Acetaminophen PBPK Model Compares Parent and Metabolite Exposure in MASLD-Related Groups
A linked parent–metabolite model explores acetaminophen exposure across obesity, fatty-liver disease and cirrhosis profiles. Predicted analgesic timing varies, while reactive-metabolite and disease-specific pharmacodynamic validation remain limited.
Disease-related exposure is modeled through linked pathways
Changes in liver physiology can alter both a drug and the metabolites formed from it. Researchers at Xuanwu Hospital, Capital Medical University, developed a physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) model of acetaminophen to examine these changes in populations related to metabolic dysfunction-associated steatotic liver disease (MASLD). The study links the parent drug to five metabolites and then maps predicted drug concentration to a pain-relief score.
The modeled profiles include overweight, obesity, fatty-liver disease and cirrhosis. The paper retains the historical nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH) labels used in the source clinical studies and software. The model compares separate physiological and clinical profiles assembled from published studies.
Healthy-population data establish the starting model
The investigators searched published single-dose acetaminophen studies available before March 2025, excluding fixed-dose combinations. GastroPlus 10.1, PBPK/PD simulation software, supplied the physiological framework. Drug-specific properties and enzyme parameters came from literature, prior models, predictions or fitting. Intravenous data first constrained distribution and clearance, followed by oral formulations and metabolite data.
The linked pathways include acetaminophen glucuronide, sulfate, the reactive intermediate N-acetyl-p-benzoquinone imine (NAPQI), cysteine and mercapturate metabolites. Parent-drug parameters were fixed before the metabolite components were progressively adjusted. Some missing physicochemical values were approximated from other compounds, and NAPQI was inferred through its upstream and downstream relationships.
The authors compared predicted concentration–time profiles with published clinical data and used a predicted-to-observed ratio of 0.5–2 as an evaluation window. Most parent and major-metabolite parameters met that window, but some metabolite exposures were overestimated. For example, 2 of 8 glucuronide measurements of area under the concentration–time curve to the last observed time were overpredicted in one healthy-population comparison. Absolute human NAPQI concentration data were insufficient to evaluate that component directly.
The disease comparison draws on four clinical studies
The established model was transferred to the software's predefined disease-related physiological profiles without manual modification. Four published studies supplied clinical comparisons with control groups. Three reported parent acetaminophen concentrations only; one NASH comparison also measured glucuronide and sulfate. The NAFLD and NASH data came from adolescents aged 12–18 years, a relevant population detail when interpreting the evidence.
In the modeled comparisons, weight- and dose-normalized exposure generally fell with higher body mass and in the NAFLD/NASH profiles. Cirrhosis produced a different parent–metabolite pattern: parent exposure increased in some comparisons, while increasing severity reduced several metabolite exposures. The authors also report that the model-derived trends did not fully match the compiled clinical-parameter trends. Sparse and heterogeneous source data limit a single consistent empirical pattern.
The model predicted low NAPQI quantities at therapeutic doses. That result depends on assumptions about metabolite formation and clearance and does not supply direct measurement of reactive-metabolite exposure. Glutathione generation and depletion were not dynamically represented, and transporters were omitted after their inclusion did not reproduce the observed profiles adequately.
Analgesic timing is a separate extrapolation
A sigmoid exposure–response model translated acetaminophen plasma concentration to a 0–10 pain-relief scale. Its pharmacodynamic benchmark came from chronic-pain patients, rather than a dedicated MASLD study. The predicted maximum score declined from 3.15 to 2.63 across increasing-body-mass profiles. Other disease comparisons retained similar maximum effects while changing their timing.
For example, the NASH profile reached its predicted peak effect at 0.93 hours, compared with 0.77 hours in its control. The corresponding cirrhosis comparison showed peak times of 0.78 hours in controls and 0.81, 0.91 and 0.97 hours for Child–Pugh A, B and C profiles. These are simulated peak-effect times, not prospectively observed analgesic outcomes in those groups.
The framework identifies hypotheses for further testing
The study separates a partially clinically checked exposure model from less directly validated reactive-metabolite and analgesic predictions. Source studies differ in demographics, sampling and assays, and several metabolite properties or enzyme relationships require approximation. The pharmacodynamic mapping additionally uses plasma concentration as a surrogate for the analgesic-site exposure.
The resulting framework can help formulate questions about disease-dependent drug disposition and trial design. Its low predicted NAPQI levels and retained modeled analgesic effect do not establish an individual dosing regimen or resolve overdose toxicity, for which glutathione dynamics would be central. Prospective disease-specific PK/PD data would test whether the modeled differences in exposure and timing translate to patients.