Prediction of drug-drug interactions and their associated variability in human populations: Application to erlotinib and its coadministration with ketoconazole and rifampicin

PDF of poster

Background: Drug–drug interactions (DDIs) mediated by cytochrome P450 enzymes are a potential cause of toxicity with co-medications. For this reason the quantitative prediction of DDIs in general as well as for the individual is of great importance. SimCYP^® is a commercially available computer-based tool developed for this purpose [1]. This software enables the prediction of the metabolic clearance of drugs (before and after administration of an inhibitor or inducer) in human populations using in vitro metabolism and inhibition data [2,3,4].

Objectives: The aim was to investigate the ability of SimCYP^® to predict clinical DDIs (i.e. AUC ratio of the substrate with and without the inhibitor/inducer) and their associated variability using erlotinib (Tarceva^TM, invented by OSI Pharmaceuticals; co-developed by OSI Pharmaceuticals, Genentech and Roche).

Methods: A SimCYP^® model for erlotinib was developed using available in vitro data. The prediction of clearance for erlotinib and the influence of inhibition by ketoconazole and induction by rifampicin on its metabolism were assessed using the model. Simulations were performed using a virtual population (demographic characteristics: North–European Caucasians aged between 20 and 50), constituting 100 trials of 10 subjects. Simulated data were analysed and compared with clinical data. Sensitivity analyses were used to determine the impact of any parameter uncertainty on the simulated results.

Results/Conclusion: The population-predicted CL/F for erlotinib was similar to observed values (obs. CL/F: 5.2-15L/hr; pred. CL/F: 9L/hr). In agreement with clinical observations, SimCYP^® predicted a mild interaction with ketoconazole (obs. AUC ratio: 1.9, mean, 1.5-2.4, 90% CI; pred. AUC ratio: 2.0, median, 1.3-3.3, 90 percentile range) and a mild induction by rifampicin (obs. AUC ratio: 0.33, mean, 0.26-0.41, 90% CI; pred. AUC ratio: 0.40, median, 0.23-0.64, 90 percentile range). The sensitivity analyses indicated that certain parameters (e.g. in vitro inhibition constant) had a greater impact on the simulation than others (e.g. absorption rate constant). This work shows that SimCYP^® has the potential to be used successfully for the prediction of DDIs and could also be used to assist in the design of clinical trials (i.e. to power a DDI study).

References:
[1]   http://www.simcyp.com/
[2]   Houston JB. Biochem Pharmacol. (47) 1994, 1469-1479
[3]   Ito K et al. Pharmacol Rev (50) 1998, 387-411.
[4]   Tucker GT et al. Eur J Pharm Sci. (13) 2001, 417-428.