Genotypes of the cytochrome p450 isoform, CYP2C9, and the vitamin K epoxide reductase complex subunit 1 conjointly determine stable warfarin dose: a prospective study.

BACKGROUND: Warfarin has a narrow therapeutic range and wide inter-individual dosing requirements that may be related to functional variants of genes affecting warfarin metabolism (i.e., CYP2C9) and activity (i.e., vitamin K epoxide reductase complex subunit 1-VKORC1). We hypothesized that variants in these two genes explain a substantial proportion of variability in stable warfarin dose and could be used as a basis for improved dosing algorithms.

METHODS: Consecutive consenting outpatients (n = 213) with stable INR (2-3) for >1 month were enrolled. Buccal DNA was extracted using a Qiagen mini-column and CYP2C9*2 and VKORC1 genotyping performed by the Taqman 3' nuclease assay. Sequencing for CYP2C9*3, genotyping was done using Big Dye v3.1 terminator chemistry Dose by genotype was assessed by linear regression.

RESULTS: Weekly warfarin dose averaged 30.8 +/- 13.9 mg/week; average INR was 2.42 +/- 0.72. CYP2C9*2/*3 genotype distribution was: CC/AA (wild-type [WT]) = 71.4%, CT/AA = 18.3%, CC/AC = 9.4%, and CT/AC = 1%; VKORC1 genotypes were CC (WT) = 36.6%, CT = 50.7%, and TT = 12.7%. Warfarin doses (mg/week) varied by genotype: for CYP2C9, 33.3 mg/week for WT (CC/AA), 27.2 mg/week for CT/AA (P = 0.04 vs. WT), 23.0 mg/week for CC/AC (P = 0.003), and 6.0 mg/week for CT/AC (P < 0.001), representing dose reductions of 18-31% for single and 82% for double variant carriers; for VKORC1: 38.4 mg/week for WT (CC), 28.6 mg/week for CT (P < 0.001 vs. WT), 20.95 mg/week for TT (P < 0.001). In multiple linear regression, genotype was the dominant predictor of warfarin dose (P = 2.4 x 10(-15)); weak predictors were age, weight, and sex. Genotype-based modeling explained 33% of dose-variance, compared with 12% for clinical variables alone.

CONCLUSION: In this large prospective study of warfarin genetic dose-determinants, carriage of a single or double CYP2C9 variant, reduced warfarin dose 18-72%, and of a VKORC1 variant by 65%. Genotype-based modeling explained almost one-half of dose-variance. A quantitative dosing algorithm incorporating genotypes for 2C9 and VKORC1 could substantially improve initial warfarin dose-selection and reduce related complications.