Physiologically-based pharmacokinetic (PBPK) models for prediction of saquinavir effect on midazolam pharmacodynamics

Abstract

Purpose: To predict the drug-drug interaction effect of saquinavir on midazolam pharmacodynamics (PD).

Methods:The absorption and pharmacokinetics (PK) of midazolam and saquinavir were simulated using a beta version of GastroPlusTM8.0 (Simulations Plus, Inc., Lancaster, CA). Independent saquinavir and midazolam models were previously validated by comparing the simulated plasma concentration-time (Cp-time) profiles with experimental data for intravenous (i.v.) and oral (p.o.) administration for each drug, as well as by comparing the prediction of the observed drug-drug interaction (DDI) of saquinavir on midazolam PK. Models were fitted for several reported PD responses after midazolam administration based on the drug’s unbound Cp-time profiles. The fitted PD models were then used with the earlier prediction of saquinavir’s effect on midazolam PK to estimate the drug-drug interaction effect of saquinavir on midazolam PD.

Results: Baseline PD responses for midazolam (i.e., without saquinavir), such as the digit symbol substitution test (DSST) and the Maddox Wing Test (MWT), were equally well described by several different types of direct or indirect PD models, each resulting in a different prediction of the magnitude of saquinavir’s effect on midazolam PD. Several models predicted the maximum response within 20% of the observed maximum response; however, the prediction error was as high as 40% with other models. In the absence of PD response data for different dose levels of midazolam, there was no obvious most relevant baseline PD model for each response.