PII-096 - DEVELOPMENT OF A PHYSIOLOGICALLY BASED PHARMACOKINETIC MODEL TO PREDICT PHARMACOKINETICS AND DRUG-DRUG INTERACTIONS FOR REPOTRECTINIB

S. Sharma¹, Z. Hu¹, L. Hamuro¹, B. Schmidt¹, L. Gaohua¹; ¹Bristol Myers Squibb, Princeton, NJ, USA.

Background: Repotrectinib is a potent small-molecule tyrosine kinase inhibitor for treating adult patients with locally advanced or metastatic ROS1-positive non-small cell lung cancer and NTRK-positive solid tumors. It is predominantly metabolized by CYP3A4, accounting for >80% of its clearance. Repotrectinib is a substrate for P-gp and possibly BCRP and MATE2-K. Based on in vitro evaluation, it has the potential to induce CYP3A4, CYP2B6, and CYP2C8/9/19, and to inhibit CYP3A4, CYP2C8, CYP2C9, and UGT1A1. A physiologically based pharmacokinetic (PBPK) model for repotrectinib was constructed to evaluate the drug-drug interaction (DDI) potential as a victim of CYP3A4 and P-gp, and as a perpetrator of CYP2B6 and CYP2C8/9/19 substrates in cancer patients.

Methods: The PBPK model was developed using Simcyp V23 through a middle-out approach by integrating available in vitro, mass-balance, and clinical data. The contributions of CYP3A4 versus P-gp to repotrectinib elimination were evaluated to capture the PK profiles of repotrectinib with strong CYP3A4/P-gp modulators (i.e. itraconazole and rifampicin) in healthy volunteers. The model was further extrapolated to predict PK in cancer population. It was validated with independent clinical studies, including single- and multiple-ascending doses, food-effect and relative bioavailability studies in healthy volunteers, and food-effect studies in cancer patients. The model was evaluated to ensure simulated-to-observed PK endpoints were within a 2-fold range of the observed data.

Results: Out of the 72 available Cmax and AUC values collected from repotrectinib clinical studies in patients and healthy volunteers, the model was able to predict 100% of them within 2-fold and 49% of them within 1.25-fold of the observed data. The model captured the terminal half-lives within 1.5-fold with CYP3A4/P-gp modulators (itraconazole and rifampicin) in healthy volunteers.

Conclusion: A PBPK model for repotrectinib was successfully developed and validated against multiple observed clinical studies. Upcoming dedicated DDI studies with selective P-gp and CYP3A4 inhibitors will be used to evaluate and update the model. The PBPK model can be used to predict DDI risk with other dual and selective CYP3A4/P-gp modulators, design clinical trials, and inform dosing recommendations.