Abstract
Responses to warfarin (Coumadin) anticoagulation therapy are affected by genetic variability in both the CYP2C9 and VKORC1 genes. Validation of pharmacogenetic testing for warfarin responses includes demonstration of analytical validity of testing platforms and of the clinical validity of testing. We compared four platforms for determining the relevant single nucleotide polymorphisms (SNPs) in both CYP2C9 and VKORC1 that are associated with warfarin sensitivity (Third Wave Invader Plus, ParagonDx/Cepheid Smart Cycler, Idaho Technology LightCycler, and AutoGenomics Infiniti). Each method was examined for accuracy, cost, and turnaround time. All genotyping methods demonstrated greater than 95% accuracy for identifying the relevant SNPs (CYP2C9 *2 and *3; VKORC1 −1639 or 1173). The ParagonDx and Idaho Technology assays had the shortest turnaround and hands-on times. The Third Wave assay was readily scalable to higher test volumes but had the longest hands-on time. The AutoGenomics assay interrogated the largest number of SNPs but had the longest turnaround time. Four published warfarin-dosing algorithms (Washington University, UCSF, Louisville, and Newcastle) were compared for accuracy for predicting warfarin dose in a retrospective analysis of a local patient population on long-term, stable warfarin therapy. The predicted doses from both the Washington University and UCSF algorithms demonstrated the best correlation with actual warfarin doses.
In 2007, the US Food and Drug Administration modified the package insert for warfarin (Coumadin) to include information on the relationship of safe and effective dosage to specific single nucleotide polymorphisms (SNPs) in two genes, cytochrome P450 2C9 (CYP2C9) and vitamin K epoxide reductase complex subunit 1 (VKORC1). Warfarin binds to and inhibits vitamin K epoxide reductase (VKOR), encoded by VKORC1, thereby inhibiting activation of vitamin K-dependent clotting factors. A single common VKORC1 SNP −1639/3673 A>G defines VKORC1 haplotypes with increased sensitivity to warfarin (group A haplotypes, H1 and H2).1,2,3,4,5,6,7,8,9 CYP2C9 is responsible for metabolism of >90% of S-warfarin, the more active enantiomer of warfarin. Two common allelic variants of CYP2C9 with reduced enzymatic activity (CYP2C9*2 and CYP2C9*3) have been associated with reduced metabolism of warfarin, lower required doses of warfarin to achieve adequate anticoagulation, and increased risk of adverse events when beginning warfarin therapy.10,11,12
As a result of increased awareness of the impact of these SNPs on warfarin dosage, it is likely that clinical laboratories will increasingly be asked to provide genotype information for the above genes. Thus, there is a need to evaluate both the analytic and clinical validity of genotyping to make dosing recommendations for warfarin. Several assays for genotyping are now commercially available. We evaluated four platforms for determining relevant SNPs in CYP2C9 and VKORC1: the Third Wave Invader Plus CYP2C9 and VKORC1 reagents, the Cepheid SmartCycler Rapid Genotyping Assay for CYP2C9 and VKORC1 (developed by ParagonDx), the Idaho Technology Warfarin Genotyping Reagents on the Roche LightCycler and the AutoGenomics Infiniti 2C9-VKORC1 assay (now marketed as Warfarin XP).
Several different dosing algorithms have been devised that incorporate VKORC1 and CYP2C9 genotype, in addition to demographic and clinical information, to predict optimal warfarin dose. We compared the accuracy of four published algorithms for determining warfarin dosage, herein called Wash U,13 UCSF,14 Louisville,15 and Newcastle,16 using retrospective data from a local patient population on long-term, stable warfarin therapy.
Materials and Methods
DNA Samples
Residual DNA from patients on warfarin therapy was generously provided by Darryl Stafford's laboratory (University of North Carolina). These samples were previously genotyped for SNPs in the CYP2C9 and VKORC1 genes using a combination of DNA sequencing and SNP genotyping by real-time polymerase chain reaction (PCR).6 We selected 20 samples that had a variety of combinations of SNPs at the sites to be genotyped to assess the accuracy of each platform for testing various genotype combinations. Each sample was analyzed by each method separately, following the manufacturer's instructions.
Genotyping Analysis
Four genotyping platforms were compared for CYP2C9 and VKORC1 genotyping: Invader Plus CYP2C9 and VKORC1 analyte-specific reagents from Third Wave Technologies (a subsidiary of Hologic, Inc., Madison, WI); Warfarin SimpleProbe Genotyping Reagents from Idaho Technology, Inc. (Salt Lake City, UT); Infiniti Warfarin XP Assay from AutoGenomics, Inc. (Carlsbad, CA); and the SmartCycler Rapid Genotyping Assay for CYP2C9 and VKORC1, developed by ParagonDx (Morrisville, NC). Reagents and instrumentation for analysis were provided by the manufacturers.
The 20 residual DNA samples were analyzed on each platform. Genotype information was compared for each of the SNPs that are included on more than one platform: CYP2C9 *2 and *3, and VKORC1 −1639 (3673) and 3730 (9041). The SNPs detected by each assay are listed in Table 1. Because VKORC1 −1639 (3673) has been shown to be in complete linkage disequilibrium with VKORC1 1173(6484),1 the presence of either SNP was denoted as VKORC1 −1639 for comparison. Each assay was run at least three times with a batch of six samples plus three controls (blank, CYP2C9 *2/*3, and VKORC1 −1639A/G) to represent a projected typical clinical run. Hands-on time was recorded using a stopwatch at each step for each assay and totaled for comparison.
Table 1.
