Skip to contents

Imetelstat (Gonzalez-Sales 2024)

Source: vignettes/articles/GonzalezSales_2024_imetelstat.Rmd
GonzalezSales_2024_imetelstat.Rmd
library(nlmixr2lib)
library(PKNCA)
#> 
#> Attaching package: 'PKNCA'
#> The following object is masked from 'package:stats':
#> 
#>     filter
library(rxode2)
#> rxode2 5.1.8 using 2 threads (see ?getRxThreads)
#>   no cache: create with `rxCreateCache()`
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(tidyr)
library(ggplot2)

Model and source

Gonzalez-Sales et al. (2024) developed a three-compartment population PK model for imetelstat (GRN163L), a first-in-class oligonucleotide telomerase inhibitor in development for hematologic malignancies (lower-risk myelodysplastic syndromes, myelofibrosis). The data set pooled 4375 plasma concentrations from 424 adults across seven clinical studies. The structural disposition is the saturable-binding / distribution model of Snoeck 1999 / Peletier 2017: free drug in central reversibly binds target sites with capacity Bmax (rates Kon, Koff), the bound complex internalises (Kint) into a deep peripheral pool, and free drug returns from that pool to central (Kback). Linear clearance (CL) from central is the only true sink. Theory-based allometric exponents (1, 0.75, -0.25) are fixed on Vc, CL, and Kback respectively; final-model covariates include sex, dose, time, and myelofibrosis / multiple myeloma malignancy on CL and Vc, plus myelofibrosis malignancy and baseline spleen volume on Bmax.

  • Citation: Gonzalez-Sales M, Lennox AL, Huang F, Pamulapati C, Wan Y, Sun L, Berry T, Kelly Behrs M, Feller F, Morcos PN. (2024). Population pharmacokinetics of imetelstat, a first-in-class oligonucleotide telomerase inhibitor. CPT Pharmacometrics Syst Pharmacol 13(7):1264-1277. doi:10.1002/psp4.13160.
  • Article: https://doi.org/10.1002/psp4.13160
  • Open-access PMC mirror: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11247122/
  • Supplementary information (NONMEM control stream PSP4-13-1264-s001.txt and appendix PSP4-13-1264-s002.docx) is available from the journal landing page and PMC.

Population

The pooled analysis dataset comprised 424 adult patients from seven Phase I-III clinical studies (Gonzalez-Sales 2024 Table 1). Mean baseline body weight was 77.2 kg (range 44.0-161 kg); 247 of 424 patients (58.3%) were male. The contributing studies (Methods, ‘Clinical studies’) were CP04-151 (chronic lymphoproliferative disease, phase I), CP05-101 (refractory solid tumors, phase I), CP14A004 (refractory multiple myeloma, phase I), CP14B013 (multiple myeloma + lenalidomide maintenance, phase II), CP14B015 (essential thrombocythemia / polycythemia vera, phase II), MYF2001 (intermediate-2 / high-risk myelofibrosis refractory to JAK inhibitors, phase II), and MDS3001 (transfusion-dependent low- or intermediate-1-risk MDS refractory to erythropoiesis-stimulating agents, phase II/III). Investigated dose levels spanned 0.4-11.7 mg/kg administered as a 2-hour IV infusion (and a 6-hour IV infusion in CP04-151) on schedules ranging from once weekly to once every 4 weeks. The registered MDS3001 dose was 7.5 mg/kg once every 4 weeks; the registered MYF3001 myelofibrosis dose is 9.4 mg/kg once every 3 weeks.

Baseline spleen volume was available only from patients with myelofibrosis (Study MYF2001) per Results, ‘Patient characteristics at baseline’; the MF-cohort median spleen volume of 3010 cm^3 (Figure 4 caption) is the covariate-effect anchor for the SPLV power term on Bmax. The 374 of 4375 plasma observations (8.55%) below the limit of quantification were handled by the Beal 2001 M3 method.

