Summary. Background: Vitamin K epoxide reductase (VKORC1) is the site of inhibition by coumarins. Several reports have shown that mutations in the gene encoding VKORC1 affect the sensitivity of the enzyme for warfarin. Recently, three main haplotypes of VKORC1; *2, *3 and *4 have been observed, that explain most of the genetic variability in warfarin dose among Caucasians. Objectives: We have investigated the main haplotypes of the VKORC1 gene in a Swedish population. Additional objective was to screen the studied population for mutations in the coding region of VKORC1 gene. Patients/methods: Warfarin doses and plasma S- and R-warfarin of 98 patients [with a target International Normalized Ratio (INR) of 2.0–3.0] have been correlated to VKORC1 haplotypes. Controls of 180 healthy individuals have also been haplotyped. Furthermore, a retrospective analysis of case records was performed to find any evidence indicating influence of VKORC1 haplotypes on warfarin response in the first 4 weeks (initiation phase) and the latest 12 months of warfarin treatment. Results and conclusions: Our result shows that VKORC1*2 is the most important haplotype for warfarin dosage. Patients with VKORC1*2 haplotype had more frequent visits than patients with VKORC1*3 or *4 haplotypes, higher coefficient of variation (CV) of prothrombin time-INR and higher percentage of INR values outside the therapeutic interval (i.e. 2.0–3.0) than patients with VKORC1*3 or *4 haplotypes. Also, there was a statistically significant difference in warfarin dose (P < 0.001) and R-warfarin plasma levels (P < 0.01) between VKORC1*2 and VKORC1*3 or 4 haplotypes. Patients with VKORC1*2 haplotype seem to require much lower warfarin doses than other patients.
It is well established that the site of inhibition by coumarins is vitamin K epoxide reductase (VKORC1); an enzyme responsible for the regeneration of vitamin K from its oxidized form . Vitamin K in its reduced form is essential for the gamma-carboxylation of several clotting factors. Warfarin is the worldwide most prescribed oral anticoagulant. Since their discovery in 1939 , the methods for monitoring and control of coumarin treatment have gradually been improved [3–12]. Supervision of the treatment is now routinely performed by repeated analysis of prothrombin complex expressed as International Normalized Ratio (INR). However, the major problem with the treatment of warfarin and other oral anticoagulants is still the significant inter-individual variation in dose requirement. A number of different factors including age, vitamin K intake, concomitant medication and genetic backgrounds have all been suggested to underlie these variations [13,14]. Of these, hereditary causes are believed to explain a substantial part of the recorded variations.
The most studied polymorphisms in warfarin's pharmacogenetics have been the microsomal CYP2C9 variant alleles, which are known for their effect on the pharmacokinetics of warfarin [15–18]. However, the pharmacodynamic site of warfarin; VKORC1, has only recently been explored [19–26], although resistance to warfarin among rats, not explained by increased liver clearance was described already in 1960 . An important explanation was evident with the identification of the gene encoding VKORC1 in the human genome [19,20]. The VKORC1 protein comprises 163 amino acids and contains at least three transmembrane alpha-helices as generated by different topology prediction programs .
The few reports so far presented indicate a strong correlation between VKORC1 mutations and warfarin response [19–26]. Initially, rare mutations leading to amino acid exchanges were reported to be associated with warfarin resistance [19,21]. However, most of the common single nucleotide polymorphisms (SNPs) related to the variability of warfarin dose requirement have been found in the non-coding regions of the VKORC1 gene [22–25]. Recently, Geisen et al.  presented a comprehensive haplotype map of the VKORC1 gene. In this study three main haplotypes were observed (VKORC1*2, VKORC1*3 and VKORC1*4) that accounted for nearly 100% of the VKORC1 genetic variability in a German population of European origin . In Asian populations, the VKORC1*2 is the dominating haplotype corresponding to 90% in Chinese Americans , 86% in Hong Kong Chinese  and 89% in Japanese populations . Haplotypes VKORC1*2 and VKORC1*3 dominate in Europeans (around 40% each) . In Africans, VKORC1*2 has a frequency of only 14% while VKORC1*1 (ancestral) and VKORC1*3 are the dominating haplotypes (31% and 43% respectively) . Haplotype VKORC1*4 is rare in Asians but is more common in European and African populations . We have investigated the VKORC1 main haplotypes and correlated them to warfarin maintenance dose and concentrations of warfarin enantiomers in plasma. Furthermore, we performed a retrospective analysis of case records and studied the influence of VKORC1 haplotypes on warfarin dosage in the initiation phase (first 4 weeks) and the most recent 12 months of warfarin treatment. The aim of the present work was to investigate the SNPs and haplotypes of the VKORC1 gene that are important for the sensitivity of warfarin in a Swedish population. Another objective was furthermore to screen the studied patient cohort for mutations in the coding region of the VKORC1 gene.
