Unlocking the Potential of Population Pharmacokinetic Models of Adalimumab in Patients with Crohn's Disease

Background/Objectives: Adalimumab (ADM) is a recombinant, fully human monoclonal antibody that exhibits pronounced inter- and intra-individual pharmacokinetic variability attributed to several factors. This study aims to externally evaluate the published ADM population pharmacokinetic models and their potential use in clinical practice, as well as to develop novel population pharmacokinetic model. Methods: Literature search was conducted using PubMed to identify ADM population pharmacokinetic models. Data from 195 patients with Crohn's disease treated at the University Medical Center "Zvezdara", Serbia, were used for the external evaluation of previously published models. In addition, the development of the new population pharmacokinetic model incorporated informative priors derived from the best-performing published model. Nonlinear mixed-effects modeling was performed in NONMEM^® (versions 7.6) for both prediction- and simulation-based diagnostics of existing models, as well as for the development of a new model. Results: Eight published pharmacokinetic models of ADM were included in the external evaluation. Although none of the models satisfied both population-level and normalized prediction distribution error (NPDE) diagnostic criteria, individual-level performance was acceptable: median prediction errors (MDPEs) were within ±20% across all studies, and median absolute prediction errors (MDAPEs) were below 30% in most cases (7 of 8 studies). The best-performing model was identified and implemented as a priori information in subsequent model development. A one-compartment model using with first-order absorption and elimination best described the data. The apparent clearance (CL/F) was estimated at 0.334 L/day, while informative priors were used for V/F and the effect of anti-drug antibodies (ADAs) on CL/F. Covariate analysis on CL/F identified C-reactive protein (CRP) and dosing regimen as statistically significant predictors (p < 0.01). Conclusions: The previous pharmacokinetic models of ADM exhibited suboptimal performance in population-level metrics and simulation-based diagnostics, while individual-level metrics showed substantial improvement. The newly developed model of ADM highlights associations among immunogenicity, drug pharmacokinetics, and inflammatory burden.