The same information is available programmatically via the model’s population metadata:

str(rxode2::rxode(readModelDb("GonzalezSales_2024_imetelstat"))$population)
#> ℹ parameter labels from comments will be replaced by 'label()'
#> List of 13
#>  $ species       : chr "human"
#>  $ n_subjects    : int 424
#>  $ n_studies     : int 7
#>  $ age_range     : chr "adults; full age range not enumerated in the main paper (no age effect on PK identified)"
#>  $ age_median    : chr "not reported in main-paper Table S1 quotation"
#>  $ weight_range  : chr "44.0-161 kg"
#>  $ weight_median : chr "77.2 kg (mean)"
#>  $ sex_female_pct: num 41.7
#>  $ race_ethnicity: chr "not enumerated in main paper text (race / ethnicity available in supplement Table S1; no race effect on PK identified)"
#>  $ disease_state : chr "Pooled adults with hematologic malignancies (myelofibrosis, lower-risk MDS, multiple myeloma, essential thrombo"| __truncated__
#>  $ dose_range    : chr "0.4-11.7 mg/kg of imetelstat sodium (MW 4896 g/mol) administered as a 2-hour or 6-hour IV infusion (per-study s"| __truncated__
#>  $ regions       : chr "Multinational (specific regions not enumerated in the main paper text)."
#>  $ notes         : chr "Baseline-covariate summary statistics quoted from Gonzalez-Sales 2024 main-paper Results (Table S1 referenced b"| __truncated__

Source trace

The per-parameter origin is recorded as an in-file comment next to each ini() entry in inst/modeldb/specificDrugs/GonzalezSales_2024_imetelstat.R. The table below collects them in one place for review. All structural and covariate values come from Gonzalez-Sales 2024 Table 2 (final population PK parameter estimates); the functional forms come from Methods (Equations 1, 4-6) and the NONMEM supplement control stream (PSP4-13-1264-s001.txt).

Equation / parameter Value Source location
CL (clearance) 1.00 L/h/70 kg Table 2, row ‘CL’
Vc (central volume) 4.08 L/70 kg Table 2, row ‘Vc’
Kback 0.0253 1/h/70 kg Table 2, row ‘Kback’
Bmax 15.0 umol/L Table 2, row ‘Bmax’
Kint 0.103 Table 2, row ‘Kint’ (units ‘L/h/70 kg’ per Table 2; interpreted as 1/h per ODE)
Kon 0.159 Table 2, row ‘Kon’ (units ’L^2/(uMh)’ per Table 2; interpreted as L/(umolh))
Koff 0.609 Table 2, row ‘Koff’ (units ‘L/h’ per Table 2; interpreted as 1/h)
Allometric exp on Vc / CL / Kback 1 / 0.75 / -0.25 Methods Equation 1; supplement S1 lines 70-74
Effect of dose on CL -0.401 (power) Table 2, row ‘Effect of dose on CL’
Effect of MF on CL 0.511 (exp) Table 2, row ‘Effect of MF on CL’
Effect of sex on CL -0.299 (exp) Table 2, row ‘Effect of sex on CL’
Effect of time on CL (t50) 5880 h Table 2, row ‘Effect of time on CL’; supplement S1 lines 107-108
Effect of MM on Vc -0.233 (exp) Table 2, row ‘Effect of MM malignancy on Vc’
Effect of sex on Vc -0.122 (exp) Table 2, row ‘Effect of sex on Vc’
Effect of spleen on Bmax (MF) 0.772 (power) Table 2, row ‘Effect of spleen volume on Bmax’; supplement S1 lines 76-80
Effect of MF on Bmax 1.44 (exp) Table 2, row ‘Effect of MF malignancy on Bmax’
IIV CL / Vc (BLOCK 2x2) var 0.189 / 0.0655 Table 2 ‘IIV on CL’ (43.7% CV) / ‘IIV on Vc’ (25.7% CV); supplement S1 $OMEGA BLOCK
Correlation r(CL,Vc) 0.545 (cov 0.0602) Table 2, row ‘Correlation between ETA on CL and Vc’
IIV Bmax (diagonal) var 0.181 Table 2 ‘IIV on Bmax’ (45.8% CV); supplement S1 $OMEGA line 196 | | Proportional residual error | 0.218 | Table 2 'Residual variability' (21.8% CV) | | ODE system | n/a (structure) | Supplement S1 `$DESblock (lines 133-138) | | Time-on-CL functional form | T50 / (T50 + t) | Supplement S1 lines 107-108, 122 (TIME_CL = TIME/(TIME+T50)) | | Spleen-on-Bmax MF gating | gated to MF subjects | Supplement S1 lines 76-80 (IF(STDY.EQ.6) … SPLEFF_BMAX = (SPLV0/3010)^THETA(9)`)

