A genetic algorithm-based approach for the prediction of metabolic drug-drug interactions involving CYP2C8 or CYP2B6

Background and objectives: A genetic algorithm (GA) approach was developed to predict drug-drug interactions (DDIs) caused by cytochrome P450 2C8 (CYP2C8) inhibition or cytochrome P450 2B6 (CYP2B6) inhibition or induction. Nighty-eight DDIs, obtained from published in vivo studies in healthy volunteers, have been considered using the area under the plasma drug concentration-time curve (AUC) ratios (i.e., ratios of AUC of the drug substrate administered in combination with a DDI perpetrator to AUC of the drug substrate administered alone) to describe the extent of DDI.

Methods: The following parameters were estimated in this approach: the contribution ratios (CR_CYP2B6 and CR_CYP2C8, i.e., the fraction of the dose metabolized via CYP2B6 or CYP2C8, respectively) and the inhibitory or inducing potency of the perpetrator drug (IR_CYP2B6, IR_CYP2C8 and IC_CYP2B6, for inhibition of CYP2B6 and CYP2C8, and induction of CYP2B6, respectively). The workflow consisted of three main phases. First, the initial estimates of the parameters were estimated through GA. Then, the model was validated using an external validation. Finally, the parameter values were refined via a Bayesian orthogonal regression using all data.

Results: The AUC ratios of 5 substrates, 11 inhibitors and 19 inducers of CYP2B6, and the AUC ratios of 19 substrates and 23 inhibitors of CYP2C8 were successfully predicted by the developed methodology within 50-200% of observed values.

Conclusions: The approach proposed in this work may represent a useful tool for evaluating the suitable doses of a CYP2C8 or CYP2B6 substrates co-administered with perpetrators.