Alirocumab full TMDD in pediatric HeFH and healthy adults (Kovalenko 2025)
Source:vignettes/articles/Kovalenko_2025_alirocumab.Rmd
Kovalenko_2025_alirocumab.RmdModel and source
- Citation: Kovalenko P, Harnisch L, Mendell J, Wang Y, Davis JD, DiCioccio AT. Target-Mediated Modeling of Alirocumab in Adolescents and Children >=8 to <12 Years of Age Using Phase II and III Data. Clin Pharmacol Drug Dev. 2025;14(5):347. doi:10.1002/cpdd.1523
- Article: Clin Pharmacol Drug Dev 14(5):347 (2025) (open access, PMC12044327)
Kovalenko et al. 2025 describe total alirocumab and free PCSK9 in children and adolescents with heterozygous familial hypercholesterolaemia (HeFH) with a full target-mediated drug disposition (TMDD) model. The analysis has two steps, and the library carries one model file for each:
-
Kovalenko_2025_alirocumab_adult– the model fitted to two densely sampled phase I studies in healthy adults (IV and SC single doses), which estimated every structural parameter (Table 2, “Adults” column). -
Kovalenko_2025_alirocumab_pediatric– the primary pediatric covariate model (Table 2, “Pediatric population” column). Bioavailability, the absorption and transit rates, and the intercompartmental rate constants are fixed to the adult estimates; weight on Vc and high-dose statins (HDS) on PCSK9 synthesis are the retained covariates.
The structure (Figure 1 of the paper) is the same in both:
- SC dose -> depot transit compartment (
ktr) -> transit compartment (ka) -> central compartment, with bioavailability F. - Free alirocumab
L, free PCSK9R, the 1:1 complexL:Rand the 1:2 complexL:R:Rall distribute to a peripheral compartment with the same rate constantskcpandkpc. - In the central compartment
L + R <-> L:RandL:R + R <-> L:R:R, each withkonandkoff(fixed to Biacore values). -
Lis eliminated atkeL, both complexes atkeLR, free PCSK9 atkeR = ksyn / R0, and PCSK9 is synthesised at the zero-order rateksyn. - Observations are total alirocumab (
L + L:R + L:R:R, the total assay) and free PCSK9 in the central compartment, each with a proportional residual error.
mod_ped <- readModelDb("Kovalenko_2025_alirocumab_pediatric")
mod_adult <- readModelDb("Kovalenko_2025_alirocumab_adult")
ui_ped <- rxode2::rxode2(mod_ped)
#> ℹ parameter labels from comments will be replaced by 'label()'
ui_adult <- rxode2::rxode2(mod_adult)
#> ℹ parameter labels from comments will be replaced by 'label()'
ui_ped
#> ── rxode2-based free-form 10-cmt ODE model ─────────────────────────────────────
#> ── Initalization: ──
#> Fixed Effects ($theta):
#> lktr lka lfdepot
#> 3.1135153 -1.0996128 -0.4231200
#> lk12 lk21 lvc
#> -1.6398971 -1.2729657 0.4946962
#> lkel lkint lksyn
#> -2.5257286 -0.9888614 1.7509375
#> lr0 lkon lkoff
#> 0.8878913 4.0587174 4.6634391
#> e_wt_vc e_conmed_statin_hi_ksyn propSd
#> 0.7240000 0.5760000 0.2470000
#> propSd_Ctarget
#> 0.4140000
#>
#> Omega ($omega):
#> etalvc etalkint etalksyn etalr0 etalkel
#> etalvc 0.011664 0.000000 0.000000 0.000000 0.000000
#> etalkint 0.000000 0.126736 0.000000 0.000000 0.000000
#> etalksyn 0.000000 0.000000 0.379456 0.000000 0.000000
#> etalr0 0.000000 0.000000 0.000000 0.076729 0.000000
#> etalkel 0.000000 0.000000 0.000000 0.000000 0.026896
#> attr(,"lotriLabels")
#> [1] "Table 2 'SD(ln[Vc])' pediatric 0.108 -> variance 0.108^2"
#> [2] "Table 2 'SD(ln[keLR])' pediatric 0.356 -> variance 0.356^2"
#> [3] "Table 2 'SD(ln[ksyn])' pediatric 0.616 -> variance 0.616^2"
#> [4] "Table 2 'SD(ln[R0])' pediatric 0.277 -> variance 0.277^2"
#> [5] "Table 2 'SD(ln[keL])' pediatric 0.164 -> variance 0.164^2"
#> attr(,"lotriFix")
#> etalvc etalkint etalksyn etalr0 etalkel
#> etalvc FALSE FALSE FALSE FALSE FALSE
#> etalkint FALSE FALSE FALSE FALSE FALSE
#> etalksyn FALSE FALSE FALSE FALSE FALSE
#> etalr0 FALSE FALSE FALSE FALSE FALSE
#> etalkel FALSE FALSE FALSE FALSE FALSE
#>
#> States ($state or $stateDf):
#> Compartment Number Compartment Name
#> 1 1 depot
#> 2 2 transit1
#> 3 3 central
#> 4 4 peripheral1
#> 5 5 target
#> 6 6 target_peripheral1
#> 7 7 complex
#> 8 8 complex_peripheral1
#> 9 9 complex_db
#> 10 10 complex_db_peripheral1
#> ── Multiple Endpoint Model ($multipleEndpoint): ──
#> variable cmt dvid*
#> 1 Cc ~ … cmt='Cc' or cmt=11 dvid='Cc' or dvid=1
#> 2 Ctarget ~ … cmt='Ctarget' or cmt=12 dvid='Ctarget' or dvid=2
#> * If dvids are outside this range, all dvids are re-numered sequentially, ie 1,7, 10 becomes 1,2,3 etc
#>
#> ── μ-referencing ($muRefTable): ──
#> theta eta level
#> 1 lvc etalvc id
#> 2 lkel etalkel id
#> 3 lkint etalkint id
#> 4 lksyn etalksyn id
#> 5 lr0 etalr0 id
#>
#> ── Model (Normalized Syntax): ──
#> function() {
#> compartmentData <- list(depot = list(analyte = "alirocumab",
#> units = "nmol", specimen = "administration site", verified = TRUE),
#> transit1 = list(analyte = "alirocumab", units = "nmol",
#> specimen = "administration site", verified = TRUE),
