Measurement of warfarin in the oral fluid of patients undergoing anticoagulant oral therapy.

BACKGROUND: Patients on warfarin therapy undergo invasive and expensive checks for the coagulability of their blood. No information on coagulation levels is currently available between two controls. METHODOLOGY: A method was developed to determine warfarin in oral fluid by HPLC and fluorimetric detection. The chromatographic separation was performed at room temperature on a C-18 reversed-phase column, 65% PBS and 35% methanol mobile phase, flow rate 0.7 mL/min, injection volume 25 µL, excitation wavelength 310 nm, emission wavelength 400 nm. FINDINGS: The method was free from interference and matrix effect, linear in the range 0.2-100 ng/mL, with a detection limit of 0.2 ng/mL. Its coefficient of variation was <3% for intra-day measurements and <5% for inter-day measurements. The average concentration of warfarin in the oral fluid of 50 patients was 2.5±1.6 ng/mL (range 0.8-7.6 ng/mL). Dosage was not correlated to INR (r = -0.03, p = 0.85) but positively correlated to warfarin concentration in the oral fluid (r = 0.39, p = 0.006). The correlation between warfarin concentration and pH in the oral fluid (r = 0.37, p = 0.009) confirmed the importance of pH in regulating the drug transfer from blood. A correlation between warfarin concentration in the oral fluid and INR was only found in samples with pH values ≥7.2 (r = 0.84, p = 0.004). CONCLUSIONS: Warfarin diffuses from blood to oral fluid. The method allows to measure its concentration in this matrix and to analyze correlations with INR and other parameters.

Prediction of hERG potassium channel affinity by the CODESSA approach.

The problem of predicting torsadogenic cardiotoxicity of drugs is afforded in this work. QSAR studies on a series of molecules, acting as hERG K+ channel blockers, were carried out for this purpose by using the CODESSA program. Molecules belonging to the analyzed dataset are characterized by different therapeutic targets and by high molecular diversity. The predictive power of the obtained models was estimated by means of rigorous validation criteria implying the use of highly diagnostic statistical parameters on the test set, other than the training set. Validation results obtained for a blind set, disjoined from the whole dataset initially considered, confirmed the predictive potency of the models proposed here, so suggesting that they are worth to be considered as a valuable tool for practical applications in predicting the blockade of hERG K+ channels.