Microsomal prediction of in vivo clearance of CYP2C9 substrates in humans

Aims To assess the utility of human hepatic microsomes for predicting in vivo intrinsic clearance (CLint ) via the use of four cytochrome P450 2C9 substrates: phenytoin, tolbutamide (S)‐ibuprofen (two pathways) and diclofenac, and to examine the role of exogenous albumin within the microsomal incubation. Methods V max, Km and CLint (defined as V max/Km ratio) were estimated under initial rate conditions for five pathways of metabolism in a bank of 15 human hepatic microsomal samples and were scaled to in vivo units using the microsomal protein index. Non‐metabolic related binding in microsomes was measured for phenytoin and tolbutamide in the presence and absence of albumin. Results Microsomal CLint values differed by over two orders of magnitude, with the means ranging from 0.18 (phenytoin) to 40.70 (diclofenac) μl min−1 mg−1 microsomal protein. When these data were scaled and compared with published in vivo studies a similar rank order was obtained, however, the actual CLint tended to be underpredicted. While the in vivo unbound Km for phenytoin, 1–5 μm is substantially lower than the value determined in microsomes based on total concentrations (56 μm ), correction for the in vitro binding reduces this value to 20 μm and 6 μm in the absence and presence of albumin, respectively. Similar trends were seen with tolbutamide Km. Conclusions An appreciation of the utility of in vitro prediction can be best achieved when the range of CLint values predicted from the individual hepatic microsomal samples are compared with the range of individual in vivo CLint values reported in the literature. The degree of underprediction is less evident using the range than the mean data and no consistent advantage in adding albumin to the incubation media is apparent.