Virtual cohort

Original observed data are not publicly available. We simulate a virtual LR-MDS cohort (DIS_MF = 0, DIS_MM = 0) receiving the registered MDS3001 dose of 7.5 mg/kg IV over 2 hours every 4 weeks, and a virtual myelofibrosis cohort (DIS_MF = 1, DIS_MM = 0) receiving the registered MYF3001 dose of 9.4 mg/kg IV over 2 hours every 3 weeks. Body weight is drawn from a normal distribution centred at the pooled-cohort mean 77.2 kg with standard deviation 15 kg (approximating the published range 44-161 kg); sex is drawn binomially with 41.7% female. Spleen volume is set to the MF-cohort median 3010 cm^3 for MF subjects and to NA (gated out) for LR-MDS subjects.

set.seed(20240518)

mw_imetelstat <- 4896  # g/mol of imetelstat sodium per Methods

# Helper: build one cohort's event table. `id_offset` shifts IDs to keep
# multi-cohort bind_rows() output disjoint.
make_imetelstat_cohort <- function(n,
                                   dose_mg_per_kg,
                                   tau_h,
                                   n_doses,
                                   is_mf,
                                   id_offset = 0L) {
  ids <- id_offset + seq_len(n)
  wt  <- pmax(40, pmin(165, rnorm(n, mean = 77.2, sd = 15)))
  sexf <- rbinom(n, size = 1, prob = 0.417)

  # Dose in umol of imetelstat sodium: amt_umol = (mg/kg * kg) / (g/mol) * 1000
  # = (mg/kg * kg * 1000) / 4896
  amt_umol <- (dose_mg_per_kg * wt * 1000) / mw_imetelstat

  # Per-subject covariates (time-invariant)
  cov_df <- tibble(
    id     = ids,
    WT     = wt,
    SEXF   = sexf,
    DIS_MF = as.integer(is_mf),
    DIS_MM = 0L,
    SPLV   = if (is_mf) 3010 else 3010,  # SPLV is gated by DIS_MF in the model
    DOSE   = amt_umol,
    amt_umol = amt_umol
  )

  # Dose events: 2-hour IV infusion at times 0, tau, 2*tau, ...
  dose_times <- (seq_len(n_doses) - 1) * tau_h
  dose_df <- cov_df |>
    tidyr::expand_grid(time = dose_times) |>
    dplyr::mutate(
      evid = 1L,
      amt  = amt_umol,
      cmt  = "central",
      rate = amt_umol / 2  # infusion rate over 2 h
    )

  # Observation grid: dense within cycle 1 to capture the rapid distribution
  # phase (~4.9 h apparent t1/2), and weekly between cycles to capture trough.
  obs_within_cycle1 <- c(0, 0.5, 1, 2, 3, 4, 6, 8, 12, 16, 24, 48, 72, 168)
  cycle_starts <- (seq_len(n_doses) - 1) * tau_h
  obs_times <- sort(unique(c(
    rep(cycle_starts, each = length(obs_within_cycle1)) +
      rep(obs_within_cycle1, n_doses),
    seq(0, n_doses * tau_h, by = 24)
  )))

  obs_df <- cov_df |>
    tidyr::expand_grid(time = obs_times) |>
    dplyr::mutate(
      evid = 0L,
      amt  = 0,
      cmt  = "central",
      rate = 0
    )

  bind_rows(dose_df, obs_df) |>
    dplyr::arrange(id, time, dplyr::desc(evid)) |>
    dplyr::select(id, time, evid, amt, cmt, rate,
                  WT, SEXF, DIS_MF, DIS_MM, SPLV, DOSE)
}

lrmds_events <- make_imetelstat_cohort(
  n              = 100,
  dose_mg_per_kg = 7.5,
  tau_h          = 24 * 28,  # Q4W = 672 h
  n_doses        = 3,
  is_mf          = FALSE,
  id_offset      = 0L
) |> dplyr::mutate(treatment = "LR-MDS 7.5 mg/kg Q4W")

mf_events <- make_imetelstat_cohort(
  n              = 100,
  dose_mg_per_kg = 9.4,
  tau_h          = 24 * 21,  # Q3W = 504 h
  n_doses        = 4,
  is_mf          = TRUE,
  id_offset      = 1000L
) |> dplyr::mutate(treatment = "MF 9.4 mg/kg Q3W")

events <- dplyr::bind_rows(lrmds_events, mf_events)
stopifnot(!anyDuplicated(unique(events[, c("id", "time", "evid")])))