#> central = list(analyte = "alirocumab (free)", units = "nmol",
#> specimen = "serum", verified = TRUE), peripheral1 = list(analyte = "alirocumab (free)",
#> units = "nmol", specimen = "not applicable", verified = TRUE),
#> target = list(analyte = "PCSK9 (free)", units = "nmol",
#> specimen = "serum", verified = TRUE), target_peripheral1 = list(analyte = "PCSK9 (free)",
#> units = "nmol", specimen = "not applicable", verified = TRUE),
#> complex = list(analyte = "alirocumab:PCSK9 (1:1 complex)",
#> units = "nmol", specimen = "serum", verified = TRUE),
#> complex_peripheral1 = list(analyte = "alirocumab:PCSK9 (1:1 complex)",
#> units = "nmol", specimen = "not applicable", verified = TRUE),
#> complex_db = list(analyte = "alirocumab:PCSK9:PCSK9 (1:2 complex)",
#> units = "nmol", specimen = "serum", verified = TRUE),
#> complex_db_peripheral1 = list(analyte = "alirocumab:PCSK9:PCSK9 (1:2 complex)",
#> units = "nmol", specimen = "not applicable", verified = TRUE))
#> covariateData <- list(WT = list(description = "Body weight at baseline",
#> units = "kg", type = "continuous", reference_category = NULL,
#> notes = "Baseline (time-fixed) weight; the paper repeated covariate selection with time-dependent weight and kept the baseline form because time-dependent covariates gave unstable OFVs (Methods, 'Population PK Analysis'). Power effect on Vc centred at 50 kg: the dataset median 50.1 kg was replaced by 50 kg because treatment groups were assigned at the 50 kg cut-off (Methods). Analysis-set weight mean 52.0 kg (SD 18.8).",
#> source_name = "weight"), CONMED_STATIN_HI = list(description = "Concomitant high-dose statin therapy",
#> units = "(binary)", type = "binary", reference_category = "0 (no statin or low-dose statin)",
#> notes = "The paper's HDS covariate (Table 2 'ksyn ~ HDS'). The paper does not state the dose threshold that defines a high-dose statin; it reports that 19 participants received HDS and that pediatric statin doses were mostly <=20 mg once daily (Discussion). Multiplicative exponential effect on PCSK9 synthesis rate, ksyn = 5.76 * exp(0.576 * HDS), i.e. 10.3 nM/day on HDS (Table 2 derived row; Results 'ksyn was 79% higher').",
#> source_name = "HDS"))
#> covariatesDataExcluded <- list(AGE = list(description = "Age",
#> units = "years", type = "continuous", reference_category = NULL,
#> notes = "Tested in the stepwise covariate search (Methods) and not retained in the final model.",
#> source_name = "age"), ALB = list(description = "Serum albumin",
#> units = "g/L", type = "continuous", reference_category = NULL,
#> notes = "Tested in the stepwise covariate search (Methods) and not retained in the final model.",
#> source_name = "albumin"), CRCL = list(description = "Creatinine clearance",
#> units = "mL/min/1.73 m^2", type = "continuous", reference_category = NULL,
#> notes = "Tested in the stepwise covariate search (Methods) and not retained in the final model.",
#> source_name = "creatinine clearance"), BMI = list(description = "Body mass index",
#> units = "kg/m^2", type = "continuous", reference_category = NULL,
#> notes = "Tested in the stepwise covariate search (Methods) and not retained in the final model.",
#> source_name = "BMI"), ADA_POS = list(description = "Anti-drug antibody positive at any time",
#> units = "(binary)", type = "binary", reference_category = "0 (ADA negative)",
#> notes = "Tested in the stepwise covariate search; not statistically significant or clinically meaningful (Results).",
#> source_name = "ADA at any time"), SEXF = list(description = "Female sex indicator",
#> units = "(binary)", type = "binary", reference_category = "0 (male)",
#> notes = "Tested in the stepwise covariate search (Methods) and not retained in the final model.",
#> source_name = "sex"), CONMED_EZE = list(description = "Concomitant ezetimibe",
#> units = "(binary)", type = "binary", reference_category = "0 (no ezetimibe)",
#> notes = "Tested in the stepwise covariate search (Methods) and not retained in the final model.",
#> source_name = "ezetimibe"))
#> description <- "Full target-mediated drug disposition (TMDD) population PK model for total alirocumab and free PCSK9 in children and adolescents (8-17 years) with heterozygous familial hypercholesterolaemia (Kovalenko 2025, primary pediatric covariate model). Two-compartment disposition of free alirocumab, free PCSK9 and both alirocumab:PCSK9 complexes (1:1 and 1:2) sharing the same intercompartmental rate constants; SC absorption through one depot-transit compartment; two-step binding of alirocumab to one and then a second PCSK9 molecule in the central compartment; zero-order PCSK9 synthesis. Weight on Vc (power, 50 kg reference) and high-dose statin on PCSK9 synthesis rate. Absorption, distribution and bioavailability fixed to the adult estimates; kon and koff fixed to Biacore values."