SNPs Detected by Each Assay
| Assay | CYP2C9 SNPs | VKORC1 SNPs |
|---|---|---|
| Third Wave Invader Plus | *2, *3 | −1639G/A |
| ParagonDx/SmartCycler | *2, *3 | 1173C/T, 3730G/A |
| Idaho Technology/LightCycler | *2, *3 | −1639G/A |
| AutoGenomics Infiniti | *2, *3, *4, *5, *6, *11 | −1639G/A, 497T/G, 698C/T, 1173C/T, 1542G/C, 2255C/T, 3462C/T, 3730G/A |
Note: −1639G>A and 1173C>T are in 100% linkage disequilibrium (see Table 2).
Third Wave Invader Plus Assay
The Third Wave Invader Plus assay was performed per manufacturer instructions with an 18-cycle multiplex PCR on a PE9700 thermal cycler (Applied Biosystems, Foster City, CA), with the resultant amplicons used in three separate Invader reactions, one each for CYP2C9 *2 and *3 and VKORC1 −1639. Fluorescence related to both wild-type (reference) sequence and SNP was measured in each well by a GENios FL plate reader (Tecan, San Jose, CA) and the results loaded into an Excel spreadsheet provided by the company.
ParagonDx/Cepheid assay
The ParagonDx/Cepheid CYP2C9/VKORC1 research use only (RUO) assay was performed per manufacturer instructions using two multiplex master mixes, each containing four primer sets and probes that were resuspended from lyophilized beads, one for CYP2C9 *2 and *3 and the other for VKORC1 1173 and 3730. Real-time PCR analysis was performed on the SmartCycler (Cepheid, Sunnyvale, CA) and the results loaded into an Excel-based interpretation file provided by ParagonDx (formerly Gentris). A Food and Drug Administration cleared version of this assay without the VKORC1 3730 SNP is to be marketed by Cepheid but is not yet commercially available at the time of this publication.
Idaho Technology LightCycler Assay
The Idaho Technology SimpleProbe Warfarin assay was performed per manufacturer instructions. Reactions for each SNP tested (CYP2C9 *2 and *3 and VKORC1 −1639 A/G) were analyzed simultaneously on a Roche LightCycler in separate capillaries and melting curves were evaluated for each sample. The assay can also be performed using a LightScanner (Idaho Technology) or other instrument for melting curve analysis. The primers are designed to match each SNP, so that the reference or wild-type sequence is an imperfect match and has a lower melting temperature.
AutoGenomics Warfarin XP Assay
The AutoGenomics Warfarin XP assay was performed per manufacturer instructions with a single multiplex PCR on a PE9700 thermal cycler, followed by allele-specific primer extension, hybridization and detection performed on the automated Infiniti instrument (AutoGenomics). The extended primers, incorporating Cy3-labeled dCTPs, are hybridized to chips using the AutoGenomics ZipCode-anti-ZipCode technology, and fluorescence at each set of the SNP-specific sites is read by the onboard confocal microscope. Onboard software analyzes the raw data, reports fluorescence values for both alleles at each of 14 SNPs (Table 1), and reports SNP determination. AutoGenomics also markets a Food and Drug Administration approved Infiniti Warfarin assay, which detects only CYP2C9 *2 and *3 and VKORC1 −1639 (3673), that was not analyzed in this comparison study.
DNA Sequencing
Selected samples from patients on high doses of warfarin were subjected to dideoxy sequencing of the promoter and coding regions of the VKORC1 gene and capillary electrophoresis on an Applied Biosystems 3130xl genetic analyzer.
Dosing Algorithm Evaluation
Patients treated at the University of North Carolina anticoagulation clinic who were on long-term, stable warfarin therapy were retrospectively analyzed for the accuracy of the dosing algorithms to predict stable warfarin dose. VKORC1 and CYP2C9 genotype and clinical information were available from a previous study.6 VKORC1 1173 (6484) genotype was assumed to be in complete linkage disequilibrium with VKORC1 −1639 (3673).1 Additional patient data necessary for each of the dosing algorithms were acquired through chart review, with appropriate informed consent per University of North Carolina Institutional Review Board approval. Sufficient information for all of the algorithms was available on 75 patients (81% Caucasian, 19% African-American; 63% male, 37% female). The required data for each of the 75 patients was entered into each algorithm and the recommended warfarin dosage calculated.
For the Wash U algorithm, data for initial warfarin dose calculation was input on the website http://www.warfarindosing.org (last accessed July 21, 2008). For the UCSF, Louisville, and Newcastle algorithms, an Excel spreadsheet was devised to calculate dose using published equations.14,15,16 The Newcastle algorithm incorporates age, height, the VKORC1 −1639 (3673) SNP, and CYP2C9 *2 and *3. The other algorithms include the same genetic variables plus additional variables. The Louisville algorithm incorporates age, gender, and weight instead of height. The Wash U algorithm incorporates age, body surface area, ethnicity, target international normalized ratio (INR), history of smoking, clinical indication for warfarin, and the presence of other medications that can affect warfarin metabolism (amiodarone). The UCSF algorithm includes the VKORC1 2255 (7566) SNP in addition to the genetic variants included in the other algorithms, along with age, height, weight, gender, ethnicity, history of smoking, and the presence of medications that can affect warfarin metabolism (amiodarone or sulfamethoxazole).
