Abstract

The success of in vitro derived K_i values for predicting drug-drug interactions in vivo has been mixed. For example, the use of hepatic input concentration of inhibitor has resolved the negative and positive interactions on the qualitative level, eliminating false negative predictions. However, several examples of false positives and a high incidence of over-predictions of true positive interactions indicated a need for incorporation of additional factors. The aim of this study was to investigate the effect of parallel elimination pathways as a possible reason for false positives and over-predictions. Simulation studies indicated that the degree of interaction (assessed by area under the plasma concentration-time curve ratio in the presence and absence of inhibitor) depends largely on the fraction of substrate metabolized by the particular P450 enzyme (fm_CYPi) that is inhibited. The current analysis focused on CYP2D6 interactions due to the well documented genetic polymorphism and the ability to estimate fm_CYP2D6 readily from in vivo data obtained in extensive and poor metabolizers. Based on either a phenotype study or an alternative regression analysis approach, the fm_CYP2D6 values of 0.37 to 0.94 and 0.25 to 0.89, respectively, were obtained for nine substrates. Prediction of 44 drug-drug interaction studies was improved by the combination of parallel pathways of elimination and their susceptibility to inhibition. The overall success of predicting positive and negative interactions was increased from 54% to 84%, and the number of over-predictions was substantially reduced. It is concluded that incorporating parallel pathways provides a valuable step forward in making quantitative predictions of drug-drug interactions from in vitro data.