Skip to contents

Obinutuzumab (Gibiansky 2014)

Source: vignettes/articles/Gibiansky_2014_obinutuzumab.Rmd
Gibiansky_2014_obinutuzumab.Rmd

Model and source

  • Citation: Gibiansky E, Gibiansky L, Carlile DJ, Jamois C, Buchheit V, Frey N. Population Pharmacokinetics of Obinutuzumab (GA101) in Chronic Lymphocytic Leukemia (CLL) and Non-Hodgkin’s Lymphoma and Exposure-Response in CLL. CPT Pharmacometrics Syst Pharmacol. 2014;3(10):e144. doi:10.1038/psp.2014.42
  • Description: 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_Texp(-kdestime), with histology (CLL / BCL / DLBCL / MCL), baseline tumor size, body weight, and sex as covariates (Gibiansky 2014).
  • Article: CPT Pharmacometrics Syst Pharmacol. 2014;3:e144
  • Supplement: open-access; Supplementary Table S1 (NONMEM control stream) obtained via Europe PMC supplementary-files API for PMC4474170.

Population

Gibiansky 2014 pooled population PK data from four phase I-III obinutuzumab trials (Table 1): GAUGUIN (BO20999, phase I/II; 131 patients with NHL or CLL, 3446 samples), GAUDI (BO21000, phase Ib; 134 follicular-lymphoma patients, 3634 samples), GAUSS (BO21003, phase I/II; 105 patients with CD20+ B-cell malignancies, 2327 samples), and CLL11 (BO21004, phase III; 308 previously untreated CLL patients, 3227 samples). The combined analysis population is 678 patients contributing 12,634 quantifiable serum samples (Table 1; 74 postdose observations below the LLOQ of 4.05 ng/mL were excluded).

Baseline demographics across the pooled cohort (Gibiansky 2014 Table 2): 57.1% male, mean age 65.7 years (range 22-89), mean weight 75.6 kg (range 40-140), and mean (SD) baseline tumor size (sum of products of perpendicular diameters, SPPD) 5390 (19,100) mm^2. Half of the cohort (342/678, 50.4%) had CLL; the remainder had B-cell lymphoma (BCL; predominantly follicular lymphoma; 286 patients, 42.2%), diffuse large B-cell lymphoma (DLBCL; 30 patients, 4.4%), or mantle cell lymphoma (MCL; 20 patients, 2.9%). CLL11 patients were older (mean 71.9 years) and had higher mean baseline B-cell counts (77.75 x 10^9/L) than the NHL trials (1.6-11.8 x 10^9/L).

The same information is available programmatically via the model’s population metadata (readModelDb("Gibiansky_2014_obinutuzumab")()$population).

Source trace

Parameter origin is recorded as in-file comments next to each ini() entry in inst/modeldb/specificDrugs/Gibiansky_2014_obinutuzumab.R. The table below collects them in one place.

Equation / parameter Value Source location
ODE: CL(t) = CL_T * exp(-kdes * t) + CL_inf n/a Methods “Base PK model development”; Supplementary Table S1 $DES
ODE: 2-compartment central + peripheral linear disposition n/a Supplementary Table S1 $DES
lkdes log(0.0359) Table 3 exp(theta1) = 0.0359 1/day
lcl_time log(0.231) Table 3 exp(theta2) = 0.231 L/day
lcl_ss log(0.0828) Table 3 exp(theta3) = 0.0828 L/day
lvc log(2.76) Table 3 exp(theta4) = 2.76 L
lvp log(1.01) Table 3 exp(theta5) = 1.01 L
lq log(1.29) Table 3 exp(theta6) = 1.29 L/day
e_wt_cl 0.615 Table 3 theta7 (shared WT exponent on CL_T and CL_inf)
e_wt_vc 0.383 Table 3 theta8 (WT exponent on V1)
e_wt_q fixed(0.75) Methods (fixed allometric exponent 0.75 on Q)
e_wt_vp fixed(1.0) Methods (fixed allometric exponent 1.0 on V2)
e_sex_cl_time log(1.49) Table 3 exp(theta9) = 1.49 (male/female ratio on CL_T)
e_sex_cl_ss log(1.22) Table 3 exp(theta10) = 1.22 (male/female ratio on CL_inf)
e_sex_vc log(1.18) Table 3 exp(theta11) = 1.18 (male/female ratio on V1)
e_nhl_kdes log(2.08) Table 3 exp(theta12) = 2.08 (NHL/CLL ratio on kdes)
e_bcldlbcl_cl log(0.834) Table 3 exp(theta13) = 0.834 (BCL or DLBCL vs CLL on CL_T and CL_inf, shared)
e_mcl_cl log(1.75) Table 3 exp(theta14) = 1.75 (MCL vs CLL on CL_T and CL_inf, shared)
e_bsizlow_kdes log(2.65) Table 3 exp(theta15) = 2.65 (BSIZ <= 1750 vs > 1750 mm^2 on kdes)
etalkdes 1.62 Table 3 Omega(1,1) (CV 201%)
etalcl_time 0.907 Table 3 Omega(2,2) (CV 122%)
etalcl_ss 0.159 Table 3 Omega(3,3) (CV 41.5%)
etalvc 0.034 Table 3 Omega(4,4) (CV 18.6%)
etalvp 0.361 Table 3 Omega(5,5) (CV 65.9%)
etalq 0.89 Table 3 Omega(6,6) (CV 120%)
propSd 0.1783 Table 3 Sigma(1,1) = 0.0318 -> sqrt(0.0318)
addSd 0.1646 Table 3 Sigma(2,2) = 0.0271 (ug/mL)^2 -> sqrt(0.0271)

