Use of Pharmacogenetics and Clinical Factors To Predict the Maintenance Dose of Warfarin.

Background Initiation of warfarin therapy using trial-and-error dosing can cause bleeding. Clinical factors explain only 20%–30% of the variability in the therapeutic dose of warfarin. Single nucleotide polymorphisms (SNPs) in the cytochrome P450 2C9 (CYP2C9) gene correlate with the clearance of S-warfarin and SNPs in the vitamin K epoxide reductase (VKORC1) gene predict warfarin sensitivity. We test the hypothesis that the combination of clinical and pharmacogenetic information can predict the therapeutic warfarin dose.

Methods We collected DNA, demographic variables, laboratory values, and medication histories from patients taking warfarin. Subjects either attended an outpatient anticoagulation clinic or participated in the PREVENT (prevention of venous thromboembolism) study. After PCR amplification, we used Pyrosequencing® to genotype DNA regions for 2 coding CYP2C9 SNPs, *2 (C430T) and *3 (A1075C), and for 4 noncoding VKORC1 SNPs: C861A, A5808C, G6853C, and G9041A. Using multiple regression, we quantified the association between therapeutic warfarin dose and clinical and genetic factors in a derivation cohort of 900 participants and a validation cohort of 100 participants.

Results The VKORC1 G6853C SNP was the first variable to enter the stepwise regression equation and was associated with a 27% decrease in the warfarin dose per allele in Caucasian patients. The VKORC1 A5808C SNP was associated with a 33% decrease per allele in warfarin dose in African-American patients. Other significant (p < 0.05) predictors of the therapeutic warfarin dose, in order of entry into the regression equation and their effect on warfarin dose were: body surface area (+12% per SD increase), CYP2C9*3 (−33% per allele), CYP2C9*2 (−20% per allele), age (−7% per decade), target INR (+8% per 0.5 unit increase), amiodarone use (−24%), African-American race (+12%), smoker (+9%), and simvastatin or fluvastatin use (−5%). A dosing equation that included these pharmacogenetic and clinical factors explained 52% of the dose variability in derivation cohort and 55% of the variability in the validation cohort.

Conclusions The therapeutic warfarin dose can be estimated from clinical and pharmacogenetic factors that can be obtained when warfarin is started. Use of this dosing equation has potential to aid in the prediction of an optimal warfarin dose, which may decrease the risk of bleeding during the initiation of warfarin therapy.