ABSTRACT
A precise and specific robotics-based liquid chromatographic (LC) method for measuring atovaquone concentrations in plasma was developed and validated, and the method was compared with an existing manual LC method. The compound was isolated from plasma by liquid-liquid extraction, separated by reversed-phase LC, and quantitated against an internal standard with UV detection. Least-squares linear regression with 1/concentration2 weighting was used as the calibration model. The range of the calibration curve for the assay under routine conditions was 0.25-50 micrograms ml-1. No endogenous interferences with the compound or the internal standard were noted in either untreated human plasma or in plasma from patients enrolled in Phase III clinical trials of atovaquone. The accuracy of the assay (determined as the percent bias) ranged from -4.8% to -9.4% in the validation runs. The intra- and interassay precisions (determined as the relative standard deviation) were less than 6.8% and 6.4%, respectively. The contribution of an internal standard on assay accuracy and precision also was examined. Interassay variability was marginally improved by the incorporation of an internal standard to the assay; accuracy and intra-assay precision were essentially unchanged. A paired t-test between estimates of atovaquone concentrations in healthy volunteer and HIV + patient human plasma samples assayed by the automated and manual methods demonstrated no significant difference (p = 0.31) between the values determined by each method.
