Skip to contents

Lorlatinib (Chen 2021)

Source: vignettes/articles/Chen_2021_lorlatinib.Rmd
Chen_2021_lorlatinib.Rmd

Model and source

  • Citation: Chen J, Houk B, Pithavala YK, Ruiz-Garcia A. (2021). Population pharmacokinetic model with time-varying clearance for lorlatinib using pooled data from patients with non-small cell lung cancer and healthy participants. CPT Pharmacometrics Syst Pharmacol 10(2):148-160. doi:10.1002/psp4.12585
  • Description: 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.
  • Article: https://doi.org/10.1002/psp4.12585 (open access)
  • Supplement: Wiley Online Library supporting information (Tables S1-S2 and NONMEM control stream ListS1); attached to the same DOI landing page.

Population

Chen 2021 pooled 5,806 lorlatinib plasma concentration measurements from 425 study participants across seven studies (Chen 2021 Table 1): 330 adults with advanced ALK-positive or ROS1-positive non-small cell lung cancer (NSCLC) enrolled in the phase I/II B7461001 dose-escalation and expansion study (NCT01970865; lorlatinib 10-200 mg q.d. or 35-100 mg b.i.d.), and 95 healthy volunteers enrolled in six phase I clinical-pharmacology studies (B7461004 mass balance, B7461005 relative bioavailability, B7461007 absolute-bioavailability crossover with a single 50 mg IV infusion vs 100 mg oral tablet, B7461008 rabeprazole and high-fat-meal effect, B7461011 rifampin CYP3A4-induction, B7461016 bioequivalence).

Baseline demographics per Chen 2021 Table 3 – cohort mean 70.53 kg body weight (SD 16.89), 49.86 years age (SD 13.20), 45 % female, race distribution 52 % White / 27 % Asian / 8 % Black / 7 % Other, mean baseline serum albumin 3.92 g/dL (SD 0.58), mean baseline creatinine clearance 98.31 mL/min (SD 32.13), 5 % on concomitant proton-pump inhibitor. Renal function stages were 57 % normal / 31 % mild impairment / 12 % moderate impairment / 0.2 % severe impairment (KDOQI). Hepatic function stages were 86 % normal / 12 % mild B1 / 2 % mild B2 / 0 % moderate or severe (NCI).

The population metadata list on the model object gives the same fields programmatically:

nlmixr2lib::readModelDb("Chen_2021_lorlatinib")$population

Source trace

Every parameter’s origin is recorded next to its ini() entry in inst/modeldb/specificDrugs/Chen_2021_lorlatinib.R. The table below collects them in one place for review. All numeric values come from Chen 2021 Table 4 “Value” column unless otherwise noted; the NONMEM ListS1 control stream in the Wiley supplement provides the encoding of covariate multipliers (linear centered vs power).