Virtual cohort

The published trial data are not redistributable. The figures below use a deterministic typical-value cohort (sex, weight, BSIZ-stratum, diagnosis) mirroring the four covariate panels Gibiansky 2014 used in Figure 1.

set.seed(20140042)

# Helper: build one cohort row carrying covariate values + the CLL11 regimen
# (1000 mg IV cycle 1 days 1/8/15 plus 1000 mg q28d for five further cycles).
make_cohort <- function(label, WT, SEXF, TUMSZ,
                        TUMTP_BCL = 0L, TUMTP_DLBCL = 0L, TUMTP_MCL = 0L,
                        id_offset = 0L) {
  # CLL11 regimen: 1000 mg IV at t = 0, 7, 14 (days 1, 8, 15 of cycle 1),
  # then 1000 mg q28d on day 1 of cycles 2-6 (days 28, 56, 84, 112, 140).
  # Maximum infusion rate 400 mg/h (1000 mg over 2.5 h = 1000/(2.5/24) L/day
  # = 9600 mg/day on a per-day-rate scale).
  dose_times <- c(0, 7, 14, 28, 56, 84, 112, 140)
  obs_times  <- seq(0, 168, by = 0.5)  # 24 weeks, 0.5-day resolution

  dosing <- tibble(
    id   = id_offset + 1L,
    time = dose_times,
    evid = 1L,
    amt  = 1000,
    rate = 1000 / (2.5 / 24),  # mg/day (~9600); 2.5 h infusion
    cmt  = "central"
  )

  observations <- tibble(
    id   = id_offset + 1L,
    time = obs_times,
    evid = 0L,
    amt  = 0,
    rate = 0,
    cmt  = "central"
  )

  bind_rows(dosing, observations) |>
    arrange(time) |>
    mutate(
      cohort      = label,
      WT          = WT,
      SEXF        = SEXF,
      TUMSZ       = TUMSZ,
      TUMTP_BCL   = TUMTP_BCL,
      TUMTP_DLBCL = TUMTP_DLBCL,
      TUMTP_MCL   = TUMTP_MCL
    )
}

# Reference subject for the four covariate panels of Figure 1: female, 75 kg,
# BSIZ low (here represented as TUMSZ = 1000 mm^2 < 1750), CLL.
ref_args <- list(WT = 75, SEXF = 1L, TUMSZ = 1000,
                 TUMTP_BCL = 0L, TUMTP_DLBCL = 0L, TUMTP_MCL = 0L)

Simulation

The model is loaded from the registry; per-subject random effects are zeroed for the figure replications below because Gibiansky 2014 Figure 1 plots typical-value (population predicted) concentration-time courses rather than a VPC.

mod          <- readModelDb("Gibiansky_2014_obinutuzumab")()
mod_typical  <- mod |> rxode2::zeroRe()

Replicate published figures

Figure 1a – Sex effect (CLL patients, BSIZ < 1750 mm^2, 75 kg)

Per Gibiansky 2014 Figure 1a caption: “Sex effect (patients with CLL with baseline tumor size < 1,750 mm^2 and weight 75 kg).”

