Skip to contents

Jackson_2022_ixekizumab

Source: vignettes/articles/Jackson_2022_ixekizumab.Rmd
Jackson_2022_ixekizumab.Rmd
library(nlmixr2lib)
library(rxode2)
#> rxode2 5.0.2 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)
library(PKNCA)
#> 
#> Attaching package: 'PKNCA'
#> The following object is masked from 'package:stats':
#> 
#>     filter

Ixekizumab paediatric population PK (IXORA-PEDS)

Simulate ixekizumab serum concentration–time profiles using the final paediatric population PK model of Jackson et al. (2022) for patients aged 6 to <18 years with moderate-to-severe plaque psoriasis enrolled in IXORA-PEDS (NCT03073200). The model is a two-compartment linear PK model with first-order subcutaneous absorption, an allometric body-weight effect on CL, Q, V2, and V3, and a log-linear ADA-titre effect on CL.

  • Citation: Jackson K, Chua L, Velez de Mendizabal N, et al. Population pharmacokinetic and exposure-efficacy analysis of ixekizumab in paediatric patients with moderate-to-severe plaque psoriasis (IXORA-PEDS). Br J Clin Pharmacol. 2022;88(3):1074-1086. doi:10.1111/bcp.15034
  • Article: https://doi.org/10.1111/bcp.15034

Population

IXORA-PEDS is a Phase 3, double-blind, placebo-controlled trial conducted in paediatric patients aged 6 to <18 years with moderate-to-severe plaque psoriasis (PASI ≥ 12, sPGA ≥ 3, BSA ≥ 10% at screening and baseline). The population PK analysis dataset comprised 558 measurable ixekizumab serum concentrations from 184 patients: 4 patients <25 kg (2.2%), 45 patients 25–50 kg (24.5%), and 135 patients >50 kg (73.4%). Overall baseline age ranged 6–17 years (median 15 years in the >50 kg group, 10 years in the 25–50 kg group, 7 years in the <25 kg group), baseline weight 21.5–136 kg, 56.5% female. The majority were white (81.5%), with 3.26% Black/African American, 3.26% Asian, 1.63% American Indian or Alaska Native, 7.61% Other, and 2.72% missing race. Regions: US 37.5%, Europe 41.3%, Rest of World 21.2%. Baseline demographics reproduced from Jackson 2022 Table 1. The approved weight-based Q4W regimens are 20 mg (<25 kg), 40 mg (25–50 kg), and 80 mg (>50 kg), after an initial 40/80/160 mg loading dose respectively.

Programmatic access: readModelDb("Jackson_2022_ixekizumab")$population.

Source trace

Equation / parameter Value Source location
Two-compartment linear model with first-order SC absorption structural Jackson 2022 Methods section 2.5 and Table 2 header
Ka 0.00801 h-1 Table 2 (Rate of absorption)
CL (reference 58.6 kg, ADA-negative) 0.0120 L/h Table 2 (Clearance)
Q (reference 58.6 kg) 0.0119 L/h Table 2 (Clearance)
V2 (reference 58.6 kg) 2.72 L Table 2 (Volume of distribution)
V3 (reference 58.6 kg) 2.11 L Table 2 (Volume of distribution)
F1 0.72 (FIXED) Table 2 + footnote c (fixed to adult model value)
Allometric exponent on CL and Q 0.989 Table 2 (Covariate weight on CL and Q)
Allometric exponent on V2 and V3 0.998 Table 2 (Covariate weight on V2 and V3)
ADA-titre coefficient on CL 0.0292 Table 2 (Covariate ADA titre on CL)
CL covariate equation CL*(WT/58.6)^0.989*(1 + 0.0292*log_e[ADA titre]) Table 2 footnote (CL_individual equation)
Q covariate equation Q*(WT/58.6)^0.989 Table 2 footnote (Q_individual equation)
V2 covariate equation V2*(WT/58.6)^0.998 Table 2 footnote (V2,individual equation)
V3 covariate equation V3*(WT/58.6)^0.998 Table 2 footnote (V3,individual equation)
IIV on CL 28.4% (ω² = log(1 + 0.284²) = 0.0776) Table 2 (Interindividual variability column) + footnote b (%IIV = 100·√(e^OMEGA − 1))
Residual error (proportional) 27.7% Table 2 (Residual error (proportional))
Approved Q4W weight-based dosing (20/40/80 mg with loading 40/80/160 mg) regimen Introduction, page 1075, and Methods section 2.1
Reported steady-state (Week 8 and Week 12) mean trough concentrations 3.20–3.33 µg/mL (25–50 and >50 kg groups) Results section 3.1, first paragraph
Reported normalized 70-kg CL / Vss 0.0144 L/h / 5.77 L Discussion, page 1083
Reported mean terminal half-life ~12 days Discussion, page 1083