Simulation 

mod <- readModelDb("GonzalezSales_2024_imetelstat")

# Stochastic VPC-style simulation (between-subject + residual variability)
sim <- rxode2::rxSolve(mod, events = events, keep = c("treatment"))
#> ℹ parameter labels from comments will be replaced by 'label()'
sim <- as.data.frame(sim)

For deterministic typical-value replication (no IIV, no residual error):

mod_typical <- mod |> rxode2::zeroRe()
#> ℹ parameter labels from comments will be replaced by 'label()'
sim_typical <- rxode2::rxSolve(mod_typical, events = events,
                               keep = c("treatment")) |> as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalvc', 'etalbmax'
#> Warning: multi-subject simulation without without 'omega'

Replicate published figures

Figure 3 – Visual predictive check (overall population)

Gonzalez-Sales 2024 Figure 3 shows a VPC of imetelstat plasma concentration versus time after dose for the overall population. The replication below shows simulated 5th / 50th / 95th percentile bands by treatment.

# Convert to ug/mL (paper Figure 3 axis) using MW = 4896 g/mol of imetelstat
# sodium: 1 umol/L = 4.896 ug/mL.
sim |>
  dplyr::filter(time > 0) |>
  dplyr::mutate(
    Cc_ugmL  = Cc * mw_imetelstat / 1000,
    time_day = time / 24
  ) |>
  dplyr::group_by(treatment, time_day) |>
  dplyr::summarise(
    Q05 = quantile(Cc_ugmL, 0.05, na.rm = TRUE),
    Q50 = quantile(Cc_ugmL, 0.50, na.rm = TRUE),
    Q95 = quantile(Cc_ugmL, 0.95, na.rm = TRUE),
    .groups = "drop"
  ) |>
  ggplot(aes(time_day, Q50)) +
  geom_ribbon(aes(ymin = pmax(Q05, 1e-3), ymax = Q95), alpha = 0.25) +
  geom_line() +
  facet_wrap(~ treatment, scales = "free_x") +
  scale_y_log10() +
  labs(
    x = "Time (days)",
    y = expression("Imetelstat plasma concentration (" * mu * "g/mL)"),
    title = "Figure 3 -- Imetelstat plasma concentration VPC",
    caption = paste(
      "Replicates Figure 3 of Gonzalez-Sales 2024 (LR-MDS) and the MYF2001 dosing arm.",
      "Ribbon: simulated 5th-95th percentiles; line: median."
    )
  ) +
  theme_minimal()

Figure 4 – Sex effect on CL

Gonzalez-Sales 2024 Figure 4 (forest plot) shows that female sex confers exp(-0.299) = 0.74-fold lower CL relative to males. Plot below compares typical-value Cmax / AUC for female vs male LR-MDS subjects at the reference dose:

# Typical-value cycle-1 trajectory for male vs female 70 kg LR-MDS subjects.
# Build via rxode2::et() then attach covariates by column assignment.
dose_umol_sex <- (7.5 * 70 * 1000) / mw_imetelstat  # 7.5 mg/kg x 70 kg

build_sex_events <- function(sexf_value, id_value) {
  ev <- rxode2::et(amt = dose_umol_sex, dur = 2, cmt = "central") |>
    rxode2::et(seq(0, 168, length.out = 80))
  ev$WT     <- 70
  ev$SEXF   <- sexf_value
  ev$DIS_MF <- 0L
  ev$DIS_MM <- 0L
  ev$SPLV   <- 3010
  ev$DOSE   <- dose_umol_sex
  ev$id     <- id_value
  ev$sex    <- ifelse(sexf_value == 1L, "Female", "Male")
  ev
}

ev_sex <- dplyr::bind_rows(
  build_sex_events(0L, 1L),
  build_sex_events(1L, 2L)
)

mod_no_re <- rxode2::zeroRe(mod)
#> ℹ parameter labels from comments will be replaced by 'label()'
sex_sim <- rxode2::rxSolve(mod_no_re, ev_sex, keep = c("sex")) |>
  as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalvc', 'etalbmax'
#> Warning: multi-subject simulation without without 'omega'