#> paper_specific_compartments <- c("complex_db", "complex_db_peripheral1")
#> population <- list(species = "human", n_subjects = 119L,
#> n_studies = 2L, n_observations = "390 total alirocumab and 531 free PCSK9 concentrations",
#> age_range = "8-17 years", age_mean = "13.1 years (SD 2.86)",
#> weight_mean = "52.0 kg (SD 18.8)", sex_female_pct = 57.5,
#> disease_state = "Children and adolescents with heterozygous familial hypercholesterolaemia (HeFH)",
#> dose_range = "Weight-banded SC alirocumab: <50 kg 30 or 40 mg Q2W, 75 or 150 mg Q4W (up-titration to 75 mg Q2W); >=50 kg 50 or 75 mg Q2W, 150 or 300 mg Q4W (up-titration to 150 mg Q2W) (Table 1)",
#> co_medication = "More than 90% received statins; 19 participants received a high-dose statin (Discussion, Results)",
#> studies = "Phase II NCT02890992 (DFI14223, 8-week dose finding, 42 treated) and phase III NCT03510884 (EFC14643, 99 treated, 20 also in phase II)",
#> notes = "121 patients received active treatment across the two studies; one had no PK observations and one had only one post-dose observation, leaving 119 in the analysis (Methods). 42.5% male (Methods). Absorption, distribution and bioavailability parameters were fixed to the adult model fitted to two dense-sampling phase I studies in healthy adults (see Kovalenko_2025_alirocumab_adult).")
#> reference <- "Kovalenko P, Harnisch L, Mendell J, Wang Y, Davis JD, DiCioccio AT. Target-Mediated Modeling of Alirocumab in Adolescents and Children >=8 to <12 Years of Age Using Phase II and III Data. Clin Pharmacol Drug Dev. 2025;14(5):347. doi:10.1002/cpdd.1523"
#> units <- list(time = "day", dosing = "mg", concentration = "mg/L")
#> vignette <- "Kovalenko_2025_alirocumab"
#> ini({
#> lktr <- fix(3.11351530921037)
#> label("Transit rate constant depot -> transit1, ktr (1/day)")
#> lka <- fix(-1.09961278900169)
#> label("Absorption rate constant transit1 -> central, ka (1/day)")
#> lfdepot <- fix(-0.423120043346885)
#> label("SC bioavailability F (fraction)")
#> lk12 <- fix(-1.63989711991881)
#> label("Central-to-peripheral rate constant kcp (1/day)")
#> lk21 <- fix(-1.27296567581289)
#> label("Peripheral-to-central rate constant kpc (1/day)")
#> lvc <- 0.494696241836107
#> label("Central volume Vc at 50 kg (L)")
#> lkel <- -2.52572864430826
#> label("Elimination rate constant of free alirocumab keL (1/day)")
#> lkint <- -0.988861424708991
#> label("Elimination rate constant of alirocumab:PCSK9 complexes keLR (1/day)")
#> lksyn <- 1.7509374747078
#> label("PCSK9 synthesis rate ksyn without high-dose statin (nM/day)")
#> lr0 <- 0.887891257352457
#> label("Baseline free PCSK9 concentration R0 (nM)")
#> lkon <- fix(4.05871738457895)
#> label("Association rate constant kon (1/nM/day)")
#> lkoff <- fix(4.66343909411207)
#> label("Dissociation rate constant koff (1/day)")
#> e_wt_vc <- 0.724
#> label("Power exponent of body weight (WT / 50 kg) on Vc (unitless)")
#> e_conmed_statin_hi_ksyn <- 0.576
#> label("Exponential effect of high-dose statin on ksyn, exp(theta * HDS) (unitless)")
#> propSd <- c(0, 0.247)
#> label("Proportional residual error, total alirocumab (fraction)")
#> propSd_Ctarget <- c(0, 0.414)
#> label("Proportional residual error, free PCSK9 (fraction)")
#> etalvc ~ 0.011664
#> label("Table 2 'SD(ln[Vc])' pediatric 0.108 -> variance 0.108^2")
#> etalkint ~ 0.126736
#> label("Table 2 'SD(ln[keLR])' pediatric 0.356 -> variance 0.356^2")
#> etalksyn ~ 0.379456
#> label("Table 2 'SD(ln[ksyn])' pediatric 0.616 -> variance 0.616^2")
#> etalr0 ~ 0.076729
#> label("Table 2 'SD(ln[R0])' pediatric 0.277 -> variance 0.277^2")
#> etalkel ~ 0.026896
#> label("Table 2 'SD(ln[keL])' pediatric 0.164 -> variance 0.164^2")
#> })
#> model({
#> mw_alirocumab <- 146000
#> nmol_per_mg <- 1e+06/mw_alirocumab
#> ktr <- exp(lktr)
#> ka <- exp(lka)
#> fdepot <- exp(lfdepot)
#> k12 <- exp(lk12)
#> k21 <- exp(lk21)
#> vc <- exp(lvc + etalvc) * (WT/50)^e_wt_vc
#> kel <- exp(lkel + etalkel)
#> kint <- exp(lkint + etalkint)
#> ksyn <- exp(lksyn + etalksyn) * exp(e_conmed_statin_hi_ksyn *
#> CONMED_STATIN_HI)
#> r0 <- exp(lr0 + etalr0)
#> kon <- exp(lkon)
#> koff <- exp(lkoff)
#> kdeg <- ksyn/r0
#> bind1 <- kon * central * target/vc - koff * complex
#> bind2 <- kon * complex * target/vc - koff * complex_db
#> d/dt(depot) <- -ktr * depot
#> d/dt(transit1) <- ktr * depot - ka * transit1
#> d/dt(central) <- ka * transit1 - kel * central - k12 *
#> central + k21 * peripheral1 - bind1
#> d/dt(peripheral1) <- k12 * central - k21 * peripheral1
#> d/dt(target) <- ksyn * vc - kdeg * target - k12 * target +
#> k21 * target_peripheral1 - bind1 - bind2
#> d/dt(target_peripheral1) <- k12 * target - k21 * target_peripheral1
#> d/dt(complex) <- bind1 - bind2 - kint * complex - k12 *
#> complex + k21 * complex_peripheral1
#> d/dt(complex_peripheral1) <- k12 * complex - k21 * complex_peripheral1
#> d/dt(complex_db) <- bind2 - kint * complex_db - k12 *
#> complex_db + k21 * complex_db_peripheral1
#> d/dt(complex_db_peripheral1) <- k12 * complex_db - k21 *
#> complex_db_peripheral1
#> target(0) <- r0 * vc
#> target_peripheral1(0) <- r0 * vc * k12/k21
#> f(depot) <- fdepot * nmol_per_mg
#> f(central) <- nmol_per_mg
#> Cc <- (central + complex + complex_db)/vc/nmol_per_mg
#> Ctarget <- target/vc
#> Cc ~ prop(propSd)
#> Ctarget ~ prop(propSd_Ctarget)
#> })
#> }Population
The pediatric model was fitted to 119 children and adolescents aged 8-17 years with HeFH from the phase II dose-finding study NCT02890992 (42 treated) and the phase III study NCT03510884 (99 treated, 20 of whom had also taken part in phase II). Mean (SD) age was 13.1 (2.86) years and weight 52.0 (18.8) kg, and 42.5% were male. Doses were weight-banded at 50 kg (Table 1 of the paper); more than 90% of patients took statins and 19 took a high-dose statin. The analysis used 390 total alirocumab and 531 free PCSK9 concentrations.