Actual dose for each patient was determined as the mean of the last five doses of warfarin on reaching target prothrombin time INR. The calculated recommended dose from each algorithm was plotted against each patient's actual dose in an XY scatter plot using Microsoft (Redmond, WA) Excel 2000 (version 9.0.3821 SR-1). A best-fit trendline, correlation coefficient (r), and coefficient of determination (r2) were determined by linear regression using Excel. The Excel linear regression calculations for the best-fit trendline are as follows: m = nΣ(xy) − ΣxΣy/nΣ(x2) − (Σx)2; b = Σy − mΣx/n; r = nΣ(xy) − ΣxΣy/[nΣ(x2) − (Σx)2] [nΣ(y2) − (Σy)2], where n is the number of data points, m is the slope, b is the y-intercept, and r is the correlation coefficient.
Results
Genotyping Method Comparison
The SNPs detected by each assay are listed in Table 1. All of the methods evaluated detect CYP2C9 *2 and *3 and at least one of the relevant VKORC1 SNPs, −1639 (3673) G>A or 1173 (6484) C>T. These two VKORC1 SNPs are in 100% linkage disequilibrium and are both associated with increased sensitivity to warfarin. The ParagonDx and AutoGenomics assays also detect VKORC1 3730(9041) G>A, which is present in group B VKORC1 haplotypes H7 and H8 and has been associated with higher dosage requirements of warfarin.1,6 The AutoGenomics Warfarin XP detects six CYP2C9 variants associated with reduced metabolism and eight VKORC1 SNPs, which enables determination of nine described VKORC1 haplotypes (Table 2).
Table 2.
VKORC1 Haplotypes
| Haplotype | −4930 (381) | −4450 (861) | −1639 (3673) | 497 (5808) | 698 (6009) | 1173 (6484) | 1542 (6853) | 2255 (7566) | 3462 (8733) | 3730 (9041) |
|---|---|---|---|---|---|---|---|---|---|---|
| H1 | C | C | A | T | C | T | C | T | C | G |
| H2 | C | C | A | G | C | T | C | T | C | G |
| H3 | C | C | G | T | C | C | C | C | C | G |
| H4 | C | C | G | T | C | C | G | T | C | G |
| H5 | T | C | A | G | C | T | C | T | C | G |
| H6 | T | C | G | T | C | C | G | C | C | G |
| H7 | T | C | G | T | C | C | G | C | C | A |
| H7* | T | C | G | T | C | C | G | C | T | A |
| H8 | T | A | G | T | C | C | G | C | C | A |
| H9 | T | A | G | T | T | C | G | C | C | G |
Shown is the VKORC1 sequence at each of the SNPs shown for each haplotype. The VKORC1 H7* haplotype is a subset of haplotype H7 with the 3462(8773) C>T SNP. SNPs in bold are typed by the AutoGenomics Warfarin XP assay. AutoGenomics cannot differentiate H2 versus H5 (can haplotype all others). Haplotypes H1 and H2 (group A) are associated with lower doses of warfarin; haplotypes H7 to H9 (group B) with higher doses.1,14
Data compiled for turnaround time (TAT) and hands-on time for each assay are shown in Table 3. The figures are based on actual time during setup and analysis of a run consisting of six samples and three controls. The ParagonDx and Idaho Technology assays had the lowest hands-on time and TAT because data are collected during or immediately after PCR onboard the instrument. The AutoGenomics required less than 1 hour of hands-on time because the allele-specific primer extension, hybridization, and detection are all automated, but had a longer TAT. The TAT increases with the number of samples because of sequential analysis by the Infiniti instrument. The Third Wave assay required the most labor because of the manipulations after PCR required for the Invader assay, but the rapidity of its detection method gave it a significantly shorter TAT than the AutoGenomics assay. Reagent list prices and instrument costs are compared in Table 3.
Table 3.
Turnaround Time, Labor, and Cost Comparison
| Supplier | TAT (hours) | Labor time (hours)* | Reagent cost per sample (list price)† | Instrumentation required | Equipment cost (approximate)‡ |
|---|---|---|---|---|---|
| Third Wave Invader | 3.2 | 1.42 | $55.00 | Thermal cycler, 96-well fluorometer | $12,000 |
| ParagonDx/Cepheid | 1.5 | 0.75 | Not available§ | Cepheid SmartCycler | $34,000 |
| Idaho Technology | 2 | 0.75 | $12.00 | Roche LightCycler | $40,000 |
| AutoGenomics | 10.6 | 0.83 | $75.00 | Thermal cycler, Infiniti analyzer | $125,000 |
Figures are based on a run of six patient samples and three controls; DNA extraction time not included.
Cost does not include DNA extraction costs, polymerase enzyme, or disposables.
Cost of thermal cycler when required not included.
The ParagonDx/Cepheid assay is not commercially available at this time.
A summary of genotype results for the 20 previously characterized samples is shown in Table 4. The table includes data for each SNP that is interrogated by more than one of the assays evaluated (CYP2C9 *2 and *3; VKORC1 −1639A>G or 1173C>T and VKORC1 3730G>A). For comparison, results for the VKORC1 SNPs −1639A>G and 1173C>T, which are in complete linkage disequilibrium, are reported as VKORC1 −1639. Some samples were not tested at all SNPs with every assay because of limited availability of reagents. Each assay demonstrated 100% accuracy, with the exception of the AutoGenomics assay, which gave one indeterminate result for CYP2C9 (95% accuracy). The accuracy of the AutoGenomics assay for genotyping additional SNPs in VKORC1 was also examined for the 20 samples by correlation with previous results for SNPs 497 (5808) T/G, 698 (6009) C/T, 1542 (6853) C/G, and 2255 (7566) T/C,6 and demonstrated 100% accuracy (data not shown).
Table 4.
