Genotype-guided warfarin therapy: Still of only questionable value two decades on.

Abstract

Despite an apparently sound pharmacological basis, clinical studies of genotype-guided warfarin dosing have yielded mixed and conflicting results, leading to reluctance in its clinical implementation. The objective of this critique is to re-evaluate key warfarin pharmacogenetic studies with a view to explaining why this may be so. Major widely-cited warfarin pharmacogenetic studies as well as recent meta-analyses were identified and a critical analysis of these was undertaken to identify factors that may account for poor clinical implementation of pre-treatment genotyping. Critical examination of major warfarin pharmacogenetic studies identified a number of methodological concerns such as marked variations in study designs with different variously-defined measures of outcome. Genotype testing involved only a limited number of CYP2C9/VKORC1 alleles. Claims of benefits of genotyping are based almost exclusively on INR-related parameters which are known to be highly time-labile and of limited value in predicting clinical risk or benefit. This is evidenced by lack of any significant effect of genotyping on rates of bleeding or thromboembolic events. Neither have the effects of potential phenoconversion or medication non-adherence in study populations been adequately investigated. Although the effect of ethnicity/race is now better characterised, studies lack the power to determine whether any benefits claimed are indication-sensitive. Since 60% of inter-individual variability in warfarin dose/response is due to other factors (many of which are non-genetic), expectations of eliminating this variability simply by CYP2C9/VKORC1 genotyping are over-optimistic and efforts cost-ineffective. Real-world studies have not always corroborated trials-based claims of clinical benefit. It is time to consider redirecting scarce resources away from the study of warfarin pharmacogenetics to pharmacogenetic research of potentially greater clinical relevance.