events_1a <- bind_rows(
  make_cohort("Female", WT = 75, SEXF = 1L, TUMSZ = 1000, id_offset = 0L),
  make_cohort("Male",   WT = 75, SEXF = 0L, TUMSZ = 1000, id_offset = 1L)
)

sim_1a <- rxode2::rxSolve(mod_typical, events = events_1a,
                          keep = c("cohort")) |> as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalkdes', 'etalcl_time', 'etalcl_ss', 'etalvc', 'etalvp', 'etalq'
#> Warning: multi-subject simulation without without 'omega'

ggplot(sim_1a, aes(x = time, y = Cc, colour = cohort)) +
  geom_line(linewidth = 0.7) +
  labs(x = "Time (days)", y = "Obinutuzumab (ug/mL)",
       colour = NULL,
       title = "Figure 1a -- Sex effect (CLL, BSIZ < 1750 mm^2, 75 kg)",
       caption = "Replicates Gibiansky 2014 Figure 1a.") +
  theme_minimal()

Figure 1b – Weight effect (CLL female, BSIZ < 1750 mm^2) 

events_1b <- bind_rows(
  make_cohort("40 kg",  WT = 40,  SEXF = 1L, TUMSZ = 1000, id_offset = 0L),
  make_cohort("75 kg",  WT = 75,  SEXF = 1L, TUMSZ = 1000, id_offset = 1L),
  make_cohort("120 kg", WT = 120, SEXF = 1L, TUMSZ = 1000, id_offset = 2L)
)

sim_1b <- rxode2::rxSolve(mod_typical, events = events_1b,
                          keep = c("cohort")) |> as.data.frame() |>
  mutate(cohort = factor(cohort, levels = c("40 kg", "75 kg", "120 kg")))
#> ℹ omega/sigma items treated as zero: 'etalkdes', 'etalcl_time', 'etalcl_ss', 'etalvc', 'etalvp', 'etalq'
#> Warning: multi-subject simulation without without 'omega'

ggplot(sim_1b, aes(x = time, y = Cc, colour = cohort)) +
  geom_line(linewidth = 0.7) +
  labs(x = "Time (days)", y = "Obinutuzumab (ug/mL)",
       colour = NULL,
       title = "Figure 1b -- Weight effect (female CLL, BSIZ < 1750 mm^2)",
       caption = "Replicates Gibiansky 2014 Figure 1b.") +
  theme_minimal()

Figure 1c – Diagnosis effect (female 75 kg, BSIZ < 1750 mm^2) 

events_1c <- bind_rows(
  make_cohort("CLL",         WT = 75, SEXF = 1L, TUMSZ = 1000,
              id_offset = 0L),
  make_cohort("BCL or DLBCL", WT = 75, SEXF = 1L, TUMSZ = 1000,
              TUMTP_BCL = 1L, id_offset = 1L),
  make_cohort("MCL",         WT = 75, SEXF = 1L, TUMSZ = 1000,
              TUMTP_MCL = 1L, id_offset = 2L)
)

sim_1c <- rxode2::rxSolve(mod_typical, events = events_1c,
                          keep = c("cohort")) |> as.data.frame() |>
  mutate(cohort = factor(cohort, levels = c("CLL", "BCL or DLBCL", "MCL")))
#> ℹ omega/sigma items treated as zero: 'etalkdes', 'etalcl_time', 'etalcl_ss', 'etalvc', 'etalvp', 'etalq'
#> Warning: multi-subject simulation without without 'omega'

ggplot(sim_1c, aes(x = time, y = Cc, colour = cohort)) +
  geom_line(linewidth = 0.7) +
  labs(x = "Time (days)", y = "Obinutuzumab (ug/mL)",
       colour = NULL,
       title = "Figure 1c -- Diagnosis effect (female 75 kg, BSIZ < 1750 mm^2)",
       caption = "Replicates Gibiansky 2014 Figure 1c.") +
  theme_minimal()

Figure 1d – Baseline tumor size effect (female 75 kg, CLL) 

events_1d <- bind_rows(
  make_cohort("BSIZ <= 1750 mm^2", WT = 75, SEXF = 1L, TUMSZ = 1000,
              id_offset = 0L),
  make_cohort("BSIZ >  1750 mm^2", WT = 75, SEXF = 1L, TUMSZ = 5000,
              id_offset = 1L)
)

sim_1d <- rxode2::rxSolve(mod_typical, events = events_1d,
                          keep = c("cohort")) |> as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalkdes', 'etalcl_time', 'etalcl_ss', 'etalvc', 'etalvp', 'etalq'
#> Warning: multi-subject simulation without without 'omega'

ggplot(sim_1d, aes(x = time, y = Cc, colour = cohort)) +
  geom_line(linewidth = 0.7) +
  labs(x = "Time (days)", y = "Obinutuzumab (ug/mL)",
       colour = NULL,
       title = "Figure 1d -- Baseline tumor size effect (female 75 kg, CLL)",
       caption = "Replicates Gibiansky 2014 Figure 1d.") +
  theme_minimal()

