that collects afferent lymph from every organ except bone and spleen and returns it to venous blood - the two most common EPTB sites. Physiological volumes, blood flows, and lymph flows are fixed literature fractions of body weight, cardiac output (5200 mL/min), and afferent lymph flow (8 L/day) respectively (Appendix S1 Tables S2 and S3). Tissue:plasma partition coefficients were computed by the Rodgers and Rowland method from the drug physicochemistry in Table S4 and are tabulated in Table S5. Only the first-order oral absorption rate ka and total systemic clearance CL were estimated, by weighted least squares against reported plasma concentrations after a 2000 mg oral dose (Appendix S1 Table S7). Clearance is split into a renal component fR * CL driven by arterial concentration and a hepatic component (1 - fR) * CL driven by the hepatic inlet concentration; the hepatic output enters the gut lumen, which for pyrazinamide is a terminal sink cleared only by faecal transit at kF = 0.252 /h (the paper assigns a non-zero gut reabsorption rate only to rifampicin). The model is a deterministic typical-value simulation: the paper reports no interindividual variability and no residual error model, so propSd is fixed at 0. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Karunajeewa_2009_sulfadoxinePyrimethamine.html">Pyrimethamine (Karunajeewa 2009)</a> </td> <td style="text-align:left;"> Population PK model for pyrimethamine (PYR) in 60 Papua New Guinean women (30 pregnant, second or third trimester; 30 age-matched nonpregnant controls) given a single oral 1,500 mg sulfadoxine / 75 mg pyrimethamine dose for intermittent presumptive treatment of malaria in pregnancy (Karunajeewa 2009). Two-compartment disposition with first-order absorption and no lag, fit as a separate NONMEM dataset from the parent SDOX/NASDOX dataset. Allometric scaling at reference WT = 70 kg is applied to all apparent volumes (exponent 1) and all apparent clearances (exponent 0.75). Pregnancy is the only retained covariate; it enters as additive terms on apparent CL/F (+0.439 L/h/70 kg), Vc/F (+76 L/70 kg) and Vp/F (+98 L/70 kg). Between-subject variability on CL/F, Vc/F and Vp/F is correlated (3x3 block, correlations 0.797 / 0.756 / 0.731 from Table 4); BSV on Q/F and ka is independent. The companion model for the co-administered sulfadoxine plus its NASDOX metabolite is shipped as 'Karunajeewa_2009_sulfadoxine' (separate NONMEM dataset, fit independently in the source publication). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ayyoub_2016_pyronaridine.html">Pyronaridine (Ayyoub 2016)</a> </td> <td style="text-align:left;"> Pooled population PK model of oral pyronaridine in 349 pediatric malaria patients (0.51-15 years, 6.8-56.2 kg) from one phase II and five phase III studies of the pyronaridine-artesunate fixed-dose combination (Pyramax). Two-compartment disposition with first-order absorption and first-order elimination from the central compartment. Body weight enters as fixed allometric scaling (exponent 0.75 on CL/F and Q/F, 1.00 on V2/F and V3/F, centred on a 20 kg reference). Age enters as a power covariate on the peripheral volume V3/F (exponent 0.624, centred on a 7 yr reference). Formulation (1 = pediatric granule sachet, 0 = tablet) increases the absorption rate Ka by 1.63-fold over the tablet baseline. Residual error is additive on the natural-log concentration scale (equivalent to proportional in linear space). Dose is encoded as pyronaridine base in mg (paper Methods: pyronaridine tetraphosphate doses are multiplied by 0.57 prior to modeling). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kang_2023_pyronaridine_artesunate_hamster_pbpk.html">Pyronaridine hamster pbpk (Kang 2023)</a> </td> <td style="text-align:left;"> Preclinical (golden hamster). PBPK (minimal, five-compartment; WinNonlin 8.3). Oral pyronaridine disposition with the antiviral target tissues -- lung and trachea -- carried as explicit perfusion-rate-limited organ compartments and all nontarget tissues lumped into a rest-of-body compartment (Kang et al. 2023, Pharmaceutics 15:838). Whole blood is the sampling matrix because pyronaridine partitions strongly into erythrocytes (reported blood-to-plasma ratio 4.9-17.8), so no blood-to-plasma conversion is applied. Tissue uptake follows the well-stirred perfusion-limited form Q_tissue * (C_blood - C_tissue / K_tissue), and a first-order bidirectional exchange (k_tl, k_lt) links lung and trachea directly. Physiological volumes and blood flows (Table 1) are fixed; the seven biochemical parameters (Table 2) were fitted to 343 pooled pyronaridine measurements by naive pooled-data analysis, so the model carries no between-subject variability and is intended for typical-value simulation of blood, lung and trachea profiles after single or daily oral dosing. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhu_2024_pyrotinib.html">Pyrotinib (Zhu 2024)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption and elimination for oral pyrotinib in Chinese patients with HER2-positive breast cancer, with a serum total protein effect on apparent clearance (Zhu 2024) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dias_2024_quetiapine_rat.html">Quetiapine rat (Dias 2024)</a> </td> <td style="text-align:left;"> Preclinical (rat, Wistar). Semimechanistic pharmacodynamic (PD-only) precursor-pool model for medial prefrontal cortex (mPFC) extracellular dopamine after a single 5 mg/kg intravenous dose of quetiapine given either as a solution (FQ) or as lipid core nanocapsules (QLNC), in naive rats and in poly(i:c) schizophrenia phenotyped rats (SPR). Dopamine precursors are synthesised into a pool at a zero-order rate and released into the extracellular dopamine compartment by a first-order rate constant that quetiapine stimulates linearly; dopamine is removed by a reuptake process that a negative-feedback modulator exacerbates, giving the fast return to baseline. Quetiapine acts through a Sheiner effect compartment driven by the unbound brain concentration, and co-administered nanoparticles competitively blunt the drug effect while decaying first-order. Disease severity enters as the continuous prepulse-inhibition score on both the dopamine baseline and the drug-effect slope. This model has NO PK layer: the unbound quetiapine brain concentration must be supplied by the user as the time-varying covariate CU_QTP_BRAIN, generated from the companion semimechanistic popPK model (Carreno et al. 2020, doi:10.1124/jpet.120.000109). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fattinger_1991_quinidine.html">Quinidine (Fattinger 1991)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral quinidine in adults treated for supraventricular or ventricular arrhythmias (Fattinger 1991). Zero-order absorption from the gastrointestinal tract with formulation-specific absorption duration: immediate-release quinidine sulphate (Chinidin sulfuricum) with a typical absorption duration of 1.37 h, and slow-release quinidine bisulphate (Kinidin duriles) with a typical absorption duration of 6.0 h and a 1.36-fold higher relative bioavailability versus quinidine sulphate. Apparent total clearance is the sum of a renal arm proportional to creatinine clearance (proportionality 0.0566 L/h per mL/min) and a non-renal arm of 12.6 L/h that is halved to 6.8 L/h in patients with severe heart failure or severe liver failure. Apparent central volume is 161 L. Inter-compartmental clearance Q is 12.6 L/h and peripheral volume V2 is 66.7 L. Inter-individual variability is assigned to total clearance (40.2% CV), central volume (75.6% CV), and the quinidine sulphate absorption duration (49.4% CV); residual variability is proportional (22% CV). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Westerhout_2013_quinidine.html">Quinidine (Westerhout 2013)</a> </td> <td style="text-align:left;"> Preclinical (rat, male Wistar WU). Systems-based pharmacokinetic (SBPK) model for quinidine intra-brain distribution following IV infusion, fit jointly to unbound plasma, brain parenchymal extracellular fluid (brain ECF), CSF in the lateral ventricle (CSF_LV) and cisterna magna (CSF_CM), and end-of-experiment total (deep) brain concentrations, with simultaneous mechanistic state compartments for CSF in the combined third + fourth ventricles (CSF_TFV) and the subarachnoid space (CSF_SAS) carrying the ventricular CSF flow from LV through SAS back to systemic plasma at the fixed physiological rate Q_CSF = 2.2 uL/min, plus a brain-ECF-to-CSF_LV flow at Q_ECF = 0.2 uL/min (Westerhout 2013, J Pharmacokinet Pharmacodyn). Two systemic peripheral compartments (V_PER1, V_PER2) with inter-compartmental clearances Q_PL-PER1, Q_PL-PER2 carry the multi-exponential plasma decline. P-glycoprotein (P-gp) activity (binary indicator CONMED_TARIQUIDAR = 0 control / 1 tariquidar-inhibited) modulates the systemic elimination CL_E (1.9-fold increase when P-gp is active per Table 4) and every transfer clearance between plasma and the brain compartments: passive influx into each brain compartment is reduced when P-gp is active (influx hindrance, subtractive) and passive efflux is increased (efflux enhancement, additive). At the BCSFB level, P-gp acts as an efflux transporter at the LV (CL_LV-PL,P-gp estimated to 0 in the combined model, with the P-gp-mediated component carried by CL_PL-LV,P-gp) and is absent at the CM (both P-gp components fixed to 0 in the combined model). The plasma-to-TFV transfer clearance is structurally assumed equal to plasma-to-LV (no TFV microdialysis sampling). Parameter values are the paper's preferred 'efflux enhancement + influx hindrance' (combined) SBPK model from Table 4, column 3 (OFV = 17,969); the two alternative P-gp mechanism variants (efflux-enhancement-only, OFV = 18,105; influx-hindrance-only, OFV = 18,030) and the simpler preliminary compartmental model (Table 3) are discussed in the validation vignette but not extracted as separate model files. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Saleh_2023_leicns_pk30_mouse_brain.html">Quinidine mouse pbpk (Saleh 2023)</a> </td> <td style="text-align:left;"> PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, NMRI). Nine-compartment CNS PBPK model predicting unbound quinidine concentrations in brain extracellular fluid (brain ECF) after a single 40 mg/kg intraperitoneal dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug's physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.2185 (taken from ref [56]; Table V). The plasma PK model is the empirical one-compartment model of Table IV (in-house NONMEM fit) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shin_2006_quinidine_QT.html">Quinidine QT (Shin 2006)</a> </td> <td style="text-align:left;"> Population pharmacodynamic Emax model for quinidine-induced QTc prolongation in 24 healthy Korean (12 M / 12 F) and 13 healthy Caucasian (7 M / 6 F) adults following a single 20 min IV infusion of quinidine gluconate 4 mg/kg (base). The Emax form is QTc(t) = E0 + DeltaEmax * Cc / (EC50 + Cc) with E0 modulated by sex (additive +34 ms in females; reference category = male) and DeltaEmax modulated by ethnicity (multiplicative x1.26 in Caucasians; reference category = Korean) plus an additive +106 ms interaction in Caucasian females only. EC50 = 3.13 uM (= 1.0155 mg/L using quinidine MW 324.42 g/mol). Source publication does not fit a popPK model; the PK driver in this file is a typical-value 1-compartment IV approximation with CL = 0.3 L/h/kg and Vc = Vss = 2.5 L/kg derived from the pooled NCA summary statistics in Shin 2006 Table 2 (see vignette Errata). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kervezee_2014_quinidine_rat.html">Quinidine rat (Kervezee 2014)</a> </td> <td style="text-align:left;"> PBPK (semi-mechanistic brain-distribution, unbound-only). Preclinical (rat). Nine-compartment brain-distribution model for i.v. quinidine in male Wistar rats (Kervezee 2014), fitted to unbound plasma, brain extracellular fluid (ECF), cerebrospinal fluid at the lateral ventricle (CSF-LV) and cisterna magna (CSF-CM), and total deep-brain tissue. Topology from Westerhout 2013: plasma (central), two peripheral tissue compartments (peripheral1, peripheral2), one deep-brain intracellular compartment (brain_deep), one brain ECF compartment (brain_ecf), and four sequential CSF sub-compartments (csf_lv, csf_tfv, csf_cm, csf_sas) draining back into plasma via CSF bulk flow. P-glycoprotein-mediated transport enters as an additional clearance component at the BBB (subtracted from PL-to-brain influx, added to brain-to-PL efflux) and on plasma elimination. Kervezee 2014 introduces the diurnal-period covariate PERIOD_ACTIVE (0 = resting / lights-on, 1 = active / lights- off) that acts on five parameters: the P-gp components of CL_DBR-PL, CL_PL-ECF, CL_ECF-PL, and CL_PL-LV, plus CSF bulk flow Q_CSF. Brain compartment volumes and Q_ECF are fixed to physiological values from Westerhout 2013 refs 38-47; plasma volume V_PL is fixed to the rat plasma volume; peripheral volumes are estimated. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Syvanen_2012_quinidine_rat.html">Quinidine rat (Syvanen 2012)</a> </td> <td style="text-align:left;"> Preclinical (rat, male Sprague-Dawley). Two-compartment plasma + brain extracellular fluid (ECF) population PK model for quinidine in rats with hippocampal microdialysis sampling, fit jointly to plasma, brain ECF, and end-of-experiment total brain concentrations (Syvanen 2012). The plasma 2-cmt system (central V1 / peripheral V2) couples to a brain ECF compartment V_Br via asymmetric BBB clearances (Q_in = f1 * Q_out into brain, Q_out out of brain, Q_out FIXED at 10.8 mL/min from the paper's bootstrap-stability analysis). A third observed output, total brain tissue concentration Cbrain_deep, is modelled as an algebraic equilibrium multiple of brain ECF (Cbrain_deep = f2 * Cbrain_csf) because the paper could not estimate separate rate constants for the deep brain compartment. Two binary covariates: CONMED_TARIQUIDAR (15 mg/kg IP tariquidar pre-administered 30 min before quinidine, a selective P-glycoprotein inhibitor) modifies CL, Q_out, Q_in, and f2; DIS_POSTSE_KAINATE (1 week post-kainate-induced status epilepticus, rat temporal-lobe-epilepsy paradigm) modifies CL, V2, and V_Br. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kloprogge_2014_quinine.html">Quinine (Kloprogge 2014)</a> </td> <td style="text-align:left;"> Population PK model for oral quinine in pregnant women with uncomplicated Plasmodium falciparum malaria in Uganda (Kloprogge 2014). First-order absorption into a two-compartment disposition model with allometric body-weight scaling on clearance and intercompartmental clearance (power 2/3) and on apparent volumes (power 1), centered at the cohort typical weight of 56 kg. Relative bioavailability F is fixed at 1 with log-normal IIV; a linear covariate effect of time-varying parasitaemia (per log10 parasites/uL, last-observation-carried-forward) increases F by 38.9% per log10 parasitaemia, and an exponential effect of admission body temperature decreases elimination clearance by ~21.6% per degC (centered at the cohort median 37.2 degC). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LeJouan_2005_quinine.html">Quinine (LeJouan 2005)</a> </td> <td style="text-align:left;"> Population PK model for oral quinine in Cameroonian children (aged 0.55-6.7 years) with uncomplicated Plasmodium falciparum malaria (Le Jouan 2005). One-compartment with first-order absorption, time-varying free fraction fu = 0.15 + 0.001*(t - 36) anchored at the literature value fu(t=36 h) = 0.15 (Babalola 1989) and clamped to its t = 72 h value beyond the studied window, and linear-in-body-weight apparent clearance CL/F = fu * 0.53 * WT and apparent volume V/F = fu * (57 + 3.8 * WT). Doses are oral quinine base in mg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sheng_2016_quinine_rat.html">Quinine rat (Sheng 2016)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two generalized Poisson (2GP) mixture PD model for bimodal lick-count data from rodent brief-access taste aversion (BATA) experiments with quinine hydrochloride dihydrate; the drug effect enters via a sigmoid emax on a logistic-transformed mixing probability between a low-count and a right-truncated high-count generalized-Poisson distribution. The fitted compound is quinine HCl dihydrate used as a model bitter stimulus. STIM_QUININE_MM is the applied sipper-tube concentration (mM); there is no PK ODE and no time evolution (each record is an 8-second presentation). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Aruldhas_2021_methadone.html">R methadone (Aruldhas 2021)</a> </td> <td style="text-align:left;"> Population pharmacokinetic model for R-methadone and its R-EDDP metabolite in post-operative paediatric surgical patients (Aruldhas 2021; 61 children aged 11-17 years given 0.1 mg/kg IV racemic methadone intra-operatively followed by 0.1 mg/kg oral racemic methadone every 12 h as either a tablet or an oral suspension). R-methadone disposition is a two-compartment model with first-order absorption (separate tablet and suspension Ka values), an estimated absolute bioavailability F common to both oral formulations, and a one-compartment R-EDDP metabolite whose central volume is set equal to the R-methadone central volume (scaling factor VF fixed to 1 due to metabolite unidentifiability). Total R-methadone clearance CL scales allometrically with body weight (fixed exponent 0.75, reference 70 kg). The fractional R-methadone clearance to R-EDDP (CLF) is a linear function of the CYP2B6 activity score and the number of active alleles of the intronic CYP3A4 SNP rs2246709. The R-methadone central volume V2 is a linear function of the plasma alpha-1 acid glycoprotein (AAG) concentration and the number of active alleles of the ORM1 SNP rs17650. Between-subject variability is estimated on CL, V2, peripheral volume, CLF, and R-EDDP clearance; residual error is proportional for both R-methadone and R-EDDP. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Darpo_2014_racSotalol.html">RacSotalol QTcF (Darpo 2014)</a> </td> <td style="text-align:left;"> Concentration-QTcF linear mixed-effects PD model for oral rac-sotalol in 39 healthy adults (28 men, 11 women) following a single 160 mg oral dose on day 1, with day-0 ECGs serving as the time-matched baseline. The endpoint is change from baseline in the Fridericia- corrected QT interval (DeltaQTcF = QT / RR^(1/3); ms). The full model is: DeltaQTcF = e0 + e_sexf_e0 * SEXF + (slope + e_sexf_slope * SEXF) * CP_RACSOTALOL_UGML + e_qtc_bl_e0 * (QTC_BL - 390), with male and median-baseline (QTC_BL = 390 ms) reference. Reference parameter values (male, median-baseline subject): e0 = -2.5 ms, slope = 24 ms per ug/mL, female increments +10.6 ms on the intercept and +7 ms per ug/mL on the slope, centred-baseline-QTcF coefficient -0.70 ms/ms. PD-only model: rac-sotalol plasma concentration is supplied as a time-varying covariate CP_RACSOTALOL_UGML (ug/mL). The source publication does not fit a population PK model -- the pharmacokinetic profile was characterised by NCA only (Cmax 1.4 ug/mL men, 1.8 ug/mL women; AUC_inf 14.6 men, 17.1 women); users wishing to drive the PD model from a simulated PK source must supply their own concentration trajectory (no rac-sotalol popPK model exists in the nlmixr2lib registry). Companion model file Darpo_2014_racSotalol_QTcI.R reports the same structure for the individually-corrected QT interval. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Darpo_2014_racSotalol.html">RacSotalol QTcI (Darpo 2014)</a> </td> <td style="text-align:left;"> Concentration-QTcI linear mixed-effects PD model for oral rac-sotalol in 39 healthy adults (28 men, 11 women) following a single 160 mg oral dose on day 1, with day-0 ECGs serving as the time-matched baseline. The endpoint is change from baseline in the individually- corrected QT interval (DeltaQTcI, ms). The full model is: DeltaQTcI = e0 + e_sexf_e0 * SEXF + (slope + e_sexf_slope * SEXF) * CP_RACSOTALOL_UGML + e_qtc_bl_e0 * (QTC_BL - 390), with male and median-baseline (QTC_BL = 390 ms) reference. Reference parameter values (male, median-baseline subject): e0 = -3.2 ms, slope = 23 ms per ug/mL, female increments +11.1 ms on the intercept and +7 ms per ug/mL on the slope, centred-baseline-QTcI coefficient -0.70 ms/ms. PD-only model: rac-sotalol plasma concentration is supplied as a time-varying covariate CP_RACSOTALOL_UGML (ug/mL). The source publication does not fit a population PK model -- the pharmacokinetic profile was characterised by NCA only (Cmax 1.4 ug/mL men, 1.8 ug/mL women; AUC_inf 14.6 men, 17.1 women); users wishing to drive the PD model from a simulated PK source must supply their own concentration trajectory (no rac-sotalol popPK model exists in the nlmixr2lib registry). Companion model file Darpo_2014_racSotalol_QTcF.R reports the same structure for the Fridericia-corrected QT interval. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ai_2024_ractopamine_goat_pbpk.html">Ractopamine goat pbpk (Ai 2024)</a> </td> <td style="text-align:left;"> PBPK (whole-body, hybrid flow- and membrane-limited; goat). Eleven-tissue physiologically based pharmacokinetic model for the beta-adrenoceptor agonist ractopamine in Liaoning cashmere goats after repeated oral gavage, comprising gastric contents, intestinal contents, liver, spleen, kidney, heart, lung, muscle, fat, brain, a lumped rest-of-body compartment, arterial plasma, venous plasma and a urinary-excretion sink. Liver, spleen, kidney, heart and lung are perfusion (flow) limited; muscle, fat, brain and the rest of the body are membrane (permeability) limited and each carry a vascular plasma sub-compartment plus a tissue sub-compartment separated by a permeability-area product. Absorption is first-order gastric emptying (Kst) into the gut lumen followed by first-order uptake (Ka) into the liver in competition with first-order fecal loss of unabsorbed drug (Kgut); elimination is hepatic (Clhe) plus renal (Clre), both acting on the unbound fraction. Built to predict edible-tissue residues and withdrawal times against Codex Alimentarius maximum residue limits (Ai 2024). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cardilin_2018_radiation_radiosensitizer_mouse.html">Radiation radiosensitizer mouse (Cardilin 2018)</a> </td> <td style="text-align:left;"> Preclinical (mouse, FaDu head-and-neck xenograft). Tumor growth inhibition model for combination therapy with ionizing radiation and a radiosensitizer (linear-quadratic radiation kill with a damage-compartment transit chain, driven by a one-compartment radiosensitizer PK). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/ArabAlameddine_2012_raltegravir.html">Raltegravir (ArabAlameddine 2012)</a> </td> <td style="text-align:left;"> Two-compartment first-order-absorption population PK model for oral raltegravir (RAL) in 145 HIV-positive adults and 19 healthy volunteers, with two HIV-status-specific absorption rate constants (ka HIV+ slower than HIV-), HIV-status-specific proportional residual error, a fixed reference bioavailability F=1 for healthy volunteers, and an estimated relative bioavailability for HIV+ subjects modified linearly by sex (female +55%), atazanavir coadministration (+39%), and total bilirubin centered at 30 umol/L (+36% per doubling), plus a -59% race effect on the central volume of distribution for Caucasian relative to non-Caucasian subjects (Arab-Alameddine 2012). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bukkems_2021_raltegravir.html">Raltegravir (Bukkems 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral raltegravir in a pooled cohort of 221 adults (healthy volunteers, non-pregnant adults living with HIV, and pregnant women living with HIV) across the 400 mg BID, 800 mg QD, and 1200 mg QD (two 600 mg tablets) regimens (Bukkems 2021). Absorption is modelled as a chain of four sequential first-order compartments: depot -> transit1 -> transit2 -> transit3 -> central, with the first three transitions governed by a shared rate constant ktr = 3 / mat (paper's mean transit time MAT parameter) and the final transit3 -> central transition governed by a separate first-order absorption rate constant ka. Disposition is a two-compartment linear model with apparent central and peripheral volumes and apparent clearance and inter-compartmental clearance. Body weight enters as allometric scaling with fixed exponents 0.75 on CL and Q and 1.0 on Vc and Vp, referenced to 70 kg. Six covariate-parameter relationships are retained in the final model: FED on MAT (+160% with any food), CONMED_ATAZANAVIR on CL (-17%), FORM_RAL_600 on F (+21%), FED_LOWFAT on F (-46%), PREG on F (-49%), and CONMED_EFV on F (-17%). Inter-individual variability is a diagonal eta on Vc plus a correlated 3-parameter block on CL, Q, and Vp with the CL-Vp off-diagonal fixed to zero. Residual error is a proportional model with a time-varying magnitude that switches at 3 h after dose (43.5% CV before, 29.0% CV after, tracking the paper's empirical time-varying residual). The paper's inter-occasion variability on F and MAT and its additional eta on the residual magnitude are omitted from this packaged model; see the validation vignette Assumptions and deviations section. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2016_raltegravir.html">Raltegravir (Lee 2016)</a> </td> <td style="text-align:left;"> Population PK model for oral raltegravir (a UGT1A1 phenotyping probe) and its glucuronide metabolite in 24 East Asian patients with advanced solid tumours receiving FOLFIRI chemotherapy (Lee 2016). Raltegravir absorption is described with a depot, a single transit compartment (the paper estimates a non-integer NN = 1.07 in the Savic 2007 transit-chain framework; the packaged model approximates this with one explicit transit compartment), and a one-compartment central compartment with first-order elimination (CL/F, V/F). Raltegravir glucuronide is described by a one-compartment metabolite compartment (central_gluc) with V_GLU fixed at 1 L (a structural identifiability anchor) and a first-order metabolite clearance CL_GLU. The formation rate constant kmet maps to the source paper's FMET, which the authors define as the formation rate of glucuronide divided by V_GLU; with V_GLU fixed at 1 L, kmet has units 1/h and drives dA_gluc / dt = kmet * V_GLU * C_RAL_central - CL_GLU * C_gluc. Bioavailability F is fixed at 1 (single oral dose; absolute F not identifiable). IIV is reported on CL/F, MTT, F, V/F, kmet (FMET), and CL_GLU with a single off-diagonal covariance between CL/F and V/F (correlation 0.567). The residual error was reported as additive on log-transformed observations for both raltegravir and glucuronide, which maps to a proportional residual on the linear-concentration scale. No baseline covariates (age, sex, weight, body surface area, serum albumin / creatinine / bilirubin / liver enzymes, ethnicity, or UGT1A1 * 6 / * 28 / * 60 and CYP3A5 * 3 genotypes) were retained in the final model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lommerse_2019_raltegravir.html">Raltegravir (Lommerse 2019)</a> </td> <td style="text-align:left;"> Integrated maternal-neonatal two-compartment first-order-absorption population PK model of oral raltegravir, coupling the maternal and fetal central compartments via a very fast intercompartmental clearance (1000 L/h) during pregnancy to enforce instantaneous placental equilibrium, and decoupling at birth (time t=0 in the model). Neonatal apparent clearance rises from nil at birth to CL_max (9.44 L/h at 25 kg) with a first-order maturation rate constant CL_tau (11.3 1/year, 90% mature by ~11 weeks); neonatal absorption rate constant rises from KA_base (0.0915 1/h) to KA_max (0.43 1/h) with a first-order maturation rate constant KA_tau (63.2 1/year, 90% mature by ~12 days). Neonate CL, Q, and volumes are allometrically scaled with fixed exponents 0.75 and 1.0 to a reference weight of 25 kg. Maternal disposition parameters (V2 3.52 L, V3 27 L, CL 9.73 L/h, Q 0.866 L/h; all at 25 kg reference) are fixed from the Rizk 2015 pediatric popPK (ref [12]); maternal KA (0.175 1/h) and bioavailability (F 0.517) are estimated. IIV is on neonate CL and KA and maternal F; residual error is combined additive (11.9 nM) and proportional (54%) (Lommerse 2019). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2011_raltegravir.html">Raltegravir (Wang 2011)</a> </td> <td style="text-align:left;"> Population PK model for plasma and intracellular (PBMC) raltegravir after a single 400 mg oral dose in six healthy male Singaporean volunteers (Wang 2011). Plasma PK is described by a one-compartment model with first-order elimination preceded by a chain of two transit compartments between depot and central (Kappelhoff 2005 transit-absorption parameterisation: MAT = (n + 1) / ktr with n = 2 transit compartments). Bioavailability F is implicit in the apparent CL/F and V/F. Intracellular (PBMC) raltegravir is described as an empirical partition of the predicted plasma concentration via the paper's accumulation ratio ACR (point estimate 11.2%) with its own inter-individual variability and exponential residual error (Wang 2011 Eq.: C_IC,obs = ACR * C_plasma,pred * exp(eps_IC)). The packaged model maps ACR to the canonical paper-named bare parameter `frac` and its log-transformed primary `lfrac`. No baseline covariates were retained in the final model. Note that the paper's term 'accumulation ratio' is a misnomer in the conventional sense -- ACR < 1 means raltegravir does NOT accumulate intracellularly, consistent with simple diffusion of unbound drug into PBMCs (the authors' Conclusions). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Blair_2004_raltitrexed.html">Raltitrexed (Blair 2004)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for intravenous raltitrexed (Tomudex) in adult patients with advanced solid tumours, with linear-additive covariate effects of Cockcroft-Gault creatinine clearance on CL and of body weight and serum albumin on central volume (Blair 2004) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/HuttonSmith_2018_ranibizumab.html">Ranibizumab (HuttonSmith 2018)</a> </td> <td style="text-align:left;"> QSP. Three-compartment (retina, vitreous, aqueous) ocular PK/PD model of intravitreal ranibizumab and free VEGF suppression in wet AMD, with bivalent VEGF-ranibizumab binding (V, R, VR, RVR species per compartment; 12 ODEs). Retina-vitreous transfer uses hydrodynamic-radius-dependent ILM and RPE permeabilities (Hutton-Smith 2017 rabbit-PK power laws); vitreous-aqueous elimination k_el is derived from the ranibizumab ocular half-life via the mass-balance of Section S2. Parameters K_D = 19000 pM and log-normal population distributions of ocular half-life and VEGF production rate were re-estimated on 31 wet-AMD patients from Saunders 2015. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mulyukov_2018_ranibizumab.html">Ranibizumab (Mulyukov 2018)</a> </td> <td style="text-align:left;"> Indirect-response PK/PD model of intravitreal ranibizumab on best-corrected visual acuity (SCORE_BCVA, ETDRS letters) in anti-VEGF-naive adults with neovascular age-related macular degeneration (Mulyukov 2018). SCORE_BCVA is driven by an indirect-response ODE in which drug concentration stimulates the SCORE_BCVA production rate (kin) through a Michaelis-Menten-like term with a time-dependent maximum effect emax(t) = emax_ss + demax_0 * exp(-kemax * t). The PK is a fixed first-order vitreous-elimination placeholder (kel = 0.077/day, vitreous volume = 4 mL, no IIV) borrowed from a previous population PK analysis (reference 20 of the paper) because vitreous PK data were not collected in the development studies. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Xu_2013_ranibizumab.html">Ranibizumab (Xu 2013)</a> </td> <td style="text-align:left;"> One-compartment population PK model for total ranibizumab in serum following intravitreal injection in adults with neovascular age-related macular degeneration (Xu et al. 2013, IOVS). Vitreous humor acts as a slow-release depot with first-order absorption Ka into a systemic central compartment with first-order clearance CL/F; a small parallel fraction of each intravitreal dose reaches the central compartment via a rapid needle-track shunt bypassing the vitreous. Covariates: Cockcroft-Gault creatinine clearance on CL/F (power) and concomitant verteporfin PDT on Ka (multiplicative). Serum ranibizumab was measured; vitreous concentration is computed algebraically from vitreous amount over an assumed 4 mL vitreous humor volume for downstream simulation and is not observed. Data pooled from two Phase 1, two Phase 1/2, and one Phase 3 trial (MARINA / FOCUS). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hawwa_2013_ranitidine.html">Ranitidine (Hawwa 2013)</a> </td> <td style="text-align:left;"> One-compartment population PK model for ranitidine in critically ill children (n = 78, age 15 days to 15.5 years, weight 1.3 to 47 kg) receiving oral and/or intravenous bolus doses for stress-ulcer or GORD prophylaxis. First-order absorption with allometric scaling of clearance (fixed exponent 0.75) and central volume (fixed exponent 1.0) to a 70 kg adult. Cardiac failure or cardiac surgery (pooled binary indicator) multiplicatively reduces clearance by 53.7%. IIVs on absorption rate constant and bioavailability were dropped during model building so the model could minimize; only CL and V carry IIV. Proportional residual error (Hawwa 2013). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ivaturi_2017_RBP_7000.html">RBP 7000 (Ivaturi 2017)</a> </td> <td style="text-align:left;"> Integrated population pharmacokinetic / PANSS pharmacodynamic model for the once-monthly long-acting subcutaneous Atrigel formulation of risperidone (RBP-7000, Indivior) in 337 adults with acute schizophrenia treated with two SC injections (90 mg or 120 mg) 28 days apart in a Phase 3 registration trial (NCT02109562). The PK sub-model is the empirical dual-absorption structure inherited from the upstream RBP-7000 SAD and MAD studies (Gomeni 2013; Laffont 2014, 2015): a fast first-order absorption rate ka1 from the SC depot to the risperidone central compartment captures the rapid release from the injection site, while a 5-compartment transit chain with rate constant ktr feeds a slow first-order absorption rate ka2 from the terminal transit compartment into central, mimicking the slow sustained release from the solidified ATRIGEL implant. Systemically available risperidone is distributed to a single peripheral compartment (rate constants krrp and krpr), eliminated by non-metabolite routes (krel), and partly converted to its equipotent 9-hydroxyrisperidone metabolite at rate kr9; the metabolite is described by a one-compartment model with first-order elimination (k9el) and an apparent volume of distribution constrained equal to the central volume of the parent V because VM was not identifiable. CYP2D6 intermediate and poor metabolizers (vs the extensive / inconclusive reference) have 76 and 94 percent lower metabolite formation rate, respectively. Plasma concentrations of risperidone and 9-OH-risperidone are converted to total active moiety in risperidone-equivalent units by AM = [risperidone] + [9-OH-risperidone] * 410/426 (molecular-weight correction). The PANSS PD sub-model combines a Weibull-shaped placebo response (PMAX, TPROG, POW), an additive linear-drift term (DRIFT) that captures the improvement-then-worsening pattern observed in some individuals, and an Emax model relating total active moiety to relative PANSS decrease, with drug and placebo effects entering additively per Predicted PANSS = BSL * (1 - PMAX * (1 - exp(-(T/TPROG)^POW)) - Emax * AM / (EC50 + AM)) + DRIFT * T (T in weeks). The proportional-odds CGI-S sub-model of Table 4 is documented in the vignette but not implemented here because rxode2's additive residual / d/dt() ODE pipeline does not natively express ordinal-logistic observation likelihoods; see the vignette Assumptions and deviations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LopezArguello_2023_pbp_binding_pseudomonas.html">Relebactam qsp (LopezArguello 2023)</a> </td> <td style="text-align:left;"> QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for relebactam (diazabicyclooctane beta-lactamase inhibitor): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Choi_2011_remifentanil.html">Remifentanil (Choi 2011)</a> </td> <td style="text-align:left;"> Combined effect-and-tolerance pharmacodynamic model for the central-nervous-system effect of remifentanil on EEG-derived temporal linear mode complexity (TLMC) in healthy volunteers (Choi 2011). The PD model captures both depression of CNS activity during infusion and rebound during recovery via a sigmoid Emax driver from an effect compartment plus an opposing sigmoid term driven by a slow tolerance compartment. The TLMC observation is baseline-normalised so the readout is dimensionless with baseline E0 = 1 by construction. The 3-compartment IV remifentanil PK underneath is fixed from the upstream Kang 2007 BJCP popPK (Kang 2007 Table 2; rate-constant parameterisation; AGE and BSA additive covariate effects on the elimination rate constant k10). Choi 2011 itself did not re-estimate PK -- it read per-subject individual PK estimates from the Kang 2007 fit as input data columns. Reported model-selection metrics across the three competing PD models in Choi 2011 (combined effect-and-tolerance, feedback, sigmoid Emax) identified the combined model implemented here as the best by AIC (-6966) and positive predictive value of rebound (100%); the feedback and sigmoid Emax models did not capture rebound (Table 3). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yang_2017_remifentanil.html">Remifentanil (Yang 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for continuous intravenous remifentanil infusion in critically ill adults receiving venoarterial extracorporeal membrane oxygenation (VA-ECMO), with sex and centrifugal-pump rotational speed as covariates on clearance (Yang 2017). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2024_remimazolam.html">Remimazolam (Chen 2024)</a> </td> <td style="text-align:left;"> Joint population PK/PD model for remimazolam, its inactive metabolite CNS 7054, and the bispectral index (BIS) in healthy Chinese adult volunteers (Chen 2024). Remimazolam is described by a three-compartment model with first-order elimination; the whole of parent clearance feeds a single transit compartment that delays the appearance of CNS 7054, which is itself described by a two-compartment model. Sedation is described by an effect compartment equilibrating with remimazolam plasma concentration and driving an inhibitory sigmoid Imax model on BIS. Body weight enters every clearance and volume term by allometric scaling with fixed exponents (0.75 and 1) and a 60 kg reference weight; no other covariate was retained on either the PK or the PD. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Luo_2024_CES1_cirrhosis.html">Remimazolam pbpk (Luo 2024)</a> </td> <td style="text-align:left;"> PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). remimazolam disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Remimazolam is a direct-acting ultrashort-acting sedative that CES1 INACTIVATES to a carboxylic-acid metabolite, so no active metabolite is tracked. It is administered intravenously only, and Table 2 reports no CLint,K, so the kidney compartment carries distribution but no elimination. The semi-PBPK circuit is three gut-wall segments, portal vein, liver, kidney and a three-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper's virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Laffont_2025_opioid_overdose_reversal_simulation.html">Respiratory physiology (Mann 2022)</a> </td> <td style="text-align:left;"> QSP. Magosso / Ursino respiratory and cerebrovascular physiology with Mann 2022 extensions for opioid-induced ventilatory depression and cardiovascular-collapse / cardiac-arrest dynamics. Encodes the 11-state physiological submodel from the FDA delaymymod.c implementation, plus the cardiac-arrest event rule (PaO2 below 15 mm Hg sustained 220 s -> cardiac output decays toward 0.01 L/min). The CAR (fraction of opioid receptors bound by an agonist) input drives reductions in wakefulness drive (W - Wmax * CAR^P3) and chemoreflex drives (factor 1 - CAR^P1). The Spencer dissociation algebra for blood gas exchange is carried inline. The original FDA implementation uses delay- differential equations for peripheral and central chemoreflex filtering with delays of roughly K_Dp/(Qb+Qt) ~ 7 s and K_Dc/(Qb+Qt) ~ 11 s; this model deploys the limit of zero delay (Plag_X = X, Clag_X = X), which preserves the steady-state structure and longer-time-scale overdose dynamics but does not reproduce the second-scale delay artefacts of the original. Composes downstream of Mann_2022_mu_receptor_binding (CAR_OPIOID input). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Roepcke_2023_rezafungin.html">Rezafungin (Roepcke 2023)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for rezafungin after weekly IV infusion in healthy subjects, hepatically impaired subjects, and patients with candidemia and/or invasive candidiasis (Roepcke 2023), with body-surface-area scaling on CL, V1, and the shared peripheral volume V23, a serum-albumin effect on V23, and a healthy-vs-diseased disease-state shift on CL and V1. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Diao_2014_rFIXFc.html">RFIXFc (Diao 2014)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for recombinant factor IX Fc fusion protein (rFIXFc, eftrenonacog alfa) in patients with severe to moderate haemophilia B aged 12-77 years (Diao 2014). Disposition is described by linear three-compartment kinetics with intravenous input and first-order elimination from the central compartment; body weight is the only retained covariate, scaling CL and V1 with estimated power exponents (not the canonical 0.75 / 1) and a reference weight of 73 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/VargasChristensen_2019_rfviii_rat.html">Rfviii rat (VargasChristensen 2019)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment quasi-steady-state target-mediated drug disposition (QSS-TMDD) population PK model of recombinant Factor VIII (rFVIII) and its carrier protein von Willebrand factor (VWF) in hemophilia A rats (Vargas Christensen 2019). Jointly describes total rFVIII plasma concentration, total VWF plasma concentration, and the rFVIII:VWF complex luminescent oxygen channeling immunoassay (LOCI) signal in cps following single IV bolus rFVIII over a 285-fold dose range (17.5 to 5000 IU/kg). Unbound rFVIII distributes to a peripheral compartment via first-order rate constants k12 / k21 and is eliminated by linear clearance CL of unbound rFVIII. VWF follows zero-order synthesis (ksyn = kdeg * R0 at steady state) and first-order degradation of the unbound species (kdeg). The rFVIII:VWF complex is eliminated with a Michaelis-Menten-shape apparent rate constant driven by the unbound rFVIII concentration (k_e,comp = Vmax,comp * Cfree / (KM,comp + Cfree)); at Cfree = 0 the complex has no direct elimination and its effective half-life reduces to that of free VWF (ln(2)/kdeg = ~22 h; Figure 2 of the paper). The rFVIII:VWF complex plasma concentration maps to the observed LOCI signal via a 4-parametric logistic (base_cps + (max_cps - base_cps) * complex^gamma / (cps50^gamma + complex^gamma)). CL scales allometrically with body weight at fixed exponent 0.75 (reference weight 0.3 kg = the cohort median; the paper reports no explicit reference weight). No covariate on V. Gender and age were tested by the paper but not retained in the final model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dodds_2005_rfxiii_cyno.html">Rfxiii cyno (Dodds 2005)</a> </td> <td style="text-align:left;"> Preclinical (cynomolgus monkey). Three-state mechanistic population PK model for recombinant Factor XIII A2 dimer (rA2) administered IV bolus, with endogenous constant-influx production of A2 dimer and B monomer, mass-action association A2 + 2 B -> A2B2 heterotetramer following the 1A2 + 2B -> 1A2B2 stoichiometry, first-order elimination of each species, and three ELISA assay outputs (total A2 = A2 + A2B2; A2B2 tetramer; free B) with proportional + proportional + additive residual error. Parameters are weight-normalised throughout (mass / kg). Estimated in NONMEM V (Dodds 2005). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Budha_2015_rg7652.html">Rg7652 (Budha 2015)</a> </td> <td style="text-align:left;"> Population PK/PD model for RG7652 (an anti-PCSK9 monoclonal antibody) in healthy hypercholesterolemic subjects (Budha 2015): one-compartment PK with first-order SC absorption and combined linear plus Michaelis-Menten elimination from the central compartment, linked to a Type 3 indirect-response model for serum low-density lipoprotein cholesterol (LDL-C) in which RG7652 stimulates LDL-C degradation through an Emax function. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mensing_2017_3D_HCV_regimen.html">Ribavirin (Mensing 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral ribavirin in HCV genotype-1 infected adults receiving the 3D + ribavirin regimen (Mensing 2017). First-order absorption, linear elimination, combined proportional + additive residual error, IIV on CL/F and a shared IIV on Vc/F + Vp/F. The author's final model retained cirrhosis, gender, and creatinine clearance as significant covariates on CL/F (and gender on Vc/F and Vp/F), but the paper does not publish point estimates for these covariate coefficients (only graphical exposure-ratio forest plots in Figure 2); the implemented model is the structural typical-value model with covariate coefficients omitted (documented in covariatesDataExcluded). Mensing 2017 reports correlated IIV on CL/F and Vc/Vp; the correlation coefficient is not given in Table 3, so this implementation encodes the random effects as independent (a documented deviation; see vignette Errata). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mulder_2025_ribavirin.html">Ribavirin (Mulder 2025)</a> </td> <td style="text-align:left;"> Integrated population PK/PD model for oral ribavirin (RBV) in solid organ transplant (SOT) recipients with chronic hepatitis E virus (HEV) infection (Mulder 2025). PK: two-compartment model with first- order absorption; ka, Vc, Q, and Vp fixed at the Wu 2015 (HCV- patient) starting-model estimates and CL re-estimated. Covariates carried over from Wu 2015: allometric weight on Vc (exponent 1.29) and Vp (exponent 0.725) with reference weight 79 kg, and a female- factor 0.732 on Vp. New covariate estimated on CL: MDRD eGFR with a capped power effect (exponent 1.32) above an estimated threshold of 57 mL/min/1.73 m^2. Haemoglobin: indirect-response (kin/kout) on an endogenous Hb pool, with RBV producing a linear concentration- proportional acceleration of the haemoglobin loss rate (1 + slope * Cc) so that haemoglobin declines with increasing RBV exposure; kin is set per subject from the baseline-Hb covariate HGB_BL so the Hb state is at steady state pre-treatment. Viral load: target-cell- limited (Baccam / Dahari) model with three paper-specific compartments (healthy hepatocytes, infected hepatocytes, virions); RBV inhibits viral replication via an Imax/IC50 sigmoidal Emax form (Imax = 0.999, IC50 = 1000 ng/L) so production of virions from infected cells is essentially fully suppressed throughout the observed RBV concentration range; healthy hepatocyte half-life and the infected:healthy hepatocyte decay-rate ratio are fixed to the literature values (Dahari 2007) and the viral elimination rate is estimated. Initial conditions for the viral compartments are computed at baseline steady state: H(0) = 1 (arbitrary unit), I(0) = rho = 0.001 (fixed), V(0) = HEV_VLOAD per subject; the synthesis rates ksyn (healthy), beta (infection), and p (virion production) are derived inside model() from these initial conditions and the rate constants so that all three viral compartments start at baseline steady state. Hb and the three viral compartments are declared paper_specific_compartments. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Saleh_2023_leicns_pk30_mouse_brain.html">Ribociclib mouse pbpk (Saleh 2023)</a> </td> <td style="text-align:left;"> PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, CD1 nude). Nine-compartment CNS PBPK model predicting unbound ribociclib concentrations in brain extracellular fluid (brain ECF) after a single 100 mg/kg oral dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug's physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.0693 (taken from ref [9]; Table V). The plasma PK model is the empirical one-compartment model of Table IV (ref [9]) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hennig_2015_rifabutin.html">Rifabutin (Hennig 2015)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model for rifabutin with simultaneous two-compartment metabolite (25-O-desacetyl rifabutin) modelling in 44 African HIV-infected adults with pulmonary tuberculosis on 300 mg daily oral rifabutin (Hennig 2015). Body weight allometrically scaled (a priori; CL exponent 0.75, V exponent 1) on all rifabutin apparent clearances and apparent volumes; sex effect on rifabutin V/F (males 1.84-fold higher than females); SLCO1B1 rs11045819 heterozygous-AC genotype increases rifabutin bioavailability F by 30.4 percent relative to homozygous-CC reference. Des-rifabutin parameters are apparent (with respect to rifabutin F and metabolite-formation fraction) and were estimated without allometric scaling, with metabolite Q and peripheral V fixed. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Barnett_2018_rifampicin.html">Rifampicin (Barnett 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model with Wilkins/Savic transit-compartment absorption for a single 600 mg oral dose of rifampicin in healthy adult males (Barnett 2018), refit from the Wilkins 2008 structural form. The rifampicin model is one of three popPK models developed jointly in Barnett 2018 to support OATP1B drug-drug-interaction modeling with coproporphyrin I and rosuvastatin; the rifampicin compartmental output is the time-varying CRIF input that drives the competitive OATP1B inhibition term in the sibling coproporphyrin I and rosuvastatin models. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chigutsa_2011_rifampicin.html">Rifampicin (Chigutsa 2011)</a> </td> <td style="text-align:left;"> Population pharmacokinetic model for oral rifampicin in adults with sputum-positive pulmonary tuberculosis in South Africa (Cape Town). One-compartment disposition with a fixed-length Erlang transit-absorption chain (NN = 19 fixed) feeding the central compartment via first-order ka. Allometric scaling of CL/F and V/F to a 70 kg reference body weight with canonical Anderson and Holford (2008) exponents (0.75 on CL, 1.0 on V; cited as Chigutsa 2011 Methods reference 3 for the allometric model). Covariate effects: female sex on V/F (-30%) and on the mean transit time MTT (+30% per Results body text page 4124 -- women have a 30% LONGER absorption delay than men; Table 2 Final-model row prints -30% with a CI bit-identical to the V/F row immediately above, which is the canonical signature of a typesetting row-duplication error; per the operator sidecar request-001 directive the body text +30% is the source of truth); high-dose-band effect on MTT (-27% for daily doses >= 600 mg vs the 450 mg reference); SLCO1B1 rs4149032 genotype-dependent oral bioavailability F (heterozygous carriers -18%; homozygous variant carriers -28%; relative to the homozygous-common-allele wild-type reference). Between-subject variability (BSV) is carried on F, CL, and MTT with the CL-MTT correlation block 0.86 from Table 2; within-subject (WSV / IOV) variability reported in Table 2 is NOT carried (forward-simulation users do not need the second-occasion IOV layer; see vignette Errata). Combined additive + proportional residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Clewe_2015_rifampicin.html">Rifampicin (Clewe 2015)</a> </td> <td style="text-align:left;"> Pharmacometric pulmonary distribution model for rifampicin in adults without tuberculosis: a one-compartment plasma PK model with single-transit oral absorption coupled to a Smythe 2012 enzyme-pool autoinduction structure (MTT, N, EMAX, EC50, kENZ all fixed from the upstream Smythe 2012 model) plus two effect compartments capturing distribution from plasma to epithelial lining fluid (ELF) and alveolar cells (AC); CL/F and Vc/F are FFM-allometrically scaled to 70 kg, the ELF and AC equilibration rate constants kELF and kAC are fixed to an equivalent 1-min half-life (instantaneous distribution at the single 4-h post-dose BAL sampling time), and only the unbound steady-state ELF/plasma and AC/plasma concentration ratios are estimated (1.28 and 5.5 after correction for the 20% rifampicin plasma free fraction). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Horita_2018_rifampicin.html">Rifampicin (Horita 2018)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with sequential zero-order then first-order absorption and first-order elimination for oral rifampin (rifampicin) in Ghanaian children with active tuberculosis (Horita 2018); allometric weight scaling on CL/F (fixed 0.75) and V/F (fixed 1.0) normalised to the cohort median 14.3 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marsot_2017_rifampicin.html">Rifampicin (Marsot 2017)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model for oral rifampicin in adult patients with staphylococcal osteoarticular infections (Marsot 2017; 62 patients, 103 steady-state plasma concentrations from routine therapeutic drug monitoring at 300 mg three times daily). Absorption uses a single-transit compartment chain (depot -> transit1 -> central) with a fixed first-order absorption rate constant ka = 1.15 1/h; first-order elimination from the central compartment. Coadministration of oral fusidic acid (500 mg three times daily in 16 % of the cohort) was retained as the sole significant covariate and reduces both apparent oral clearance and apparent central volume of distribution: typical CL/F is 5.1 L/h (with fusidic acid) vs 13.7 L/h (without) and typical V/F is 23.8 L (with) vs 61.1 L (without), an interaction attributed to CYP3A4 inhibition and altered plasma protein binding by fusidic acid. Inter-individual variability is reported on CL/F (72.9 % CV) and V/F (59.1 % CV); residual variability is additive with standard deviation 2.256 mg/L. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schipani_2016_rifampicin.html">Rifampicin (Schipani 2016)</a> </td> <td style="text-align:left;"> Simultaneous population pharmacokinetic model for oral rifampicin in a mixed Malawian cohort of adults (n=115) and children (n=50) with tuberculosis. One-compartment disposition with first-order absorption (depot to central). Allometric scaling of CL/F and V/F to a 70 kg reference body weight with canonical Anderson and Holford (2008) exponents (0.75 on CL, 1.0 on V; both fixed). Estimated power-form effect of age on CL/F (exponent 0.517) centered at AGE_median. Children (defined as body weight 5-29 kg, age < 15 y per Schipani 2016 Results) carry a relative bioavailability factor F = 0.517 vs adults (F fixed at 1). Inter-individual variability is carried on CL/F (46.6% approx CV) and V/F (87.4% approx CV); ka and F have no estimated IIV. Proportional residual error 48%. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sloan_2017_rifampicin.html">Rifampicin (Sloan 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral rifampin in Malawian adults with smear-positive pulmonary tuberculosis (Sloan 2017), developed using a two-stage NONMEM workflow: stage 1 fit a one-compartment + Savic 2007 transit-compartment absorption chain (NN, MTT, Ka) to 47 intensively-sampled patients, then stage 2 fit CL/F and V/F (plus IIVs and a multiplicative sex effect on CL) to 174 sparsely-sampled patients with absorption parameters fixed at the stage 1 estimates; F is fixed at 1, between-subject variability is on CL/F, V/F, and (fixed from stage 1) MTT, and an allometric weight model with fixed exponents 0.75 / 1.0 is referenced to 70 kg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Smythe_2012_rifampicin.html">Rifampicin (Smythe 2012)</a> </td> <td style="text-align:left;"> Semimechanistic population PK / enzyme-turnover autoinduction model for oral rifampicin in adult tuberculosis patients (Smythe 2012). One-compartment disposition with single-transit absorption (N = 1 FIX) feeds the central compartment; rifampicin plasma concentration drives a nonlinear Emax production-rate increase on a unitary-baseline enzyme pool, which in turn multiplies apparent oral clearance. CL/F and V/F are Anderson-Holford normal-fat-mass (NFM) allometrically scaled to a 70-kg patient with separate estimated Ffat contributions on CL/F and V/F. HIV-positive status increases V/F by 29.6%. IIV is on CL/F (correlated with V/F at 91.1%), V/F, and EC50; interoccasion variability is on MTT and bioavailability F. Residual error is combined additive + proportional. The same one-compartment + single-transit + autoinduction structural backbone (and the autoinduction parameters MTT, N, Emax, EC50, kENZ) is inherited verbatim by Clewe 2015 and Svensson 2016 -- see modellib('Clewe_2015_rifampicin') and modellib('Svensson_2016_rifampicin'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Svensson_2016_rifampicin.html">Rifampicin (Svensson 2016)</a> </td> <td style="text-align:left;"> Combined population PK/PD model for rifampicin in adults with drug-susceptible pulmonary tuberculosis: a one-compartment, single-transit, oral PK model with first-order plasma-concentration-driven autoinduction of clearance via an enzyme-pool turnover (structure from Smythe 2012) linked to the Multistate Tuberculosis Pharmacometric (MTP) three-state bacterial disease model (fast-, slow-, and nonmultiplying Mycobacterium tuberculosis states; structure from Clewe 2016) with rifampicin drug effects as fixed-at-100% on/off inhibition of fast-multiplying bacterial growth plus second-order plasma-concentration-driven death of slow- and nonmultiplying bacteria; all PK parameters and all MTP transfer/growth rates are fixed to the upstream-paper estimates, while the system carrying capacity Bmax (with 152% CV IIV) and the two second-order death rates SDk and NDk are re-estimated against 19 patients from a 1966-1977 Kenyan rifampicin monotherapy trial. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vinnard_2017_rifampicin.html">Rifampicin (Vinnard 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral rifampicin in HIV/TB patients in Botswana (Vinnard 2017), with a Savic 2007 analytical transit-compartment absorption chain feeding a virtual depot, oral bioavailability fixed at 1, between-subject variability on CL, F, MTT, and the (non-integer) number of transit compartments NN, and inter-occasion variability on F across two sampling visits (pre-ART vs after approximately 4 weeks of ART). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wicha_2018_rifampicin.html">Rifampicin (Wicha 2018)</a> </td> <td style="text-align:left;"> Preclinical-to-clinical translational Multistate Tuberculosis Pharmacometric (MTP) framework for high-dose oral rifampicin in adults with pulmonary tuberculosis. The Svensson 2018 HIGHRIF1 plasma PK model (Erlang transit absorption + Michaelis-Menten clearance + enzyme-pool autoinduction + dose-dependent bioavailability anchored at 450 mg) is coupled via the Clewe 2015 epithelial lining fluid (ELF) effect compartment to a new post-antibiotic-effect (PAE) compartment with saturable Michaelis-Menten elimination, driving the Clewe 2016 three-state MTP model (fast-, slow-, and nonmultiplying Mycobacterium tuberculosis substates) at human-specific carrying capacity Bmax = 2.42e8/mL and fast-multiplying growth rate kG = 0.206/day. Time unit is days; all PK rates from Svensson 2018 (reported in 1/h) and the ELF kELF from Clewe 2015 are multiplied by 24 to bring to days. All structural parameters are fixed at the Wicha 2018 Table 1 typical values; only the Svensson 2018 IIV is carried (IOV is omitted because the EBA forward simulation models a single 14-day monotherapy course). The model predicts early bactericidal activity (EBA0-2 / EBA0-5 / EBA0-14) for clinical rifampicin doses 2.5-50 mg/kg without re-estimating any parameter from clinical EBA data. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ramachandran_2023_tuberculosis_eptb_pbpk.html">Rifampicin pbpk (Ramachandran 2023)</a> </td> <td style="text-align:left;"> PBPK (whole-body, 18 ODEs, MATLAB ode15s). Oral rifampicin disposition at extrapulmonary tuberculosis (EPTB) sites in a 70-kg reference adult male (Ramachandran and Gadgil 2023, CPT Pharmacometrics Syst Pharmacol). Seventeen perfusion-limited well-stirred tissue compartments connected by blood and lymph flow, plus a gut-lumen state for enterohepatic recycling. The model's two novel states are a pleural-fluid compartment (a filtrate of the lung that drains via lymphatics) and a consolidated lymph-node compartment that collects afferent lymph from every organ except bone and spleen and returns it to venous blood - the two most common EPTB sites. Physiological volumes, blood flows, and lymph flows are fixed literature fractions of body weight, cardiac output (5200 mL/min), and afferent lymph flow (8 L/day) respectively (Appendix S1 Tables S2 and S3). Tissue:plasma partition coefficients were computed by the Rodgers and Rowland method from the drug physicochemistry in Table S4 and are tabulated in Table S5. Only the first-order oral absorption rate ka and total systemic clearance CL were estimated, by weighted least squares against reported plasma concentrations after a 450 mg oral dose (Appendix S1 Table S7). Clearance is split into a renal component fR * CL driven by arterial concentration and a hepatic component (1 - fR) * CL driven by the hepatic inlet concentration; the hepatic output enters the gut lumen, from which rifampicin is reabsorbed at kr = 0.17 /h (enterohepatic circulation) or lost by faecal transit at kF = 0.252 /h. The model is a deterministic typical-value simulation: the paper reports no interindividual variability and no residual error model, so propSd is fixed at 0. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zvada_2010_rifapentine.html">Rifapentine (Zvada 2010)</a> </td> <td style="text-align:left;"> Parent-metabolite population pharmacokinetic model for single-dose 900 mg oral rifapentine (RFP) and its primary active metabolite 25-O-desacetyl rifapentine (25-DRFP) in 34 healthy adult male volunteers, with characterization of food effect on bioavailability for four meal types (high-fat English breakfast A, low-fat bulky maize porridge B, high-fat bulky maize porridge with lard C, and low-fat high-fluid chicken noodle soup D) relative to fasted reference (meal E). Parent RFP is described by a one-compartment model with Savic transit absorption (NN = 10.9, MTT = 1.45 h) and a step-function autoinduction of apparent clearance at MTIME = 43 h (CL1/F = 2.14 to CL2/F = 3.22 L/h). All RFP is assumed to convert to 25-DRFP; metabolite disposition is two-compartment with its own step-function clearance switch at MTIME_M = 46.8 h (CLM1/F = 1.81 to CLM2/F = 4.63 L/h). Meal effects multiply the typical bioavailability via TVF = 1 * (1 + sum of per-meal fractional changes). Inter-individual variability is a 3x3 block on CL/F, MTT, and F (correlations rho_F_MTT = 0.65 and rho_CL_MTT = -0.56; cov(CL, F) assumed 0 since not reported); a single shared eta on CL/F applies to both CL1 and CL2. Inter-occasion variability (Table 2 IOV columns) is omitted because nlmixr2lib does not standardize an OCC encoding; the cross-over IOV magnitudes are noted in the vignette Errata. Residual variability is combined additive + proportional on plasma RFP (additive 0.206 mg/L, proportional 10.6%) and on plasma 25-DRFP (additive 0.211 mg/L, proportional 19.1%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2016_rilotumumab.html">Rilotumumab (Zhang 2016)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for rilotumumab (fully human anti-HGF IgG2 monoclonal antibody) in patients with MET-positive gastric or gastroesophageal-junction adenocarcinoma receiving rilotumumab in combination with epirubicin / cisplatin / capecitabine (ECX). The structural model and parameter values were inherited from the previously developed population PK analysis of rilotumumab (Zhu et al. 2014, J Pharm Sci 103:328-336); Zhang 2016 reports the typical-value point estimates and IIV %CV from that prior model and uses it as the reference for an external visual predictive check assessing whether ECX co-administration alters rilotumumab PK. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Aouri_2017_rilpivirine.html">Rilpivirine (Aouri 2017)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral rilpivirine in HIV-1-infected adults (Aouri 2017), with zero-order absorption from the gastrointestinal tract directly into the central compartment (duration D1 = 4 h; derived mean absorption time D1/2 = 2 h), apparent clearance CL/F = 11.7 L/h, apparent volume of distribution V/F = 401 L, combined proportional plus additive residual error (21.6% and 9.8 ng/mL), and inter-individual variability on CL/F only (33% CV). No demographic, clinical, or genetic covariates (sex, body weight, height, age, race, AST, ALT, HCV, HBV, comedications, CYP3A4*22, CYP3A5*3, CYP2C19*2, CYP2C19*17, UGT1A1*28, UGT1A4*2) were retained in the final covariate model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marcantonio_2022_efa.html">Risankizumab (Marcantonio 2022)</a> </td> <td style="text-align:left;"> QSP. One-compartment monospecific anti-ligand mechanistic PKPD model of risankizumab-p19 binding in adults with plaque psoriasis (Marcantonio 2022 Early Feasibility Assessment, Case Study 3). Risankizumab binds the p19 subunit of IL-23 (specific to IL-23; distinguished from ustekinumab which binds the shared p40). Bivalent binding (assumed valency = 2 as a monoclonal antibody). Structure identical to the Marcantonio 2022 anti-ligand family; parameters FIXED from paper Table S4. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Suleiman_2018_risankizumab.html">Risankizumab (Suleiman 2018)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of risankizumab (anti-IL-23 mAb) with first-order SC absorption in subjects with moderate-to-severe plaque psoriasis and moderate-to-severe Crohn's disease (Suleiman 2018 phase I-II) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Suleiman_2019_risankizumab.html">Risankizumab (Suleiman 2019)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of risankizumab (anti-IL-23 mAb) with first-order SC absorption in healthy subjects and patients with moderate-to-severe plaque psoriasis (Suleiman 2019) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Thakre_2022_risankizumab.html">Risankizumab (Thakre 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK model of risankizumab (anti-IL-23 mAb) with first-order SC absorption in patients with active psoriatic arthritis (Thakre 2022) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cleary_2023_risdiplam.html">Risdiplam (Cleary 2023)</a> </td> <td style="text-align:left;"> Population PK model for risdiplam (Evrysdi) in healthy adults and patients with spinal muscular atrophy aged 2 months to 61 years (Cleary 2023 Clin Pharmacokinet Table 2, final PPK model). Three transit absorption compartments feed a linear two-compartment disposition model. Apparent clearance and intercompartmental clearance scale allometrically with time-varying body weight (estimated exponent 0.276, reference 33 kg); apparent volumes use a separate estimated exponent (0.860). Sigmoidal maturation functions of time-varying postnatal age act on CL/F (Age50 0.877 y) and Vc/F (Age50 0.322 y), and healthy adults carry a higher CL/F than patients with SMA. The proportional residual error switches between venous and capillary blood samples. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Feng_2008_risperidone.html">Risperidone (Feng 2008)</a> </td> <td style="text-align:left;"> Adult one-compartment parent-plus-metabolite population PK model for oral risperidone and its active metabolite 9-OH-risperidone in 490 subjects pooled across the CATIE-AD (n = 110, behavioural symptoms of Alzheimer disease, mean age 78.3 years) and CATIE-SZ (n = 380, schizophrenia, mean age 40.6 years) trials (Feng 2008). First-order absorption with Ka fixed at 1.7 1/h into a single central compartment with first-order elimination; the fraction of risperidone metabolized to 9-OH-risperidone (KF) feeds a single metabolite compartment whose apparent volume of distribution is set equal to the parent apparent volume per the paper's identifiability constraint. A mixture model with three CYP2D6 metabolizer subpopulations (poor PM, intermediate IM, extensive EM) yields subpopulation-specific apparent oral clearances (CL/F) and metabolite formation fractions (KF); CL/F in IM (36 L/h) and KF in IM (1) are fixed per Table 3 to stabilize the mixture estimation. Age is the only retained subject-level covariate, acting on 9-OH-risperidone apparent clearance (CLM/F) via a power model with exponent -0.378 referenced at a nominal median age of 45 years. Inter-individual variability is reported separately for CL/F in PM and EM (no IIV is reported for CL/F in IM or for CLM/F), for Ka (despite a fixed typical value), and for the shared Vd/F; combined additive-plus-proportional residual error is reported separately for risperidone and 9-OH-risperidone plasma concentrations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Sherwin_2012_risperidone.html">Risperidone (Sherwin 2012)</a> </td> <td style="text-align:left;"> One-compartment parent-plus-metabolite population PK model for oral risperidone and its active metabolite (+/-)-9-hydroxyrisperidone in 45 children and adolescents (aged 3-18.3 years, 16.8-110 kg) with neuropsychiatric disorders treated with maintenance oral risperidone (Sherwin 2012). First-order absorption (Ka fixed) into a single central compartment with first-order elimination; the fraction of risperidone metabolized to (+/-)-9-hydroxyrisperidone (KF) feeds a single metabolite compartment whose apparent volume of distribution is set equal to the parent apparent volume per Table 2 footnote (a). A mixture model with three CYP2D6 metabolizer subpopulations (poor PM, intermediate IM, extensive EM) yields subpopulation-specific apparent oral clearances and metabolite formation fractions; KF in IM subjects is fixed at 1 to stabilize the model per the paper's Mixture Model section. Allometric scaling (exponent 0.75 for CL/F and CLM/F, exponent 1 for Vd/F, reference 70 kg) is applied to all three subpopulations' clearance estimates and to the shared apparent volume. Inter-individual variability is reported separately for each subpopulation's CL/F (PM, IM, EM), for the metabolite CLM/F, and for the shared Vd/F; a combined additive-plus-proportional residual error is reported separately for risperidone and (+/-)-9-hydroxyrisperidone plasma concentrations. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2024_risperidone_lai_switching.html">Risperidone consta (Wang 2024)</a> </td> <td style="text-align:left;"> Population PK model for the risperidone active moiety (risperidone + 9-OH-risperidone) after intramuscular gluteal injection of RISPERDAL CONSTA, the reference biweekly long-acting injectable risperidone formulation, in adults with schizophrenia or schizoaffective disorder (Wang 2024). Same structure as the companion Rykindo model: one-compartment disposition fed by three parallel release pathways out of the injection site, namely an immediate zero-order release of fraction F2 into the central compartment over duration D2, a middle release of fraction F3 = 1 - F1 - F2 entering a second depot as a zero-order input of duration D3 after lag ALAG3 and then absorbed first-order with rate K32, and a main first-order release of fraction F1 beginning ALAG1 = 27 days after injection. The long main-release lag reproduces the well-known 3-week delay that obliges 3 weeks of oral risperidone supplementation when Consta is started. Because the active moiety displays flip-flop kinetics, the elimination rate constant K was set equal to KA, so the apparent central volume is derived as V = CL/KA. Apparent clearance is a single value across trials and sexes; only the main-release absorption rate constant differs between the two trials. The companion Rykindo model is modellib('Wang_2024_risperidone_rykindo'). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PillaReddy_2013_panss_subscales.html">Risperidone panss subscales (PillaReddy 2013)</a> </td> <td style="text-align:left;"> Population PK/PD model for risperidone against the three PANSS subscales (positive, negative, general) in adults with schizophrenia from Pilla Reddy 2013 Part II. The driving exposure variable is the active moiety (parent risperidone + the equipotent metabolite 9-hydroxy-risperidone), following the Part I (PMID 23473810) and Vermeulen 2007 (Eur J Clin Pharmacol 63:1063-1077) methodology. The PK sub-model is a simplified one-compartment representation of the active moiety: the published Part I model for risperidone is a two-compartment parent-plus-metabolite system with a lag-time and a consecutive zero-then-first-order absorption process, and clearance stratified by CYP2D6 phenotype (poor / medium / fast metabolizers). This nlmixr2lib representation uses a single CL/F that maps a typical oral risperidone dose to the active-moiety steady-state concentration Css consumed by the PD model: CL_AM/F = 6.3 L/h (derived from Part II Table 3 effective-dose / effective-Css pair 0.8 mg/day / 5.3 ng/mL at 30% PANSS reduction, CL/F = Dose / (Css * tau)), Vc = 144 L (parent central volume from Part I Table 2; approximate active-moiety value), and Ka = 2.37 1/h (parent first- order rate from Part I Table 2). The zero-order absorption duration DUR = 0.47 h and lag time ALAG1 = 0.16 h are omitted in this Part-II simplification (Css is approximately invariant to short-window absorption details at steady state). The PD sub-model has three outputs that share the Weibull placebo time-course form Pplacebo = Pmax * (1 - exp(-(t/TD)^POW)) but each subscale carries its own placebo Pmax, TD, POW (Part II Table 1) and risperidone's own Emax / EC50 / KT triplet per subscale (Part II Table 2). The KT for risperidone PANSS positive and general subscales (0.048 / 0.035 1/day) was estimated as a common value across all atypical antipsychotic drugs (Part II Methods, 'A common model ... was developed for PANSS positive and general scales'); the KT for the negative subscale (0.16 1/day) was estimated separately per drug because the cross-drug pooled fit did not converge. The exponential time-to-event dropout sub-model from Part II Table 4 is documented in population$dropout_model but is not encoded in this model body.
Risperidone rykindo (Wang 2024) Population PK model for the risperidone active moiety (risperidone + 9-OH-risperidone) after intramuscular gluteal injection of RYKINDO (LY03004), a biweekly long-acting injectable risperidone formulation, in adults with schizophrenia or schizoaffective disorder (Wang 2024). One-compartment disposition fed by three parallel release pathways out of the injection site: (1) an immediate zero-order release of fraction F2 directly into the central compartment over duration D2, (2) a middle release of fraction F3 = 1 - F1 - F2 delivered as a zero-order input of duration D3 into a second depot beginning ALAG3 after injection and then absorbed first-order with rate K32, and (3) a main first-order release of fraction F1 from the primary depot beginning ALAG1 after injection with absorption rate constant KA. Because the active moiety displays flip-flop kinetics, the elimination rate constant K was set equal to KA, so the apparent central volume is derived as V = CL/KA. Apparent clearance carries a sex effect (females about 18 percent lower) and a between-study effect (0.675-fold in the two US relative-bioavailability trials relative to the single-ascending-dose trial); KA is higher in the multiple-dose trial. The companion RISPERDAL CONSTA model from the same paper is modellib(‘Wang_2024_risperidone_consta’).
Ritlecitinib base (Wojciechowski 2023) Base model (iteration 1 of 3) for oral ritlecitinib: two-compartment disposition with first-order absorption and a direct-response non-stationary (autoinhibitory) Imax effect of the peripheral-compartment concentration on both apparent clearance and bioavailability. Fitted to intensively sampled healthy participants plus sparsely sampled rheumatoid arthritis and alopecia areata patients (186 individuals, 2174 concentrations). Carries rheumatoid arthritis and alopecia areata effects on CL/F, high-fat-meal and 800 mg dose effects on ka, and inflammatory-disease scaling of the IIV and proportional residual error magnitudes. Allometric weight scaling with fixed exponents 0.75 on clearances and 1 on volumes referenced to 70 kg.
Ritlecitinib final (Wojciechowski 2023) Final model (iteration 3 of 3) for oral ritlecitinib: two-compartment disposition with first-order absorption and a direct-response non-stationary (autoinhibitory) Imax effect of the peripheral-compartment concentration on both apparent clearance and bioavailability. Parameters were re-estimated with the NONMEM PRIOR (NWPRI) frequentist-prior subroutine, using the updated model as the prior, on an evaluation dataset dominated by sparsely sampled phase IIb/III alopecia areata patients plus healthy participants and severe renal impairment participants (601 individuals, 2944 concentrations). Carries an alopecia areata effect on CL/F, a severe renal impairment effect on F, and inflammatory-disease scaling of the IIV and proportional residual error magnitudes. Allometric weight scaling with fixed exponents 0.75 on clearances and 1 on volumes referenced to 70 kg. This is the iteration that underwrote the approved Litfulo (ritlecitinib) product label.
Ritlecitinib updated (Wojciechowski 2023) Updated model (iteration 2 of 3) for oral ritlecitinib: two-compartment disposition with first-order absorption and a direct-response non-stationary (autoinhibitory) Imax effect of the peripheral-compartment concentration on both apparent clearance and bioavailability. Base-model parameters were re-estimated on the pooled healthy participant, rheumatoid arthritis, ulcerative colitis, alopecia areata, vitiligo and moderate hepatic impairment data (668 individuals, 5187 concentrations) and a stepwise covariate analysis was run. Carries rheumatoid arthritis, ulcerative colitis, alopecia areata and vitiligo effects on CL/F; ulcerative colitis and moderate hepatic impairment effects on F; high-fat-meal, 800 mg dose and capsule effects on ka; an over-encapsulated-capsule effect on loss from the depot; and inflammatory-disease scaling of the IIV and proportional residual error magnitudes. Allometric weight scaling with fixed exponents 0.75 on clearances and 1 on volumes referenced to 70 kg.
Ritonavir (Kappelhoff 2005) One-compartment population PK model with first-order absorption, an absorption lag time, and first-order elimination for oral ritonavir in HIV-1-infected adults (186 patients, 1228 plasma concentrations; Kappelhoff 2005). Concomitant lopinavir is the only retained covariate and multiplies apparent oral clearance by 2.72-fold (power form: CL/F = exp(lcl) * 2.72^CONMED_LPV). Inter-individual variability on apparent CL/F, V/F, and ka, with correlated etas for V and ka (rho = 0.868). Residual error has a single 15.4% proportional component and a mixture-model additive component (subpopulation P1, 64.8% of subjects: 0.0600 mg/L; subpopulation P2, 35.2%: 0.199 mg/L), gated by the binary covariate MIX_LARGE_RUV. Interoccasion variability on apparent bioavailability (59.1% in the source) is not propagated – see the validation vignette Assumptions and deviations section.
Ritonavir (Mensing 2017) One-compartment population PK model for oral ritonavir (co-dosed with paritaprevir as a CYP3A4 pharmacokinetic enhancer) in HCV genotype-1 infected adults receiving the 3D regimen (Mensing 2017). First-order absorption, linear elimination, combined proportional + additive residual error, IIV on CL/F only. The author’s final model retained gender, creatinine clearance, and HCV genotype (1a vs 1b) as significant covariates on CL/F, but the paper does not publish point estimates for these covariate coefficients (only graphical exposure-ratio forest plots in Figure 2); the implemented model is the structural typical-value model with covariate coefficients omitted (documented in covariatesDataExcluded).
Rituximab (Candelaria 2018) Two-compartment population PK model of rituximab (and its biosimilar RTXM83) with linear distribution and linear elimination from the central compartment in patients with diffuse large B-cell lymphoma (DLBCL) treated with rituximab-CHOP or RTXM83-CHOP (Candelaria 2018; pooled-arm fit, all 5341 concentrations from both treatment arms)
Rituximab (Chen 2024) Two-compartment population PK of rituximab coupled to a CD19+ B-lymphocyte turnover PD model with saturable stimulation of B-cell loss, in pediatric patients with frequent-relapsing or steroid-dependent nephrotic syndrome
Rituximab (Liang 2024) Two-compartment quasi-steady-state target-mediated drug disposition (TMDD) population PK/PD model for rituximab in adults with primary membranous nephropathy, with CD20 as the target; outputs are rituximab concentration and total CD20 (the molar surrogate for CD20+ B cell count) (Liang 2024)
Rituximab (Lioger 2017) Two-compartment population PK model of rituximab in rheumatoid arthritis patients, parameterised with first-order distribution (k12, k21) and elimination (k10) rate constants (per Lioger 2017 Methods ‘Structural pharmacokinetics model design’: rate-constant parameterisation gave lower AIC, shrinkages, and Vc/elimination correlation than CL/Vc). Final covariate model: BSA, sex, and rituximab treatment course on V1; time-varying CD19+ B-cell count and IgG serum concentration on k10 (Lioger 2017 Table 2).
Rituximab (Rozman 2017) Two-compartment population PK model of rituximab in adults with diffuse large B-cell lymphoma (DLBCL) receiving R-CHOP; total CL is the sum of a time-stationary non-specific (IgG-catabolic) component cl_exp_inf and a mono-exponentially decaying target-mediated component cl_exp_component * exp(-cl_exp_kdes * time), with age and body weight on cl_exp_inf, sex on V1 (central volume), and a post-hoc progression-free-survival event indicator (PFS_EVENT) on cl_exp_kdes (Rozman 2017).
Rituximab acr (Williams 2016) Longitudinal ordered-categorical PD model of American College of Rheumatology (ACR) response (ACR20 / ACR50 / ACR70) over time for rituximab (both the proposed biosimilar PF-05280586 and the two rituximab reference products sourced from the EU and US) in adults with rheumatoid arthritis on background methotrexate with prior inadequate response to one or more TNF-antagonist therapies (Williams 2016). The model is a cumulative-probit latent-variable formulation with three thresholds (ACR20, ACR50, ACR70 as an ordered categorical score in {0, 1, 2, 3}) and an exponential-onset time course for the population-mean latent variable. Unlike the companion DAS28cfb model, no rituximab exposure effect is included (Williams 2016 Results: ‘rituximab exposure effects could not be supported in the ACR model’). The covariate model includes seven baseline disease-activity components (TJ68, SJ66, log CRP+1, PGA_PT, PhGA, PAIN, HAQ-DI) and two treatment-arm indicators (PF-05280586 vs rituximab-EU, rituximab-US vs rituximab-EU) additive on the log scale on both the maximum-effect parameter PMAX and the onset half-life parameter, with rituximab-EU as the reference arm. Outputs three continuous probabilities pACR20, pACR50, pACR70 = probability of achieving at least ACR20/50/70 response at the observation time; a single subject-level additive probit-scale eta captures between-subject variability.
Rituximab das28cfb (Williams 2016) Longitudinal PD model of Disease Activity Score in 28 joints change from baseline (DAS28cfb) for rituximab (both the proposed biosimilar PF-05280586 and the two rituximab reference products sourced from the EU and US) in adults with rheumatoid arthritis on background methotrexate with prior inadequate response to one or more TNF-antagonist therapies (Williams 2016). The model is an inhibitory-Emax-on-log-scale exposure-response for the concentration-driven drug effect superimposed on an exponential-onset placebo/background time course, with the outer transformation DAS28cfb = 1 - exp(fnon-C(t) + fC(C)) so that improvement is a negative change from baseline. The typical maximum placebo/background effect PMAX and typical maximum drug effect Emax are log-scale additive parameters with correlated between-subject etas; a separate additive subject-level eta on the outer scale (etaCS) captures the within-subject correlation induced by computing change from baseline. Covariate model includes the individual components of baseline disease activity (TJ28, SJ28, log CRP+1, patient global assessment) and treatment-arm indicators on each of the three main structural parameters (PMAX, kp, Emax) with rituximab-EU as the reference arm. This is a PD-only extraction: the rituximab plasma concentration Cc is supplied as a time-varying covariate CP_RITUXIMAB_UGML (the Williams 2016 popPK model - the source of the individual predicted concentrations Cij feeding the DAS28cfb model - was a two-compartment structural model whose parameter values are not reported in the paper or its supplements; users wanting to drive this PD model from a simulated PK source must supply their own concentration trajectory).
Rituximab pla2r (Liang 2024) Empirical mono-exponential decline of the serum anti-PLA2R autoantibody titer after rituximab in adults with primary membranous nephropathy; fitted independently of the CD20+ B cell count, so it is a separate model from the companion QSS TMDD PK/PD model (Liang 2024)
Rivaroxaban (Willmann 2018) Paediatric population PK model for oral rivaroxaban in children aged 0.5-18 years (Willmann 2018, EINSTEIN-Jr phase I). Linear two-compartment model with first-order absorption from a depot and first-order elimination from the central compartment; CL and central V allometrically scaled to body weight (CL exponent 0.323 estimated; V exponent 1 fixed); ka shifted between the undiluted-oral-suspension formulation and the tablet / diluted-oral-suspension reference; relative bioavailability F1 reduced for the 20 mg-equivalent body-weight-adjusted dose relative to the 10 mg-equivalent reference.
Rivaroxaban (Willmann 2021) Pediatric population PK model for rivaroxaban developed on the integrated EINSTEIN-Jr phase I / I-II / II / III dataset and interim PK from part A of the UNIVERSE study (524 children, 1988 plasma concentrations, age birth to <18 years, body weight 2.7-194 kg). Two-compartment disposition with first-order absorption and first-order elimination from the central compartment. Body weight enters as estimated allometric scaling on CL, Q, Vc, and Vp, centred on the 82.48 kg median of the integrated adult popPK analysis (a shared exponent is used for Vc and Vp). The undiluted ready-to-use oral suspension has a lower first-order absorption rate constant ka than the other three formulations (tablet, granules for oral suspension, and diluted ready-to-use oral suspension), which share a common ka. Relative oral bioavailability decreases with dose per body weight following an exponential function carried over from the integrated adult popPK analysis (anchored to F1 = 1 at 10 mg / 82.48 kg = 0.1213 mg/kg). Inter-individual variability is on CL and F1 only (no IIV on Ka, Vc, Vp, or Q); residual error is proportional. Age, eGFR (Schwartz and Rhodin), serum creatinine, comedications (CYP3A4 inhibitors / inducers, P-gp inhibitors), and Fontan status were tested and not retained.
Rivaroxaban (Willmann 2022) Adapted pediatric population PK model for rivaroxaban in 76 post-Fontan congenital heart disease patients aged 2-8 years (body weight 9.8-25.3 kg) from the UNIVERSE study. Structural framework is inherited from the EINSTEIN-Jr popPK model of Willmann 2021 (already extracted as Willmann_2021_rivaroxaban): 2-compartment disposition with first-order absorption from a depot, allometric body-weight scaling of CL, Vc, Vp, and Q centred on the 82.48 kg adult reference weight, with all structural parameters (ka, Vc, Vp, Q, and the three allometric exponents) fixed to the Willmann 2021 EINSTEIN-Jr estimates. The Willmann 2021 dose-dependent relative bioavailability function is replaced by a bimodal age-binned F1 (post-Fontan patients aged <5 years vs. >=5 years). Apparent CL, the two F1 values, IIV on CL and F1, and the proportional residual error are re-estimated on the 76-patient UNIVERSE dataset. The refined CL is 6.07 L/h at 82.48 kg vs. 8.02 L/h in EINSTEIN-Jr (24% lower in post-Fontan patients). This model was used to bridge doses for thromboprophylaxis in post-Fontan patients aged 9 years or older or >=30 kg (the target extrapolation population), leading to the US label of 7.5 mg once daily (30-<50 kg) and 10 mg once daily (>=50 kg).
Rivipansel (Tammara 2017) Three-compartment IV population PK model for rivipansel in adults and adolescents with sickle cell disease (SCD) and in healthy adult volunteers (Tammara 2017). Rivipansel is a pan-selectin antagonist given as a 20-minute IV infusion; renal excretion of unchanged drug is the primary clearance mechanism. The integrated population PK model pools 109 subjects across three phase I studies (rivipansel studies 101, 102, 103) and one phase II SCD study (NCT01119833, Telen 2015). Clearance is a power function of creatinine clearance (CRCL, raw Cockcroft-Gault mL/min reference 150) with an additive 23.4% shift in the phase II SCD cohort (STUDY_RIV201) attributed to glomerular hyperfiltration. The central, first peripheral, and second peripheral volumes share a single estimated body-weight exponent (0.569, reference 70 kg). The additive and proportional residual error magnitudes differ between the phase I and phase II cohorts and are selected per observation via STUDY_RIV201.
Rlyb212 (MocWilleford 2024) Target-mediated drug disposition (TMDD) model of RLYB212, a recombinant human anti-HPA-1a IgG1 monoclonal antibody, with simultaneous fit of RLYB212 pharmacokinetics and HPA-1a-positive platelet dynamics in HPA-1b/b healthy volunteers (Moc Willeford 2024). Two-compartment SC PK for RLYB212 (central Vc, peripheral Vp, ka, CL) shares its central volume with the HPA-1a-positive platelet distribution (target central), which itself has a peripheral compartment (Vp_target) and first-order natural degradation kdeg. RLYB212 and free receptor form a drug-receptor complex in the central compartment via second-order kon and first-order koff (both fixed at internal / literature values). A novel threshold-gated phagocytic elimination pathway removes coated platelets (both free receptor and complex) once receptor occupancy exceeds a fixed threshold thres = 10%. Parameters and structure per Moc Willeford 2024 Table 2 and the supplement’s differential-equation appendix.
Roblitinib (Wilbaux 2022) Two-compartment population PK model for oral roblitinib (FGF401), a selective FGFR4 inhibitor, in adults with hepatocellular carcinoma or other FGF19-FGFR4-expressing solid tumors (Wilbaux 2022). The paper describes a delayed zero-order absorption directly into the central compartment (lag time Tlag before absorption starts, then duration Tk0 of the zero-order input rate) with linear elimination. Categorical covariate effects on CL/F and V1/F for female sex (SEXF) and Asian race (RACE_ASIAN) with non-Asian male as the reference; on Tk0 for fed vs fasted (FED). Continuous power-form covariate effects: body weight on V1/F (exponent 0.332), BMI on Tk0 (exponent -1.66), and administered dose on Tk0 (exponent 0.983). The source paper reports these covariates as identified but not clinically relevant based on simulated exposure metrics. Diagonal IIV on Tlag, Tk0, CL/F, V1/F, and V2/F (Q/F fixed at 0 IIV per Table 1). Combined additive + proportional residual error.
Rocatinlimab (Okada 2025) Two-compartment population PK model with parallel linear and time-dependent saturable (Michaelis-Menten) clearance and first-order subcutaneous absorption for rocatinlimab (anti-OX40 mAb) in adults; covariates body weight, albumin, plaque-psoriasis disease state, and healthy-volunteer cohort indicator (Okada 2025)
Roflumilast (Facius 2018) Integrated population PK model for oral roflumilast and its primary active metabolite roflumilast N-oxide in adult patients with severe chronic obstructive pulmonary disease (COPD) (Facius 2018). The structural model is the joint parent-metabolite model previously developed by Lahu 2010 on 21 phase I + 2 phase II/III studies: a two-compartment parent disposition with first-order absorption and a shared lag time, and a two-compartment N-oxide disposition with first-order absorption from a separate pre-systemic dose compartment (relative bioavailability F5) plus complete first-order conversion from the parent central compartment. All structural disposition parameters and the F5 / KAm-to-KAp ratio are fixed to the Lahu 2010 base-model estimates re-applied to OPTIMIZE via a Bayesian feedback MAXEVAL = 0 step; only the phase II-III dichotomous patient effects (on KA, parent CL, N-oxide CL, and N-oxide central volume), the covariate effects, the between-subject variability on parent and N-oxide clearance (with a Box-Cox-shape transformation), and the log-additive residual errors were estimated on the combined OPTIMIZE and REACT phase III dataset of 1238 + 461 patients. Covariates retained are body weight on all volume terms and on N-oxide CL, smoking (current vs not-current) on parent and N-oxide CL, age on parent and N-oxide CL, and sex on N-oxide CL. tPDE4i (total phosphodiesterase-4 inhibitory activity), the exposure metric used in the paper’s downstream PK/adverse-event and PK/time-to-event models, is a per-dosing-interval summary derived from the predicted average plasma concentrations and is not embedded in the ODE system; see the vignette for the derivation.
Roflumilast (Lahu 2010) Joint parent-metabolite population PK model for oral roflumilast and its primary active metabolite roflumilast N-oxide in adult healthy volunteers and patients with moderate-to-severe COPD (Lahu 2010). Roflumilast is described by a two-compartment model with first-order absorption and a lag time; the absolute parent bioavailability is not identifiable and is fixed at F1 = 1. Roflumilast N-oxide is described by a one-compartment model with zero-order absorption (duration D1) and a lag time, with relative bioavailability Frel fixed at 1 for the null-covariate reference (also non-identifiable). Retained covariates on roflumilast parameters are food on tlag and ka, sex / smoking / race-Black / race-Hispanic / COPD on CL, and COPD on V1. Retained covariates on roflumilast N-oxide parameters are food on D1; age / sex / smoking / COPD on CL; body weight and COPD on Vd; and age / sex / race-Black / race-Hispanic on Frel. Inter-individual variability is reported on parent tlag, ka, CL, V1, Q, V2 (with a Q-V2 covariance) and on N-oxide D1, CL, Vd (with a full 3x3 covariance block); no IIV is reported on N-oxide tlag or on Frel. Residual error is proportional on the linear- concentration scale (additive on the log-transformed observation) for both observed analytes, fitted on the phase I dataset (the more data-rich layer).
Rolofylline (Stroh 2013) Simultaneous three-output population PK model for IV rolofylline (adenosine A1 receptor antagonist) and both M1-trans and M1-cis active hydroxyl metabolites in 36 healthy adult male volunteers after single 1-60 mg IV infusions over 1-2 h (Stroh 2013, study KW-3902 IV-EU01). Parent rolofylline disposition is two-compartment with linear total clearance CL1 directing the entire parent loss to metabolite formation; the fraction FM of formed material is converted directly to M1-cis while (1 - FM) is converted to M1-trans. M1-trans disposition is two-compartment with distributional clearance CL4 and an additional unidirectional stereochemical interconversion clearance CL3 from M1-trans to M1-cis. M1-cis disposition is one-compartment with first-order clearance CL5. Random effects were not estimable for parent and M1-trans distributional clearances (CL2, CL4) or for the M1-cis central volume (V5) and were fixed to zero per the source. No covariates were retained: the cohort was a single Phase 1 dose-escalation in white male volunteers and the paper screened no demographic effects. Structural identifiability of the final model was confirmed via the DAISY software tool (Bellu et al.).
