PII-095 - A POPULATION PHARMACOKINETIC (PK) MODEL OF MILVEXIAN, A SMALL MOLECULE FACTOR XIA INHIBITOR, IN HEALTHY VOLUNTEERS.

H. Back¹, A. Ajavon-Hartmann¹, W. Chen¹, W. Zhou², P. Zannikos², S. Merali¹; ¹Bristol Myers Squibb, ²Janssen R&D.

Background: Milvexian is a potential first-in-class oral factor XIa inhibitor being developed for prevention and treatment of major thrombotic events. Milvexian demonstrated over-proportionality at low doses, and under-proportionality at high doses, with proportionality across the therapeutic range when administered in a fasted state. Twice daily (BID) PK profile of milvexian deviates from linearity, due to dose dependent food effect. The objective of this study was to characterize milvexian PK using a population PK model, incorporating intrinsic and extrinsic drivers of PK in healthy volunteers to serve as the basis for understanding sparsely sampled patient data from Phase 2, to support Phase 3 dosing and study design.

Methods: Population PK modeling was performed using 6 Phase 1 clinical studies in healthy volunteers (n= 245) using NONMEM (version 7.5). A two-compartment with multiple absorption model was tested to characterize milvexian PK and covariate analysis was done using Stepwise Covariate Modeling (SCM). Goodness of fit (GOF) and Visual Predictive Check (VPC) were performed to evaluate the model performance.

Results: The absorption of milvexian was reasonably explained using a sequential zero and first-order absorption model and an exponential model was used for the inter-individual variability (IIV) and residual variability. The final parameter estimates and covariates are presented in Table 1. Food was selected as a major covariate that affected multiple parameters in addition to other covariates [age, dose, formulation, sex, weight, study (NCT04844424)]. The final model showed dose-dependent food effect with delayed and decreased exposure at doses less than or equal to 100 mg but increased exposure at doses greater than 100 mg. GOF plots showed no visual bias for the predictions and VPCs appropriately predicted and captured the observed milvexian PK across the dose range.

Conclusion: A population PK model was successfully developed, using healthy volunteer data. The model characterized milvexian PK well and identified key covariates such as nonlinearity and variability combined with the dose-dependent food effect. The final model and PK predictions will be utilized to characterize an exposure-response relationship of milvexian in patients and facilitate dose selection for Phase 3 studies.