Optimizing plasma sampling of oral (PO) midazolam (MDZ) as a biomarker for CYP3A activity

Abstract

Background MDZ AUC is used to measure CYP3A activity and requires 5 to 8 blood samples. The aim of this study was to determine if minimized plasma sampling can be used to assess baseline CYP3A activity with PO MDZ. Methods 45 healthy adults taking no drugs known to affect CYP3A were included (data from 4 previous studies). Subjects received PO MDZ 0.075mg/kg. Plasma concentrations were collected at various time points (Chr) and analyzed by LC/MS/MS. MDZ AUC0-∞ was determined by non-compartmental analysis. MDZ C0.5, 1, 2, 4, and 6 and AUC0-∞ from 20 randomly selected subjects were log-transformed into normally distributed data and used to generate models to predict MDZ AUC (AUCpred). Models were validated using the MDZ data of the remaining 25 subjects and evaluated for bias and precision using mean prediction error (MPE) and root mean square error (RMSE) with acceptable limits of ≤5% and ≤15%, respectively. Results (see Table) Conclusions Two limited sampling models, using MDZ C0.5 and 6 or MDZ C0.5, 2, and 6 were acceptable and predictive of MDZ AUC. These models could reduce costs and inconvenience of greater numbers of samples when PO MDZ is used as a CYP3A biomarker. Clinical Pharmacology & Therapeutics (2005) 77, P32–P32; doi: 10.1016/j.clpt.2004.12.015 Sampling Time (hr) Log of AUCpred R2 %MPE %RMSE 0.5, 6 1.67 + 0.21(logC0.5) + 0.66(logC6) 0.97 −1.0 12.1 0.5, 2, 6 1.11 + 0.21(logC0.5) + 0.43(logC2) + 0.3(logC6) 0.99 −3.5 13.1