Bayesian Methods for Pharmacokinetic/Pharmacodynamic Modeling of Pazopanib‐Induced Increases in Blood Pressure and Transaminases

Abstract

Relationships between plasma pazopanib concentrations and the probability of elevations in blood pressure, a marker of vascular endothelial growth factor receptor inhibition, and alanine aminotransferase (ALT) were investigated with logistic regression models. Data from a Phase I dose-escalation study in cancer patients (n = 57) were examined to determine the relationship between steady-state trough plasma pazopanib concentrations (Cτ ) and a clinically significant blood pressure increase, using a Bayesian logistic regression model. Data from 5 monotherapy studies in cancer patients (n = 344) were pooled to investigate the relationship between Cτ and maximum ALT ≥ 3× the upper limit of normal (ULN), using a Bayesian logistic regression model incorporating an asymptote. Both models were fit using WinBUGS. The median (95% credible interval, CrI) Cτ at which the probability of a clinically significant increase in blood pressure was 50% (EC50 ) was 12.3 μg/mL (6.12, 18.4). The median (95% CrI) EC50 for the maximum probability of ALT ≥ 3 × ULN was 15.4 μg/mL (3.8, 41.2) and the median (95% CrI) maximum probability of ALT ≥ 3 × ULN was 21% (14.5, 43.1). Results suggest that dose adjustments could be useful in managing the potential for hepatotoxicity.