The adult model was fitted to two phase I single-dose studies in healthy adults (NCT01026597, NCT01074372) with dense sampling. Kovalenko 2025 does not report their subject counts or demographics.
str(mod_ped()$population)
#> List of 13
#> $ species : chr "human"
#> $ n_subjects : int 119
#> $ n_studies : int 2
#> $ n_observations: chr "390 total alirocumab and 531 free PCSK9 concentrations"
#> $ age_range : chr "8-17 years"
#> $ age_mean : chr "13.1 years (SD 2.86)"
#> $ weight_mean : chr "52.0 kg (SD 18.8)"
#> $ sex_female_pct: num 57.5
#> $ disease_state : chr "Children and adolescents with heterozygous familial hypercholesterolaemia (HeFH)"
#> $ dose_range : chr "Weight-banded SC alirocumab: <50 kg 30 or 40 mg Q2W, 75 or 150 mg Q4W (up-titration to 75 mg Q2W); >=50 kg 50 o"| __truncated__
#> $ co_medication : chr "More than 90% received statins; 19 participants received a high-dose statin (Discussion, Results)"
#> $ studies : chr "Phase II NCT02890992 (DFI14223, 8-week dose finding, 42 treated) and phase III NCT03510884 (EFC14643, 99 treate"| __truncated__
#> $ notes : chr "121 patients received active treatment across the two studies; one had no PK observations and one had only one "| __truncated__
str(mod_adult()$population)
#> List of 7
#> $ species : chr "human"
#> $ n_studies : int 2
#> $ disease_state: chr "Healthy adults"
#> $ dose_range : chr "Single IV and single SC doses of alirocumab with dense PK sampling"
#> $ studies : chr "Phase I NCT01026597 (R727-CL-0902) and NCT01074372 (R727-CL-0904)"
#> $ co_medication: chr "No high-dose statins (healthy volunteers), so no statin covariate was tested"
#> $ notes : chr "Kovalenko 2025 reports the adult parameter estimates (Table 2, 'Adults' column) but not the adult subject count"| __truncated__Source trace
All values are from Table 2 of Kovalenko 2025 unless stated;
kon and koff come from Supplemental Table 11
(they are fixed in every model of the paper). Variances in
ini() are the squares of the reported SD of the
log-parameter.
| Parameter | Pediatric | Adult | Source location |
|---|---|---|---|
lktr (ktr, 1/day) |
22.5 (fixed) | 22.5 | Table 2 ‘ktr’ |
lka (ka, 1/day) |
0.333 (fixed) | 0.333 | Table 2 ‘ka’ |
lfdepot (F) |
0.655 (fixed) | 0.655 | Table 2 ‘F’ |
lk12 (kcp, 1/day) |
0.194 (fixed) | 0.194 | Table 2 ‘kcp’ |
lk21 (kpc, 1/day) |
0.280 (fixed) | 0.280 | Table 2 ‘kpc’ |
lvc (Vc, L) |
1.64 at 50 kg | 3.01 at 75 kg | Table 2 ‘Vc (L)’ and footnote |
lkel (keL, 1/day) |
0.0800 | 0.0698 | Table 2 ‘keL’ |
lkint (keLR, 1/day) |
0.372 | 0.320 | Table 2 ‘keLR’ |
lksyn (ksyn, nM/day) |
5.76 (HDS = No) | 4.97 | Table 2 ‘ksyn’ |
lr0 (R0, nM) |
2.43 | 2.28 | Table 2 ‘R0’ |
lkon (kon, 1/nM/day) |
57.9 (fixed) | 57.9 (fixed) | Table S11 ‘kon’; Methods (Biacore) |
lkoff (koff, 1/day) |
106 (fixed) | 106 (fixed) | Table S11 ‘koff’; Methods (Biacore) |
e_wt_vc |
0.724 | 0.787 | Table 2 ‘Vc ~ weight’ |
e_conmed_statin_hi_ksyn |
0.576 | – | Table 2 ‘ksyn ~ HDS’ |
etalvc (SD) |
0.108 | 0.241 | Table 2 ‘SD(ln[Vc])’ |
etalkint (SD) |
0.356 | 0.271 | Table 2 ‘SD(ln[keLR])’ |
etalksyn (SD) |
0.616 | 0.354 | Table 2 ‘SD(ln[ksyn])’ |
etalr0 (SD) |
0.277 | 0.214 | Table 2 ‘SD(ln[R0])’ |
etalkel (SD) |
0.164 | 0.242 | Table 2 ‘SD(ln[keL])’ |
etalka (SD) |
– | 0.578 | Table 2 ‘SD(ln[ka])’ |
etalktr (SD) |
– | 1.01 | Table 2 ‘SD(ln[ktr])’ |
propSd |
0.247 | 0.147 | Table 2 ‘SD(alirocumab)’; Table S1 ‘b2’ (proportional) |
propSd_Ctarget |
0.414 | 0.249 | Table 2 ‘SD(PCSK9)’; Table S1 ‘b5’ (proportional) |
| Structure | n/a | n/a | Figure 1; Methods ‘Population PK Analysis’ |
kdeg = ksyn / r0 |
n/a | n/a | Methods (‘keR = ksyn/R0’) |
| Covariate forms | n/a | n/a | Methods (log-linear power for weight, exp(theta * HDS)
for HDS) |
Checks against the paper’s derived parameters
Table 2 reports derived linear-PK parameters. They follow from the
model’s rate constants (CL = keL * Vc,
Q = kcp * Vc, Vp = Vc * kcp / kpc, and the
terminal half-life of the linear two-compartment system), and
ksyn with HDS is 5.76 * exp(0.576). These are
closed-form checks on the packaged values.
