Mechanistic Prediction of HIV Drug-Drug Interactions in Virtual Populations from in vitro Enzyme Kinetic Data: Ritonavir and Saquinavir.
Almond L1, Rowland Yeo K1, Howgate EM1, Tucker GT1,2, Rostami-Hodjegan A1,2
Simcyp Limited1 John Street Sheffield S2 4SU, UK and University of Sheffield, Academic Unit of Clinical Pharmacology2, Royal Hallamshire Hospital, Sheffield S10 2JF, UK.
Objectives: Quantitative prediction of metabolic drug-drug interactions (mDDI) and the ability to identify patients most likely to experience such interactions is of clinical benefit. Although this is challenging, particularly in patient groups receiving multiple drugs, (e.g. HIV-infected individuals), the extrapolation of in vivo ADME properties from in vitro data (IVIVE) in simulated virtual populations provides a useful framework to assess mDDI. Such prediction requires estimates of the inhibition constant, Ki, and of the concentration of inhibitor [I] at the enzyme active site. Ritonavir (RTV), a potent inhibitor of cytochrome P450 3A4 (CYP3A4), is used in many regimens as a pharmacoenhancer of other protease inhibitors metabolised by CYP3A4, including saquinavir (SQV). Here, we report prediction of the magnitude of the interaction observed in a multiple dose study [1] of healthy volunteers taking either SQV alone (800mg b.d. fortovase; n=8) or SQV/RTV (800/300mg b.d.).
Methods: In vitro kinetic data (Ki) were taken from multiple sources and corrected for non-specific microsomal binding. Virtual trials (n = 10), mimicking the study design of the in vivo study (n = 8), were simulated (Simcyp Version 7.0, www.simcyp.com) assuming the absence and presence of CYP3A4 induction by ritonavir, as well as its inhibitory effect.
Results: Observed changes in maximum saquinavir plasma concentration (Cmax) and the area under the concentration-time curve (AUC) were 10.0 and 21.5-fold, respectively. The corresponding predicted values for the 10 simulated trials ranged from 6.9 to 13.4 and from 9.6 to 19.9, respectively. The mean fold error in prediction of Cmax ratios was 0.9 and all values were within 2-fold of corresponding observed values. The mean fold error in prediction of AUC ratios was 0.7 with 9/10 predicted values within 2-fold of the observed data. Assuming simultaneous induction of CYP3A4 by RTV led to under prediction of the interaction (only 4 of 10 trials predicted the fold change in Cmax and none of predicted change in AUCs were within 2-fold of observed changes).
Conclusions: IVIVE accurately predicted the interaction of SQV (fortovase) with RTV, indicating that IVIVE is a useful tool for assessing HIV mDDI. The simulations did not indicate a significant contribution to the interaction from enzyme induction at a RTV dose of 300mg.
References:
[1] Buss N et al. (2001) Saquinavir and ritonavir pharmacokinetics following combined ritonavir and saquinavir (soft gelatin capsules) administration. Br J Clin Pharmacol. 52(3):255-64.
