Failure of pharmacogenetic-based dosing algorithms to identify older patients requiring low daily doses of warfarin.

INTRODUCTION: Warfarin doses vary greatly among patients and warfarin administration is accompanied by risk of bleeding. Genes responsible for its metabolism (CYP2C9) and effect on clotting (VKORC1) have been identified. It has been suggested that genotyping for variants in these genes can improve warfarin dosing and decrease bleeding complications. We evaluated performance of pharmacogenetic-based warfarin dosing estimation algorithms in old and very old patients. METHODS: Cross-sectional study of stable patients older than 65 years receiving warfarin with therapeutic International Normalized Ratio (INR) anticoagulation. Medical and laboratory data were reviewed and genotyping for CYP2C9 and VKORC1 performed. Warfarin dose estimates with and without genotype information were compared to clinically established therapeutic doses. RESULTS: Sixty-nine patients (32 men, 37 women; 41 nursing home residents; 28 senior care community residents) aged 81.4 ± 8.3 (mean ± S.D) years; ethnicity: Caucasian in 53, Asian in 10, Hispanic in 4, and African American in 2, received 3.3 ± 1.7 mg/d (range 0.7-9) warfarin achieving target INRs of 2.5 ± 0.2. Pharmacogenetic-based dose estimates (in combination with age, weight, height, smoking history, ethnicity/race, history of liver disease, selected co-medications such as amiodarone and enzyme inducers, baseline INR, clinical indication, and target INR), explained 50% of variability (P < .0001) compared with 12% without genetic data (P = .003). However, doses were overestimated in 15 of 16 patients requiring less than 2 mg/d (2.6 ± 0.9 mg/d compared with observed 1.5 ± 0.3 mg/d, P = .0001). Renal disease was a potential variable contributing to low warfarin requirements. DISCUSSION: The role of pharmacogenetic testing in the management of warfarin administration in patients is undergoing evaluation. Currently available pharmacogenetic- based warfarin dose estimation algorithms reduce variability in estimates for groups of older patients but consistently overestimate doses for older patients requiring the lowest doses of warfarin. CONCLUSIONS: Pharmacogenetic data add to our understanding of variability in warfarin dosing requirements but do not accurately identify older patients requiring the lowest warfarin doses. Therefore, the most prudent approach to warfarin therapy in older patients should include low initial doses in the absence of genotype variants associated with very low warfarin sensitivity and careful monitoring of INR responses.