sex_sim |>
  dplyr::mutate(Cc_ugmL = Cc * mw_imetelstat / 1000) |>
  dplyr::filter(time > 0) |>
  ggplot(aes(time, Cc_ugmL, colour = sex)) +
  geom_line(linewidth = 1) +
  scale_y_log10() +
  labs(
    x = "Time after first dose (h)",
    y = expression("Imetelstat (" * mu * "g/mL)"),
    title = "Figure 4 -- Sex effect on imetelstat exposure",
    caption = paste(
      "Typical-value (no IIV) simulation; 70-kg LR-MDS subject, 7.5 mg/kg",
      "single 2-h IV infusion. Females have exp(-0.299) = 0.74-fold lower CL,",
      "yielding modestly higher exposure."
    )
  ) +
  theme_minimal()

PKNCA validation

Compute Cmax, Tmax, AUC, and apparent half-life with PKNCA on the typical-value LR-MDS cohort.

# Use the LR-MDS arm of the stochastic VPC cohort, restricted to cycle 1
# (0 to 672 h) so the result is comparable to the published cycle-1 NCA.
nca_window_h <- 672  # 28 days

sim_lrmds_cycle1 <- sim |>
  dplyr::filter(treatment == "LR-MDS 7.5 mg/kg Q4W",
                time <= nca_window_h) |>
  dplyr::mutate(Cc_ugmL = Cc * mw_imetelstat / 1000)

# Time-zero guarantee: bind a Cc = 0 row at t = 0 for every subject.
sim_lrmds_nca <- sim_lrmds_cycle1 |>
  dplyr::filter(!is.na(Cc_ugmL)) |>
  dplyr::select(id, time, Cc_ugmL, treatment)

sim_lrmds_nca <- dplyr::bind_rows(
  sim_lrmds_nca,
  sim_lrmds_nca |> dplyr::distinct(id, treatment) |>
    dplyr::mutate(time = 0, Cc_ugmL = 0)
) |>
  dplyr::distinct(id, treatment, time, .keep_all = TRUE) |>
  dplyr::arrange(id, treatment, time) |>
  dplyr::rename(Cc = Cc_ugmL)

dose_lrmds_nca <- events |>
  dplyr::filter(treatment == "LR-MDS 7.5 mg/kg Q4W",
                evid == 1, time == 0) |>
  dplyr::mutate(amt_mg = amt * mw_imetelstat / 1000) |>
  dplyr::select(id, time, amt = amt_mg, treatment)
conc_obj <- PKNCA::PKNCAconc(sim_lrmds_nca,
                             Cc ~ time | treatment + id,
                             concu = "ug/mL",
                             timeu = "h")

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

intervals <- data.frame(
  start       = 0,
  end         = nca_window_h,
  cmax        = TRUE,
  tmax        = TRUE,
  auclast     = TRUE,
  aucinf.obs  = TRUE,
  half.life   = TRUE
)

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

Comparison against published NCA

Gonzalez-Sales 2024 Results, ‘Evaluation of clinical relevance of covariates’, reports secondary PK parameters from cycle-1 simulations of LR-MDS patients at 7.5 mg/kg as geometric mean (CV%): Cmax = 89.5 ug/mL (27.3%), AUC0-28d = 559 h*ug/mL (43.2%), and apparent t1/2 = 4.9 h (43.2%).

published <- tibble::tribble(
  ~treatment,               ~cmax, ~tmax, ~auclast, ~half.life,
  "LR-MDS 7.5 mg/kg Q4W",   89.5,  2.0,   559,      4.9
)

cmp <- nlmixr2lib::ncaComparisonTable(
  simulated     = nca_res,
  reference     = published,
  by            = "treatment",
  units         = c(cmax = "ug/mL", auclast = "h*ug/mL",
                    tmax = "h", half.life = "h"),
  tolerance_pct = 20
)

knitr::kable(
  cmp,
  caption = paste(
    "Simulated (typical 7.5 mg/kg Q4W LR-MDS cohort, cycle 1) vs. Gonzalez-Sales",
    "2024 published cycle-1 NCA. * differs from reference by >20%."
  ),
  align = c("l", "l", "r", "r", "r")
)
Simulated (typical 7.5 mg/kg Q4W LR-MDS cohort, cycle 1) vs. Gonzalez-Sales 2024 published cycle-1 NCA. * differs from reference by >20%.
NCA parameter treatment Reference Simulated % diff
Cmax (ug/mL) LR-MDS 7.5 mg/kg Q4W 89.5 101 +12.9%
Tmax (h) LR-MDS 7.5 mg/kg Q4W 2 2 +0.0%
AUClast (h*ug/mL) LR-MDS 7.5 mg/kg Q4W 559 721 +29.1%*
t½ (h) LR-MDS 7.5 mg/kg Q4W 4.9 42.4 +764.7%*