Equation / parameter Value Source location
lcl (initial CL, CLI) log(9.035) Table 4 theta_CLI = 9.035 L/h
lcl_exp_inf (steady-state induced CL, CLMX) log(14.472) Table 4 theta_CLMX = 14.472 L/h
lcl_exp_kdes (auto-induction rate constant) log(0.020) Table 4 theta_IND = 0.020 h^-1; abstract “~7.25 d to functional steady state”
lvc (V2) log(120.511) Table 4 theta_V2 = 120.511 L
lvp (V3) log(154.905) Table 4 theta_V3 = 154.905 L
lq log(22.002) Table 4 theta_Q = 22.002 L/h
lka log(3.113) Table 4 theta_ka = 3.113 h^-1
ld1 (zero-order duration) log(1.148) Table 4 theta_D1 = 1.148 h
lfdepot (F1) log(0.759) Table 4 theta_F = 0.759; anchored by B7461007 IV vs oral crossover
e_wt_cl (allometry on CL, fixed) 0.75 Methods “Model development”; ListS1 TVCLI = THETA(1)*(BWT/70)**0.75
e_wt_vc (allometry on V2, fixed) 1.0 Methods “Model development”; ListS1 TVV2 = THETA(2)*(BWT/70)
e_alb_cl (linear centered) 0.00670 per g/L Table 4 theta_BALB_on_CL = 0.067 per g/dL (paper units) – converted to per g/L for canonical ALB
e_dose_lor_cl (linear centered) 0.00100 per mg/day Table 4 theta_TDOSE_on_CL = 0.001 centered at 100 mg/day
e_crcl_cl (power) 0.235 Table 4 theta_WNCL_on_CL = 0.235 centered at 100 mL/min
e_ppi_ka (linear indicator) -0.675 Table 4 theta_PPI_on_ka = -0.675
IIV block CL/F var 0.030, cov -0.006, var 0.022 Table 4 omega^2 CL / omega_F omega_CL / omega^2 F; ListS1 $OMEGA BLOCK(2) line 102-104
IIV block V2/V3 var 0.086, cov -0.017, var 0.101 Table 4 omega^2 V2 / omega_V2 omega_V3 / omega^2 V3; ListS1 $OMEGA BLOCK(2) line 105-107
IIV ka var 2.329 Table 4 omega^2 ka
propSd (oral residual) 0.438 Table 4 theta_Res_Error_for_PO = 0.438; ListS1 line 11
Auto-induction ODE CL(t) = CLI + (CLMX - CLI) * (1 - exp(-cl_exp_kdes * t)) Methods “Lorlatinib clearance estimation” (continuous canonical form; see Assumptions)
CL covariate composite CLCOV = CLBALB * CLTDOSE * CLWNCL, applied to both CLI and CLMX NONMEM ListS1 line 38
ka covariate multiplier ka * (1 + e_ppi_ka * CONMED_PPI) NONMEM ListS1 line 14-15 (KAPPI)

Virtual cohort

The Chen 2021 individual observed concentrations are not publicly available. The figures below use a virtual cohort whose covariate distribution approximates the Chen 2021 pooled analysis dataset (Table 3). Two dose regimens are simulated to illustrate model behavior:

  • 100 mg q.d. – the labelled recommended phase II dose (Chen 2021 Study B7461001 phase II arm).
  • 10 mg q.d. – the lowest phase I escalation dose (Chen 2021 Table 1), exercising the low end of the dose-nonlinearity range.

Cohort size is capped at 100 subjects per arm (below the 200-per-arm ceiling), which is ample for the illustrative VPC below.

set.seed(20260724L)

n_per_arm <- 100L
sim_duration_days <- 15L
tau <- 24  # dosing interval in hours (q.d.)

make_cohort <- function(n, dose_mg, treatment, id_offset = 0L) {
  # Baseline covariate sampler matching Chen 2021 Table 3 marginals.
  # Weight is truncated at 35 kg lower / 130 kg upper to avoid
  # unphysical tails; distributions are treated as independent
  # because Chen 2021 does not report a covariance matrix.
  subj <- tibble::tibble(
    id           = id_offset + seq_len(n),
    WT           = pmin(pmax(rnorm(n, mean = 70.53, sd = 16.89), 35), 130),
    ALB          = pmin(pmax(rnorm(n, mean = 39.2,  sd = 5.8),   25), 55),  # g/L (from Chen 2021 mean 3.92 g/dL x 10)
    CRCL         = pmin(pmax(rnorm(n, mean = 98.31, sd = 32.13), 30), 180),
    CONMED_PPI   = rbinom(n, size = 1, prob = 0.05),
    DOSE_LOR_MGD = dose_mg,
    treatment    = treatment
  )

  # Dose records: q.d. dosing, day 1 through day sim_duration_days.
  # Sequential zero-first order absorption is exercised via
  # rate = -2 (rxode2 convention: use the model's dur(depot)).
  dose_times <- seq(0, by = tau, length.out = sim_duration_days)
  doses <- tidyr::crossing(subj, time = dose_times) |>
    dplyr::mutate(evid = 1L, amt = dose_mg, cmt = "depot", rate = -2)

