High-throughput, semi-automated determination of a cyclooxygenase II inhibitor in human plasma and urine using solid-phase extraction in the 96-well format and high-performance liquid chromatography with post-column photochemical derivatization-fluorescence detection

Compound I, 5-chloro-3-(4-methanesulfonylphenyl)-6'-methyl-[2,3']bipyridinyl, has been found to be a specific inhibitor of the enzyme cyclooxygenase II (COX II). The anti-inflammatory properties of this compound are currently being investigated. HPLC assays for the determination of this analyte in human plasma and human urine have been developed. Isolation of I and the internal standard (II) was achieved by solid-phase extraction (SPE) in the 96-well format. A C-8 SPE plate was used for the extraction of the drug from human plasma (recovery >90%) while a mixed-mode (C-8/Cation) SPE plate was used to isolate the analytes from human urine (recovery approximately 71%). The analyte and internal standard were chromatographed on a Keystone Scientific Prism-RP(R) guard column (20x4.6 mm) connected to a Prism-RP(R) analytical column (150x4.6 mm), using a mobile phase consisting of 45% acetonitrile in 10 mM acetate buffer (pH=4); the analytes eluted at retention times of 5.2 and 6.9 min for I and II, respectively. Compounds I and II were found to form highly fluorescent products after exposure to UV light (254 nm). Thus, the analytes were detected by fluorescence (lambda (ex) = 260 nm, lambda (em) = 375 nm) following post-column photochemical derivatization. Eight point calibration curves over the concentration range of 5-500 ng/ml fur human plasma and human urine yielded a linear response (R-2>0.99) when a 1/y weighted linear regression model was employed. Based on the replicate analyses (n=5) of spiked standards, the within-day precision for both assays was better than 7% C.V. at all points on the calibration curve; within-day accuracy was within 5% of nominal at all standard concentrations. The between-run precision and accuracy of the assays, as calculated from the results of the analysis of quality control samples, was better than 8% C.V. and within 8% of nominal. I was found to be stable in human plasma and urine for at least 8 and 2 months, respectively. In addition, the human plasma assay was semi-automated in order to improve sample throughput by utilizing a Packard liquid handling system and a Tom-Tee Quadra 96 SPE system. The precision and accuracy of the semi-automated procedure were comparable to the manual procedure. Over 5000 clinical samples have been analyzed successfully using these methods. (C) 2001 Elsevier Science B.V. All rights reserved.