A strategy for early-risk predictions of clinical drug–drug interactions involving the GastroPlus^TM DDI module for time-dependent CYP inhibitors

Abstract

1. A set of reference compounds for time-dependent inhibition (TDI) of cytochrome P450 with available literature data for k_inact and K_I was used to predict clinical implications using the GastroPlus^TM software. Comparisons were made to in vivo literature interaction data. 2. The predicted AUC ratios (AUC_+inhibitor/AUC_control) could be compared with the observed ratios from literature for all compounds with detailed information about in vivo administration, pharmacokinetics and in vivo interactions (N = 21). For this dataset, the difference between predicted and observed AUC ratios for interactions with midazolam was within twofold for all compounds except one (telaprevir, for which non-CYP-mediated metabolism likely plays a role after multiple dosing). 3. The sensitivity, specificity and accuracy of the GastroPlus^TM predictions using a binary classification as no-to-weak interaction versus moderate-to-strong interaction for all compounds with available in vivo interaction data, were 80%, 82% and 81%, respectively (N = 31). 4. As a result of our evaluations of the DDI module in GastroPlus^TM, we have implemented an early TDI risk assessment decision tree for our drug discovery projects involving in vitro screening and early GastroPlus^TM predictions. Shifted IC₅₀ values are determined and k_inact/K_I estimated (by using a regression line established with in house-shifted IC₅₀ values and literature k_inact/K_I ratios), followed by GastroPlus^TM predictions.