  # Observations: dense sampling over the last dosing interval to
  # capture Cmax_ss / AUCtau_ss, plus sparse pre-dose points over
  # the induction ramp to visualise CL(t) rise.
  last_dose <- max(dose_times)
  obs_dense <- last_dose + c(0, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 12, 16, 20, 24)
  obs_sparse <- c(1, 3, 6, 10, 24 * (2:sim_duration_days) - 0.5)  # troughs
  obs_times  <- sort(unique(c(obs_dense, obs_sparse)))
  obs <- tidyr::crossing(subj, time = obs_times) |>
    dplyr::mutate(evid = 0L, amt = NA_real_, cmt = "central", rate = NA_real_)

  dplyr::bind_rows(doses, obs) |>
    dplyr::arrange(id, time, dplyr::desc(evid))
}

events <- dplyr::bind_rows(
  make_cohort(n_per_arm, dose_mg = 100, treatment = "100 mg q.d.", id_offset =   0L),
  make_cohort(n_per_arm, dose_mg =  10, treatment = "10 mg q.d.",  id_offset = n_per_arm)
)

# Cheap regression guard against silently-collapsed subject IDs across cohorts.
stopifnot(!anyDuplicated(unique(events[, c("id", "time", "evid")])))

Simulation 

mod <- nlmixr2lib::readModelDb("Chen_2021_lorlatinib")

sim <- rxode2::rxSolve(
  mod,
  events = events,
  keep   = c("treatment", "WT", "ALB", "CRCL", "CONMED_PPI", "DOSE_LOR_MGD")
) |>
  as.data.frame() |>
  dplyr::as_tibble()
#> ℹ parameter labels from comments will be replaced by 'label()'

Replicate published behavior

Time-varying clearance rise (Chen 2021 Results paragraph 3)

Chen 2021 states that the individual clearance ramps from CLI = 9.04 L/h at single-dose to CLMX = 14.5 L/h at steady state, with an induction rate constant IND = 0.020 h^-1 corresponding to a ~1.45 d induction half-life and ~7.25 d to functional steady state (~5 induction half-lives). The typical-value CL(t) trajectory for a 100 mg q.d. subject is:

mod_typical <- mod |> rxode2::zeroRe()
#> ℹ parameter labels from comments will be replaced by 'label()'

typical_events <- events |>
  dplyr::filter(id == 1L)  # a single 100 mg q.d. subject with 70-kg-ish covariates

sim_typical <- rxode2::rxSolve(
  mod_typical, events = typical_events,
  keep = c("treatment")
) |>
  as.data.frame() |>
  dplyr::as_tibble()
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalfdepot', 'etalvc', 'etalvp', 'etalka'

sim_typical |>
  dplyr::filter(time <= 24 * 15) |>
  ggplot(aes(time / 24, cl)) +
  geom_line(color = "steelblue", linewidth = 0.9) +
  geom_hline(yintercept = 9.035, linetype = 2, color = "grey40") +
  geom_hline(yintercept = 14.472, linetype = 2, color = "grey40") +
  annotate("text", x = 0.3, y = 9.035, label = "CLI = 9.04 L/h", vjust = -0.5, hjust = 0, size = 3.2) +
  annotate("text", x = 0.3, y = 14.472, label = "CLMX = 14.5 L/h", vjust = -0.5, hjust = 0, size = 3.2) +
  scale_x_continuous(breaks = c(0, 1.45, 3, 5, 7.25, 10, 15)) +
  labs(x = "Time (days)", y = "Typical clearance CL(t) (L/h)",
       title = "Auto-induction of lorlatinib CL to steady state",
       caption = "Reproduces the Chen 2021 abstract narrative (CLI = 9.04, CLMX = 14.5 L/h, ~7.25 d to functional steady state).")

