Prediction of CYP2C9-mediated drug-drug interactions: a comparison using data from recombinant enzymes and human hepatocytes.

The IC50 values of 14 drugs were determined in recombinantly expressed CYP2C9 (rCYP2C9) and human hepatocytes and the data used to simulate clinical area under the plasma concentration-time curve (AUC) changes upon coadministration with prototypic CYP2C9 substrates. There was an excellent correlation between IC(50, apparent) values determined using diclofenac and naproxen as CYP2C9 substrates (r2 = 0.82, p < 0.0001), with values being generally higher in the naproxen assay. After correcting for nonspecific binding, the IC(50, unbound) values were similar between the assays, for the majority of compounds. Two compounds, amiodarone and benzbromarone, demonstrated substrate-specific differences, activating naproxen O-demethylase to approximately 250% of control activity at 1 mM and 1 microM, respectively, while inhibiting diclofenac 4'-hydroxylation with IC(50, apparent) values of 3 microM and 0.04 microM, respectively. CYP2C9 IC(50, apparent) values generated in human hepatocytes were systematically higher than those determined with rCYP2C9. After correcting for nonspecific binding, there was an excellent correlation of IC(50, unbound) values generated in the different milieu (r2 = 0.88, p < 0.0001). The ratio of inhibitor concentration at the entrance to the liver to the inhibition constant ([I]in/Ki) was used to simulate clinical deltaAUC changes and compared with that observed in vivo. Where [I]in, total/Ki, apparent) was used, there were zero false negatives (observed deltaAUC >or=2, predicted deltaAUC <2), eight correct assignations, and seven false positives (observed deltaAUC <or=2, predicted deltaAUC >2. Where [I]in, unbound/Ki, unbound was used, there was one false negative, 14 correct assignations, and zero false positives. In summary, the data presented here suggest that for CYP2C9 interactions, the use of total liver inhibitor concentrations may indeed avoid false negatives, but more realistic predictions may be achieved using unbound liver inhibitor concentrations and unbound in vitro inhibition parameters.