Influence of Sampling on the Determination of Warfarin and Warfarin Alcohols in Oral Fluid

BACKGROUND AND OBJECTIVE The determination of warfarin, RS/SR- and RR/SS-warfarin alcohols in oral fluid may offer additional information to the INR assay. This study aimed to establish an optimized sampling technique providing the best correlation between the oral fluid and the unbound plasma concentrations of these compounds. MATERIALS AND METHODS Samples of non-stimulated and stimulated oral fluid, and blood were collected from 14 patients undergoing warfarin therapy. After acidification, analytes were extracted with a dichloromethane/hexane mixture and determined by HPLC with fluorescence detection. Plasma samples were also ultrafiltered for the determination of the unbound fraction. The chromatographic separation was carried out in isocratic conditions with a phosphate buffer/methanol mobile phase on a C-18 reversed-phase column. The absence of interfering compounds was verified by HPLC-ESI-Q-TOF. RESULTS Stimulation generally increased the oral fluid pH to values close to blood pH in about 6 minutes. The concentration of warfarin and RS/SR-warfarin alcohols in oral fluid followed the same trend, whereas the concentration of RR/SS-warfarin alcohols was not affected. Six minute stimulation with chewing gum followed by collection with a polyester swab was the best sampling procedure, with a good repeatability (RSD < 10%) and relatively low inter-subject variability (RSD  = 30%) of the oral fluid to plasma ratio. This procedure provided strong correlations between the measured oral fluid and unbound plasma concentration of warfarin (r  =  0.92, p < 0.001) and RS/SR-warfarin alcohols (r  =  0.84, p < 0.001), as well as between stimulated oral fluid and total plasma concentration of warfarin (r  =  0.78, p < 0.001) and RS/SR-warfarin alcohols (r  =  0.81, p < 0.001). CONCLUSION The very good correlation between oral fluid and unbound plasma concentration of warfarin and RS/SR-warfarin alcohols suggests that oral fluid analysis could provide clinically useful information for the monitoring of anticoagulant therapy, complementary to the INR assay.