derived <- function(ui) {
th <- ui$theta
vc <- exp(th[["lvc"]])
kel <- exp(th[["lkel"]])
k12 <- exp(th[["lk12"]])
k21 <- exp(th[["lk21"]])
s <- kel + k12 + k21
beta <- (s - sqrt(s^2 - 4 * kel * k21)) / 2
c(cl = kel * vc, q = k12 * vc, vp = vc * k12 / k21, thalf_beta = log(2) / beta)
}
der <- rbind(Pediatric = derived(ui_ped), Adult = derived(ui_adult))
published_der <- rbind(
Pediatric = c(cl = 0.131, q = 0.318, vp = 1.14, thalf_beta = 15.8),
Adult = c(cl = 0.210, q = 0.584, vp = 2.09, thalf_beta = 17.9)
)
ksyn_hds <- exp(ui_ped$theta[["lksyn"]] + ui_ped$theta[["e_conmed_statin_hi_ksyn"]])
data.frame(
model = rep(rownames(der), each = 4),
parameter = rep(colnames(der), 2),
packaged = signif(c(t(der)), 3),
published = c(t(published_der))
) |>
dplyr::rename(
"Model" = model,
"Derived parameter" = parameter,
"From packaged values" = packaged,
"Table 2" = published
) |>
knitr::kable(caption = "Derived parameters (CL, Q in L/day; Vp in L; t1/2 beta in days).")| Model | Derived parameter | From packaged values | Table 2 |
|---|---|---|---|
| Pediatric | cl | 0.131 | 0.131 |
| Pediatric | q | 0.318 | 0.318 |
| Pediatric | vp | 1.140 | 1.140 |
| Pediatric | thalf_beta | 15.800 | 15.800 |
| Adult | cl | 0.210 | 0.210 |
| Adult | q | 0.584 | 0.584 |
| Adult | vp | 2.090 | 2.090 |
| Adult | thalf_beta | 17.900 | 17.900 |
# Table 2 prints three significant figures, so the closed forms agree within
# rounding.
stopifnot(
all(abs(der / published_der - 1) < 0.01),
abs(ksyn_hds / 10.3 - 1) < 0.01
)The packaged pediatric ksyn on HDS is 10.25 nM/day
(Table 2: 10.3 nM/day, 79% above the no-HDS value).
Baseline steady state
Free PCSK9 starts at R0 in the central compartment and
at the matching distribution equilibrium in the peripheral compartment,
and keR = ksyn / R0. Without drug, free PCSK9 must
therefore stay at R0 for any covariates.
ev_nodrug <- rxode2::et(seq(0, 168, by = 7), cmt = "central") |>
as.data.frame() |>
dplyr::mutate(dvid = 1L, WT = 50, CONMED_STATIN_HI = 1)
ss_ped <- rxode2::rxSolve(rxode2::zeroRe(mod_ped), ev_nodrug, returnType = "data.frame")
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalvc', 'etalkint', 'etalksyn', 'etalr0', 'etalkel'
r0_ped <- exp(ui_ped$theta[["lr0"]])
stopifnot(max(abs(ss_ped$Ctarget / r0_ped - 1)) < 1e-4, max(abs(ss_ped$Cc)) < 1e-8)
range(ss_ped$Ctarget)
#> [1] 2.43 2.43Virtual pediatric cohort
The cohort reproduces the four phase III regimens (Table 1, NCT03510884): below 50 kg 150 mg Q4W or 75 mg Q2W, at or above 50 kg 300 mg Q4W or 150 mg Q2W, each for 24 weeks without up-titration. Weight is drawn from a log-normal distribution with the reported mean (52.0 kg) and SD (18.8 kg), truncated to 25-110 kg, and split at 50 kg; 16% (19/119) of subjects take a high-dose statin.