Covariate column naming

Source column Canonical column used here
WT (baseline body weight, kg) WT
ADA titre (continuous reciprocal dilution; 1:10, 1:20, …, 1:2560) ADA_TITER (per inst/references/covariate-columns.md). ADA-negative samples encoded as ADA_TITER = 1 so that log_e(1) = 0 cancels the covariate effect (85.8% of the dataset).

Virtual cohort

The IXORA-PEDS patient-level data are not publicly available, so the figures below use a virtual paediatric cohort whose weight, weight-group, and ADA distributions approximate Jackson 2022 Table 1 and the ADA incidence reported in Results section 3.1.

set.seed(20220301) # IXORA-PEDS paper month
n_subj <- 184 # match dataset size

# Weight-group proportions from Jackson 2022 Table 1 (overall column)
# <25 kg: 4/184 = 2.2%; 25-50 kg: 45/184 = 24.5%; >50 kg: 135/184 = 73.4%
u <- runif(n_subj)
weight_group <- cut(
  u,
  breaks = c(-Inf, 4 / 184, 49 / 184, Inf),
  labels = c("<25 kg", "25-50 kg", ">50 kg"),
  right  = TRUE
)

# Sample body weights inside each group to match the reported mean +/- SD
# from Jackson 2022 Table 1. Clip to the reported per-group ranges.
simulate_weight <- function(group) {
  switch(as.character(group),
    "<25 kg"   = pmin(22.6, pmax(21.5, rnorm(1, 21.9, 0.520))),
    "25-50 kg" = pmin(50.0, pmax(25.0, rnorm(1, 37.7, 8.10))),
    ">50 kg"   = pmin(136.0, pmax(50.1, rnorm(1, 71.3, 19.5)))
  )
}
WT <- vapply(weight_group, simulate_weight, numeric(1))

# ADA titre: 85.8% ADA-negative (ADA_TITER = 1); 14.2% TE-ADA-positive with
# titres in three bands per Results section 3.1:
#   - Low    (<1:160)        : 52.6% of ADA+ samples
#   - Moderate (>=1:160, <1:1280): 38.5%
#   - High     (>=1:1280)    :  9.0%
ada_pos <- runif(n_subj) < 0.142
ada_titre <- rep(1, n_subj)
n_pos <- sum(ada_pos)
if (n_pos > 0) {
  band <- sample(c("low", "mod", "high"),
    size    = n_pos,
    replace = TRUE,
    prob    = c(0.526, 0.385, 0.090)
  )
  titre_sample <- function(b) {
    switch(b,
      low  = sample(c(10, 20, 40, 80), 1),          # 1:10 to <1:160
      mod  = sample(c(160, 320, 640), 1),           # 1:160 to <1:1280
      high = sample(c(1280, 2560), 1)               # >=1:1280
    )
  }
  ada_titre[ada_pos] <- vapply(band, titre_sample, numeric(1))
}

cohort <- tibble(
  ID           = seq_len(n_subj),
  weight_group = weight_group,
  WT           = WT,
  ADA_TITER    = ada_titre
)

summary_tbl <- cohort |>
  group_by(weight_group) |>
  summarise(
    n           = dplyr::n(),
    mean_WT     = mean(WT),
    median_WT   = median(WT),
    pct_ada_pos = 100 * mean(ADA_TITER > 1),
    .groups     = "drop"
  )
knitr::kable(summary_tbl, digits = 2,
             caption = "Simulated cohort summary (compare with Jackson 2022 Table 1 weight-group means and Results 3.1 ADA incidence).")
Simulated cohort summary (compare with Jackson 2022 Table 1 weight-group means and Results 3.1 ADA incidence).
weight_group n mean_WT median_WT pct_ada_pos
<25 kg 9 21.91 21.86 22.22
25-50 kg 43 38.99 40.04 11.63
>50 kg 132 74.64 73.96 12.12

Dosing dataset

Weight-based Q4W regimen per IXORA-PEDS: loading dose at Week 0 (40/80/160 mg for <25/25–50/>50 kg), then maintenance Q4W (20/40/80 mg). Simulate through Week 24 (6 maintenance doses) so that the trough at Week 12 (the co-primary endpoint timepoint) reflects steady state. Times are in hours (model unit).

# Weeks -> hours conversion
hours_per_week <- 7 * 24

maint_doses <- function(group) {
  switch(as.character(group), "<25 kg" = 20, "25-50 kg" = 40, ">50 kg" = 80)
}
load_doses <- function(group) {
  switch(as.character(group), "<25 kg" = 40, "25-50 kg" = 80, ">50 kg" = 160)
}

dose_weeks    <- c(0, 4, 8, 12, 16, 20)
dose_times_hr <- dose_weeks * hours_per_week
obs_weeks     <- sort(unique(c(seq(0, 24, by = 0.5), 1, 9))) # dense early + the Week 1 and Week 9 PK addendum timepoints
obs_times_hr  <- obs_weeks * hours_per_week

d_dose <- cohort |>
  tidyr::crossing(dose_idx = seq_along(dose_times_hr)) |>
  mutate(
    TIME = dose_times_hr[dose_idx],
    AMT  = ifelse(dose_idx == 1,
                  vapply(weight_group, load_doses, numeric(1)),
                  vapply(weight_group, maint_doses, numeric(1))),
    EVID = 1,
    CMT  = "depot",
    DV   = NA_real_
  ) |>
  select(-dose_idx)

d_obs <- cohort |>
  tidyr::crossing(TIME = obs_times_hr) |>
  mutate(
    AMT  = 0,
    EVID = 0,
    CMT  = "central",
    DV   = NA_real_
  )

d_sim <- bind_rows(d_dose, d_obs) |>
  arrange(ID, TIME, desc(EVID)) |>
  as.data.frame()

Simulation 

mod <- readModelDb("Jackson_2022_ixekizumab")
set.seed(42)
sim <- rxode2::rxSolve(mod, events = d_sim, returnType = "data.frame")
#> ℹ parameter labels from comments will be replaced by 'label()'

Replicate published figures

Figure 1 — Mean ixekizumab concentration vs time by weight group (Weeks 0–12)

Jackson 2022 Figure 1 plots mean (±SD) trough-only observed concentrations in the 25–50 kg and >50 kg groups through Week 12. The simulated curves below show the mean (and 5th/95th percentile ribbon) of simulated ixekizumab concentrations over the same window, stratified by baseline weight group.

sim_plot <- sim |>
  as.data.frame() |>
  left_join(cohort |> select(ID, weight_group), by = c("id" = "ID")) |>
  filter(time > 0, time <= 12 * hours_per_week, !is.na(Cc)) |>
  mutate(week = time / hours_per_week)

fig1 <- sim_plot |>
  group_by(weight_group, week) |>
  summarise(
    mean_Cc = mean(Cc, na.rm = TRUE),
    Q05     = quantile(Cc, 0.05, na.rm = TRUE),
    Q95     = quantile(Cc, 0.95, na.rm = TRUE),
    .groups = "drop"
  )

ggplot(fig1, aes(x = week, y = mean_Cc, colour = weight_group, fill = weight_group)) +
  geom_ribbon(aes(ymin = Q05, ymax = Q95), alpha = 0.15, colour = NA) +
  geom_line(linewidth = 0.9) +
  scale_y_continuous() +
  scale_x_continuous(breaks = seq(0, 12, by = 4)) +
  labs(
    x = "Time (weeks)",
    y = "Ixekizumab concentration (\u03bcg/mL)",
    title = "Figure 1 replication: mean concentration by weight group",
    subtitle = "Simulated Weeks 0\u201312 (loading + Q4W maintenance); mean and 5\u201395th percentile",
    caption = "Replicates Figure 1 of Jackson 2022."
  ) +
  theme_bw()

Figure 3A — Week 12 trough concentration vs body weight 

wk12_time <- 12 * hours_per_week
sim_wk12 <- sim |>
  as.data.frame() |>
  filter(abs(time - wk12_time) < 1e-6) |>
  left_join(cohort |> select(ID, weight_group, WT), by = c("id" = "ID"))

ggplot(sim_wk12, aes(x = WT, y = Cc, colour = weight_group)) +
  geom_point(alpha = 0.5) +
  scale_y_log10() +
  labs(
    x = "Baseline body weight (kg)",
    y = "Week 12 trough concentration (\u03bcg/mL)",
    title = "Figure 3A replication: Week 12 model-predicted trough vs body weight",
    caption = "Replicates Figure 3A of Jackson 2022."
  ) +
  theme_bw()

Steady-state trough by weight group (Weeks 8 and 12)

Jackson 2022 Results section 3.1 reports mean trough concentrations at predicted steady state (Weeks 8 and 12) in the 25–50 kg and >50 kg groups in the range 3.20–3.33 µg/mL. Insufficient PK data were available to summarise the <25 kg group.

trough_weeks <- c(8, 12)
trough_times <- trough_weeks * hours_per_week

trough_tbl <- sim |>
  as.data.frame() |>
  filter(time %in% trough_times) |>
  left_join(cohort |> select(ID, weight_group), by = c("id" = "ID")) |>
  group_by(weight_group, week = time / hours_per_week) |>
  summarise(
    mean_trough  = mean(Cc, na.rm = TRUE),
    sd_trough    = sd(Cc, na.rm = TRUE),
    median_trough = median(Cc, na.rm = TRUE),
    .groups      = "drop"
  )

knitr::kable(trough_tbl, digits = 2,
             caption = paste(
               "Simulated mean trough concentrations at Weeks 8 and 12.",
               "Compare with Jackson 2022 Results 3.1: reported means 3.20\u20133.33 \u03bcg/mL",
               "in the 25\u201350 kg and >50 kg groups; insufficient data in the <25 kg group."
             ))
Simulated mean trough concentrations at Weeks 8 and 12. Compare with Jackson 2022 Results 3.1: reported means 3.20–3.33 μg/mL in the 25–50 kg and >50 kg groups; insufficient data in the <25 kg group.
weight_group week mean_trough sd_trough median_trough
<25 kg 8 3.27 1.07 3.39
<25 kg 12 3.04 1.04 3.14
25-50 kg 8 3.11 1.33 2.98
25-50 kg 12 2.88 1.26 2.76
>50 kg 8 3.27 1.60 2.90
>50 kg 12 3.03 1.52 2.66

PKNCA validation

Run PKNCA on the steady-state dosing interval between the Week 8 and Week 12 doses (i.e., from the Week 8 dose to the Week 12 dose, a 4-week window). This is the interval over which Jackson 2022 reports steady-state trough concentrations.

ss_start <- 8  * hours_per_week
ss_end   <- 12 * hours_per_week

nca_conc <- sim |>
  as.data.frame() |>
  filter(time >= ss_start, time <= ss_end, Cc > 0) |>
  left_join(cohort |> select(ID, weight_group), by = c("id" = "ID")) |>
  mutate(time_rel = time - ss_start,
         treatment = paste0("IXE_Q4W_", weight_group)) |>
  rename(ID = id) |>
  select(ID, time_rel, Cc, treatment)

nca_dose <- cohort |>
  mutate(time_rel = 0,
         AMT      = vapply(weight_group, maint_doses, numeric(1)),
         treatment = paste0("IXE_Q4W_", weight_group)) |>
  select(ID, time_rel, AMT, treatment)

conc_obj <- PKNCAconc(nca_conc, Cc ~ time_rel | treatment + ID)
dose_obj <- PKNCAdose(nca_dose, AMT ~ time_rel | treatment + ID)

data_obj <- PKNCAdata(
  conc_obj,
  dose_obj,
  intervals = data.frame(
    start   = 0,
    end     = 4 * hours_per_week,
    cmax    = TRUE,
    tmax    = TRUE,
    cmin    = TRUE,
    auclast = TRUE,
    cav     = TRUE
  )
)

nca_results <- pk.nca(data_obj)
nca_summary <- summary(nca_results)
knitr::kable(
  nca_summary,
  digits  = 2,
  caption = paste(
    "Simulated steady-state (Week 8\u2013Week 12) NCA by weight group.",
    "Cmin is the trough at Week 12; compare with Jackson 2022 Results 3.1:",
    "reported mean trough 3.20\u20133.33 \u03bcg/mL in the 25\u201350 and >50 kg groups."
  )
)
Simulated steady-state (Week 8–Week 12) NCA by weight group. Cmin is the trough at Week 12; compare with Jackson 2022 Results 3.1: reported mean trough 3.20–3.33 μg/mL in the 25–50 and >50 kg groups.
start end treatment N auclast cmax cmin tmax cav
0 672 IXE_Q4W_<25 kg 9 3620 [23.1] 7.77 [16.3] 2.88 [37.8] 168 [84.0, 168] 5.39 [23.1]
0 672 IXE_Q4W_>50 kg 132 3740 [37.0] 8.43 [30.6] 2.69 [53.2] 168 [84.0, 168] 5.57 [37.0]
0 672 IXE_Q4W_25-50 kg 43 3600 [31.9] 8.08 [25.3] 2.61 [49.2] 168 [84.0, 168] 5.35 [31.9]

Comparison against published troughs 

pub <- tibble(
  weight_group = c("<25 kg", "25-50 kg", ">50 kg"),
  published_mean_trough = c(
    NA_real_,  # insufficient data in <25 kg group
    mean(c(3.20, 3.33)),
    mean(c(3.20, 3.33))
  ),
  published_range = c(
    "Insufficient data",
    "3.20-3.33",
    "3.20-3.33"
  )
)
sim_trough <- trough_tbl |>
  filter(week == 12) |>
  select(weight_group, simulated_mean_trough = mean_trough)

comparison <- pub |>
  left_join(sim_trough, by = "weight_group") |>
  mutate(pct_diff = 100 * (simulated_mean_trough - published_mean_trough) / published_mean_trough)

knitr::kable(comparison, digits = 2,
             caption = "Week 12 mean trough: simulated vs Jackson 2022 Results 3.1.")
Week 12 mean trough: simulated vs Jackson 2022 Results 3.1.
weight_group published_mean_trough published_range simulated_mean_trough pct_diff
<25 kg NA Insufficient data 3.04 NA
25-50 kg 3.27 3.20-3.33 2.88 -11.79
>50 kg 3.27 3.20-3.33 3.03 -7.09

Simulated Week 12 mean troughs track the published 3.20–3.33 µg/mL range within a reasonable margin; any residual difference is driven by differences between the virtual cohort’s weight distribution and the actual IXORA-PEDS cohort.

Assumptions and deviations

  • Patient-level data unavailable. Weight distributions within each group are drawn from the Jackson 2022 Table 1 mean ± SD and clipped to the reported range; actual IXORA-PEDS per-subject weights are not published.
  • ADA-titre sampling. 14.2% of patients flagged TE-ADA positive per Results section 3.1; the per-band split (52.6% low / 38.5% moderate / 9.0% high titre) applies the paper’s band-level percentages uniformly at the patient level, whereas the paper reports them at the sample level. ADA-negative patients (85.8%) carry ADA_TITER = 1 so the log-linear effect is zero.
  • Time-varying ADA not modelled. The paper uses time-matched ADA titre but does not publish per-subject longitudinal trajectories; each virtual subject carries a single titre for the full follow-up.
  • Baseline weight only. Jackson 2022 used baseline weight (mean percent change 1.6% over the first 12 weeks; 23% had ≥±10% change over 108 weeks). This vignette uses a single baseline weight per subject, matching the paper.
  • Residual error only on the central compartment. The paper reports one proportional residual error (27.7%); additive residual error is not reported and is not added here.
  • Site of injection not modelled. The paper evaluated injection site as a covariate but did not retain it in the final paediatric model.
  • Dosing regimen. IXORA-PEDS used the approved Q4W regimens (loading dose followed by Q4W maintenance). Ongoing open-label periods beyond Week 12 are not simulated beyond Week 24.

Reference

  • Jackson K, Chua L, Velez de Mendizabal N, et al. Population pharmacokinetic and exposure-efficacy analysis of ixekizumab in paediatric patients with moderate-to-severe plaque psoriasis (IXORA-PEDS). Br J Clin Pharmacol. 2022;88(3):1074-1086. doi:10.1111/bcp.15034