Romiplostim (Petrov 2024) Population PK/PD model for romiplostim in adults with chronic immune thrombocytopenia (ITP). One-compartment first-order subcutaneous PK plus an Emax stimulation of platelet precursor production into a 4-transit-compartment Friberg-style chain feeding circulating platelets, with first-order platelet degradation. PK/PD backbone is the healthy-volunteer population PK/PD model (Makarenko 2024); ITP-specific platelet production (kin) and degradation (kdeg) constants and IIV(kdeg) come from Petrov 2024 supplement Table S1. Default parameters are non-splenectomized ITP patients with mechanism 1 (increased platelet degradation, normal precursor production); see vignette for the other 3 subpopulation variants (non-splenectomized mechanism 2; splenectomized mechanism 1; splenectomized mechanism 2).
Romiplostim (Wang 2010) Population PK/PD model for romiplostim in healthy subjects (Wang 2010 AAPS J). Pharmacodynamics-mediated drug disposition (PDMDD, a TMDD subtype) two-compartment quasi-equilibrium PK with first-order SC absorption, parallel linear (kel) and target-mediated (kint) elimination, coupled to a Krzyzanski-style cytokinetic precursor + platelet lifespan PD model with NP=10 megakaryocyte and NPLT=10 platelet age-compartments. Romiplostim free serum concentration stimulates platelet precursor production via a Hill function (Smax, SC50). The total c-Mpl receptor concentration is taken proportional to the circulating platelet count (Rtot = xi * PLT). Wang 2010 fit the model to MEAN PK and platelet-count data from 32 healthy subjects after single IV (0.3, 1, 10 ug/kg) or SC (0.1, 0.3, 1, 2 ug/kg) doses; no IIV was estimated (the population approach failed for this complex model).
Romosozumab (Stein 2018) Two-compartment QSS TMDD typical-value fit for romosozumab (anti-sclerostin mAb) used to illustrate the critical concentration (Ccrit) for nonlinear PK (Stein and Peletier 2018 Table 1)
Ropivacaine (Ollier 2015) Semi-mechanistic population PK model for free and total ropivacaine after transversus abdominis plane (TAP) nerve block in adult patients undergoing liver resection surgery (Ollier 2015). One-compartment first-order absorption disposition for free ropivacaine (Rfree) with apparent clearance Cl/F and apparent volume V/F. Free ropivacaine binds reversibly to a latent unbound-binding-site pool (target) that approximates alpha-1 acid glycoprotein (AAG) via 1:1 mass-action kinetics: binding rate kbind (fixed at 100 uM^-1 h^-1 per the paper’s reported insensitivity of kb over 10^2 - 10^15 uM^-1 h^-1) and dissociation constant Kd (fixed at 0.557 uM, prior-pinned in the paper and reported without RSE). The bound species (complex) plus free species give total ropivacaine. Binding-site production rate kin switches on at 12 h post-incision (the 2nd TAP bolus timepoint, representing the postoperative-inflammation-driven onset of AAG acute-phase response). Two retained covariates: allometric power on Vc for body weight (reference 70 kg, exponent 1.28); multiplicative exponential effect on Cl for major hepatic resection LIVER_RESECT_MAJOR = 1 (>=3 segments) vs 0 (2 segments), reducing free ropivacaine clearance from 1310 L/h to 620 L/h (53% drop). The paper’s third retained covariate, a per-subject postoperative fibrinogen-slope effect on kin (beta = 0.422), was dropped in this packaged model because the population mean fibrinogen slope used to centre the covariate was not reported in any source on disk; see vignette Assumptions and deviations. Ropivacaine is dosed as 5 TAP boluses of 3 mg/kg (10.9 umol/kg) at 0, 12, 24, 36, 48 h post-incision by protocol. Concentrations throughout are in molar units (uM) matching the paper.
Rosuvastatin (Barnett 2018) Two-compartment population PK model with first-order oral absorption for a single 5 mg dose of rosuvastatin in healthy adult males (Barnett 2018), refit from the Tzeng 2008 structural form with simultaneous plasma + urine fitting. The model includes separable biliary (CLb,RSV) and renal (CLr,RSV) clearance components from the central compartment, competitive rifampicin OATP1B inhibition of the biliary clearance via KiRSV driven by the instantaneous plasma rifampicin concentration, and a binary RIF-coadministration covariate that captures paper-reported reductions of V1, V2, and Q during the rifampicin phase (Barnett 2018 Table 1: V1 430 -> 2.98 L, V2 865 -> 128 L, Q 45.3 -> 5.03 L/h on RIF). Companion to modellib(‘Barnett_2018_coproporphyrin_I’); both share the rifampicin perpetrator parameterisation in modellib(‘Barnett_2018_rifampicin’).
Rosuvastatin (Kakara 2014) PD-only indirect-response Imax model for LDL-cholesterol lowering by rosuvastatin (Kakara 2014). One LDL-C compartment with zero-order synthesis Kin inhibited by Imax * DOSE / (ID50 + DOSE), where DOSE is the current daily rosuvastatin dose (mg/day) supplied as a time-varying covariate column. An additive 0.109 contribution to the inhibition fraction is applied when ezetimibe is coadministered (CONMED_EZE = 1; estimated on the n=12 rosuvastatin + ezetimibe subgroup in this paper). The LDL-C synthesis-elimination loop is set up at steady state by enforcing Kin = Baseline * Kout (Kout derived inside model() as Kin / Baseline). Baseline LDL-C is age-scaled as 152 * (AGE/62)^(-0.240). Imax (0.567), Kin (32.8 mg/dL/day), Baseline (152 mg/dL), the age power exponent (-0.240), the ezetimibe INH contribution (0.109), and the IIV magnitudes are shared with Kakara_2014_atorvastatin and Kakara_2014_pitavastatin (one joint NONMEM 7.2 FOCE-INTER fit across 378 patients). Rosuvastatin ID50 = 1.04 mg per Kakara 2014 Table 2.
Rosuvastatin (Macpherson 2015) Two-compartment population PK model with first-order oral absorption for rosuvastatin in pediatric patients (aged 6 to <18 years) with heterozygous familial hypercholesterolemia (Macpherson 2015 Eur J Clin Pharmacol). Apparent clearance scales with body weight (estimated power exponent 0.352, reference 42 kg) and is 1.41-fold higher in males than females. Residual error is proportional and switches between intensive and sparse PK sampling phases.
Rosuvastatin mbma (Yang 2010) MBMA. Literature-based meta-analysis simple Emax dose-response model for percentage reduction in low-density lipoprotein cholesterol (LDL-C) from baseline in adult hypercholesterolemia patients receiving rosuvastatin. Operates at the study-arm level over 14 dose-ranging trials (46 study-arm-mean effect samples; 9 Western trials and 5 Asian trials, total N substantially larger than 46 because each arm pools many patients). Output Cc is the study-arm mean percent LDL-C reduction from baseline (unsigned: Cc = 50 means a 50 percent reduction). The placebo intercept E0 (-0.802 percent, a small expected LDL-C increase under placebo) and the Hill / sigmoidicity exponent (1) are fixed at the values used by the source paper – E0 from prior literature [Mandema 2005, ref 15] and gamma after the sigmoidal Emax fit produced unstable estimates. Race (Asian vs Western reference) is the only retained covariate and acts on ED50: ED50_Asian = ED50_Western * 0.564 (i.e. roughly twofold-lower ED50 in Asians). Between-trial variability is encoded as a single study-arm-level eta on the predicted output (SD 3.0 percent); residual error is additive (SD 3.1 percent). Baseline LDL-C was screened but not retained. Suitable simulation scope is study-arm-mean percent LDL-C reduction, NOT individual-subject LDL-C trajectories. The model also predicts only the steady-state effect (paper restricted to arms with at least 4 weeks of treatment).
Rucaparib (Wang 2015) Three-compartment IV population PK model coupled to a direct-effect Emax PK/PD model for inhibition of poly(ADP-ribose) polymerase (PARP-1) activity in peripheral blood lymphocytes (PBL) by rucaparib (AG-014699 / PF-01367338) in adult cancer patients (Wang 2015 Phase 1 study A4991002), with a power covariate effect of baseline PBL PARP activity on the residual maximum-inhibition parameter Emin.
Rutin rat (Chen 2024) Preclinical (rat). Two-compartment intravenous pharmacokinetic model for rutin (RT), one of eight constituents of guhong injection (GHI) quantified in plasma, in male Sprague-Dawley rats subjected to 30 min left-anterior-descending ligation followed by 1 h reperfusion (myocardial ischemia/reperfusion, MI/R) (Chen 2024). GHI was given as a single tail-vein injection of 2.5, 5 or 10 mL/kg; the rutin dose is the GHI volume dose times its content in GHI (11.7 ug/mL), i.e. 29.25 ug/kg, 58.5 ug/kg, 117 ug/kg. Disposition was fitted separately in each dose group with Drug and Statistics (DAS) v3.2.6, so V1, V2, CL1 and Q are selected from the covariate DOSE_GHI_MLKG rather than through a dose-covariate function the authors did not fit. Direct-effect sigmoid-Emax models link the rutin plasma concentration to the GHI-minus-model-group difference in creatine kinase-MB (CK-MB) and cardiac troponin I (cTn I) (E = Emax * C^gamma / (ED50^gamma + C^gamma); Tables 12, 14). Chen 2024 fitted a PK/PD model only for the analyte/biomarker/dose combinations whose PLSR coefficient was negative, so the effect of an unfitted combination is returned as zero rather than extrapolated. No between-subject variability or residual error was reported; every parameter is fixed at the published mean and the residual SDs are fixed at zero.
Rwj416457 (Schmidt 2009) In vitro (Staphylococcus aureus MRSA strain OC2878). Mechanism-based PD model of bacterial-killing time-kill curves for RWJ-416457, an investigational oxazolidinone (Schmidt 2009). Susceptibility-based two-subpopulation structure: an active self-replicating susceptible pool with logistic carrying-capacity limit and a dormant persister pool that is insusceptible to killing; first-order S->P conversion (P->S held fixed at 0), natural-death loss from both pools, exponential turn-on of growth and of drug-induced killing, and Emax killing of the susceptible subpopulation by the antibiotic. Drug concentration in the Mueller-Hinton broth (MHB) declines first-order at the published 10%-over-24-h degradation rate; for dynamic syringe-replacement experiments the user adds the dilution-equivalent rate to kdeg via rxSolve(…, params = c(kdeg = )). The same joint fit is shared with Schmidt_2009_linezolid (only EC50 and kdeg differ).
S ketamine (Flint 2017) Joint two-compartment S-ketamine + one-compartment S-norketamine population PK model for continuous intravenous S-ketamine infusion during prolonged sedation in pediatric intensive care patients aged 0.02-12.5 years (Flint 2017). The parent S-ketamine has two-compartment disposition (CL = 112 L/h, V1 = 7.73 L, Q = 196 L/h, V2 = 545 L at 70 kg) and feeds the active metabolite S-norketamine, modelled as one apparent central compartment with Clsnk/Fm = 53.2 L/h and Vsnk/Fm = 1 L (fixed; Fm is not identifiable). Body weight is allometrically scaled with fixed exponents 0.75 for clearances and 1.0 for volumes referenced to 70 kg; time after the first S-ketamine dose acts as a linear positive multiplier on Clsnk (0.870 percent per hour), the only retained covariate at backward elimination.
S methadone (Aruldhas 2021) Population pharmacokinetic model for S-methadone and its S-EDDP metabolite in post-operative paediatric surgical patients (Aruldhas 2021; 61 children aged 11-17 years given 0.1 mg/kg IV racemic methadone intra-operatively followed by 0.1 mg/kg oral racemic methadone every 12 h as either a tablet or an oral suspension). S-methadone disposition is a two-compartment model with first-order absorption (separate tablet and suspension Ka values), an estimated absolute bioavailability F common to both oral formulations, and a one-compartment S-EDDP metabolite whose central volume is set equal to the S-methadone central volume (scaling factor VF fixed to 1 due to metabolite unidentifiability). Total S-methadone clearance CL scales allometrically with body weight (fixed exponent 0.75, reference 70 kg). The fractional S-methadone clearance to S-EDDP (CLF) is a linear function of the CYP2B6 activity score and the number of active alleles of the intronic CYP3A4 SNP rs2246709 (Table 2 final model chosen in preference to the alternative rs11882424 model because rs2246709 is more commonly genotyped in clinical laboratories). The S-methadone central volume V2 is a linear function of the plasma alpha-1 acid glycoprotein (AAG) concentration and the number of active alleles of the ORM1 SNP rs17650. Between-subject variability is estimated on CL, V2, peripheral volume, CLF, and S-EDDP clearance; residual error is proportional for both S-methadone and S-EDDP.
S33138 (Bertrand 2011) Joint parent-metabolite population PK model for the investigational antipsychotic S33138 (parent) and its active metabolite S35424 in adults with schizophrenia (Bertrand 2011). The final selected structural model is a two-compartment back-transformation form with a presystemic dose apportionment Fp into the parent depot vs (1 - Fp) into the metabolite depot, a shared first-order absorption rate Ka, and four linear elimination / interconversion clearances: parent elimination via other pathways (CLpo), parent-to-metabolite formation (CLpm), metabolite elimination via other pathways (CLmo), and metabolite back-transformation to parent (CLmp). The two volumes (parent Vp and metabolite Vm) are set equal to a single volume V for identifiability per the source paper. CYP2D6 poor metabolizers carry a 34% decrease in CLmo (the genetic covariate retained in the final model). Linear dose-level effects on the bioavailability f and the parent-fraction Fp are encoded with a 10 mg reference dose. Parameter values are from Table IV ‘With the genetic covariate’ column (SAEM in MONOLIX, closed-form coding, N = 99 patients with available CYP2D6 genotyping).
Sacituzumab (Sathe 2024) Coupled three-analyte population PK model for sacituzumab govitecan (SG, the ADC; output Cc), free SN-38 (released payload; output Cc_sn38), and total antibody (tAB; output Cc_tab) in adults with metastatic triple-negative breast cancer and other solid tumors (Sathe 2024). All three analytes are described by two-compartment models with body-weight allometric scaling. SG carries IIV on CL and a baseline-albumin power covariate on CL. Free SN-38 is generated from SG by a first-order release rate KREL with apparent volumes fixed to literature values (Klein 2002). tAB has time-dependent CL (asymptotic onset, max ~17% reduction at t1/2 ~48 days), correlated IIV on CL and V1, and covariates of baseline albumin (CL), tumor type (CL), and sex (V1). Simulation requires dosing two compartments simultaneously (central and central_tab) for each SG infusion event.
Sacituzumab (Sathe 2025) Updated coupled three-analyte population PK model for sacituzumab govitecan (SG, the ADC; output Cc), free SN-38 (released payload; output Cc_sn38), and total antibody (tAB; output Cc_tab) after pooling data from TROPiCS-02 (HR+/HER2- mBC) with IMMU-132-01 and ASCENT (mTNBC + mUC + HR+/HER2- mBC + other solid tumors) (Sathe 2025). Structure is identical to Sathe 2024 (see Sathe_2024_sacituzumab): three two-compartment models with body-weight allometric scaling; SG has IIV on CL and a baseline-albumin power covariate on CL; free SN-38 is generated from SG central by a first-order release rate KREL with apparent V1 and V2 fixed to literature values; tAB has time-dependent CL (asymptotic onset), correlated IIV on CL and V1, and covariates of baseline albumin (CL), tumor type (CL), and sex (V1). Parameter values are the pooled 3-study updated estimates from Sathe 2025 Tables 2, 3, and 4. Simulation requires dosing two compartments simultaneously (central and central_tab) for each SG infusion event.
SAL003 (Peng 2024) Two-compartment population PK model for SAL003, a novel anti-PCSK9 IgG4 monoclonal antibody, with first-order SC absorption (with lag time), saturable Michaelis-Menten elimination from the central compartment, and a body-weight effect on central volume, in Chinese healthy volunteers and patients with hyperlipidemia (Peng 2024)
Salbutamol (Heuberger 2018) Semi-physiological PK simulation model for inhaled and oral salbutamol with its sulphate metabolite (S-SAL) in adult elite athletes. Eight compartments (gut, two-compartment parent disposition, parent plasma metabolite arm, cumulative parent urine, cumulative S-SAL urine, cumulative urine volume) with allometric scaling on disposition and physiological scaling on the cardiac-output-driven urine production rate, synthesised from literature (Auclair 2000 dog model, Morgan 1986 renal CL, Holt 1968 cardiac output, Moerkeberg 2009 haematocrit) and calibrated to Haase 2009 inhaled-salbutamol data (Heuberger 2018).
Salbutamol (Marques 2024) One-compartment population PK model for inhaled salbutamol (600 ug single dose, unit-dose dry-powder inhaler) in healthy adults, with first-order absorption, no lag time, and linear elimination. Encodes the EXTERNAL-VALIDATION fit to real clinical data (GSK study NCT01984086, n = 30; Marques 2024 Table 4) – NOT the paper’s headline virtual-patient model (Table 3), whose central volume of 0.02e-9 L is not solvable at any ODE tolerance and overshoots the paper’s own observed Cmax by four orders of magnitude. The authors declare a two-compartment structure, but their clinical-data fit returned Q = 1.30e-7 L/h and V2 = 0 exactly, both with relative standard errors above 1e6 percent (the authors’ own footnote calls this overparametrization); the peripheral compartment therefore carries no flux and k21 = Q/V2 is undefined, so the model is encoded honestly as one compartment. No covariate effects are encoded: the paper reports which covariates were significant but publishes no coefficients, functional forms, or centering values anywhere in the article or its supplement (see covariatesDataExcluded and the vignette Errata).
Sapropterin (Feillet 2008) Two-compartment population PK model with first-order oral absorption, an absorption lag, linear elimination, and an additive endogenous BH4 baseline for sapropterin dihydrochloride in adolescent and adult patients with BH4-responsive phenylketonuria (Feillet 2008).
Sapropterin (Muntau 2017) One-compartment population PK model with first-order oral absorption, an absorption lag, linear elimination, and an additive endogenous BH4 baseline for sapropterin dihydrochloride in pediatric patients <4 years with BH4-responsive phenylketonuria or mild hyperphenylalaninemia (Muntau 2017 SPARK trial).
Sapropterin (Qi 2014) One-compartment population PK model with first-order oral absorption, an absorption lag, linear elimination, and an additive endogenous BH4 baseline for sapropterin dihydrochloride in pediatric and adult patients with phenylketonuria (Qi 2014).
Saquinavir (vonHentig 2009) One-compartment first-order-absorption population PK model for oral ritonavir-boosted saquinavir (1000/100 mg BID) in 136 HIV-1-infected adults including 13 pregnant women. Apparent oral clearance CL/F is modulated by two retained covariates: a binary atazanavir-coadministration indicator (CONMED_ATAZANAVIR; 49 of 136 patients on ATV 300 mg QD) as a power-of-binary multiplier 0.703^CONMED_ATAZANAVIR (30% CL reduction when atazanavir is coadministered), and the per-subject ritonavir 12 h AUC (CONMED_RTV_AUC_12h, cohort median 6.70355 mg*h/L) as a normalised power form (CONMED_RTV_AUC_12h / 6.70355)^(-0.403). Saquinavir formulation (Invirase hard gel vs Fortovase soft gel) was tested and not retained. Inter-individual variability is estimated on CL/F (53.1% CV) and V/F (54.8% CV); IIV on ka was rejected during model building. Residual error was reported as an additive-error model but the additive SD value is not reported anywhere in the paper – addSd is encoded as fixed(0) and the vignette Errata documents the omission (von Hentig & Loetsch 2009).
Sarilumab (Xu 2019) Two-compartment population PK model for sarilumab in adults with rheumatoid arthritis (Xu 2019), with first-order SC absorption and parallel linear plus Michaelis-Menten (target-mediated) elimination from the central compartment.
Sarilumab anc (Ma 2020) Indirect-response PopPK/PD model for absolute neutrophil count (ANC) following subcutaneous sarilumab in adults with rheumatoid arthritis (Ma 2020). Sarilumab concentrations drive stimulation of ANC elimination (margination); PK backbone is Xu 2019.
Sarilumab das28crp (Ma 2020) Indirect-response PK/PD model of sarilumab on the 28-joint disease activity score by C-reactive protein (DAS28-CRP) in adults with rheumatoid arthritis (Ma 2020). Sarilumab inhibits the DAS28-CRP production rate (kin) via a sigmoid emax function that includes a background DMARD placebo component (PLB). The PK driver is the two-compartment, parallel linear + Michaelis-Menten model of Xu 2019 evaluated at its typical covariate-reference values (adult female, 71 kg, ADA-negative, commercial drug product, ALBR = 0.78, CrCl = 100 mL/min/1.73 m^2, baseline CRP = 14.2 mg/L).
Saxagliptin rat (Wang 2024) Preclinical (rat). Two-compartment PK model with first-order absorption for orally (intragastric) administered saxagliptin (SAX), directly linked to a sigmoidal Emax model WITHOUT a baseline term for the parent-drug contribution to plasma dipeptidyl peptidase-4 (DPP-4) inhibition, in streptozotocin plus high-fat-diet induced type 2 diabetic male Sprague-Dawley rats. Wang 2024 Eq 1-3 write the disposition as dA1/dt = FKaAa + (CL2/V2)A2 - (CL2/V)A1 - (CL/V)A1 and dA2/dt = (CL2/V)A1 - (CL2/V2)A2, i.e. CL2 is a conventional inter-compartmental clearance (k12 = CL2/V, k21 = CL2/V2) and is encoded here as lq. The PD model is Wang 2024 Eq 8, E = Emax C^Gam / (EC50^Gam + C^Gam), applied directly to the plasma SAX concentration with no effect compartment; the paper states explicitly that no hysteresis was observed. IMPORTANT: the fitted DPP-4 inhibition ratio for the parent drug is a DERIVED observation, not a directly measured one. Wang 2024 substituted the measured 5-OH SAX concentrations obtained after intragastric SAX into the separately fitted metabolite PK/PD model (see modellib(‘Wang_2024_5hydroxysaxagliptin_rat’)) and subtracted the resulting metabolite inhibition from the measured total inhibition: In_SAX = In_(SAX+5OHSAX) - In_5OHSAX. The resulting EC50 ratio EC50_5OHSAX = 0.46 * EC50_SAX is the paper’s central finding, confirming in vivo the roughly two-fold weaker potency of the metabolite reported in vitro. All volumes and clearances are body-weight normalised (mL/kg, mL/h/kg), so ODE state amounts are in ng per kg body weight and the 10 mg/kg dose is supplied as amt = 1e7. Parameters are the arithmetic means of three individually fitted rats (WinNonlin 8.1, naive individual fitting); no population model and therefore no OMEGA was estimated, so the model carries no eta terms and is intended for typical-value simulation. The PK residual error was described as additive but its magnitude was never reported, so addSd is FIXED at 0; the PD residual SD (stdev0) IS reported and is used.
Scopolamine (AlvarezJimenez 2016) Two-compartment linear intravenous PK model for scopolamine hydrobromide in healthy adults (18-78 years), coupled with 13 indirect-response EMAX PD models describing neurophysiological, psychomotor, and cognitive tests: 0-back reaction time, saccadic inaccuracy, saccadic peak velocity, adaptive tracker performance, EEG alpha/delta/theta band power in Fz-Cz and Pz-Oz leads, and N-back 0/1/2 correct-answer percentages (log-odds). Age and body weight modify clearance; body weight modifies peripheral volume; age modifies kIN for 0-back RT and delta/theta EEG baselines and EC50 for saccadic peak velocity.
Scopolamine (LiemMoolenaar 2011) Two-compartment population PK model for IV scopolamine in healthy adults (Liem-Moolenaar 2011, Table 2) with ten parallel effect-compartment linear-concentration-effect PK/PD models for central-nervous-system endpoints (Table 3): heart rate, saccadic peak velocity, adaptive tracking, VAS external perception, body sway, VAS alertness, VAS internal perception, smooth pursuit, VAS feeling high, and finger tapping (with an additive time-trend on finger tapping). PK was fit to 90 healthy male volunteers given a single 0.5 mg scopolamine i.v. infusion over 15 minutes; the ten PD endpoints were fit independently as effect-compartment linear-slope models on the empirical-Bayes individual PK profiles. PD parameter sets are grouped in Table 3 by equilibration half-life (heart rate <0.5 h; saccadic peak velocity and adaptive tracking 1-1.5 h; VAS external, body sway, VAS alertness, VAS internal, smooth pursuit 2.5-3.5 h; VAS feeling high and finger tapping >8 h).
Selexipag (Krause 2017) Joint two-compartment parent + two-compartment metabolite population PK model for oral selexipag and its active metabolite ACT-333679 in adults with pulmonary arterial hypertension (Krause 2017, GRIPHON study). First-order absorption with a fixed 0.668 h absorption lag delivers selexipag into a two-compartment disposition with linear total clearance CL/F (apparent total clearance, of which the rate constant kmet describes the fraction converted to ACT-333679); the metabolite has its own two-compartment disposition with first-order elimination via km. Body weight (allometric on V_p/F and CL/F; on V_m/F), total bilirubin (power on CL/F), sex (multiplicative on km), and a four-level PAH-comedication categorical (naive / ERA only / PDE5 inhibitor only / ERA + PDE5 combined; multiplicative on km) were retained as statistically significant covariates.
Selumetinib (Patel 2017) Sequential two-compartment population PK model for oral selumetinib (AZD6244, ARRY-142886) and its active metabolite N-desmethyl-selumetinib in adults with advanced solid tumors pooled with children with recurrent low-grade glioma (Patel 2017). Selumetinib disposition uses sequential zero-order (release into the gut compartment over duration D1 with lag ALAG1) and first-order (rate Ka) absorption with bioavailability anchored at 1 under fasted conditions and reduced by an additive food-effect coefficient under fed conditions; D1 and ALAG1 carry additive food-effect coefficients. Body surface area (power on CL/F and Vc/F), age (power on Vc/F), and alanine aminotransferase (negative power on CL/F) modify selumetinib parameters; BSA (negative power) modifies the fraction metabolized to N-desmethyl-selumetinib. The metabolite is two-compartment with its central volume fixed equal to the parent central volume to resolve identifiability; metabolite clearance and intercompartmental clearance are apparent values.
Semaglutide (CarlssonPetri 2018) One-compartment population PK model for once-weekly subcutaneous semaglutide (GLP-1 receptor agonist) in adults with type 2 diabetes, pooled across five SUSTAIN phase III trials (Carlsson Petri 2018).
Semaglutide (Overgaard 2019) Two-compartment population PK model for subcutaneous semaglutide (GLP-1 receptor agonist) with first-order absorption and first-order elimination, pooled across nine clinical pharmacology trials in healthy volunteers and adults with type 2 diabetes (Overgaard 2019).
Semimechanistic antibiotic pd (Nielsen 2007) In vitro (Streptococcus pyogenes M12 NCTC P1800). Semimechanistic time-kill pharmacokinetic/pharmacodynamic model of five antibiotics (benzylpenicillin, cefuroxime, erythromycin, moxifloxacin, vancomycin) against S. pyogenes. The bacterial system has two states: a growing drug-susceptible population (bact_susceptible) and a resting drug-insusceptible population (bact_resting) that captures phenotypic persister-cell switching. Bacteria grow in the susceptible state at first-order rate kgrowth, die in both states at first-order rate kdeath, and transfer from susceptible to resting at rate kSR = (kgrowth - kdeath) * (bact_susceptible + bact_resting) / Bmax so the total population reaches Bmax at stationary phase (reverse transfer kRS is fixed to 0). Each drug is dosed into its own PK compartment (pen, cxm, ery, mxf, van; the compartment state IS the bath concentration in mg/L per the in-vitro convention). Drug decays first-order via degradation (kdeg fixed from stability experiments; nonzero only for benzylpenicillin and cefuroxime). A biophase (effect) compartment (pen_e, cxm_e, ery_e, mxf_e, van_e) equilibrates with the PK compartment at first-order rate ke and drives the killing effect through a sigmoidal Emax function DRUG = Emax * Ce^gamma / (Ce^gamma + EC50^gamma). DRUG adds to the natural death rate on susceptible bacteria only (paper equation 6). Multi-drug DRUG contributions sum; in monotherapy (as fitted) only one drug is active. The starting inoculum is treated as mix1 (all bacteria in the growing state) per the paper’s mixture-model dominant mode; parameters fmix1 (0.747) and fpers (0.0529) are retained as fixed documentation for the mix2 alternative starting condition.
Sepantronium (Aoyama 2012) One-compartment IV population PK model for sepantronium bromide (YM155), a small-molecule survivin suppressant administered as a 7-day continuous IV infusion every 21 days, with power-form covariate effects of creatinine clearance and alanine aminotransferase and proportional cancer-type effects (hormone-refractory prostate cancer and melanoma vs non-small cell lung cancer) on clearance, in adults with NSCLC, HRPC, or unresectable stage III/IV melanoma (Aoyama 2012)
Sertraline (Cooper 2015) One-compartment first-order absorption population PK model for sertraline in overdose (Cooper 2015). Apparent clearance is increased 1.92-fold in subjects who received single-dose activated charcoal; the model holds relative bioavailability F at 1 and a shifted lag time at 1 h, with between-subject variability on F, ts_lag, ka, Vc, and CL absorbing the overdose-specific dose-amount and dose-time uncertainty.
Sertraline (Zhang 2024) One-compartment first-order absorption population PK model for sertraline in Chinese psychiatric inpatients (Zhang 2024). Apparent oral clearance decreases linearly with age around the 22-year cohort median (CL/F = 76.1 * [1 - 0.0068 * (AGE - 22)] L/h); the absorption rate constant is held at 0.098 1/h taken from Li 2013 because the therapeutic-drug-monitoring dataset contained trough samples only and the absorption phase was not identifiable.
Setrobuvir (Canini 2018) Integrated PK / viral-kinetic (VK) model for setrobuvir, a non-nucleoside HCV NS5B polymerase inhibitor, in adults chronically infected with HCV genotype 1a or 1b (Canini 2018). PK is a two-compartment model with first-order absorption and an absorption lag time, parameterised as rate constants (ka, ke, k12, k21) with central volume Vc (Table 3, source rates in /h). PD is a sigmoid Emax inhibition of virion production by central-compartment concentration, with genotype-specific EC50 and Hill coefficient. VK is the standard Neumann-style HCV model reduced to two ODE states (productively infected cells and free virus) under the short-treatment assumption that uninfected target cells remain at their pretreatment steady state; the reduction is dI/dt = dcV - dI and dV/dt = (1 - e(t))I - cV (equivalent to the paper’s target-fixed form with production rate p normalised to 1, so I represents cV0 at baseline and its value is a virion-production-rate surrogate rather than an absolute infected-cell count). Genotype (HCV_GT1B binary) switches EC50, Hill, and viral clearance rate c between the GT1a and GT1b typical values and their independent IIVs. Data source: 77 subjects across 4 studies (three healthy-volunteer PK studies A/B/C and one Phase 1 HCV-infected patient PK+VK study D; treated patients received 200, 400, or 800 mg BID for 3 days). Suitable for simulating BID or ascending-dose setrobuvir regimens and the resulting biphasic HCV RNA decline over 3 to 14 days by genotype, and for reproducing the paper’s Figure 3 14-day projections. NOT suitable beyond the on-treatment window studied (no drug-resistant-variant emergence submodel, no long-term target-cell repopulation).
Sevoflurane (Shin 2014) Pharmacodynamic sigmoid Emax model for the probability of recovery of consciousness (ROC) vs end-tidal sevoflurane concentration (vol %) during emergence from general anesthesia in pediatric dental-surgery patients (Shin 2014). Mentality (intact vs severely mentally disabled, MENT_DISABLED) stratifies both the concentration at 50% probability of ROC (C50) and the Hill coefficient. NONMEM Bernoulli likelihood in the source paper; this implementation exposes the typical-value probability with a placeholder additive residual error (see vignette Assumptions and deviations).
Sglt qsp (Lu 2014) QSP. Mechanistic systems pharmacology model of renal glucose reabsorption by SGLT1 and SGLT2 along the proximal tubules in humans, with optional competitive inhibition by an SGLT2 inhibitor (calibrated to dapagliflozin; evaluated against canagliflozin). The proximal convoluted tubules (PCT) are divided into six sub-segments (PCT1-6, SGLT2-mediated reabsorption) and the proximal straight tubules into three (PST1-3, SGLT1-mediated). Filtrate drains into a urinary bladder. Plasma glucose (GLU, mmol/L) and plasma inhibitor (CINH, nmol/L) enter as time-varying regressors through glomerular filtration. Calibrated by hand-tuning in Berkeley Madonna v8.3.18 against the DeFronzo et al. (2013) urinary glucose excretion data; evaluated against Polidori et al. (2013), Mogensen (1971), and Wolf et al. (2009). 23 ODE states; no fitted IIV or residual error (typical-individual mechanism model fit to mean per-step data).
SHetA2 dog (Sharma 2018) Preclinical (beagle dog). Two-compartment PK model with a 7-compartment gastrointestinal (GI) transit absorption process for SHetA2 (a flexible heteroarotinoid anti-cancer / chemoprevention drug) in beagle dogs after intravenous (5 mg/kg) and oral (100, 400, 1500 mg/kg) administration. Drug transits through 7 serial GI segments (stomach G1 = depot, then transit1..transit6 = G2..G7) at a common transit rate kAT; absorption occurs only from G2 (transit1, rate kA) and G7 (transit6, rate kA2). Disposition (CL, V1, V2, CLD) is reported as absolute total values for a typical 6.4-11.2 kg dog, fit by naive-pooled simultaneous IV+oral least-squares (1/y^2 weighting) in Phoenix WinNonlin. No IIV was reported. F, kA, kA2, and kAT are all dose-dependent; this file encodes the 100 mg/kg parameter set (F=11.2%, kA=1.12/h, kA2=0.929/h, kAT=0.532/h) as the typical value because 100 mg/kg is the new NOAEL used to derive the first-in-human dose. Higher-dose parameter sets (400 and 1500 mg/kg) are documented in the vignette. Parameter values from Sharma 2018 Table 3.
SHetA2 human (Sharma 2018) Allometrically-scaled human (70 kg) projection. Two-compartment intravenous PK model for SHetA2 (a flexible heteroarotinoid anti-cancer / chemoprevention drug) with disposition parameters scaled from preclinical mouse / rat / dog data via simple allometry (CL = a * BW^b) on a log-log plot (R^2 = 0.91-0.99 across CL, V1, V2, CLD; Sharma 2018 Fig 5). Clearance uses the maximum-life-span-potential (MLP) correction to account for SHetA2’s hepatic metabolism (CL_MLP = 17.3 L/h vs 41.0 L/h by simple allometry). The model carries no parametric oral absorption – the source paper simulated the oral profile (Fig 6) externally by linking these disposition parameters to the Advanced Compartmental Absorption and Transit (ACAT) model in GastroPlus 9.5, because the preclinical kA values did not correlate across species and could not be projected to humans. The predicted human oral bioavailability was 18.8% (range 7.4-42%) at 10 mg/kg, very close to the maximum extent of absorption observed in preclinical species at doses <100 mg/kg (18.6%). Parameter values from Sharma 2018 Results (Allometric scaling) and Prediction of human pharmacokinetics.
SHetA2 mouse (Sharma 2018) Preclinical (mouse, CD2F1 female). Two-compartment PK model with first-order absorption for SHetA2 (a flexible heteroarotinoid anti-cancer / chemoprevention drug) in non-tumor-bearing CD2F1 female mice after intravenous (20 mg/kg) and oral (20, 60 mg/kg) administration. Disposition (CL, V1, V2, CLD) is reported as absolute total values for a typical 20-28 g mouse, fit by naive-pooled simultaneous IV+oral least-squares (1/y^2 weighting) in Phoenix WinNonlin. No IIV was reported by the authors. Bioavailability F is dose-dependent (17.7% at 20 mg/kg, 19.5% at 60 mg/kg) but kA is shared across doses. Parameter values from Sharma 2018 Table 3.
SHetA2 rat (Sharma 2018) Preclinical (rat, Crl:CD Sprague-Dawley). Two-compartment PK model with first-order absorption for SHetA2 (a flexible heteroarotinoid anti-cancer / chemoprevention drug) in Crl:CD (SD) rats after intravenous (5 mg/kg single dose) and oral (100, 500, 2000 mg/kg/day for 28 days) administration. Disposition (CL, V1, V2, CLD) is reported as absolute total values for a typical 260-347 g rat, fit by naive-pooled simultaneous IV+oral least-squares (1/y^2 weighting) in Phoenix WinNonlin. No IIV was reported. The absorption is slow with flip-flop kinetics at the higher oral doses; kA was estimated as a single value across doses (0.0755 1/h) while F is dose-dependent (1.03% at 100 mg/kg, 1.57% at 500 mg/kg, 0.560% at 2000 mg/kg). Parameter values from Sharma 2018 Table 3.
Sibutramine (Han 2015) Two-compartment population PK for the active mono-desmethyl metabolite M1 plus a one-compartment PK for the downstream di-desmethyl metabolite M2 of the appetite-suppressant prodrug sibutramine, combined with an asymptotic exposure-response weight-loss PD model in Korean obese adults with metabolic syndrome. Sibutramine is dosed orally and assumed to convert entirely to M1 during absorption; M1 is then metabolised entirely to M2 and M2 is the only elimination pathway. Drug effect inhibits the rate of weight gain via a sigmoid Emax function of the steady-state sum AUC of M1 and M2 (AUC_ss,sum, computed from the current daily dose and the individual M1 and M2 clearances). A constant placebo effect is acknowledged only in female subjects and scales with mean-normalised baseline BMI.
Sifalimumab (Narwal 2013) Two-compartment population PK model for sifalimumab (anti-IFN-alpha IgG1) in adult patients with systemic lupus erythematosus (Narwal 2013)
Sifalimumab (Zheng 2016) Two-compartment population PK model for sifalimumab (anti-IFN-alpha human IgG1 monoclonal antibody) in adults with systemic lupus erythematosus following repeat fixed intravenous doses (Zheng 2016).
Siltuximab (Cao 2013) Second-generation minimal physiologically-based PK (mPBPK) model for siltuximab in adults (Cao 2013 Model A; clearance from plasma)
Siltuximab (Nikanjam 2019) Two-compartment population PK model for siltuximab (anti-IL-6) in adults pooled across healthy volunteers and oncology cohorts including Castleman’s disease, smoldering multiple myeloma, and other tumor types (Nikanjam 2019)
Simvastatin (Jin 2014) Joint two-compartment population PK model for orally administered simvastatin (lactone parent) and its active metabolite simvastatin acid (open beta-hydroxyacid), describing atypical multiple-peak absorption via three parallel mixed zero-and-first-order absorption processes and non-equilibrium reversible interconversion between the two species (Jin 2014). The simvastatin lactone is delivered into three depot compartments with fractional bioavailabilities F1, F2, F3 (sum = 1) parameterised through two relative-bioavailability constants BA1 and BA2, each depot has its own first-order absorption rate constant Ka1, Ka2, Ka3, zero-order infusion duration D1, D2, D3, and absorption lag-times ALAG1 = 0, ALAG2, ALAG3, the lactone disposes via a 2-compartment system with apparent clearance CL and inter-compartmental clearance Q, the fraction FM of total parent CL is converted to simvastatin acid (V_acid central fixed at 1 L for identifiability), and a reverse clearance Q64 returns acid to the parent central compartment. Age, body weight, and height were tested as covariates and not retained in the final model. The source publication analysed data in molar units; this packaged model preserves that choice – doses are expressed in nmol and concentrations in nmol/L. The validation vignette demonstrates the standard milligram-to-nanomole conversion using the simvastatin lactone molecular weight.
Sirolimus (Golubovic 2019) Two-compartment population PK model for sirolimus in adult kidney transplant recipients on triple immunosuppressive therapy (sirolimus + mycophenolate mofetil + corticosteroids) developed from routine therapeutic-drug-monitoring trough data with the NONMEM informative-prior functionality (Golubovic 2019). Covariate effects on CL/F: aspartate aminotransferase greater than 37 IU/L as a binary indicator of elevated liver enzymes (-37 percent multiplicative effect via power form 0.63^AST_HIGH) and age as a linear-deviation effect on CL/F with reference age 44 years (coefficient -0.388 on AGE/44, reproducing the 49 percent CL/F decrease from age 16 to age 64 reported in the Discussion).
Sirolimus (Jiao 2009) One-compartment population PK model for oral sirolimus in Chinese adult de novo renal transplant recipients on triple immunosuppression with ciclosporin and corticosteroids (Jiao 2009). First-order absorption with ka fixed at the literature value 0.752 1/h. Covariate effects on apparent clearance: linear-deviation effects of total cholesterol and whole-blood ciclosporin trough concentration centred on the cohort medians, multiplicative power-form effects of concomitant silymarin and glycyrrhizin co-therapy in hepatically impaired patients, and a power-form effect of the current sirolimus daily dose centred at 2 mg. Apparent volume of distribution carries a linear-deviation effect of ciclosporin trough concentration.
Sirolimus (Wu 2012) Two-compartment population PK model for oral sirolimus with saturable Michaelis-Menten absorption in patients with advanced cancer (Wu 2012). Hematocrit power covariate on apparent oral clearance.
Sirukumab (Xu 2011) Two-compartment population PK model for sirukumab (anti-IL-6 human IgG1 kappa monoclonal antibody, CNTO 136) in healthy adults following a single intravenous infusion, with first-order elimination from the central compartment and allometric body-weight scaling (Xu 2011).
Sitafloxacin (Rodjun 2023) Population PK model for oral sitafloxacin as encoded by Rodjun 2023 for Monte Carlo probability-of-target-attainment simulation against carbapenem-, multidrug- and colistin-resistant Acinetobacter baumannii. One compartment with first-order absorption in the fasted state. Apparent clearance is entirely proportional to creatinine clearance (CL/F = 2.58 x CrCL, with CrCL in L/h) and apparent volume is proportional to body weight (V/F = 1.72 L/kg). Unbound concentration is returned as Ccu using the reported unbound fraction of 0.388, because the paper’s PK/PD target is the unbound AUC ratio fAUC/MIC > 30.
SKL10406 (Park 2014) Two-compartment first-order oral absorption population PK with effect-compartment Emax PK-PD model for striatal serotonin transporter (SERT) occupancy by SKL10406 (a triple monoamine reuptake inhibitor candidate) in healthy adult volunteers (Park 2014; EME variant, Table 3)
Sn38 organoid (Zhu 2023) In vitro (patient-derived colorectal-cancer tumour organoids, PDTOs). Exponential organoid-growth model with a sigmoidal Emax (Hill) killing term describing the 96 h concentration-viability relationship of SN-38, the active metabolite of irinotecan. A vehicle-control organoid and a treated organoid grow in parallel; the reported readout is cell viability, the ratio of treated to control organoid volume. This is the in vitro half of the Zhu 2023 in vitro-to-in vivo translation: the same Emax / EC50 / Hill parameters are carried into the human irinotecan minimal-PBPK PK/PD model, see modellib(‘Zhu_2023_irinotecan’).
Snake venom (Sanhajariya 2018) Exploratory population PK meta-analysis of snake venom in humans (Sanhajariya 2018): one-compartment model with zero-order input (duration D1 = 1 h, fixed) and first-order elimination, fit in NONMEM 7.2 to 218 timed venom concentrations from 145 snakebite patients pooled across 24 published case reports / series. Snake family (Elapidae vs Viperidae) modifies F1; Viperidae is the reference (F1 = 1, fixed). Authors describe the model as a preliminary prior for future snake-envenoming PK modelling; F1 also absorbs the large bite-to-bite variability in injected venom mass.
Sodium nitrite qsp (VegaVilla 2013) QSP. Mechanistic systems pharmacology model of the NO metabolome (nitrite, nitrate) and methemoglobin (MetHb) in healthy adults receiving a 48-hour intravenous infusion of sodium nitrite. Nine ODEs covering plasma/RBC/tissue nitrite and nitrate, MetHb, NO and methemoglobin reductase activity; nonlinear nitrite/nitrate renal clearance (linear slope), entero-salivary nitrate-to-nitrite recycling, and indirect-response stimulation of MetHb reductase. Time in minutes; amounts in umol; concentrations in umol/L.
