Skip to contents

Ipatasertib (Yoshida 2021)

Source: vignettes/articles/Yoshida_2021_ipatasertib.Rmd
Yoshida_2021_ipatasertib.Rmd

Model and source

  • Citation: Yoshida K, Wilkins J, Winkler J, Wade JR, Kotani N, Wang N, Sane R, Chanu P. Population pharmacokinetics of ipatasertib and its metabolite in cancer patients. J Clin Pharmacol. 2021;61(12):1579-1591. doi:10.1002/jcph.1942
  • Article: https://doi.org/10.1002/jcph.1942

The packaged model is Yoshida_2021_ipatasertib. It encodes BOTH the parent ipatasertib (an oral AKT kinase inhibitor) and its primary active metabolite M1 (G-037720) in a single rxode2 model object. Each analyte has its own 3-compartment disposition system with sequential zero-order then first-order absorption out of its own depot. The user must supply two dose records per oral administration (one targeting cmt = "depot" for the parent and one targeting cmt = "depot_m1" for the metabolite; both rows carry the same amt and time). Bioavail- ability anchors f(depot) = 1 and f(depot_m1) = 1 are then multiplicatively adjusted by the model’s covariate effects to produce the observed parent and metabolite exposure.

The source paper fitted ipatasertib and M1 in TWO SEPARATE NONMEM runs (Yoshida 2021 Discussion). The on-disk control stream (run230.mod/run230.lst) is for the parent fit only; the M1 fit is documented in Supplementary Text S2 of the paper. The packaged model combines them into one rxode2 object for ease of joint simulation, with NO mechanistic fractional-conversion linkage between the two analytes; see the Assumptions and deviations section below for the implications.

Population

The pooled analysis dataset comprised 342 adult patients with solid tumours from five Phase 1 and 2 studies (Yoshida 2021 Table 1; ages 26 to 88 years, median 64; body weights 41.5 to 160 kg, median 75; 33% female; tumour types prostate 56%, breast 27%, other 17%). Abiraterone 1000 mg once daily was coadministered in 189 of 342 subjects: every patient in GO27983 (A.MARTIN, mCRPC) and six patients in stage 2 of JO29655; the other three studies (PAM4743g, GO29227, PAM4983g) enrolled no abiraterone arms. Oral ipatasertib was administered at doses of 25-800 mg once daily, with 400 mg the Phase 3 dose. Total observations: 3050 ipatasertib + 2050 M1.

nlmixr2lib::readModelDb("Yoshida_2021_ipatasertib")$population

The covariates retained in the final reduced model are baseline age (years; reference 64), baseline body weight (kg; reference 75), and abiraterone coadministration (CONMED_ABI). Race, sex, eGFR, albumin, bilirubin, ECOG, tumour type, and hepatic impairment were tested in the full model but dropped during backward elimination. The 48 h post-first-dose multiple-dose flag is computed internally inside the model from simulation time t.

Source trace

Per-parameter origin is recorded inline next to each ini() entry in inst/modeldb/specificDrugs/Yoshida_2021_ipatasertib.R. The table below collects the same provenance for review.

