We have previously described a midazolam limited sampling strategy employing a population pharmacokinetic (PK) approach to estimate constitutive cytochrome P450 (CYP) 3A activity. This study evaluated expansion of this approach to estimate CYP3A constitutive, inhibitory, and induction activities. Midazolam concentrations (n = 4441) from adults (n = 152) were obtained from previous studies after single, oral, or intravenous administration with intensive sample collection. Data were fit to a 2-compartment population PK model that incorporated CYP3A conditions as covariates for clearance (CL), volume of distribution, and bioavailability (F). Limited sampling models using single- or 2-time point concentrations were compared with full PK profiles using the empiric Bayesian post hoc estimations of midazolam area under the plasma concentration-time curve derived from the population PK model. Ketoconazole, rifampin, and pleconaril were significant covariates of CL, while ketoconazole, rifampin, and grapefruit juice were significant covariates for F. Typical midazolam CL and F estimates were 32.9 L/h and 0.31 for the constituent state, while the ratio of inducer/inhibitor for midazolam CL and CL/F for the induced/inhibited (rifampin/ketoconazole) states were 14.2 and 85.3. Upon comparison to the population PK model, the majority of evaluated single- and 2-time point limited sampling models estimated area under the plasma concentration-time curve had unacceptable r2 and/or unacceptable bias and precision. Exclusively during CYP3A inhibitory conditions, the 4- and 6-hour limited sampling model had acceptable limits of r2 , bias, and precision. Consequently, development of a single- or 2-time point midazolam limited sampling model for general, widespread use to simultaneously evaluate various CYP3A conditions remains elusive.
Keywords: cytochrome P450 3A; drug-drug interaction; midazolam; phenotyping.
© 2019, The American College of Clinical Pharmacology.