Translational pharmacogenomics: a study of warfarin dosing

Abstract

Warfarin is one of the most commonly used oral anticoagulants worldwide and is highly efficacious for the treatment and prevention of thromboembolic disorders. However, due to its narrow therapeutic index, large inter-individual variability in dose requirements, and extensive drug and food interactions, warfarin remains a challenging drug to prescribe. Genetic factors (CYP2C9 and VKORC1), together with clinical factors (age and body weight), account for up to 60% of warfarin dose variance, whereas ~40% variability remains poorly understood. Several warfarin dosing algorithms, comprising genetic and non-genetic covariates have been published over the years. However, none of the published algorithms included patients from Southern Italy. We therefore conducted a candidate-gene study to develop an algorithm for predicting warfarin maintenance dose in patients from the Campania Region (n=266) in Southern Italy. Our pharmacogenetic dosing algorithm consisted of six variables (age, body surface area, amiodarone intake, CYP2C9*2, CYP2C9*3, and VKORC1 -1639G>A). It led to the accurate prediction of warfarin maintenance dose in 44% of patients (mean absolute error 7.41 mg/week). The prediction accuracy of the pharmacogenetic algorithm was superior to three previously published pharmacogenetic algorithms derived from patients in Northern and Central Italy. Given that previous studies suggested a role for miR-133a in warfarin response, we conducted a pilot study comparing baseline serum levels of miRNA in patients who achieved warfarin stable dose (n=10) to those who did not achieve warfarin stability (n=10), using the Affymetrix miRNA array. No association was found between miRNA-133a and warfarin response. Interestingly, circulating levels of miR-548a-3p were observed to be higher in patients who did not achieve warfarin stability (P=0.0053, fold change =1.66) compared to patients who achieved stable dose. In silico analyses showed that several target genes of miR-548a-3p are involved in the coagulation pathway. Work is currently underway to validate and replicate these findings in a larger cohort of prospectively recruited patients initiated onto warfarin therapy (n=980) using TaqMan miRNA real-time quantitative PCR. Pharmacogenetic algorithms have shown that common variants in CYP2C9 and VKORC1 genes cannot fully explain the extreme dose requirements in individuals sensitive and resistant to warfarin. To investigate the role of other genetic variants in these patients with extreme phenotypes, we performed a genome-wide association study (GWAS) comprising of warfarin sensitive patients (≤1.5 mg/day, n=55), warfarin resistant patients (≥10 mg/day, n=51), and healthy controls from the National Blood Service (NBS, n=2,501). Our results suggested that an intergenic variant on chromosome 10, rs4918797, could be involved in warfarin sensitivity. Intronic SNPs in MIR6873 on chromosome 6 (rs114213056) and PIGN on chromosome 18 (rs10163900, rs76455916, rs77118150, and rs79434376) showed suggestive association with warfarin resistance. The findings of this thesis showed that a multitude of factors affect warfarin dosing, some of which still need further investigations. Insights of the roles of other factors such as non-coding RNA and rare genetic variants will hopefully improve dose prediction and drug efficacy and ultimately patient outcomes. The work being undertaken with warfarin acts as a pathfinder project, the concepts from which could be applied to other drugs with variable dose requirement. [edited by author]