CYP2C9 and VKORC1 Genotyping Results
| Sample |
CYP2C9 |
VKORC1 |
||||||
|---|---|---|---|---|---|---|---|---|
| Third Wave | Idaho Tech | ParagonDx | AutoGenomics | Third Wave* | Idaho Tech* | ParagonDx† | AutoGenomics | |
| 01 | *3/*3 | *3/*3 | *3/*3 | *3/*3 | −1639/−1639 | −1639/−1639 | −1639/−1639 | −1639/−1639 |
| 02 | wt/wt | wt/wt | ND | wt/wt | wt/wt | wt/wt | 3730/3730 | 3730/3730 |
| 03 | *2/wt | *2/wt | *2/wt | *2/wt | wt/wt | wt/wt | ND | 3730/3730 |
| 04 | wt/wt | wt/wt | wt/wt | wt/wt | wt/wt | wt/wt | ND | 3730/wt |
| 05 | wt/wt | wt/wt | ND | wt/wt | −1639/wt | −1639/wt | −1639/wt | −1639/wt |
| 06 | *2/wt | *2/wt | *2/wt | IND | −1639/−1639 | −1639/−1639 | ND | −1639/−1639 |
| 07 | *3/wt | *3/wt | *3/wt | *3/wt | −1639/wt | −1639/wt | −1639/wt | −1639/wt |
| 08 | wt/wt | wt/wt | ND | wt/wt | −1639/wt | −1639/wt | −1639/3730 | −1639/3730 |
| 09 | *3/wt | *3/wt | *3/wt | *3/wt | −1639/wt | −1639/wt | −1639/3730 | −1639/3730 |
| 10 | *2/wt | *2/wt | *2/wt | *2/wt | −1639/−1639 | −1639/−1639 | ND | −1639/−1639 |
| 11 | wt/wt | wt/wt | ND | wt/wt | wt/wt | wt/wt | 3730/3730 | 3730/3730 |
| 12 | *2/*3 | *2/*3 | *2/*3 | *2/*3 | −1639/wt | −1639/wt | −1639/3730 | −1639/3730 |
| 13 | wt/wt | wt/wt | ND | wt/wt | −1639/−1639 | −1639/−1639 | −1639/−1639 | −1639/−1639 |
| 14 | *2/wt | *2/wt | ND | *2/wt | wt/wt | wt/wt | 3730/3730 | 3730/3730 |
| 15 | wt/wt | wt/wt | ND | wt/wt | wt/wt | wt/wt | 3730/3730 | 3730/3730 |
| 16 | wt/wt | wt/wt | ND | wt/wt | −1639/−1639 | −1639/−1639 | −1639/−1639 | −1639/−1639 |
| 17 | *3/wt | *3/wt | *3/wt | *3/wt | −1639/wt | −1639/wt | ND | −1639/3730 |
| 18 | wt/wt | wt/wt | ND | wt/wt | wt/wt | wt/wt | 3730/wt | 3730/wt |
| 19 | *2/wt | *2/wt | *2/wt | *2/wt | −1639/wt | −1639/wt | ND | −1639/3730 |
| 20 | wt/wt | wt/wt | wt/wt | wt/wt | −1639/−1639 | −1639/−1639 | −1639/−1639 | −1639/−1639 |
| Accuracy | 100% | 100% | 100% | 95% | 100% | 100% | 100% | 100% |
Method does not genotype VKORC1 3730.
ParagonDx method genotypes VKORC1 1173, reported here as −1639 assuming 100% LD for clarity. ND, sample not tested at this locus because of limited reagents available at the time of testing. IND, indeterminate results, sample inadequate to repeat. wt, wild type. All results were confirmed by sequencing or SNP analysis.6
The advantages and disadvantages of each assay are summarized in Table 5. The technically simplest assays were the Idaho Technology and ParagonDx assays because they use established, one-tube methods. The Third Wave Invader Plus assay required more manipulation in transferring the PCR products into the Invader reaction and then into the fluorometer. The AutoGenomics is a relatively new instrument and uses novel technology, which required specialized training. Capital instrumentation requirements differ for each of the assays. The Third Wave assay requires only a fluorometer, the ParagonDx requires a Cepheid SmartCycler, and the Idaho Technology assay requires a Roche LightCycler 2.0 or another melting curve analysis instrument. The largest capital investment requirement is the Infiniti instrument from AutoGenomics.
Table 5.
Advantages and Disadvantages of Each Method
| Method | Advantages | Disadvantages |
|---|---|---|
| Third Wave Invader | Low capital expense | Post-PCR manipulations |
| Established methodology | Hands-on time | |
| Readily scalable to larger run sizes | ||
| ParagonDx/Cepheid SmartCycler | Short TAT | Not commercially available |
| Ease of use | Capital investment | |
| Established methodology | Small run size | |
| Low hands-on time | ||
| Idaho Technology LightCycler | Short TAT | Capital investment |
| Ease of use | ||
| Established methodology | ||
| Low hands-on time | ||
| AutoGenomics Infiniti | Number of SNPs offered | Long TAT |
| Warfarin XP | Post-PCR automated | Large capital investment |
| Easily adaptable to include additional SNPs | Less established methodology |
Dosing Algorithm Comparison
A comparison of the four algorithms for determining warfarin dosage is shown in Figure 1. The Wash U algorithm requires more clinical information than the other three, which limited the number of patients we were able to analyze in this retrospective study. Only those patients for whom all four algorithms could be run were included in the analysis (n = 75). Overall, the subpopulation of patients for which we were able to run the algorithms was similar demographically and genotypically to the total study population (n = 117)6 and therefore it is unlikely that the data are biased toward any particular group (see Supplemental Table S1 at http://jmd.amjpathol.org). The study population is similar demographically to the total patient population followed at the University of North Carolina. The recommended dosage from the Wash U and UCSF algorithms had the best correlation with stable dosage in the patient population, with correlation coefficients (r) of 0.60 compared with 0.58 for the Louisville algorithm and 0.48 for the Newcastle algorithm. The difference between the Wash U, UCSF, and Louisville algorithms is not significant, whereas the Newcastle algorithm was significantly less predictive of stable dose. The Louisville algorithm calculated a high warfarin dose of greater than 12 mg/day for two patients whose actual warfarin dose was less than 9 mg/day.
