Optimizing higher throughput methods to assess drug-drug interactions for CYP1A2, CYP2C9, CYP2C19, CYP2D6, rCYP2D6, and CYP3A4 in vitro using a single point IC(50)

Abstract

Drug-drug interactions involving cytochrome P(450) (CYP) are an important factor in whether a new chemical entity will survive through to the development stage. Therefore, the identification of this potential as early as possible in vitro could save considerable future unnecessary investment. In vitro CYP interaction screening data generated for CYP2C9, CYP2D6, and CYP3A4 were initially analyzed to determine the correlation of IC(50) from 10- and 3-point determinations. A high correlation (r = 0.99) prompted the further assessment of predicting the IC(50) by a single value of percent inhibition at either 10, 3, or 1 microM. Statistical analysis of the initial proprietary compounds showed that there was a strong linear relationship between log IC(50) and percent inhibition at 3 microM, and that it was possible to predict a compound's IC(50) by the percent inhibition value obtained at 3 microM. Additional data for CYP1A2, CYP2C19, and the recombinant CYP2D6 were later obtained and used together with the initial data to demonstrate that a single statistical model could be applicable across different CYPs and different in vitro microsomal systems. Ultimately, the data for all five CYPs and the recombinant CYP2D6 were used to build a statistical model for predicting the IC(50) with a single point. The 95% prediction boundary for the region of interest was about +/- 0.37 on log(10) scale, comparable to the variability of in vitro determinations for positive control IC(50) data. The use of a single inhibitor concentration would enable determination of more IC(50) values on a 96-well plate and result in more economical use of compounds, human liver or expressed enzyme microsomes, substrates, and reagents. This approach would offer the opportunity to increase screening for CYP-mediated drug-drug interactions, which may be important given the challenges provided by the generation of orders of magnitude more new chemical entities in the field of combinatorial chemistry. In addition, the algorithmic approach we propose would obviously be applicable for other in vitro bioactivity and therapeutic target enzyme and receptor screens.