We utilized measurement of warfarin enantiomers to compensate possible confounding factors (such as patient compliance and increased clearance of the drug) that would arise if the maintenance dose alone was relied on. The bioavailability and absorption of warfarin can also be reduced in some cases [31,32]. Measurement of the actual concentration of warfarin enantiomers in plasma should therefore be a useful variable together with the maintenance dose.
Sconce et al.  recently described a model for dosing regimen. They studied the contribution of age, body size, CYP2C9 and VKORC1 genotypes, and found that these variables together accounted for nearly 55% of the variability in warfarin dose requirement. In that study, patients were genotyped for the c.-1639G > A polymorphism that is associated with low dose requirement [33,34]. The homozygous A allele of c.-1639 position corresponds to one of the haplotypes (VKORC1*2) that we studied in our work. In this study, previous knowledge of VKORC1 is further extended and analysis of case records was performed to demonstrate the impact of VKORC1 haplotypes on dose, warfarin enantiomers and warfarin response.
Materials and methods
Patients and healthy controls
Blood samples were drawn from 31 females and 67 male patients (n = 98, range 22–89 years old) on stable oral anticoagulation at anticoagulation clinics at the hospitals in Eksjö, Linköping, Motala, Värnamo and Västervik, all in southeastern Sweden. The target INR of patients was 2.0–3.0. For DNA and warfarin analysis, the samples were collected in 5-mL evacuated tubes containing ethylenediaminetetraacetic acid (EDTA). For coagulation analysis, 2.7 mL tubes containing 0.5 mL of 0.13 mol L−1 buffered trisodiumcitrate were used.
DNA samples (n = 180) from randomly collected healthy individuals from the southeastern region of Sweden formed the control group. Both patient and control groups consisted exclusively of Swedish Caucasians.
Ethical permission was obtained from the regional board of Ethical Review (Linköping, Sweden).
Polymerase chain reaction (PCR)
Genomic DNA was extracted from samples of whole blood on GenoVision 48 (BioRobot® M48; Qiagen, Hilden, Germany) by using the MagAttract DNA Blood Mini M48 Kit (Qiagen) according to the manufacturer's instructions. VKORC1 exons and intron/exon boundaries as well as specific regions in intron 1 were amplified by PCR. Primers were designed for VKORC1 by Primer3 PCR Primer program (http://frodo.wi.mit.edu). Primer sequences, amplified regions and other PCR-conditions are shown in Table 1.
Table 1. Primers used for amplification and sequencing of the vitamin K epoxide reductase (VKORC1) exons and SNPs in intron 1. Fragment size, polymerase chain reaction (PCR) conditions such as Mg2+ concentration and annealing temperature are shown. Oven temperature used for detection of PCR fragments by denaturing high-performance liquid cromatography (DHPLC) is also pointed out
The PCRs were generally carried out in 20 μL volume reactions in a Mastercycler® EP gradient S (Eppendorf AG, Hamburg, Germany). Each sample contained 20 pmol of each primer (Invitrogen, Carlsbad, CA, USA), 1X PCR buffer (Qiagen), 1.5–2 mm MgCl2 (Qiagen), nuclease-free water (Sigma-Aldrich, St Louis, MO, USA), 0.2 mm deoxyribonucleoside triphosphate (dNTP) (ABgene, Epsom, UK) and 0.5 U HotStar Taq® DNA Polymerase (Qiagen) (see Table 1). The amount of template DNA used for the PCR-reaction was approximately 35 ng. For exons 1–3, a total of 35 cycles of PCR were performed (after an initial hot start activation at 94 °C for 15 min) in a step down protocol (five cycles at each centigrade at 63–61 °C annealing and the remaining 20 cycles at 60 °C annealing temperature). Denaturation and extensions were performed at 94 °C for 30 s and 72 °C for 60 s, respectively and a final extension at 72 °C for 5 min completed the temperature cycles. For the intronic sequences, the temperature cycling conditions (after an initial hot start activation at 94 °C for 15 min) was: denaturation at 94 °C for 30 s, annealing at 59 °C for 30 s and extension at 72 °C for 50 s, for 35 cycles with a final extension step at 72 °C for 5 min. Successful PCR reactions were controlled under ultraviolet (UV)-light after electrophoresis on 2% agarose gels containing 0.5 μg mL−1 ethidium bromide.