Figure 1.
Comparison of four algorithms for determining warfarin dosage. Shown is the actual mean warfarin dose (mg/day) plotted against the warfarin dose predicted by each of the dosing algorithms for patients on long-term, stable warfarin therapy (n = 75). Blue diamonds represent each patient analyzed. Red squares represent outlier patients with high warfarin doses not predicted by any of the dosing algorithms. For each graph, solid dark lines represent the best-fit trendline plotted in Excel using linear regression and dashed light lines represent the line of identity. The linear correlation coefficient (r) for each algorithm is shown.
Two of the patients had much higher actual warfarin doses than predicted by any of the algorithms (Figure 1, square symbols). Both of these outlier patients were African-American females with an actual stable warfarin dose of >12 mg/day. Removal of these two data points improved the correlation coefficient for each algorithm, particularly the Wash U and UCSF algorithms (r = 0.67) (Table 6). All four algorithms performed less well among the African-American population, even after removal of the outlier patients (Table 7). However, the sample size was too small (n = 14) to make conclusions about the efficacy of the dosing algorithms in this population.
Table 6.
Evaluation of Warfarin Dosage Algorithms
| All patients* | Less two outliers† | |||
|---|---|---|---|---|
| Algorithm | r, r2 | Trendline equation | r, r2 | Trendline equation |
| Wash U | 0.60, 0.36 | y = 0.9131x + 0.9966 | 0.67, 0.45 | y = 0.8346x + 1.1584 |
| UCSF | 0.60, 0.36 | y = 1.0665x + 0.6290 | 0.67, 0.45 | y = 0.9725x + 0.8346 |
| Louisville | 0.58, 0.33 | y = 0.6985x + 1.7400 | 0.62, 0.38 | y = 0.6088x + 2.0118 |
| Newcastle | 0.48, 0.23 | y = 0.8926x + 1.8266 | 0.54, 0.29 | y = 0.8147x + 1.9115 |
Comparison of correlation coefficient (r), coefficient of determination (r2), and trendline equations for each algorithm.
n = 75; 81% Caucasian, 19% African-American; 63% male, 37% female.
n = 73; 84% Caucasian, 16% African-American; 64% male, 36% female.
Table 7.
Evaluation of Warfarin Dosage Algorithms in African-American Patients
| All African-American patients* |
Less two “outliers† |
|||
|---|---|---|---|---|
| Algorithm | r, r2 | Trendline equation | r, r2 | Trendline equation |
| Wash U | 0.36, 0.13 | y = 1.1131x + 0.8642 | 0.52, 0.27 | y = 0.8447x + 1.1604 |
| UCSF | 0.33, 0.11 | y = 1.2723x + 0.1863 | 0.62, 0.38 | y = 1.2580x − 1.0661 |
| Louisville | 0.30, 0.09 | y = 0.4872x + 3.7178 | 0.56, 0.31 | y = 0.4750x + 2.4520 |
| Newcastle | 0.34, 0.12 | y = 0.9184x + 3.0793 | 0.54, 0.29 | y = 0.7451x + 2.5902 |
Comparison of correlation coefficient (r), coefficient of determination (r2), and trendline equations for each algorithm.
n = 14; 64% male, 36% female.
n = 12; 75% male, 25% female.
For the two outlier patients, sequencing of the coding region of VKORC1 was performed to examine for mutations, and previously performed SNP analysis6 was examined to determine VKORC1 haplotypes. The first patient had an actual mean warfarin dose of 17.4 mg/day. This individual was CYP2C9 *1/*1 (wild type) and had VKORC1 SNPs 2255 (7566) C/T and 3730 (9041) G/A, indicating that the patient is heterozygous for VKORC1 haplotype H4 and a group B haplotype (H7 or H8). The VKORC1 2255C>T SNP is seen in group A haplotypes, which are associated with increased warfarin sensitivity, as well as in haplotype H4, a less common haplotype for which there is little information on warfarin dose correlation. The 3730 G>A SNP is associated with group B haplotypes, which have been correlated with higher warfarin doses,1,5 but it is not included in any of the dosing algorithms. In addition, DNA sequencing demonstrated a novel single nucleotide change in the promoter, VKORC1 −477 (4835) G/A, which to our knowledge has not been published previously. The impact of this novel change is unknown.
The second outlier patient had an actual mean warfarin dose of 13.8 mg/day. This individual was also CYP2C9 *1/*1 and was heterozygous for VKORC1 −1639 (3673) G/A, 1173 (6484) C/T, and 2255 (7566) C/T and homozygous for 1542 (6853) C/C (VKORC1 haplotypes H1/H3). Haplotype H1 is a group A haplotype associated with increased sensitivity to warfarin. Haplotype H3 is a less common haplotype with little information on warfarin dose correlation. No VKORC1 coding region mutations were found by DNA sequencing and chart review revealed nothing remarkable that could account for the high warfarin dose requirement in either patient.
