Review of an Article: The international Warfarin Pharmacogenetics Consortium (2009). Estimation of the warfarin dose with clinical and pharmacogenetic data. NEJM 360 (8): 753-64

Abstract

This article examines the use of clinical and genetic data to determine a dose algorithm for each, as well as a combined algorithm for dosing. Data for 4,043 patients were used. Only stable doses, not initial doses were considered. A validation cohort of 1,009 subjects was used to evaluate the clinical value of each of the three algorithms. Dosing ranges evaluated were those of 21 mg of warfarin per week or less and those requiring 49 mg of warfarin per week or more to achieve the target international normalized ratio (INR). It was felt that clinical factors, demographic variables and variations in the CYP2C9 and VKORC1 genes contribute significantly to variations in dose requirements. With this information in mind, researchers recruited 21 research groups from 9 countries. The groups contributed clinical and genetic data for 5,700 patients treated with warfarin. Analysis included those patients whose target INR was reached at 2-3. Variables such as smoking and vitamin K intake were not available. The researchers used 80% of the eligible patients (4,043) who had a stable INR of 23 to develop dose prediction models. Their modeling strategies are too complex for discussion in this review. The study included a pharmacogenetic, clinical and fixed-dose algorithm. The pharmacogenetic model provided dose estimates that were significantly closer to the actual doses than the estimates of the other two models. The addition of genetic information to the clinical information decreased the error in the dose estimate, above that of the clinical or fixed-dose approach. This finding was true for both the high and the low doses of warfarin (which was 46% of the entire cohort). However, the study and analysis did not address the issue of a precise initial dose of warfarin and its effect on clinical endpoints. The authors suggest a prospective trial to detect the benefits of incorporating pharmacogenetic information into the dose algorithm for patients who require high or low doses of warfarin. The subgroups in this study differed significantly from those based on the clinical algorithm. The authors list several limitations. They did not have enough data to include factors such as smoking, vitamin K intake or inherited genetic factors other than the ones included in this study. They had no data reporting adverse events before dose stabilization. In addition, they restricted their algorithms to those with a target INR of 2-3. Finally, only a small number of the patients were under age 40. Their conclusions: A pharmacogenetically devised dose algorithm for warfarin that used the genotypes mentioned and clinical variables predicts the stable therapeutic dose of warfarin better than a clinical or fixed-dose algorithm. The pharmacogenetically devised dose algorithm was most helpful in patients receiving high- and low-dose warfarin. Future dosing possibly will be guided by additional genetic data.