Abstract
Quantitative risk assessment prediction of drug-drug interactions (DDIs) caused by metabolic enzyme inhibition is essential in both new drug development and clinical practice. Regulatory guidance recommends three approaches: cutoff criteria that conservatively evaluate the potential for DDIs using minimal data and quantitative prediction using mechanistic static pharmacokinetic (MSPK) models and physiologically based pharmacokinetic (PBPK) models. In clinical settings, the CR-IR method is also applied, which predicts DDIs using the contribution ratio of cytochrome P450 (CYP) isoforms to substrate clearance together with the inhibition ratio of an inhibitor toward the corresponding CYP.
This study examined 27 clinically reported DDIs, including 20 cases of reversible CYP inhibition and seven of time-dependent inhibition, by comparing observed increases in the area under the plasma concentration-time curve (AUC) with those predicted by the three approaches. The ratio of predicted to observed increases in AUC was within twofold (0.5 – 2.0) in approximately half of the cases for both the MSPK and PBPK models, whereas this criterion was met in 26 cases using the CR-IR method. PBPK models can accommodate analyses that involve flexible dosing regimens, and even a simple PBPK model was able to adequately predict increases in AUC. Thus, the PBPK model analysis is suggested to be applicable in clinical settings.
The CR-IR method showed the most accurate prediction of AUC increase, suggesting that this method can be applicable for the prediction of AUC increases in clinical settings. Nevertheless, further investigations are required to confirm its accuracy for other DDIs.
