Translate Pharmacokinetics of PD-1/PD-L1 Monoclonal Antibodies from Cynomolgus Monkey to Human: Comparison of Different Approaches.

ABSTRACT

Antibodies blocking programmed death-1 (PD-1) and its natural ligand programmed death-ligand 1 (PD-L1) have been proved to be promising strategies in recent years. Hundreds of PD-1/PD-L1 antibodies are under development worldwide. Prediction of human pharmacokinetics (PK) in the preclinical stage is critical for designing dosing regimens in first-in-human studies. This study aims to predict the PK of PD-1/PD-L1 antibodies in human by scaling of monkey data. A systematic literature search of published articles on the PK of PD-1/PD-L1 antibodies in cynomolgus monkey and in human was conducted. Allometric scaling (AS), the species time-invariant (STIV) method, as well as physiologically based pharmacokinetic (PBPK) modeling were investigated. Six antibodies (avelumab, atezolizumab, nivolumab, pembrolizumab, cemiplimab, and zimberelimab) were included for investigation. The exponents used in this study were 0.85 and 1 for clearance (CL) and distribution volume (V), respectively, both for AS and STIV methods. The generic PBPK model for macromolecules in PK-Sim was used without further modifications. The dissociation constant of the antibody for binding to FcRn (KD) in endosome space for human was assumed to be two-fold of that for monkey. Predicted human CLs for the majority of drugs were within the observed range, while Vs were not well predicted using the AS method. The STIV method and the generic PBPK model can be employed to translate concentration-time curves of PD-1/PD-L1 antibodies from cynomolgus monkey to human with comparable efficacy. The results of this study provide reference for the early development of PD-1/PD-L1 antibodies.