Warfarin: Inclusion of genotypes important for African Americans leads to better dose prediction

Published: December 15th, 2014

Category: Stories

Warfarin is used widely to prevent thromboembolic events. Numerous studies have demonstrated that CYP2C9 and VKORC1 genotypes influence warfarin dosage requirements. However, the Clarification of Optimal Anticoagulation through Genetics (COAG) trial recently found that use of a pharmacogenomic dosing algorithm incorporating genetic and clinical information (www.warfarindosing.org [WD]) led to an increased likelihood of warfarin overdosing in African American patients compared to an algorithm with only clinical information. These results have raised questions regarding use of genotype-guided warfarin dosing algorithms in African Americans, specifically regarding the accuracy of those that do not include alleles important in this patient population.

Drozda* and colleagues examined the relationship between the CYP2C9*5, CYP2C9*6, CYP2C9*8, and CYP2C9*11 alleles and rs12777828 G > A genotype and warfarin dose prediction error with dosing algorithms used in clinical trials. Investigators genotyped 274 African Americans (by self report) who were on a stable warfarin dose (therapeutic INR for at least 2 consecutive clinic visits for at least 14 days apart) and compared this dose to that predicted with the WD algorithm used in the COAG trial and the International Warfarin Pharmacogenetics Consortium (IWPC) algorithm.

The mean warfarin maintenance dose in the study cohort was 6.6 ± 2.5 mg daily. Overall, the use of the IWPC algorithm predicted lower doses (6.1 ± 1.3 mg/day; P < 0.0001) and the WD algorithm predicted higher doses (7.2 ± 1.8 mg/day; P < 0.0001) than those observed.

When observed and predicted doses were compared within each genotype strata, the WD algorithm overestimated doses by a median of 1.2 mg/day in rs12777823 heterozygotes (P < 0.001); 2.0 mg/day in rs12777823 homozygotes (P = 0.004); and 2.2 mg/day in CYP2C9 variant carriers (P < 0.001). The IWPC algorithm underdosed warfarin by 0.8 mg/day in patients with rs12777823 GG genotype (P < 0.001) and overdosed warfarin by 0.7 mg/day in CYP2C9 variant carriers (P = 0.04). Modification of the WD algorithm to include these variants led to significantly better dose prediction.

Clinical Implications

Up to 50% of African Americans have a genotype associated with a lower dose requirement that was not accounted for in the COAG trial. Results of the current study demonstrate that dosing algorithms that were used in recent large, randomized controlled trials did not perform well in African Americans who have such genetic variants. “Although it had been suggested that the limited number of genotypes considered in the [COAG] trial contributed to poor outcomes in African Americans, our data are the first to support this and to show the degree to which failure to account for variants important in African Americans affects dosing error,” study authors wrote.

To account for African-specific genotypes when dosing warfarin, authors recommend using an algorithm derived from an African American population that contains the studied variants. Alternatively, authors suggested use of the WD or IWPC algorithms with further dose adjustments made to account for genotypes important in this population.

Find more information about warfarin pharmacogenomics and links to the WD and IWPC dosing algorithms in the Pharmacogenomics Knowledgebase.


Drozda K et al. Poor warfarin dose prediction with pharmacogenetic algorithms that exclude genotypes important for African Americans. Pharmacogenet Genomics. 2014 Dec 1. [Epub ahead of print]

*Editorial note: UF Health Personalized Medicine Program Associate Director Larisa Cavallari is an author on this manuscript.