Discussion
Warfarin is one of the most commonly prescribed medications in the U.S. Because of its narrow therapeutic range and the wide variation in warfarin dose requirement in different individuals, there is a significant risk of bleeding or thromboembolic events during warfarin therapy. The recent understanding that genetic variability in CYP2C9 and VKORC1 can account for some of the variability in individual warfarin dose requirements has prompted the development of commercially available genotyping assays and of dosing algorithms that incorporate genotype information. We compared four different analysis platforms and four dosing algorithms for determining warfarin dosage in a local patient population.
Genotyping Test Platforms
All four platforms performed well and gave accurate genotyping results. The selection of genotyping platform must take into consideration needs, demands, and available resources of each clinical laboratory. Although the Idaho Technology and ParagonDx methods have the shortest TAT and hands-on time of the four, the additional SNPs available on the AutoGenomics Infiniti may be valuable in collecting data in non-Caucasian patients and those with less common genotypes. The Third Wave assay offers comparable TAT and the advantages of low capital equipment investment and scalability to medium test volumes. Although most of the platforms demonstrated 100% accuracy, the AutoGenomics assay gave one indeterminate result for CYP2C9. In our experience, indeterminate results on the AutoGenomics assay are usually resolved with repeat testing. A recent publication compared two of these platforms plus two different 2C9-VKOR genotyping platforms and also found very good genotyping accuracy by each of the commercial platforms.17
TAT may be a crucial factor for warfarin genotyping and for pharmacogenomic (PGx) testing in general. Many of the adverse events associated with warfarin occur during the initiation period. Ideally, genetic variability that affects drug response would be known before drug administration. This may not always be feasible for drugs such as warfarin, which may need to be administered quickly, although in some clinical situations a faster-acting anticoagulant such as heparin may be used before warfarin. Some investigators have advocated using a loading dose of warfarin and incorporating genotype into subsequent dose calculations.18 Nonetheless, although the need for rapid TAT of warfarin genotyping has not been demonstrated, the demand for clinical laboratories to provide useful and timely information that impacts warfarin initiation or dose modification may require assays with low failure rates and rapid TAT. As clinical information grows to accommodate dosing recommendations using the full spectrum of known SNPs, the added genetic information may justify the longer TAT and cost of the AutoGenomics platform or may dictate development of more efficient assays with more complete genomic information.
Dosing Algorithms
Development of testing in a clinical laboratory includes not only analytic validation of testing platform, but also test interpretation and guidance for clinicians. Several groups have devised dosing algorithms through regression analysis of the relative effect of CYP2C9 and VKORC1 variants and various demographic and clinical factors on warfarin dose. These studies have indicated that VKORC1 and CYP2C9 genotype may account for up to 40% of warfarin dose variance, and genotype in addition to other known factors may account for 50 to 80% of warfarin dose variance seen in patients with effective anticoagulation.14,15,16,19,20,21,22,23 Thus, genetic variability appears at least as important as other factors that are currently considered by clinicians when dosing warfarin.
To incorporate warfarin response genotyping at our institution, we analyzed whether four available dosing algorithms could accurately predict warfarin dose in a local patient population. There was a good linear relationship between actual warfarin dose and predicted dose for each of the algorithms. The UCSF and Wash U algorithms demonstrated the best correlation with actual warfarin dose, with correlation coefficients (r) of 0.60 in the total patient population and 0.67 after the removal of two outlier patients. Both of these algorithms include ethnicity, smoking status, and the presence of a CYP2C9 inhibitor as factors whereas the other algorithms do not, which may be why they performed somewhat better. The Newcastle algorithm performed the least well and tended to underdose patients, perhaps because it includes height but not body mass index or weight. Wu and colleagues14 compared several dosing algorithms in a multiethnic American population and found a similar correlation with actual warfarin dose as in our study (r = 0.64 to 0.68) with similar worse correlation for the Newcastle algorithm (r = 0.55). Although the Wash U algorithm requires that more clinical information be gathered on each patient, it did not perform better than the UCSF algorithm in our population. The web-based format of the Wash U algorithm is convenient and easily accessible, which may make it more easily adaptable by a wide group of clinicians.
After removal of the two outlier data points, 45% of warfarin dose variance (r2) was explained by the genetic and clinical factors included in the Wash U and UCSF algorithms and less than 40% by the other two algorithms. This is more than the variance in dose that has been attributed to clinical factors alone, but less than that reported by other groups. It is unclear why these algorithms explain less of the warfarin dose variance in our study population than in the populations studied by the groups that derived the algorithms. There may be factors unique to different geographic or demographic groups that affect the efficacy of dosing algorithms derived from a specific patient population. Wu and colleagues14 predictably found the best correlation with warfarin dose using the training set of patients that was used to derive their own algorithm. Because ours was a small retrospective analysis, there may have been missing or confounding factors that were not taken into account, or the results may be skewed by individual patients. This may simply indicate the better performance of such an approach in a large group of patients than in an individual patient.