Somatropin human (Thorsted 2016) Translational (allometrically-scaled rat-to-human) population PKPD model for recombinant human growth hormone (rhGH / somatropin) in growth-hormone-deficient adult males. Structural parameter values are derived from the Thorsted 2016 hypophysectomized-rat PKPD fit by allometric scaling to a 70 kg reference subject (Table 3 of the source paper): clearance terms (CL, Q) and Vmax with exponent 0.75; distribution volumes (Vc, Vp) with exponent 0.9 (the empirically-selected best-fit exponent for human i.v. data); first-order absorption rate constants (ka1, ka2) and kout with exponent -0.25; KM unscaled; Emax and EC50 unscaled. The s.c. absorption model is the corrected form (Table 3 / Figure 5): bioavailability of the ka2 path reduced from 0.833 (rat) to 0.500, and one transit compartment added to the ka1 path. The IGF-1 indirect response uses kin = kout * R0 with R0 fixed to 65 ng/mL (human population mean per Laursen 1996) and is driven directly by plasma rhGH (no effect-delay chain - the rat CPLAG chain is intentionally dropped for the human prediction). Bodyweight gain is not included in the human model. Variability is inherited from the rat PKPD fit; residual error is fixed at the values used for the human-simulation validation (Methods).
Somatropin rat (Thorsted 2016) Preclinical (hypophysectomized Sprague-Dawley rat). Mixed-effects PKPD model for recombinant human growth hormone (rhGH / somatropin) describing PK as a two-compartment model with parallel linear (CL) and Michaelis-Menten (Vmax, KM) elimination, parallel first-order subcutaneous absorption (ka1 direct path, ka2 delayed through one transit compartment, with bioavailabilities F1 and F2), an indirect response model for IGF-1 induction (stimulation of kin via a three-compartment effect-delay chain feeding an Emax/EC50 stimulation), and a linear bodyweight-gain model driven by IGF-1 above baseline. Reference rat body weight is 0.1 kg (100 g) and the allometric exponents (0.75 / 1.0) are fixed.
Sonidegib (Goel 2016) Two-compartment population PK model for sonidegib (LDE225) in healthy subjects and patients with advanced solid tumors with first-order absorption, lag time, linear elimination, and dose-dependent bioavailability (Goel 2016)
Sorafenib (Jain 2011) One-compartment population PK model for orally administered sorafenib in patients with solid tumours (Jain 2011). Absorption is described by an Erlang-style chain of four catenary GI transit compartments downstream of an upstream absorption depot, all linked by a single first-order rate constant ka (mean absorption transit time MAT = 5 / ka). Enterohepatic recirculation is modelled by routing a fraction Fent of the drug leaving the central compartment into a gallbladder reservoir, with periodic release back to the most distal transit compartment gated by a smooth Hill switch Ehc = tad^40 / (tad^40 + t’^40), where tad is the time since the most recent dose; release becomes essentially full once tad exceeds the gallbladder-emptying onset time t’. The irreversible elimination rate constant ke equals the biliary excretion rate constant kb (= CL/V) per the published assumption kb = ke. Body weight is the only retained covariate (allometric exponent fixed to 1 on V/F, reference weight 80 kg).
Sotatercept (AitOudhia 2024) Two-compartment population pharmacokinetic model for sotatercept in healthy post-menopausal women and patients with pulmonary arterial hypertension (Ait-Oudhia 2024). First-order subcutaneous absorption with logit-scale bioavailability (about 66%) and linear elimination from the central compartment; time-varying body weight enters as a power covariate on clearance (exponent 0.814) and central volume (exponent 1.02) with a 70 kg reference, and baseline serum albumin as a power covariate on clearance (exponent -0.849) with a 4.5 g/dL reference. Separate log-scale residual error magnitudes are used for healthy participants and for patients with PAH. Intravenous doses go to the central compartment and subcutaneous doses to the depot.
Sparsentan (Wada 2023) Two-compartment population PK model with lagged first-order absorption for oral sparsentan (a single-molecule dual endothelin / angiotensin II receptor antagonist, DEARA) pooled over nine phase I-III studies in 446 subjects: 236 healthy volunteers, 16 subjects with hepatic impairment, and 194 patients with primary or genetic focal segmental glomerulosclerosis (FSGS). Apparent oral clearance carries a first-order CYP3A auto-induction term that steps CL/F from 3.88 L/h on the first dosing day to 5.11 L/h thereafter (induction half-life fixed at 0.001 day, i.e. effectively instantaneous), multiplied by power effects of alkaline phosphatase and creatinine clearance and by log-additive effects of male sex and of moderate / strong CYP3A4 inhibitor coadministration. Relative bioavailability is a power function of dose clamped at 200 mg, producing less-than-dose-proportional exposure (Frel = 1.41, 1.00, 0.71 at 200, 400, 800 mg). Race shifts the apparent central volume, and formulation (whole tablet or crushed tablet against the capsule reference) shifts both the absorption rate constant and the absorption lag time. Residual error is combined proportional plus additive, with the additive SD fixed at the 2 ng/mL assay lower limit of quantitation.
Spectinamide 1599 mouse pbpk (Parmar 2023) Minimal PBPK (mPBPK; author-coded ODEs, fitted in Monolix 2021R1). Preclinical (BALB/c mouse, 20 g). Spectinamide 1599, an anti-tuberculosis spectinomycin analogue, after intravenous, subcutaneous, and intrapulmonary-aerosol dosing (Parmar 2023, Pharmaceutics). Venous and arterial blood plus five tissues (lung, spleen, liver, kidney, and a lumped ‘other’), each tissue split into a rapid-equilibrium extracellular pool (vascular + interstitial, in instantaneous equilibrium with blood through the blood-to-plasma ratio) and a slow cellular pool coupled by a first-order influx K(I->C) acting on the unbound fraction and a first-order back flux K(C->I). The lung sits in series between venous and arterial blood; the spleen drains portally into the liver; elimination is glomerular filtration (GFR x fu) acting on the kidney extracellular pool. Subcutaneous doses reach venous blood through depot with ka / F; intrapulmonary-aerosol doses reach an epithelial-lining-fluid compartment (elf) through depot2 with its own ka / F, and the ELF exchanges with the lung cellular and extracellular pools. Tissue concentrations (lung, spleen, liver, kidney) are simultaneous outputs alongside venous plasma Cc. Two mass-balance defects in the published equations are corrected here (venous outflow to the lung, and hepatic outflow of the portal spleen flow); see the vignette Errata.
Spectinamide 1599 rat pbpk (Parmar 2023) Minimal PBPK (mPBPK; author-coded ODEs, fitted in Monolix 2021R1). Preclinical (Sprague-Dawley rat, 225 g). Spectinamide 1599, an anti-tuberculosis spectinomycin analogue, after a single intravenous dose (Parmar 2023, Pharmaceutics). Rat-physiology counterpart of Parmar_2023_spectinamide_1599_mouse_pbpk: the same structure and the same jointly estimated distribution rate constants, with the Sprague-Dawley physiology and rat-specific blood-to-plasma ratio and unbound fraction substituted. Venous and arterial blood plus five tissues (lung, spleen, liver, kidney, and a lumped ‘other’), each split into a rapid-equilibrium extracellular pool (vascular + interstitial) and a slow cellular pool coupled by a first-order influx K(I->C) on the unbound fraction and a back flux K(C->I). Lung in series between venous and arterial blood, spleen draining portally into the liver, and elimination by glomerular filtration (GFR x fu) on the kidney extracellular pool. Intravenous only: the paper reports no subcutaneous or aerosol rat data and estimated Ka / F from mouse data alone, so no absorption depot is included. Two mass-balance defects in the published equations plus a flow-rounding inconsistency are corrected here; see the vignette Errata.
Spectinamide 1810 mouse (Wagh 2021) Preclinical (BALB/c mouse, Mycobacterium tuberculosis infection). Population PK + PK/PD model for subcutaneous spectinamide 1810 in a murine TB efficacy / dose-fractionation study. PK is a two-compartment first-order absorption model with all volumes and clearances expressed per kg body weight (mg/kg dosing, mg/L plasma); IIV is carried on CL/F only (13.2% CV in infected animals). PK/PD couples the plasma central concentration Cc to a hypothetical PAE (post-antibiotic effect) compartment that tracks Cc whenever Cc exceeds the PAE concentration and otherwise decays first-order at rate K_PAE; the PAE concentration drives bacterial killing through a sigmoidal Emax (K_kill_max, EC50, Hill g) on a one-population logistic-growth Mycobacterium tuberculosis model (K_gs net growth rate, log10 N_max carrying capacity, log10 baseline CFU at aerosol infection time). A binary STUDY_WAGH_2 covariate switches K_kill_max from the study 1 typical value to the study 2 value via a 1.15 multiplicative factor; study-specific log10 CFU residual SDs are exposed as parameters.
Spectinamide 1810 mouse pbpk (Parmar 2023) Minimal PBPK (mPBPK; author-coded ODEs, fitted in Monolix 2021R1). Preclinical (BALB/c mouse, 20 g). Spectinamide 1810, a structurally close analogue of spectinamide 1599, after intravenous and subcutaneous dosing (Parmar 2023, Pharmaceutics). Same structure as Parmar_2023_spectinamide_1599_mouse_pbpk with the drug-specific blood-to-plasma ratio and unbound fraction updated and the distribution rate constants, Ka, F and residual errors re-estimated for 1810 (Table 7). Venous and arterial blood plus five tissues (lung, spleen, liver, kidney, and a lumped ‘other’), each split into a rapid-equilibrium extracellular pool (vascular + interstitial) and a slow cellular pool coupled by a first-order influx K(I->C) on the unbound fraction and a back flux K(C->I). Lung in series between venous and arterial blood, spleen draining portally into the liver, and elimination by glomerular filtration (GFR x fu) on the kidney extracellular pool. Subcutaneous doses reach venous blood through depot with ka / F. No intrapulmonary-aerosol route: the paper studied 1810 by the intravenous and subcutaneous routes only. Two mass-balance defects in the published equations plus a flow-rounding inconsistency are corrected here, and Table 7’s three intravenous tissue residual errors are unrecoverable transcription duplicates so they are fixed at zero; see the vignette Errata.
Spectinamide 1810 rat pbpk (Parmar 2023) Minimal PBPK (mPBPK; author-coded ODEs, fitted in Monolix 2021R1). Preclinical (Sprague-Dawley rat, 225 g). Spectinamide 1810 after a single intravenous dose (Parmar 2023, Pharmaceutics). Rat-physiology counterpart of Parmar_2023_spectinamide_1810_mouse_pbpk: the same structure and the same distribution rate constants, with the Sprague-Dawley physiology and rat-specific blood-to-plasma ratio and unbound fraction substituted. Venous and arterial blood plus five tissues (lung, spleen, liver, kidney, and a lumped ‘other’), each split into a rapid-equilibrium extracellular pool (vascular + interstitial) and a slow cellular pool coupled by a first-order influx K(I->C) on the unbound fraction and a back flux K(C->I). Lung in series between venous and arterial blood, spleen draining portally into the liver, and elimination by glomerular filtration (GFR x fu) on the kidney extracellular pool. Intravenous only, and plasma is the only endpoint with a reported residual error: the rat study sampled plasma alone. Two mass-balance defects in the published equations plus a flow-rounding inconsistency are corrected here; see the vignette Errata.
SPI 62 (Wu 2023) Simultaneous population target-mediated drug disposition (TMDD) PK / PD model for the small-molecule 11-beta-hydroxysteroid dehydrogenase type 1 (HSD-1) inhibitor SPI-62 (formerly ASP3662) in healthy adults (Wu 2023). Oral absorption through a chain of four identical first-order transit steps (depot -> transit1 -> transit2 -> transit3 -> central, all governed by a single Ktr), two-compartment linear disposition (central + peripheral1) with clearance CL and distribution flow Q, and explicit second-order binding of SPI-62 in the central compartment to its pharmacological target HSD-1 (association rate constant Kon, dissociation rate constant Koff) to form a drug-target complex. The total target amount Rtotal is held constant – no target synthesis or degradation is included – so free target is Rtotal minus the complex amount. This saturable, high-affinity / low-capacity binding is what produces SPI-62’s striking nonlinear PK: extremely low plasma exposure after the first low doses (most of the dose is trapped on target), turning into linear PK with unusually high accumulation ratios once the target is saturated by repeated dosing. Hepatic HSD-1 activity, measured as the urinary ratio of (tetrahydrocortisol + allotetrahydrocortisol) to tetrahydrocortisone expressed as a percentage of each subject’s own baseline, is driven directly (no effect delay) by the central-compartment SPI-62 concentration through an inhibitory sigmoid Imax function with half-maximal inhibitory concentration IC50 and power coefficient gamma. Fit by simultaneous PK-PD estimation in NONMEM 7.4.3 (FOCEI, ADVAN13) to pooled data from the SPI-62 first-in-human single-ascending-dose (1-10 mg) and multiple-ascending-dose (0.2-2 mg once daily) phase 1 trials. All PK parameters are apparent (per unit bioavailability) because F is unknown. Exponential inter-individual variability on Vcentral, CL, Ktr, Koff, Rtotal, and IC50; proportional residual error on both the plasma concentration and the HSD-1 activity endpoint. No covariates were retained: age, sex, body weight, and race showed no meaningful relationship with any parameter, so the final model equals the base model. Amounts are carried internally in nmol so that A / V is directly in nmol/L (= nM), the units of Kon, Koff, and IC50; convert an mg dose to nmol by dividing by the SPI-62 molecular weight of 424.4 g/mol and multiplying by 1e6.
Spironolactone qsp (Meid 2024) QSP. Whole-body plasma-potassium homeostasis with renal handling along the nephron and aldosterone feedback, pharmacologically modulated by the mineralocorticoid-receptor antagonist spironolactone. Sixteen ODE states: a three-transit oral spironolactone absorption chain plus a spironolactone central concentration, a two-compartment canrenone (active metabolite) disposition, an extracellular and an intracellular potassium pool, luminal potassium amounts and principal-cell potassium concentrations in the distal convoluted tubule (DCT), connecting tubule / cortical collecting duct (CNT/CCD) segments, cumulative urinary potassium excretion, and a mineralocorticoid-receptor occupancy state. Potassium is filtered at the glomerulus, reabsorbed in the proximal tubule / loop of Henle in proportion to sodium reabsorption, secreted into DCT / CNT / CCD lumen by Goldman-Hodgkin-Katz passive flux coupled to Na/K-ATPase active basolateral uptake, and partly reabsorbed in the medullary collecting duct. Aldosterone rises exponentially with plasma potassium and falls with sodium intake, and scales luminal potassium permeability; canrenone inhibits that pathway through an Emax mineralocorticoid-receptor occupancy model. Five parameters carry inter-individual variability and were the ones the source estimated per patient by Bayesian maximum-a-posteriori updating from electronic health records: potassium intake, sodium intake, extracellular fluid volume, mineralocorticoid-receptor abundance, and a hyperaldosteronism effect. The structural model is inherited unchanged from Maddah and Hallow 2022; Meid 2024 Appendix C prints the complete operationalized rxode2 source that this file transcribes.
Sugammadex rocuronium (Kleijn 2011) Integrated population PK-PD model for sugammadex-mediated reversal of rocuronium-induced neuromuscular blockade (Kleijn 2011). Both sugammadex and rocuronium have two-compartment PK from IV bolus dosing into the central compartment; the sugammadex-rocuronium inclusion complex has its own two-compartment PK with parameters set equal to free sugammadex (Bom 2002 framework). Complex formation is dynamic with fixed equilibrium dissociation constant kd = 0.0559 uM and estimated dissociation rate k2 = 0.034 1/min (association k1 = k2 / kd = 0.61 1/(minuM)). The rocuronium central concentration drives an effect compartment via ke0 = 0.134 1/min; neuromuscular blockade (T4/T1 twitch ratio x 100) follows a sigmoid Emax form with Emax set equal to E0 so the readout decreases monotonically from baseline E0 ~ 104 toward 0 as the effect-compartment rocuronium concentration rises. Sugammadex-mediated reversal enters as an additional first-order elimination of rocuronium from the effect compartment driven by the central free-sugammadex concentration (ks = 0.033 1/(minuM)). Both plasma assays measured total drug (free + complex), so the Cc and Cc_roc outputs return total sugammadex and total rocuronium. Allometric scaling on all volumes (exponent 1), flows (0.75), and rate constants (-0.25) at reference WT = 70 kg; sugammadex CL is NOT allometrically scaled (creatinine-clearance covariate replaces size scaling). All units are molar inside the model (dose in umol, concentrations in uM); see vignette for mg-to-umol conversion with rocuronium MW = 529.78 g/mol and sugammadex MW = 2178.01 g/mol (octasodium salt).
Sugemalimab (Wang 2024) Two-compartment population PK model with sigmoidal-cl_hill_max time-varying clearance for intravenous sugemalimab (anti-PD-L1 IgG4) in adults with advanced solid tumours or lymphomas across nine Phase I-III trials (Wang 2024)
Sulbactam durlobactam (Cammarata 2024) Combined four-compartment (two compartments per drug) population PK model for the sulbactam-durlobactam combination, fitted simultaneously to 373 subjects and 5,188 plasma concentrations pooled from six Phase 1 studies, one Phase 2 cUTI study, and one Phase 3 study in Acinetobacter baumannii-calcoaceticus complex infections (Cammarata 2024). Total clearance of each drug is the sum of a renal arm scaled by baseline BSA-normalized creatinine clearance and a non-renal arm, split by a fixed fraction excreted renally; total CL carries a further proportional downward shift in subjects with CLcr < 30 mL/min/1.73 m^2. Body weight acts as a power function on CL and central volume of both drugs; infection type and East Asian region act as proportional shifts. The paper’s two sub-models are integrated into the same file: a time-varying hemodialysis gate multiplies total CL 6.24-fold (durlobactam) and 8.19-fold (sulbactam) while a session is running, and epithelial lining fluid concentrations are returned as instantaneous ratios of the plasma concentrations (37.2% and 53.3% of total drug). Durlobactam uses the unsuffixed canonical compartment / parameter set; sulbactam carries the sibling-drug suffix _sbt throughout.
Sulbactam qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for sulbactam (beta-lactamase inhibitor): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Sulfadoxine (Karunajeewa 2009) Population PK model for sulfadoxine (SDOX) and its primary N-acetylsulfadoxine (NASDOX) metabolite in 60 Papua New Guinean women (30 pregnant, second or third trimester; 30 age-matched nonpregnant controls) given a single oral 1,500 mg sulfadoxine / 75 mg pyrimethamine dose for intermittent presumptive treatment of malaria in pregnancy (Karunajeewa 2009). SDOX is described by first-order absorption (no lag) into a 2-compartment disposition with separate non-metabolic clearance CL/F (renal excretion) and metabolic formation clearance CLM/F that drains SDOX into a 1-compartment NASDOX disposition. NASDOX elimination clearance is fixed at 10 times the structural SDOX non-metabolic CL/F (rapid formation-rate-limited renal excretion of the metabolite, Bell 1985). Allometric scaling is applied to all apparent volumes (exponent 1) and all apparent clearances (exponent 0.75) at reference WT = 70 kg. Pregnancy is the only retained covariate, entering as an additive term on the structural SDOX non-metabolic CL/F (+0.0181 L/h/70 kg). The companion model for the co-administered pyrimethamine is shipped as ‘Karunajeewa_2009_pyrimethamine’ (separate NONMEM dataset, fit independently in the source publication).
SulfadoxinePyrimethamine (deKock 2017) Joint popPK model for the antimalarial fixed-dose combination of sulfadoxine (1500 mg) and pyrimethamine (75 mg) as intermittent preventive treatment during pregnancy (IPTp) and after delivery in 98 women from Mali, Mozambique, Sudan, and Zambia (de Kock 2017). Sulfadoxine has 2-compartment disposition with first-order absorption; pyrimethamine has 3-compartment disposition with first-order absorption. Apparent volumes and flow rates are allometrically scaled with total body weight (exponents 1 and 0.75 respectively, reference WT = 60 kg). Whole-blood predictions are derived from plasma predictions using hematocrit and an estimated RBC-to-plasma partition ratio per drug. Pregnancy effects on apparent CL differ by drug: sulfadoxine uses a sigmoidal time-after-delivery effect (asymptotic -75.7%, T50 = 6.35 weeks, gamma = 4.90), while pyrimethamine uses a step contrast (+21.2% postpartum). Pyrimethamine apparent CL is additionally -20.2% in the Mozambique site. Residual country-specific scaling on the observed whole-blood concentrations is fitted with Mali as the reference.
SulfadoxinePyrimethamine (Odongo 2015) Joint popPK model for the antimalarial fixed-dose combination of sulfadoxine (1500 mg) and pyrimethamine (75 mg) administered as a single oral dose for intermittent preventive treatment of malaria during pregnancy (IPTp) in 34 non-pregnant and 87 pregnant Ugandan women dosed in the second trimester, of whom 78 were redosed in the third trimester (Odongo 2015). Each drug is described by a two-compartment model with first-order absorption and an absorption lag time, with bioavailability fixed at 1. Covariates on apparent CL/F (additive in L/h): pregnancy status (both drugs), serum albumin (sulfadoxine only), and subject age (pyrimethamine only). Covariates on apparent central volume V2/F (exponential per-unit): gestational age at dose (both drugs) and body weight (pyrimethamine only). Inter-individual variability is log-normal and is not estimated on V2/F or V3/F for sulfadoxine, nor on Q/F for pyrimethamine, in line with the paper’s over-parameterisation control.
Sumatriptan (Lee 2015) One-compartment population PK model for oral sumatriptan in healthy Korean male volunteers (Lee 2015): two parallel absorption routes (first-order absorption with lag time, and a transit-compartment chain with the Savic 2007 analytical input form) into a single central compartment with linear elimination. Captures the multiple-peaks absorption phenomenon reported in oral sumatriptan.
Sunitinib (Ait-Oudhia 2016) Joint population PK/PD model for sunitinib and its equipotent active metabolite SU12662 in adults with advanced hepatocellular carcinoma (HCC) receiving 37.5 mg sunitinib PO QD. Parent drug and metabolite each follow a 2-compartment oral PK structure with first-order absorption; each oral sunitinib dose deposits Dose into the parent depot and fM * Dose (fM = 0.21 fixed, Houk 2009) into the SU12662 depot. The active free (unbound) drug concentration ACub = (1 - fb_D) * Cc + (1 - fb_M) * Cc_su12662 (fb_D = 0.9, fb_M = 0.95 fixed, free fractions 0.1 and 0.05) inhibits the zero-order production rate of plasma sVEGFR2, captured with an indirect-response model dsVEGFR2/dt = kin / (1 + alpha * INH) - kout * sVEGFR2 with INH = ACub / (kd + ACub) (kd = 4 ug/L fixed, Mendel 2003) and kin = R0 * kout. Tumor volume follows a first-order growth dTG/dt = kg * (1 - H(t)) * TG with kg derived from baseline tumor volume by kg = ln(2) / (114 * TG0^0.14) (Taouli 2005) and H(t) = Imax * dsVEGFR2 / (dsVEGFR2 + dIC50) with Imax = 1 fixed and dsVEGFR2 = R0 - sVEGFR2(t). The paper reports a significant covariate effect of the DCE-MRI volume-transfer constant Ktrans on dIC50 (power coefficient 2.12) but the cohort-median Ktrans required to centre that effect is not reported in the paper or supplements on disk; the effect is omitted from model() and documented in the vignette. A Cox-style time-to-tumor progression hazard h(t) = b0 * exp(b1 * dAUC24h_sVEGFR2) is described in the paper but evaluated post-simulation in the vignette, not encoded as an ODE.
Sunitinib (vanErp 2010) One-compartment population PK model for oral sunitinib in cancer patients with a mechanism-specific grapefruit-juice (GJ) drug-interaction module. Sunitinib is absorbed first-order (ka, tlag) into a single central compartment with linear elimination (CL/F, Vd/F). A paper-specific intestinal CYP3A4-activity state (baseline 1, recovery first-order with t1/2 = 23 h fixed from Greenblatt 2003) is fully depleted to 0 by each GJ ingestion event. The relative bioavailability is F = 1 + deltaF * (1 - cyp3a4), so simultaneous GJ + sunitinib intake gives F = 1.11 (deltaF = 0.11) and the GJ-induced increase in sunitinib exposure decays back to baseline with the CYP3A4 recovery half-life (8.9% at 7 h, 5.3% at 24 h, 1.3% at 72 h, 0.07% at 1 week after the last GJ dose). No covariates were retained in the final model. Eight metastatic-cancer patients (1 female / 7 male, age 41-78 years) on chronic sunitinib 25-50 mg once daily contributed 268 plasma concentrations.
Sunitinib (Yu 2015) Integrated semi-physiological population PK model for oral sunitinib and its equipotent active metabolite N-desethyl sunitinib (SU12662) in adult cancer patients (n = 70 across three studies). Sunitinib is absorbed first-order into a hypothetical hepatic enzyme compartment that sits algebraically in equilibrium with the sunitinib central compartment via hepatic blood flow Qh (fixed at 80 L/h for a 70 kg subject). Clearance CL of sunitinib acts at the enzyme site (Cliv = (ka * depot + Qh / Vc * central) / (Qh + CL)); fraction fm = 0.21 (fixed from Houk 2009) of the cleared sunitinib appears as SU12662 input into the metabolite central compartment, the rest is true (non-SU12662) sunitinib clearance. SU12662 follows a 2-compartment disposition with its own central and peripheral volumes and inter-compartmental clearance. Body-weight allometric scaling with fixed exponents 0.75 (clearance / flow: CL_sun, Qh, CL_SU12662, Qi_SU12662) and 1.0 (volumes: Vc_sun, Vc_SU12662, Vp_SU12662) is applied a priori with reference WT = 70 kg. IIV is modelled with a 4 x 4 OMEGA BLOCK on Vc_sun, Vc_SU12662, CL_SU12662, CL_sun with the paper’s reported correlations 0.48, 0.45, 0.53 and remaining off-diagonals fixed to zero. Per-study residual proportional error from Yu 2015 Table 2 is simplified to a single propSd / propSd_su12662 pair populated from the Study 1 sigma^2 estimates (the largest cohort: 50 patients, 703 samples); the smaller Study 2 + 3 residuals are noted in the vignette.
Sunitinib irinotecan mouse (Wilson 2015) Preclinical (mouse with HT-29 colorectal-cancer xenograft). Mechanistic tumor-growth PD model for the antiangiogenic agent sunitinib (reduces vascular carrying capacity) combined with the cytotoxic agent irinotecan (three-stage transit-cell-death chain following Simeoni et al. 2004) and an empirical interaction term (Wilson 2015 Equation 4) in which the irinotecan transit-death rate kC depends on the cumulative pre-irinotecan sunitinib exposure. Drug input is K-PD (no pharmacokinetic data; each oral sunitinib or 5-min IV irinotecan dose enters its drug-amount compartment with normalized magnitude 1).
Sunitinib OS (Hansson 2013) Parametric overall-survival (Weibull TTE) model in adults with imatinib-resistant gastrointestinal stromal tumours (GIST) on sunitinib. The hazard for death is a Weibull baseline (lam_haz, alfa_haz) modulated log-linearly by the model-predicted relative change in soluble VEGFR-3 (sVEGFR-3) from individual baseline and by observed baseline tumour size (sum of longest diameters, SLD). The sVEGFR-3 time course is simulated in-model as a one-compartment indirect-response turnover with simple-Imax inhibition of Kin driven by the per-cycle exposure summary auc = DOSE / CLI. A parallel Weibull censoring hazard (lam_cens, alfa_cens) is included so the model can drive prospective Kaplan-Meier simulations with censoring per the paper’s published procedure. The model has no PK ODE and consumes individual posthoc upstream-PD parameters (BAS_SVEGFR3, MRT_SVEGFR3, EC50_SVEGFR3) and posthoc upstream-PK clearance (CLI) plus observed baseline tumour size (TUMSZ, mm) as data covariates. No IIV reported in the source for the OS or censoring hazard parameters.
Tacrine (Holford 1992) Population pharmacodynamic disease-progression model for the cognitive subscale of the Alzheimer’s Disease Assessment Scale (ADAS-cog, 0-70 score) in patients with probable Alzheimer’s disease treated with tacrine. Linear disease progression (baseline S0 + alpha*time) with a tacrine effect on the location of the progression curve (effect compartment driven by IBW-normalised daily dose rate, no estimable PK clearance because the response is slow relative to the 2-hour tacrine plasma half-life) and a placebo effect with asymmetric onset / elimination / tolerance dynamics (placebo response builds up during treatment, dissipates after treatment ends, and develops tolerance during continued treatment). Estimated by Holford and Peace 1992 on 909 patients (5253 ADAS-cog observations) pooled from two clinical trials of identical design: US protocol 970-01 (n = 632) and French protocol 970-04 (n = 277). The French cohort takes multiplicative scale factors on baseline status (FS04 = 1.08), placebo potency (Fpp4 = 1.76), and placebo elimination half-time (Ft1/2el-p4 = 2.78). Inter-individual variability is correlated across baseline S0, progression rate alpha, and tacrine potency beta_a (block of three) with diagonal IIV on placebo potency beta_p; the time constants of the effect compartments are typical-value only. Residual error is proportional. NOTE: the lead Holford 1992 PNAS 89:11471-11475 ‘Results and validation’ paper supplies all parameter values but the exact ODE form of the placebo dynamics is described in the companion methodology paper (PNAS 89:11466-11470) which was not available on disk at extraction time; the ODE form here is the field-standard reconstruction (asymmetric on/off placebo compartment plus multiplicative tolerance) and is documented in the validation vignette’s Assumptions and deviations section.
Tacrolimus (AbdelJalil 2013) One-compartment population PK model for oral tacrolimus in paediatric liver transplant recipients (Abdel Jalil 2013), with first-order absorption (ka fixed at the literature value 4.5 1/h), an apparent volume of distribution fixed at the literature value 30 L/kg, allometric (WT/13.2 kg)^0.75 scaling on apparent clearance with the theory-based exponent fixed, multiplicative exponential effects of time post-transplantation (days) and CYP3A5*1 carrier status on CL/F, exponential (log-normal) inter-individual variability on CL/F, and proportional residual error.
Tacrolimus (Andrews 2017) Two-compartment population PK model with first-order absorption and an absorption lag time for twice-daily oral immediate-release tacrolimus (Prograft and Modigraf) in paediatric renal transplant recipients during the first 6 weeks post-transplantation (Andrews 2017). Apparent oral clearance CL/F and apparent inter-compartmental clearance Q/F scale allometrically with body weight at a fixed exponent of 0.75 referenced to a 70 kg adult; apparent central volume V1/F and apparent peripheral volume V2/F scale at a fixed exponent of 1.0; ka has no body-weight scaling. CL/F additionally varies with CYP3A5 expresser status (1.04 multiplier for 3/3 or unknown genotype, 1.98 multiplier for 1/1 or 1/3 carriers; pooled with unknown because Andrews 2017 explicitly groups 3/3 with unknown in the final equation), donor source (0.74 multiplier for living-donor recipients vs deceased-donor reference; equivalent to deceased-donor recipients having ~35% higher CL/F), eGFR (power exponent 0.19 centred at the cohort median 69 mL/min/1.73 m^2 of adapted-Schwartz eGFR), and a piecewise hematocrit effect (power exponent -0.44 centred at 0.3 L/L applied only when HCT < 0.3 L/L). Inter-individual variability is diagonal on ka, CL/F, V1/F, and V2/F. Residual error is a combined additive + proportional model with separate immunoassay and LC-MS/MS magnitudes selected by the per-sample IMMUNOASSAY indicator. Inter-occasion variability (IOV) on CL/F (18% CV) and V2/F (35% CV) reported by Andrews 2017 Table 2 is NOT encoded structurally here (per the Brooks 2021 tacrolimus precedent) – the source paper does not define an operational occasion column for the model-library use case; downstream users who want to simulate IOV can add an OCC indicator and a per-occasion eta in rxode2.
Tacrolimus (Antignac 2007) One-compartment population PK model for oral and intravenous tacrolimus in adult kidney transplant recipients (Antignac 2007). First-order absorption (ka fixed at 4.5 1/h from Jusko 1995), linear elimination, simultaneous fit of IV and oral data to estimate bioavailability. Clearance increases sigmoidally with days postoperation from a baseline CLmin (at POD = 0) to 2 * CLmin asymptotically, with half-recovery at TCL50 = 3.81 days and Hill exponent 2.54; clearance is multiplied by (1 + theta_PRD) when concomitant prednisone dose exceeds 25 mg/day. No covariates retained on V or F.
Tacrolimus (Benkali 2010) Two-compartment population PK model with Erlang-distributed transit absorption (3 transit compartments) for once-daily extended-release oral tacrolimus (Advagraf) in stable adult renal transplant recipients more than 6 months post-transplant who were switched from twice-daily ciclosporin (Benkali 2010), with a multiplicative CYP3A5*1-carrier (expresser) effect on apparent clearance and combined additive + proportional residual error.
Tacrolimus (Bergmann 2014) Two-compartment population PK model for oral tacrolimus in adult kidney transplant recipients (Bergmann 2014), with first-order absorption after a lag time, allometric (WT/70 kg)^0.75 scaling on apparent clearance, multiplicative CYP3A5*1-carrier effect on CL/F, linear hematocrit and post-transplant-day effects on CL/F, linear free prednisolone Cmax effect on V1/F, correlated inter-individual variability across V1/F, ka, and V2/F, and proportional residual error.
Tacrolimus (Brooks 2021) Two-compartment population pharmacokinetic model for IV continuous-infusion tacrolimus in pediatric and young adult patients undergoing allogeneic hematopoietic cell transplantation (Brooks 2021). Allometric weight scaling on all PK parameters with fixed theoretic exponents (0.75 on CL and Q, 1.0 on V and V2; reference weight 70 kg); a structural ratio Fact fixed at 2.0 links Q to CL and V2 to V; and a multiplicative azole-antifungal (voriconazole or posaconazole) factor of 0.8 on CL captures the CYP3A4/5 inhibitor co-treatment effect.
Tacrolimus (Chen 2017) One-compartment population PK model with first-order absorption and absorption lag for low-dose oral tacrolimus (FK506, Prograf 0.5 mg capsules) in Chinese adult and paediatric myasthenia-gravis (MG) patients (Chen 2017). The absorption parameters ka and tlag are fixed at values obtained from a supplementary dataset of healthy volunteers, because the sparse-trough MG dataset is not informative about the absorption phase. Apparent oral clearance CL/F (3.6 L/h typical) is modulated by hematocrit and blood urea nitrogen through a multiplicative power-of-covariate-ratio form referenced to cohort medians (HCT median 38.4 %, exponent 4.31; BUN median 4.2 mmol/L, exponent 1.42). Apparent volume V/F is 1700 L typical with no retained covariate effects (high-dose IV immunoglobulin treatment was tested as a covariate on V/F but did not survive backward elimination). Inter-individual variability is diagonal on CL/F (141.6% CV) and V/F (72.4% CV); no IIV is estimated on ka or tlag. Residual variability is a pure proportional model (35.8% CV) on whole-blood tacrolimus concentrations.
Tacrolimus (CirrincioneDall 2011) One-compartment population PK model with first-order absorption for oral tacrolimus in pediatric liver transplant recipients (Cirrincione-Dall 2011 ACOP poster, Metrum Research Group). Apparent oral clearance CL/F (25.8 L/h at a 70 kg reference) and apparent volume V/F (2490 L at a 70 kg reference) are estimated; allometric body-weight scaling is fixed at exponent 0.75 on CL/F and 1.0 on V/F. The first-order absorption rate constant ka is fixed at 4.48 1/h from literature because the sparse therapeutic-drug-monitoring sampling could not identify it. CL/F additionally varies (full covariate model) with post-operative day as (POD/7)^0.409, with CYP3A5 expresser status as 1.24^CYP3A5_EXPR (missing genotype data imputed as non-expressers), with AST as (AST/510.5)^-0.0364, with albumin as (ALB/28)^-0.357, with hematocrit as 0.993^HCT (HCT entered as a fraction L/L, not as percent), and with age as (AGE/2)^-0.0310. Inter-individual random variation on CL/F and V/F was modeled exponentially with an estimated covariance of the two random effects per the poster text; the off-diagonal covariance value itself is not reported in the poster Table 2 so this implementation encodes uncorrelated diagonal IIVs and documents the gap in the vignette Errata. Residual error is a combined additive (SD 2.508 ng/mL) + proportional (SD 0.3674 fraction) model on whole-blood tacrolimus concentrations.
Tacrolimus (Dunlap 2025) Two-compartment population pharmacokinetic model for oral immediate-release tacrolimus in adult allogeneic hematopoietic cell transplant (allo-HCT) recipients (Dunlap 2025): first-order absorption with bioavailability fixed at 1; allometric (TBW/70 kg) scaling fixed at 0.75 on CL/F and Q/F and at 1 on V1/F and V2/F; exponential CYP3A5 intermediate / normal metabolizer phenotype effect on CL/F (CYP3A5 IM or NM have ~2.14-fold higher CL/F than CYP3A5 PM); exponential reduced-intensity-conditioning effect on CL/F (RIC recipients have ~37% lower CL/F than myeloablative-conditioning recipients); inter-individual variability on V1/F, CL/F, and V2/F; and an additive residual error of 2.51 ng/mL on the linear concentration scale.
Tacrolimus (Grover 2011) Two-compartment population PK model for oral tacrolimus in adult Native American kidney transplant recipients (Grover 2011), with first-order absorption after a lag time, no covariate effects (the Native American cohort showed no association of age, sex, weight, BMI, or post-transplant duration with PK parameters), and a placeholder proportional residual error model (residual error was not reported in the short communication).
Tacrolimus (Hao 2018) One-compartment population PK model with first-order absorption (no lag) and first-order elimination for twice-daily oral immediate-release tacrolimus (Prograf) in paediatric nephrotic-syndrome patients aged 2.7-17.3 years (Hao 2018). Apparent oral clearance CL/F scales allometrically with body weight at a fixed exponent of 0.75 referenced to a 70 kg adult; apparent volume of distribution V/F scales linearly with body weight at a fixed exponent of 1.0 referenced to 70 kg; ka has no body-weight scaling. CL/F additionally varies with CYP3A5 expresser status (multiplicative factor 1.60 for 1/1 or 1/3 carriers vs the 3/3 nonexpresser reference). Inter-individual variability is diagonal on ka, V/F, and CL/F (exponential / log-normal model). Residual unexplained variability is proportional (paper text: ‘The proportional model best described residual variability’; Table 2 reports it under the ‘Residual variability (exponential)’ label, which is the standard NONMEM additive-on-log-scale parameterisation equivalent to proportional in linear space).
Tacrolimus (JacoboCabral 2015) Two-compartment population PK model for oral tacrolimus in Mexican paediatric renal-transplant recipients (Jacobo-Cabral 2015): first-order absorption with a lag time, no allometric scaling, three-level CYP3A5 genotype effect on apparent oral clearance (3/3 reference, 1/3 +50%, 1/1 +93%), formulation-type effects on Ka and on relative bioavailability F (pooled Prograf + Framebin + Tenacrine reference vs Limustin generic vs unrecorded), an exponential per-dose effect on F centred at 2 mg, exponential inter-patient variability on Ka, V/F and F, and a residual error described in the paper as additive on the natural-log concentration scale (encoded as proportional residual error in linear space, which is the standard nlmixr2 equivalent for SD <= 0.15).
Tacrolimus (Ji 2018) One-compartment population pharmacokinetic model for oral tacrolimus in Korean adult living-donor liver-transplant recipients during the first 14 days post-transplantation (Ji 2018). First-order absorption with ka fixed at 4.48 1/h from prior reports; CL/F = 6.33 * POD^0.257 multiplied by a combinational CYP3A5 recipient-and-donor categorical factor (2.314 if both recipient and donor are CYP3A5 expressers; 1.523 if the recipient is a CYP3A5 expresser and the donor is a nonexpresser; 1.0 otherwise); V/F = 465 * POD^0.322; exponential IIV on CL/F and V/F; combined proportional + additive residual error on whole-blood tacrolimus concentration.
Tacrolimus (Kassir 2014) Two-compartment population PK model with first-order absorption and an absorption lag time for twice-daily oral tacrolimus in paediatric liver transplant recipients (Kassir 2014). Apparent oral clearance CL/F and apparent inter-compartmental clearance Q2/F scale allometrically with body weight at a fixed exponent of 0.75 referenced to the cohort median weight of 20 kg; apparent central volume V1/F and apparent peripheral volume V2/F scale at a fixed exponent of 1.0 to the same 20 kg reference; the first-order absorption rate constant ka carries an allometric exponent of -0.25 per Anderson and Holford theory. Apparent peripheral volume V2/F was fixed to 290 L during estimation to stabilise the model (Kassir 2014 Table 4 footnote). Inter-individual variability is diagonal on CL/F, V1/F, and Q2/F (no IIV on ka, tlag, or V2/F). Residual error is a proportional model. No covariates beyond body weight were retained after stepwise covariate analysis – age, sex, type of transplant, age of liver donor, time post-transplantation, liver function tests, albumin, renal function (serum creatinine and creatinine clearance), haematocrit, use of steroids, presence of clinically relevant CYP3A4 inhibitors, and drug formulation were all screened and dropped (Kassir 2014 Results ‘Analysis of covariates and sources of variability’).
Tacrolimus (Kim 2018) Integrated population PK model of the tacrolimus (TAC) - mycophenolate mofetil (MMF) drug-drug interaction in healthy Korean male volunteers (Kim 2018, final integrated model). TAC follows a two-compartment model with first-order absorption and a lag time; apparent oral clearance (CL/F) is increased 1.48-fold in CYP3A5 expressers and is suppressed by co-administered mycophenolic acid (MPA) through an inverse-exponential interaction (CL/F = 13.8 / exp(0.0294Cmpa) 1.48^CYP3A5). MPA (the active moiety of MMF) follows a two-compartment model with first-order absorption; MPA is metabolised to MPAG (7-O-glucuronide; 85% of metabolism) and AcMPAG (acyl glucuronide; 15%). MPAG undergoes enterohepatic recirculation via a gallbladder compartment that empties into the MPA absorption compartment during a meal-triggered window. Tacrolimus concentrations are in ng/mL; MPA, MPAG and AcMPAG are in ug/mL.
Tacrolimus (Kirubakaran 2022) Two-compartment population pharmacokinetic model for oral immediate-release tacrolimus (Prograf) in adult heart transplant recipients (Kirubakaran 2022): first-order absorption; FFM-allometric scaling on CL/F and Q/F (exponent 0.75) and on V2/F and V3/F (exponent 1.0); haematocrit power effect on CL/F; and a state-dependent typical CL/F (without vs with concomitant azole antifungal, 21.1 vs 4.2 L/h) with a state-dependent CL/F BSV magnitude (61% vs 89.5% CV). Structural PK was estimated with NONMEM PRIOR (NWPRI) support from the published Sikma 2017 thoracic-transplant tacrolimus popPK model.
Tacrolimus (Lu 2015) Two-compartment population PK model with first-order absorption and lag time for oral tacrolimus in pooled Chinese healthy volunteers and adult orthotopic liver-transplant recipients (Lu 2015). Apparent peripheral volume V3/F is fixed at the healthy-volunteer-only estimate (916 L). Apparent clearance CL/F is reduced multiplicatively in liver-transplant recipients and further modulated by an exponential serum ALT effect that applies only to the transplant cohort.
Tacrolimus (Moes 2016) Two-compartment population pharmacokinetic model for oral once-daily tacrolimus (Advagraf) in stable adult liver transplant recipients (Moes 2016), with first-order elimination from the central compartment and a delayed first-order absorption phase described by three sequential transit compartments sharing the absorption rate constant ka, a fixed oral bioavailability F = 0.23, a categorical donor + recipient CYP3A53 combination effect on apparent oral clearance (reference both nonexpressers; donor nonexpresser + recipient 1 carrier +33%; donor 1 carrier + recipient nonexpresser +33%; both 1 carriers +71%), independent log-normal IIV on CL, Vc, and ka, and proportional residual error on whole-blood concentration.
Tacrolimus (MohammedAli 2023) Two-compartment population PK model for once-daily extended-release LCP-Tac tacrolimus (MeltDose technology, Envarsus) in stable adult renal transplant recipients (Mohammed Ali 2023), parameterized in apparent elimination clearance CL/F, apparent distributional clearance CLD/F and apparent central and peripheral volumes Vc/F and Vp/F. Delayed absorption is described by a Savic 2007 transit-compartment chain with the number of transit compartments fixed at NN = 2 (mean transit time MTT = 2.91 h) feeding a first-order absorption step ka into the central compartment. A combined CYP3A4/CYP3A5 cluster phenotype (high, intermediate and poor metabolizer, reconstructed inside model() from the recipient CYP3A5 expresser status and the CYP3A4*22 rs35599367 carrier indicator) gives three distinct typical CL/F values. Inter-individual variability is carried on CL/F, Vc/F, Vp/F and MTT, with inter-occasion variability on CL/F and a proportional residual error on whole-blood concentrations.