Concentration-time profile at 100 mg q.d. 

sim |>
  dplyr::filter(!is.na(Cc), Cc > 1e-3) |>  # log-scale plot; drop pre-first-dose zeros only
  dplyr::group_by(time, treatment) |>
  dplyr::summarise(
    Q05 = quantile(Cc, 0.05, na.rm = TRUE),
    Q50 = quantile(Cc, 0.50, na.rm = TRUE),
    Q95 = quantile(Cc, 0.95, na.rm = TRUE),
    .groups = "drop"
  ) |>
  ggplot(aes(time / 24, Q50, color = treatment, fill = treatment)) +
  geom_ribbon(aes(ymin = Q05, ymax = Q95), alpha = 0.2, color = NA) +
  geom_line(linewidth = 0.7) +
  scale_y_log10() +
  facet_wrap(~treatment, scales = "free_y") +
  labs(x = "Time (days)", y = "Lorlatinib plasma concentration Cc (ng/mL)",
       title = "Simulated plasma concentration profiles",
       caption = "Median and 5th-95th percentile ribbon over 100 virtual subjects per arm.") +
  theme(legend.position = "none")

PKNCA validation at steady state

Chen 2021 Methods (“Sensitivity analysis of covariate effects on exposure”) defines the reference typical individual (70 kg, no PPI, WNCL = 100 mL/min, BALB = 4 g/dL, 100 mg q.d.) as having steady-state Cmax = 606 ng/mL and AUCtau = 5180 ng.h/mL over the last (day 15) dosing interval. The PKNCA validation below reproduces these steady-state exposure metrics from the virtual cohort at 100 mg q.d.

# PKNCA input: use ONLY !is.na(Cc) so the time-zero row is preserved
# (pknca-recipes.md "Time-zero records (mandatory)").
sim_nca <- sim |>
  dplyr::filter(!is.na(Cc)) |>
  dplyr::select(id, time, Cc, treatment)

# Defensive time-zero row per (id, treatment). Extravascular pre-dose = 0.
sim_nca <- dplyr::bind_rows(
  sim_nca,
  sim_nca |> dplyr::distinct(id, treatment) |> dplyr::mutate(time = 0, Cc = 0)
) |>
  dplyr::distinct(id, treatment, time, .keep_all = TRUE) |>
  dplyr::arrange(id, treatment, time)

conc_obj <- PKNCA::PKNCAconc(
  sim_nca, Cc ~ time | treatment + id,
  concu = "ng/mL", timeu = "h"
)

dose_df <- events |>
  dplyr::filter(evid == 1L) |>
  dplyr::select(id, time, amt, treatment)

dose_obj <- PKNCA::PKNCAdose(
  dose_df, amt ~ time | treatment + id,
  doseu = "mg"
)

# Steady-state interval: the last dosing interval (day 14 to day 15).
last_dose <- max(dose_df$time)
end_ss    <- last_dose + tau

intervals <- data.frame(
  start   = last_dose,
  end     = end_ss,
  cmax    = TRUE,
  tmax    = TRUE,
  auclast = TRUE,
  cav     = TRUE,
  ctrough = TRUE
)

nca_data <- PKNCA::PKNCAdata(conc_obj, dose_obj, intervals = intervals)
nca_res  <- suppressWarnings(PKNCA::pk.nca(nca_data))

Comparison against the published typical-individual exposure 

# Published typical-individual exposure at 100 mg q.d. (Chen 2021 Methods
# 'Sensitivity analysis of covariate effects on exposure'). Chen 2021 does
# not report a published 10 mg q.d. Cmax/AUCtau in the Results text; only
# the 100 mg q.d. row can be cross-checked.
published <- tibble::tribble(
  ~treatment,     ~cmax, ~auclast, ~tmax,
  "100 mg q.d.",   606,  5180,     2.0
)

cmp <- nlmixr2lib::ncaComparisonTable(
  simulated     = nca_res,
  reference     = published,
  by            = "treatment",
  units         = c(cmax = "ng/mL", auclast = "ng*h/mL", tmax = "h"),
  tolerance_pct = 25
)

