Comparison of Different Algorithms for Predicting Clinical Drug-Drug Interactions, Based on the Use of CYP3A4 in Vitro Data: Predictions of Compounds as Precipitants of Interaction

Abstract

ABSTRACT:Cytochrome P450 3A4 (CYP3A4) is the most important enzyme in drug metabolism and because it is the most frequent target for pharmacokinetic drug-drug interactions (DDIs) it is highly desirable to be able to predict CYP3A4-based DDIs from in vitro data. In this study, the prediction of clinical DDIs for 30 drugs on the pharmacokinetics of midazolam, a probe substrate for CYP3A4, was done using in vitro inhibition, inactivation, and induction data. Two DDI prediction approaches were used, which account for effects at both the liver and intestine. The first was a model that simultaneously combines reversible inhibition, time-dependent inactivation, and induction data with static estimates of relevant in vivo concentrations of the precipitant drug to provide point estimates of the average magnitude of change in midazolam exposure. This model yielded a success rate of 88% in discerning DDIs with a mean -fold error of 1.74. The second model was a computational physiologically based pharmacokinetic model that uses dynamic estimates of in vivo concentrations of the precipitant drug and accounts for interindividual variability among the population (Simcyp). This model yielded success rates of 88 and 90% (for "steady-state" and "time-based" approaches, respectively) and mean -fold errors of 1.59 and 1.47. From these findings it can be concluded that in vivo DDIs for CYP3A4 can be predicted from in vitro data, even when more than one biochemical phenomenon occurs simultaneously.

CYP‐mediated drug–drug interactions with evacetrapib, an investigational CETP inhibitor: in vitro prediction and clinical outcome

“…Similarly, using the EMA guidance, this equation is [I]/ K i , where [I] is the maximum dose taken at one occasion/250 ml. In addition to the previously described basic model equations, the more recent mechanistic static models of CYP inhibition proposed by Fahmi 14 were also used in accordance with the FDA and EMA guidances for drug interaction studies 12, 13. The mechanistic static model is described by the following equation: $\text{AUCR}=\left(1true/\left(\left({\mathrm{A}}_{\mathrm{g}}\times {\mathrm{B}}_{\mathrm{g}}\times {\mathrm{C}}_{\mathrm{g}}\right)\times \left(1-{\mathrm{F}}_{\mathrm{g}}\right)+{\mathrm{F}}_{\mathrm{g}}\right)\right)\times true(1true/\left(\left({\mathrm{A}}_{\mathrm{h}}\times {\mathrm{B}}_{\mathrm{h}}\times {\mathrm{C}}_{\mathrm{h}}\right)\times {\mathrm{f}}_{\mathrm{m}}+\left(1-{\mathrm{f}}_{\mathrm{m}}\right)\right)$ …”

Confidence Assessment of the Simcyp Time-Based Approach and a Static Mathematical Model in Predicting Clinical Drug-Drug Interactions for Mechanism-Based CYP3A Inhibitors

“…Several retrospective analyses were conducted to compare the prediction accuracy of the Simcyp models with that of static models (Einolf, 2007;Fahmi et al, 2009). However, only a small portion of the entire dataset in these studies involved mechanism-based inhibitors of CYP3A.…”

“…The static mathematical model developed by Mayhew et al (2000) is a commonly used approach to predict mechanism-based drug interactions from in vitro estimated inactivation parameters. Several modifications to the original model have been made to incorporate the effects of intestinal wall metabolism (Wang et al, 2004b), competitive inhibition, and induction (Fahmi et al, 2009). Nonetheless, these static models are only capable of predicting the average magnitude of drug interactions across a population, assuming that the steady state of enzyme inhibition has been reached.…”

Substrate-dependent modulation of the catalytic activity of CYP3A by erlotinib

“…According to a recent publication, the reversible inhibition portion performed the best when the unbound portal vein concentration was used for [I] in vivo , while for irreversible inactivation and induction the unbound systemic concentration was the best. Thus, in this research, the unbound portal vein concentration (0.16, 0.18, and 0.31 μmol/L) was used for the reversible inhibition portion (for 50, 100, and 150 mg/d doses, respectively), while the unbound systemic concentration (0.07, 0.19, and 0.13 μmol/L) was adopted to avoid over-prediction of irreversible inactivation [30] . The values for [I] in vivo were derived from references [17,31] .…”