set.seed(20250501)
n_per_arm <- 100L
draw_weights <- function(n, band) {
sdlog <- sqrt(log(1 + (18.8 / 52.0)^2))
meanlog <- log(52.0) - sdlog^2 / 2
w <- numeric(0)
while (length(w) < n) {
x <- rlnorm(4 * n, meanlog, sdlog)
x <- x[x >= 25 & x <= 110]
x <- if (band == "<50 kg") x[x < 50] else x[x >= 50]
w <- c(w, x)
}
w[seq_len(n)]
}
arms <- tibble::tribble(
~arm, ~band, ~amt, ~ii,
"<50 kg: 150 mg Q4W", "<50 kg", 150, 28,
"<50 kg: 75 mg Q2W", "<50 kg", 75, 14,
">=50 kg: 300 mg Q4W", ">=50 kg", 300, 28,
">=50 kg: 150 mg Q2W", ">=50 kg", 150, 14
)
obs_times <- sort(unique(c(seq(0, 168, by = 1), seq(140, 168, by = 0.25))))
make_arm <- function(i, id_offset) {
a <- arms[i, ]
subj <- tibble::tibble(
id = id_offset + seq_len(n_per_arm),
WT = draw_weights(n_per_arm, a$band),
CONMED_STATIN_HI = rbinom(n_per_arm, 1, 19 / 119),
arm = a$arm
)
dose_times <- seq(0, 168 - a$ii, by = a$ii)
doses <- tidyr::expand_grid(subj, time = dose_times) |>
dplyr::mutate(evid = 1L, amt = a$amt, cmt = "depot")
# Two declared endpoints (Cc, Ctarget): observation rows carry dvid = 1 on
# the central compartment, and every row returns both observables.
obs <- tidyr::expand_grid(subj, time = obs_times) |>
dplyr::mutate(evid = 0L, amt = 0, cmt = "central", dvid = 1L)
dplyr::bind_rows(doses, obs)
}
events <- dplyr::bind_rows(lapply(seq_len(nrow(arms)), function(i) make_arm(i, (i - 1L) * n_per_arm))) |>
dplyr::arrange(id, time, dplyr::desc(evid))
stopifnot(!anyDuplicated(unique(events[, c("id", "time", "evid")])))
events |>
dplyr::distinct(id, arm, WT, CONMED_STATIN_HI) |>
dplyr::group_by(arm) |>
dplyr::summarise(
n = dplyr::n(),
median_wt = median(WT),
hds_pct = 100 * mean(CONMED_STATIN_HI),
.groups = "drop"
) |>
dplyr::rename("Arm" = arm, "N" = n, "Median weight (kg)" = median_wt, "HDS (%)" = hds_pct) |>
knitr::kable(digits = 1)| Arm | N | Median weight (kg) | HDS (%) |
|---|---|---|---|
| <50 kg: 150 mg Q4W | 100 | 38.8 | 19 |
| <50 kg: 75 mg Q2W | 100 | 38.9 | 20 |
| >=50 kg: 150 mg Q2W | 100 | 61.9 | 16 |
| >=50 kg: 300 mg Q4W | 100 | 61.2 | 17 |
Simulation
rxode2::rxSetSeed(20250501)
sim <- rxode2::rxSolve(mod_ped, events = events, keep = "arm", returnType = "data.frame") |>
dplyr::mutate(arm = factor(arm, levels = arms$arm))
#> ℹ parameter labels from comments will be replaced by 'label()'Total alirocumab (Figure 2A)
sim |>
dplyr::group_by(arm, time) |>
dplyr::summarise(
Q05 = quantile(Cc, 0.05), Q50 = median(Cc), Q95 = quantile(Cc, 0.95),
.groups = "drop"
) |>
ggplot(aes(time, Q50)) +
geom_ribbon(aes(ymin = Q05, ymax = Q95), alpha = 0.25, fill = "steelblue") +
geom_line(colour = "steelblue4") +
facet_wrap(~arm) +
labs(
x = "Time (day)", y = "Total alirocumab (mg/L)",
title = "Simulated total alirocumab, phase III regimens",
caption = "Median and 90% prediction interval; compare with Figure 2A of Kovalenko 2025."
)
Figure 2A of the paper pools all phase II and III sequences; its observed and predicted medians lie at roughly 10-25 mg/L during the first 24 weeks, the same range as the simulated regimens above.
Free PCSK9 (Figure 3A)
# The paper pairs R0 = 2.43 nM with a baseline free PCSK9 of 0.176 mg/L
# (Results), which is used here only to label a mg/L axis for comparison with
# Figure 3A.
mgL_per_nM <- 0.176 / 2.43
sim |>
dplyr::group_by(arm, time) |>
dplyr::summarise(
Q05 = quantile(Ctarget, 0.05), Q50 = median(Ctarget), Q95 = quantile(Ctarget, 0.95),
.groups = "drop"
) |>
ggplot(aes(time, Q50)) +
geom_ribbon(aes(ymin = Q05, ymax = Q95), alpha = 0.25, fill = "darkorange") +
geom_line(colour = "darkorange4") +
facet_wrap(~arm) +
scale_y_continuous(sec.axis = sec_axis(~ . * mgL_per_nM, name = "Free PCSK9 (mg/L)")) +
labs(
x = "Time (day)", y = "Free PCSK9 (nM)",
title = "Simulated free PCSK9, phase III regimens",
caption = "Median and 90% prediction interval; compare with Figure 3A of Kovalenko 2025."
)
As in Figure 3A, free PCSK9 falls from about 0.2 mg/L at baseline to near the assay floor within days of dosing and recovers partially at the end of each Q4W interval.
