Osimertinib Optimal Dosing Through Population Pharmacokinetics in Virtual Patients with NSCLC: An Approach Integrating Multiple Studies

Background/Objectives: Osimertinib is a third-generation tyrosine kinase inhibitor targeting epidermal growth factor receptor (EGFR) T790M mutations in metastatic non-small cell lung cancer (NSCLC). A comprehensive population pharmacokinetic (popPK) model may inform dosing across patient populations. The objective was to develop an integrated popPK model for osimertinib in virtual patients with NSCLC and to evaluate the influence of patient characteristics on PK variability. Methods: Data were extracted from four published popPK studies of osimertinib in patients with NSCLC. A virtual cohort of 1000 patients from each study was simulated using the corresponding final popPK model, with covariate distributions reconstructed from the reported demographics, resulting in a total virtual cohort of 4000 patients. The primary outcome was the development of a popPK model using nonlinear mixed-effects modeling. Demographic and clinical covariates were evaluated for effects on PK variability. Results: A parent-metabolite one-compartment model best described osimertinib popPK. Body weight and serum albumin were significant covariates, with lower body weight and lower albumin concentrations associated with increased osimertinib plasma exposure. However, simulations demonstrated that the standard 80 mg once-daily dose maintained plasma concentrations within the literature-informed reference interval across all body weight and albumin subgroups. Conclusions: This popPK model suggests that despite significant effects of body weight and albumin on osimertinib exposure, routine dose adjustment may not be required. The standard 80 mg once-daily regimen maintains concentrations within the reference interval across diverse virtual patient populations with NSCLC, supporting current dosing recommendations. These model-informed findings warrant prospective validation with real-world clinical data to confirm generalizability.