Drug-drug interaction (DDI) assessments of ruxolitinib, a dual substrate of CYP3A4 and CYP2C9, using a verified physiologically based pharmacokinetic (PBPK) model to support regulatory submissions

Abstract

Ruxolitinib is mainly metabolized by cytochrome P450 (CYP) enzymes CYP3A4 and CYP2C9 followed by minor contributions of other hepatic CYP enzymes in vitro. A physiologically based pharmacokinetic (PBPK) model was established to evaluate the changes in the ruxolitinib systemic exposures with co-administration of CYP3A4 and CYP2C9 perpetrators. The fractions metabolized in the liver via oxidation by CYP enzymes (fm,CYP3A4 = 0.75, fm,CYP2C9 = 0.19, and fm,CYPothers = 0.06) for an initial ruxolitinib model based on in vitro data were optimized (0.43, 0.56, and 0.01, respectively) using the observed exposure changes of ruxolitinib (10 mg) with co-administered ketoconazole (200 mg). The reduced amount of fm,CYP3A4 was distributed to fm,CYP2C9. For the initial ruxolitinib model with co-administration of ketoconazole, the area under the curve (AUC) increase of 2.60-fold was over-estimated compared with the respective observation (1.91-fold). With the optimized fm values, the predicted AUC ratio was 1.82. The estimated AUC ratios of ruxolitinib by co-administration of the moderate CYP3A4 inhibitor erythromycin (500 mg) and the strong CYP3A4 inducer rifampicin (600 mg) were within a 20% error compared with the clinically observed values. The PBPK modeling results may provide information on the labeling, i.e. supporting a dose reduction by half for co-administration of strong CYP3A4 inhibitors. Furthermore, an AUC increase of ruxolitinib in the absence or presence of the dual CYP3A4 and CYP2C9 inhibitor fluconazole (100-400 mg) was prospectively estimated to be 1.94- to 4.31-fold. Fluconazole simulation results were used as a basis for ruxolitinib dose adjustment when co-administering perpetrator drugs. A ruxolitinib PBPK model with optimized fm,CYP3A4 and fm,CYP2C9 was established to evaluate victim DDI risks. The previous minimal PBPK model was supported by the FDA for the dose reduction strategy, halving the dose with the concomitant use of strong CYP3A4 inhibitors and dual inhibitors on CYP2C9 and CYP3A4, such as fluconazole at ≤200 mg. Fluconazole simulation results were used as supportive evidence in discussions with the FDA and EMA about ruxolitinib dose adjustment when co-administering perpetrator drugs. Thus, this study demonstrated that PBPK modeling can support characterizing DDI liabilities to inform the drug label and might help reduce the number of clinical DDI studies by simulations of untested scenarios, when a robust PBPK model is established.