摘要：Established guidelines have recommended a number of methods based on in vitro data to assess the CYP3A induction risk of new chemical entities in clinical practice. In this study, we evaluated the predictability of various assessment methods. We collected in vitro parameters from a variety of literature that includes data on 19 batches of hepatocytes. Clinical CYP3A induction was predicted using 3 direct approaches—the fold-change, basic model, and mechanistic static models—as well as 5 correlation approaches, including the relative induction score (RIS) and the relative factor (RF) method. These predictions were then compared with data from 30 clinical inductions. Collected in vitro parameters varied greatly between hepatocyte batches. Direct assessment methods using fixed cut-off values provided a lot of false predictions due to hepatocyte variability, which can overlook induction risk or lead to needless clinical drug–drug interaction (DDI) studies. On the other hand, correlation methods with the cut-off values set for each batch of hepatocytes accurately predicted the induction risk. Among these, the AUC u /inducer concentrations for half the maximum induction (EC 50 ) and the RF methods which use the area under the curve ( AUC ) of the unbound inducers for calculating induction potential showed an especially good correlation with clinical induction. Correlation methods were better at predicting clinical induction risk than the other methods, regardless of hepatocyte variability. The AUC u /EC 50 and the RF methods in particular had a small number of false predictions, and can therefore be used to assess induction risk along with the other correlation methods recommended in guidelines.