knitr::kable(
  cmp,
  caption = "Simulated (median across 100 virtual subjects) vs. Chen 2021 typical-individual steady-state exposure at 100 mg q.d. Rows differing from reference by >25% receive a trailing asterisk. AUC0-tau over the day 14-15 dosing interval is compared with the paper's `AUCtau` because the model's continuous-CL formulation and stochastic residual error do not target the exact typical-value trajectory the paper's forest-plot simulation reported."
)
Simulated (median across 100 virtual subjects) vs. Chen 2021 typical-individual steady-state exposure at 100 mg q.d. Rows differing from reference by >25% receive a trailing asterisk. AUC0-tau over the day 14-15 dosing interval is compared with the paper’s AUCtau because the model’s continuous-CL formulation and stochastic residual error do not target the exact typical-value trajectory the paper’s forest-plot simulation reported.
NCA parameter treatment Reference Simulated % diff
Cmax (ng/mL) 100 mg q.d. 606 540 -11.0%
Tmax (h) 100 mg q.d. 2 1.5 -25.0%
AUClast (ng*h/mL) 100 mg q.d. 5180 5300 +2.4% 

fn <- attr(cmp, "footnote")
if (!is.null(fn)) cat(fn, sep = "\n")

Steady-state summary at both doses 

nca_tbl <- as.data.frame(nca_res$result) |>
  dplyr::filter(PPTESTCD %in% c("cmax", "auclast", "tmax", "cav", "ctrough")) |>
  dplyr::group_by(treatment, PPTESTCD) |>
  dplyr::summarise(
    median = median(PPORRES, na.rm = TRUE),
    q05    = quantile(PPORRES, 0.05, na.rm = TRUE),
    q95    = quantile(PPORRES, 0.95, na.rm = TRUE),
    .groups = "drop"
  ) |>
  dplyr::mutate(
    Parameter = dplyr::recode(PPTESTCD,
                              cmax    = "Cmax,ss (ng/mL)",
                              auclast = "AUC0-tau,ss (ng*h/mL)",
                              tmax    = "Tmax,ss (h)",
                              cav     = "Cavg,ss (ng/mL)",
                              ctrough = "Ctrough,ss (ng/mL)"),
    median = signif(median, 3),
    q05    = signif(q05, 3),
    q95    = signif(q95, 3)
  ) |>
  dplyr::select(treatment, Parameter, median, q05, q95) |>
  dplyr::rename(
    "Treatment"          = treatment,
    "NCA parameter"      = Parameter,
    "Median"             = median,
    "5th percentile"     = q05,
    "95th percentile"    = q95
  )

knitr::kable(
  nca_tbl,
  caption = "Simulated steady-state NCA distribution over the day 14-15 dosing interval."
)
Simulated steady-state NCA distribution over the day 14-15 dosing interval.
Treatment NCA parameter Median 5th percentile 95th percentile
10 mg q.d. AUC0-tau,ss (ng*h/mL) 597.0 323.0 1010.0
10 mg q.d. Cavg,ss (ng/mL) 24.9 13.4 42.0
10 mg q.d. Cmax,ss (ng/mL) 55.1 26.6 97.7
10 mg q.d. Ctrough,ss (ng/mL) NA NA NA
10 mg q.d. Tmax,ss (h) 1.5 1.5 4.0
100 mg q.d. AUC0-tau,ss (ng*h/mL) 5300.0 3310.0 8960.0
100 mg q.d. Cavg,ss (ng/mL) 221.0 138.0 373.0
100 mg q.d. Cmax,ss (ng/mL) 540.0 265.0 880.0
100 mg q.d. Ctrough,ss (ng/mL) NA NA NA
100 mg q.d. Tmax,ss (h) 1.5 1.5 6.0

