Predicting warfarin dose

Recent evidence has shown that genetic polymorphisms in genes affecting warfarin metabolism and mechanism of action influence its dose. It has been suggested that dosing algorithms incorporating genetics might outperform usual care but it is still unclear whether such an approach has a definitive clinical advantage.

In this article we review: the current knowledge on the clinical and genetic determinants of warfarin dosing and summarize how such determinants have been incorporated in clinical practice; the main pharmacogenetics-based predictive models and their performance; and the potential clinical role of recently developed alternative predictive models not incorporating genetic information. Finally, we provide an overview on the pending issues in this area.

In spite of the widespread enthusiasm, pharmacogenetics-based models for predicting warfarin requirements only account for 35-50% of the variability in its dose. Recently, forecasting models using early responses to standardized fixed-dose warfarin initiation nomograms have been developed with performances similar or superior to pharmacogenetics-based models and without the need for costly genetic testing. Further studies are needed to compare these strategies. On the basis of current evidence, we cannot recommend the routine use of pharmacogenetics-based models in clinical practice.