Skip to contents

Anti-tryptase MTPS9579A (Rymut 2023)

Source: vignettes/articles/Rymut_2023_anti_tryptase.Rmd
Rymut_2023_anti_tryptase.Rmd
library(nlmixr2lib)
library(rxode2)
#> rxode2 5.1.2 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

MTPS9579A serum + airway mechanistic PK/PD model

MTPS9579A is a humanized IgG4 monoclonal antibody that inhibits the serine protease tryptase by binding both monomeric tryptase (the predominant form in serum) and active tetrameric tryptase (the predominant form in the airway) with sub-nanomolar affinity and rapidly disrupting the active tetramer into inactive monomers. Rymut et al. (2023) developed a two-layer mechanistic population PK/PD model from a Phase 1 healthy-subject study (n = 106) and applied it prospectively to forecast active-tryptase suppression in adults with moderate-to-severe asthma:

  1. A systemic two-compartment quasi-equilibrium (QE) target-mediated drug-disposition (TMDD) model in NONMEM 7.4.3 (SAEM) fits the serum MTPS9579A and serum total-tryptase data jointly with body weight as an allometric covariate on linear CL and central volume (Table 1, Text S1; OFV = 7593.39).
  2. A mechanistic airway-tissue compartment layered on top of the systemic model receives free mAb and mAb-monomer complex from circulation via lymph flow with vascular reflection coefficients (Text S2 mrgsolve). In the airway interstitial fluid (ISF), tryptase is secreted as the active tetramer, spontaneously dissociates into inactive monomers, and is rapidly disrupted when bound by MTPS9579A (kbreak); free mAb binds both tetramer and monomer with the same KD.

The packaged model combines both layers into a single nine-state ODE system with the systemic parameters estimated and the airway parameters fixed from physiological literature and a healthy-subject visual fit of the upper-airway biodistribution coefficient at 3% (Methods + Figure 4 + Figure S2).

  • Citation: Rymut SM, Henderson LM, Poon V, Staton TL, Cai F, Sukumaran S, Rhee H, Owen R, Ramanujan S, Yoshida K. A mechanistic PK/PD model to enable dose selection of the potent anti-tryptase antibody (MTPS9579A) in patients with moderate-to-severe asthma. Clin Transl Sci. 2023;16(4):694-703. doi:10.1111/cts.13483
  • Article: https://doi.org/10.1111/cts.13483

Population

The model-building dataset is a Phase 1 single + multiple ascending dose study of MTPS9579A in 106 healthy adults (Methods):

  • Single-ascending-dose (SAD) cohorts: 30, 100, 300 mg SC and 300, 900, 1800, 3600 mg IV (single dose).
  • Multiple-ascending-dose (MAD) cohorts: 150, 300, 750 mg SC and 1350, 1800, 3600 mg IV every 4 weeks (q4w).
  • Body weight: 49.9-113.3 kg (SAD) and 47.1-97.7 kg (MAD); reference 70 kg in the allometric model.

Prospective simulations of moderate-to-severe asthma used an independent observational cohort of n = 15 adults (mean age 42.7 years, 100% White, 26.7% female; Table S1) whose nasosorption samples established the elevated baseline-tryptase constants (serum total tryptase 10 ng/mL vs 7 ng/mL in healthy; nasal active tryptase 4 ng/mL vs 0.4 ng/mL in healthy; nasal total tryptase 140 ng/mL vs 12 ng/mL in healthy; Figure 5). These observational data did not contribute to parameter estimation.

The same metadata is available programmatically through readModelDb("Rymut_2023_anti_tryptase")$population.

Source trace

Per-parameter origin is recorded as in-file comments next to each ini() entry in inst/modeldb/specificDrugs/Rymut_2023_anti_tryptase.R. The table collects them for review.

Equation / parameter Value (paper) Value (file) Source
lka (SC absorption rate) 0.239 1/day log(0.239) 1/day Table 1
lcl (linear clearance at 70 kg) 0.128 L/day log(0.128) L/day Table 1
lvc (central volume V2 at 70 kg) 3.33 L log(3.33) L Table 1
lq (inter-compartmental clearance) 0.408 L/day log(0.408) L/day Table 1
lvp (peripheral volume V3) 2.28 L log(2.28) L Table 1
lfdepot (SC bioavailability Fsc) 0.661 log(0.661) Table 1
lkss (QE dissociation constant Kss = KD) 0.0448 nM log(0.0448) nM Table 1
lbase (baseline total serum tryptase) 0.223 nM log(0.223) nM Table 1
lkdeg (tryptase degradation rate) 20.7 1/day log(20.7) 1/day Table 1
lclint (internalisation clearance of mAb-tryptase complex) 0.398 L/day log(0.398) L/day Table 1
e_wt_cl (allometric exponent of WT/70 on CL) 0.820 (RSE 23%) 0.820 Table 1
e_wt_vc (allometric exponent of WT/70 on Vc) 0.808 (RSE 18%) 0.808 Table 1
etalka omega^2 = 0.214 0.214 Table 1
etalcl + etalvc BLOCK(2) var_CL = 0.0809, cov = 0.0342, var_Vc = 0.0387 c(0.0809, 0.0342, 0.0387) Table 1 + Text S1
etalbase omega^2 = 0.197 0.197 Table 1
etalkdeg omega^2 = 0.103 0.103 Table 1
etalclint omega^2 = 0.135 0.135 Table 1
propSd (proportional residual on serum MTPS9579A) 0.103 0.103 Table 1
addSd (additive residual on serum MTPS9579A; thetarized with SIGMA = 1) 2.72 nM (NONMEM units) 2.72 * 155 / 1000 = 0.4216 ug/mL Table 1 (unit-converted)
propSd_TotalSerumTryptase 0.245 0.245 Table 1
kp_free / kp_bound (biodistribution coefficient to airway ISF) 3% (visual best fit) fixed(0.03) Methods + Figure 4 + Figure S2
refl_lymph (lymphatic reflection coefficient) 0.2 fixed(0.2) Table S2 (Shah and Betts 2012)
klf_tissue (ISF lymph turnover) 1.2 1/h fixed(1.2 * 24) 1/day Table S2; Methods (0.2% of 182 L/h / 0.3 L)
bl_tryptase_isf (baseline ISF total tryptase, healthy) 40 ng/mL fixed(40) ng/mL Table S2
kel_tryp_isf (tryptase elim rate; t1/2 = 2 h) 8.31 1/day fixed(log(2)/(2/24)) Methods + Table S2
kdiss_tet_isf (spontaneous tetramer dissociation; t1/2 = 30 min) 33 1/day fixed(log(2)/(0.5/24)) Methods (also Table S2 by terminal half-life)
sum_kel_isf / f_tet_isf (precomputed kel + kdiss and kel/(kel+kdiss)) derived fixed (numeric) Rate-balance partition (Text S2)
kbreak_tet (mAb-induced tetramer disruption; t1/2 = 1 min) 1000 1/day fixed(1000) Methods + Table S2
kel_mono_ab (monomer-mAb complex elim; t1/2 = 100 h) log(2)/(100/24) 1/day fixed(log(2)/(100/24)) Table S2 (assumed)
kon_isf (mAb-tryptase association rate) 7.62e5 1/M/s fixed(7.62e5/1e9*86400) 1/nM/day Table S2 (in vitro)
kd_isf (equilibrium dissociation constant in ISF) 0.0448 nM (= serum Kss) fixed(0.0448) Table S2 (4.88e-11 M) and Table 1
mw_ab / mw_mono / mw_tet 155 / 32 / 128 ug/nmol fixed Text S2 mrgsolve constants
Eq. d/dt(central) = ka * depot + Q * (peripheral1/Vp) - Q * cfree - CL * cfree - CLint * total_target * cfree / (Kss + cfree) n/a n/a Text S1 NONMEM $DES
Eq. d/dt(total_target) = ksyn - kdeg * total_target - (CLint/Vc - kdeg) * cfree * total_target / (Kss + cfree) n/a n/a Text S1 NONMEM $DES
Eq. d/dt(mab_isf) = klf * (1 - refl_free) * cfree - klf * (1 - refl_lymph) * mab_isf - kon * mab_isf * (target_isf + monomer_isf) + koff * (complex_isf + complex_monomer_isf) n/a n/a Text S2 mrgsolve $ODE
Eq. d/dt(target_isf) = vin_tryp - kel * target_isf - kdiss * target_isf - kon * mab_isf * target_isf + koff * complex_isf n/a n/a Text S2 mrgsolve $ODE
Eq. d/dt(complex_isf) = kon * mab_isf * target_isf - koff * complex_isf - kbreak * complex_isf n/a n/a Text S2 mrgsolve $ODE
Eq. d/dt(monomer_isf) = -kel * monomer_isf - kon * mab_isf * monomer_isf + koff * complex_monomer_isf + 3 * kbreak * complex_isf + 4 * kdiss * target_isf n/a n/a Text S2 mrgsolve $ODE
Eq. d/dt(complex_monomer_isf) = kon * mab_isf * monomer_isf - koff * complex_monomer_isf + kbreak * complex_isf - kel_mono_ab * complex_monomer_isf + klf * (1 - refl_bound) * cbound - klf * (1 - refl_lymph) * complex_monomer_isf n/a n/a Text S2 mrgsolve $ODE
QE-TMDD free drug quadratic: cfree = 1/2 * ((ctot - total_target - Kss) + sqrt((ctot - total_target - Kss)^2 + 4 * Kss * ctot)) n/a n/a Text S1 NONMEM $DES

Virtual cohort and simulation

The published Phase 1 subject-level concentration data are not publicly available; the simulation below reproduces the typical-value PK/PD trajectories of Figure 2 (serum) and Figure 3 (airway active tryptase) at the body-weight reference (70 kg). Between-subject variability is zeroed out via rxode2::zeroRe() so each dose level is represented by its typical-value response.

mod <- rxode2::rxode(readModelDb("Rymut_2023_anti_tryptase"))
#> ℹ parameter labels from comments will be replaced by 'label()'
mod_t <- rxode2::zeroRe(mod)
# Helper: build a single-cohort event table (one dose level, one route).
# Dosing into cmt = "depot" for SC; cmt = "central" for IV. Sample times
# in days. WT is supplied as a covariate column.
make_cohort <- function(dose_mg, route = c("SC", "IV"), n_doses = 1L,
                        ii_days = 28, sample_grid,
                        wt_kg = 70, id_offset = 0L) {
  route <- match.arg(route)
  cmt_dose <- if (route == "SC") "depot" else "central"
  dose_times <- seq(0, by = ii_days, length.out = n_doses)
  dose_rows <- data.frame(
    id   = id_offset + 1L,
    time = dose_times,
    amt  = dose_mg,
    evid = 1,
    cmt  = cmt_dose,
    WT   = wt_kg,
    dose_mg  = dose_mg,
    route    = route,
    n_doses  = n_doses,
    treatment = sprintf("%d mg %s%s", dose_mg, route,
                        if (n_doses > 1) " q4w" else "")
  )
  obs_rows <- data.frame(
    id   = id_offset + 1L,
    time = sample_grid,
    amt  = 0,
    evid = 0,
    cmt  = "Cc",
    WT   = wt_kg,
    dose_mg  = dose_mg,
    route    = route,
    n_doses  = n_doses,
    treatment = sprintf("%d mg %s%s", dose_mg, route,
                        if (n_doses > 1) " q4w" else "")
  )
  dplyr::bind_rows(dose_rows, obs_rows)
}

Single-dose SAD simulation (Figure 2a + Figure 3a) 

sad_doses_sc <- list(c(30, "SC"), c(100, "SC"), c(300, "SC"))
sad_doses_iv <- list(c(300, "IV"), c(900, "IV"), c(1800, "IV"), c(3600, "IV"))
sad_doses    <- c(sad_doses_sc, sad_doses_iv)

# Observation grid: dense up to 35 days post first dose to mirror Figure 2a/3a
sad_grid <- c(seq(0.01, 2, by = 0.1), seq(2.5, 35, by = 0.5))

sad_events <- dplyr::bind_rows(lapply(seq_along(sad_doses), function(i) {
  d <- sad_doses[[i]]
  make_cohort(dose_mg = as.numeric(d[1]), route = d[2], n_doses = 1L,
              sample_grid = sad_grid, id_offset = (i - 1L) * 1L)
}))

# rxSolve treats id as the subject key; the multi-cohort id_offset above
# ensures distinct subject ids per dose group so concentrations are not
# inadvertently summed.
stopifnot(!anyDuplicated(unique(sad_events[, c("id", "time", "evid")])))

sad_sim <- rxode2::rxSolve(
  mod_t, events = sad_events,
  keep = c("dose_mg", "route", "treatment")
) |> as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalka', 'etalcl', 'etalvc', 'etalbase', 'etalkdeg', 'etalclint'
#> Warning: multi-subject simulation without without 'omega'

# Serum MTPS9579A (Figure 2a). Use log y to match the published scale.
ggplot(sad_sim,
       aes(time, pmax(Cc, 1e-3), colour = treatment)) +
  geom_line() +
  scale_y_log10() +
  facet_wrap(~ route, scales = "free_y") +
  labs(x = "Time (day)", y = "Serum MTPS9579A (ug/mL, log scale)",
       title = "Figure 2a equivalent: SAD serum profiles by dose",
       colour = "Dose",
       caption = "Typical-value (no IIV) PK simulation; 70 kg reference subject. Replicates the shape of Rymut 2023 Figure 2a.")

# Active airway tryptase (Figure 3a) -- relative to baseline.
ggplot(sad_sim,
       aes(time, ActiveAirwayTryptase, colour = treatment)) +
  geom_line() +
  facet_wrap(~ route) +
  labs(x = "Time (day)", y = "Active airway tryptase (fraction of baseline)",
       title = "Figure 3a equivalent: SAD active-tryptase suppression in airway ISF",
       colour = "Dose",
       caption = "Kp = 3% biodistribution to airway ISF (Methods). Lower values indicate stronger inhibition; 1.0 = no inhibition.")

Multiple-dose MAD simulation (Figure 2b + Figure 3b) 

mad_doses <- list(
  c(150,  "SC"), c(300,  "SC"), c(750,  "SC"),
  c(1350, "IV"), c(1800, "IV"), c(3600, "IV")
)

# Five q4w doses over 140 days, dense sampling to capture the full multiple-
# dose accumulation envelope used in Rymut 2023 Figure 2b / Figure 3b.
mad_grid <- c(seq(0.1, 7, by = 0.5), seq(8, 140, by = 1))

mad_events <- dplyr::bind_rows(lapply(seq_along(mad_doses), function(i) {
  d <- mad_doses[[i]]
  make_cohort(dose_mg = as.numeric(d[1]), route = d[2], n_doses = 5L,
              ii_days = 28, sample_grid = mad_grid,
              id_offset = (i - 1L) * 1L)
}))
stopifnot(!anyDuplicated(unique(mad_events[, c("id", "time", "evid")])))

mad_sim <- rxode2::rxSolve(
  mod_t, events = mad_events,
  keep = c("dose_mg", "route", "treatment")
) |> as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalka', 'etalcl', 'etalvc', 'etalbase', 'etalkdeg', 'etalclint'
#> Warning: multi-subject simulation without without 'omega'

ggplot(mad_sim,
       aes(time, pmax(Cc, 1e-3), colour = treatment)) +
  geom_line() +
  scale_y_log10() +
  facet_wrap(~ route, scales = "free_y") +
  labs(x = "Time (day)", y = "Serum MTPS9579A (ug/mL, log scale)",
       title = "Figure 2b equivalent: MAD serum profiles by dose",
       colour = "Dose",
       caption = "Five q4w doses; 70 kg reference subject. Replicates the shape of Rymut 2023 Figure 2b.")

ggplot(mad_sim,
       aes(time, ActiveAirwayTryptase, colour = treatment)) +
  geom_line() +
  facet_wrap(~ route) +
  labs(x = "Time (day)", y = "Active airway tryptase (fraction of baseline)",
       title = "Figure 3b equivalent: MAD active-tryptase suppression in airway ISF",
       colour = "Dose",
       caption = "Higher / more frequent dosing achieves sustained near-complete inhibition (< 5% remaining).")

Comparison against published Table 2

Rymut 2023 Table 2 reports model-based percentage inhibition of active airway tryptase at the steady-state trough concentration for adults with moderate-to-severe asthma. The asthma scenario differs from the healthy fit in two ways: the baseline ISF total tryptase is 3.5x larger (40 ng/mL healthy -> 140 ng/mL asthma per Figure 5b) and the baseline serum tryptase is 1.5x larger (0.223 nM -> 0.335 nM per Figure 5c, Methods). Both are overrideable parameters on the loaded model.

# Override baseline tryptase to the asthma values per Methods + Figure 5b/5c.
# `bl_tryptase_isf` is the airway baseline (ng/mL). `lbase` is the systemic
# baseline (log nM); 1.5x = log(0.223 * 1.5) = log(0.3345). rxSolve replaces
# the entire parameter vector when `params` is supplied, so we start from the
# typical-value theta of mod_t and overwrite only the two asthma-specific
# baselines.
asthma_pars <- mod_t$theta
asthma_pars["bl_tryptase_isf"] <- 140
asthma_pars["lbase"]           <- log(0.223 * 1.5)

# Build a single 5-dose q4w regimen per Table 2 row; sample trough only
# at the start of cycle 5 (day 112, just before the cycle-5 dose).
table2_doses <- list(
  c(300,  "SC"), c(600,  "SC"),
  c(900,  "IV"), c(1800, "IV"), c(3600, "IV")
)

trough_grid <- c(seq(0.1, 7, by = 1), seq(28, 140, by = 1))
table2_events <- dplyr::bind_rows(lapply(seq_along(table2_doses), function(i) {
  d <- table2_doses[[i]]
  make_cohort(dose_mg = as.numeric(d[1]), route = d[2], n_doses = 5L,
              ii_days = 28, sample_grid = trough_grid,
              id_offset = (i - 1L) * 1L)
}))
stopifnot(!anyDuplicated(unique(table2_events[, c("id", "time", "evid")])))

# Apply parameter overrides through rxSolve(..., params = ...).
table2_sim <- rxode2::rxSolve(
  mod_t, events = table2_events,
  params = asthma_pars,
  keep = c("dose_mg", "route", "treatment")
) |> as.data.frame()
#> ℹ omega/sigma items treated as zero: 'etalka', 'etalcl', 'etalvc', 'etalbase', 'etalkdeg', 'etalclint'
#> Warning: multi-subject simulation without without 'omega'

# Trough = lowest ActiveAirwayTryptase in the final dosing interval (days 112-140).
trough <- table2_sim |>
  dplyr::filter(time >= 112, time <= 140) |>
  dplyr::group_by(treatment) |>
  dplyr::summarise(
    trough_active_frac = min(ActiveAirwayTryptase, na.rm = TRUE),
    .groups = "drop"
  ) |>
  dplyr::mutate(
    pct_inhibition_sim   = 100 * (1 - trough_active_frac),
    pct_inhibition_paper = c(`300 mg SC q4w` = 79.2, `600 mg SC q4w` = 94.0,
                             `900 mg IV q4w` = 97.6, `1800 mg IV q4w` = 98.9,
                             `3600 mg IV q4w` = 99.5)[treatment]
  ) |>
  dplyr::select(treatment, pct_inhibition_paper, pct_inhibition_sim)
knitr::kable(trough, digits = 2,
             caption = "Steady-state active-tryptase inhibition (%) in asthma at the lowest concentration of dosing interval 5; simulated values vs Rymut 2023 Table 2 (no-exacerbation column).")
Steady-state active-tryptase inhibition (%) in asthma at the lowest concentration of dosing interval 5; simulated values vs Rymut 2023 Table 2 (no-exacerbation column).
treatment pct_inhibition_paper pct_inhibition_sim
1800 mg IV q4w 98.9 99.56
300 mg SC q4w 79.2 90.41
3600 mg IV q4w 99.5 99.78
600 mg SC q4w 94.0 96.60
900 mg IV q4w 97.6 99.08

PKNCA validation

Run PKNCA on the simulated serum MTPS9579A SAD profiles for non- compartmental confirmation that the structural disposition matches typical mAb behaviour (slow CL, two-compartment back-extrapolated half-life on the order of weeks at supra-saturation doses).

sad_nca_input <- sad_sim |>
  dplyr::filter(!is.na(Cc))

conc_obj <- PKNCA::PKNCAconc(
  sad_nca_input,
  Cc ~ time | treatment + id
)

dose_df <- sad_events |>
  dplyr::filter(evid == 1) |>
  dplyr::select(id, time, amt, treatment)

dose_obj <- PKNCA::PKNCAdose(dose_df, amt ~ time | treatment + id)

intervals <- data.frame(
  start      = 0,
  end        = 35,
  cmax       = TRUE,
  tmax       = TRUE,
  auclast    = TRUE,
  half.life  = TRUE
)

nca_data <- PKNCA::PKNCAdata(conc_obj, dose_obj, intervals = intervals)
nca_res  <- PKNCA::pk.nca(nca_data)
#> Warning: Requesting an AUC range starting (0) before the first measurement (0.01) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.01) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.01) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.01) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.01) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.01) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (0.01) is not allowed
nca_summary <- as.data.frame(nca_res$result) |>
  dplyr::select(treatment, PPTESTCD, PPORRES) |>
  tidyr::pivot_wider(names_from = PPTESTCD, values_from = PPORRES)
knitr::kable(nca_summary, digits = 3,
             caption = "Simulated SAD NCA parameters by dose / route (typical-value PK; 0-35 day window).")
Simulated SAD NCA parameters by dose / route (typical-value PK; 0-35 day window).
treatment auclast cmax tmax tlast lambda.z r.squared adj.r.squared lambda.z.time.first lambda.z.time.last lambda.z.n.points clast.pred half.life span.ratio
100 mg SC NA 10.009 6.50 35 0.077 1 1 30.0 35 11 2.111 9.046 0.553
1800 mg IV NA 539.672 0.01 35 0.024 1 1 14.0 35 43 130.030 29.494 0.712
30 mg SC NA 2.996 6.50 35 0.085 1 1 17.5 35 36 0.366 8.142 2.149
300 mg IV NA 89.945 0.01 35 0.034 1 1 30.0 35 11 18.083 20.204 0.247
300 mg SC NA 30.052 7.00 35 0.041 1 1 31.5 35 8 12.026 17.007 0.206
3600 mg IV NA 1079.344 0.01 35 0.023 1 1 14.5 35 42 264.319 30.488 0.672
900 mg IV NA 269.836 0.01 35 0.025 1 1 12.5 35 46 62.900 27.650 0.814

Assumptions and deviations

  • Biodistribution coefficient (Kp). Rymut 2023 chose Kp = 3% as the visual best fit across SAD + MAD cohorts (Figure S2) and the observed nasal-lining-fluid:serum concentration ratio (Figure 4). The same Kp is used for free mAb and for the mAb-tryptase complex per Methods. The packaged model exposes kp_free and kp_bound as overrideable fixed parameters; users may rerun simulations at 1% or 10% (Figure S2) by passing alternate values via rxSolve’s params = list(kp_free = 0.01, kp_bound = 0.01) argument.

  • Asthma scenario via parameter overrides. The fit cohort is healthy. To simulate adults with moderate-to-severe asthma the user overrides the two baseline-tryptase constants per Methods

    • Figure 5: bl_tryptase_isf = 140 (airway total tryptase, ng/mL) and lbase = log(0.223 * 1.5) = log(0.3345) (systemic total tryptase, log nM). This is the recipe used to reproduce Table 2; no additional re-fitting was performed by the paper authors.

  • Exacerbation scenarios. Rymut 2023 Table 2 columns 2-4 (10x, 100x, 1000x tryptase) simulate an acute exacerbation by transiently scaling the tetramer production rate. The packaged model does not embed a time-window covariate for the exacerbation; users implement the scenario by overriding bl_tryptase_isf to the scaled value for the duration of the exacerbation window (e.g. via a time- varying covariate in the event table).

  • Active tryptase has no residual error. The paper reports active airway tryptase only through visual comparison of model prediction vs nasosorption observations (Figure 3, Figure S2); the fit was not stochastic in this output. The packaged model emits ActiveAirwayTryptase as a derived ratio (no ~ error term) so that downstream users see a deterministic typical-value trajectory.

  • Unit conversion of the additive serum-mAb residual error. The NONMEM TMDD model worked internally in nM (the dose AMT column was pre-converted from mg to nmol so that A(2)/V2 = Kss = Base units were aligned; Text S1 confirms Kss = 0.0448 nM and Base = 0.223 nM). The packaged model reports Cc in ug/mL to match Rymut 2023 Figure 2; the additive residual addSd = 2.72 nM in the NONMEM output is therefore converted to ug/mL as 2.72 * 155 / 1000 = 0.4216 using mAb MW 155 kDa (Text S2). The proportional residual propSd = 0.103 and the serum total tryptase residual propSd_TotalSerumTryptase = 0.245 are unitless fractions and transfer unchanged.

  • Convention deviations (compartments). Three compartment names (mab_isf, monomer_isf, complex_monomer_isf) are not in the pre-existing nlmixr2lib::compartments register; they were added to R/conventions.R in this PR with rationale notes. The mAb- monomer complex and the free monomer in the ISF do not fit the pre-existing target_<loc> / complex_<loc> pattern because this model carries the target in two distinct oligomeric forms (active tetramer + inactive monomer) each with its own bound- complex state.

  • Convention deviations (parameter names). The mechanistic parameters (kss, kdeg, clint, base, kp_free, kp_bound, refl_lymph, klf_tissue, bl_tryptase_isf, kel_tryp_isf, kdiss_tet_isf, sum_kel_isf, f_tet_isf, kbreak_tet, kel_mono_ab, kon_isf, kd_isf, mw_ab, mw_mono, mw_tet) are paper-named and may surface as convention warnings; they follow the same pattern used by Hayashi_2007_omalizumab.R, PerezRuixo_2025_posdinemab.R, and Papachristos_2020_bevacizumab_qss.R.

  • rxode2 parser workaround. The rxode2 mu-reference parser flags <ini_param> + <ini_param> arithmetic in model() as a between- subject-variability block declaration even when the line uses <-. The aggregate rate constants sum_kel_isf (= kel + kdiss) and the rate-balance fraction f_tet_isf (= kel / (kel + kdiss)) are therefore precomputed as fixed numerics in ini() rather than being assembled at simulation time from the two component rates. The numeric values are identical; the parameterisation is split only to bypass the parser ambiguity.