Assumptions and deviations

  • Continuous auto-induction canonical form vs Chen 2021 piecewise SD-flag implementation. Chen 2021 estimated the auto-induction parameters using a hard NONMEM switch on the SD (single-dose) data-column flag: CL = CLI when SD = 1, and CL = CLMX * (1 - exp(-IND * TIME)) when SD = 0 (NONMEM ListS1 line 63-64). This file implements the continuous canonical form the paper’s narrative describes: CL(t) = CLI + (CLMX - CLI) * (1 - exp(-k_ind * t)), rising smoothly from CLI at t = 0 (first-dose regime) to CLMX at t -> infinity (steady state). The two forms are numerically equivalent at the paper’s estimation anchors (single-dose CL at t = 0 and post-day-15 steady-state CL, where exp(-0.020 * 336) < 0.002); the continuous form additionally supports arbitrary dosing histories in downstream simulation without requiring a SD covariate column.
  • Baseline serum albumin unit conversion. Chen 2021 reports the paper’s theta_BALB_on_CL = 0.067 per g/dL centered at BALB = 4 g/dL (the US clinical-lab unit for albumin). The canonical nlmixr2lib covariate ALB is in g/L (SI). The model file rescales the coefficient by a factor of 10 (e_alb_cl = 0.00670 per g/L) and centers at 40 g/L. The rescaled term reproduces the paper’s +/-5.3% CL shift at the 10th-percentile ALB = 3.2 g/dL = 32 g/L and +/-4% CL shift at the 90th-percentile ALB = 4.6 g/dL = 46 g/L.
  • Weight-normalized (WNCL) vs BSA-normalized (CRCL) creatinine clearance. Chen 2021 derives WNCL from the Cockcroft-Gault BCCL by weight- normalization: WNCL = BCCL / (BWT / 70)^0.75 (approximate; the paper cites Rhodin et al. 2009 for the exact form and the exponent choice). The canonical CRCL column in nlmixr2lib holds BSA-normalized creatinine clearance in mL/min/1.73 m^2 (per inst/references/covariate-columns.md). For the reference 70-kg adult with BSA ~ 1.73 m^2, weight-normalized and BSA-normalized creatinine clearance are numerically nearly identical, so the CRCL column with reference 100 mL/min is a compatible substitute. Downstream users applying the model to obese / underweight cohorts should either (a) precompute WNCL exactly per the Chen 2021 derivation and supply it in the CRCL column, or (b) accept the small numerical discrepancy that arises from the BSA-vs-weight normalization when using BSA-normalized CRCL from routine clinical data.
  • Route-specific residual error simplified to the oral route. Chen 2021 fits separate log-scale residual SDs for the IV route (theta = 0.115; Study B7461007 absolute-bioavailability arm) and the oral route (theta = 0.438) via a NONMEM IF(ROUT.EQ.28) branch (ListS1 line 77-81). Because the labelled lorlatinib route of administration is oral and the simulation library targets clinical q.d. tablet dosing, this file encodes only the oral residual SD as propSd = 0.438. The IV residual SD is documented here for reviewers examining absolute-bioavailability scenarios; a future model file could add a route indicator if IV simulation becomes a priority use case.
  • Inter-occasion variability (IOV) omitted. The Chen 2021 NONMEM control stream ListS1 does not enable an occasion-level $OMEGA block, so no IOV is encoded here.
  • Covariate joint distribution assumed independent. Chen 2021 Table 3 reports the marginal distributions of each covariate but not their joint distribution. The virtual cohort in this vignette samples WT, ALB, CRCL, and CONMED_PPI independently from their marginals; downstream users who need to preserve the empirical BWT-BSA-CRCL correlation observed in real clinical cohorts should either digitize the covariate correlation matrix from a separate source or perform copula-based sampling.
  • B7461001 phase I dose-escalation regimens (10-200 mg q.d. and 35-100 mg b.i.d.) are not exercised in the vignette. The illustrative simulation above uses only the 100 mg q.d. labelled dose and a 10 mg q.d. low-end contrast; the full escalation grid is best exercised via a parametric sweep in a downstream simulation script, not embedded in the vignette narrative.

Errata

No errata for Chen 2021 CPT Pharmacometrics Syst Pharmacol 10(2):148-160 were located as of 2026-07-24 (PubMed Central author-correction feed and Wiley Online Library corrections listing consulted).