Determination of midazolam and its hydroxy metabolites in human plasma and oral fluid by liquid chromatography/electrospray ionization ion trap tandem mass spectrometry

Abstract

Midazolam (MDZ), a short-acting benzodiazepine, is a widely accepted probe drug for CYP3A phenotyping. Published methods for its analysis have used either therapeutic doses of MDZ, or, if employing lower doses, were mostly unable to quantify the two hydroxy metabolites. In the present study, a sensitive and specific liquid chromatography/electrospray ionization tandem mass spectrometry method was developed and validated for the quantitative determination of MDZ and two of its metabolites (1'-hydroxymidazolam (1'-OHMDZ) and 4-hydroxymidazolam (4-OHMDZ)) in human plasma and oral fluid. After liquid-liquid extraction with hexane/dichloromethane (73:27, v/v), the analytes were separated on a Luna C18(2) (100 x 2.1 mm) analytical column using gradient elution. Detection was achieved using tandem mass spectrometry on an ion trap mass spectrometer. Midazolam-d6 was used as internal standard for quantification. The calibration curves were linear (R2 >0.998) between 0.05 and 20 ng/mL for MDZ and both metabolites in both matrices. Using 1 mL samples, the limit of detection was 0.025 ng/mL and the limit of quantification was 0.05 ng/mL for MDZ and the hydroxy metabolites in both matrices. Intra- and inter-day accuracies, determined at three different concentrations, were between 92.1 and 102.3% and the corresponding coefficients of variation were <7.3%. The average recoveries were 90.6%, 86.7% and 79.0% for MDZ, 1'-OHMDZ and 4-OHMDZ in plasma and 95.3%, 96.6% and 86.8% for MDZ, 1'-OHMDZ and 4-OHMDZ, respectively, in oral fluid. The method was successfully applied to a pharmacokinetic study, showing that MDZ and its hydroxy metabolites can be determined precisely in in vivo samples obtained following a single oral or intravenous dose of 2 mg MDZ. The method appears to be useful for CYP3A phenotyping in plasma using sub-therapeutic MDZ doses, but larger studies are needed to test this assumption.