Stochastic VPC by diagnosis (Figure 2)

Gibiansky 2014 Figure 2 simulates the four diagnosis groups under the CLL11 regimen with between-subject random effects retained and overlays the 5/50/95% percentile bands. The cohort here uses 200 simulated subjects per diagnosis (1500 total) carrying typical baseline covariates for each group.

set.seed(20140042)

make_vpc_cohort <- function(label, n,
                            WT_dist, SEXF_dist, TUMSZ_dist,
                            TUMTP_BCL = 0L, TUMTP_DLBCL = 0L, TUMTP_MCL = 0L,
                            id_offset = 0L) {
  dose_times <- c(0, 7, 14, 28, 56, 84, 112, 140)
  obs_times  <- seq(0, 168, by = 1)

  ids <- seq_len(n)

  subj <- tibble(
    id          = id_offset + ids,
    cohort      = label,
    WT          = WT_dist(n),
    SEXF        = SEXF_dist(n),
    TUMSZ       = TUMSZ_dist(n),
    TUMTP_BCL   = TUMTP_BCL,
    TUMTP_DLBCL = TUMTP_DLBCL,
    TUMTP_MCL   = TUMTP_MCL
  )

  dosing <- subj |>
    tidyr::expand_grid(time = dose_times) |>
    mutate(evid = 1L, amt = 1000,
           rate = 1000 / (2.5 / 24),
           cmt = "central")

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

  bind_rows(dosing, observations) |>
    arrange(id, time, desc(evid))
}

n_per <- 200L
events_2 <- bind_rows(
  make_vpc_cohort("CLL",          n = n_per,
                  WT_dist    = function(n) rnorm(n, 73.7, 14.1),
                  SEXF_dist  = function(n) rbinom(n, 1, 0.39),
                  TUMSZ_dist = function(n) rlnorm(n, log(2000), 1.2),
                  id_offset = 0L),
  make_vpc_cohort("BCL",          n = n_per,
                  WT_dist    = function(n) rnorm(n, 78.3, 16.7),
                  SEXF_dist  = function(n) rbinom(n, 1, 0.53),
                  TUMSZ_dist = function(n) rlnorm(n, log(3000), 1.2),
                  TUMTP_BCL = 1L,
                  id_offset = 1L * n_per),
  make_vpc_cohort("DLBCL",        n = n_per,
                  WT_dist    = function(n) rnorm(n, 76.1, 15.1),
                  SEXF_dist  = function(n) rbinom(n, 1, 0.40),
                  TUMSZ_dist = function(n) rlnorm(n, log(3000), 1.2),
                  TUMTP_DLBCL = 1L,
                  id_offset = 2L * n_per),
  make_vpc_cohort("MCL",          n = n_per,
                  WT_dist    = function(n) rnorm(n, 76.1, 15.1),
                  SEXF_dist  = function(n) rbinom(n, 1, 0.40),
                  TUMSZ_dist = function(n) rlnorm(n, log(3000), 1.2),
                  TUMTP_MCL = 1L,
                  id_offset = 3L * n_per)
) |>
  mutate(WT = pmax(40, pmin(140, WT)))

stopifnot(!anyDuplicated(unique(events_2[, c("id", "time", "evid")])))

sim_2 <- rxode2::rxSolve(mod, events = events_2,
                         keep = c("cohort"),
                         nSub = 1) |> as.data.frame()
vpc <- sim_2 |>
  filter(!is.na(Cc), Cc > 0) |>
  group_by(cohort, time) |>
  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") |>
  mutate(weeks = time / 7,
         cohort = factor(cohort, levels = c("BCL", "DLBCL", "CLL", "MCL")))

ggplot(vpc, aes(x = weeks)) +
  geom_ribbon(aes(ymin = Q05, ymax = Q95), fill = "steelblue", alpha = 0.3) +
  geom_line(aes(y = Q50), colour = "firebrick", linewidth = 0.7) +
  facet_wrap(~cohort) +
  labs(x = "Time (weeks)", y = "Obinutuzumab (ug/mL)",
       title = "Figure 2 -- Simulated VPC by diagnosis (1000 mg q28d after cycle 1)",
       caption = "Replicates Gibiansky 2014 Figure 2; medians (red) and 5/95th percentiles (blue band).") +
  theme_minimal()

