Propagation of pharmacogenetic differences in cytochrome P450 into pharmacokinetic and pharmacodynamic measures

Literature reports of studies that investigate the impact of CYP polymorphisms on drug pharmacokinetics and response are often conflicting and the importance of genetic variation in drug metabolism for drug response remains unclear. Johnsson & Sheiner (2002) have advocated the need for 'smarter clinical trial design' and showed that simulation techniques can help in this process by integrating all the available information. However, current examples of clinical trial simulation rely heavily on data already available from in vivo studies and there is a need for utilising pharmacokinetic information gathered earlier on during drug development. The aim of the current work was to integrate early preclinical data on drug metabolism into a clinical trial simulation paradigm in order to investigate (A) the impact of genetic polymorphisms in the cytochromes P450 on the pharmacokinetics and pharmacodynamics of 5 model drugs: dextromethorphan, (S)-warfarin, midazolam, omeprazole and tolbutamide, and (B) the predicted power of studies to detect the effects of such polymorphisms. SimcypS algorithms incorporate information on in vitro metabolism and in vivo kinetics with interindividual variability in the genetics of drug metabolising enzymes and other physiological and demographic features. In the current study these algorithms were linked to pharmacokinetic-pharmacodynamic models to describe the time course of concentration and effect of the model drugs in virtual populations of subjects. The probability of detecting a statistically significant difference in the pharmacokinetics or response between CYP phenotypes/genotypes was assessed and the power of studies to detect such differences was calculated. Various aspects of study design (study size and enrichment) and drug characteristics (active metabolites, PD variability etc) were investigated in each case. The study powers calculated from the simulations where largely consistent with the observed in vivo outcomes and helped to explain the aforementioned literature discrepancies. In conclusion, the simulations described have demonstrated the usefulness of clinical trial simulations, incorporating preclinical information on the genetics of drug metabolism for the prediction of drug pharmacokinetics and dynamics in virtual populations of individuals of varying drug metabolizing capability. In the future, clinical trial simulation may increasingly use prior in vitro data.