Major determinants of outcome and dosing in warfarin treatment

Abstract

The aim of this thesis was to identify factors that determine individual patients sensitivity to the anticoagulant warfarin, and to quantify the effect of such factors on different measures of anticoagulation control. As part of this project, a large bio-bank was prospectively collected, with DNA and clinical data from more than 1500 patients starting warfarin treatment (the WARG cohort). To facilitate data retrieval and monitoring in this multi-centre trial, we developed two internet-based study interfaces, a browser-based protocol for manual data entry and a protocol for automated data extraction from an existing medical record system. All data was stored in a central database and monitoring of the data was performed in real-time via a browser-based monitoring interface. The Internet-based study tools proved efficient and safe, with a potential to improve quality and cost-effectiveness of future multi-centre studies.

In the WARG cohort, the incidence of first-time severe bleeding (according to the WHO criteria for severe adverse drug reactions) was 2.3 per 100 patient-years. Male gender and use of drugs potentially interacting with warfarin both increased the risk of severe bleeding, with odds ratios of 2.8 and 2.3, respectively. The incidence of severe bleeding was 2.4 times higher during the first month of treatment, compared to any time hereafter.

To evaluate the influence of genetic factors on the outcome of warfarin treatment, we analysed the DNA samples from the WARG cohort. Out of 29 polymorphic genes analysed, only two were clearly associated with warfarin dose requirements and anticoagulation control. These genes were VKORC1 (coding for warfarin s target molecule), and CYP2C9 (important for the elimination of S-warfarin). During initiation of therapy, polymorphisms in VKORC1 and CYP2C9 increased the risk of over-anticoagulation significantly. The effect was most pronounced in the individuals homozygous for the CYP2C9*3 allele, with a hazard ratio of 21.8. The corresponding hazard ratio in patients homozygous for the VKORC1 haplotype A was 4.6. Carriers of VKORC1 haplotype A also reached therapeutic levels of anticoagulation more rapidly than others. An extended analysis of the association between CYP2C9 genotype and the risk of over-anticoagulation showed that the association was strong during the first two weeks of warfarin therapy, but abolished by the third week. Both VKORC1 and CYP2C9 genotypes were nominally associated with anticoagulation stability, measured as time spent within the therapeutic INR interval. However, this association was no longer significant after statistical correction for multiple testing.

The association between CYP2C9 genotype and warfarin dose requirements was further investigated in a meta-analysis, pooling data from 39 published studies to provide precise estimates of the gene-dose effect. Compared to the wild-type genotype (CYP2C9*1/*1), individuals with the *1/*2 genotype required doses that were 19.6% lower. Corresponding values were 33.7% for the *1/*3 genotype, 36.0% for *2/*2, 56.7% for *2/*3, and 78.1% for *3/*3. Although warfarin has been on the market for more than 50 years, the knowledge of the drug s effects in clinical use is still expanding, as demonstrated in our studies.