VKORC1, CYP2C9 and CYP4F2 pharmacogenetics: relevance in warfarin maintenance dose prediction among italian patients.

Background. Warfarin is a widely used anticoagulant with a narrow therapeutic index and relevant inter-individual variability in dosing requirements. The present study was conducted to ascertain the influence of pharmacogenetics on the maintenance dose of warfarin. Methods. We retrospectively selected 437 Italian patients (277 males and 160 females, mean age 71.9yrs, range 28.9-92.8 yrs) under stable warfarin therapy (target INR=2.5). The indications for oral anti-coagulation were: atrial fibrillation and venous thromboembolism. Clinico-demographic features as well as information on smoking habit, alcohol and coffee consumption were collected. The gene polymorphisms of VKORC1 (-1639G>A SNP), CYP2C9 (*1,*2 and *3 alleles)§ and CYP4F2 (*1 and *3 alleles)§ were analyzed by means of PCR with TaqMan fluorescent chemistry starting from whole blood extracted DNA. In plasma samples ALT, GGT and albumin levels were measured. Results. The mean weekly warfarin dose (WWD) (range 3.75 - 80 mg), higher in males than females (t=3.47; p<0.001), was associated inversely with age (F=10.96; p<0.0001) and directly with Body Surface Area (BSA) (F=15.39; p<0.0001), coffee consumption (t=2.50; p<0.05) and ALT levels (r=0.199, p<0.0001). WWD was associated with VKORC1, CYP2C9 and CYP4F2 polymorphisms (F=127.68; p=<0.0001, F=28.80; p<0.0001 and F=6.52; p<0.005 respectively) being lower in VKORC1 and CYP2C9 mutated homozygotes, intermediate in heterozygotes and higher in wild type homozygotes (CYP2C9*1*1 or VKORC1 G/G). By contrast it was lower in CYP4F2*1*1 homozygotes compared to subjects bearing at least one CYP4F2*3 mutated allele. Multiple linear regression analysis was performed and a pharmacogenetic algorithm for warfarin maintenance dose prediction was derived and validated: the R2 adj was 65.4% and 55.5% while the mean Absolute Error (± SE) was 6,80 ± 0,38 and 7,03 ± 0,58 mg/week in the derivation and validation cohorts respectively. The algorithm included as significant predictors VKORC1, CYP2C9 and CYP4F2 polymorphisms, BSA and age. The percentage of cases whose predicted WWD was within 20% of the actual maintenance dose was 51.8% considering patients overall; taking into account patients requiring low (43.75 mg/week) maintenance dose, the percentages were 36.2%, 66.2% and 55.4% respectively. Overall the pharmacogenetic algorithm could correctly assign 73.8% and 63.2% of patients to the low and high dose regimens respectively. Conclusions. The pharmacogenetic algorithm developed in this study, new with respect to others in that it comprises CYP4F2 polymorphisms, allows the correct prediction of low or high warfarin dose regimens in the great majority of cases. Patients in these categories may therefore benefit greatly from a pre-treatment application of the maintenance dose prediction. § according to the Human Cytochrome P450 (CYP) Allele Nomenclature Committee (http://www.imm.ki.se/CYPalleles/).