Development of a PK/PD-Efficacy Modeling Framework for Covalent Inhibitors

Background/Objectives: Covalent inhibitors often demonstrate prolonged pharmacological effects even after their disappearance from the site of action because target recovery depends on target turnover. This disconnect between pharmacokinetics (PK) and pharmacodynamics (PD) complicates the development of such inhibitors since plasma exposure coverage of in vitro potency cannot be used for compound selection and human dose projections. In this study, we describe the development of a PK/PD modeling framework for covalent inhibitors. Methods: The model was developed for KRAS G12C inhibitors using pre-clinical data on sotorasib. The model was validated using data from both internal Amgen compounds and published data for several KRAS G12C inhibitors. The applicability of the framework was extended to EGFR covalent inhibitors by incorporating PK/PD and the efficacy of osimertinib and its active metabolite AZ5104. Results: The model successfully captured the pharmacokinetics, KRAS G12C target occupancy, inhibition of phosphorylation of ERK protein, and tumor growth inhibition following the administration of sotorasib in mice bearing MIA PaCa-2 xenografts. External validation with several internal Amgen compounds as well as publicly available data on KRAS G12C inhibitors showed the robustness of the model. The application of this framework to EGFR inhibitor Osimertinib and AZ5104 captured p-EGFR dynamics and resultant tumor growth inhibition. Conclusions: A modeling framework for covalent inhibitors was developed that links exposure to target occupancy, downstream signaling, and tumor efficacy. This framework could be useful for compound optimization and human dose projections.