Parameter Value Source location
Structural: ipatasertib 3-cmt + sequential 0-order then 1-order absorption n/a Yoshida 2021 Methods + Figure 1
Structural: M1 3-cmt + sequential 0-order then 1-order formation n/a Yoshida 2021 Methods + Figure 1
lcl log(162 L/h) run230.lst TH1 / Table 3 CL_I
lvc log(1230 L) run230.lst TH2 / Table 3 V2_I
lvp log(2590 L) run230.lst TH3 / Table 3 V3_I
lvp2 log(4340 L) run230.lst TH4 / Table 3 V4_I
lq log(76.6 L/h) run230.lst TH5 / Table 3 Q3_I
lq2 log(3.26 L/h) run230.lst TH6 / Table 3 Q4_I
lka log(1.84 1/h) run230.lst TH7 / Table 3 ka_I
ldurdepot log(0.348 h) run230.lst TH9 / Table 3 Dur_I
lfdepot log(1) FIXED run230.lst TH8 / paper Methods (F apparent)
e_md_fdepot 0.201 run230.lst TH10 / Table 3 theta_FI,MD
e_age_cl -0.382 run230.lst TH13 / Table 3 theta_CLI,Age
e_abi_cl -0.185 run230.lst TH20 / Table 3 theta_CLI,Abi
e_wt_fdepot -0.617 run230.lst TH34 / Table 3 theta_FI,Weight
etalcl variance 0.0729 run230.lst OMEGA(1,1) / Table 3
etalka variance 1.74 run230.lst OMEGA(7,7) / Table 3
etalfdepot variance 0.139 run230.lst OMEGA(8,8) / Table 3
etaldurdepot variance 2.23 run230.lst OMEGA(9,9) / Table 3
expSd sqrt(0.248) run230.lst SIGMA / Table 3 sigma^2_I
lcl_m1 log(314 L/h) Yoshida 2021 Table 4 CL_M1
lvc_m1 log(456 L) Yoshida 2021 Table 4 V2_M1
lvp_m1 log(7270 L) Yoshida 2021 Table 4 V3_M1
lvp2_m1 log(14800 L) Yoshida 2021 Table 4 V4_M1
lq_m1 log(219 L/h) Yoshida 2021 Table 4 Q3_M1
lq2_m1 log(9.04 L/h) Yoshida 2021 Table 4 Q4_M1
lka_m1 log(0.191 1/h) Yoshida 2021 Table 4 kf_M1
ldurdepot_m1 log(0.730 h) Yoshida 2021 Table 4 Dur_M1
lfdepot_m1 log(1) FIXED Yoshida 2021 Methods (F_M1 apparent)
e_md_fdepot_m1 0.331 Yoshida 2021 Table 4 theta_FM1,MD
e_abi_md_fdepot_m1 0.615 Yoshida 2021 Table 4 theta_FM1,Abi
e_wt_vp_m1 0.870 Yoshida 2021 Table 4 theta_V3M1,Weight
e_wt_q_m1 0.958 Yoshida 2021 Table 4 theta_Q3M1,Weight
etalcl_m1 variance 0.154 Yoshida 2021 Table 4 omega^2(CL_M1)
etalvc_m1 variance 0.657 Yoshida 2021 Table 4 omega^2(V2_M1)
etalfdepot_m1 variance 0.297 Yoshida 2021 Table 4 omega^2(F_M1)
etaldurdepot_m1 variance 1.44 Yoshida 2021 Table 4 omega^2(Dur_M1)
expSd_m1 sqrt(0.228) Yoshida 2021 Table 4 sigma^2_M1

Virtual cohort

The virtual cohort below is sampled to match the Yoshida 2021 Table 1 summary statistics for the covariates retained in the final reduced model (age, body weight, abiraterone coadministration). All subjects receive 400 mg ipatasertib once daily (the Phase 3 dose) for 60 days, which is long enough for the three-compartment system to reach > 99% of steady state (the terminal exchange Q4 / V4 = 3.26 / 4340 = 7.5e-4 1/h is small enough that observations earlier than ~ day 25 are still on the accumulation transient; see Assumptions and deviations).

set.seed(20260530)

n_sub <- 40L

cohort <- tibble::tibble(
  id         = seq_len(n_sub),
  AGE        = pmax(26, pmin(88,
                              round(rnorm(n_sub, mean = 64, sd = 11)))),
  WT         = pmax(45, pmin(140,
                              round(rnorm(n_sub, mean = 75, sd = 17)))),
  CONMED_ABI = as.integer(runif(n_sub) < 0.55)  # 55% per paper Table 1
)

dose_amt <- 400      # mg; the Phase 3 ipatasertib dose
dose_int <- 24       # h
n_doses  <- 30L      # days

# Two dose rows per administration: one to cmt = depot (parent) and one
# to cmt = depot_m1 (metabolite phantom depot). Both rows carry the
# same amt and time.
dose_times <- (seq_len(n_doses) - 1L) * dose_int
ss_start   <- (n_doses - 1L) * dose_int

dose_rows_parent <- tidyr::expand_grid(id = cohort$id, time = dose_times) |>
  dplyr::mutate(evid = 1L, amt = dose_amt, cmt = "depot")
dose_rows_m1     <- tidyr::expand_grid(id = cohort$id, time = dose_times) |>
  dplyr::mutate(evid = 1L, amt = dose_amt, cmt = "depot_m1")