Free PCSK9 suppression at trough
The paper reports that “free trough PCSK9 levels were suppressed by more than 90% in at least 1 treatment regimen among 2 treatments available for each pediatric weight cohort”. Using the typical-value model at a representative weight in each band (40 kg and 65 kg, no HDS) and the trough at week 24:
typical_trough <- function(wt, amt, ii) {
n_dose <- 168 / ii
ev <- rxode2::et(amt = amt, cmt = "depot", ii = ii, addl = n_dose - 1) |>
rxode2::et(c(0, 168), cmt = "central") |>
as.data.frame() |>
dplyr::mutate(dvid = ifelse(evid == 0, 1L, NA_integer_), WT = wt, CONMED_STATIN_HI = 0)
s <- rxode2::rxSolve(rxode2::zeroRe(mod_ped), ev, returnType = "data.frame")
100 * (1 - s$Ctarget[s$time == 168] / s$Ctarget[s$time == 0])
}
trough <- arms |>
dplyr::mutate(
wt = ifelse(band == "<50 kg", 40, 65),
suppression = mapply(typical_trough, wt, amt, ii)
)
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalvc', 'etalkint', 'etalksyn', 'etalr0', 'etalkel'
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalvc', 'etalkint', 'etalksyn', 'etalr0', 'etalkel'
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalvc', 'etalkint', 'etalksyn', 'etalr0', 'etalkel'
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalvc', 'etalkint', 'etalksyn', 'etalr0', 'etalkel'
trough |>
dplyr::select(arm, wt, suppression) |>
dplyr::rename("Arm" = arm, "Weight (kg)" = wt, "Free PCSK9 suppression at week-24 trough (%)" = suppression) |>
knitr::kable(digits = 1)| Arm | Weight (kg) | Free PCSK9 suppression at week-24 trough (%) |
|---|---|---|
| <50 kg: 150 mg Q4W | 40 | 86.5 |
| <50 kg: 75 mg Q2W | 40 | 91.5 |
| >=50 kg: 300 mg Q4W | 65 | 91.3 |
| >=50 kg: 150 mg Q2W | 65 | 94.3 |
In each weight band the Q2W regimen keeps the typical trough suppression above 90%, matching the paper’s statement; at the end of a Q4W interval the suppression is 86.5% (<50 kg, 150 mg) and 91.3% (>=50 kg, 300 mg).
PKNCA: steady-state interval (week 20-24)
The paper reports no NCA, so this section summarises the simulated exposure over the last dosing interval for reference, by arm. No published values are available for a side-by-side comparison.
last_dose <- events |>
dplyr::filter(evid == 1) |>
dplyr::group_by(id) |>
dplyr::slice_max(time, n = 1) |>
dplyr::ungroup() |>
dplyr::select(id, time, amt, arm)
tau <- arms$ii[match(last_dose$arm, arms$arm)]
conc_ped <- sim |>
dplyr::filter(!is.na(Cc)) |>
dplyr::select(id, time, Cc, arm) |>
dplyr::mutate(arm = as.character(arm))
conc_obj <- PKNCA::PKNCAconc(conc_ped, Cc ~ time | arm + id)
dose_obj <- PKNCA::PKNCAdose(
events |> dplyr::filter(evid == 1) |> dplyr::select(id, time, amt, arm),
amt ~ time | arm + id
)
intervals_ped <- last_dose |>
dplyr::mutate(start = time, end = time + tau, cmax = TRUE, cmin = TRUE, auclast = TRUE, cav = TRUE) |>
dplyr::select(arm, id, start, end, cmax, cmin, auclast, cav)
nca_ped <- PKNCA::pk.nca(PKNCA::PKNCAdata(conc_obj, dose_obj, intervals = intervals_ped))
nca_ped_summary <- as.data.frame(nca_ped$result) |>
dplyr::group_by(arm, PPTESTCD) |>
dplyr::summarise(median = median(PPORRES, na.rm = TRUE), .groups = "drop") |>
tidyr::pivot_wider(names_from = PPTESTCD, values_from = median) |>
dplyr::mutate(arm = factor(arm, levels = arms$arm)) |>
dplyr::arrange(arm)
nca_ped_summary |>
dplyr::select(arm, cmax, cmin, cav, auclast) |>
dplyr::rename(
"Arm" = arm,
"Cmax (mg/L)" = cmax,
"Cmin (mg/L)" = cmin,
"Cav (mg/L)" = cav,
"AUCtau (mg*day/L)" = auclast
) |>
knitr::kable(digits = 2, caption = "Median steady-state NCA of total alirocumab over the last dosing interval.")| Arm | Cmax (mg/L) | Cmin (mg/L) | Cav (mg/L) | AUCtau (mg*day/L) |
|---|---|---|---|---|
| <50 kg: 150 mg Q4W | 40.57 | 9.37 | 24.15 | 676.14 |
| <50 kg: 75 mg Q2W | 28.22 | 16.46 | 23.08 | 323.17 |
| >=50 kg: 300 mg Q4W | 57.18 | 14.93 | 35.48 | 993.33 |
| >=50 kg: 150 mg Q2W | 43.19 | 25.47 | 35.28 | 493.87 |
# Within a weight band the Q2W and Q4W arms deliver the same mg per 4 weeks.
# At these concentrations total alirocumab clearance is dominated by the linear
# keL pathway, so the median average concentration over the interval must agree
# between the two schedules (about 1-5% apart when rendered; the 15% bound
# leaves room for the sampling error of a 100-subject median).
cav <- setNames(nca_ped_summary$cav, as.character(nca_ped_summary$arm))
stopifnot(
abs(log(cav[["<50 kg: 75 mg Q2W"]] / cav[["<50 kg: 150 mg Q4W"]])) < log(1.15),
abs(log(cav[[">=50 kg: 150 mg Q2W"]] / cav[[">=50 kg: 300 mg Q4W"]])) < log(1.15)
)Adult model
The adult model is shown for single 150 mg and 300 mg SC doses at 75 kg and, to exercise the IV path, an illustrative 300 mg IV dose given over one hour. (The adult studies’ actual dose levels are not given in Kovalenko 2025.)
