Quantitative prediction of macrolide drug-drug interaction potential from in vitro studies using testosterone as the human cytochrome P4503A substrate

Objective Macrolide antibiotics are mechanism-based inactivators of CYP3A enzymes that exhibit varying degrees of inhibitory potency. Our aim was to predict quantitatively the drug-drug interaction (DDI) potential of five macrolides from in vitro studies using testosterone as the CYP3A substrate, and to compare the predictions generated from human liver microsomal and recombinant CYP3A4 data. Methods The in vitro kinetic constants of CYP3A inactivation (K _I and k _inact) were estimated by varying the time of pre-incubation and the concentration of troleandomycin, erythromycin, clarithromycin, roxithromycin or azithromycin. CYP3A activity was determined from the measurement of testosterone 6β-hydroxylation with human liver microsomes (HLM) and recombinant CYP3A4 as the enzyme sources. The mechanism-based pharmacokinetic model was fitted with inactivation data to predict the increase in oral area under the plasma concentration-time curve (AUC) for midazolam. Results All five macrolides inactivated testosterone 6β-hydroxylation by HLM and recombinant CYP3A4 with k _inact values in the range of 0.023 to 0.058 min⁻¹. The potency of inactivation (K _I) was higher using recombinant CYP3A4 as the enzyme source. The oral AUCs for midazolam were predicted from HLM data to increase 16.6, 5.3, 4.6, 1.6 and 1.2-fold due to the inhibition of metabolic clearance by troleandomycin, erythromycin, clarithromycin, roxithromycin and azithromycin, respectively. These results are within the range of the AUC ratios reported for clinical DDI studies. The predicted AUC increases generated using recombinant CYP3A4 overestimated the magnitude of the DDIs. Conclusions The DDI potential of five macrolide antibiotics was quantitatively predicted from in vitro studies using testosterone as the CYP3A substrate with HLM as the enzyme source.