Screening for mutations was performed with denaturing high-performance liquid chromatography on a Wave DNA Fragment Analysis System (Transgenomic Inc, Omaha, NB, USA). The PCR products (5 μL) were injected onto a DNA Sep HT Column (Transgenomic Inc.), eluted on a linear acetonitrile gradient [0.1 m triethylammonium acetate (TEAA)/0.1 m TEAA with 25% acetonitrile] and hetero- and homoduplexes were detected by UV-absorbance at 260 nm.
Potential mutations and polymorphisms were confirmed by DNA sequencing on a fluorescence-based capillary sequencer, MegaBACETM500 (Amersham Biosciences, Piscataway, NJ, USA). Prior to sequencing, the PCR products were treated with hydrolytic enzymes (ExoSAP-IT®; USB Corporation, Cleveland, OH, USA) according to manufacturer's protocol to remove unconsumed dNTPs and primers. Labeling of samples was made using a Dye-ET Terminator Cycle Sequencing Premix (Amersham Biosciences). The sequencing primers (2 μm of working solution) were those used in the PCR amplifications and the genomic sequence for VKORC1 was obtained from the National Center for Biotechnology Information (GenBank acc. no. AY587020http://www.ncbi.nlm.nih.gov/).
Haplotypes for the VKORC1 gene have been identified in the German (European) population using 14 biallelic markers . By using markers (M) discriminating the studied haplotypes, we categorized our patient cohort into VKORC1*2 (M17), VKORC1*3 (M23), and VKORC1*4 (M16) . The wild type haplotype VKORC1*1 is found in less than 0.1% of Caucasians and was not expected to be found in considerable numbers of the studied population.
Patient records at the anticoagulation clinics in Linköping and Motala were used to carry out a retrospective case study. Initiation phase was defined as the first 4 weeks following the start of warfarin treatment. The most recent 12 months was defined as the latest 12 months during which each patient regularly received warfarin treatment. No case had the initiation phase within this period. Values of prothrombin time (PT)-INR for each visit, both in the initiation phase and in the most recent 12 months, were collected and processed to calculate the number of visits by each patient, coefficients of variation (CV) of PT-INR and the number of INR values outside the therapeutic range (i.e. INR of 2–3). Age of each subject was also recorded and known confounding factors (provided they were noted in the patient records) such as drug interactions were taken into consideration.
Analysis of warfarin enantiomers in plasma was performed according to the high-performance liquid chromatographic method we described before . Patients who did not have S- and R-warfarin enantiomers in their plasma were excluded from the study.
Prothrombin complex analysis
All PT tests were performed after basic plasma predilution (1 + 6) and at a final plasma dilution of 1 + 20 utilizing the combined thromboplastin GHI-131 from Medirox AB (Studsvik, Sweden) on an ACL Futura, (Instrumentation Laboratories, Milan, Italy). Calibration was performed as we described before  utilizing plasmas with assigned INR from Equalis AB (Uppsala, Sweden).
Plasma was analyzed on an ACL Futura according to the manufacturer's instruction as a one stage clotting assay utilizing deficient plasma (Diagnostica Stago, Asnières, France) and thromboplastin (Medirox, Nyköping, Sweden).
Deviation of the studied SNPs from Hardy–Weinberg equilibrium was investigated with chi-squared test using a significance level of P < 0.05. Mann–Whitney U-test was used for tests containing two groups. Kruskal–Wallis test was employed for groups more than two. Inter-quartile illustration was performed with Box and Whisker plots.
In this study, 98 anticoagulated patients and 180 healthy controls had initially participated. Two patients were excluded from the study as we could not detect warfarin in their plasma. Samples of additional four patients could not be analyzed for different practical reasons, making the final number of patients 92.
