A coding VKORC1 Asp36Tyr polymorphism predisposes to warfarin resistance

Reduced warfarin dose requirements are dictated by the CYP2C9*2 and *3 genetic variants due to decreased S-warfarin metabolic clearance and by the VKORC1*2 haplotype (or H1 and H2) with decreased VKORC1 activity and lower ambient reduced vitamin K levels.^1–7  Variants affecting the intermediate range of warfarin doses include the VKORC1*3 and *4 (or H7, H8, and H9) haplotypes.^4–5  At the other end of the dose spectrum, several rare VKORC1 mutations have been described, including the original report on 4 distinct mutations in warfarin-resistant individuals,⁸  a rare Val66Met mutation associated with high warfarin doses,^9–10  and an Asp36Tyr mutation described in 2 patients with doses of 40 to 50 mg/wk¹¹  However, genetic effects on the highest warfarin doses (more than 80 mg/wk), which are met in clinical practice, have been less well characterized. In an attempt to identify novel markers of warfarin resistance, we performed sequence analysis of all the coding exons in the VKORC1 gene in a selected group of warfarin-resistant and warfarin-sensitive patients, including an analysis of known polymorphisms in CYP2C9 and VKORC1. We validated our findings in a series of unselected warfarin-treated patients described in our previous studies.^12–13 

Patients were at stable anticoagulation (therapeutic international normalized ratio [INR] in 4 clinic visits). Resistant patients were defined by warfarin dose requirements of at least 80 mg/wk in the absence of known dose-increasing factors and sensitive patients by doses of 13 mg/wk or less. The controls were 99 previously described patients recruited as an unselected consecutive series. We recorded patient sex, age, weight, indications for warfarin therapy, additional medical problems, and concurrent medications. Blood samples for genetic analyses were extracted upon signing of an informed consent approved by the Sheba Medical Center Institutional Ethical Review Board.

VKORC1 exonic fragments were polymerase chain reaction (PCR) amplified using primers (available upon request) and standard protocols. Sequence analysis incorporating the same primers was performed using the automated ABI Prism 3100 Avant Genetic Analyzer (Applied Biosystems, Foster City, CA). CYP2C9*2 and *3 genotypes were determined by direct sequencing of the tag–single nucleotide polymorphisms (tag-SNPs), rs1799853 and rs1057910, respectively.

Haplotype analysis was performed using Sequenom mass spectrograph technology (Sequenom, San Diego, CA). VKORC1 haplotypes were determined according to the following tag-SNPs: 6853G>C (rs8050894) for VKORC1*2, 9041G>A (rs7294) for VKORC1*3, 6009C>T (rs17708472) for VKORC1*4, and 5417G>T for Asp36Tyr. Lack of these markers was considered as VKORC1*1 haplotype. Primers enabling multiplex PCR reactions were designed using Sequenom software. PCR fragments were amplified using a standard protocol. Sequenom analysis was performed according to the manufacturer-provided protocol.

Frequency of the Asp36Tyr polymorphism was determined in 4 ethnic groups from the Israeli Jewish population, including Ashkenazi (European origin), Yemenite (South Arabian), North African, and Ethiopian (East African) groups, each consisting of 100 DNA samples (200 chromosomes) retrieved anonymously from the Prenatal Genetic Screening Program depository at the Genetics Institute, Sheba Medical Center. Population screening for Asp36Tyr was performed by RFLP analysis using RsaI restriction endonuclease.

Warfarin doses (mean ± SD) were compared across groups by analysis of variance (ANOVA). Frequencies of CYP2C9, VKORC1, and Asp36Tyr variants in the warfarin-resistant and -sensitive groups and in the control group were compared by χ² and Fisher exact tests. In the controls, logistic regression was used to evaluate the effect of age, weight, and genetic variants on warfarin doses defined categorically by percentiles of the dose distribution as less than 20 mg/wk (25th percentile) and more than 70 mg/wk (75th percentile). Odds ratios (ORs) are presented with 95% confidence limits (CL). Multiple regression was used to evaluate the effects of CYP2C9*2 and *3 combined, and VKORC1*3 and *4 together on doses within the intermediate (interquartile) 20 to 70 mg/wk range. These analyses were performed using the GB-STAT software package (Dynamic Microsystems, Silver Spring, MD). Frequencies of VKORC1 and Asp36Tyr alleles were determined by the χ² test and frequencies of VKORC1 haplotypes by expectation-maximization (EM) algorithm using the PHASE software version 2.1 (University of Chicago, IL).

Resistant patients were younger compared with the sensitive group (F = 9.79; P = .002). Weight, target INR values, and indications for warfarin therapy were not different. In the controls, dose range was 8 to 105 mg/wk.