African-American Patients
None of the algorithms predicted warfarin dose very well in the African-American subgroup, although the data were skewed by the two outlier patients on high doses of warfarin. However, even after removal of the two outlier African-American patients, the predicted warfarin doses correlated less well with actual warfarin dose in this subgroup than for the total population. The UCSF algorithm, which was derived in a multiethnic population, performed somewhat better in this subgroup, although the number of patients was too small to make statistically rigorous conclusions. Only the Wash U and UCSF algorithms incorporate ethnicity as a factor. Although part of difference in warfarin response in different ethnic groups may be explained by genetic factors, the spectrum of causative genetic differences is not well understood and there are almost certainly important environmental factors that differ between ethnic groups. Accordingly, a recent study indicated that African-American patients require higher warfarin maintenance doses even after adjusting for known clinical and CYP2C9-VKORC1 genetic covariants.24
Because many of the studies of the effect of genotype on warfarin dose have been performed primarily in European Caucasians, genetic variants more common in other ethnic groups may be ignored. Thus, CYP2C9 variants other than 2C9 *2 and *3 that affect warfarin metabolism are not incorporated (eg, CYP2C9 *4, *5, *6, and *11), nor are VKORC1 haplotypes other than those common in Caucasians, such as H3 to H6 and other rarer haplotypes that have been described.14 Many of these variants are more common in African-Americans. The UCSF multiethnic study, which included Caucasian, Asian, Hispanic, and African-American populations, found a significant effect of the VKORC1 SNP 2255(7566) C, which is seen in haplotypes H3 and H6 to H9 and was associated with higher doses of warfarin.14 However haplotypes H3 to H6 were grouped together and were not analyzed individually because of their low frequency. Further studies are needed in African-American and other non-European Caucasian populations to understand the relative effect of genetic and other parameters on warfarin dose requirements.
Warfarin Resistance
The explanation for the high warfarin dose requirement in the two African-American outlier patients is not known. VKORC1 SNP analysis indicated that each was heterozygous for VKORC1 H3 or H4, haplotypes reported to be more common in African-Americans and which may require higher doses of warfarin.14 However, this is unlikely to explain the extremely high doses of warfarin required by these patients. At least six different mutations in the VKORC1 coding region have been described that are associated with warfarin resistance, possibly because of altered binding of warfarin.25,26,27,28,29,30,31,32 Examination of the VKORC1 coding region for mutations in these patients was negative. The effect of the novel promoter region variant found in one of the patients is unknown, although it is possible that this change could result in higher expression of VKOR and higher dose requirements.
Currently, the only genetic variants that are included in the Food and Drug Administration warfarin package insert and incorporated by most of the dosing algorithms are those associated with increased sensitivity to warfarin (CYP2C9 *2 and *3, VKORC1 −1639/3673 G>A). Therefore patients with VKORC1 variants associated with high dose requirements of warfarin may not be captured by current practice. The −1639(3673) SNP, which is seen in VKORC1 group A haplotypes (H1 and H2) as well as H5, has consistently been associated with significantly lower dose requirements of warfarin, possibly because of decreased expression of the VKOR protein.1 It is possible that VKORC1 −1639 (or a linked SNP) is the only relevant functional SNP and explains all of the variability in warfarin response attributable to VKORC1 haplotype. However, other possible explanations are that the effect of other variants did not reach statistical significance in some studies because they were rare in the populations studied or because they were not analyzed. Wu and colleagues14 found that elimination of cases with rare genotypes improved the correlation between actual warfarin dose and dose predicted by various algorithms, indicating that rare variants may have an effect on warfarin dose that is currently not incorporated in such dosing algorithms.
The significance of nongroup A VKORC1 haplotypes requires further investigation. Group B haplotypes (H7 to H9) have been correlated with higher dose requirements of warfarin.1,14 Haplotypes H3, H4, and H6, which are more common in African-Americans, have also been correlated with higher doses of warfarin,14 but require further study in a larger group of patients with these haplotypes. In addition, patients with certain VKORC1 coding region mutations may require very high warfarin doses exceeding 10 mg/day.25,26,27,28,29,30,31,32 One of these, the Asp36Tyr mutation, has been reported to be a common variant in Ashkenazi and Ethiopian Jews.29,33 We sequenced the VKORC1 coding region for all patients on high doses of warfarin (>10 mg/day) from the 117 patients in the University of North Carolina study and from a similar study of Brazilian patients. Although none of the nine University of North Carolina patients on >10 mg/day of warfarin had coding region mutations, 2 of 11 Brazilian patients on >10 mg/day of warfarin had a Val66Met mutation (unpublished data), which has previously been reported to be associated with warfarin resistance.26,32 Thus, variants in the VKOR coding region associated with extreme warfarin resistance may be common in certain populations, but are not currently included in any of the available genotyping test platforms or dosing algorithms.
Future Directions
The question remains—how will genetic information affecting warfarin response be used clinically? Although there is a well-documented relationship between specific CYP2C9 and VKORC1 variants and increased sensitivity to warfarin, the clinical utility of testing for these variants in determining optimal warfarin dose has not been well established. It remains unclear whether dosing algorithms that predict final warfarin dose in patients on stable long-term therapy will be effective in patients initiating warfarin therapy. Several clinical trials have been performed to investigate whether incorporation of genotype in warfarin dosing results in better patient management. Genotype-guided dosage has been associated with fewer dosage changes, faster time to target INR, more time in target INR, and a decrease in out of range INRs in some studies.12,13,23,34,35,36,37,38 However, overall the results of these studies have been inconclusive and warfarin genotyping has not yet been incorporated into daily practice. It is unclear whether complicated dosing algorithms will be widely adopted by the bulk of prescribing physicians. Perhaps a more practical approach would be for clinicians to take genotype information into consideration along with other factors when dosing warfarin empirically, for example using an incremental reduction in empirical starting dose based on the number of sensitivity variants present. Those patients that will require the greatest dose modification are those that are homozygous for or have multiple relevant genetic variants, which represent a small proportion of the population.