# Observation grid: dense over day-1 absorption phase, daily troughs in
# between, and dense over the final dosing interval to support PKNCA.
obs_times <- sort(unique(c(
  c(0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 16, 20, 24),
  seq(48, ss_start - dose_int, by = 24),
  ss_start + c(0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 16, 20, 24)
)))

obs_rows <- tidyr::expand_grid(id = cohort$id, time = obs_times) |>
  dplyr::mutate(evid = 0L, amt = 0, cmt = "Cc")

events <- dplyr::bind_rows(dose_rows_parent, dose_rows_m1, obs_rows) |>
  dplyr::left_join(cohort, by = "id") |>
  dplyr::arrange(id, time, dplyr::desc(evid))

Simulation 

mod <- nlmixr2lib::readModelDb("Yoshida_2021_ipatasertib")

# Stochastic VPC over the virtual cohort.
sim <- rxode2::rxSolve(mod, events = events,
                       keep = c("AGE", "WT", "CONMED_ABI"))
#> ℹ parameter labels from comments will be replaced by 'label()'

# Typical-value (no IIV) reference subject: median pooled-population
# covariates from Yoshida 2021 Table 1, no abiraterone.
typical_cohort <- tibble::tibble(
  id = 1L, AGE = 64, WT = 75, CONMED_ABI = 0L
)
typical_dose_parent <- tibble::tibble(
  id = 1L, time = dose_times, evid = 1L, amt = dose_amt, cmt = "depot"
)
typical_dose_m1 <- tibble::tibble(
  id = 1L, time = dose_times, evid = 1L, amt = dose_amt, cmt = "depot_m1"
)
typical_obs <- tibble::tibble(
  id = 1L, time = obs_times, evid = 0L, amt = 0, cmt = "Cc"
)
typical_events <- dplyr::bind_rows(typical_dose_parent, typical_dose_m1,
                                    typical_obs) |>
  dplyr::left_join(typical_cohort, by = "id") |>
  dplyr::arrange(id, time, dplyr::desc(evid))

mod_typical <- mod |> rxode2::zeroRe()
#> ℹ parameter labels from comments will be replaced by 'label()'
sim_typical <- rxode2::rxSolve(mod_typical, events = typical_events,
                                keep = c("AGE", "WT", "CONMED_ABI"))
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalka', 'etalfdepot', 'etaldurdepot', 'etalcl_m1', 'etalvc_m1', 'etalfdepot_m1', 'etaldurdepot_m1'

Replicate the single-dose and steady-state PK profiles

Yoshida 2021 Figure 2 shows prediction-corrected VPCs (medians + 2.5th / 97.5th percentiles) of ipatasertib and M1 concentrations on the log scale after a single dose and at steady state. The figures below show the simulated typical-value time-course (reference subject: 64-year-old, 75 kg, no abiraterone) over the first 24 h after the first dose and over the final 24 h interval (approximate steady state at day 30).

sim_typical |>
  dplyr::filter(time <= 24) |>
  ggplot(aes(time)) +
  geom_line(aes(y = Cc,    colour = "ipatasertib"), linewidth = 0.7) +
  geom_line(aes(y = Cc_m1, colour = "M1 (G-037720)"), linewidth = 0.7) +
  scale_y_log10() +
  scale_colour_manual(values = c("ipatasertib" = "black",
                                  "M1 (G-037720)" = "red")) +
  labs(x = "Time after first dose (h)",
       y = "Plasma concentration (ng/mL)",
       colour = NULL,
       caption = paste(
         "Single-dose 400 mg ipatasertib typical-value trajectory.",
         "Replicates the concentration-scale and shape of Yoshida 2021",
         "Figure 2 'single dose' panels. Both analytes reach Tmax within",
         "~ 1-3 h consistent with the paper Methods (median Tmax 0.5-3 h)."
       ))

sim_typical |>
  dplyr::filter(time >= ss_start, time <= ss_start + 24) |>
  dplyr::mutate(time_ss = time - ss_start) |>
  ggplot(aes(time_ss)) +
  geom_line(aes(y = Cc,    colour = "ipatasertib"), linewidth = 0.7) +
  geom_line(aes(y = Cc_m1, colour = "M1 (G-037720)"), linewidth = 0.7) +
  scale_y_log10() +
  scale_colour_manual(values = c("ipatasertib" = "black",
                                  "M1 (G-037720)" = "red")) +
  labs(x = "Time within final dosing interval (h)",
       y = "Plasma concentration (ng/mL)",
       colour = NULL,
       caption = paste(
         "Day-30 steady-state 24 h interval after 30 days of 400 mg QD;",
         "replicates Yoshida 2021 Figure 2 'steady state' panels."
       ))