DNA samples were sequenced for detection of possible mutations in the coding region and exon/intron boundaries of the VKORC1 gene. One novel mutation in exon 2, 6648C > T (c.202, His 68Tyr) and one synonymous SNP in exon 1  5347G > A (c.36, Arg12Arg.) were identified in the coding region. The heterozygous point mutation, His68Tyr (H68Y) is to our knowledge not identified earlier and represents a rare genetic variant, as it was not detected among 180 individuals in a control population. We screened DNA samples from parents of this subject for mutations in the VKORC1 and for the H68Y mutation in particular, after their informed consent. The heterozygous point mutation was identified as a maternal heredity. No close relative was treated with warfarin.
The following VKORC1 gene positions (VKORC1:c) were studied; c.173 + 1000 (haplotype VKORC1*2), c.492 + 134 (haplotype VKORC1*3), and c.173 + 525 (haplotype VKORC1*4). A fourth variant, c.36 (VKORC1*3D) was found in only four heterozygous cases with intermediate maintenance doses. Allele frequencies of the investigated SNPs both in the control samples and in the patients were estimated by gene counting (Table 2). The wild type haplotype (VKORC1*1) was found neither in the patients nor in the control group. There were no significant deviations of the genotyped SNPs from the Hardy–Weinberg equilibrium with a significance level of P < 0.05 using chi-squared test. Analysis of R- and S-warfarin, prothrombin levels and prothrombin complex (INR) were performed on patient samples and information of maintenance dose was collected (Table 3). The inter-quartile ranges of R- and S-warfarin for the three haplotypes are shown in Fig. 1. Patients with VKORC1*2 had significantly lower R-warfarin levels compared to haplotypes of either VKORC1*3 or VKORC1*4 (P < 0.01). Differences in S-warfarin were not significant and might be explained by the faster clearance of S-warfarin compared to its R counterpart. The difference between the three haplotypes was even more apparent when warfarin doses were compared (Fig. 2A). Patients with VKORC1*2 had much lower doses than those with VKORC1*3 or *4 haplotypes (P < 0.001). No statistically significant difference was shown between VKORC1*3 and *4 haplotypes in warfarin dosage. According to the sequencing results, more than half of the patients (n = 52) exhibited a mixture of two different heterozygous polymorphisms (Fig. 2B). Ten patients were heterozygous both in c.173 + 525 and in c.173 + 1000 gene positions, 22 individuals in c.173 + 525 and in c.492 + 134, while heterozygous SNP combinations in c.173 + 1000 and in c.492 + 134 gene positions were found in 20 patients. The control group had similar distributions. As shown in Fig. 2, heterozygous combinations of VKORC1*3 and VKORC1*4 seem to have at least equal impact on the maintenance dose as homozygotes of either VKORC1*3 or VKORC1*4. On the other hand, patients with VKORC1*2 had significantly lower warfarin doses than those with any of the three heterozygous combinations (P < 0.01). Contrarily, no significant difference in warfarin doses was shown between VKORC1*3 or VKORC1*4 haplotypes and the heterozygous mixtures.
Table 2. Allele frequency, nucleotide and gene positions of the studied haplotypes of VKORC1. A new rare point mutation found in a patient is also shown. Haplotype VKORC1*1 was not found in the studied population
c., cDNA sequence; c. + 1 is the A of the ATG initiation codon for translation. dbSNP, identity code from the public database of single nucleotide polymorphisms.
6484 C > T
c.173 + 1000
9041 G > A
c.492 + 134
6009 C > T
c.173 + 525
6648 C > T
New rare mutation
Table 3. Statistical summary of plasma warfarin concentrations, maintenance dose, prothrombin level and prothrombin complex (International Normalized Ratio, INR) for the three haplotypes of VKORC1 analyzed in the patients
R-warfarin (μg mL−1)
S-warfarin (μg mL−1)
Dose (mg week−1)
Prothrombin (IU mL−1)
Prothrombin time (PT) (INR)
R-warfarin (μg mL−1)
S-warfarin (μg mL−1)
Dose (mg week−1)
Prothrombin (IU mL−1)
R-warfarin (μg mL−1)
S-warfarin (μg mL−1)
Dose (mg week−1)
Prothrombin I (IU mL−1)
Records of 10 patients homozygous for VKORC1*2 and 11 patients homozygous for either VKORC1*3 or VKORC1*4 were studied at the anticoagulation clinic in Linköping. Table 4 summarizes the result from this study. Patients with VKORC1*2 had more variations in PT-INR than those with VKORC1*3 or *4 haplotypes both in the initiation phase (mean CV of 32.7 for haplotype *2 vs. 24.0 for haplotypes *3 or *4) and in the most recent 12 months (mean CV of 20.1 for haplotype *2 vs. 12.8 for haplotypes *3 or *4) despite similar mean INR. Although the number of visits in the initiation phase was similar in both groups, many more visits were required for VKORC1*2 patients than patients with haplotypes *3 or *4 in the most recent 12 months (average control visits/group: 21.2 for haplotype *2 patients vs. 11.8 for patients with *3 or *4).
