The relative contribution of phenotypic measures and CYP2C9-vitamin K epoxide reductase complex subunit 1 (VKORC1) polymorphisms to warfarin dose requirements at day 14 was determined in 132 hospitalized, heavily medicated patients. Phenotypic measures were (1) the urinary losartan metabolic ratio before the first dose of warfarin, (2) the S:R-warfarin ratio at day 1, and (3) a dose-adjusted international normalized ratio (INR) at day 4. CYP2C9 and VKORC1 genotypes were determined by gene chip analysis. In multivariate analyses, the dose-adjusted INR at day 4 explained 31% of variability observed in warfarin doses at day 14, whereas genotypic measures (CYP2C9-VKORC1) contributed 6.5%. When S:R-warfarin ratio was used, genotypes contributed more significantly (23.5%). Finally, urinary losartan metabolic ratio was of low predictive value. The best models obtained explained 51% of intersubject variability in warfarin dose requirements. Thus, combination of a phenotypic measure to CYP2C9-VKORC1 genotypes represents a useful strategy to predict warfarin doses in patients receiving multiple drugs (11 +/- 4 drugs/day).

CYP2C9 metabolizes a wide variety of drugs, including losartan and (S)-warfarin. CYP2C9 is polymorphic and correlation between warfarin clearance and CYP2C9 genotypes has been demonstrated in healthy volunteers and selected patients. The overall objective of our study was to determine, in patients with a multiple drug regimen, correlations between required doses of warfarin and 1) CYP2C9 genotypes, or 2) CYP2C9 phenotype (losartan metabolic ratio). Losartan and its main metabolite EXP 3174 were analysed by HPLC in 6-hour urine samples collected from 77 subjects after a single 12.5 mg oral dose of losartan. The three most common CYP2C9 allelic variants were analysed by PCR-RFLP using genomic DNA of 121 patients. Mean urinary metabolic ratios of losartan were 3.9±3.7 for CYP2C9*1/*1 (n=49), 4.0±5.4 for *1/*2 (n=13), 5.4±4.5 for *1/*3 (n=8), 5.2 for *2/*2 (n=2) and 12.8 for *2/*3 (n=1). A multiple linear regression analysis model was developed using phenotype, age, weight, gender and amiodarone treatment as cofactors. This model explains up to 40% of variability in warfarin dose. In contrast, a genotype analysis correlated with phenotype values only in patients carrying two copies of variant alleles. Our results indicate that in more than 90% of patients, a genotypic approach does not predict required doses of warfarin. On the other hand, CYP2C9 phenotype, when feasible, could represent a more favourable approach to explain intersubject variability in warfarin disposition. Clinical Pharmacology & Therapeutics (2004) 75, P60–P60; doi: 10.1016/j.clpt.2003.11.230