Tacrolimus (Passey 2011) Steady-state apparent-clearance regression model for oral tacrolimus trough concentrations in adult kidney-transplant recipients (Passey 2011). Encoded as a 1-compartment IV continuous-infusion model with a nominal fixed central volume of distribution: at steady state, Cc = dose-rate / CL/F is independent of V, so the rxode2 simulation reproduces the paper’s regression-style trough prediction. Apparent clearance CL/F is multiplied by five covariate factors: an ordered-categorical days-post-transplant effect (3-5 = reference, 6-10 = 0.86, 11-180 = 0.71), three-level CYP3A5 genotype (CYP3A53/3 = reference, CYP3A51/3 = 1.70, CYP3A51/1 = 2.00), steroid-sparing immunosuppression protocol (0.70), a power-form age effect ((Age/50)^-0.40), and concomitant calcium channel blocker coadministration (0.94).
Tacrolimus (Pei 2023) One-compartment population PK model with first-order absorption and elimination for oral tacrolimus in adult heart transplant recipients (Pei 2023 Pharmaceutics, Phoenix NLME 8.3). Apparent oral clearance CL/F carries four covariate effects: a power effect of total bilirubin, an exponential effect of concomitant voriconazole, an exponential three-level CYP3A53 (rs776746) genotype effect with the 3/*3 nonexpresser as reference, and an exponential IL-10 G-1082A (rs1800896) heterozygote effect. Absorption rate constant fixed at 0.30 1/h; no covariate was retained on Vd/F. Exponential IIV on both CL/F and Vd/F, proportional residual error. This is the companion top-down model to the whole-body PBPK model of the same paper (Pei_2023_tacrolimus_pbpk.R); the two identify different covariates, which the paper’s Discussion addresses explicitly.
Tacrolimus (Prytula 2016) Two-compartment population PK model with first-order absorption and a fixed absorption lag time for twice-daily oral tacrolimus (Prograft) in stable paediatric renal transplant recipients at least one year after kidney transplantation (Prytula 2016). All apparent-PK parameters (CL/F, Q/F, V1/F, V2/F, ka) scale allometrically with body weight at fixed exponents (0.75 on CL/F and Q/F, 1 on V1/F and V2/F, -0.25 on ka) referenced to a 70 kg adult; V2/F is fixed at 1090 L/70 kg during covariate analysis; CL/F additionally varies with CYP3A51 carrier status (1+0.45-fold higher in carriers vs 3/3 nonexpressers), gamma-glutamyltransferase (power -0.21, centred at 13 U/L), and haematocrit (power -0.59, centred at 0.34); eta_Q is perfectly correlated with eta_CL and is constructed as iiv_q_scale etalcl (iiv_q_scale = 2.0; the ‘IIV-CL-Q’ parameter in Table 2); inter-individual variability is a 3x3 correlated block on (ka, CL/F, V1/F); proportional residual error.
Tacrolimus (Rower 2017) One-compartment population pharmacokinetic model for oral / enteral tacrolimus in paediatric heart transplant recipients (Rower 2017): first-order absorption with fixed Ka = 3.43 1/h; AGE power effect on apparent volume with exponent 0.775 and reference 5.7 years; creatinine-clearance power effect on apparent elimination rate with exponent 0.850 and reference 122.4 mL/min/1.73 m^2; concomitant fluconazole reduces apparent elimination by 34%. Originally parameterised in NONMEM ADVAN2 TRANS1 on (ke, V); converted here to the canonical (CL/F, V/F) form via CL/F = ke * V, so the AGE effect propagates to CL/F with the same exponent as on V/F.
Tacrolimus (Storset 2014) Theory-based two-compartment population pharmacokinetic model for oral tacrolimus in adult kidney-transplant recipients (Storset 2014): plasma-based disposition with first-order absorption and a lag time, allometric scaling on fat-free mass, CYP3A5-expresser effects on plasma clearance and oral bioavailability, a sigmoid-Emax prednisolone-driven reduction in bioavailability, a first-day-post-transplant bioavailability spike with subject-level random effect, and a saturable haematocrit-dependent red-blood-cell-binding equation that maps plasma concentration to whole-blood concentration.
Tacrolimus (Wang 2024) One-compartment population PK model with first-order absorption for twice-daily oral tacrolimus (Prograf) trough concentrations in Chinese adult renal transplant recipients (Wang 2024). The absorption rate constant ka is held at 3.86 1/h because only pre-dose trough concentrations (C0) were available. Apparent oral clearance CL/F carries two exponential covariate effects: the CYP3A5 3-allele count (0 for 1/1, 1 for 1/3, 2 for 3/3; coefficient -0.348, so CL/F falls to 70.6% and 49.9% of the 1/1 value for 1/3 and 3/*3 respectively) and hematocrit expressed as a volume fraction (coefficient -0.122). Apparent central volume V/F carries no covariate. Inter-individual variability is exponential and diagonal on CL/F and V/F. Residual error is additive on log-transformed concentrations, which is proportional error on the linear concentration scale. Wang 2024 also builds MLP, SVR, and XGBoost machine-learning predictors on top of this model’s post hoc individual predictions; those are not ODE models and are outside the scope of this model file.
Tacrolimus (Woillard 2011) Two-compartment population PK model with Erlang-distributed transit absorption (3 transit compartments) for oral tacrolimus in adult renal transplant recipients pooled across the twice-daily immediate-release Prograf formulation and the once-daily prolonged-release Advagraf formulation (Woillard 2011), with multiplicative CYP3A5*1-carrier (expresser) and power-scaled haematocrit effects on apparent clearance, multiplicative formulation effects on the Erlang transit rate constant and on apparent central volume, and combined additive plus proportional residual error.
Tacrolimus (Zhu 2014) Two-compartment population PK model for oral tacrolimus in Chinese adult liver transplant recipients (Zhu 2014), with first-order absorption, a power-form joint DOSE x POD covariate effect on apparent clearance, log-normal IIV on CL/F, V2/F, Q/F, V3/F, and ka, and proportional residual error. Bioavailability was not estimated; the structural disposition parameters are apparent values (CL/F, V/F, Q/F).
Tacrolimus industry meta (Lu 2019) Industry meta-analysis. Two-compartment population PK model for oral tacrolimus immediate-release (IR-T; Prograf, twice daily) and prolonged-release (PR-T; Advagraf / Astagraf XL, once daily) formulations in adult and paediatric liver, kidney, and heart transplant recipients (Lu 2019). Pooled individual-patient data from 8 Astellas Phase II studies (n = 408 patients, 23,176 whole-blood concentration records). Structural model: first-order absorption with formulation-dependent Ka (PR-T ~50% slower than IR-T), fixed absorption lag time, and two-compartment disposition with first-order elimination. Significant covariates: Asian race on CL/F (+59% vs Whites); log-AST on CL/F, Vc/F, Vp/F, and F1 (power normalised at LAST = 3.15, i.e., AST ~= 23.3 IU/L); female sex on Vc/F (-44.6% vs males); albumin on Vc/F and F1; and Asian / Black race on F1 (Asians > Whites > Blacks). Type of organ transplanted and adult-vs-paediatric population had no significant effect on PK parameters.
Tacrolimus metaanalysis (Nanga 2019) MBMA. Two-compartment population PK meta-model for oral tacrolimus in solid organ transplantation (Nanga 2019), built from pooled individual-patient data across 7 historical NONMEM datasets (n = 281 paediatric + adult liver and kidney transplant recipients). Structural model: first-order absorption with fixed lag time, time-varying first-order elimination, allometric (WT/50 kg) scaling on apparent clearance and apparent central volume, multiplicative reduction of CL/F in hepatic-graft recipients, sigmoidal post-transplant-day recovery of CL/F, and reduced relative bioavailability for the oral syrup formulation. The literature-review summary table (Nanga 2019 Table 2: 76 published popPK models) is not used for parameter fitting and is not reproduced here.
Tacrolimus pbpk (Pei 2023) PBPK (whole-body, SimBiology 5.8.2). Perfusion-limited 15-compartment physiologically based model for oral tacrolimus in adult heart transplant recipients (Pei 2023 Pharmaceutics). Thirteen well-stirred flow-limited tissues (gut, spleen, pancreas, liver, muscle, kidney, brain, heart, lung, skin, tendon, other, adipose) plus arterial and venous blood, on the Levitt PKQuest standard-human physiology (organ weights and per-kg perfusions from Table S1, 70 kg / 25 percent body fat reference). Gut, spleen and pancreas drain into the liver via the portal vein; the liver is the only eliminating organ. Rodgers and Rowland tissue-to-plasma partition coefficients (Table S2) are converted to tissue-to-blood by the blood-to-plasma ratio BPR and multiplied by a Kp scaling factor. BPR is hematocrit-dependent through the red-cell binding capacity Bmax and affinity constant KD (Eq 5). Hepatic blood clearance follows the well-stirred extraction ratio (Eq 3) scaled by the CYP3A5 / CYP3A4 metabolic fractions and genotype-specific activity levels (Eq 6), with an optional reversible voriconazole inhibition term (supplement Eq 1-2) driven by a user-supplied voriconazole whole-blood concentration. Absorption is first-order from a depot into the gut compartment with a fixed absorbed fraction Fg = 0.2 (Eq 2). Ka, KD, Bmax and CLint are the four fitted parameters (Table 3, heart-transplant model-building column); no IIV or residual-error variance was reported. Observation is venous whole blood. NOTE: the paper’s printed Kp scaling factor of 350 is refuted by the paper’s own Tables S5 and 6 by 4-11 fold; the model uses 9.15 = 11.9 / 1.3, the mouse-to-human average-Kp ratio the Methods sentence motivates. See the vignette Errata.
Tacrolimus thoracic (Sikma 2020) Two-compartment population pharmacokinetic model for oral whole-blood tacrolimus in 30 adult thoracic organ transplant recipients (10 heart, 20 lung) during the first 6 postoperative days at the University Medical Center Utrecht intensive care unit (Sikma 2020 EJDMP). Apparent clearance CL/F, apparent volumes V1/F and V2/F, inter-compartmental clearance Q/F, and first-order absorption rate ka are estimated; bioavailability F is fixed at 1. Only the inter-individual variability of CL/F was identifiable in the source dataset; all other IIV elements were not estimated. Inter-occasion (dose-to-dose) variability dominated the variance structure but is not encoded structurally in this extraction. No covariates were retained in the final model.
Tacrolimus unbound plasma (Sikma 2020) Two-compartment population PK model for whole-blood (Cc), unbound plasma (Cupc), and total plasma (Ctpc) tacrolimus in 30 adult thoracic-organ (10 heart + 20 lung) transplant recipients during the first 6 postoperative days (Sikma 2020). First-order oral absorption with ka, F, and the within-PK fixed-parameter variabilities inherited from a previously estimated tacrolimus model; non-linear saturable binding of tacrolimus to erythrocytes (UPC = WBC * Kd / (Bmax * HCT - WBC)) with the maximum erythrocyte binding capacity Bmax scaled by hematocrit, and a linear non-specific plasma binding constant Nplasma linking unbound to total plasma (TPC = Nplasma * UPC).
Tafenoquine (Charles 2007) One-compartment first-order-absorption population PK model for oral tafenoquine in adult Australian soldiers on weekly malaria prophylaxis (Charles 2007)
Tafenoquine (Edstein 2001) One-compartment population PK model for oral tafenoquine in 135 male Thai soldiers receiving 400 mg base for malaria prophylaxis (monthly n=104 or weekly n=31). The final model carries correlated IIV on apparent clearance and apparent volume of distribution (rho ~ 0.71) plus separate IIV on the first-order absorption rate constant; no covariates retained (centred age and weight on V/F and a prior-malaria indicator on CL/F were screened but not deemed to have sufficient clinical impact to alter the base model).
Tak 079 (Roepcke 2018) Preclinical (cynomolgus monkey). Two-compartment QSS-TMDD population PK model plus three PK-PD lymphocyte-depletion models (NK-cell turnover with Emax on elimination; B-cell four-transit-compartment chain with Emax on the circulating depletion rate; T-cell direct-response with Emax) for the cytolytic anti-CD38 human IgG1-kappa monoclonal antibody TAK-079 pooled from eight monkey studies (dose 0.03-100 mg/kg IV / SC). PK covariate: route of administration on Vc (SC vs IV).
Tamibarotene pediatric (Azechi 2024) Two-compartment population PK model with first-order absorption and an absorption lag time for oral tamibarotene (synthetic retinoid RAR-alpha/beta agonist) in pediatric and young-adult patients (4-23 years) with recurrent or refractory solid tumors (Azechi 2024). Apparent oral clearance CL/F, apparent central volume V1/F, and apparent peripheral volume V2/F scale linearly with body surface area (BSA) referenced to the cohort mean of 0.995 m^2 (Table 1); inter-compartmental clearance Q/F, the absorption rate constant ka, and the absorption lag time tlag have no covariate effects. tlag was held fixed at 0.95 h in the published final model (the authors judged the post-covariate Tlag estimate of ~1.8 h to have low physiological validity and fell back to the pre-covariate value). Residual error is proportional with a 42.4% magnitude. The Methods section specifies an exponential IIV model on all five PK parameters but the paper reports no per-parameter omega magnitudes and no supplement exists; per operator decision (sidecar request-001 q1 = A, 2026-06-21) the five eta terms are encoded as fixed(0) so the published structural IIV declaration is preserved while remaining faithful to the absence of reported variance values. See the vignette Assumptions and deviations section for the resulting limitations on VPC-style validation.
Tamsulosin (Tsuda 2010) One-compartment population PK model for oral modified-release tamsulosin hydrochloride in paediatric patients (2-16 years) with neuropathic and non-neuropathic bladder (Tsuda 2010), with first-order absorption after a lag time, allometric (WT/70)^0.75 on apparent clearance and (WT/70)^1 on apparent central volume (allometric exponents fixed at theory values), a power-form alpha-1-acid glycoprotein (AAG/20 uM) effect on both CL/F and V/F, correlated inter-individual variability on CL/F and V/F, independent IIV on ka, and a combined additive + proportional residual error.
Taranabant (Li 2010) Three-compartment population PK model for oral taranabant in healthy and obese adults (Li 2010)
Taspoglutide mbma (Li 2015) MBMA. Coupled PD model-based meta-analysis of taspoglutide (long-acting human glucagon-like peptide-1 analogue, once-weekly SC) net efficacy on fasting plasma glucose (FPG) and glycosylated hemoglobin (HbA1c) in type 2 diabetes. Each endpoint is the sum of an exponential-to-asymptote placebo response (Pmax, Kp) and a saturable Emax drug response (Dmax, IC50, Kdrug) approached exponentially over time. The FPG drug effect is driven by the study-arm-mean taspoglutide concentration between weeks 2 and 4 (Cavg; supplied as the METRIC_TASPO_C covariate: 0 / 59.85 / 119.7 pmol/L for placebo / 10 mg / 20 mg QW). The HbA1c drug effect is driven by the model-predicted drug-induced FPG change (i.e. the placebo-adjusted FPG response feeds the HbA1c Emax). Estimated on digitised study-arm-mean PD data from 8 published clinical trials of taspoglutide monotherapy or add-on therapy in type 2 diabetes (3,702 patients pooled, 8-52 week treatment durations); a ninth trial (Rosenstock 2013) was held out for external validation. Placebo Pmax and Kp were fitted on the placebo-only subset first and held fixed in the final combined PD model. Between-trial variability (ITV) is encoded as study-level etas (one eta per parameter); the model is suitable for simulating study-arm-mean PD outcomes and is NOT suitable for individual-subject simulation. Residual error is a proportional/power model on each endpoint (the small power-correction term is simplified to a plain proportional error in this implementation; see vignette Assumptions and deviations).
Taurine rat (Catalan-Latorre 2018) Preclinical (rat). Population PK model for taurine (2-aminoethylsulphonic acid) in male Wistar rats after IV bolus or oral gavage administration (1, 10, or 100 mg per animal). Two-compartment disposition (central and peripheral1) with zero-order endogenous formation Q0, first-order passive oral absorption ka, first-order inter-compartmental distribution (K12, K21), and non-linear renal elimination described as two parallel Michaelis-Menten processes: saturable tubular secretion (Vms, Kms) and saturable tubular reabsorption (Vmr, Kmr), with net elimination = secretion - reabsorption. Oral bioavailability was modelled as 100% (passive diffusion; not altered by nutritional status). Protein-energy undernutrition (MAL_NOURISH = 1) reduces the secretion Vmax by 9.4% relative to well-nourished animals; no other PK parameter depends on nutritional status. Initial conditions in the central and peripheral compartments are set from the analytic positive root of the no-dose steady-state quadratic so that the endogenous taurine concentration is reproduced at t = 0.
Tazobactam (CohenWolkowiez 2014) One-compartment population PK model for tazobactam in premature and term infants under 61 days postnatal age (Cohen-Wolkowiez 2014); linear body-weight scaling on CL and V (fixed exponent = 1), and PMA, serum creatinine and concomitant gentamicin coadministration as covariates on CL.
Tazobactam (Nichols 2016) One-compartment population PK model for tazobactam in critically ill children (1-9 years) receiving extended-infusion piperacillin-tazobactam (Nichols 2016); IV zero-order input, first-order elimination, a multiplicative female-sex effect on CL, and a linear-additive WT effect on CL centered at the cohort median 18 kg.
Tazobactam qsp (LopezArguello 2023) QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for tazobactam (beta-lactamase inhibitor): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0).
Tebipenem (Ganesan 2023) Two-compartment population PK model with two absorption transit compartments for tebipenem, the active moiety of the oral carbapenem pro-drug tebipenem pivoxil hydrobromide (TBP-PI-HBr), pooled across three phase 1 studies and the phase 3 ADAPT-PO trial in adults with complicated urinary tract infection / acute pyelonephritis (Ganesan 2023; 746 subjects, 3448 plasma concentrations). Apparent oral clearance is split into an additive non-renal arm (power function of creatinine clearance) and a renal arm (sigmoidal Hill function of creatinine clearance) that drives a cumulative urine compartment; the summed CL/F carries a linear body-surface-area effect. Central volume scales with height and peripheral volume with body surface area, both shifted by infection status. The absorption rate constant switches on fed status, is shifted by infection status, and carries a dose effect confined to the crossover thorough-QT study (study 104). Interindividual variability on CL/F is cohort-specific (healthy subjects vs infected patients) and Ka carries both IIV and two-occasion interoccasion variability.
Tefibazumab (Cao 2013) Second-generation minimal physiologically-based PK (mPBPK) model for tefibazumab in adults (Cao 2013 Model A; clearance from plasma)
Teicoplanin (Chen 2023) Two-compartment IV infusion population PK model for teicoplanin in critically ill adults with sepsis in the intensive care unit, with CKD-EPI estimated glomerular filtration rate as a power covariate on clearance (Chen 2023)
Teicoplanin (Wi 2017) Two-compartment IV bolus population PK model for teicoplanin in adult patients receiving venoarterial extracorporeal membrane oxygenation (VA-ECMO) for cardiogenic shock, with binary within-subject ECMO indicators on the central volume of distribution (V1) and inter-compartmental clearance (Q) and a binary CRRT indicator on the peripheral volume of distribution (V2) (Wi 2017)
Teicoplanin (Zhao 2015) Two-compartment IV-injection population PK model for teicoplanin in 85 children with malignant hematological disease (Zhao 2015). Body weight enters Vc and Vp with the fixed allometric exponent 1 and enters CL and Q with the fixed allometric exponent 0.75; Schwartz-formula creatinine clearance enters CL via a power exponent estimated at 0.606. Reference subject: WT = 27.1 kg, CRCL = 179 mL/min. The published model was used to derive age-band mg/kg dosing (18 mg/kg for infants, 14 mg/kg for children, 12 mg/kg for adolescents) and a patient-tailored daily dose (target AUC * CL_i) to attain the AUC(0,24 h) target of 750 mg.L/h.
Telapristone (Morris 2011) Population PK model for telapristone (CDB-4124, a selective progesterone-receptor antagonist developed for endometriosis and uterine fibroids) and its active monodemethylated metabolite CDB-4453 (Morris 2011). Parent is a two-compartment model with first-order oral absorption (no lag); metabolite is a one-compartment model with apparent volume V3/F fixed to 1 L for identifiability (Fmet is not separately identifiable from V3, so the estimated fmetest is interpreted as the ratio Fmet / V3 in 1/L). A NONMEM $MIXTURE block splits parent CL/F into a high-CL fast-eliminator subpopulation (CL/F = 11.6 L/h, population fraction P = 0.251) and a low-CL slow-eliminator subpopulation (CL/F = 3.34 L/h, P = 0.749); the mechanism is hypothesized in the Discussion to be polymorphic CYP3A5 activity but not directly tested. The mixture assignment is supplied as the binary covariate MIX_FAST_ELIM (1 = fast eliminator, 0 = slow eliminator) drawn per subject from a Bernoulli(0.251). The only retained clinical covariate is moderate renal impairment (RENALIMP_MOD), which produces a 74% proportional decrease in the telapristone absorption rate constant Ka relative to the healthy / mild-renal-impaired reference cohort. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Luo_2024_CES1_cirrhosis.html">Temocapril pbpk (Luo 2024)</a> </td> <td style="text-align:left;"> PBPK (semi-mechanistic, custom WinNonlin 8.1 implementation). Joint temocapril + temocaprilat disposition in healthy adults and in liver cirrhosis (Child-Pugh A/B/C). Temocapril is an inactive ester prodrug hydrolysed by hepatic CES1 to the active diacid temocaprilat. Temocaprilat is eliminated by BOTH renal and biliary routes, with biliary clearance twice the renal clearance (Section 3.1.5); biliary excretion is MRP2-mediated and is scaled in cirrhosis by the Table 1 MRP2 content ratio. The semi-PBPK circuit is stomach, three small-intestinal lumen segments (duodenum / jejunum / ileum), the matching three gut-wall segments, portal vein, liver, kidney and a one-compartment systemic compartment. Cirrhosis is applied by switching the Child-Pugh-specific physiology of Table 1 (organ blood flows, functional liver volume, GI transit rates, GFR, plasma-binding protein concentrations, hepatic CES1 content) and rescaling CLint, CLint,K, Peff and Vsys through Eq 1-6. Deterministic: the paper's virtual populations are uniform 80-120% draws on the drug parameters, not lognormal random effects, so no IIV is encoded. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mazzocco_2015_temozolomide.html">Temozolomide (Mazzocco 2015)</a> </td> <td style="text-align:left;"> Tumour growth inhibition (TGI) model for low-grade glioma (LGG) treated with first-line temozolomide chemotherapy (Mazzocco 2015): three tumour-tissue compartments (proliferative, non-damaged quiescent, damaged quiescent) coupled to a K-PD virtual drug compartment, with logistic proliferative growth (carrying capacity K fixed at 100 mm), treatment-induced damage of both proliferative and quiescent tissues, time-dependent acquired resistance of the proliferative tissue only, and tumour-genotype covariate effects of TP53 mutation status on TMZ efficacy and 1p/19q codeletion status on the damaged-quiescent-to-proliferative repair rate. Observation is mean tumour diameter (MTD = P + Q + Qp) in millimetres. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Baheti_2011_tenofovir.html">Tenofovir (Baheti 2011)</a> </td> <td style="text-align:left;"> Two-compartment first-order-absorption population PK model for plasma tenofovir (TFV) in HIV-1-infected adults on once-daily tenofovir disoproxil fumarate (TDF) coupled with a stimulatory indirect-response (Dayneka 1993) model for intracellular tenofovir diphosphate (TFV-DP) in peripheral blood mononuclear cells; plasma TFV drives TFV-DP formation through a sigmoidal Emax stimulation function. Creatinine clearance enters CL/F and Vc/F via a power covariate. Fitted sequentially (PK first, PD with PK individual post-hoc Bayes estimates fixed). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Thoueille_2023_tenofovir_ckd.html">Tenofovir alafenamide (Thoueille 2023)</a> </td> <td style="text-align:left;"> Joint two-analyte population PK model fitting tenofovir alafenamide and its active moiety tenofovir simultaneously in people living with HIV, with complete irreversible prodrug conversion and creatinine clearance, age, ethnicity and potent P-glycoprotein-inhibitor comedication on apparent tenofovir clearance. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lu_2016_tenofovir.html">Tenofovir combined (Lu 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and an absorption lag time for tenofovir (300 mg oral TDF once daily) in HIV-1-uninfected African adults receiving once-daily preexposure prophylaxis (Lu 2016, Partners PrEP Study). Combined variant: parameters estimated using a combined data set in which patient-reported dosing records were replaced with MEMS electronic adherence monitoring records where available. Absorption rate constant Ka is fixed at 1.5 /h; absorption lag time ALAG1 = 0.41 h. Apparent oral clearance (CL/F) carries a power-form covariate effect on creatinine clearance (raw Cockcroft-Gault, mL/min) centred at the cohort mean 106 mL/min. Diagonal IIV on CL/F, V1/F, and Ka; combined additive + proportional residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2016_tenofovir_emtricitabine.html">Tenofovir emtricitabine (Chen 2016)</a> </td> <td style="text-align:left;"> Linked population PKPD model for daily oral co-administered tenofovir (TFV, given as the prodrug TDF 300 mg = TFV 136 mg) and emtricitabine (FTC 200 mg) in HIV-positive and HIV-negative adults (Chen 2016 Cell-PrEP study). Each parent drug is described by a two-compartment first-order-absorption plasma popPK model. Each parent feeds a hybrid first-order-formation + saturation link into its intracellular triphosphate anabolite in peripheral blood mononuclear cells (TFV-DP, FTC-TP), modelled with a two-compartment 'recycle' elimination structure where a fraction R of the eliminated drug re-enters the central intracellular compartment. Each anabolite inhibits the zero-order production rate of two endogenous deoxynucleoside triphosphates via an Emax indirect-response model with Kout fixed to 1/day and Emax fixed to 1: TFV-DP inhibits dATP and dGTP (deoxypurines); FTC-TP inhibits dCTP and TTP (deoxypyrimidines). The dGTP effect waned over time and is described by an additional 1/(1+t^gamma) time factor. Sex is a covariate on FTC plasma Vc/F and HIV-infection status is a covariate on FTC-TP Kf. Intended for simulating analog:dNTP molar ratios (TFV-DP:dATP, FTC-TP:dCTP) for various dosing strategies, e.g., the IPERGAY on-demand PrEP regimen. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Thoueille_2023_tenofovir_ckd.html">Tenofovir full (Thoueille 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model for tenofovir after oral tenofovir alafenamide in people living with HIV, with creatinine clearance, age, ethnicity and potent P-glycoprotein-inhibitor comedication on apparent clearance (full final covariate model). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Granda_2024_kidney_pbpk.html">Tenofovir pbpk (Granda 2024)</a> </td> <td style="text-align:left;"> PBPK (mechanistic kidney, 35 states). Individualised prediction of tenofovir renal clearance from per-subject biomarker measurements of kidney blood flow and OAT1/3-mediated tubular secretory capacity, on top of tracer-measured GFR (Granda 2024, Clin Transl Sci). The kidney is resolved into 11 longitudinal subsegments (proximal tubule S1-S3, descending and ascending loop of Henle, distal tubule, and five collecting-duct subsegments), each with a tubular lumen, a tubular epithelial cell and a peritubular blood subspace (33 states), plus a systemic blood compartment and a bladder. Mechanisms: unbound glomerular filtration, OAT1/3-mediated active secretion in the proximal tubule only, pH-dependent bidirectional passive diffusion along the whole nephron, and CKD tubular-flow adaptation. The structural equations, segment volumes, surface areas, tubular pH profile and adaptive flow factors are the Huang & Isoherranen 2018 / 2020 framework; Granda 2024 replaces the framework population defaults for kidney blood flow and secretory clearance with per-subject measured / fitted values. Predicts each subject's tenofovir renal clearance from their measured GFR and kidney blood flow plus the kynurenic-acid-derived secretory capacity, scaled by 0.033. Reproduces the Predicted TFV CLr column of Table 4. Deterministic typical-value model: the paper reports no IIV and no residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lu_2016_tenofovir.html">Tenofovir prdi (Lu 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption for tenofovir (300 mg oral TDF once daily) in HIV-1-uninfected African adults receiving once-daily preexposure prophylaxis (Lu 2016, Partners PrEP Study). PRDI variant: parameters estimated using patient-reported dosing information with a steady-state assumption. Apparent oral clearance (CL/F) carries a power-form covariate effect on creatinine clearance (raw Cockcroft-Gault, mL/min) centred at the cohort median 106 mL/min. Diagonal IIV on CL/F only; combined additive + proportional residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Thoueille_2023_tenofovir_ckd.html">Tenofovir reduced (Thoueille 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model for tenofovir after oral tenofovir alafenamide in people living with HIV, with creatinine clearance on apparent clearance (reduced model, used by the authors for all chronic-kidney-disease simulations). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hamren_2008_tesaglitazar.html">Tesaglitazar (Hamren 2008)</a> </td> <td style="text-align:left;"> Mechanistic parent + acyl-glucuronide population PK model for tesaglitazar (a dual PPAR alpha/gamma agonist) in 41 adult subjects with varying degrees of renal function (Hamren 2008). Parent tesaglitazar follows a two-compartment disposition with first-order oral absorption (ka fixed at 1.5 1/h, F fixed at 1); renal clearance CLrt = 0.027 L/h directs parent to a cumulative urine compartment, and metabolic clearance CLmt = 1.91 L/h generates the acyl glucuronide metabolite. The metabolite follows a one-compartment disposition (Vcm = 8.5 L) with saturable Michaelis-Menten renal clearance (Vmax = 0.188 umol/h, Km = 0.041 umol/L) routing to a cumulative urine compartment, linear non-renal clearance (CLnrm = 1.2 L/h), and biliary excretion (kbm = 11.7 1/h) into a paper-specific gut compartment. The gut compartment releases interconverted parent tesaglitazar back into the parent central compartment at rate kicv = 0.79 1/h, completing the futile-cycle interconversion loop that the source paper proposes as the mechanism for increased tesaglitazar exposure in renal-impairment subjects. Covariates: BSA-normalized renal function CRCL (iohexol-clearance-measured GFR, mL/min/1.73 m^2; linear centered slope on CLrt and direct linear normalised scaling on metabolite Vmax), per-subject free fraction FU (% by ultrafiltration; linear centered slope on CLmt), sex SEXF (women have 31% lower CLrt than men), concomitant probenecid CONMED_PROBENECID (75% reduction of both CLrt and metabolite Vmax), and body weight WT (shared centered linear slope on Vct and Vpt). Concentrations are molar (umol/L) and amounts are molar (umol) throughout to match the Michaelis-Menten parameterisation of the acyl-glucuronide renal elimination; the user converts mg-of-tesaglitazar doses to umol using the molecular weight of 408.45 g/mol (1 mg = 2.45 umol). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lehr_2010_tesofensine.html">Tesofensine (Lehr 2010)</a> </td> <td style="text-align:left;"> Joint parent (tesofensine) + metabolite (M1, CYP3A4-formed) population PK and effect-compartment PK/PD model in mild Alzheimer's disease (Lehr 2010 Phase IIa fit; 62 patients across two 4-week placebo-controlled studies). Parent is one-compartment with first-order absorption (ka FIXED from upstream Phase I popPK) and parallel elimination through a metabolite-formation arm (CL_met = parent -> M1 flux) and a non-formation arm (CL_non-met = elimination via routes other than M1 formation). M1 is one-compartment with apparent volume FIXED at 0.768-fold of the parent apparent volume (mouse-derived ratio, Lehr 2010 ref 17). Tesofensine and M1 each drive their own effect compartment (shared keo FIXED at a small value, equivalent to a long effect-equilibration half-life); the combined drug effect on ADAS-Cog uses an extended Emax with competitive interaction in which the M1 effect-compartment concentration is divided by 5 to reflect the in-vivo M1 potency one-fifth that of the parent (Lehr 2010 Methods, ref 17). The ADAS-Cog observation equals the sum of drug, placebo, and disease-progression contributions (change from each subject's baseline). The placebo bi-exponential model (onset rate keq, offset rate kel_pla, scaling beta_pla) is fully FIXED to literature values from a published large-AD-cohort placebo model (Lehr 2010 ref 34); the linear disease-progression slope is FIXED at 6 ADAS-Cog points/year (Lehr 2010 ref 26). Emax is negative (a clinically meaningful ADAS-Cog improvement is a score reduction); the sign is applied inside model() while |Emax| is the ini-scale magnitude carried with multiplicative IIV. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Matsumoto_2005_TF_505.html">TF 505 (Matsumoto 2005)</a> </td> <td style="text-align:left;"> Two-compartment first-order-absorption population PK model for the oral 5-alpha-reductase inhibitor TF-505 coupled to an indirect-response PD model for plasma dihydrotestosterone (DHT, expressed as percent of basal) in which the DHT synthesis rate kin is modulated by a 24-h circadian cosine; fit to single- and multiple-dose data from healthy adult male Japanese volunteers (Matsumoto 2005). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Rovei_1982_theophylline.html">Theophylline (Rovei 1982)</a> </td> <td style="text-align:left;"> One-compartment oral PK model for theophylline tablets (Rovei 1982): first-order absorption with lag time in healthy adult volunteers across single oral doses of 125-500 mg. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Suda_2008_theophylline.html">Theophylline (Suda 2008)</a> </td> <td style="text-align:left;"> Steady-state population PK model for oral theophylline in 52 Japanese premature neonates and infants with apnea (Suda 2008). One-compartment first-order absorption structure; oral clearance CL/F is the only structural parameter the paper estimates (steady-state trough analysis Css = R / CL/F). Body-weight allometric scaling and a binary indicator for the Apnecut formulation (vs the in-house theophylline-alcohol comparator) on CL/F. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yano_1993_theophylline.html">Theophylline (Yano 1993)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for theophylline (Yano 1993 Paper II) in 55 adult inpatients with stable chronic airway obstruction; clearance and volume of distribution are log-linear functions of arterial PaCO2 and a binary hepatic-dysfunction indicator. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/elDesoky_1997_theophylline_pediatric_asthma.html">Theophylline pediatric asthma (elDesoky 1997)</a> </td> <td style="text-align:left;"> One-compartment IV PK model for theophylline in 15 Egyptian pediatric patients (age 2-12 yr, weight 12-30 kg) treated for an acute asthma attack (elDesoky 1997). Aminophylline given as a 30-min loading infusion (6 mg/kg) followed by 12 hr of continuous maintenance infusion (1 mg/kg/h); theophylline concentrations measured at 0.75, 7, and 13.25 hr. Parameter values taken from the Standard Calculations (SC) column of Table 2, which is independent of the Bayesian-prior population data and is treated by the authors as the reference (true) values. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Almquist_2016_ticagrelor.html">Ticagrelor (Almquist 2016)</a> </td> <td style="text-align:left;"> Preclinical (mouse, C57Bl/6 male). Mechanistic interaction PK model for ticagrelor, its active metabolite (TAM, AR-C124910XX), and the ticagrelor-neutralising Fab antibody fragment MEDI2452 in mouse (Almquist 2016). Three-compartment disposition for ticagrelor and TAM (shared plasma V, tissue V1, V2; V1 in instantaneous equilibrium with V); MEDI2452 lives in plasma V only and reversibly binds the free fractions of ticagrelor and TAM with rate kon and dissociation constant Kd; both free MEDI2452 and the two MEDI2452-drug complexes are eliminated together at the Fab clearance Cl_f (no recycling). Naive-pooled fit (no IIV); multiplicative log-normal residual error on five plasma assays. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LopezArguello_2023_pbp_binding_pseudomonas.html">Ticarcillin qsp (LopezArguello 2023)</a> </td> <td style="text-align:left;"> QSP. In vitro (Pseudomonas aeruginosa PAO1). Whole-cell penicillin-binding protein (PBP) covalent-binding model for ticarcillin (penicillin): seven coupled ODEs for the rate of net influx of drug across the outer membrane into periplasm and the competitive, mass-balanced acylation of six PBPs (1a, 1b, 2, 3, 4, 5/6) counted as molecules per bacterial cell. The intact parameter switches between the intact whole-cell assay (penetration-limited; drug enters periplasm at Rate_Influx/access) and the lysed isolated-membrane assay (no outer membrane; a vast excess of drug molecules is present at time 0). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2023_tilmicosin.html">Tilmicosin (Chen 2023)</a> </td> <td style="text-align:left;"> Preclinical (piglet, Duroc x Landrace x Yorkshire crossbred). Sigmoidal Emax PK/PD-integration model for the antibacterial effect of orally administered tilmicosin against Pasteurella multocida serovar D:7 (strain C44-15, MIC = 0.25 ug/mL) in subcutaneously implanted tissue-cage fluid (TCF). Chen 2023 Section 2.6 parameterises the antibacterial effect over a 24 h dosing interval as E = E0 + (Emax - E0) * Ce^N / (EC50^N + Ce^N), where E is the MAGNITUDE of the log10(CFU/mL) reduction accrued over that interval (positive = kill), E0 is the corresponding change in the untreated control, and Ce is a PK/PD index. The packaged model uses the paper's best-correlating index AUC24h/MIC (R^2 = 0.92, versus 0.90 for Cmax/MIC and 0.83 for %T>MIC), formed as the per-interval covariate AUC_TILM divided by the parameter mic. Parameters from Chen 2023 Table 2: Emax = 1.09 log10(CFU/mL), E0 = 0.003 log10(CFU/mL), EC50 = 26.66 h, Hill N = 2.69. The bacterial density bact (linear CFU/mL) is integrated as d/dt(bact) = -ln(10) * (E / 24) * bact so that log10(bact) falls by exactly E across each 24 h interval, reproducing the paper's per-interval model exactly at every time the paper actually counted bacteria (24 h boundaries only). There is NO PK component: tilmicosin exposure enters as the externally supplied per-interval AUC_TILM, because Chen 2023 analysed the tissue-cage-fluid concentrations non-compartmentally in WinNonlin (Table 1 reports NCA AUC0-24h, Cmax and T>MIC per dose group per dosing day) and reported no structural PK model. Neither between-subject variability nor a residual error magnitude was reported, so no eta parameters are present and addSd is FIXED at 0; the model is intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PerezRuixo_2006_tipifarnib.html">Tipifarnib (PerezRuixo 2006)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for oral and IV tipifarnib in healthy subjects and adult cancer patients (Perez-Ruixo 2006). Sequential zero-order release into the depot (duration D1) followed by first-order absorption (Ka) into the central compartment, with absorption lag time, linear elimination, two peripheral compartments, and bioavailability fixed at 26.7 percent. Covariate effects retained in the final model are total bilirubin on CL (power exponent -0.103 centred at 9 umol/L) and body weight on V2 (linear scaling, exponent fixed at 1, centred at 70 kg); healthy-vs-cancer cohort multipliers apply to CL, V2, Q4, V4, and Ka; a solution-vs-solid formulation indicator scales D1, Ka, and tlag. The mixture-model lag-time subpopulation (71.7 percent subpop 1 vs 28.3 percent subpop 2) is collapsed to the typical subpop-1 lag time for library simulation use; correlated IIVs with paper-reported correlation 1 (Q3-V3, CL-Q4, CL-V4) are encoded as derived etas via the published variance-expansion factors. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Stein_2019_Tisagenlecleucel.html">Tisagenlecleucel (Stein 2019)</a> </td> <td style="text-align:left;"> Cellular kinetic model for tisagenlecleucel CAR-T cells in pediatric and young adult patients with relapsed or refractory B-cell acute lymphoblastic leukemia (Stein 2019). Single-infusion expansion-then-biexponential-decline analytical model: transgene levels grow exponentially at rate rho up to Tmax, after which effector cells decline at rate alpha and a fraction FB transitions to memory cells declining at rate beta. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Budha_2023_tislelizumab.html">Tislelizumab (Budha 2023)</a> </td> <td style="text-align:left;"> Three-compartment population PK model for intravenous tislelizumab (anti-PD-1 IgG4) in patients with advanced tumors (Budha 2023) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/McLaughlin_2024_tld_1.html">Tld 1 (McLaughlin 2024)</a> </td> <td style="text-align:left;"> Joint parent (entrapped) + free + metabolite (doxorubicinol) population PK model for TLD-1, a novel small-diameter pegylated liposomal doxorubicin, in 30 adults with advanced solid tumours (McLaughlin 2024 phase I dose-escalation, SAKK 65/16 / NCT03387917). Structure: one-compartment liposomal-entrapped reservoir (V1) with linear release into the free-doxorubicin central compartment (release rate krel = CL1/V1); free doxorubicin disposition is two- compartment (Vc, Vp, Q) with linear metabolism-to-doxorubicinol clearance (CL); doxorubicinol is one-compartment with linear elimination (Vc_doxol, CL_doxol). Body surface area (BSA, reference 1.75 m^2) is the only retained covariate, entering as a power model on the free-doxorubicin central (V2 exponent 4.47) and peripheral (V3 exponent 11.5) volumes. Inter-individual variability is fitted on CL1 (release), V1 (shared-eta scale 0.643 of ome_CL1), CL2 (free->doxol) and CL4 (doxol elimination); inter-occasion variability on CL1, CL2, V1, V2 from Table 2 is documented but not encoded structurally here (nlmixr2lib has no canonical occasion-column convention; see Hempel 2003 / Hong 2006 precedents). Residual error is log-transformed-both-sides additive on the log scale -- equivalent to proportional in nlmixr2's linear space and encoded here as separate propSd per analyte. Distinct from Hempel 2003 (paediatric liposomal daunorubicin, total drug only) and Varatharajan 2016 (free daunorubicin + daunorubicinol in adult AML, no liposomal reservoir). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hennig_2013_tobra.html">Tobra (Hennig 2013)</a> </td> <td style="text-align:left;"> Two-compartment intravenous population PK model for tobramycin in adults and children with and without cystic fibrosis (Hennig 2013); fat-free mass allometric scaling on CL/Q (estimated exponent) and on V1/V2 (linear), sex-specific reference CL and V1, piecewise-linear age effect on CL with breakpoint at 18 years, and a power effect of the SCR_mean/SCR ratio on CL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Conil_2010_tobramycin.html">Tobramycin (Conil 2010)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for tobramycin in adult ICU patients receiving once-daily aminoglycoside therapy for nosocomial Gram-negative infections (Conil 2010); additive linear covariate effects of Cockcroft-Gault creatinine clearance and height on CL, with no IIV on Q or V2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hennig_2008_tobramycin.html">Tobramycin (Hennig 2008)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for once-daily IV tobramycin in paediatric cystic fibrosis patients (Hennig 2008), with allometric weight scaling on CL, Q, Vc, and Vper (reference 70 kg, exponent 3/4 for clearances and 1 for volumes), full-block correlated between-subject variability on CL/Vc/Vper, a fixed 30 min infusion duration into the central compartment, and an estimated lag time between infusion start and drug entry into the patient's vein. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Livio_2014_tobramycin.html">Tobramycin (Livio 2014)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for systemic tobramycin released from an implanted calcium-sulfate bone-graft substitute (Osteoset T) in adults undergoing orthopedic surgery (Livio 2014); clearance equated to Cockcroft-Gault creatinine clearance under the assumption that absorbed tobramycin is exclusively eliminated by glomerular filtration, and absolute bioavailability differing between the 10 g (262 mg tobramycin) and 20 g (524 mg tobramycin) Osteoset T cast cohorts. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ting_2014_tobramycin_inhaled.html">Tobramycin inhaled (Ting 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for inhaled tobramycin powder (TIP / TOBI Podhaler) in cystic fibrosis patients (Ting 2014), with first-order absorption from a depot compartment and apparent (post-bioavailability) clearance and volumes. Body mass index (BMI) and baseline FEV1 percent-predicted are power-form covariates on apparent central volume of distribution (reference 18.8 kg/m^2 and 62.1 % respectively). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marier_2002_tobramycin_rat_conventional.html">Tobramycin rat conventional (Marier 2002)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment population PK model for the conventional (non-liposomal) formulation of tobramycin (Tobi inhalation solution, PathoGenesis) after a single 1,200 ug intratracheal dose to male Sprague-Dawley rats with chronic Burkholderia cepacia (strain BC 1368) pulmonary infection. NONMEM ADVAN4 (depot, central, peripheral) parameterised in rate-constant form: first-order absorption ka into a lung central compartment carrying drug amount (not concentration -- volumes of distribution were not fitted because the dependent variable was the amount of tobramycin recovered from homogenised lung tissue, calculated as the measured tissue concentration times the lung volume per animal), inter-compartmental rate constants k12 and k21 between central and peripheral, first- order elimination kel from central, and a fitted lung bioavailability FL accounting for the fraction of the intratracheal dose actually reaching the lung tissue compartment. Comparator arm for Marier_2002_tobramycin_rat_liposomal; the conventional formulation shows faster absorption, faster elimination, and ~8-fold lower lung AUC than the liposomal formulation in the source paper (Table 1, Results). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marier_2002_tobramycin_rat_liposomal.html">Tobramycin rat liposomal (Marier 2002)</a> </td> <td style="text-align:left;"> Preclinical (rat). Two-compartment population PK model for the liposomal formulation of tobramycin (DPPC:DMPG 10:1 phospholipids, 230-400 nm extruded) after a single 1,200 ug intratracheal dose to male Sprague-Dawley rats with chronic Burkholderia cepacia (strain BC 1368) pulmonary infection. NONMEM ADVAN4 (depot, central, peripheral) parameterised in rate- constant form: first-order absorption ka into a lung central compartment carrying drug amount (not concentration -- volumes of distribution were not fitted because the dependent variable was the amount of tobramycin recovered from homogenised lung tissue, calculated as the measured tissue concentration times the lung volume per animal), inter-compartmental rate constants k12 and k21 between central and peripheral, first-order elimination kel from central, and a fitted lung bioavailability FL accounting for the fraction of the intratracheal dose actually reaching the lung tissue compartment. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bastida_2018_tocilizumab.html">Tocilizumab (Bastida 2018)</a> </td> <td style="text-align:left;"> One-compartment population PK model for intravenous tocilizumab in adults with rheumatoid arthritis (Bastida 2018), with parallel first-order linear and Michaelis-Menten elimination from the central compartment; total body weight and time-varying C-reactive protein on linear CL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Frey_2010_tocilizumab.html">Tocilizumab (Frey 2010)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for tocilizumab in adults with moderate-to-severe rheumatoid arthritis (Frey 2010), with parallel first-order linear and Michaelis-Menten elimination from the central compartment. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Frey_2013_tocilizumab.html">Tocilizumab (Frey 2013)</a> </td> <td style="text-align:left;"> Indirect-response PK/PD model of tocilizumab on the 28-joint Disease Activity Score (DAS28) in adults with rheumatoid arthritis (Levi/Grange/Frey 2013, OPTION + TOWARD phase III pool, n = 1703 patients with 12,618 DAS28 observations). Tocilizumab inhibits the DAS28 production rate kin via a sigmoid emax function whose driving concentration is the sum of circulating tocilizumab and a constant DMARD background term expressed in tocilizumab concentration units. The PK driver is the two-compartment, parallel linear + Michaelis-Menten model of Frey 2010 (PMID 20097931), reused unchanged for the exposure-response analysis. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jorga_2000_tolcapone.html">Tolcapone fluctuators (Jorga 2000)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption (no lag) for tolcapone in parkinsonian patients with fluctuating levodopa response, with effects of lean body weight and serum protein on clearance, lean body weight and dose group on central volume, serum albumin and dose group on peripheral volume, and concomitant food on bioavailability (Jorga 2000, fluctuator dataset, n=215) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jorga_2000_tolcapone.html">Tolcapone nonfluctuators (Jorga 2000)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and absorption lag for tolcapone in parkinsonian patients with stable (non-fluctuating) levodopa response, with effects of creatinine clearance on clearance, serum protein on central volume, and concomitant food on bioavailability (Jorga 2000, nonfluctuator dataset, n=60) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yamamoto_2023_tominersen.html">Tominersen (Yamamoto 2023)</a> </td> <td style="text-align:left;"> Six-compartment population PK model for tominersen (2'-O-methoxyethyl antisense oligonucleotide targeting huntingtin mRNA) following intrathecal lumbar-puncture administration in adults with Huntington's disease (Yamamoto 2023). Two coupled three-compartment subsystems: the intrathecal bolus enters the central CSF compartment, which exchanges with two CNS-tissue compartments and drains unidirectionally into the plasma central compartment via CL_CSF (bioavailability F1 and F2 both fixed to 1); the plasma central compartment has two peripheral compartments and first-order elimination. Baseline CSF total protein, age, and antidrug antibodies are covariates on CL_CSF; body weight is an allometric power covariate on all plasma clearances and volumes, and antidrug antibodies and female sex are covariates on plasma CL. Residual error is additive on the log scale (lognormal) for both CSF and plasma, each with interindividual variability on the residual magnitude. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ahmed_2015_topiramate.html">Topiramate (Ahmed 2015)</a> </td> <td style="text-align:left;"> Population PK/PD model of topiramate (TPM) and its acute effect on phonemic generative fluency (Controlled Oral Word Association, COWA) in healthy adult volunteers given single oral or intravenous doses of 50-100 mg (Ahmed 2015). Two-compartment popPK with first-order absorption and elimination, oral bioavailability ~108%, allometric body-weight scaling on CL/Q (fixed 3/4) and Vc/Vp (fixed 1); separate proportional residual errors for oral and IV cohorts. PD: COWA = baseline * practice_factor * exp(-KE * Cc), where the practice factor inflates baseline by 12% beginning with the fourth (and subsequent) COWA test administration and KE = 0.157 L/mg gives a 14.5% drop in COWA per 1 mg/L rise in plasma TPM. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2024_topiramate.html">Topiramate (Lee 2024)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption and elimination for topiramate (TPM) in Korean adults with epilepsy undergoing routine therapeutic drug monitoring (Lee 2024). Apparent clearance CL/F carries an additive enzyme-inducing antiseizure-medication term (separate coefficients in L/h for phenytoin, carbamazepine, oxcarbazepine and phenobarbital), a power effect of creatinine clearance normalized to 90 mL/min, and a power effect of the topiramate daily dose normalized to 100 mg/day; apparent volume Vd/F scales allometrically with body weight (exponent fixed at 1) centred at 62 kg. Absorption rate constant fixed at 2 /h. Inter-individual variability on CL/F only (31.0 %CV) with a proportional residual error of 27.8%. The structural model was carried from Bae 2016 and all parameters were re-estimated on the present therapeutic-drug-monitoring dataset. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vuu_2016_topiramate_dog.html">Topiramate dog (Vuu 2016)</a> </td> <td style="text-align:left;"> Preclinical (dog). Population two-compartment intravenous PK model for topiramate (TPM) in dogs with naturally-occurring epilepsy (Vuu 2016). Stable-labelled TPM was given as a 5-min IV infusion at 10 mg/kg (n = 4) or 20 mg/kg (n = 3); pooled across the low- and high-dose data, a two- compartment model with first-order elimination from the central compartment described the disposition best. Concomitant phenobarbital (CONMED_PB) was identified as an enzyme-inducer covariate on systemic clearance via an exponential effect (Cl = tvCl * exp(dCl * CONMED_PB)), yielding a 5.64-fold higher CL in PB-coadministered dogs. Per-kg structural parameters (Vc, Vp, CL, Q) are scaled to absolute units by individual body weight (WT, kg) inside the model; the dose in the event table is therefore absolute mg (mg/kg dose times WT). IIV is exponential on Vc and CL; residual error is proportional (~15%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Leger_2004_topotecan.html">Topotecan (Leger 2004)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral and intravenous topotecan in adult cancer patients, with first-order absorption + lag time for the oral route, additive linear creatinine-clearance plus linear-ordinal WHO performance-status effects on CL, and linear body-weight effect on the central volume of distribution (Leger 2004) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Roberts_2016_topotecan.html">Topotecan (Roberts 2016)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model for oral topotecan lactone in infants and very young children with primary central nervous system tumours (Roberts 2016). First-order absorption into a depot compartment is followed by first-order elimination from a central compartment. Apparent volume of distribution (V/F) and apparent clearance (CL/F) are scaled by body surface area as power functions centred on the cohort median (0.57 m^2); the ABCG2 rs4148157 G>A variant (heterozygous AG or homozygous AA carriers pooled vs the GG reference) carries an exponential covariate effect on the absorption rate constant Ka, yielding an approximately 2-fold higher Ka in carriers than in GG homozygotes. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Saleh_2023_leicns_pk30_mouse_brain.html">Topotecan mouse pbpk (Saleh 2023)</a> </td> <td style="text-align:left;"> PBPK (LeiCNS-PK3.0 CNS physiologically-based model, mouse version). Preclinical (mouse, CD1 nude). Nine-compartment CNS PBPK model predicting unbound topotecan concentrations in brain extracellular fluid (brain ECF) after a single 4 mg/kg intravenous dose. The CNS structure is the mouse re-parameterisation of LeiCNS-PK3.0: brain microvasculature (brain_vascular), brain ECF, a phospholipid brain-cell-membrane binding compartment (brain_cell_membrane), brain intracellular fluid (brain_icf), lysosomes (brain_lysosome), and the four CSF compartments (lateral ventricles, third + fourth ventricles, cisterna magna, subarachnoid space) draining in series back to plasma. Transport across the BBB and BCSFB is the sum of a paracellular clearance (Qp, charged + neutral drug) and a transcellular clearance (Qt, neutral drug only, scaled by pH-dependent neutral fractions PHF and by asymmetry factors AF that encode active transport). Every CNS parameter is fixed to mouse physiology (Table III) or derived from the drug's physicochemical properties (Table I); none was fitted here. The BBB asymmetry factors are back-calculated from Kp,uu,BBB = 0.21 (taken from ref [57]; Table V). The plasma PK model is the empirical two-compartment model of Table IV (in-house NONMEM fit) and acts purely as a forcing function: brain uptake does not deplete plasma, exactly as published. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jeong_2022_torsemide.html">Torsemide (Jeong 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for oral torsemide in healthy Korean adult males (Jeong 2022), with first-order absorption after a lag time, proportional residual error, and categorical genotype covariates: OATP1B1 *15 haplotype (intermediate / poor transporter) reduces apparent central volume, and CYP2C9 extensive-metabolizer phenotype increases apparent oral clearance and apparent inter-compartmental clearance. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Baverel_2015_tralokinumab.html">Tralokinumab (Baverel 2015)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for tralokinumab in adolescent (12-17 y) and adult subjects with asthma or healthy volunteers (Baverel 2015), with parallel subcutaneous absorption (first-order with lag plus zero-order over a fixed duration), allometric body-weight scaling on disposition parameters, and an additional 15% lower clearance in adolescents. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Soehoel_2022_tralokinumab.html">Tralokinumab (Soehoel 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for tralokinumab (Soehoel 2022) in adults with moderate-to-severe atopic dermatitis, with SC first-order absorption and allometric body-weight effects. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mercier_2014_tramadol_tapentadol.html">Tramadol tapentadol mbma (Mercier 2014)</a> </td> <td style="text-align:left;"> MBMA. Longitudinal model-based meta-analysis pain-intensity time-course model comparing tramadol and tapentadol in adults with chronic non-malignant pain (osteoarthritis, back pain, neuropathic pain, and other chronic non-malignant pain), fit to arm-level summary data from 45 double-blind Phase II/III randomized clinical trials representing 81 treatment arms and approximately 12,985 patients. Pain intensity on a normalized 0-10 scale is described by a logistic (expit) latent-scale model: PI(t) = 10 * expit(Base + R * (1 - exp(-k*t))), where the extent-of-reduction term R has an Emax-in-dose form for tramadol: R = R_Pla * (1 + theta_Base * logit(PI0/10) + theta_Trm * Dose_trm / (Dose_trm + ED50) * TRAMADOL + theta_Tap * TAPENTADOL). Between-study-arm variability enters additively on Base (SD 0.313 on the latent scale) and exponentially on R (SD 0.065). The residual variance is inversely proportional to per-arm sample size N; the paper reports two scale-dependent residual SDs (VAS/continuous SD 0.260 and categorical SD 0.205) and this file exposes the dominant VAS SD as the primary residual, with the categorical SD documented in ini() and in the vignette Assumptions/Errata. Suitable simulation scope is study-arm-mean pain-intensity time-course over 0-15 weeks (one 52-week trial contributed but the paper's simulations use 12 weeks); NOT individual-patient predictions. Tramadol was studied over a wide dose range (adequate to characterize a dose-response); tapentadol was only studied over 100-250 mg bid (no dose-response estimable), so the tapentadol effect is a single per-arm indicator effect. The paper also fit separate logistic MBMA sub-models for adverse events (constipation, nausea, vomiting, dizziness, somnolence) and drop-outs (due to adverse event, lack of efficacy); those sub-models are reported only as graphical odds ratios in Figs 3-4 with no tabulated logistic intercepts or slopes, so they are NOT extracted here (see vignette Errata for the omission). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Dunn_2025_tranexamicAcid.html">TranexamicAcid (Dunn 2025)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for tranexamic acid (TXA) with parallel first-order intramuscular and first-order oral absorption (oral lag time) and first-order elimination, in pregnant individuals receiving IV, IM, or oral TXA for prevention or treatment of postpartum hemorrhage (Dunn 2025). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bruno_2005_trastuzumab.html">Trastuzumab (Bruno 2005)</a> </td> <td style="text-align:left;"> Two-compartment linear population PK model for intravenous trastuzumab in adults with HER2-positive metastatic breast cancer (MBC) or advanced solid tumors; covariate effects of number of metastatic sites (>= 4) and baseline HER2 shed extracellular domain (ECD) on clearance, and body weight and ECD on central volume (Bruno 2005, first published trastuzumab popPK). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/LeTilly_2021_trastuzumab.html">Trastuzumab (LeTilly 2021)</a> </td> <td style="text-align:left;"> Two-compartment serum/CSF population PK model for trastuzumab after intrathecal and intravenous administration in adults with HER2+ breast cancer leptomeningeal metastases (Le Tilly 2021); zero-order serum-to-CSF transfer plus first-order CSF-to-serum return, with a Friberg-style chain of latent target (HER2) transit compartments and irreversible binding-driven elimination of trastuzumab in the CSF compartment. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marcantonio_2022_efa.html">Trastuzumab (Marcantonio 2022)</a> </td> <td style="text-align:left;"> QSP. Two-compartment monospecific anti-receptor mechanistic PKPD model of trastuzumab-HER2 binding in adults with HER2-overexpressing breast cancer (Marcantonio 2022 Early Feasibility Assessment, Case Study 2/3). Drug administered IV as a bivalent (valency = 2) antibody that binds membrane HER2 (in central and peripheral compartments) and soluble HER2 (the shed extracellular domain) via independent binding events with identical Kd. All species eliminate first-order; drug bound to membrane receptor eliminates at the receptor's rate; drug bound to soluble receptor eliminates at the drug's rate. Only soluble species (Ab_00, Ab_0S, Ab_S0, Ab_SS, S1) distribute between central and peripheral; membrane-bound drug forms (Ab_0R, Ab_R0, Ab_RR, Ab_RS, Ab_SR) stay in their compartment. The endogenous HER2 ligand is not resolved (paper sets its concentration to effectively zero) so L1 and L1R1 species are omitted here. Parameters FIXED from paper Table S7 and JSON run file; disease and toxicity compartments are omitted per paper Case Study 2 text. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Quartino_2016_trastuzumab.html">Trastuzumab (Quartino 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with parallel linear and Michaelis-Menten nonlinear elimination from the central compartment and first-order subcutaneous absorption (with bioavailability) for trastuzumab (Herceptin) administered IV or as a fixed 600 mg manual-syringe SC dose in women with HER2-positive early breast cancer; covariates body weight (on CL, Vc, Vp) and ALT (on CL) (Quartino 2016, HannaH study) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Quartino_2019_trastuzumab.html">Trastuzumab (Quartino 2019)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with parallel linear and Michaelis-Menten nonlinear elimination from the central compartment for intravenous trastuzumab (Herceptin) in patients with metastatic breast cancer, early breast cancer, advanced gastric cancer, or other solid tumors (Quartino 2019) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Reijers_2016_trastuzumab.html">Trastuzumab (Reijers 2016)</a> </td> <td style="text-align:left;"> Three-compartment population PK model with parallel linear and Michaelis-Menten nonlinear elimination from the central compartment for intravenous trastuzumab in healthy male volunteers from a phase I biosimilarity trial of the FTMB biosimilar vs Herceptin reference product (Reijers 2016, combined model on all dose levels 0.49-6.44 mg/kg); covariates are lean body mass on central volume of distribution V1 and BMI on the linear elimination rate constant ke. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/deVriesSchultink_2018_cardiotoxicity.html">Trastuzumab LVEF (deVriesSchultink 2018)</a> </td> <td style="text-align:left;"> Effect-compartment PD model for left-ventricular ejection fraction (LVEF) decline during adjuvant trastuzumab treatment in HER2-positive early breast cancer (de Vries Schultink 2018). Trastuzumab pharmacokinetics are an inlined deterministic forcing function from the previously published Bruno 2005 two-compartment linear popPK (de Vries Schultink 2018 Methods: 'The trastuzumab PK profiles were obtained using fixed effect parameters from a previously published PK model for HER2-positive breast cancer patients [18]'); typical population values plus WT / HER2_ECD / MET_GE4 covariate effects are retained from Bruno 2005 Table 3. Cardiac damage is generated by cumulative trastuzumab concentration via an effect compartment Ceff that integrates plasma Ctrastuzumab and decays at rate log(2)/T1/2rec; LVEF declines through a sigmoid Emax expression LVEF = LVEF0 * (1 - Ceff / (Ceff + EC50)), and the EC50 is modulated by the per-subject peak post-anthracycline troponin T (TROPONIN_T_MAX), with a higher TROPONIN_T_MAX lowering EC50 and increasing sensitivity to trastuzumab-induced cardiotoxic decline. Companion file `deVriesSchultink_2018_anthracycline_troponinT.R` supplies the upstream K-PD anthracycline-troponin T model whose peak output is used here as a covariate. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/FehlingKaschek_2019_trastuzumab_skbr3.html">Trastuzumab skbr3 (FehlingKaschek 2019)</a> </td> <td style="text-align:left;"> In vitro (SKBR3 cell line). Mechanistic ODE model of trastuzumab-induced HER2 receptor internalization with two cell-membrane phenotypes (ruffled vs flat); Model B of Fehling-Kaschek 2019, no recycling or degradation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Yin_2021_trastuzumabDeruxtecan.html">TrastuzumabDeruxtecan (Yin 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for intact trastuzumab deruxtecan (T-DXd, DS-8201, anti-HER2 antibody-drug conjugate) with linear elimination and covariate effects of body weight, albumin, baseline tumor size, sex, and Japan-country indicator in patients with HER2-positive breast cancer or other HER2-expressing solid tumors (Yin 2021) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Asiimwe_2025_trastuzumab_ADCs_mbma.html">TrastuzumabDeruxtecan mbma (Asiimwe 2025)</a> </td> <td style="text-align:left;"> MBMA. Two-compartment linear population PK model of trastuzumab deruxtecan (T-DXd, DS-8201, Enhertu, anti-HER2 antibody-drug conjugate) fitted by model-based meta-analysis to summary-level concentration-time data digitised from 4 published clinical trials in patients with HER2-positive breast, non-small-cell lung, gastric/GEJ, colorectal and other solid tumors. Between-study variability (BSV) is a study-level random effect on CL and Vc (block correlation 0.915) representing differences in inclusion criteria across trials and dose-escalation arms treated as separate studies. Residual error is proportional + additive on serum concentration (ng/mL) and was weighted by the square root of each trial's sample size during fitting; the tabulated parameter estimates are the unweighted values and simulation of a study of N subjects should scale the residual SD by 1/sqrt(N). Suitable simulation scope is study-arm-mean cycle-1 concentration-time profiles, NOT individual-patient concentrations. Parameter values are the T-DXd column of Asiimwe 2025 Table 1 (Monolix 2024R1). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lu_2014_trastuzumabemtansine.html">Trastuzumabemtansine (Lu 2014)</a> </td> <td style="text-align:left;"> Linear two-compartment population PK model of trastuzumab emtansine (T-DM1, anti-HER2 antibody-drug conjugate) with first-order elimination from the central compartment in patients with HER2-positive locally advanced or metastatic breast cancer (Lu 2014) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Asiimwe_2025_trastuzumab_ADCs_mbma.html">TrastuzumabEmtansine mbma (Asiimwe 2025)</a> </td> <td style="text-align:left;"> MBMA. Two-compartment linear population PK model of trastuzumab emtansine (T-DM1, anti-HER2 antibody-drug conjugate) fitted by model-based meta-analysis to summary-level concentration-time data digitised from 14 published clinical trials in patients with HER2-positive metastatic breast cancer and other HER2-positive solid tumors. Between-study variability (BSV) is a study-level random effect on CL and Vc (block correlation 0.825) representing differences in inclusion criteria and the pooling of Phase I dose-escalation arms as separate 'studies'. Residual error is proportional + additive on serum concentration (ng/mL) and was weighted by the square root of each trial's sample size during fitting; the tabulated parameter estimates are the unweighted values and simulation of a study of N subjects should scale the residual SD by 1/sqrt(N). Suitable simulation scope is study-arm-mean cycle-1 concentration-time profiles, NOT individual-patient concentrations. Parameter values are the T-DM1 column of Asiimwe 2025 Table 1 (Monolix 2024R1). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bender_2014_trastuzumabEmtansine_mechanistic.html">TrastuzumabEmtansine mechanistic (Bender 2014)</a> </td> <td style="text-align:left;"> Mechanistic DAR0-DAR7 catenary deconjugation PK model for trastuzumab emtansine (T-DM1) in cynomolgus monkeys (default) and rats (Bender 2014): each DAR moiety distributes into a shared three-compartment backbone and deconjugates sequentially toward naked trastuzumab (DAR0); uses five shared upper-chain rate constants (k7->3) plus separate k_2->1 and k_1->0. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bender_2014_trastuzumabEmtansine_reduced.html">TrastuzumabEmtansine reduced (Bender 2014)</a> </td> <td style="text-align:left;"> Reduced three-compartment population PK model for trastuzumab emtansine (T-DM1) and naked trastuzumab (DAR0) in cynomolgus monkeys (default) and rats (Bender 2014): single lumped T-DM1 conjugate species deconjugates into DAR0 via a single deconjugation clearance; both species share V1/V2/V3 and distributional clearances. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hwang_2022_tremelimumab.html">Tremelimumab (Hwang 2022)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for tremelimumab (anti-CTLA-4 IgG2 kappa) with regimen-dependent sigmoidal time-varying clearance in adults with advanced solid tumours, dosed as monotherapy or in combination with durvalumab (Hwang 2022) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Danielak_2017_treosulfan.html">Treosulfan (Danielak 2017)</a> </td> <td style="text-align:left;"> Two-compartment IV-infusion population PK model for treosulfan (TREO) in pediatric patients undergoing conditioning prior to hematopoietic stem cell transplantation (Danielak 2017). Allometric body-weight scaling normalised to a 70 kg adult typical value with exponents fixed at 0.75 on CL and 1 on V1 and V2; Q has no weight covariate. Correlated IIV on CL and V1 (Cl-V1 correlation 0.714); independent IIV on Q. Proportional residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Park_2014_triflusal.html">Triflusal (Park 2014)</a> </td> <td style="text-align:left;"> One-compartment population PK with first-order metabolite-formation kinetics for the active triflusal metabolite hydroxy-4-(trifluoromethyl) benzoic acid (HTB) in healthy Korean male volunteers, with a binary probability PD model for inhibition of platelet aggregation (IPA). Triflusal is an antiplatelet prodrug; only HTB is measured analytically. NONMEM ADVAN2 TRANS2 is used by the source paper -- the canonical depot compartment carries triflusal and the canonical first-order rate constant (here `lka`) plays the role of the paper's HTB formation rate constant kf (0.341 1/h). Apparent oral clearance CL/F (0.200 L/h at 71.65 kg) and apparent oral volume V/F (8.300 L at 71.65 kg) describe HTB disposition; F absorbs the unknown fraction of triflusal converted to HTB. Body weight is the only retained covariate and enters as a power on CL/F (exponent 0.845) and direct proportionality on V/F (exponent fixed to 1). PD endpoint is binary IPA = 1 when platelet aggregation < 74% else 0; the instantaneous probability of IPA is a sigmoid Hill function of HTB concentration, prob_ipa = Cc^gamma / (EC50^gamma + Cc^gamma), with EC50 = 84.9 ug/mL and gamma = 19.2 (BSV on gamma fixed to 0). The Hill exponent is very steep (quantal-like concentration-response). Parameter values from Park 2014 Table 2 Estimates column. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2023_tripegfilgrastim.html">Tripegfilgrastim (Lee 2023)</a> </td> <td style="text-align:left;"> Semi-mechanistic population PK/PD model for tripegfilgrastim (a PEGylated long-acting recombinant human G-CSF) and absolute neutrophil count (ANC) in healthy Korean adults and Korean pediatric patients with solid tumors receiving chemotherapy. Subcutaneous drug enters a depot compartment and is absorbed at first-order rate KSC into a total-drug compartment. Pharmacodynamics-mediated drug disposition (PDMDD) uses a quasi-equilibrium quadratic between total drug and the circulating G-CSF-receptor pool, giving free drug (FDC) and bound complex (RDC). Free drug is cleared linearly (CLD/VD) and bound drug is internalised (KINT). Granulopoiesis is a five-state receptor chain (stem -> mitotic -> post-mitotic I -> post-mitotic II -> circulating blood receptors) with baselines KP/KTR and KP/KC; ANC = 1000 * RB / SR. Drug binding stimulates receptor production (ST1 = 1 + STM1*driver) and bone-marrow transit (ST2 = 1 + STM2*driver), where the driver is the fraction of receptors bound by exogenous drug plus endogenous G-CSF. Endogenous G-CSF is carried as its own turnover compartment (fixed KIN, KEL, GCSF0 from Quartino 2014) and a negative-feedback term FB = (RB0/RB)^GAM modulates receptor production. Chemotherapy is a KPD virtual compartment with lag LAG; its output (KCHM*CHM) scaled by CHMSL adds to the mitotic-cell elimination rate. Study population (DIS_HEALTHY) shifts VD and KD, age scales KSC (exponent -0.97, reference 18.5 y), body weight scales KINT (exponent 1.7, reference 55.1 kg), and baseline ANC scales KP (exponential, reference 2106 cells/uL). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Tornoe_2006_HPG_axis.html">Triptorelin (Tornoe 2006)</a> </td> <td style="text-align:left;"> Population PK/PD model of the hypothalamic-pituitary-gonadal (HPG) axis after a single 3.75 mg subcutaneous (s.c.) depot injection of the GnRH agonist triptorelin in healthy adult males. PK is a two-compartment disposition model with a combined zero-order burst (fraction Fr of dose released over duration t into central) and a two-step first-order s.c. absorption (lymphatic delay) for the remaining (1 - Fr) fraction. PD is a four-state HPG-axis feedback model (feedback compartment F, LH pool P, LH, testosterone Te) with sigmoidal Emax stimulation of LH pool release by triptorelin and a negative interaction (F^-1) from the feedback compartment on LH synthesis and release; testosterone secretion is stimulated by LH via a sigmoidal Emax model. ke_LH, ke_F, lambda, LH_base, and Te_base are the triptorelin-study-specific values from Table 4. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Grimm_2023.html">Trontinemab (Grimm 2023)</a> </td> <td style="text-align:left;"> Trontinemab PK model in non-human primates (Grimm 2023): two-compartment plasma PK with Michaelis-Menten elimination and brain-region effect-compartment distribution (brain_cerebellum, brain_hippocampus, brain_striatum, brain_cortex, choroid plexus, CSF). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Pouzin_2022_tusamitamab.html">Tusamitamab (Pouzin 2022)</a> </td> <td style="text-align:left;"> Integrated multi-analyte semi-mechanistic population PK model of tusamitamab ravtansine (SAR408701, anti-CEACAM5 IgG1-SPDB-DM4 ADC) in adults with advanced solid tumors (Pouzin 2022): explicit two-compartment disposition for DAR1-DAR8 ADC species and a separate naked-antibody (NAB) chain sharing Vc/Vp/Q, irreversible first-order DAR_n -> DAR_(n-1) deconjugation feeding a one-compartment DM4 catabolite that converts to MeDM4. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2023_tylosin.html">Tylosin Apleuropneumoniae healthy (Lee 2023)</a> </td> <td style="text-align:left;"> Preclinical (pig, crossbred Duroc x (Landrace x Yorkshire)). Ex vivo sigmoidal Emax PK/PD-integration model for the antibacterial effect of tylosin against Actinobacillus pleuropneumoniae (challenge isolate BA2000013) in plasma drawn from healthy pigs after a single 20 mg/kg intramuscular dose. Lee 2023 Section 2.8 parameterises the drug effect over a 24 h ex vivo incubation as E = E0 - (Emax * C^gamma) / (C^gamma + EC50^gamma), where E is the SIGNED log10(CFU/mL) difference between 0 and 24 h of incubation (positive = net growth, negative = net kill), E0 is that same 24 h change in the drug-free control, and C is the PK/PD index AUC24h/MIC. Parameters from Lee 2023 Table 3, column 'A. pleuropneumoniae / Healthy': Emax = 7.19 log10 CFU/mL, EC50 = 1.21 h, E0 = 2.84 log10 CFU/mL, gamma = 2.76. The parameterisation was confirmed numerically against the paper's own independently printed rows: Emax - E0 reproduces exactly (4.35), and solving E = 0 and E = -3 returns AUC24h/MIC = 1.037 and 2.057 h versus the published 0.98 and 1.97 h. The bacterial density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * (E / 24) * bact so that log10(bact) changes by exactly E across each 24 h window, reproducing the paper's model at the only times bacteria were counted. There is NO PK component: Lee 2023 analysed the plasma concentrations non-compartmentally in WinNonlin 8.3 (Section 2.7; Table 2 reports only Cmax, Tmax, T1/2, AUC, Vz/F, Cl/F and MRT) and published no structural PK model, so exposure enters as the externally supplied covariate AUCMIC_TYLO. That covariate carries the AUC24h/MIC RATIO directly because Lee 2023 never reports the challenge isolates' own MIC -- only that strains 'with MIC values similar to the MIC90' were chosen -- so no MIC can be sourced to split the ratio into numerator and denominator. Neither between-subject variability nor a residual error magnitude was reported (the +/- values in Table 3 are standard deviations of point estimates across the n = 3 animals whose plasma was used, not estimated variance components), so there are no eta parameters and addSd is FIXED at 0; the model is intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2023_tylosin.html">Tylosin Apleuropneumoniae infected (Lee 2023)</a> </td> <td style="text-align:left;"> Preclinical (pig, crossbred Duroc x (Landrace x Yorkshire)). Ex vivo sigmoidal Emax PK/PD-integration model for the antibacterial effect of tylosin against Actinobacillus pleuropneumoniae (challenge isolate BA2000013) in plasma drawn from co-infected pigs after a single 20 mg/kg intramuscular dose. Lee 2023 Section 2.8 parameterises the drug effect over a 24 h ex vivo incubation as E = E0 - (Emax * C^gamma) / (C^gamma + EC50^gamma), where E is the SIGNED log10(CFU/mL) difference between 0 and 24 h of incubation (positive = net growth, negative = net kill), E0 is that same 24 h change in the drug-free control, and C is the PK/PD index AUC24h/MIC. Parameters from Lee 2023 Table 3, column 'A. pleuropneumoniae / Infected': Emax = 7.34 log10 CFU/mL, EC50 = 1.33 h, E0 = 2.76 log10 CFU/mL, gamma = 1.86. The parameterisation was confirmed numerically against the paper's own independently printed rows: Emax - E0 reproduces exactly (4.58), and solving E = 0 and E = -3 returns AUC24h/MIC = 1.013 and 2.666 h versus the published 1.03 and 2.54 h. The bacterial density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * (E / 24) * bact so that log10(bact) changes by exactly E across each 24 h window, reproducing the paper's model at the only times bacteria were counted. There is NO PK component: Lee 2023 analysed the plasma concentrations non-compartmentally in WinNonlin 8.3 (Section 2.7; Table 2 reports only Cmax, Tmax, T1/2, AUC, Vz/F, Cl/F and MRT) and published no structural PK model, so exposure enters as the externally supplied covariate AUCMIC_TYLO. That covariate carries the AUC24h/MIC RATIO directly because Lee 2023 never reports the challenge isolates' own MIC -- only that strains 'with MIC values similar to the MIC90' were chosen -- so no MIC can be sourced to split the ratio into numerator and denominator. Neither between-subject variability nor a residual error magnitude was reported (the +/- values in Table 3 are standard deviations of point estimates across the n = 3 animals whose plasma was used, not estimated variance components), so there are no eta parameters and addSd is FIXED at 0; the model is intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2023_tylosin.html">Tylosin Pmultocida healthy (Lee 2023)</a> </td> <td style="text-align:left;"> Preclinical (pig, crossbred Duroc x (Landrace x Yorkshire)). Ex vivo sigmoidal Emax PK/PD-integration model for the antibacterial effect of tylosin against Pasteurella multocida (challenge isolate BA1700127) in plasma drawn from healthy pigs after a single 20 mg/kg intramuscular dose. Lee 2023 Section 2.8 parameterises the drug effect over a 24 h ex vivo incubation as E = E0 - (Emax * C^gamma) / (C^gamma + EC50^gamma), where E is the SIGNED log10(CFU/mL) difference between 0 and 24 h of incubation (positive = net growth, negative = net kill), E0 is that same 24 h change in the drug-free control, and C is the PK/PD index AUC24h/MIC. Parameters from Lee 2023 Table 3, column 'P. multocida / Healthy': Emax = 6.98 log10 CFU/mL, EC50 = 1.17 h, E0 = 3.49 log10 CFU/mL, gamma = 4.78. The parameterisation was confirmed numerically against the paper's own independently printed rows: Emax - E0 reproduces exactly (3.49), and solving E = 0 and E = -3 returns AUC24h/MIC = 1.170 and 2.009 h versus the published 1.10 and 1.99 h. The bacterial density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * (E / 24) * bact so that log10(bact) changes by exactly E across each 24 h window, reproducing the paper's model at the only times bacteria were counted. There is NO PK component: Lee 2023 analysed the plasma concentrations non-compartmentally in WinNonlin 8.3 (Section 2.7; Table 2 reports only Cmax, Tmax, T1/2, AUC, Vz/F, Cl/F and MRT) and published no structural PK model, so exposure enters as the externally supplied covariate AUCMIC_TYLO. That covariate carries the AUC24h/MIC RATIO directly because Lee 2023 never reports the challenge isolates' own MIC -- only that strains 'with MIC values similar to the MIC90' were chosen -- so no MIC can be sourced to split the ratio into numerator and denominator. Neither between-subject variability nor a residual error magnitude was reported (the +/- values in Table 3 are standard deviations of point estimates across the n = 3 animals whose plasma was used, not estimated variance components), so there are no eta parameters and addSd is FIXED at 0; the model is intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lee_2023_tylosin.html">Tylosin Pmultocida infected (Lee 2023)</a> </td> <td style="text-align:left;"> Preclinical (pig, crossbred Duroc x (Landrace x Yorkshire)). Ex vivo sigmoidal Emax PK/PD-integration model for the antibacterial effect of tylosin against Pasteurella multocida (challenge isolate BA1700127) in plasma drawn from co-infected pigs after a single 20 mg/kg intramuscular dose. Lee 2023 Section 2.8 parameterises the drug effect over a 24 h ex vivo incubation as E = E0 - (Emax * C^gamma) / (C^gamma + EC50^gamma), where E is the SIGNED log10(CFU/mL) difference between 0 and 24 h of incubation (positive = net growth, negative = net kill), E0 is that same 24 h change in the drug-free control, and C is the PK/PD index AUC24h/MIC. Parameters from Lee 2023 Table 3, column 'P. multocida / Infected': Emax = 7.04 log10 CFU/mL, EC50 = 1.06 h, E0 = 3.41 log10 CFU/mL, gamma = 2.90. The parameterisation was confirmed numerically against the paper's own independently printed rows: Emax - E0 reproduces exactly (3.63), and solving E = 0 and E = -3 returns AUC24h/MIC = 1.037 and 2.359 h versus the published 1.12 and 2.36 h. The bacterial density bact (linear CFU/mL) is integrated as d/dt(bact) = ln(10) * (E / 24) * bact so that log10(bact) changes by exactly E across each 24 h window, reproducing the paper's model at the only times bacteria were counted. There is NO PK component: Lee 2023 analysed the plasma concentrations non-compartmentally in WinNonlin 8.3 (Section 2.7; Table 2 reports only Cmax, Tmax, T1/2, AUC, Vz/F, Cl/F and MRT) and published no structural PK model, so exposure enters as the externally supplied covariate AUCMIC_TYLO. That covariate carries the AUC24h/MIC RATIO directly because Lee 2023 never reports the challenge isolates' own MIC -- only that strains 'with MIC values similar to the MIC90' were chosen -- so no MIC can be sourced to split the ratio into numerator and denominator. Neither between-subject variability nor a residual error magnitude was reported (the +/- values in Table 3 are standard deviations of point estimates across the n = 3 animals whose plasma was used, not estimated variance components), so there are no eta parameters and addSd is FIXED at 0; the model is intended for typical-value simulation. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zuo_2016_UDCA.html">UDCA (Zuo 2016)</a> </td> <td style="text-align:left;"> Systems model. Enterohepatic recirculation of ursodeoxycholic acid (UDCA) and its glycine (GUDCA) and taurine (TUDCA) conjugates in healthy adults, with adaptation to primary biliary cirrhosis (PBC). 19 ODEs across stomach, intestine, portal vein, blood, liver, biliary system, and feces compartments per analyte; oral square-wave absorption (0.5 h) and meal/snack-modulated biliary-to-intestinal flux. No IIV or residual error - typical-value mechanistic simulation only. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kim_2016_udenafil.html">Udenafil (Kim 2016)</a> </td> <td style="text-align:left;"> Parent-metabolite population PK model for oral udenafil and its active metabolite DA-8164 in healthy subjects and patients with mild (Child-Pugh A) and moderate (Child-Pugh B) hepatic impairment (Kim 2016). Two-compartment udenafil with first-order absorption and an absorption lag time, two parallel parent-side clearances (CLp/F = non-metabolic apparent clearance, CLpm/F = apparent formation clearance to DA-8164) feeding a two-compartment metabolite. Central and peripheral apparent volumes are assumed equal for parent and metabolite (the fraction metabolised f_m and the metabolite volume of distribution are not separately identifiable from this dataset). Mass-balance is preserved by multiplying the formation flux into the metabolite central compartment by the molecular-weight ratio Rpm = MW(DA-8164) / MW(udenafil) = 405.4 / 516.66. Prothrombin time expressed as INR (PT) acts on CLpm/F via a power covariate normalised to the cohort median 1.13: CLpm/F = theta1 * (PT/1.13)^theta10 with theta10 = -1.65 (decrease in CLpm/F with increasing PT). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Galluppi_2021_ulotaront.html">Ulotaront (Galluppi 2021)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order oral absorption for ulotaront (SEP-363856), a trace amine-associated receptor 1 (TAAR1) agonist with 5-HT1A agonist activity in phase III development for schizophrenia. Pooled analysis of nine studies (seven phase I, one phase II acute, one 6-month open-label extension) in 404 adult subjects (99 healthy volunteers and 305 patients with schizophrenia). Body weight was estimated as a power-form covariate on the clearance parameters (CL/F, Q/F) and the volume parameters (Vc/F, Vp/F); disease status, sex, race (Asian vs non-Asian), and age were retained as full-model covariates on CL/F only. IIV on CL/F, Vc/F, ka, Vp/F is modelled as a full 4x4 correlated BLOCK. Residual error is proportional-only per AIC/BIC (Galluppi 2021). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/AlSallami_2016_unfractionatedHeparin.html">UnfractionatedHeparin (AlSallami 2016)</a> </td> <td style="text-align:left;"> One-compartment population PK + linear pharmacodynamic model for unfractionated heparin (UFH) in paediatric patients receiving a single high intravenous bolus dose during cardiac angiography (Al-Sallami 2016). Fat-free mass (FFM) scales heparin clearance linearly and total body weight (WT) scales the central volume of distribution linearly. The PD layer is a linear concentration-effect model relating activated partial thromboplastin time (aPTT) to plasma heparin concentration (E0 + slope x Cc). The IV bolus was modelled in the source paper as a 0.1 h zero-order input (theta_D1 = 0.1 h, fixed); reproduce this in simulation by dosing the central compartment with rate = -2 to engage the model-defined duration. PD parameters were estimated sequentially via PPP&D with the PK parameters fixed at the values reported in Table 2. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jia_2015_unfractionatedHeparin.html">UnfractionatedHeparin (Jia 2015)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order elimination for unfractionated heparin (UFH) administered as multiple intravenous bolus injections during cardiopulmonary bypass (CPB) in adult Chinese cardiac surgery patients (Jia 2015). Plasma UFH exposure was inferred from anti-FIIa chromogenic activity. No covariates were retained in the final model (age, body weight, and sex were tested via forward inclusion / backward elimination and none met the p < 0.001 retention threshold). Concentrations are reported in IU/mL of anti-FIIa activity; doses are in IU (1 mg UFH = 125 IU). The published model also describes instantaneous neutralization of central-compartment UFH at protamine sulfate dosing (see vignette for the simulation pattern); the structural ODEs here are the standard two-compartment IV bolus form. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Klunder_2017_upadacitinib.html">Upadacitinib (Klunder 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and an absorption lag time for oral upadacitinib (ABT-494), a selective JAK1 inhibitor, in healthy adults and adults with rheumatoid arthritis (Klunder 2017, pooled phase I + phase IIb analysis). Statistically significant covariates retained in the final model: population (RA vs healthy) on CL/F, sex on CL/F and Vc/F, baseline creatinine clearance on CL/F (raw Cockcroft-Gault, not BSA-normalized), and total body weight on Vc/F. ISV is reported separately for healthy subjects and RA patients on CL/F and Vc/F, and is encoded here as paired healthy / RA structural means with cohort-specific log-normal random effects gated by DIS_HEALTHY. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/vandenBerg_2021_uprifosbuvir.html">Uprifosbuvir pbpk (vandenBerg 2021)</a> </td> <td style="text-align:left;"> PBPK-PD (minimal PBPK; well-stirred liver + Qgut gut wall + intrahepatic UXP/CXP nucleotide cycle) for uprifosbuvir (MK-3682), an HCV NS5B nucleoside polymerase inhibitor, and its plasma metabolites M5 and M6, with a sigmoid Emax link from intrahepatic UXP to viral inhibition (epsilon). 22 ODE compartments (gut absorption + transit chains, hepatic + portal-vein + central + two peripherals for parent, intrahepatic UXP/CXP cycle, M5 and M6 central + M5 gut transit chain, pseudo-M4 gut pool). All parameters fixed at the final NONMEM point estimates (Tables 1-6 of van den Berg 2021). Supports i.v. (25 mg), tablet oral (150-750 mg) and capsule oral (50-400 mg) dosing in HV and HCV patients, with optional itraconazole DDI encoded via covariate multipliers. All amounts are in nmol; concentrations in nmol/L. Framework adapted from the Brill et al. 2016 midazolam mPBPK with metabolite. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Aguiar_2021_ustekinumab.html">Ustekinumab (Aguiar 2021)</a> </td> <td style="text-align:left;"> Population pharmacokinetic-pharmacodynamic model for ustekinumab in adults with Crohn's disease (Aguiar 2021): two-compartment quasi-equilibrium TMDD model for ustekinumab and the unbound IL-12/IL-23 p40 target, linked to fecal calprotectin via an indirect-response model with target-driven stimulation of FC production. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Marcantonio_2022_efa.html">Ustekinumab (Marcantonio 2022)</a> </td> <td style="text-align:left;"> QSP. One-compartment monospecific anti-ligand mechanistic PKPD model of ustekinumab-p40 binding in adults with plaque psoriasis (Marcantonio 2022 Early Feasibility Assessment, Case Study 3). Ustekinumab binds the shared p40 subunit of IL-12 and IL-23 with bivalent binding (valency = 2 per Benson 2011); model treats p40 as the soluble target and its cognate receptor (IL-12R-beta1) as the membrane binding partner. Structure identical to the Marcantonio 2022 anti-ligand family; parameters FIXED from paper Table S3. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vezina_2010_valganciclovir.html">Valganciclovir (Vezina 2010)</a> </td> <td style="text-align:left;"> One-compartment population PK model for ganciclovir following oral valganciclovir prophylaxis in pediatric solid organ transplant recipients at risk for Epstein-Barr virus disease (Vezina 2010). First-order absorption with no covariates retained in the final model; doses are mg of valganciclovir uncorrected for molecular weight, and the apparent CL/F and V/F absorb both oral bioavailability and the molar conversion from valganciclovir to ganciclovir. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Vezina_2014_valganciclovir.html">Valganciclovir (Vezina 2014)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for ganciclovir after oral valganciclovir prophylaxis in paediatric and adult solid organ transplant recipients (Vezina 2014). First-order absorption with fixed lag time and rate, allometric (WT/70 kg) scaling on apparent CL/F and Q/F (exponent 0.75) and on V2/F and V3/F (exponent 1.0), and a power-form effect of body-weight-adjusted creatinine clearance on CL/F (reference 60 mL/min). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2023_valproic_acid_protein_binding.html">Valproic acid base (Zhang 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for total plasma valproic acid in Chinese children with epilepsy (Zhang 2023 base model). Linear clearance with no covariates on CL/F or V/F; formulation-specific absorption rate constants FIXED from the literature (syrup 2.64 1/h reference, conventional tablet 1.57 1/h, sustained-release tablet 0.46 1/h). This is the reference model against which the paper's five protein-binding non-linearity strategies are compared; see modellib('Zhang_2023_valproic_acid_exponent') for the daily-dose power model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2023_valproic_acid_protein_binding.html">Valproic acid ddemax (Zhang 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for total plasma valproic acid in Chinese children with epilepsy (Zhang 2023 Model III, the dose-dependent maximum effect model). Apparent clearance rises with the patient's own total daily dose through a sigmoid-Emax term, CL/F = CLp/F * (1 + Emax * DD^gamma / (DD50^gamma + DD^gamma)), with Emax, gamma and DD50 all FIXED to the values of Ding 2015. Formulation-specific absorption rate constants FIXED from the literature (syrup 2.64 1/h reference, conventional tablet 1.57 1/h, sustained-release tablet 0.46 1/h). The weakest of the five non-linearity strategies the authors compare - its objective function value is worse than the linear base model's - and it is packaged for fidelity to the published comparison; see modellib('Zhang_2023_valproic_acid_nonsaturable') for the strategy the authors recommend. DD50 is not reported by Zhang 2023 and is carried over from Ding 2015; see the vignette Errata. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2023_valproic_acid_protein_binding.html">Valproic acid exponent (Zhang 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for total plasma valproic acid in Chinese children with epilepsy (Zhang 2023 Model V, the simple exponent model). Apparent clearance follows a power function of the patient's own daily dose per kilogram, CL/F = CLp/F * (DD/25)^0.658, with DD in mg/kg/day and a reference daily dose of 25 mg/kg/day; this is the empirical dose-dependence strategy the authors contrast against mechanistic protein-binding models. Formulation-specific absorption rate constants FIXED from the literature (syrup 2.64 1/h reference, conventional tablet 1.57 1/h, sustained-release tablet 0.46 1/h). Best prediction-based performance of the five strategies (MDPE 1.50%, MAPE 17.68%) but the authors conclude it does not describe the underlying non-linearity; see modellib('Zhang_2023_valproic_acid_base') for the reference model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2023_valproic_acid_protein_binding.html">Valproic acid langmuir (Zhang 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for valproic acid in Chinese children with epilepsy (Zhang 2023 Model II, the Langmuir protein-binding model). The disposition is carried on the UNBOUND concentration Cu = central/(V/F) with linear unbound clearance; the observed TOTAL plasma concentration is reconstructed as Cc = Cu + Cb, where the bound concentration follows the single-site Langmuir isotherm Cb = Bm * Cu / (Kd + Cu) with Bm and Kd FIXED to the literature values of Ueshima 2008. Formulation-specific absorption rate constants FIXED from the literature (syrup 2.64 1/h reference, conventional tablet 1.57 1/h, sustained-release tablet 0.46 1/h). One of five protein-binding non-linearity strategies the authors compare; see modellib('Zhang_2023_valproic_acid_nonsaturable') for the linear non-saturable extension the authors ultimately preferred. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2023_valproic_acid_protein_binding.html">Valproic acid nonsaturable (Zhang 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for valproic acid in Chinese children with epilepsy (Zhang 2023 Model IV, the linear non-saturable binding model - the strategy the authors conclude is best suited to valproate protein-binding non-linearity). The disposition is carried on the UNBOUND concentration Cu = central/(V/F) with linear unbound clearance; the observed TOTAL plasma concentration is reconstructed as Cc = Cu + Cb, where the bound concentration adds a linear non-saturable term to the Langmuir isotherm, Cb = Bm * Cu / (Kd + Cu) + NS * Cu, with Bm, Kd and NS FIXED to the values of Gu 2021. Formulation-specific absorption rate constants FIXED from the literature (syrup 2.64 1/h reference, conventional tablet 1.57 1/h, sustained-release tablet 0.46 1/h). Best simulation-based behaviour of the five strategies compared; see modellib('Zhang_2023_valproic_acid_langmuir') for the pure Langmuir form it extends. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2023_valproic_acid_protein_binding.html">Valproic acid onebindingsite (Zhang 2023)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order absorption for valproic acid in Chinese children with epilepsy (Zhang 2023 Model I, the one-binding-site protein-binding model). The disposition is carried on the UNBOUND concentration Cu = central/(V/F) with linear unbound clearance; the observed TOTAL plasma concentration is reconstructed as Cc = Cu + Cb, where the albumin-bound concentration follows the one-binding-site isotherm Cb = N * K * Cu * ALB / (1 + K * Cu) with N and K FIXED to the adult literature values of Dutta 2007. Formulation-specific absorption rate constants FIXED from the literature (syrup 2.64 1/h reference, conventional tablet 1.57 1/h, sustained-release tablet 0.46 1/h). One of five protein-binding non-linearity strategies the authors compare; see modellib('Zhang_2023_valproic_acid_base') for the linear reference model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Williams_2012_valproic_acid_pediatric.html">Valproic acid pediatric (Williams 2012)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for valproic acid in pediatric patients with epilepsy (Williams 2012). Allometric weight scaling on CL/Q (fixed 0.75) and Vc/Vp (fixed 1.0); estimated age power (-0.267) on Vc; reference weight 70 kg, reference age 8.5 years. Default first-order oral absorption is for divalproex sodium enteric-coated sprinkle (Ka 1.2 1/h, ALAG 1 h, FIXED); other formulations (syrup K0=410 mg/h, capsule Ka=2 1/h, tablet Ka=4.1 1/h with ALAG=2 h) require overriding lka/ltlag at simulation time. Direct IV dosing into the central compartment is supported. Residual error defaults to the TDM-subset proportional SD (CV 34.8%); paper also reports a TRIAL-subset SD (CV 4.6%) for the IV-infusion subset. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Kim_2015_valsartan.html">Valsartan (Kim 2015)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for valsartan with zero-order absorption in healthy adult Korean male volunteers (Kim 2015) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Alqahtani_2018_vancomycin.html">Vancomycin (Alqahtani 2018)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for vancomycin used as prophylactic antibiotic in 28 adult patients undergoing open heart surgery with cardiopulmonary bypass (Alqahtani 2018). Clearance scales by power exponent with Cockcroft-Gault creatinine clearance (raw mL/min, reference 83.5) and serum albumin (g/L, reference 35.5); central volume scales by power exponent with body weight (kg, reference 79.6). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Buelga_2005_vancomycin.html">Vancomycin (Buelga 2005)</a> </td> <td style="text-align:left;"> One-compartment IV intermittent-infusion population PK model for vancomycin in adult patients with hematological malignancies (Buelga 2005). CL is a purely multiplicative function of Cockcroft-Gault creatinine clearance (CL [L/h] = 1.08 x CLCR [L/h]) and V is a purely multiplicative function of total body weight (V [L] = 0.98 x TBW [kg]). Exponential inter-individual variability on CL and V with an estimated CL-V correlation; additive residual error in mg/L. The AML-1 and AML-2 subpopulation-specific models from the same paper are not packaged here; only the general final model (Table 4) is implemented. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chung_2013_vancomycin.html">Vancomycin (Chung 2013)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for vancomycin in Korean adults with normal serum creatinine (Chung 2013). CL and V are described by centered-linear additive deviations on age, total body weight, serum creatinine (CL only), and sex, plus a power-law effect of serum cystatin C on CL (reference 0.91 mg/L, exponent -0.78); cystatin C is the dominant CL covariate, accounting for ~62% of the inter-individual variability in CL even within the SCr <= 1.2 mg/dL inclusion window. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Goti_2018_vancomycin.html">Vancomycin (Goti 2018)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for vancomycin in hospitalized adults with and without intermittent hemodialysis (Goti 2018). Volumes scaled allometrically to body weight (reference 70 kg, fixed linear exponent), CL scaled by Cockcroft-Gault creatinine clearance with a power exponent (reference 120 mL/min), and intermittent hemodialysis acts as a multiplicative factor on CL (0.7) and central volume (0.5). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Grimsley_1999_vancomycin.html">Vancomycin (Grimsley 1999)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for vancomycin in neonates and young infants (Grimsley 1999). Developed from routine therapeutic-drug-monitoring data in 59 neonates (347 concentrations). Clearance scales linearly with body weight and inversely with serum creatinine concentration (CL = 3.56 * WT / CREAT, L/h, WT in kg, CREAT in umol/L); central volume scales linearly with body weight (V = 0.669 * WT, L/kg). The covariate-coupled CL form (no separately estimated exponents) is reported by the paper as the entire structural model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ji_2017_vancomycin.html">Vancomycin (Ji 2017)</a> </td> <td style="text-align:left;"> One-compartment IV (intermittent-infusion) population PK model for vancomycin in Chinese adult patients (Ji 2017). Clearance is scaled by raw Cockcroft-Gault creatinine clearance (centered linear term, reference 80 mL/min) and by age (power of (75/age), reference 75 years); the volume of distribution is a single typical value. Developed from steady-state trough therapeutic-drug-monitoring data. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Jung_2024_vancomycin.html">Vancomycin (Jung 2024)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for vancomycin in critically ill children (90 days to <18 years) in the PICU who are not on extracorporeal therapy (Jung 2024). Clearance and intercompartmental clearance scale allometrically with body weight (exponent 0.75, reference 20 kg) and clearance additionally scales as a power function of bedside-Schwartz estimated GFR (exponent 0.5259, reference 141 mL/min/1.73 m^2); central and peripheral volumes scale linearly with body weight (exponent 1, reference 20 kg). Residual variability is a log error model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Li_2018_vancomycin.html">Vancomycin (Li 2018)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for vancomycin in critically ill Chinese ICU neonates (Li 2018). CL scales allometrically with body weight (reference 2.9 kg, exponent 1.55) and as an inverse power of serum creatinine (reference 23.3 umol/L, exponent 0.337 on the SCr_ref/SCr ratio). V scales allometrically with body weight (reference 2.9 kg, exponent 1.05). IIV is on CL only; residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/MarquesMinana_2010_vancomycin.html">Vancomycin (MarquesMinana 2010)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for vancomycin in neonates (Marques-Minana 2010). Developed from 70 NICU neonates (postmenstrual age 25.1-48.1 weeks; weight 0.7-3.7 kg). Weight-normalized clearance is linear in postmenstrual age and increased by concomitant amoxicillin-clavulanic acid; weight-normalized volume of distribution is decreased by concomitant spironolactone. Additive interindividual variability on CL and V per the paper's Step 4 error-model selection; additive residual error. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Moore_2016_vancomycin.html">Vancomycin (Moore 2016)</a> </td> <td style="text-align:left;"> Two-compartment IV population PK model for vancomycin in adult patients on extracorporeal membrane oxygenation (ECMO) therapy (Moore 2016). Linear (additive) covariate effects on CL (Cockcroft-Gault creatinine clearance), Vc, and Vp (body weight), each centered on the cohort median (CRCL 84 mL/min; WT 95 kg). Proportional residual error; IIV on CL and Vc only (Q and Vp had no IIV). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Nielsen_2011_antibacterial_efficacy.html">Vancomycin (Nielsen 2011)</a> </td> <td style="text-align:left;"> In vitro (Streptococcus pyogenes M12 NCTC P1800). Semimechanistic PKPD model of vancomycin time-kill kinetics; two-stage bacterial life-cycle (proliferating drug-sensitive S and non-growing drug-insensitive R) with sigmoidal Emax killing of S via an effect compartment; first-order drug elimination (ke set per in vitro kinetic-system flow rate); drug-specific degradation kdeg fixed at zero. Parameter values are from the combined static and dynamic estimation in Table 3. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Revilla_2010_vancomycin.html">Vancomycin (Revilla 2010)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for vancomycin in critically ill adult medical ICU patients (Revilla 2010). Clearance is the sum of a renal arm proportional to weight-normalised creatinine clearance and a non-renal arm scaling as AGE^-0.24; central volume of distribution is per kg with a >2-fold increase when serum creatinine exceeds 1 mg/dL. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Roberts_2011_vancomycin.html">Vancomycin (Roberts 2011)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for vancomycin administered by continuous infusion in adult septic critically ill ICU patients (Roberts 2011). Volume of distribution scales linearly with total body weight (1.53 L/kg); clearance scales linearly with BSA-normalized 24-hour urinary creatinine clearance referenced to 100 mL/min/1.73 m^2 (4.58 L/h at the reference). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Shen_2024_vancomycin.html">Vancomycin (Shen 2024)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for vancomycin in 386 Southern Chinese children (Shen 2024). Clearance uses an age-cutoff structure at 2 years: a separate typical clearance AND a separate body-weight allometric exponent are estimated in each age stratum (2.59 L/h with exponent 0.38 for age > 2 years; 1.98 L/h with exponent 0.739 for age <= 2 years), both normalized to a 12 kg reference weight, with a shared Cockcroft-Gault creatinine-clearance power effect (exponent 0.517, reference 75 mL/min). Central volume scales linearly with body weight (22.4 L at 12 kg). Additive residual error is estimated separately for each of the two study centers. Both strata share the volume, the CLcr exponent and the clearance IIV in a single joint NONMEM fit. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Staatz_2005_gentamicin_vancomycin.html">Vancomycin (Staatz 2005)</a> </td> <td style="text-align:left;"> One-compartment IV population PK model for vancomycin in adult cardiothoracic-surgery patients with unstable renal function (Staatz 2005). Clearance scales linearly with raw Cockcroft-Gault creatinine clearance centred at the population median (57 mL/min); volume of distribution scales linearly with body weight (typical-value reported per kg). Vancomycin did not benefit from the Wahlby 2004 baseline-CrCl + change-from-baseline (BCOV+DCOV) decomposition in the paper -- the simpler covariate form was retained as the final vancomycin model -- so this implementation reproduces the paper's published vancomycin final model exactly. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhao_2014_vancomycin.html">Vancomycin (Zhao 2014)</a> </td> <td style="text-align:left;"> One-compartment IV-infusion population PK model for vancomycin in 70 children with malignant hematological disease (Zhao 2014). Clearance scales with body weight by power exponent (reference 20.2 kg, exponent 0.677) and with Schwartz-formula creatinine clearance by power exponent (reference 191 mL/min/1.73 m^2, exponent 1.03); central volume scales with body weight by power exponent (reference 20.2 kg, exponent 0.838). Vancomycin clearance was substantially higher than in pediatric populations without cancer; the published patient-tailored daily dose is target AUC * CL_i. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Ravva_2009_varenicline.html">Varenicline (Ravva 2009)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order absorption and lag-time for varenicline in adult smokers (Ravva 2009): apparent clearance scales with creatinine clearance and race; central volume scales with body weight, age, and race; peripheral disposition uses fixed allometric exponents on weight. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wang_2014_vatalanib.html">Vatalanib (Wang 2014)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral vatalanib in adults with myelodysplastic syndrome (CALGB 10105); apparent oral clearance carries a first-order auto-induction term that rises from a pre-induction value toward a steady-state post-induction value over the first 7 days of therapy, lagged first-order absorption, log-normal residual error on the natural-log-transformed concentration; no covariates retained in the final model (Wang 2014). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Rosario_2015_vedolizumab.html">Vedolizumab (Rosario 2015)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for vedolizumab (humanised anti-alpha4-beta7 integrin IgG1 monoclonal antibody) with parallel linear and Michaelis-Menten elimination in adults with moderately-to-severely active ulcerative colitis or Crohn's disease and healthy volunteers (Rosario 2015). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Waterhouse_2024_vedolizumab.html">Vedolizumab (Waterhouse 2024)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with first-order (linear) elimination for vedolizumab (humanised anti-alpha4-beta7 integrin IgG1 monoclonal antibody) as acute graft-versus-host disease (aGvHD) prophylaxis in adults undergoing allogeneic hematopoietic stem cell transplantation (allo-HSCT) (Waterhouse 2024). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Syvanen_2011_verapamil_rat.html">Verapamil rat (Syvanen 2011)</a> </td> <td style="text-align:left;"> Preclinical (rat, male Sprague-Dawley). Population mixed-effects popPK model for (R)-[11C]verapamil in plasma and whole-brain PET tissue, fit by Syvanen et al. (2011, BMC Med Imaging) as part of a PET study comparing P-glycoprotein (P-gp) functionality at the blood-brain barrier between kainate-induced post-status-epilepticus rats (n = 22) and saline-treated controls (n = 20), with paired tariquidar (15 mg/kg IV) vs vehicle co-administration arms. The structural model is a three-compartment plasma disposition (central + 2 peripherals) coupled to a two-compartment brain model (brain_csf = fast-exchange brain compartment connected to plasma via Qin in / Qout out; brain_deep = deep-brain compartment exchanging with brain_csf via Qbr). Plasma curves are complete-metabolite- corrected before fitting, so the model describes intact (R)-[11C]verapamil kinetics only. Body weight is the only continuous covariate (allometric on plasma CL, reference weight 0.3084 kg). Tariquidar co-administration multiplies Vp1 by 1.20, Vbr1 by 2.41, and Qin by 12.0; the kainate-induced post-SE state multiplies Vbr1 by 1.32 (no significant effect on Qin or Qout); both categorical effects use the paper's theta^COV multiplicative form (Equation 5). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Landersdorfer_2012_vildagliptin.html">Vildagliptin (Landersdorfer 2012)</a> </td> <td style="text-align:left;"> Mechanism-based population PK plus DPP-4 activity model for vildagliptin in patients with type 2 diabetes. Target-mediated drug disposition with capacity-limited slow-tight binding of vildagliptin to DPP-4 in plasma and tissue and partial hydrolysis of vildagliptin by DPP-4. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Schmitt_2018_vinflunine.html">Vinflunine (Schmitt 2018)</a> </td> <td style="text-align:left;"> Combined population PK / PD model for IV vinflunine in adult cancer patients (Schmitt 2018, 18 phase I/II trials, n=372). Four-compartment IV-infusion popPK with creatinine clearance, body surface area, body weight, and PEGylated liposomal doxorubicin co-administration covariates, plus a five-compartment Friberg-style semi-mechanistic myelosuppression PD model for absolute neutrophil count (proliferation + 3 transit + circulation; linear drug effect 1 - slope*Cc on proliferation; (circ0/circ)^gamma feedback). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Petric_2023_vinpocetine.html">Vinpocetine (Petric 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for apovincaminic acid (AVA), the active de-esterified metabolite of vinpocetine, in healthy adult male volunteers dosed with vinpocetine (Petric 2023). Only AVA (not the parent vinpocetine) is modelled; the reported CL/F, V1/F, Q/F, V2/F apparent parameters fold the fraction of the vinpocetine dose that appears in plasma as AVA and the vinpocetine oral bioavailability into the /F term. Absorption is described as a zero-order input of duration Tk0 into the central compartment preceded by an absorption lag Tlag. Formulation is the only significant covariate: the sustained-release beta-cyclodextrin complex Ultra Vinca is the reference; the Cavinton immediate-release tablet and the extemporaneous 10 mg / 5 mL oral solution enter as log-additive shifts on Tk0 and V1/F. Between-subject variability is placed on Tlag, Tk0, CL/F, V1/F, Q/F, V2/F with a correlation of 0.72 between CL/F and V1/F. Residual error is proportional. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Cao_2013_visilizumab.html">Visilizumab (Cao 2013)</a> </td> <td style="text-align:left;"> Second-generation minimal physiologically-based PK (mPBPK) model for visilizumab in adults (Cao 2013 Model A; clearance from plasma) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lu_2015_vismodegib.html">Vismodegib (Lu 2015)</a> </td> <td style="text-align:left;"> Semi-mechanism-based one-compartment population pharmacokinetic model for vismodegib (GDC-0449, oral Hedgehog pathway inhibitor) in adults with advanced solid tumors and healthy volunteers. First-order absorption, first-order elimination of unbound drug, and saturable fast-equilibrium binding to alpha-1-acid glycoprotein (AAG) jointly describe total and unbound plasma vismodegib concentrations. AAG is supplied as a time-varying covariate (uM); covariates retained on disposition are age (power on CLunbound, reference 60 years) and body weight (power on Vc, reference 75 kg); formulation (Phase I dry-blend capsule vs Phase II wet-granulation commercial capsule) and population (healthy volunteer vs patient) shift Ka and relative bioavailability F (Lu 2015). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bauer_2023_vonWillebrandFactor.html">VonicogAlfa (Bauer 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for von Willebrand factor:ristocetin cofactor activity (VWF:RCo) after intravenous recombinant von Willebrand factor (rVWF, vonicog alfa) coupled to an indirect-response PK/PD model for endogenous factor VIII activity (FVIII:C), from Bauer 2023. VWF:RCo disposition is linear two-compartment with first-order elimination from the central compartment plus an additive endogenous background activity E_VWF, fixed at half the assay LLOQ (0.5 IU/dL) for von Willebrand disease (VWD) type 3. CL and Q are allometrically scaled by body weight with a fixed exponent of 0.75, Vc and Vp with a fixed exponent of 1, both referenced to 75 kg; Vc additionally decreases with hematocrit as (HCT/40)^-0.334. FVIII:C is a turnover pool with zero-order production kin = FVIII0 * kout and first-order removal kout (15.9 1/h) that VWF:RCo inhibits through an Imax function, 1 - Imax * VWF:RCo / (IC50 + VWF:RCo), so that rising VWF:RCo protects FVIII from clearance; FVIII0 is the theoretical baseline FVIII:C at VWF:RCo = 0 IU/dL and scales with hematocrit as (HCT/39)^-0.571. Fitted to 1664 VWF:RCo samples from 79 patients across four studies (VWD types 1, 2, 3 and severe hemophilia A) and 686 FVIII:C samples from the 41 patients of the two phase 3 VWD studies. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Akbar_2025_voriconazole.html">Voriconazole (Akbar 2025)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with first-order elimination for intravenous voriconazole in adult and pediatric Pakistani cancer patients receiving therapeutic drug monitoring (Akbar 2025); creatinine clearance and primary cancer diagnosis are covariates on clearance </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Chen_2015_voriconazole.html">Voriconazole (Chen 2015)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with first-order elimination for intravenous voriconazole in Chinese adult critically ill patients with pulmonary disease (Chen 2015); direct bilirubin enters as a power-form covariate on clearance. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Friberg_2012_voriconazole.html">Voriconazole (Friberg 2012)</a> </td> <td style="text-align:left;"> Integrated population pharmacokinetic model for voriconazole in children, adolescents, and adults (Friberg 2012). Two-compartment with first-order oral absorption and mixed linear plus nonlinear (Michaelis-Menten with time-dependent Vmax) elimination; allometric scaling on all clearance terms (exponent 0.75) and on volumes (exponent 1.0) with 70 kg reference; population-specific Vmax,inh, Q, ka, and Alag for children, adolescents, and adults; CYP2C19 heterozygous extensive or poor metabolizer adults have fully blocked nonlinear clearance (Vmax,inh = 100%). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Han_2010_voriconazole.html">Voriconazole (Han 2010)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model with first-order absorption and first-order elimination for intravenous and oral voriconazole in adult lung transplant recipients during the early postoperative period (Han 2010). Bioavailability is estimated for the oral route. The base structural model is reported as the primary result; three separate single-covariate sub-models -- cystic fibrosis (CF) and postoperative time (POT) on bioavailability, and body weight (WT) on peripheral volume -- are reported in the paper but were not combined into a final model; the base-model typical-value parameter estimates are encoded here, and the three covariate sub-models are reproduced in the validation vignette. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Karlsson_2009_voriconazole.html">Voriconazole (Karlsson 2009)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model with Michaelis-Menten elimination for voriconazole in pediatric patients aged 2 to <12 years (Karlsson 2009), pooled from three open-label intravenous and oral studies; first-order oral absorption with bioavailability, no lag time; all disposition parameters proportional to body weight; CYP2C19 metabolizer status (heterozygous extensive metabolizers pooled with poor metabolizers) and alanine aminotransferase as covariates on clearance; residual error stratified by CYP2C19 metabolizer group </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lin_2018_voriconazole.html">Voriconazole (Lin 2018)</a> </td> <td style="text-align:left;"> One-compartment population pharmacokinetic model with first-order absorption for intravenous and oral voriconazole in Chinese adult renal transplant recipients receiving therapeutic drug monitoring (Lin 2018); CYP2C19 phenotype enters as a covariate on clearance, postoperative time as a covariate on oral bioavailability, and body weight as a power-form covariate on volume of distribution. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Muto_2015_voriconazole.html">Voriconazole (Muto 2015)</a> </td> <td style="text-align:left;"> Two-compartment population pharmacokinetic model with first-order absorption (lag time, oral bioavailability) and parallel linear plus time-dependent Michaelis-Menten elimination for voriconazole in 21 immunocompromised Japanese pediatric subjects (Muto 2015). Vmax declines with time after the first dose toward Vmax * (1 - Vmax_inh) with half-time T50; the maximum inhibition fraction Vmax_inh is fixed to 1 (full inhibition) for CYP2C19 heterozygous-extensive-metabolizer or poor-metabolizer subjects and modeled on the logit scale otherwise. Allometric scaling on all clearances (exponent 0.75) and all volumes (exponent 1) to a 70 kg reference; oral bioavailability F1 is modeled on the logit scale with a Manly-transformed log-normal random effect. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Wahlby_2004_time_varying_covariates.html">Voriconazole (Wahlby 2004)</a> </td> <td style="text-align:left;"> Pediatric (2-11 years) two-compartment population PK model for intravenous voriconazole in 35 children, demonstrating Wahlby 2004's extended covariate-model formulation. All disposition parameters scale linearly with body weight. Final-model clearance depends on the time-varying log-ratio (log(ALP/ALP_BASE), 'log(DALKP)' in the source) and on log(ALT) with individual variability in both covariate-effect coefficients (Wahlby 2004 Eq 3 demonstrated). A binary CYP2C19 non-extensive-metabolizer indicator (PM + heterozygous-EM versus homozygous-EM) multiplicatively modifies CL. Underlying structural PK comes from Walsh TJ et al. (Antimicrob Agents Chemother 2004;48(6):2166-2172) and the Karlsson 1995 (J Pharmacokin Biopharm 1998;26(2):207-246) sigma-IIV residual-error pattern is approximated in this entry. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Naik_2016_vortioxetine.html">Vortioxetine (Naik 2016)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for vortioxetine in adult patients with major depressive disorder or generalized anxiety disorder, with first-order oral absorption, region-specific oral clearance, and linear creatinine-clearance and height effects on CL/F (Naik 2016) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Qi_2024_vosoritide.html">Vosoritide (Qi 2024)</a> </td> <td style="text-align:left;"> One-compartment population PK model with first-order elimination and change-point first-order absorption (the absorption rate constant switches from Ka1 to Ka2 at an estimated time after each dose) for subcutaneous vosoritide (BMN 111, a C-type natriuretic peptide analog) in children with achondroplasia aged 0.95-15 years, pooled from five clinical trials (Qi 2024). Body weight is a power covariate on CL/F (exponent 0.356) and on V/F (exponent 1.09), both referenced to 20 kg. Relative bioavailability rises exponentially with time on treatment and is 56% higher for the 0.2 mg/mL dosing solution used only in study 111-202. Study-level random effects nested inside the subject-level IIV on CL/F and V/F reproduce the paper's secondary study identity number (SIDN) hierarchy, and separate log-scale residual errors are carried for the ELISA and the electrochemiluminescence assays. The model was used to derive the eight-band weight-band dosing regimen of Qi 2024 Table 6. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Huang_2017_vrc01.html">Vrc01 (Huang 2017)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for VRC01 (HIV-1 broadly neutralizing IgG1 monoclonal antibody) in healthy adults after IV or SC administration (Huang 2017) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Huynh_2026_VRC07523LS.html">VRC07523LS (Huynh 2026)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with zero-order subcutaneous absorption, allometric weight scaling, and binary effects of age (adult vs infant) and repeat dosing for the broadly neutralizing HIV-1 monoclonal antibody VRC07-523LS in healthy adults and HIV-exposed infants (Huynh 2026). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Bauer_2023_vonWillebrandFactor.html">VwfFviii humateP (Bauer 2023)</a> </td> <td style="text-align:left;"> Two-compartment population PK model for von Willebrand factor:ristocetin cofactor activity (VWF:RCo) after intravenous plasma-derived von Willebrand factor / factor VIII concentrate (pdVWF/FVIII; Humate-P, VWF:RCo/FVIII:C 2.4:1) coupled to an indirect-response PK/PD model for factor VIII activity (FVIII:C), from Bauer 2023. The structure is the one developed for recombinant VWF (see modellib('Bauer_2023_vonicogAlfa')) refitted to the plasma-derived product: linear two-compartment VWF:RCo disposition with first-order elimination from the central compartment plus an additive endogenous background E_VWF fixed at half the assay LLOQ (0.5 IU/dL) for von Willebrand disease (VWD) type 3, allometric body weight on CL and Q (exponent 0.75) and on Vc and Vp (exponent 1) with a 75 kg reference, and (HCT/40)^-0.334 on Vc. FVIII:C is a turnover pool whose first-order removal kout is inhibited by VWF:RCo through 1 - Imax * VWF:RCo / (IC50 + VWF:RCo). Because the product delivers FVIII as well as VWF, a volume of distribution for FVIII (V FVIII = 32.9 dL) is added so that the administered FVIII:C dose enters the FVIII pool; the elimination of plasma-derived FVIII is assumed identical to that of endogenous FVIII and is therefore carried by kout. The system-specific parameters FVIII0, kout and the hematocrit effect on FVIII0 are fixed to the recombinant-VWF model estimates. VWF:RCo clearance is roughly twice that of recombinant VWF (4.14 vs 2.10 dL/h), giving a 1.76-fold shorter mean residence time. Fitted to 281 VWF:RCo and FVIII:C samples from 20 patients with VWD type 3. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Savic_2010_warfarin.html">Warfarin (Savic 2010)</a> </td> <td style="text-align:left;"> Population PKPD model for orally dosed warfarin in adult subjects, presented as the worked illustration of MONOLIX 3.1's SAEM algorithm for ordered-categorical PD data. PK: one-compartment with first-order absorption and a lag time. PD link: effect compartment driven by central amount via rate constant ke0. PD endpoint: a three-category recoding of percent prothrombin complex activity (PCA) with cutoffs 50% and 33% (Y=0 if PCA > 50%, Y=1 if 33% <= PCA <= 50%, Y=2 if PCA < 33%), described by a proportional- odds (cumulative-logit) model with random intercept driven by effect-site warfarin concentration. The PD categorisation is acknowledged by the authors (Page 6) as 'done for illustration purpose only ... not recommended in the real analysis'; this extraction is the registry's founding example of an ordered- categorical PD likelihood and the authors' caveat applies. All parameter values are from the MONOLIX output in Fig. 4. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Xia_2024_warfarin.html">Warfarin (Xia 2024)</a> </td> <td style="text-align:left;"> K-PD warfarin PK/PD model for adult Han Chinese (Alfalfa-Warfarin-PPK/PD; Xia 2024). PK parameters fixed from the Hamberg model; PD EC50 re-estimated, with VKORC1 -1639 G/A and CYP2C9 *1/*2/*3 allele-specific contributions, body-weight power scaling, and amiodarone effect on EC50. Two parallel coagulation-factor transit chains drive INR. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Falkenhagen_2023_warfarin_qsp.html">Warfarin qsp (Falkenhagen 2023)</a> </td> <td style="text-align:left;"> QSP-derived mechanism-based warfarin/INR pharmacodynamic model (Falkenhagen 2023). Obtained by systematically reducing the 62-ODE / 174-parameter Wajima 2009 blood-coagulation quantitative systems pharmacology model down to 6 ODEs and 11 structural parameters, while guaranteeing under 10% relative INR error for at least 95% of a diverse virtual population. One-compartment oral warfarin PK inhibits vitamin K hydroquinone (VKH2) synthesis through an Imax function; VKH2 in turn drives the synthesis of coagulation Factors II, VII, and X, each modelled as a turnover pool holding its own pre-stimulus steady state. The INR is recovered algebraically as a power law in the product of the three relative factor concentrations, INR = INR0 * (II/II0 * VII/VII0 * X/X0)^gamma with gamma = -0.1975. CYP2C9 *1/*2/*3 allele counts set warfarin clearance and VKORC1 -1639 G/A allele counts set IC50, both as per-allele sums. All parameter values are the Wajima 2009 reference parameterization carried through the reduction; no parameter was estimated from clinical data in this paper. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hamberg_2007_warfarin_pkpd_pgx.html">Warfarin r (Hamberg 2007)</a> </td> <td style="text-align:left;"> R-warfarin population PK (1-compartment, first-order absorption) with age as the only structural covariate on CL_R (Hamberg 2007). R-warfarin was not found to contribute (additive or competitive) to the INR PD; the companion file Hamberg_2007_warfarin_s carries the S-warfarin PK and the INR PD model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lane_2011_warfarin.html">Warfarin r (Lane 2011)</a> </td> <td style="text-align:left;"> R-warfarin population PK (1-compartment, first-order absorption) in adults on long-term warfarin therapy (Lane 2011). Bodyweight, age, CYP2C19 rs3814637 genotype, and CYP3A4 rs2242480 (CYP3A4*1G) genotype influence apparent clearance; volume of distribution carries no covariates. Block correlation between random effects on CL and V. S-warfarin is reported separately in the same paper (modellib('Lane_2011_warfarin_s')). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Hamberg_2007_warfarin_pkpd_pgx.html">Warfarin s (Hamberg 2007)</a> </td> <td style="text-align:left;"> S-warfarin population PK (2-compartment, first-order absorption) coupled to an inhibitory-Emax INR PD model with two parallel transit-compartment chains (6 + 1) driving the anticoagulant response (Hamberg 2007). CYP2C9 genotype and age are predictors for S-warfarin clearance; VKORC1 -1639G>A genotype is a predictor for INR sensitivity (EC50). R-warfarin is reported separately (modellib('Hamberg_2007_warfarin_r')). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Lane_2011_warfarin.html">Warfarin s (Lane 2011)</a> </td> <td style="text-align:left;"> S-warfarin population PK (1-compartment, first-order absorption) in adults on long-term warfarin therapy (Lane 2011). Bodyweight, age, sex, and CYP2C9 diplotype influence apparent clearance; volume of distribution carries no covariates. Block correlation between random effects on CL and V. R-warfarin is reported separately in the same paper (modellib('Lane_2011_warfarin_r')). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhou_2016_warfarin_vk2.html">Warfarin vk2 (Zhou 2016)</a> </td> <td style="text-align:left;"> Two-drug population PK/PD model for warfarin and intravenous vitamin K2 (menatetrenone) in Japanese adults with atrial fibrillation undergoing catheter ablation. Warfarin and vitamin K2 each have a 1-compartment PK with fixed volumes-of-distribution (Vd1 = 0.183 L/kg for warfarin from Sato 2006; Vd3 = 0.051 L/kg for vitamin K2 from the Eisai product information) and fixed warfarin elimination rate (k10 = 0.0129 1/h); only the vitamin K2 elimination rate (k30) and the indirect-response PD parameters (ks, kd, IC50, Emax, EC50) were estimated from 579 INR observations in 100 patients. Warfarin inhibits clotting-factor synthesis (Emax = 1 - Cp1/(Cp1 + IC50)) while vitamin K2 stimulates it (1 + Emax_vk2 * Cp3/(Cp3 + EC50)); a binary renal-impairment indicator (CREAT >= 1.1 mg/dL in men or >= 0.8 mg/dL in women) reduces IC50 to 61.4% of normal. The model predicts thrombotest (TT, %); INR is recovered from TT via the Gogstad 1986 quadratic conversion (Equation 4). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Zhang_2024_XZP5610.html">XZP5610 (Zhang 2024)</a> </td> <td style="text-align:left;"> One-compartment oral PK model for the non-steroidal FXR agonist XZP-5610 (NASH) in healthy Chinese adults, forward-predicted from SD rat and beagle dog preclinical PK by allometric scaling (Zhang 2024): first-order absorption from a depot with bioavailability, linear elimination from central. Every parameter is a cross-species prediction used for first-in-human dose selection, not an estimate fitted to human data; the paper reports no IIV and no residual error for it. The companion whole-body PBPK model in the same paper was built in PK-Sim v11.2 and is not reproducible from the published sources (no ODEs, organ volumes, blood flows, or per-tissue partition coefficients are reported). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Adams_1998_zalcitabine.html">Zalcitabine (Adams 1998)</a> </td> <td style="text-align:left;"> One-compartment first-order-absorption population PK model for oral zalcitabine (ddC) in HIV-infected adults (Adams 1998). Apparent clearance (CL/F = 14.8 L/h) and apparent volume of distribution (V/F = 87.6 L) were estimated from sparse-sampling clinic data; the absorption rate constant was not estimable in Adams 1998 (paper Discussion p. 412) and is fixed in this model to ka = 2.5 /h from primary single-dose PK data (Klecker 1988). No baseline covariates (age, sex, total body weight, calculated creatinine clearance, food, concomitant zidovudine) improved the basic fit and none were retained in the final model. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/deVriesSchultink_2020_zenocutuzumab.html">Zenocutuzumab (deVriesSchultink 2020)</a> </td> <td style="text-align:left;"> Two-compartment population PK model with parallel linear and Michaelis-Menten non-linear elimination from the central compartment for intravenous zenocutuzumab (MCLA-128), a bispecific IgG1 (anti-HER2 x anti-HER3) monoclonal antibody, in patients with various advanced solid tumors (de Vries Schultink 2020) </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Fauchet_2013_zidovudine.html">Zidovudine (Fauchet 2013)</a> </td> <td style="text-align:left;"> One-compartment population PK model for oral zidovudine (ZDV) and its glucuronide metabolite 3'-azido-3'-deoxy-5'-glucuronylthymidine (G-ZDV) in HIV-1-infected children, infants, and adolescents (Fauchet 2013, retrospective Paris-area therapeutic-drug-monitoring cohort, n = 247, age 0.5-18 years). First-order absorption with a fixed ka = 2.86 1/h (inherited from Panhard 2007) delivers ZDV into a one-compartment central compartment with apparent total clearance CL_p/F and apparent volume V/F. The metabolite is described by a single G-ZDV state driven by a lumped metabolic formation rate constant CL_m/V_m and a first-order metabolite elimination rate constant k_el. The metabolite distribution volume V_m is not identifiable from plasma data alone and is set structurally to 1 L (same convention used by Lee 2016 for raltegravir glucuronide). Body weight enters as an estimated power-allometric covariate on CL_p/F (exponent 0.858) and on V/F (exponent 0.534), centered on the cohort median 32.2 kg; age, sex, dosage form, and antiretroviral cotreatments (3TC, ddI, ABC, LPV, RTO, NFV, NVP, EFV) were all tested and none was retained at p < 0.01. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/Mitra_2026_ziftomenib.html">Ziftomenib (Mitra 2026)</a> </td> <td style="text-align:left;"> Sequential two-stage population PK model for oral ziftomenib (a potent, selective, oral menin inhibitor for R/R NPM1-mutated acute myeloid leukemia) and its two active metabolites KO-739 and KO-516 (Mitra 2026 Kura Oncology KOMET-001 + KO-MEN-003). Parent PK is a 2-compartment model with first-order absorption, absorption lag time, and linear elimination from the central compartment; oral bioavailability F1 is fixed at 0.129 (identifiability constraint from the human ADME + absolute-BA study KO-MEN-005). Each metabolite is 2-compartment with linear elimination; the metabolic clearance is split between KO-739 and KO-516 by a fixed 1:1 in-vitro-anchored biotransformation ratio (FM_KO516 = 0.5), with the total metabolized fraction FM held fixed at 0.535 after an initial identifiability-limited estimation. Covariate effects retained in the final model: FED and PPI on parent F1 (logit-scale shifts +3.21 fed; -0.520 PPI = 6.09x and 0.627x multipliers on F1), PPI on parent Ka (log-scale shift -0.485 = 0.616x), FED on parent absorption lag time (log-scale shift +0.322 = 1.38x), strong CYP3A4 inhibitor on parent CL/F (log-scale -0.778 = 0.459x), healthy-volunteer status on parent CL/F (log-scale +0.950 = 2.59x), healthy-volunteer status on FM (logit-scale -1.62 = 0.348x multiplier on FM), strong CYP3A4 inhibitor on KO-739 CL (log-scale -1.64 = 0.195x), strong CYP3A4 inhibitor on KO-516 CL (log-scale -0.802 = 0.449x), healthy-volunteer status on KO-739 Vc (log-scale -1.62 = 0.197x), and healthy-volunteer status on KO-516 Vc (log-scale -1.87 = 0.154x). No effect of NPM1-m vs KMT2A-r mutational status, body weight, sex, race, age, mild/moderate renal or hepatic impairment, or P-gp inhibitor coadministration on ziftomenib PK. IIV: parent 47.3% CV on CL and 120% CV on Vc; metabolites 74.7% (KO-739 CL), 110% (KO-739 Vc), 162% (KO-739 Q), 31.2% (KO-516 CL), 191% (KO-516 Vc), 118% (KO-516 Q), and 56.8% CV on FM (all independent diagonals). Inter-occasion variability on F1 (Omega 1.06 corresponding to 137.3% CV) reported in the parent NONMEM run across 3 occasions is not encoded structurally here (no operational occasion column is defined for the model-library use case; see vignette Assumptions and deviations). Residual error: proportional 43.7% CV on parent Cc; proportional 45.2% CV plus additive 0.128 ng/mL on Cc_ko739; proportional 36.4% CV on Cc_ko516. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PiresdeMello_2018_zika_FAV_HFIM.html">Zika FAV HFIM (PiresdeMello 2018)</a> </td> <td style="text-align:left;"> In vitro (HUH-7 human hepatoma cells, hollow-fiber infection model). Refined translational mechanism-based pharmacodynamic (MBM) model of Zika virus replication and inhibition by favipiravir (FAV) under dynamic, human-like FAV concentration-time profiles. Twelve-state model: uninfected host cells (uninfected) with logistic-growth replication limited by carrying capacity HOSTmax; five sequential infected host cell stages (infected1..infected5) representing the delay from infection to virus release; five intracellular virus transit compartments (vi1..vi5) for viral maturation; and extracellular virus (vextra) as the observation output (log10 PFU/mL). FAV inhibits viral RNA release between vi4 and vi5 via a simple Imax/IC50 inhibition function (Eq 8). FAV concentration is a time-varying covariate (CONC_FAV_UM) driven externally by the user-supplied clinical PK profile. Parameters are the HFIM column of Table 1; drug-effect parameters (Imax_FAV, IC50_FAV) and the additive residual SD are shared estimates with the parallel plate assay co-fit reported in the same paper. </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PiresdeMello_2018_zika_FAV_IFN_RBV.html">Zika FAV IFN RBV (PiresdeMello 2018)</a> </td> <td style="text-align:left;"> In vitro (Vero cells). Translational mechanism-based pharmacodynamic (MBM) model of Zika virus replication and inhibition by favipiravir (FAV), interferon alpha (IFN), and ribavirin (RBV) as monotherapy and in two-drug combinations. Eight-state model: uninfected (U) and infected (I) host cells, five intracellular virus transit compartments (vi1..vi5) capturing maturation delay, and extracellular virus (vextra) as the observation output (log10 PFU/mL). IFN inhibits cellular infection via a sigmoidal Hill function; FAV and RBV both inhibit the vi4 -> vi5 maturation transit; RBV additionally causes first-order cytotoxicity to both uninfected and infected host cells. FAV+RBV antagonism is encoded via a competitive-interaction factor PSI (= 1 monotherapy, = 1.37 combination). Drug concentrations are static covariates -- the in vitro experiment fixes nominal concentrations for the 4-day window. All parameters fixed at the Table 1 point estimates; the between-curve CVs reported in Table 1 are not encoded as etas (typical-value mechanism). </td> </tr> <tr> <td style="text-align:left;"> <a href="https://nlmixr2.github.io/nlmixr2lib/articles/PillaReddy_2013_panss_subscales.html">Ziprasidone panss subscales (PillaReddy 2013)</a> </td> <td style="text-align:left;"> Population PK/PD model for ziprasidone against the three PANSS subscales (positive, negative, general) in adults with schizophrenia from Pilla Reddy 2013 Part II. The PK sub-model is the one-compartment ziprasidone structural model from Part I (PMID 23473810) Table 2: first-order absorption ka = 0.07 1/h, apparent oral clearance CL/F = 54 L/h, apparent central volume of distribution Vc/F = 87.5 L. The PD sub-model has three outputs that share the Weibull placebo time- course form Pplacebo = Pmax * (1 - exp(-(t/TD)^POW)) but each subscale carries its own placebo Pmax, TD, POW (Part II Table 1) and ziprasidone's own Emax / EC50 / KT triplet per subscale (Part II Table 2). The KT for ziprasidone PANSS positive and general (0.048 and 0.035 1/day) is the common-across-atypical-antipsychotic value; the KT for the negative subscale (0.0073 1/day) was estimated separately per drug and is the slowest of any compared drug, consistent with Part II's report that ziprasidone has the longest onset for negative symptoms (more than 3 weeks half-time vs 5 days for haloperidol). The exponential time-to-event dropout sub-model from Part II Table 4 is documented in population$dropout_model but not encoded in the model body.
