Comprehensive Analysis of Allometric Scaling Exponents for Human CL and Vdss Using PXB Mice and a Proposed Method for...

Reliable pharmacokinetic (PK) translation into humans is a cornerstone of first-in-human (FIH) dose selection, providing the exposure estimates needed to design efficient and informative early clinical trials. To predict human PK parameters conventional approaches include in vitro-in vivo extrapolation (IVIVE) and animal allometric scaling. The emergence of humanized liver PXB mice provides a new translational platform for predicting human clearance (CL) and steady-state volume of distribution (Vdss), typically by applying allometric scaling with fixed exponents of 0.75 for CL and 1.0 for Vdss. Previous evaluations generally applied fixed exponents without rigorous, data-driven validation of their robustness. In this study, we performed a systematic, data-science-driven assessment of the allometric exponents for human CL and Vdss derived from PXB mice data by iteratively varying dataset combinations and optimizing exponent values. Our results showed that the CL allometric relationship was highly robust across dataset combination and that the conventional 0.75 exponent is reliable. In contrast, Vdss predictions were markedly dataset dependent and the conventional 1.0 exponent was not stable. Based on these results, we investigated a new method for predicting Vdss. The compounds were grouped according to readily available data of PXB mice, and then subgroup specific allometric exponents were determined. One of these predictive models - grouping by fingerprints - substantially improved accuracy compared with the conventional approach; this was proven by repeated cross-validation. Hence, combining conventional CL and proposed Vdss prediction approach enhances confidence in FIH dosing decisions and offers a practical framework to improve translational pharmacokinetic prediction in drug discovery. SIGNIFICANCE STATEMENT: Using a systematic, data-science driven assessment, this study confirmed the 0.75 exponent in the allometric equation for CL via PXB mice is robust, whereas the 1.0 Vdss exponent is shown to be unstable. By determining subgroup-specific allometric exponents through fingerprint grouping, the proposed method significantly improved prediction accuracy-as validated by repeated cross-validation-offering a more precise, quantitative framework for informing human FIH dose selection compared to conventional fixed-exponent approaches.