(T-036) A Novel Empirical Drug Induced Autoinduction Model to Characterize the Population Pharmacokinetics of Repotrectinib with Advanced Solid Tumors Harboring ALK, ROS1, or NTRK1-3 Rearrangements

Jun Shen, NA – Director, BMS; Zheyi Hu, NA – Director, BMS; Lora Hamuro, NA – Senior Director, BMS; Justine Lam, NA – NA, NA; Ming Lu, NA – NA, NA; Li Zhu, NA – Executive Director, BMS; Amit Roy, NA – Head consulting, PUMAS; Anna Kondic, NA – Executive Director, BMS

Objective: Characterize the population pharmacokinetics (PPK) of repotrectinib in adults and pediatrics with a model to account for time-changing clearance resulting from drug induced auto-induction.

Methods: Repotrectinib is a tyrosine kinase inhibitor being developed for treatment of adults harboring ROS1 positive non small cell lung cancer (NSCLC) and adult or pediatric NTRK positive solid tumors. A dataset (with healthy volunteers and patients) including 7 studies in adults (N=620) and 1 study in pediatrics (N=24) were available for model development in NONMEM 7.3. The PPK model is a two-compartment model in first order absorption with an absorption lag time. The clearance was modeled as a time-dependent and concentration-dependent autoinduction process to account for the increased clearance and reduced drug concentration over time. The time-dependent piece was modeled as an Emax function. The concentration-dependent piece was parameterized as an empirical autoinduction model driven by model predicted trough concentration that is modeled as a separate Emax function where clearance increases with increasing Ctrough. Food effects and formulation effects were considered on absorption rate (KA) and bioavailability (F1) parameters. Baseline body weight (BW) and age effects were incorporated on clearance related parameters. The BW effects were also added to the volume of distribution parameters. The continuous covariates effects (e.g. body weight, age etc.) are modeled as a power function. The categorical covariate effects (e.g. disease status, formulation etc.) modeled as a linear function. The model was applied to simulate exposures in adults and adolescents (12-18 yrs) by body weight groups with dose regimens of interest.

Results: The PPK model provided an adequate description of the PK data for both adults and pediatric subjects. The empirical Ctrough driven autoinduction model described the clearance change over time reasonably well to avoid abrupt change of model predicted repotrectinib concentration which is typically expected from a dose driven autoinduction model. The Ctrough model does not need intensive computation of differential equations that are required for a typical semi-mechanism enzyme turnover autoinduction model. The typical maximum drug induced clearance (CLMAX) was estimated 4.9 fold over the baseline clearance. The time to achieve half of CLMAX (TC50) was estimated 47.1 hours. The trough concentration needed to achieve half of CLMAX (EC50) was estimated 77 ng/mL. The BW effects on clearance and volume of distribution were estimated 0.477 and 0.962, respectively as allometric scaling exponents. Age has an effect on CLMAX where the younger patients tend to have higher CLMAX. Based on the simulation, the model predicted the same flat dosing regimen (e.g. 160 mg QD 14 days followed by 160 mg BID) can produce comparable exposures in adult and adolescents patients.

Citations: NA