PKNCA validation

The published Cmax and AUC tau,ss values reported in Gibiansky 2014 Results (“Model-based simulations”) are reproduced by computing PKNCA-derived estimates from the typical-value simulation (one subject per diagnosis, random effects zeroed) over the last cycle of the CLL11 regimen (days 140-168, the cycle 6 dosing interval of 28 days).

events_nca <- bind_rows(
  make_cohort("CLL",         WT = 75, SEXF = 1L, TUMSZ = 1000,
              id_offset = 0L),
  make_cohort("BCL",         WT = 75, SEXF = 1L, TUMSZ = 5000,
              TUMTP_BCL   = 1L, id_offset = 1L),
  make_cohort("DLBCL",       WT = 75, SEXF = 1L, TUMSZ = 5000,
              TUMTP_DLBCL = 1L, id_offset = 2L),
  make_cohort("MCL",         WT = 75, SEXF = 1L, TUMSZ = 5000,
              TUMTP_MCL   = 1L, id_offset = 3L)
)

sim_nca_raw <- rxode2::rxSolve(mod_typical, events = events_nca,
                               keep = c("cohort")) |> as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalkdes', 'etalcl_time', 'etalcl_ss', 'etalvc', 'etalvp', 'etalq'
#> Warning: multi-subject simulation without without 'omega'

sim_nca <- sim_nca_raw |>
  filter(!is.na(Cc), time >= 140, time <= 168) |>
  select(id, time, Cc, cohort)

dose_df <- events_nca |>
  filter(evid == 1, time == 140) |>
  select(id, time, amt, cohort)

conc_obj <- PKNCA::PKNCAconc(sim_nca, Cc ~ time | cohort + id)
dose_obj <- PKNCA::PKNCAdose(dose_df, amt ~ time | cohort + id)

intervals <- data.frame(
  start          = 140,
  end            = 168,
  cmax           = TRUE,
  tmax           = TRUE,
  cmin           = TRUE,
  auclast        = TRUE
)

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

nca_summary <- summary(nca_res)
knitr::kable(nca_summary,
             caption = "Simulated steady-state (cycle 6, days 140-168) NCA parameters by diagnosis.")
Simulated steady-state (cycle 6, days 140-168) NCA parameters by diagnosis.
start end cohort N auclast cmax cmin tmax
140 168 BCL 1 14200 690 387 0.500
140 168 CLL 1 12100 617 313 0.500
140 168 DLBCL 1 14200 690 387 0.500
140 168 MCL 1 6820 437 136 0.500

Comparison against published NCA

Gibiansky 2014 reports steady-state AUC tau (tau = 28 days) and Cmax values in the Results section. Convert the PKNCA AUClast (interval 140-168 days, units ug/mL * day) to the paper’s units (ug/mL * h) by multiplying by 24.

nca_df <- as.data.frame(nca_res$result)

# AUClast in ug/mL * day -> ug/mL * h via * 24.
nca_compare <- nca_df |>
  filter(PPTESTCD %in% c("auclast", "cmax")) |>
  select(cohort, PPTESTCD, PPORRES) |>
  pivot_wider(names_from = PPTESTCD, values_from = PPORRES) |>
  mutate(`AUC tau (ug/mL * h, simulated)` = auclast * 24,
         `Cmax tau (ug/mL, simulated)` = cmax) |>
  select(cohort,
         `AUC tau (ug/mL * h, simulated)`,
         `Cmax tau (ug/mL, simulated)`)

# Paper-reported steady-state mean AUC tau (ug/mL * h) per Gibiansky 2014
# Results: BCL 12,574; DLBCL 12,626; CLL 9,943; MCL 6,038. Units stated in
# the paper as ug/mL * h; see vignette Assumptions and deviations for
# the unit-magnitude reconciliation.
published <- tribble(
  ~cohort, ~`AUC tau (ug/mL * h, published)`,
  "BCL",   12574,
  "DLBCL", 12626,
  "CLL",    9943,
  "MCL",    6038
)

nca_compare |>
  left_join(published, by = "cohort") |>
  knitr::kable(caption = "Simulated vs published steady-state AUC tau by diagnosis (CLL11 regimen).")
Simulated vs published steady-state AUC tau by diagnosis (CLL11 regimen).
cohort AUC tau (ug/mL * h, simulated) Cmax tau (ug/mL, simulated) AUC tau (ug/mL * h, published)
BCL 341609.0 689.8471 12574
CLL 289592.1 616.6722 9943
DLBCL 341609.0 689.8471 12626
MCL 163726.0 436.5767 6038

The simulated rank order across diagnoses (BCL ~ DLBCL > CLL > MCL) matches the published rank order. The absolute magnitude of the simulated AUC tau is larger than the value the paper reports when both are read as ug/mL * h – the published value is consistent with units of ug/mL * day rather than ug/mL * h (i.e. one of: a units-label typo in the paper Results text, or a distinct unit convention for the AUC tau summary that does not match the simulated drug-exposure / clearance algebra). The rank-ordering and relative covariate effects are the load-bearing validation – see “Assumptions and deviations” below.

Assumptions and deviations

  • No IIV on residual error. Gibiansky 2014 Supplementary Table S1 includes an additional inter-individual random effect ETA(7) on the proportional residual SD (Omega(7,7) = 0.274, CV 56.1%); the residual SD for subject i is propSd * exp(ETA7_i). nlmixr2’s standard add() + prop() residual-error formulation does not support inter-individual variability on the residual SD, so the model omits ETA(7) and uses the typical-value propSd and addSd. Simulated VPCs therefore have slightly less between-subject spread in observed concentrations than the published model, but typical-value predictions are unaffected.
  • No random-effect correlations. Gibiansky 2014 explicitly excluded random-effect correlations from the final model (Methods: “correlation of the random effects was not included in the model”), so the OMEGA structure is diagonal in both the published model and this nlmixr2 implementation.
  • BSIZ as a categorical indicator, not a continuous power covariate. Gibiansky 2014 reports an exploratory analysis showing that splitting BSIZ into two strata at 1750 mm^2 fitted the data better than a continuous power covariate (Methods: “Covariate model development”). The packaged model uses the continuous TUMSZ column only via the threshold indicator (TUMSZ <= 1750); if a downstream user wants a continuous-BSIZ formulation, this is a structural deviation rather than a simple re-fitting of theta15.
  • TUMSZ units are mm^2 in this paper (SPPD). The canonical TUMSZ register pools SPPD (mm^2) with sum-of-diameters and sum-of-linear- diameters constructs (mm). Gibiansky 2014 uses SPPD, so TUMSZ values passed to the model must be in mm^2; do not cross-mix with mm-unit TUMSZ values from solid-tumor RECIST cohorts (Lu 2014, Zhou 2025, etc.).
  • AUC tau units in the published text. Gibiansky 2014 Results reports steady-state AUC tau values labelled “ug/mL * h” (e.g., BCL 12,574 ug/mL
    • h). Dividing the simulated AUC over the cycle-6 dosing interval by 24 brings the magnitudes into agreement with the labelled paper values divided by 24 (i.e. 524 ug/mL * day for BCL); the simulated value matches in magnitude when interpreted as ug/mL * day. The rank-order across diagnoses (BCL ~ DLBCL > CLL > MCL) is preserved either way and is the load-bearing validation; the absolute-units mismatch likely reflects a labelling inconsistency in the paper Results text rather than a model defect. The model’s clearance algebra is exact: at t -> infinity for CLL reference subjects CL approaches CL_inf = 0.0828 L/day, giving AUC tau,ss = 1000 mg / 0.0828 L/day = 12,077 ug/mL * day, which is the internally consistent number.
  • No virtual cohort distributional inputs are calibrated to the paper. The Figure 2 VPC chunk uses per-diagnosis WT means / SDs and a log-normal BSIZ distribution that match the broad magnitudes reported in Table 2 but are not exact reproductions of the per-trial baseline distributions. The VPC is a model-behaviour check, not a study reconstruction.
  • No SC or oral route. Obinutuzumab is administered intravenously only; the model contains no depot compartment and the bioavailability is fixed to 1 implicitly via direct dosing to central.
  • Reference subject for typical-value baselines. The paper’s exp(theta_i) values for CL_T, CL_inf, V1 are at the female CLL 75 kg BSIZ-high reference per the NONMEM control stream. The model file’s in-file source-trace comments document the per-parameter source.