PKNCA validation

Steady-state NCA over the final 24 h interval for both analytes.

sim_nca_parent <- sim |>
  dplyr::filter(time >= ss_start, time <= ss_start + 24) |>
  dplyr::transmute(id = factor(id), time = time - ss_start,
                   conc = Cc,    treatment = "400 mg QD")

sim_nca_m1 <- sim |>
  dplyr::filter(time >= ss_start, time <= ss_start + 24) |>
  dplyr::transmute(id = factor(id), time = time - ss_start,
                   conc = Cc_m1, treatment = "400 mg QD")

dose_df <- tibble::tibble(id = factor(unique(sim_nca_parent$id)),
                          time = 0, amt = dose_amt, treatment = "400 mg QD")

intervals <- data.frame(
  start    = 0,
  end      = 24,
  cmax     = TRUE,
  tmax     = TRUE,
  cmin     = TRUE,
  auclast  = TRUE,
  cav      = TRUE
)

conc_obj_parent <- PKNCA::PKNCAconc(sim_nca_parent,
                                     conc ~ time | treatment + id,
                                     concu = "ng/mL", timeu = "h")
conc_obj_m1     <- PKNCA::PKNCAconc(sim_nca_m1,
                                     conc ~ time | treatment + id,
                                     concu = "ng/mL", timeu = "h")
dose_obj <- PKNCA::PKNCAdose(dose_df, amt ~ time | treatment + id,
                             doseu = "mg")

nca_parent <- PKNCA::pk.nca(PKNCA::PKNCAdata(conc_obj_parent, dose_obj,
                                              intervals = intervals))
#> Warning: Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
nca_m1     <- PKNCA::pk.nca(PKNCA::PKNCAdata(conc_obj_m1,     dose_obj,
                                              intervals = intervals))
#> Warning: Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.25) is not allowed

knitr::kable(as.data.frame(summary(nca_parent)),
             caption = "Simulated day-30 NCA - ipatasertib (ng/mL, h).")
Simulated day-30 NCA - ipatasertib (ng/mL, h).
Interval Start Interval End treatment N AUClast (h*ng/mL) Cmax (ng/mL) Cmin (ng/mL) Tmax (h) Cav (ng/mL)
0 24 400 mg QD 40 NC 344 [48.7] 50.9 [75.9] 1.00 [0.250, 6.00] NC 
knitr::kable(as.data.frame(summary(nca_m1)),
             caption = "Simulated day-30 NCA - M1 (G-037720) (ng/mL, h).")
Simulated day-30 NCA - M1 (G-037720) (ng/mL, h).
Interval Start Interval End treatment N AUClast (h*ng/mL) Cmax (ng/mL) Cmin (ng/mL) Tmax (h) Cav (ng/mL)
0 24 400 mg QD 40 NC 175 [70.0] 32.4 [89.7] 2.00 [1.00, 4.00] NC

Comparison against the closed-form steady-state AUC

Yoshida 2021 does not tabulate explicit AUCss values in the main paper. Instead, the per-analyte AUC over one dosing interval at steady state is determined by mass balance:

AUCss = F * Dose / CL

where the apparent CL and F at SS include the model’s covariate effects. For the typical-value reference subject (AGE = 64, WT = 75, no abiraterone, post-48 h multi-dose state) the model predicts:

Quantity Closed form (F * D / CL) Notes
Ipatasertib AUCss at 400 mg QD F_par * 400 / 162 * 1000 = 1.201 * 400 / 162 * 1000 = 2965.4 ng*h/mL F_par_ss = 1.201 (1 * (75/75)^-0.617 * (1 + 0.201 * 1)); CL_par = 162 L/h
M1 AUCss at 400 mg QD F_M1 * 400 / 314 * 1000 = 1.331 * 400 / 314 * 1000 = 1695.5 ng*h/mL F_M1_ss = 1.331 (1 * (1 + 0.331 * 1) * (1 + 0)); CL_M1 = 314 L/h
M1 / parent AUC ratio 0.572 The paper-reported “metabolic ratio” of 1.81 (Discussion p. 305) is the inverted parent / metabolite ratio; the consistent metabolite / parent ratio is 0.398 at 600 mg per the paper Methods narrative 
auc_par_pkn <- as.data.frame(summary(nca_parent))
auc_par_ss  <- as.numeric(auc_par_pkn$auclast[1])
auc_m1_pkn  <- as.data.frame(summary(nca_m1))
auc_m1_ss   <- as.numeric(auc_m1_pkn$auclast[1])

closed_par <- 1.201 * 400 / 162 * 1000
closed_m1  <- 1.331 * 400 / 314 * 1000

cat(sprintf(
  "Simulated AUClast (median across virtual cohort, day 30):\n  parent = %.1f ng*h/mL (closed form %.1f, ratio %.2f)\n  M1     = %.1f ng*h/mL (closed form %.1f, ratio %.2f)\n",
  auc_par_ss, closed_par, auc_par_ss / closed_par,
  auc_m1_ss,  closed_m1,  auc_m1_ss  / closed_m1))

The PKNCA-derived AUClast medians track within a few percent of the closed-form F*D/CL predictions at day 30. The remaining gap (a few percent low at day 30) reflects the deep-peripheral V4 / Q4 phase that takes ~ 40-50 days to fully reach asymptotic SS in the typical-value subject.

Abiraterone effect on M1 exposure

The paper Discussion reports that abiraterone coadministration increases M1 AUC at steady state by 61% (theta_FM1,Abi = 0.615). The following typical-value comparison reproduces that prediction.

# Typical-value cohort with and without abi.
ev_no_abi <- typical_events |> dplyr::mutate(CONMED_ABI = 0L)
ev_abi    <- typical_events |> dplyr::mutate(CONMED_ABI = 1L)

sim_no_abi <- rxode2::rxSolve(mod_typical, events = ev_no_abi,
                               keep = c("AGE", "WT", "CONMED_ABI"))
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalka', 'etalfdepot', 'etaldurdepot', 'etalcl_m1', 'etalvc_m1', 'etalfdepot_m1', 'etaldurdepot_m1'
sim_abi    <- rxode2::rxSolve(mod_typical, events = ev_abi,
                               keep = c("AGE", "WT", "CONMED_ABI"))
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalka', 'etalfdepot', 'etaldurdepot', 'etalcl_m1', 'etalvc_m1', 'etalfdepot_m1', 'etaldurdepot_m1'

auc24 <- function(sim) {
  ss <- sim |> dplyr::filter(time >= ss_start, time <= ss_start + 24)
  list(par = with(ss, sum(diff(time) * (head(Cc, -1) + tail(Cc, -1))) / 2),
       m1  = with(ss, sum(diff(time) * (head(Cc_m1, -1) + tail(Cc_m1, -1))) / 2))
}

no_abi <- auc24(sim_no_abi)
with_abi <- auc24(sim_abi)
cat(sprintf("No abi: parent AUCss = %.0f, M1 AUCss = %.0f\n",
            no_abi$par, no_abi$m1))
#> No abi: parent AUCss = 2915, M1 AUCss = 1656
cat(sprintf("With abi: parent AUCss = %.0f, M1 AUCss = %.0f\n",
            with_abi$par, with_abi$m1))
#> With abi: parent AUCss = 3565, M1 AUCss = 2675
cat(sprintf("Parent AUCss ratio (abi / no abi) = %.2f (paper: 1.22)\n",
            with_abi$par / no_abi$par))
#> Parent AUCss ratio (abi / no abi) = 1.22 (paper: 1.22)
cat(sprintf("M1 AUCss ratio (abi / no abi)     = %.2f (paper: 1.61)\n",
            with_abi$m1 / no_abi$m1))
#> M1 AUCss ratio (abi / no abi)     = 1.61 (paper: 1.61)

Assumptions and deviations

  • Joint vs separate fits. Yoshida 2021 explicitly fitted ipatasertib and M1 in TWO SEPARATE NONMEM runs (Discussion p. 305: “A joint model for ipatasertib and M1 would have been mechanistically more appropriate … but not utilised because of the execution time, computational complexity, and model instability”). The packaged model collapses both fits into a single rxode2 object for ease of joint simulation, but there is NO mechanistic fractional-conversion pathway between parent central and M1 central. Each analyte’s apparent absorption / clearance / volume parameters are the paper-reported apparent values, and the “M1 depot” is a mathematical phantom: the user supplies a second dose row that triggers the M1 sub-system, and the M1 sub-system’s apparent F absorbs the fraction-formed and first-pass-survival terms. The simulated metabolite / parent AUC ratio at steady state therefore equals the ratio of apparent F over apparent CL between the two sub-systems (= 1.331 * 162 / (1.201 * 314) = 0.572 at the typical-value reference), consistent with the paper-reported metabolic ratio (0.398 at 600 mg per the paper Methods narrative; the Discussion text’s “1.81” is the inverted ratio).

  • Time-of-first-dose convention. The model’s multiple-dose flag is computed as mdflag <- (t > 48) * 1.0. The user’s event table is therefore expected to place the FIRST oral administration at time = 0. The Yoshida 2021 paper applies the SSEFFF1 / SSEFFKA switch using the NONMEM SSFLAG column at the same 48 h threshold (paper Simulations section).

  • Apparent F’s anchored at 1. Both lfdepot and lfdepot_m1 are wrapped in fixed() because the absolute parent bioavailability and the fraction of parent metabolised to M1 are not separately identifiable from oral data alone. The IIV terms etalfdepot and etalfdepot_m1 then carry the entire between-subject variability of the apparent bioavailability and metabolite formation yield.

  • Steady state takes ~ 25-40 days to reach. Although the paper reports an effective terminal half-life of 31.9-53 h, the three-compartment model’s deep V4 / Q4 phase (V4 = 4340 L, Q4 = 3.26 L/h, hence k42 = 7.5e-4 1/h) gives a much slower asymptotic gamma half-life that means observations earlier than ~ day 25 are still on the accumulation transient. The vignette therefore simulates 30 days of QD dosing rather than the 14 days the paper used internally. The paper’s reported accumulation ratio of 1.8-2.4 at the 200-800 mg dose range (paper Introduction) is measured at the visible terminal phase rather than at full asymptotic SS, and the model reproduces this at day 14 (parent trough at day 14 ~ 95% of asymptotic SS at day 60).

  • NONMEM minimization caveat. The on-disk run230.lst reports MINIMIZATION SUCCESSFUL with the standard HOWEVER, PROBLEMS OCCURRED WITH THE MINIMIZATION caveat (NSIG = 3.3 of 3 requested; gradient values were large but the covariance step ran successfully and the FINAL PARAMETER ESTIMATE values match the published Table 3 within the precision the paper reports). The paper’s authors clearly accepted these as final estimates.

  • No NCA against published observed data. Yoshida 2021 does not publish absolute AUC / Cmax tables; the Discussion narrative reports fold-changes (1.81 metabolic ratio, 22% increase in parent AUC with abiraterone, 61% increase in M1 AUC with abiraterone, age effect of -19% to +9% across the 2.5th-97.5th percentile age range, weight effect of -22% to +32% across the 2.5th-97.5th percentile weight range for parent). The vignette’s self-consistency checks are therefore against the closed-form F*D/CL prediction and against the paper-reported fold-changes.

  • Race and renal function covariates dropped. The paper Discussion notes that Asian race appeared associated with higher exposure in preliminary analysis (likely a body-weight confound), but no race effect was retained after backward elimination; mild and moderate renal impairment and mild hepatic impairment were also tested and dropped. The packaged model therefore does NOT require RACE_, EGFR, or CONMED_ hepatic indicators as covariate columns.

  • Single-dose subjects under-predicted by ~ 25% at low dose. The paper notes (Discussion + Supplementary Figure S2) that the 25 mg and 50 mg cohorts (3 subjects each) showed biased predictions in the diagnostic plots, likely reflecting dose non-linearity at the lowest doses. The packaged structural model is the final reduced linear fit and does not encode that non-linearity. The model is therefore best applied to the clinically relevant 200-800 mg dose range.