There is still much to be done in the field of pharmacogenomics of warfarin dosage. Further studies to elucidate the effect of genotypes more common in non-Caucasian ethnic groups and to identify other genes that may have an effect on warfarin response are needed. In addition, there is a need for additional clinical trials to demonstrate whether there exists clear clinical utility of genotype-guided dosing and if so, the formulation of easily implementable recommendations for dosing modification.
Acknowledgements
We thank Third Wave Technologies, ParagonDx, Idaho Technology, and AutoGenomics for providing reagents and/or instrumentation for this study; ParagonDx for performing DNA sequencing of selected samples; Darryl Stafford for providing DNA samples; and Lisa Susswein and Cristina Santos for help with clinical chart review.
Footnotes
Supported by the University of North Carolina Department of Pathology and Laboratory Medicine.
Financial Disclosure: K.W. is a consultant for ParagonDx, and H.M. is a consultant for Third Wave Technologies.
Supplemental material can be found online at http://jmd.amjpathol.org.
Current address of M.R.L.: Expression Analysis, Durham, NC.
Supplementary data
References
- 1.Rieder MJ, Reiner AP, Gage BF, Nickerson DA, Eby CS, McLeod HL, Blough DK, Thummel KE, Veenstra DL, Rettie AE. Effect of VKORC1 haplotypes on transcriptional regulation and warfarin dose. N Engl J Med. 2005;352:2285–2293. doi: 10.1056/NEJMoa044503. [DOI] [PubMed] [Google Scholar]
- 2.Wadelius M, Chen LY, Downes K, Ghori J, Hunt S, Eriksson N, Wallerman O, Melhus H, Wadelius C, Bentley D, Deloukas P. Common VKORC1 and GGCX polymorphisms associated with warfarin dose. Pharmacogenomics J. 2005;5:262–270. doi: 10.1038/sj.tpj.6500313. [DOI] [PubMed] [Google Scholar]
- 3.D'Andrea G, D'Ambrosio RL, Di Perna P, Chetta M, Santacroce R, Brancaccio V, Grandone E, Margaglione M. A polymorphism in the VKORC1 gene is associated with an interindividual variability in the dose-anticoagulant effect of warfarin. Blood. 2005;105:645–649. doi: 10.1182/blood-2004-06-2111. [DOI] [PubMed] [Google Scholar]
- 4.Yuan HY, Chen JJ, Lee MT, Wung JC, Chen YF, Charng MJ, Lu MJ, Hung CR, Wei CY, Chen CH, Wu JY, Chen YT. A novel functional VKORC1 promoter polymorphism is associated with inter-individual and inter-ethnic differences in warfarin sensitivity. Hum Mol Genet. 2005;14:1745–1751. doi: 10.1093/hmg/ddi180. [DOI] [PubMed] [Google Scholar]
- 5.Geisen C, Watzka M, Sittinger K, Steffens M, Daugela L, Seifried E, Muller CR, Wienker TF, Oldenburg J. VKORC1 haplotypes and their impact on the inter-individual and inter-ethnical variability of oral anticoagulation. Thromb Haemost. 2005;94:773–779. doi: 10.1160/TH05-04-0290. [DOI] [PubMed] [Google Scholar]
- 6.Li T, Lange LA, Li X, Susswein L, Bryant B, Malone R, Lange EM, Huang TY, Stafford DW, Evans JP. Polymorphisms in the VKORC1 gene are strongly associated with warfarin dosage requirements in patients receiving anticoagulation. J Med Genet. 2006;43:740–744. doi: 10.1136/jmg.2005.040410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Aquilante CL, Langaee TY, Lopez LM, Yarandi HN, Tromberg JS, Mohuczy D, Gaston KL, Waddell CD, Chirico MJ, Johnson JA. Influence of coagulation factor, vitamin K epoxide reductase complex subunit 1, and cytochrome P450 2C9 gene polymorphisms on warfarin dose requirements. Clin Pharmacol Ther. 2006;79:291–302. doi: 10.1016/j.clpt.2005.11.011. [DOI] [PubMed] [Google Scholar]
- 8.Obayashi K, Nakamura K, Kawana J, Ogata H, Hanada K, Kurabayashi M, Hasegawa A, Yamamoto K, Horiuchi R. VKORC1 gene variations are the major contributors of variation in warfarin dose in Japanese patients. Clin Pharmacol Ther. 2006;80:169–178. doi: 10.1016/j.clpt.2006.04.010. [DOI] [PubMed] [Google Scholar]
- 9.Vecsler M, Loebstein R, Almog S, Kurnik D, Goldman B, Halkin H, Gak E. Combined genetic profiles of components and regulators of the vitamin K-dependent gamma-carboxylation system affect individual sensitivity to warfarin. Thromb Haemost. 2006;95:205–211. doi: 10.1160/TH05-06-0446. [DOI] [PubMed] [Google Scholar]
- 10.Margaglione M, Colaizzo D, D'Andrea G, Brancaccio V, Ciampa A, Grandone E, Di Minno G. Genetic modulation of oral anticoagulation with warfarin. Thromb Haemost. 2000;84:775–778. [PubMed] [Google Scholar]
- 11.Higashi MK, Veenstra DL, Kondo LM, Wittkowsky AK, Srinouanprachanh SL, Farin FM, Rettie AE. Association between CYP2C9 genetic variants and anticoagulation-related outcomes during warfarin therapy. JAMA. 2002;287:1690–1698. doi: 10.1001/jama.287.13.1690. [DOI] [PubMed] [Google Scholar]
- 12.Voora D, Eby C, Linder MW, Milligan PE, Bukaveckas BL, McLeod HL, Maloney W, Clohisy J, Burnett RS, Grosso L, