Skip to contents

Model and source

  • Citation: Abouelhassan Y, Shen Y, Chen A, Ye X, Nicolau DP, Kuti JL. Ex vivo assessment of sulbactam-durlobactam clearance during continuous renal replacement therapy to guide dosing recommendations. Antimicrob Agents Chemother. 2025;69(1):e01674-23. doi:10.1128/aac.01674-23. CRRT clearance regression from the Results (‘CRRT clearance studies’); PK parameters and dispersion from supplement Tables S11 and S12.
  • Description (sulbactam): Two-compartment population PK simulation model for sulbactam (given as sulbactam-durlobactam) in critically ill adults with ventilator-associated bacterial pneumonia, with and without continuous renal replacement therapy (CRRT) (Abouelhassan 2025). Without CRRT, total clearance is a non-renal arm plus a creatinine-clearance-driven renal arm, scaled by body weight and an infection-type shift (supplement Table S11). On CRRT the renal arm is replaced by the transmembrane clearance measured in an ex vivo CRRT circuit, a linear function of the effluent flow rate (CLTM = 0.114 + 0.973 x effluent, L/h), added to the non-renal clearance (supplement Table S12). Unbound concentration uses a fixed 38% protein binding.
  • Description (durlobactam): Two-compartment population PK simulation model for durlobactam (given as sulbactam-durlobactam) in critically ill adults with hospital-acquired or ventilator-associated bacterial pneumonia, with and without continuous renal replacement therapy (CRRT) (Abouelhassan 2025). Without CRRT, total clearance is a non-renal arm plus a creatinine-clearance-driven renal arm, scaled by body weight (supplement Table S11). On CRRT the renal arm is replaced by the transmembrane clearance measured in an ex vivo CRRT circuit, a linear function of the effluent flow rate (CLTM = 0.134 + 0.760 x effluent, L/h), added to the non-renal clearance (supplement Table S12). Unbound concentration uses a fixed 10% protein binding.
  • Article: https://doi.org/10.1128/aac.01674-23

Abouelhassan et al. measured the transmembrane clearance (CLTM) of sulbactam and durlobactam in an ex vivo continuous renal replacement therapy (CRRT) circuit and regressed it on the effluent flow rate. They then replaced the renal clearance of an established two-compartment population PK model with CLTM and ran Monte Carlo simulations to choose CRRT doses. The paper’s two drugs were modelled and simulated independently, so the library carries one model per drug (Abouelhassan_2025_sulbactam, Abouelhassan_2025_durlobactam). Each model holds both clearance models of the paper and switches between them on RRT_CRRT_STATUS:

  • RRT_CRRT_STATUS = 0 – the reference population of ventilator-associated bacterial pneumonia (VABP) patients not on CRRT (supplement Table S11).
  • RRT_CRRT_STATUS = 1 – CRRT patients, with clearance equal to the non-renal clearance plus CLTM at the prescribed effluent flow (supplement Table S12).

Population

No patient was studied in this paper. The ex vivo experiments circulated one litre of heparinized bovine blood (haematocrit 38.7%, albumin 3.6 g/dL) through a Prismaflex circuit at a blood flow of 200 mL/min, in CVVH and CVVHD modes, with M100 and HF1400 hemofilters and effluent flows of 1, 2 and 3 L/h, each in triplicate (Methods, ‘Ex vivo CRRT model’). The base population PK model was fitted to 162 patients from the Phase 2 and Phase 3 sulbactam-durlobactam trials; its demographics are not reported in this paper. The simulated populations were VABP patients with a mean creatinine clearance of 80 mL/min (range 30-129) and weight 76.4 +/- 4.7 kg, and CRRT patients of the same weight with effluent flows of 1-5 L/h (supplement Tables S11 and S12).

The same information is available programmatically via readModelDb("Abouelhassan_2025_sulbactam")()$population.

Source trace

The per-parameter origin is recorded as an in-file comment next to each ini() entry in inst/modeldb/specificDrugs/Abouelhassan_2025_sulbactam.R and inst/modeldb/specificDrugs/Abouelhassan_2025_durlobactam.R.

Parameter Sulbactam Durlobactam Source location
lcl_nonren (non-renal CL, L/h) 4.79 3.18 Table S11 ‘Total Body CL’; Table S12 ‘CL_NR’; supplement Methods
lcl (population PK CL, L/h) 13.6 9.36 Table S11 ‘Total Body CL’
f_renal (renal fraction) 0.648 0.66 Table S11 ‘Total Body CL’
e_crcl_cl_renal (exponent on CLcr/90) 0.932 0.694 Table S11 ‘Total Body CL’
e_wt_cl (exponent on WT/75, non-CRRT CL) 1.03 0.664 Table S11 ‘Total Body CL’
e_vabp_cl (VABP shift on CL, non-CRRT) -0.294 – Table S11 footnote a
lvc (Vc at 75 kg, L) 12.1 12.5 Table S11 ‘Vc’
e_wt_vc (exponent on WT/75) 0.828 0.518 Table S11 ‘Vc’
e_vabp_vc / e_habp_vabp_vc (infection shift on Vc) 1.38 1.52 Table S11 footnote b; Table S12 ‘Vc’
lq (Q, L/h) 7.82 4.42 Tables S11 and S12 ‘Q’
lvp (Vp, L) 6.98 5.83 Tables S11 and S12 ‘Vp’
lcl_crrt (CLTM intercept, L/h) 0.114 0.134 Results ‘CRRT clearance studies’; Fig. S1
e_effl_cl_crrt (CLTM slope on effluent, L/h per L/h) 0.973 0.760 Results ‘CRRT clearance studies’; Table 2; Fig. S1
etalvp (variance) 0.647 (CV 95.4%) 0.597 (CV 90.4%) Tables S11/S12 ‘Vp’ mean +/- SD
etalcl_nonren (variance) 0.0862 (CV 30%) 0.0862 (CV 30%) supplement Methods; Table S12 ‘CL_NR’
etalcl_crrt (variance) 0.0143 (CV 12%) 0.0143 (CV 12%) supplement Methods
fu (unbound fraction) 0.62 0.90 supplement Methods (protein binding 38% / 10%)
propSd 0 (fixed) 0 (fixed) not reported (simulation study)
Non-CRRT CL equation – – Table S11 ‘Total Body CL’
CRRT CL = CL_NR + CL_TM – – Table S12 ‘Total Body CL’; supplement Methods
Two-compartment ODEs – – supplement Methods (‘two compartment population pharmacokinetic analysis’)

Transmembrane clearance against effluent flow

The CRRT clearance is a deterministic function of effluent flow. Solving the typical-value models at the effluent flows of supplement Tables S3 and S4 reproduces the ‘ER only’ column of those tables.

mod_sul <- readModelDb("Abouelhassan_2025_sulbactam")
mod_dur <- readModelDb("Abouelhassan_2025_durlobactam")

effluent_grid <- c(1, 2, 2.5, 3, 4, 5)

typical_cltm <- function(mod, drug) {
  ev <- data.frame(
    id = seq_along(effluent_grid),
    time = 0,
    evid = 0,
    amt = 0,
    cmt = "central",
    WT = 75,
    CRCL = 90,
    DIS_HABP = 0,
    DIS_VABP = 1,
    RRT_CRRT_STATUS = 1,
    RRT_CRRT_EFFLUENT_FLOW = effluent_grid * 1000
  )
  sim <- rxode2::rxSolve(rxode2::zeroRe(mod), ev, returnType = "data.frame")
  data.frame(drug = drug, effluent = effluent_grid, cl_crrt = sim$cl_crrt)
}

cltm <- bind_rows(
  typical_cltm(mod_sul, "Sulbactam"),
  typical_cltm(mod_dur, "Durlobactam")
) |>
  mutate(
    published = c(
      1.09, 2.06, 2.55, 3.03, 4.01, 4.98, # Table S3 'ER only'
      0.89, 1.65, 2.03, 2.41, 3.17, 3.93 # Table S4 'ER only'
    )
  )
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalvp', 'etalcl_nonren', 'etalcl_crrt'
#> Warning: multi-subject simulation without without 'omega'
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalvp', 'etalcl_nonren', 'etalcl_crrt'
#> Warning: multi-subject simulation without without 'omega'

# Same parameters on both sides: the only difference is the table's rounding.
stopifnot(all(abs(cltm$cl_crrt - cltm$published) < 0.006))

cltm |>
  mutate(cl_crrt = signif(cl_crrt, 4)) |>
  rename(
    "Drug" = drug,
    "Effluent flow (L/h)" = effluent,
    "Model CL_TM (L/h)" = cl_crrt,
    "Published CL_TM (L/h)" = published
  ) |>
  knitr::kable(caption = "Typical CL_TM vs supplement Tables S3 and S4 ('ER only').")
Typical CL_TM vs supplement Tables S3 and S4 (‘ER only’).
Drug Effluent flow (L/h) Model CL_TM (L/h) Published CL_TM (L/h)
Sulbactam 1.0 1.087 1.09
Sulbactam 2.0 2.060 2.06
Sulbactam 2.5 2.547 2.55
Sulbactam 3.0 3.033 3.03
Sulbactam 4.0 4.006 4.01
Sulbactam 5.0 4.979 4.98
Durlobactam 1.0 0.894 0.89
Durlobactam 2.0 1.654 1.65
Durlobactam 2.5 2.034 2.03
Durlobactam 3.0 2.414 2.41
Durlobactam 4.0 3.174 3.17
Durlobactam 5.0 3.934 3.93
ggplot(cltm, aes(effluent, cl_crrt, colour = drug)) +
  geom_line() +
  geom_point(aes(y = published), shape = 1, size = 2.5) +
  labs(x = "Effluent flow rate (L/h)", y = "Transmembrane clearance (L/h)", colour = NULL) +
  theme_bw()
Replicates Fig. S1 of Abouelhassan 2025: CL_TM against effluent flow rate (lines: model; points: Tables S3 and S4).

Replicates Fig. S1 of Abouelhassan 2025: CL_TM against effluent flow rate (lines: model; points: Tables S3 and S4).

Virtual cohort

The paper simulated 1,000 patients per scenario; the cohorts below use 100 per scenario. Weight is drawn from N(76.4, 4.7) kg for every scenario, and the non-CRRT VABP creatinine clearance from a uniform distribution over 30-129 mL/min (the paper reports a mean of 80 and the range only). Each scenario is a 3-h infusion of 1 g of each drug at the stated interval.

rxode2::rxSetSeed(20250101)
set.seed(20250101)

n_per <- 100L

scenarios <- tibble::tribble(
  ~scenario,          ~crrt, ~effluent, ~tau,
  "No CRRT, 1g q6h",  0,     0,         6,
  "1 L/h, 1g q8h",    1,     1,         8,
  "1 L/h, 1g q12h",   1,     1,         12,
  "2 L/h, 1g q6h",    1,     2,         6,
  "2 L/h, 1g q8h",    1,     2,         8,
  "2.5 L/h, 1g q6h",  1,     2.5,       6,
  "2.5 L/h, 1g q8h",  1,     2.5,       8,
  "3 L/h, 1g q6h",    1,     3,         6,
  "3 L/h, 1g q8h",    1,     3,         8,
  "4 L/h, 1g q6h",    1,     4,         6,
  "4 L/h, 1g q8h",    1,     4,         8,
  "5 L/h, 1g q6h",    1,     5,         6,
  "5 L/h, 1g q4h",    1,     5,         4
) |>
  mutate(
    scenario = factor(scenario, levels = scenario),
    # Table S5/S6 footnote b: q12h was evaluated over 48-72 h, all others 24-48 h
    win_start = ifelse(tau == 12, 48, 24),
    win_end = win_start + 24
  )

subjects <- scenarios |>
  slice(rep(seq_len(n()), each = n_per)) |>
  mutate(
    id = row_number(),
    WT = rnorm(n(), 76.4, 4.7),
    CRCL = ifelse(crrt == 0, runif(n(), 30, 129), 90),
    DIS_HABP = 0,
    DIS_VABP = 1,
    RRT_CRRT_STATUS = crrt,
    RRT_CRRT_EFFLUENT_FLOW = effluent * 1000
  )

dose_rows <- subjects |>
  group_by(id) |>
  reframe(
    time = seq(0, win_end - tau, by = tau),
    across(c(scenario, WT, CRCL, DIS_HABP, DIS_VABP, RRT_CRRT_STATUS, RRT_CRRT_EFFLUENT_FLOW))
  ) |>
  mutate(evid = 1, amt = 1000, rate = 1000 / 3, cmt = "central")

obs_rows <- subjects |>
  group_by(id) |>
  reframe(
    time = sort(unique(c(0, seq(win_start, win_end, by = 0.1)))),
    across(c(scenario, WT, CRCL, DIS_HABP, DIS_VABP, RRT_CRRT_STATUS, RRT_CRRT_EFFLUENT_FLOW))
  ) |>
  mutate(evid = 0, amt = 0, rate = 0, cmt = "central")

events <- bind_rows(dose_rows, obs_rows) |>
  arrange(id, time, desc(evid)) |>
  as.data.frame()

stopifnot(!anyDuplicated(events[events$evid == 1, c("id", "time")]))

Simulation

simulate_drug <- function(mod, drug) {
  rxode2::rxSolve(mod, events, keep = "scenario", returnType = "data.frame") |>
    mutate(drug = drug)
}
sim <- bind_rows(
  simulate_drug(mod_sul, "Sulbactam"),
  simulate_drug(mod_dur, "Durlobactam")
) |>
  left_join(select(scenarios, scenario, tau, win_start, win_end), by = "scenario")
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ parameter labels from comments will be replaced by 'label()'
sim |>
  filter(
    scenario %in% c("No CRRT, 1g q6h", "2 L/h, 1g q8h", "3 L/h, 1g q6h"),
    time >= 24, time <= 48
  ) |>
  group_by(drug, scenario, time) |>
  summarise(
    p05 = quantile(Cc, 0.05), p50 = median(Cc), p95 = quantile(Cc, 0.95),
    .groups = "drop"
  ) |>
  ggplot(aes(time, p50, colour = scenario, fill = scenario)) +
  geom_ribbon(aes(ymin = p05, ymax = p95), alpha = 0.15, colour = NA) +
  geom_line() +
  facet_wrap(~drug) +
  labs(x = "Time (h)", y = "Total plasma concentration (mg/L)", colour = NULL, fill = NULL) +
  theme_bw() +
  theme(legend.position = "bottom")
Median and 5th-95th percentile total plasma concentrations over the 24-48 h window for the regimens the paper recommends (1g q8h for effluent < 3 L/h, 1g q6h for 3-5 L/h) and the non-CRRT reference.

Median and 5th-95th percentile total plasma concentrations over the 24-48 h window for the regimens the paper recommends (1g q8h for effluent < 3 L/h, 1g q6h for 3-5 L/h) and the non-CRRT reference.

PKNCA: total-drug AUC over the evaluation window

# A 0.25-h subsample of the simulation grid keeps PKNCA fast; the linear
# trapezoidal AUC moves by well under 1% against the full grid.
conc_df <- sim |>
  filter(!is.na(Cc), abs(time * 4 - round(time * 4)) < 1e-8) |>
  select(id, time, Cc, drug, scenario)
dose_df <- events |>
  filter(evid == 1) |>
  select(id, time, amt, scenario) |>
  tidyr::expand_grid(drug = c("Sulbactam", "Durlobactam"))

intervals <- scenarios |>
  distinct(scenario, win_start, win_end) |>
  transmute(scenario, start = win_start, end = win_end, aucint.last = TRUE)

conc_obj <- PKNCA::PKNCAconc(conc_df, Cc ~ time | drug + scenario / id)
dose_obj <- PKNCA::PKNCAdose(dose_df, amt ~ time | drug + scenario + id)
nca_res <- PKNCA::pk.nca(PKNCA::PKNCAdata(conc_obj, dose_obj, intervals = intervals))

auc_ind <- as.data.frame(nca_res) |>
  filter(PPTESTCD == "aucint.last") |>
  select(drug, scenario, id, auc = PPORRES)

The paper reports the mean, SD and 25th/75th percentiles of the total-drug AUC24-48 (Tables S5 and S6), not the median that ncaComparisonTable() would compute from individual results, so the simulated results are first aggregated to the per-scenario mean.

published <- tibble::tribble(
  ~scenario,          ~Sulbactam, ~Durlobactam,
  "No CRRT, 1g q6h",  466, 452,
  "1 L/h, 1g q8h",    533, 735,
  "1 L/h, 1g q12h",   362, 505,
  "2 L/h, 1g q6h",    607, 838,
  "2 L/h, 1g q8h",    456, 630,
  "2.5 L/h, 1g q6h",  563, 777,
  "2.5 L/h, 1g q8h",  420, 581,
  "3 L/h, 1g q6h",    526, 726,
  "3 L/h, 1g q8h",    399, 544,
  "4 L/h, 1g q6h",    469, 636,
  "4 L/h, 1g q8h",    348, 478,
  "5 L/h, 1g q6h",    420, 566,
  "5 L/h, 1g q4h",    622, 844
) |>
  pivot_longer(-scenario, names_to = "drug", values_to = "aucint.last") |>
  mutate(group = paste(drug, scenario, sep = ": ")) |>
  select(group, aucint.last)

sim_mean <- auc_ind |>
  group_by(drug, scenario) |>
  summarise(aucint.last = mean(auc), .groups = "drop") |>
  mutate(group = paste(drug, scenario, sep = ": ")) |>
  select(group, aucint.last)

cmp <- nlmixr2lib::ncaComparisonTable(
  simulated = sim_mean,
  reference = published,
  by = "group",
  units = c(aucint.last = "mg*h/L"),
  tolerance_pct = 20
)
#> Warning: ncaParamLabel(): unknown PKNCA code(s) returned as-is: 'aucint.last'
knitr::kable(
  cmp,
  caption = "Mean total-drug AUC24-48 (AUC48-72 for q12h): simulated vs supplement Tables S5 and S6. * differs by more than 20%."
)
Mean total-drug AUC24-48 (AUC48-72 for q12h): simulated vs supplement Tables S5 and S6. * differs by more than 20%.
NCA parameter group Reference Simulated % diff
aucint.last (mg*h/L) Sulbactam: No CRRT, 1g q6h 466 466 -0.0%
aucint.last (mg*h/L) Durlobactam: No CRRT, 1g q6h 452 467 +3.3%
aucint.last (mg*h/L) Sulbactam: 1 L/h, 1g q8h 533 532 -0.2%
aucint.last (mg*h/L) Durlobactam: 1 L/h, 1g q8h 735 705 -4.1%
aucint.last (mg*h/L) Sulbactam: 1 L/h, 1g q12h 362 357 -1.5%
aucint.last (mg*h/L) Durlobactam: 1 L/h, 1g q12h 505 494 -2.1%
aucint.last (mg*h/L) Sulbactam: 2 L/h, 1g q6h 607 577 -5.0%
aucint.last (mg*h/L) Durlobactam: 2 L/h, 1g q6h 838 833 -0.6%
aucint.last (mg*h/L) Sulbactam: 2 L/h, 1g q8h 456 431 -5.4%
aucint.last (mg*h/L) Durlobactam: 2 L/h, 1g q8h 630 612 -2.9%
aucint.last (mg*h/L) Sulbactam: 2.5 L/h, 1g q6h 563 526 -6.6%
aucint.last (mg*h/L) Durlobactam: 2.5 L/h, 1g q6h 777 757 -2.6%
aucint.last (mg*h/L) Sulbactam: 2.5 L/h, 1g q8h 420 411 -2.1%
aucint.last (mg*h/L) Durlobactam: 2.5 L/h, 1g q8h 581 578 -0.6%
aucint.last (mg*h/L) Sulbactam: 3 L/h, 1g q6h 526 532 +1.1%
aucint.last (mg*h/L) Durlobactam: 3 L/h, 1g q6h 726 708 -2.5%
aucint.last (mg*h/L) Sulbactam: 3 L/h, 1g q8h 399 388 -2.7%
aucint.last (mg*h/L) Durlobactam: 3 L/h, 1g q8h 544 535 -1.6%
aucint.last (mg*h/L) Sulbactam: 4 L/h, 1g q6h 469 457 -2.6%
aucint.last (mg*h/L) Durlobactam: 4 L/h, 1g q6h 636 623 -2.0%
aucint.last (mg*h/L) Sulbactam: 4 L/h, 1g q8h 348 343 -1.3%
aucint.last (mg*h/L) Durlobactam: 4 L/h, 1g q8h 478 470 -1.7%
aucint.last (mg*h/L) Sulbactam: 5 L/h, 1g q6h 420 410 -2.3%
aucint.last (mg*h/L) Durlobactam: 5 L/h, 1g q6h 566 557 -1.5%
aucint.last (mg*h/L) Sulbactam: 5 L/h, 1g q4h 622 635 +2.1%
aucint.last (mg*h/L) Durlobactam: 5 L/h, 1g q4h 844 838 -0.7%

pct <- merge(sim_mean, published, by = "group", suffixes = c("_sim", "_pub")) |>
  mutate(pct_diff = 100 * (aucint.last_sim / aucint.last_pub - 1))
stopifnot(
  nrow(pct) == 26,
  # A mis-transcribed clearance, slope or dose shifts every scenario of a drug
  # by tens of percent.
  abs(median(pct$pct_diff)) < 10,
  quantile(abs(pct$pct_diff), 0.9) < 20
)

Probability of target attainment

The targets are 50% fT>MIC for sulbactam and fAUC/MIC >= 10 for durlobactam, evaluated on the unbound concentration Cu over the same 24-h window.

mics <- c(2, 4, 8, 16, 32)

sim_win <- sim |>
  filter(time >= win_start, time <= win_end) |>
  arrange(drug, id, time)

fauc_ind <- sim_win |>
  group_by(drug, scenario, id) |>
  summarise(
    fauc = sum(diff(time) * (head(Cu, -1) + tail(Cu, -1)) / 2),
    .groups = "drop"
  )

# On the uniform grid, the share of grid intervals whose left end is above the
# MIC approximates fT>MIC.
ft_ind <- sim_win |>
  filter(time < win_end) |>
  select(drug, scenario, id, Cu) |>
  tidyr::expand_grid(mic = mics) |>
  group_by(drug, scenario, id, mic) |>
  summarise(ft_above = mean(Cu > mic), .groups = "drop")

pta_ind <- ft_ind |>
  left_join(fauc_ind, by = c("drug", "scenario", "id")) |>
  mutate(
    attained = ifelse(drug == "Sulbactam", ft_above >= 0.5, fauc / mic >= 10)
  )

pta <- pta_ind |>
  group_by(drug, scenario, mic) |>
  summarise(pta = mean(attained), .groups = "drop")

pta |>
  mutate(mic = paste0("MIC ", mic)) |>
  pivot_wider(names_from = mic, values_from = pta) |>
  rename("Drug" = drug, "Scenario" = scenario) |>
  knitr::kable(digits = 2, caption = "Simulated PTA (compare supplement Tables S5 and S6).")
Simulated PTA (compare supplement Tables S5 and S6).
Drug Scenario MIC 2 MIC 4 MIC 8 MIC 16 MIC 32
Durlobactam No CRRT, 1g q6h 1 1 1.00 1.00 0.98
Durlobactam 1 L/h, 1g q8h 1 1 1.00 1.00 1.00
Durlobactam 1 L/h, 1g q12h 1 1 1.00 1.00 0.92
Durlobactam 2 L/h, 1g q6h 1 1 1.00 1.00 1.00
Durlobactam 2 L/h, 1g q8h 1 1 1.00 1.00 1.00
Durlobactam 2.5 L/h, 1g q6h 1 1 1.00 1.00 1.00
Durlobactam 2.5 L/h, 1g q8h 1 1 1.00 1.00 1.00
Durlobactam 3 L/h, 1g q6h 1 1 1.00 1.00 1.00
Durlobactam 3 L/h, 1g q8h 1 1 1.00 1.00 0.98
Durlobactam 4 L/h, 1g q6h 1 1 1.00 1.00 1.00
Durlobactam 4 L/h, 1g q8h 1 1 1.00 1.00 0.98
Durlobactam 5 L/h, 1g q6h 1 1 1.00 1.00 1.00
Durlobactam 5 L/h, 1g q4h 1 1 1.00 1.00 1.00
Sulbactam No CRRT, 1g q6h 1 1 0.99 0.10 0.00
Sulbactam 1 L/h, 1g q8h 1 1 0.98 0.24 0.00
Sulbactam 1 L/h, 1g q12h 1 1 0.58 0.01 0.00
Sulbactam 2 L/h, 1g q6h 1 1 1.00 0.31 0.00
Sulbactam 2 L/h, 1g q8h 1 1 0.91 0.01 0.00
Sulbactam 2.5 L/h, 1g q6h 1 1 1.00 0.22 0.00
Sulbactam 2.5 L/h, 1g q8h 1 1 0.85 0.01 0.00
Sulbactam 3 L/h, 1g q6h 1 1 0.97 0.17 0.00
Sulbactam 3 L/h, 1g q8h 1 1 0.79 0.00 0.00
Sulbactam 4 L/h, 1g q6h 1 1 0.99 0.02 0.00
Sulbactam 4 L/h, 1g q8h 1 1 0.63 0.00 0.00
Sulbactam 5 L/h, 1g q6h 1 1 0.94 0.00 0.00
Sulbactam 5 L/h, 1g q4h 1 1 1.00 0.62 0.00

# The paper's headline claim: every simulated regimen achieves PTA > 0.9 at the
# 4/4 mg/L susceptibility breakpoint.
stopifnot(all(pta$pta[pta$mic == 4] > 0.9))

Durlobactam attains its target in nearly every subject at every MIC, as in Table S6. For sulbactam the discriminating MICs are 8 and 16 mg/L:

pta_sul_pub <- tibble::tribble(
  ~scenario,          ~mic8, ~mic16,
  "No CRRT, 1g q6h",  0.14,  0,
  "1 L/h, 1g q8h",    1,     0.21,
  "1 L/h, 1g q12h",   0.53,  0,
  "2 L/h, 1g q6h",    1,     0.41,
  "2 L/h, 1g q8h",    1,     0.04,
  "2.5 L/h, 1g q6h",  1,     0.271,
  "2.5 L/h, 1g q8h",  0.98,  0.01,
  "3 L/h, 1g q6h",    1,     0.16,
  "3 L/h, 1g q8h",    0.93,  0,
  "4 L/h, 1g q6h",    1,     0.02,
  "4 L/h, 1g q8h",    0.71,  0,
  "5 L/h, 1g q6h",    1,     0,
  "5 L/h, 1g q4h",    1,     0.54
)

pta |>
  filter(drug == "Sulbactam", mic %in% c(8, 16)) |>
  mutate(scenario = as.character(scenario), mic = paste0("sim", mic)) |>
  pivot_wider(names_from = mic, values_from = pta) |>
  select(-drug) |>
  left_join(pta_sul_pub, by = "scenario") |>
  select(scenario, mic8, sim8, mic16, sim16) |>
  rename(
    "Scenario" = scenario,
    "Published PTA, MIC 8" = mic8,
    "Simulated PTA, MIC 8" = sim8,
    "Published PTA, MIC 16" = mic16,
    "Simulated PTA, MIC 16" = sim16
  ) |>
  knitr::kable(digits = 2, caption = "Sulbactam PTA (50% fT>MIC) at MIC 8 and 16 mg/L: simulated vs supplement Table S5.")
Sulbactam PTA (50% fT>MIC) at MIC 8 and 16 mg/L: simulated vs supplement Table S5.
Scenario Published PTA, MIC 8 Simulated PTA, MIC 8 Published PTA, MIC 16 Simulated PTA, MIC 16
No CRRT, 1g q6h 0.14 0.99 0.00 0.10
1 L/h, 1g q8h 1.00 0.98 0.21 0.24
1 L/h, 1g q12h 0.53 0.58 0.00 0.01
2 L/h, 1g q6h 1.00 1.00 0.41 0.31
2 L/h, 1g q8h 1.00 0.91 0.04 0.01
2.5 L/h, 1g q6h 1.00 1.00 0.27 0.22
2.5 L/h, 1g q8h 0.98 0.85 0.01 0.01
3 L/h, 1g q6h 1.00 0.97 0.16 0.17
3 L/h, 1g q8h 0.93 0.79 0.00 0.00
4 L/h, 1g q6h 1.00 0.99 0.02 0.02
4 L/h, 1g q8h 0.71 0.63 0.00 0.00
5 L/h, 1g q6h 1.00 0.94 0.00 0.00
5 L/h, 1g q4h 1.00 1.00 0.54 0.62

The CRRT rows follow Table S5: high attainment at 8 mg/L that falls with effluent flow under q8h dosing, and low attainment at 16 mg/L. The simulated attainment at 8 mg/L under q8h is somewhat lower than published at the higher effluent flows (for example about 0.8 vs 0.93 at 3 L/h), consistent with the untruncated log-normal dispersion used here (see Assumptions and deviations).

The non-CRRT row is not reproduced: Table S5 gives a PTA of 0.14 at 8 mg/L, the simulation gives close to 1. The published row is hard to reconcile with the same table: its mean AUC24-48 of 466 mgh/L implies an average unbound concentration of 0.62 x 466 / 24 = 12 mg/L over the window, and the 2 L/h q8h CRRT row, with a lower mean AUC (456 mgh/L), reports a PTA of 1 at the same MIC. The mean AUC of this row is reproduced to within a few percent, so the clearance is right; the PTA difference must come from the shape of the profile or the PTA computation for this row, which the paper does not describe. The model was not adjusted to match it.

Assumptions and deviations

  • Dispersion as log-normal. The authors simulated with the Pmetrics non-parametric simulator from the means and SDs of Tables S11 and S12, with parameter limits (for example Vp 1-30 L, CL_NR 2-7 L/h for sulbactam). The library carries the SDs as log-normal between-subject variability, omega^2 = log(CV^2 + 1), and does not truncate. The very wide Vp dispersion (CV 95% and 90%) therefore yields some large Vp values the authors’ limits would have clipped; this affects the shape of the profile, not the AUC.
  • No variability on Vc and Q. Tables S11 and S12 print ranges for Vc and Q (for example Vc 20-45 L for sulbactam) but no SD, so no between-subject variability is carried on them; Vc varies only through body weight.
  • Variability on the non-renal clearance applies on CRRT only. Table S11 gives the non-CRRT clearance no dispersion, so the 30% CV of the supplement Methods enters only the CRRT clearance; in the non-CRRT arm the spread in clearance comes from creatinine clearance and weight alone.
  • Creatinine clearance distribution. The paper gives only the mean (80 mL/min) and range (30-129 mL/min) of the simulated VABP population; the cohort above draws it uniformly over the range, whose mean is 79.5 mL/min.
  • Infection type. The base model’s infection-type effects are printed only for the simulated population: a VABP shift of -0.294 on sulbactam clearance and HABP/VABP shifts of 1.38 (sulbactam) and 1.52 (durlobactam) on Vc. They are encoded on DIS_VABP (sulbactam) and DIS_HABP + DIS_VABP (durlobactam). The sulbactam clearance shift for HABP is not printed, so HABP is not encoded for sulbactam. Set DIS_VABP = 1 to reproduce the paper; the reference category (DIS_VABP = 0) of the unpublished base model is not identified in the source.
  • Base population PK model. The base model (162 Phase 2/3 patients) is not published separately in this paper and is cited as data on file. It is close to, but not the same as, the later published model of Cammarata 2024 (Cammarata_2024_sulbactam_durlobactam, 373 subjects, creatinine clearance BSA-normalized with reference 100 mL/min/1.73 m^2). The non-renal clearances of 4.79 and 3.18 L/h are numerically equal to (1 - f_renal) x CL of the base model.
  • Final CLTM model only. The multiple regressions on filter type, CRRT mode and replacement-fluid dilution (Table 2 and Table S2) were used by the authors only for sensitivity analyses (Tables S7 and S8); the final model that drives the dose recommendations uses effluent flow alone and is the one encoded.
  • No residual error. The source is a simulation study with no residual error model; propSd is fixed to 0.
  • PTA definitions. fT>MIC is computed as the fraction of the 0.1-h grid intervals with an unbound concentration above the MIC, and fAUC by the linear trapezoidal rule, over the 24-h evaluation window. The paper does not say over which window PTA was assessed; the AUC window of Tables S5 and S6 is used.
  • Sulbactam PTA without CRRT not reproduced. Supplement Table S5 reports a PTA of 0.14 at 8 mg/L for the non-CRRT 1g q6h reference; the model gives close to 1 while reproducing that row’s mean AUC. See the discussion under “Probability of target attainment”.
  • No erratum or correction notice was found for this article (Europe PMC, checked 2026-11-02).