Physiologically Based Pharmacokinetic Models Can Be Used to Predict the Unique Nonlinear Absorption Profiles of Vismodegib

Abstract

Predicting human pharmacokinetics (PK) during the drug discovery phase is valuable to assess doses required to reach therapeutic exposures. For orally administered compounds, however, this can be especially difficult since the absorption process is complex. Vismodegib is a compound with unique nonlinear oral PK characteristics in humans. Oral physiologically-based pharmacokinetic (PBPK) models were built using preclinical in vitro and in vivo data and successfully predicted the oral PK profiles in rats, dogs, and monkeys. Simulated drug exposures (AUC0-inf and Cmax), following oral administration were within 2-fold of observed values for the dog and monkey, and close to 2-fold for the rat, providing validation to the model structure. Adaptation of this oral PBPK model to humans, using human physiological parameters coupled with predicted human PK, resulted in underpredictions of vismodegib exposure following both single and multiple doses. When observed human PK was used to drive the oral PBPK model, oral PK profiles in humans were well predicted with fold errors in predicted vs observed drug exposures being close to 1. Importantly, the oral PBPK model captured the unique nonlinear, non-dose dependent PK of vismodegib at steady-state. The mechanism responsible for nonlinearity was consistent with oral absorption being influenced by nonsink permeation conditions. We introduce a new parameter, the permeation gradient factor, to characterize the effect of nonsink conditions on permeation. Using vismodegib as an example, we demonstrate the value of using oral PBPK models in drug discovery to predict the oral PK of compounds with nonlinear absorption characteristics in human. Significance Statement A physiologically-based pharmacokinetic model was built to demonstrate the value of these models early in the drug discovery stage for the prediction of human PK for compounds with unusual oral pharmacokinetics. In this study, our model could successfully capture the unique steady-state oral pharmacokinetics of our model compound, vismodegib. The mechanism for nonlinearity can be attributed to nonsink permeation conditions in vivo. We introduce the permeation gradient factor as a parameter to assess this effect.