Table 4. Retrospective analysis of case records for 10 patients homozygous to VKORC1*2 and 11 patients homozygous to VKORC1*3 or *4 haplotypes
VKORC1*3 or 4
INR, International Normalized Ratio; CV, coefficient of variation.
Number of patients
Mean age (years)
Mean INR in the initiation phase
Mean INR in the most recent 12 months
Percent INR > 3 in the initiation phase
Percent INR < 2 in the initiation phase
Percent INR > 3 in the most recent 12 months
Percent INR < 2 in the most recent 12 months
Mean CV in the initiation phase
Mean CV in the most recent 12 months
Mean number of visits in the initiation phase
Mean number of visits in the most recent 12 months
A point mutation at position c.202 in exon 2 leading to an amino acid exchange was found in one patient. The resulting amino acid exchange, H68Y, could not be associated with any warfarin resistance although H68 is in the vicinity of a previously reported mutation, V66M that is associated with warfarin resistance . Both H68 and V66 have been suggested to be located in the third loop of VKORC1 on the cytoplasmic side . At this point it is not possible to speculate about the importance of H68Y, but functional analysis may disclose the role of the mutation for VKORC1 enzyme function.
The three haplotypes studied in this work (VKORC1*2, *3, and *4) explain most of the genetic variability in Caucasians. The most important haplotype among these three is according to our results homozygous VKORC1*2. Individuals with VKORC1*2 had considerably lower maintenance doses of warfarin compared to those homozygous for VKORC1*3 and *4 (P < 0.001) and to heterozygous mixtures (P < 0.01). We think that the patients with VKORC1*2 haplotype could be at increased risk for bleeding complications if their initial maintenance doses are overestimated, causing over-anticoagulation. The retrospective analysis of case records, which we carried out, supports this statement (Table 4). Average age of the two groups was similar; 73.9 (VKORC1*2) and 73.0 years (VKORC1*3 or 4) respectively. Homozygotes of VKORC1*2 had larger PT-INR variations in both their initiation phase and in their most recent 12 months than homozygotes of VKORC1*3 or *4. Although the number of visits in the initiation phase was similar in both groups, homozygotes of VKORC1*2 had almost twice as many visits as homozygotes of VKORC1*3 or *4 in the most recent 12 months, indicating some difficulties for VKORC1*2 patients even in the long term. It seems likely that knowledge prior to initiation of warfarin treatment, about which VKORC1 haplotype the patients have would improve the treatment in the initiation phase. However, it might be questionable to suggest such testing. Provided that the risk of serious bleeding is about 2% per patient and year, the number of patients needed to be tested and hence the cost to avoid just one serious bleeding event would probably be very high. However, it is well known that if more time is spent outside the therapeutic range, the risk of adverse events increase. The cost of warfarin treatment is mainly dependent on the adverse events and the number of control visits. In the near future there will probably be alternative oral anticoagulant drugs available. If these in general are neither better nor worse than warfarin, in treatment effects and adverse events, such drugs are likely to have an advantage for patients homozygous for VKORC1*2. This should also be taken into consideration when designing clinical trials. Overall, among the studied haplotypes, VKORC1*2 can be considered to be the most important and most predictive haplotype of clinical relevance for warfarin treatment.
Annette Molbaek is gratefully acknowledged for her help with the mutation and SNP analysis. We are also grateful to Kerstin Gustafsson for valuable technical assistance in prothrombin and warfarin measurements. Yvonne Thornberg is also acknowledged for her help with patient records. The authors also want to thank Sören and Inger Hanssen, Christer Kihlström, Leif Engquist, Torbjörn Wallén and their staffs and patients at the anticoagulation clinics at the hospitals in Eksjö, Linköping, Motala, Värnamo and Västervik. This study and establishment of the population based control group for genetic analysis, was supported by the Medical Research Council of Southeast Sweden (FORSS).
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.