Zolbetuximab (Yamada 2025) Two-compartment population PK model of zolbetuximab (anti-CLDN18.2 IgG1 mAb) with zero-order IV input and time-dependent clearance in patients with locally advanced unresectable or metastatic gastric/gastroesophageal junction (G/GEJ) adenocarcinoma (Yamada 2025)
ZoledronicAcid (Mori 2018) Kinetic-pharmacodynamic (K-PD) PK / bone-turnover-marker / lumbar-spine BMD model for once-yearly intravenous zoledronic acid (ZOL) 5 mg in Japanese patients with primary osteoporosis (ZONE study). A virtual effect-site amount A receives the administered dose and decays first-order at rate KD; the drug-effect signal KDA enters a sigmoidal Imax factor with Hill coefficient Gamma and half-effect EKD50 that inhibits the zero-order synthesis Kin of the serum bone-resorption marker (tartrate-resistant acid phosphatase 5b, TRACP-5b), which is eliminated first-order at Kout. The observed marker carries a multiplicative disease-progression / supplementation drift (1 + Slope t + Emax * t / (T50 + t)) capturing the daily oral calcium + vitamin D + magnesium supplementation effect pooled with natural osteoporosis progression (the two effects could not be separated because all subjects received the supplements). Lumbar-spine BMD follows an effect-compartment ODE with rate Ke0 whose target is BMD0 + Scale * (marker - Marker0), where Scale (negative) is the marker-to-BMD coupling. Baseline TRACP-5b (TRACP5B_BL) enters EKD50, Slope, T50, and (active-arm-only) Scale as a power-model covariate centred on the cohort reference 400 mU / dL.
Zonisamide (Hashimoto 1994) Steady-state Michaelis-Menten population PK model for zonisamide in 68 Japanese epileptic patients (pediatric + adult) on chronic oral zonisamide. A power-of-weight body-size factor scales both volume of distribution and Vmax; concomitant carbamazepine multiplicatively increases Vmax (Hashimoto 1994 Eqs. 1-4).

DDMoRe

name description
Alzheimer (Conrado 2014) Updated Alzheimer’s disease progression model for ADAS-Cog total score (0-70) over time, fit to ~25,000 ADAS-Cog observations from 4,494 subjects across 15 studies in the Coalition Against Major Diseases (CAMD) Alzheimer’s database. The expected score is described by a Richards three-parameter logistic growth function whose asymptote is the upper boundary score 70; the residual distribution on the bounded ADAS-Cog/70 scale is a beta distribution with precision parameter TAU. Subject-level baseline and slope inter-individual variability are correlated through a 2x2 BLOCK; covariates of sex on baseline, APOE4-allele count on baseline and slope, age on slope, and concomitant Alzheimer’s-symptomatic medication on slope are reproduced from the source. The original publication adds a third-level (study) random effect on baseline and slope plus study-1131-specific scalers; nlmixr2 does not natively support multi-level random effects, so those layers are dropped here and documented in the validation vignette’s Errata section.
Artesunate (Birgersson 2019) Joint parent-metabolite population PK model for oral artesunate and dihydroartemisinin in pregnant and non-pregnant women with uncomplicated Plasmodium falciparum malaria, with a 3-compartment transit-absorption chain into a 1-compartment artesunate disposition model and complete in-vivo conversion to a 1-compartment DHA disposition model. Allometric body-weight scaling on CL (exponent 0.75) and V (exponent 1) is applied to both parent and metabolite. Covariate effects: pregnancy on DHA elimination clearance, and admission alanine aminotransferase and log-asexual-parasite-count on relative bioavailability of artesunate.
Bedaquiline (Svensson 2016) Three-compartment population PK model for the antimycobacterial bedaquiline (BDQ) and its one-compartment N-desmethyl metabolite M2 in adult patients with multidrug-resistant tuberculosis (MDR-TB), with two-transit-compartment first-order oral absorption, time-varying body-weight allometric scaling, time-varying serum-albumin power effects on disposition and metabolite formation/elimination, and additional Black-race and linear age covariate effects on bedaquiline and M2 clearance.
Capecitabine (Henin 2009) Longitudinal Markov-proportional-odds model for hand-and-foot syndrome (HFS) toxicity grades 0-2 in cancer patients receiving capecitabine. Capecitabine exposure is described with a kinetic-pharmacodynamic (K-PD) one-compartment delay; the per-week effective drug rate drives a sigmoid Emax that shifts the cumulative log-odds for the next HFS grade conditional on the previous grade. Baseline-Cockcroft-Gault creatinine clearance is the only structural covariate.
Ciprofloxacin (Khan 2015) Mechanism-based six-compartment PK/PD model for ciprofloxacin against Escherichia coli K-12 wild-type and quinolone-resistant single-step mutants in static in vitro time-kill experiments (Khan et al. 2015 J Antimicrob Chemother). Two co-existing bacterial subpopulations (susceptible-growing, plus a small pre-existing less-susceptible subpopulation seeded at MUT*1e-6 of the inoculum) each carry growing, resting, and drug-induced non-colony-forming states; drug effect is a Hill-Emax term on bacterial kill with strain-specific EC50 (LM202 estimated; LM347, LM378, LM534, LM625, LM693, LM707 fixed at the published per-strain values), a separate EC502 for the resistant subpopulation, density-dependent active->resting flux, and a model-time gate that switches off the active->non-colony-forming transition after THETA(19) hours.
Cladribine (Novakovic 2017) Item Response Theory (IRT) model of EDSS disability progression in patients with multiple sclerosis treated with cladribine (Novakovic 2017). Eight EDSS functional system subscores (Pyramidal, Cerebellar, Brainstem, Sensory, Bowel/Bladder, Visual, Mental, Ambulation) are linked to a latent disability variable that follows a power-law disease progression in time. The model embeds an exposure-dependent symptomatic drug effect (Emax on cumulative cladribine dose adjusted for creatinine clearance) and an exposure-independent fractional protective effect on disease progression, plus full Random Effects on covariates (FREM) for Age, months since diagnosis (MSD), and exacerbation rate baseline (EXNB).
Colistin (Leuppi-Taegtmeyer 2019) Six-structural-compartment population PK model for colistimethate sodium (CMS, prodrug) and colistin (Col, active metabolite) during continuous renal replacement therapy (CRRT) in critically ill adults. CMS converts to Col by first-order metabolism (CL1M); Col is cleared by metabolism (CL2M). Both CMS and Col exchange with a CRRT filter compartment (driven by blood flow QBL and patient hematocrit HCT) and a downstream cartridge / effluent compartment (driven by effluent flow QEFF and species-specific sieving coefficients SC_CMS and SC_COL). Filter and cartridge priming volumes are fixed device constants (V_filter = 0.2 L, V_cart = 0.3 L). Bioavailability factor F = 1155.5/1749.8 on the dose compartment converts mg of CMS sodium administered into mg colistin-base equivalents (the molar-mass ratio of colistin to CMS sodium); central concentrations are reported as mg/L colistin-base equivalents.
Colistin meropenem (Mohamed 2016) In vitro time-kill PK/PD model for colistin and meropenem alone and in combination against P. aeruginosa wild-type ATCC 27853 (BACT=2) and meropenem-resistant clinical isolate ARU552 (BACT=1); proliferating/resting bacterial subpopulations with strain-specific drug effects, colistin adaptive resistance, and a meropenem-resistant mutant subpopulation
DDMoRe: lidocaine Population PK model for lidocaine and three sequential metabolites (monoethylglycinexylidide [MEGX], glycinexylidide [GX], and 2,6-xylidide [2,6-XYL]) using the source’s NONMEM ADVAN5 / TRANS1 general-linear rate-constant parameterisation. Lidocaine in the central compartment is metabolised in parallel to MEGX (rate constant k_megx, fixed at 0.03) and to 2,6-XYL (k_xyl, fixed at 0.007); MEGX is sequentially metabolised to GX (k_gx, estimated). GX and 2,6-XYL each carry a typical-value elimination rate constant (kel_gx, kel_xyl) modulated by binary stratifications of dose-level (DLVL > 2), bilirubin (BIL > 0.53), creatinine clearance (CRCL <= 52.7), CYP1A2-modifying co-medications (S1A2 == 3), body-mass index (BMI > 27.93), serum ALT (SGPT > 11), and lactate dehydrogenase (LDH > 195). Lidocaine apparent central volume V1 is also dose-level-stratified (DLVL > 2); the three metabolite compartments share a fixed apparent volume of 100. Distributed in the DDMORE Foundation Model Repository as DDMODEL00000281; no linked publication is identified in the bundle’s Model_Accommodations.txt or in the .ctl / .res headers. Final parameter estimates come from the bundle’s Output_real_data_original_final_run249.res listing after the MINIMIZATION SUCCESSFUL block (FOCE estimation, OBJ = 10219.922, covariance step succeeded). The bundle’s .ctl does not declare time, dose, or concentration units explicitly; units$time = 'h', units$dosing = 'mg', and units$concentration = 'mg/L' are operator-default placeholders chosen so the values flow through unit- checking consistently. See the vignette Errata for the unit ambiguity and a per-time-point self-consistency check against the bundle’s Simulated_Lid_B04_ddmore.csv.
DDMoRe: miridesap Population PK/PD model for CPHPC (miridesap, GSK2315698, Ro 63-8695) and serum amyloid P (SAP) in healthy volunteers (CPH113776) and patients with systemic amyloidosis (CPH114527). Two-compartment PK for CPHPC (with first-order subcutaneous absorption from a depot in addition to IV infusion); two-compartment turnover model for SAP with first-order endogenous production and elimination; bimolecular CPHPC + free SAP -> complex binding (treated as effectively irreversible because internalization is fast relative to dissociation) followed by complex internalization. Final-model covariates from Sahota 2015 Eq. 1 and Eq. 2: creatinine clearance (CRCL) modifies CPHPC clearance below an 80 mL/min threshold, hepatic amyloid involvement (DIS_AMYLOID_LIVER) multiplies SAP intercompartmental clearance Q4, whole-body amyloid load (DIS_AMYLOID_LOAD: 0-3) multiplies SAP peripheral volume V4 in two categorical steps, and biological sex (SEXF) multiplies baseline plasma SAP. Distributed in the DDMORE Foundation Model Repository as DDMODEL00000262.
DDMoRe: paracetamol Mechanistic OGTT model from the DDMORE Foundation Model Repository (DDMODEL00000228) that uses paracetamol as a gastric-emptying tracer to drive a coupled paracetamol PK + glucose + GLP-1 + GIP system. Fifteen compartments span paracetamol stomach / intestine / central / peripheral (with saturable first-pass loss), glucose stomach / duodenum / jejunum / ileum / central / peripheral, two effect compartments for glucose-on- production and insulin-on-elimination delays, a cumulative first-pass- loss tally for paracetamol, and indirect-response states for the incretin hormones GLP-1 and GIP. The gastric-emptying rate KS is modulated downwards by duodenal glucose via a Hill function (IGD50 / GAM) and gated by a logistic lag(T-T50) profile; glucose absorption from each small-intestine segment is Michaelis-Menten in segment amount (KMG, RAMAXD / RAMAXJ / RAMAXI). Plasma insulin (INS) is a time-varying regressor that enters the central glucose compartment through a one- compartment effect delay (KIE). Type 2 diabetes mellitus (DIS_DIAB) is encoded as a binary indicator switching the glucose baseline (GSSH / GSSD), glucose clearance (CLGH / CLGD), insulin-dependent glucose clearance (CLGIH / CLGID), glucose bioavailability (FPGH / FPGD), and the empirical glucose-on-production exponent (GPRG = -2.79 healthy, 0 DIS_DIAB). Body weight (WT) scales the central glucose volume linearly (VG * WT / 70). Outputs are observed on the linear scale: paracetamol concentration plus baseline noise (Cc, uM), glucose concentration (Cglu, mM), GLP-1 concentration (CGLP1), and GIP concentration (CGIP). Source listing reports the FOCEI step terminated due to rounding errors (NSIG = 0.5); see vignette Errata for the implication on parameter-precision claims.
DDMoRe: sunitinib Semi-mechanistic PK/PD/tumor-growth model for sunitinib in non-small cell lung cancer (NSCLC) patients (DDMORE Foundation Model Repository entry DDMODEL00000231; MDL/PharmML deposit; no linked publication identified). Parent and metabolite each follow a 2-compartment oral PK model with first-order absorption from a separate depot; both depots receive the dose, with effective bioavailability split (1 - fp) to the parent and fp to the metabolite (fp = 0.21 hard-coded). Four indirect-response PD biomarker compartments (biom1, biom2, biom3, biom4) are driven by parent (and optionally metabolite) plasma concentration through 1/(1 + pd*conc) inhibition factors; biom1 is on Kout-modulation, biom2-4 are on Kin-modulation. Tumor volume is described by a sphere-volume state (radius is the observation) with a doubling-time-capped exponential growth term modulated by a delayed parent-concentration memory and a lambda-feedback growth-rate state. A resistance-accumulator state, a parent-concentration integrator, and a delayed-signal compartment combine into the tumor’s effective growth rate. The model preserves four hard-coded structural placeholders from the MDL: fp = 0.21 (metabolite-formation fraction), th1..4 = 1 (parent-only drive on biomarkers; metabolite drive disabled), thettum = 1 (parent-only drive on tumor; metabolite drive disabled), dres = 0 (no decay of resistance accumulator).
DDMoRe: tte gompertz Parametric time-to-event base hazard model for Event 1 in the BAST PTTE 2017 four-event teaching dataset (DDMODEL00000243). The .mod $PROBLEM line names this a ‘Gompertz hazard model’ but the equation has no time-varying alphat term, so the realised hazard is constant: h(t) = (lam/1000) exp((coef_neut/10000)(NEUT-4133)) exp((coef_age/100)*(AGE-55)). The BAST guiding-document text (Figure 2-1, page 13) confirms an exponential distribution was selected for Event 1; the .mod / file name retain the ‘Gompertz’ label per the source $PROBLEM line and the operator’s selected option NA_NA_tte_gompertz.R.
DDMoRe: tte gompertz ev2 Parametric time-to-event Gompertz hazard model for Event 2 in the BAST PTTE 2017 four-event teaching dataset (DDMODEL00000243). Hazard h(t) = (lam/1000) * exp((alpha/1000)t) exp((coef_auc/1000)*(AUC_BAST_FW - 3065.5)). Event 2 in the bundle’s simulated dataset is interval-censored (CENSORING = 2), assessed at scheduled visits rather than observed exactly; the BAST guiding-document Section 2.4.1 (Figure 2-2) selected Gompertz as the base distribution by AIC, then Section 2.4.2 (Table 2-3) retained first-week AUC as the only covariate.
DDMoRe: tte loglogistic Parametric time-to-event log-logistic hazard model for Competing Event 2 in the BAST PTTE 2017 four-event teaching dataset (DDMODEL00000243). Hazard h(t) = alpha * (lam^alpha) * t^(alpha-1) / (1 + (lam*t)^alpha), where lam = lambda/1000 and alpha is unscaled. No covariates were retained. Note: the BAST guiding-document text (Section 2.4.1, Figure 2-4) selected log-normal as the base distribution for Competing Event 2, but the executable supplied with the bundle (Executable_runCOMPEV2_005.mod, $PROBLEM ‘Log_logistic model’) and the final-fit listing in the bundle are log-logistic, not log-normal. This file follows the executable; the publication-vs-bundle distribution discrepancy is documented in the validation vignette’s Errata.
DDMoRe: tte lognormal Parametric time-to-event log-normal hazard model for Competing Event 1 in the BAST PTTE 2017 four-event teaching dataset (DDMODEL00000243). Hazard h(t) = val * pdf(t) / (1 - Phi(log((t+DEL)/alpha) / lambda)), where pdf(t) is the log-normal probability density with shape lambda (sigma) and time-scale alpha, val = exp((coef_age/1000) * (AGE - 55)) is the multiplicative AGE effect, and Phi is the standard normal CDF. Competing Event 1 is interval-censored; the BAST guiding-document Section 2.4.1 (Figure 2-3) selected log-normal as the base distribution, then Section 2.4.2 (Table 2-4) retained AGE as the only covariate.
Disufenton (Jonsson 2005) Two-compartment intravenous PK model for disufenton sodium (NXY-059) in adult patients with acute ischaemic or haemorrhagic stroke (Jonsson 2005), as packaged in DDMORE Foundation Model Repository entry DDMODEL00000245. Continuous IV infusion (1-h loading + 71-h maintenance) with a piecewise-linear creatinine-clearance effect on CL (no effect at CLCR <= 40 mL/min, linear above 40) and a linear weight effect on the central volume of distribution (centered at 76 kg). Correlated inter-individual variability on CL and Vc and a log-transform-both-sides residual error model.
Docetaxel (Netterberg 2017) Friberg-style semi-mechanistic myelosuppression PD model for docetaxel-induced neutropenia in adult cancer patients (DDMODEL00000224, Netterberg 2017 / Kloft 2006). The bundle’s NM-TRAN .mod fixes parameter values at the Kloft 2006 docetaxel myelosuppression analysis (per the .mod’s ; Parameter estimates as according to Kloft et al., 2006 block; MAXEVALS=0 in the bundle’s $ESTIMATION confirms no re-fit) and Netterberg 2017 reuses the model unchanged to evaluate frequent-monitoring ANC prediction methodology. Structurally, the model has a self-renewing proliferation pool plus three transit compartments and a circulating compartment; docetaxel exposure is supplied as a time-varying plasma-concentration covariate (CP_MGL, mg/L) that drives a linear drug effect (1 - SL * CP_MGL) on proliferation, with a feedback term (BA / circ)^PO. Covariate effects on baseline ANC (sex, ECOG performance status, prior anticancer therapy, alpha-1 acid glycoprotein) and on the drug-effect slope (alpha-1 acid glycoprotein) follow Kloft 2006. Output is the absolute neutrophil count ANC in 10^9 cells/L.
Dupilumab ddmore (Kovalenko 2016) Dupilumab population PK as encoded in DDMORE Foundation Model Repository entry DDMODEL00000273. Two-compartment model with parallel linear and Michaelis-Menten elimination from the central compartment, first-order absorption from a SC depot, and a non-standard body-weight covariate on the central volume in which weight enters log(V2) multiplicatively. Final estimates here come from the bundle’s Output_simulated_.lst (a SAEM/IMP fit on the bundle’s Simulated_Dupilumab.CSV); no Output_real_.lst is shipped, so these values do NOT match Table 2 of the publication. The publication-faithful encoding (Eq. 1, Eq. 2, Table 2 estimates) is the replicate_of counterpart at inst/modeldb/specificDrugs/Kovalenko_2016_dupilumab.R.
Ethambutol ddmore (Jonsson 2011) DDMoRE-source replicate of Jonsson_2011_ethambutol. Two-compartment population PK model for oral ethambutol in adult South African pulmonary tuberculosis patients (Jonsson 2011), with one transit compartment preceding first-order absorption, allometric scaling on clearance and volume terms (theory-based exponents on a 50 kg reference), an HIV-status effect on bioavailability, and 4-occasion inter-occasion variability on apparent oral clearance. Parameter values are taken from the DDMORE bundle Output_real_run32150.lst FINAL PARAMETER ESTIMATE block (DDMODEL00000220); see inst/modeldb/specificDrugs/Jonsson_2011_ethambutol.R for the publication-Table-2-sourced replicate.
Gemtuzumab (Jager 2011) Mechanism-based PKPD model for the antibody-drug conjugate gemtuzumab ozogamicin (GO; Mylotarg) in patients with acute myeloid leukemia, as packaged in DDMORE Foundation Model Repository entry DDMODEL00000229. The model couples drug pharmacokinetics with explicit binding to the cell-surface antigen CD33: free drug in a central compartment binds free CD33 receptor to form a drug-receptor complex that is internalized; the toxic ozogamicin component then drives linear depletion of leukemic blast cells. The DDMORE entry extends the original Jager 2011 PK structure by adding a peripheral drug compartment and re-estimating all parameters simultaneously in Monolix, so it is not a literal reproduction of the published model – see the validation vignette for the comparison.
Gentamicin (Germovsek 2016) Three-compartment population PK model for gentamicin in neonates and infants (Germovsek 2016), as packaged in DDMORE Foundation Model Repository entry DDMODEL00000238.
Glucose (Bizzotto 2016) Mechanistic model of glucose tracer kinetics in humans driven by time-varying plasma insulin and glucose regressors (Bizzotto 2016). Glucose uptake is a Michaelis-Menten function of glucose at the site of action whose maximum rate Vmax is itself a Hill (sigmoidal) function of insulin at the site of action; this captures the observation that hyperglycemia suppresses the glucose-clearance response to hyperinsulinemia. Two-compartment delays smooth plasma insulin and glucose into their site-of-action analogues, and a heart-lung block plus a three-channel periphery block give the tracer disposition. Distributed in the DDMORE Foundation Model Repository (DDMODEL00000227) as a simulation-only implementation; the linked publication fits the same equations to real data from 123 subjects spanning normal-tolerant, impaired-glucose-tolerance, and type 2 diabetic adults.
Il21 (Elishmereni 2011) Preclinical (mouse). Seven-compartment population PK model for recombinant murine interleukin-21 (rIL-21) administered by intraperitoneal (ip), intravenous (iv), or subcutaneous (sc) routes, extracted from the DDMORE Foundation Model Repository (DDMODEL00000230). Linear elimination from the central compartment plus a direct first-pass loss from the ip depot, parallel ip and sc absorption routes each with a single intermediate transit compartment, and two peripheral distribution compartments. Saturable transfer terms present in the published model (sa0, sa1, sa2) and the sc-depot direct elimination rate (k30) are held at zero in the DDMORE deposit, leaving an effectively linear system. Per-subject random effects on CL, central volume, and the ip and sc absorption rates are correlated as a 4 x 4 block. Dose route is encoded by the cmt column: ip = depot, iv = central, sc = depot2. Units are not declared in the DDMORE bundle (no IL_21_PK.csv shipped); see the validation vignette Errata.
Levetiracetam (Schoemaker 2018) Combined adult / pediatric population PK-PD count model for levetiracetam (LEV) used in Schoemaker 2018 to scaffold a pediatric brivaracetam (BRV) extrapolation. The fitted compound is LEV; CAV is LEV plasma concentration (mg/L). Negative-binomial seizure-count likelihood with two-component mixture (responder vs non-responder), Box-Cox-transformed inter-individual variability on log baseline rate, and a Markovian dependence on the previous-period seizure count entering as a per-record covariate (PDV). The DDMORE bundle is a PD-only $PRED model: drug exposure enters as the data column CAV, so there is no PK ODE in this file. Adults contribute monthly (~28-day) counts (PDV unused, sentinel -99 in the source dataset, CHILD = 0); pediatrics contribute daily counts (NDAYS = 1, PDV = previous-day observed count, CHILD = 1).
Meropenem (Li 2006) Two-compartment population PK model for meropenem in adult patients (Li 2006), as packaged in DDMORE Foundation Model Repository entry DDMODEL00000213.
Meropenem (Themans 2019) Three-compartment population PK model for meropenem in adults with severe pneumonia, with parallel ELF (epithelial lining fluid) sampling (Themans 2019), as packaged in DDMORE Foundation Model Repository entry DDMODEL00000301.
Midazolam (vanRongen 2018) Two-compartment population PK model for midazolam with a five-transit-compartment first-order oral absorption chain (KA = KTR), supporting oral and intravenous dosing, in 19 obese adolescents (median total body weight 102.7 kg) and 20 morbidly obese adults (median 144 kg). Adult and adolescent typical clearances are estimated as separate intercepts; total body weight enters as a power covariate on clearance only in adolescents (reference 104.7 kg) and on peripheral volume only in adults (reference 141.8 kg).
Midazolam (Vet 2016) Two-compartment population PK model for IV midazolam in critically ill children (Vet 2016) with body-weight allometric scaling on CL and V1 (reference 5 kg), a CRP power effect on CL (reference 32 mg/L), per-stratum typical CL values for the number of failing organs (ORG_FAIL_COUNT strata 0 / 1 / 2 / 3 / >=4; the source NMTRAN dataset names this column ORGF), inter-individual variability on CL and V1, and inter-occasion variability on CL across six daily occasions. Packaged in DDMORE Foundation Model Repository entry DDMODEL00000249.
Morphine (Knibbe 2009) Joint parent-metabolite population PK model for morphine and its glucuronide metabolites M3G and M6G in preterm neonates, infants and toddlers <3 years, with bodyweight allometric scaling and a postnatal-age-stratified glucuronidation step at PNA = 10 days
Morphine (Valitalo 2017) Item response theory (IRT) pharmacodynamic model for morphine analgesia in mechanically ventilated preterm neonates undergoing endotracheal suctioning. A latent pain variable is driven by suctioning state (pre/during/after), a linear morphine plasma-concentration effect, and a linear study-time effect. The 8 ordinal pain-item scores (COMFORT-B alertness/calmness/respiratory/body movement/facial tension; PIPP brow bulge/eye squeeze/nasolabial furrow; NIPS total) and 1 continuous VAS score map onto the latent variable via a graded-response IRT model. Morphine plasma concentration is supplied as a time-varying covariate CP_MORPH_NGML from an upstream PK model (Knibbe 2009; see modellib(‘Knibbe_2009_morphine’)). The IRT discrimination and difficulty parameters were estimated separately in a graded-response sub-model and are FIX’d here per the source NONMEM control stream. Population: 140 preterm neonates (mean weight 1.4 kg, mean postmenstrual age 211 days).
Morphine (Wang 2013) Two-compartment population PK model for morphine across the entire paediatric age range and adults using a bodyweight-dependent allometric exponent (BDE) on clearance, with adolescent-specific intercompartmental clearance and central volume and an adult-stratum oral-bioavailability adjustment, as packaged in DDMORE Foundation Model Repository entry DDMODEL00000269 (Wang 2013 Model I).
Nivolumab ddmore (Bajaj 2017) Two-compartment population PK model for nivolumab (anti-PD-1 IgG4) with time-varying clearance (sigmoid Emax of time since first dose) in patients with advanced solid tumors (Bajaj 2017; DDMORE Foundation Model Repository entry DDMODEL00000284). DDMORE-source replicate of inst/modeldb/specificDrugs/Bajaj_2017_nivolumab.R; parameter values are taken from the DDMORE bundle’s Output_real_Nivo-PPK.lst FINAL PARAMETER ESTIMATE block and time is kept in hours to mirror the bundle directly.
Osteoprotegerin (Zierhut 2008) Population PK/PD model for Fc-osteoprotegerin (Fc-OPG, AMG 162 / AMGN-0007 precursor) in healthy postmenopausal women (Zierhut 2008): two-peripheral-compartment IV/SC PK with parallel linear and Michaelis-Menten elimination from the central compartment, first-order absorption from the SC depot with a logistic-style bioavailability F = FSC / (1 + FSC), and an indirect-response biomarker turnover model for urinary N-telopeptide (uNTX) where Fc-OPG inhibits bone-resorption-driven NTX synthesis via an Imax = 1, IC50 sigmoidal Hill term.
Paclitaxel (Friberg 2002) Semi-mechanistic Friberg-style myelosuppression PK/PD model for paclitaxel in adult cancer patients (Friberg 2002, leukocyte arm of DDMODEL00000186). Paclitaxel exposure is driven by per-subject empirical-Bayes PK estimates supplied as data columns (CL_INDIV, VC_INDIV, VP_INDIV) with intercompartmental clearance Q fixed at 204 L/h. Leukocyte response is described by a self-renewing proliferating pool plus three transit compartments and a circulating compartment, with a linear drug effect (1 - SLOPU * Cc) on proliferation and a feedback term (CIRC0 / circ)^GAMMA. Output is total circulating leukocytes in 10^9 cells/L.
Paclitaxel (Terranova 2018) Preclinical xenograft-mouse Dynamic Energy Budget tumor-growth-inhibition (DEB-TGI) PK/PD model with paclitaxel-induced tumor kill and tumor-driven plus drug-driven host cachexia (Terranova 2018; DDMODEL00000274 paclitaxel scenario). Two-compartment paclitaxel PK (rate constants K10/K12/K21 and central volume V1 fixed from upstream popPK) drives a Simeoni-style tumor inhibition arm (proliferating tumor VU1 plus three damaged-cell transit compartments VU2/VU3/VU4) coupled to a host energy budget (structural body component Z, enzyme density EN). Body weight is W = density_V * (1 + xi * EN) * Z; tumor weight is Wu = density_Vu * (VU1 + VU2 + VU3 + VU4). The host-tumor coupling makes the structural-body dynamics piecewise: three switch branches (SWITCH1 / SWITCH2 thresholds with a delta_Vmax cap) determine whether the tumor draws from host enzymes preferentially, from the structural body component, or hits the catabolic body-loss cap.
Pain (Plan 2012) Markov Integer Model for placebo time-course of Likert (0-10) pain scores in adults; pooled placebo arm of three Phase III neuropathic-pain trials (Plan 2012; DDMODEL00000194)
Paracetamol (Allegaert 2015) Eight-compartment population PK model for IV propacetamol/paracetamol (APAP) and its glucuronide and sulphate metabolites in young women (Allegaert 2015), distributed in the DDMORE Foundation Model Repository as DDMODEL00000267. The structural model carries a three-compartment plasma disposition for parent APAP (central + two peripherals), two plasma metabolite compartments (APAP-glucuronide and APAP-sulphate, each with a metabolite-specific volume V_meta = 0.18 * V_central), and three cumulative-urine compartments (urine APAP, urine APAP-glucuronide, urine APAP-sulphate). Pregnancy state, time post partum, term-vs-preterm birth, oral-contraceptive use, and time-varying urine flow rate enter as covariates on the parent and metabolite-formation clearances; an OCC-conditional residual-error model gives non-pregnant volunteers on birth control a combined proportional + additive plasma error while every other occasion uses a proportional-only plasma error.
Paracetamol (Cook 2016) Population PK model for paracetamol (APAP) and its glucuronide and sulphate conjugates with cumulative urinary excretion in term and preterm newborns (Cook 2016), as packaged in DDMORE Foundation Model Repository entry DDMODEL00000271.
Paracetamol (vanWijk 2019) PRECLINICAL (zebrafish): two-compartment paracetamol PK model fit to zebrafish (Danio rerio) larvae continuously exposed to a 1 mM paracetamol bath at 3, 4, or 5 days post-fertilization (van Wijk 2019, DDMODEL00000294). The medium reservoir (compartment 1) is held at constant amount, so K12 acts as a zero-order absorption rate from the bath into the larva; elimination from the larva (compartment 2) is first-order with rate K25. Larval age in dpf enters as a step factor on K12 (~2.06x at >= 4 dpf vs 3 dpf) and a per-day power factor on K25 (+17.4% per day post-fertilization), consistent with maturation of paracetamol absorption and elimination capacity across the 3-5 dpf window.
Paracetamol (Zurlinden 2016) Whole-body physiologically-based pharmacokinetic (PBPK) model for paracetamol (acetaminophen, APAP) and its conjugated metabolites APAP-glucuronide (AG) and APAP-sulfate (AS) in healthy adults (Zurlinden & Reisfeld 2016, DDMODEL00000237 Scenario 4 = 1000 mg single oral dose). Each chemical is distributed across nine flow-limited tissue compartments (fat, kidney, muscle, rapidly perfused, slowly perfused, liver, arterial blood, venous blood) with a separate hepatic sub-compartment for the conjugates. Liver metabolism uses Michaelis-Menten kinetics with partial substrate inhibition for CYP-mediated NAPQI formation, sulfation by SULT (cofactor PAPS), and glucuronidation by UGT (cofactor UDP-glucuronic acid, GA); both cofactors are tracked as relative-amount states with zeroth-order resynthesis. Renal elimination is linear, scaled by body weight. Oral absorption is encoded as a bi-exponential gastric-emptying rate function (Tg, Tp) added directly to the liver compartment, with dose-dependent bioavailability fa = 0.0005*Dose_mg + 0.37 (Dose < 1000 mg) or 0.88 (>= 1000 mg). The model is deterministic typical-value (no IIV, no residual error) – the DDMORE bundle exposes only the Bayesian posterior-mean parameters from Forward_APAP1.in, not the per-individual variability or measurement-error distributions reported in the publication.
Phenobarbital (Voller 2017) One-compartment first-order-absorption population PK model for phenobarbital in preterm and term newborns (Voller 2017), as packaged in DDMORE Foundation Model Repository entry DDMODEL00000256.
Pimasertib (Girard 2012) Joint K-PD / cumulative-logit Markov / Weibull-TTE-dropout model for ocular adverse events and treatment discontinuation in advanced solid-tumour and hematological-malignancy patients dosed with the MEK inhibitor pimasertib in two phase I dose-escalation studies (Girard 2012; DDMODEL00000215)
Prostate (Wilbaux 2015) Joint semi-mechanistic kinetic-pharmacodynamic (K-PD) model of circulating tumour cell (CTC) count and prostate-specific antigen (PSA) longitudinal kinetics during chemotherapy and/or hormonotherapy in adults with metastatic castration-resistant prostate cancer (mCRPC) (Wilbaux 2015 / DDMODEL00000261). The structural model couples (i) two parallel first-order K-PD compartments for chemotherapy and hormonotherapy (no PK data – virtual unit doses per cycle), (ii) a latent, dimensionless tumour-burden variable LV(t) governed by an indirect-response ODE with saturable Emax inhibition by both treatment compartments and a steady-state-anchored production rate, (iii) a CTC count that is the difference between two integrals of K0 * LV separated by the cell-lifespan delay LS (cell lifespan model implemented via a parallel delayed copy of the LV dynamics), and (iv) a PSA concentration following a non-steady-state indirect-response ODE driven by the same delayed LV. The published likelihood combines a negative-binomial count distribution for CTC observations (mean = alpha * CTC_total, overdispersion OVDP) with an exponential residual error on log-transformed PSA observations (W1 = 0.30); the full BLOCK(9) correlated $OMEGA across the nine PD parameters is preserved verbatim. nlmixr2’s parser cannot natively express the .mod’s F_FLAG=2 (-2ln-likelihood) negative-binomial branch, so this implementation provides typical-value mechanistic outputs (NCTC = CTC alpha, PSA, log_PSA) with placeholder additive residual errors on each – see vignette Assumptions and deviations.
Remoxipride (Stevens 2012) Mechanism-based PK/PD model for the prolactin response to remoxipride in rats: 3-compartment plasma + brain-ECF + peripheral PK with parallel intranasal absorption (systemic and direct nose-to-brain), feeding a pool model for prolactin synthesis (with positive feedback), storage in lactotrophs, release into plasma, and elimination, where remoxipride brain-ECF concentration drives an Emax stimulation of prolactin release
Ribavirin (Laouenan 2015) Hemoglobin turnover (indirect-response) model describing ribavirin-induced anemia in HCV genotype-1 cirrhotic patients on telaprevir- or boceprevir-based triple therapy (Laouenan 2015). Hemoglobin (g/dL) follows a kin/kout indirect-response ODE in which ribavirin inhibits hemoglobin synthesis with an Imax = 1, EC50 form. The ribavirin concentration time-course is reconstructed analytically from per-subject empirical-Bayes regressors (CSS_RBV, K_RBV) supplied as data columns from a separately fitted Laouenan 2015 upstream ribavirin popPK fit; this PD model does not instantiate the PK ODE itself. Distributed in the DDMORE Foundation Model Repository as DDMODEL00000285; the linked publication fits the same equations to 15 ANRS-CO20-CUPIC patients (9 telaprevir, 6 boceprevir).
Rifampicin (Clewe 2016) Multistate Tuberculosis Pharmacometric (MTP) model for in vitro M. tuberculosis H37Rv natural growth (no drug effect): three bacterial states (fast-multiplying, slow-multiplying, non-multiplying) with Gompertz growth on F and time-varying F->S transfer (Clewe 2016; DDMODEL00000240, scenario 4 – natural-growth backbone of the framework that the publication then couples with rifampicin exposure-response; the bundled .mod ships only the natural-growth scaffold)
Rifampicin (Clewe 2018) Multistate Tuberculosis Pharmacometric (MTP) model coupled with the General Pharmacodynamic Interaction (GPDI) model for the triple combination of rifampicin (CONMED_RIF_CC), isoniazid (CONMED_INH_CC), and ethambutol (CONMED_EMB_CC) against in vitro Mycobacterium tuberculosis B1585 (Clewe 2018, scenario = 4). Three bacterial subpopulations (fast-multiplying Fbugs, slow-multiplying Sbugs, non-replicating Nbugs) exchange via first-order rates and a time-dependent F-to-S transfer; CONMED_INH_CC adaptive resistance is captured by a two-state ARON / AROFF system that dynamically shifts the CONMED_INH_CC EC50 on the F and S subpopulations. Each drug acts on each subpopulation through a Hill or hyperbolic exposure-response, combined across drugs via Bliss independence on Fbugs and linear addition on Sbugs and Nbugs; pairwise GPDI interaction parameters shift the Emax / EC50 of each affected drug-effect term. Drug exposures (CONMED_RIF_CC, CONMED_INH_CC, CONMED_EMB_CC) are time-fixed in vitro concentrations supplied as data covariates.
Rifampicin (Svensson 2018) One-compartment population PK model for high-dose oral rifampicin in adult pulmonary tuberculosis patients (Svensson 2018, HIGHRIF1), with closed-form transit-compartment absorption (mean transit time and Erlang shape estimated), Michaelis-Menten clearance scaled by an auto-induced enzyme turnover compartment, fat-free-mass allometric scaling on Vmax (0.75) and central volume (1.0), and a saturable dose-dependent bioavailability anchored at the 450 mg reference dose.
Rifampicin (Wilkins 2008) One-compartment population PK model for oral rifampicin in adult South African pulmonary tuberculosis patients (Wilkins 2008), with an analytical transit-compartment chain (Savic 2007 form) preceding first-order absorption, multiplicative formulation effects of single-drug-combination (vs fixed-dose-combination reference) on apparent oral clearance and on mean transit time, IIV on CL/V (correlated)/Ka/MTT/NN, and 6-occasion inter-occasion variability on log-CL and log-MTT.
Sibrotuzumab (Kloft 2004) Two-compartment population PK model for sibrotuzumab in adults with metastatic FAP-positive cancer (Kloft 2004), with parallel linear and Michaelis-Menten elimination from the central compartment and a fixed linear body-weight covariate (centered at 75 kg) on linear CL, central and peripheral volumes, and Vmax.
Sultiame (Dao 2020) Population PK model for sultiame in healthy adult volunteers with non-linear distribution into erythrocytes (saturable binding to a putative red-blood-cell carrier). Four-compartment structure: depot (oral absorption, KA fixed at 1/h), central (plasma), erythrocytes (drug bound to a saturable carrier in red blood cells parameterised by KON, KOFF, BTOT), and urine (cumulative urinary excretion as a fraction QREN of total elimination). Drug binding to erythrocytes is written in mass-action form on amounts (KON in 1/(h*mg)). DDMORE Foundation Model Repository entry DDMODEL00000298, fit on 4 healthy volunteers (433 observations) by NONMEM ADVAN13 FOCEI.
Sunitinib (Hansson 2013a) Population PD biomarker model for sunitinib in adults with imatinib-resistant gastrointestinal stromal tumours (GIST). Four indirect-response compartments for the soluble biomarkers VEGF, sVEGFR-2, sVEGFR-3, and sKIT, each driven by a per-cycle drug-exposure summary AUC = DOSE / CLI. Sigmoid Imax inhibition with a Hill coefficient applies to VEGF (Kout) and sVEGFR-2 (Kin); simple Imax inhibition applies to sVEGFR-3 (Kin) and sKIT (Kin). A linear disease-progression term increases the baseline of VEGF and sKIT over time. The PD model has no PK ODE: the user supplies DOSE (current daily sunitinib dose, mg, time-varying with on/off cycling) and CLI (subject-specific posthoc total plasma clearance, L/h, from an upstream popPK fit) as data columns. No covariates other than the two exposure inputs.
Sunitinib (Hansson 2013b) Population PD tumor growth inhibition model for sunitinib in adults with imatinib-resistant gastrointestinal stromal tumours (GIST). The longitudinal sum of longest tumor diameters (SLD) is modelled as exponential growth (KG) with three additive shrinkage drivers – exposure-driven (KDRUG * AUC), and the model-predicted relative-from-baseline changes in soluble KIT (sKIT) and soluble VEGFR-3 (sVEGFR-3) acting through the rate constants KSKIT and KVEGFR3 – modulated by an exponential time-dependent resistance term (LAMBDA). The two soluble-biomarker time-courses are simulated in-model as 3 indirect-response compartments (treated sKIT, placebo / untreated sKIT, sVEGFR-3) driven by simple-Imax inhibition of Kin with the per-cycle exposure summary AUC = DOSE / CLI; the placebo-arm sKIT compartment carries a linear disease-progression term. The PD model has no PK ODE and consumes individual posthoc upstream-PD parameters (BAS_SKIT, MRT_SKIT, EC50_SKIT, SLOPE_SKIT, BAS_SVEGFR3, MRT_SVEGFR3, EC50_SVEGFR3) plus posthoc upstream-PK clearance (CLI) and observed baseline tumor size (TUMSZ, mm) as data covariates. IIV on tumor growth (KG), drug effect (KDRUG), and the sKIT-driven shrinkage rate (KSKIT); LAMBDA’s IIV is held at zero in the source; the IPP-style baseline-residual eta is fixed-variance 1 with proportional residual scaling (Dansirikul / Silber / Karlsson 2008).
Sunitinib (Hansson 2013c) Population PD model of fatigue (NCI-CTC grades 0 / 1 / 2 / 3+) in adults with imatinib-resistant gastrointestinal stromal tumours (GIST) on sunitinib. A first-order Markov + proportional-odds (PO) likelihood describes the fatigue-grade transition probabilities at each scheduled visit, conditional on the previous fatigue grade. The cumulative-logit baselines are shifted per starting state by a placebo coefficient on the relative change in plasma soluble VEGFR-3 (sVEGFR-3) from baseline; sVEGFR-3 itself follows an indirect-response turnover driven by the per-cycle drug-exposure summary AUC = DOSE / CLI. The PD model has no PK ODE and consumes individual posthoc upstream-PD parameters (BAS_SVEGFR3, MRT_SVEGFR3, EC50_SVEGFR3) and posthoc upstream-PK clearance (CLI) as data covariates. Random effects are diagonal across the four per-state baseline logits.
Sunitinib (Schindler 2016) Joint pharmacodynamic model for sunitinib in advanced GIST coupling a five-lesion indirect-response model of [18F]FDG-PET SUVmax with a per-subject sum-of-longest-diameters (SLD) tumor-growth-inhibition module and constant-baseline-hazard Weibull time-to-event sub-models for overall survival and study dropout (Schindler 2016 / DDMODEL00000221). Sunitinib exposure enters via an effect compartment driven by a per-day AUC = DOSE / CLI; the OS hazard depends on the per-subject week-1 maximum across-lesion relative SUVmax change from baseline, RCFB1MAX.
Survival (Zecchin 2016) Time-to-event model for overall survival (OS) in advanced epithelial ovarian cancer (Zecchin 2016 / DDMODEL00000218): Weibull baseline hazard with covariate effects of normalised baseline SLD (TUM_SLD / 70 mm), tumour-size-ratio TSR(t) capped at week 12, time-varying new-lesion indicator (NEW_LESION), and binary ECOG performance status, with the underlying SLD trajectory (subject-specific tumour-growth and drug-cytotoxicity rate constants from the upstream Zecchin 2016 SLD model) integrated inline.
Tamoxifen (TerHeine 2014) Joint parent-metabolite population PK model for tamoxifen and endoxifen at steady state in adult breast-cancer patients, with CYP2D6 and CYP3A4/5 individual-activity covariates on the endoxifen-formation clearance
Tgfbinhibitor (Lestini 2015) One-compartment first-order absorption PK with indirect-response biomarker turnover (E represents fractional inhibition of TGF-beta signalling) for a small-molecule TGF-beta inhibitor in oncology, simplified by Lestini 2015 from Bueno et al. for use as a population PK/PD test bench in adaptive-design simulations.
Tumorovarian (Zecchin 2016) Tumour-size dynamics model (sum of longest diameters, SLD) for advanced epithelial ovarian cancer with independent additive carboplatin and gemcitabine cytotoxic effects (Zecchin 2016 / DDMODEL00000217): exponential SLD growth with two drug-exposure-driven death-rate terms (no resistance, no synergy), additive residual error, and M3-method handling of below-LLOQ observations in the source NONMEM run.