We identified a coding VKORC1 5417G>T (Asp36Tyr) polymorphism in 7 of 15 resistant patients but not in the 8 sensitive patients (P = .026). In the control group, Asp36Tyr was present in 8 of 99 patients (all in the upper dose quartile) requiring 80.9 ± 10.1 mg/wk compared with the noncarriers (n = 91) requiring 42.7 ± 7.5 mg/wk (F = 9.28; P < .001). Patients with Asp36Tyr and other dose-reducing markers (CYP2C9*2, *3, and VKORC1*2) still required high warfarin doses (Table 1, patients 4, 5, 12 in the warfarin-resistant group). The Asp36Tyr polymorphism was significantly associated with the highest warfarin dose category (more than 70 mg/wk: OR, 13.0; CL, 1.3 to 124.2) with no effect on the median and low-dose categories (Table 2).

The CYP2C9*2 and *3 alleles were overrepresented in the warfarin-sensitive (8 of 16) compared with the warfarin-resistant group (2 of 30) (P = .001) as well as the VKORC1*2 haplotypes in the sensitive (12 of 16) versus the resistant group (4 of 30) (P < .001). In the controls (Table 2), warfarin sensitivity was associated with CYP2C9*2 and *3 markers (OR, 2.4; CL, 1.3 to 4.6) and marginally with age.

In control patients of this category (n = 51), doses were significantly determined by the Asp36Tyr polymorphism (partial r² = 0.11), age (r² = 0.10), CYP2C9*1, *2, and *3 (r² = 0.08), and VKORC1*2 (r² = 0.05). Considering the entire control group (n = 99), Asp36Tyr was the major determinant of warfarin doses (r² = 0.18) together with VKORC1 and CYP2C9 (r² = 0.13 and 0.12, respectively), and age and body weight (r² = 0.12 and 0.07, respectively) accounting for a total of 62% of the variability in warfarin dose requirements.

Haplotype analysis of the control group using tag-SNPs of the known VKORC1 haplotypes showed that all carriers of Asp36Tyr had the tag-SNPs of the wild-type VKORC1*1, suggesting the possibility of a new configuration. The frequency of this putative Asp36Tyr/*1 haplotype in this group (after verification of Hardy-Weinberg equilibrium) was 4%. The other haplotype frequencies were 41% for VKORC1*2, 37% for VKORC1*3, 18% VKORC1*4, and 1% for VKORC1*1, consistent with the profiles characteristic of Caucasians.^4–5  Analysis of the ethnic groups showed that Asp36Tyr was overrepresented in individuals of Ethiopian origin (15%) (25 heterozygotes and 3 homozygotes) and was also common in the Ashkenazi (4%) but less common in the North African and Yemenite groups (both 0.5%). Direct sequence analysis of 2 Asp36Tyr homozygotes showed that both were also homozygotes for VKORC1*1, reconfirming the association between Asp36Tyr and VKORC1*1 haplotype.

The present study describes a new VKORC1 marker of warfarin resistance, which may improve our understanding of genetic factors affecting warfarin dose requirements. While the known CYP2C9 and VKORC1 markers enable classification of warfarin dose requirements as low (7 to 20 mg/wk) or intermediate (20 to 70 mg/wk), respectively, the Asp36Tyr marker is specifically indicative of high dose requirements and is dominant over the dose-reducing effect of CYP2C9*2, *3, and VKORC1*2. These findings suggest that future models for prediction of warfarin dose requirements should include Asp36Tyr in addition to the presently known CYP2C9 and VKORC1 markers.

Warfarin dose requirements vary across ethnic groups, with African Americans requiring higher and Japanese and Chinese lower doses than Caucasians.^14–16  Ethnic stratification of the known CYP2C9 and VKORC1 genetic variants provides only partial explanation of these differences.^10,17–22  The major limitation of this study is the small number of patients and its cross-sectional nature. Nonetheless, the new Asp36Tyr/VKORC1*1 haplotype associated with warfarin resistance may serve to explain epidemiologic observations of increased dose requirements in Jewish and non-Jewish population groups in which VKORC1*1 is characteristic (eg, African Americans).⁵  Future prospective studies focusing on larger cohorts of warfarin-treated patients will be needed to define the true predictive value of this new marker.

Contribution: R.L. was responsible for patient recruitment, data analysis, and manuscript preparation; I.D. and M.V. performed the laboratory work; A.L. contributed to patient recruitment; Y.C., N.A., and G.R. were responsible for the Sequenom analyses; G.K.-D. took part in the data analyses; S.A. assisted in the laboratory work and data analyses; H.H. was responsible for the study design, data analysis, and manuscript preparation; and E.G. was the director of this research.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Eva Gak, Institute of Human Genetics, Sheba Medical Center, Tel Hashomer 52621, Israel; e-mail: eva.gak@sheba.health.gov.il.

We thank Gregory Livshits from the Department of Epidemiology and Preventive Medicine at the Sackler School of Medicine, Tel Aviv University, for constructive suggestions.