A simplified approach to predict CYP3A-mediated drug–drug interactions at early drug discovery: validation with clinical data

Abstract

1. The present study evaluates which factors should be incorporated into a simplified approach to reasonably predict CYP3A-mediated drug–drug interaction (DDI) at an early drug discovery stage.2. CYP3A IC50 values were obtained using human liver microsomes (HLM) and hepatocytes. Plasma and microsomal protein binding and in vitro hepatocyte partition coefficient (Kp) were also determined for 10 drugs. Therapeutic human maximum plasma concentrations (Cmax) were retrieved from the literature. DDI predictions were performed using an equation incorporating the fraction of the substrate metabolized by CYP3A with the total or free plasma Cmax, with or without correction for hepatocyte Kp.3. Based on the Ki data from HLM, the use of total Cmax provided a prediction of DDI within 2-fold of the observed clinical values for 9 out of 10 drugs.4. In comparison, free drug corrections for both Cmax and Ki values from HLM led to an underprediction of DDI (>3-fold error for five drugs).5. Data from hepatocytes showed, in general, lower prediction accuracy than data from HLM.6. CYP3A-mediated DDIs can be predicted with a high level of accuracy based on Ki estimates from HLM data and the total therapeutic plasma Cmax of the inhibitors. This approach should be widely applicable to the assessment of clinically significant DDIs risk in early drug discovery programs