Assumptions and deviations

  • IIV on residual variability dropped. Gonzalez-Sales 2024 Table 2 reports a between-subject variability on the residual error standard deviation (omega^2 = 0.302 on log(W); CV ~54.6%) in addition to the typical proportional CV of 21.8%. The marginal residual variance therefore exceeds the typical value. This implementation collapses the residual to a single proportional error propSd = 0.218 (matching the typical-value CV) and does not estimate per-subject residual scales. Downstream simulations consequently understate the residual spread slightly; for typical-value predictions and PKNCA validation against geometric-mean exposures this is inconsequential.

  • Unit interpretation of Kint, Kon, Koff. Table 2 reports Kint = 0.103 L/h/70 kg, Kon = 0.159 L^2/(uM*h), and Koff = 0.609 L/h. The NONMEM ODE system (supplement $DES) treats the bound-complex state (A(3)) and the deep-peripheral state (A(2)) as concentration-like states with an implicit unit volume of 1 L – so that BMAX - A(3) is a dimensionally consistent subtraction in uM. Under that convention the rate constants act as 1/time (Kint, Koff) or 1/(uM * time) for Kon. The published unit labels in Table 2 are consistent with the bound-state-as-concentration convention but include an extra factor of L that is conventional rather than dimensionally derived. This implementation reproduces the NONMEM equations directly; numerical results are unchanged regardless of the unit-label ambiguity.

  • No supplement Figure S1 / S2 covariate-pairwise screen. The paper’s supplement Figures S1 and S2 (scatter and box plots of pairwise covariate correlations) were not available when this model was built; the final-model covariates retained after backward elimination are those reported in Table 2 and reproduced here.

  • Population body-weight distribution. The published Table S1 (full baseline-demographics table) was not available when this model was built; the virtual cohort uses a normal weight distribution centred at the main-paper mean (77.2 kg) with a standard deviation chosen to span the published 44.0-161 kg range. This is conservative for the cycle-1 NCA comparison because the allometric weight scaling on CL and Vc, plus the dose-on-CL effect at fixed mg/kg dosing, are not strongly sensitive to the exact weight distribution.

  • Reference-disease group. The reference disease group for the malignancy-effect coefficients on CL, Vc, and Bmax is “solid tumors and other hematologic malignancies excluding MF and MM” – the supplement MTYPE = 0 value. The LR-MDS cohort (MTYPE = 2) is also in the reference group for the model’s MF-on-CL, MF-on-Bmax, and MM-on-Vc effects, which drop out for non-MF and non-MM subjects.

  • Simulated AUClast exceeds the published value by ~29%. The PKNCA comparison flags simulated AUClast 721 hug/mL vs published 559 hug/mL with * for >20% difference (Cmax matches within 13%, Tmax matches exactly). The discrepancy reflects three sources of simulation-vs-publication divergence and is not a transcription error in the model parameters: (1) the virtual cohort’s body-weight distribution is approximated (normal, mean 77.2 kg, SD 15 kg) rather than drawn from the actual MDS3001 patient-level demographics, so per-subject doses (7.5 mg/kg) span a wider range than the published cohort; (2) the PKNCA summary across 100 simulated subjects with log-normal IIV on CL, Vc, and Bmax (between-subject CV approximately 43.7% / 25.7% / 45.8%) is right-skewed, inflating arithmetic and even geometric means above the typical-value AUC; (3) the published cycle-1 AUC0-28d may have been computed from individual post-hoc Bayesian parameter estimates rather than a forward simulation from typical-value parameters. Parameter values match Gonzalez-Sales 2024 Table 2 exactly, so the structural model and parameter transcription are correct; the AUC discrepancy is a simulation-design feature rather than a model bug.

  • No erratum found. A search of PubMed for “Gonzalez-Sales 2024 imetelstat erratum” returned no results; the main-paper Table 2 values are used as final estimates without correction.