adult_arms <- tibble::tribble(
~arm, ~amt, ~cmt, ~dur,
"150 mg SC", 150, "depot", 0,
"300 mg SC", 300, "depot", 0,
"300 mg IV (1 h)", 300, "central", 1 / 24
)
adult_times <- sort(unique(c(seq(0, 2, by = 1 / 24), seq(2, 84, by = 0.5))))
make_adult <- function(i, n, id_offset) {
a <- adult_arms[i, ]
subj <- tibble::tibble(id = id_offset + seq_len(n), WT = 75, arm = a$arm)
dose <- subj |> dplyr::mutate(time = 0, evid = 1L, amt = a$amt, cmt = a$cmt, dur = a$dur)
obs <- tidyr::expand_grid(subj, time = adult_times) |>
dplyr::mutate(evid = 0L, amt = 0, cmt = "central", dvid = 1L, dur = 0)
dplyr::bind_rows(dose, obs)
}
events_adult <- dplyr::bind_rows(lapply(seq_len(nrow(adult_arms)), function(i) make_adult(i, 50L, (i - 1L) * 50L))) |>
dplyr::arrange(id, time, dplyr::desc(evid))
stopifnot(!anyDuplicated(unique(events_adult[, c("id", "time", "evid")])))
sim_adult <- rxode2::rxSolve(mod_adult, events = events_adult, keep = "arm", returnType = "data.frame")
#> ℹ parameter labels from comments will be replaced by 'label()'
sim_adult |>
dplyr::select(id, time, arm, Cc, Ctarget) |>
tidyr::pivot_longer(c(Cc, Ctarget), names_to = "output") |>
dplyr::mutate(output = dplyr::recode(output, Cc = "Total alirocumab (mg/L)", Ctarget = "Free PCSK9 (nM)")) |>
dplyr::group_by(arm, output, time) |>
dplyr::summarise(Q05 = quantile(value, 0.05), Q50 = median(value), Q95 = quantile(value, 0.95), .groups = "drop") |>
ggplot(aes(time, Q50, colour = arm, fill = arm)) +
geom_ribbon(aes(ymin = Q05, ymax = Q95), alpha = 0.15, colour = NA) +
geom_line() +
facet_wrap(~output, scales = "free_y") +
labs(x = "Time (day)", y = NULL, colour = NULL, fill = NULL,
title = "Adult model, single doses at 75 kg",
caption = "Median and 90% prediction interval (50 subjects per arm).")
conc_adult <- sim_adult |>
dplyr::filter(!is.na(Cc)) |>
dplyr::select(id, time, Cc, arm)
conc_adult <- dplyr::bind_rows(
conc_adult,
conc_adult |> dplyr::distinct(id, arm) |> dplyr::mutate(time = 0, Cc = 0)
) |>
dplyr::distinct(id, arm, time, .keep_all = TRUE) |>
dplyr::arrange(id, arm, time)
nca_adult <- PKNCA::pk.nca(PKNCA::PKNCAdata(
PKNCA::PKNCAconc(conc_adult, Cc ~ time | arm + id),
PKNCA::PKNCAdose(events_adult |> dplyr::filter(evid == 1) |> dplyr::select(id, time, amt, arm), amt ~ time | arm + id),
intervals = data.frame(start = 0, end = 84, cmax = TRUE, tmax = TRUE, auclast = TRUE)
))
as.data.frame(nca_adult$result) |>
dplyr::group_by(arm, PPTESTCD) |>
dplyr::summarise(median = median(PPORRES, na.rm = TRUE), .groups = "drop") |>
tidyr::pivot_wider(names_from = PPTESTCD, values_from = median) |>
dplyr::select(arm, cmax, tmax, auclast) |>
dplyr::rename("Arm" = arm, "Cmax (mg/L)" = cmax, "Tmax (day)" = tmax, "AUC0-84 day (mg*day/L)" = auclast) |>
knitr::kable(digits = 2, caption = "Median single-dose NCA of total alirocumab (adult model).")| Arm | Cmax (mg/L) | Tmax (day) | AUC0-84 day (mg*day/L) |
|---|---|---|---|
| 150 mg SC | 14.78 | 4.00 | 257.19 |
| 300 mg IV (1 h) | 95.91 | 0.04 | 1046.37 |
| 300 mg SC | 32.81 | 4.00 | 710.62 |
Assumptions and deviations
-
Alirocumab molecular weight. The paper estimates
its parameters in nM and reports pediatric concentrations in mg/L
without stating the mass conversion. Both models use 146,000 g/mol, the
approximate molecular weight on the Praluent label (the same value as
Nolain_2022_alirocumab), to turn mg doses into nmol and total alirocumab back into mg/L. Free PCSK9 is returned in nM, the unit ofR0andksyn. -
Binding stoichiometry. Figure 1 labels both binding
steps (
L + R <-> L:RandL:R + R <-> L:R:R) with the samekonandkoffand the paper prints no equations, so both steps usekonandkoffwithout statistical (valency) factors. The 1:2 complex is eliminated atkeLR, as drawn in Figure 1. -
Distribution of every species. Figure 1 draws free
alirocumab, free PCSK9 and both complexes moving to a peripheral
compartment with the same
kcpandkpc, and binding only in the central compartment. The peripheral states are carried as amounts on the central-volume scale, which reproduces the paper’s derivedVp = Vc * kcp / kpc(Table 2). -
One transit compartment. Figure 1 shows a single
“Depot Transit -> Transit” chain (
ktrthenka); the text mentions that changing the number of transit compartments changed onlyktr. -
High-dose statin definition. The paper does not
define the dose threshold for HDS; the covariate is mapped to
CONMED_STATIN_HIwith that gap recorded in the model’s covariate notes. - Residual error. The Methods describe exploring a combined additive and proportional error, but the final tables report one proportional term per analyte (Table S1 ‘b2’, ‘b5’), which is what is encoded. The paper handled BLQ data with the M4 method; this affects estimation only.
-
Adult population. Kovalenko 2025 does not report
the adult subject count, demographics or dose levels, so the adult
populationmetadata carries only what the paper states; the adult IV dose in this article is illustrative. - Virtual cohort. Weight follows a log-normal distribution with the published mean and SD; the HDS fraction is 19/119. Phase II regimens and the week-12 up-titration of phase III are not simulated.
- No erratum or correction notice was found for this article (EuropePMC search, 2026-11-04).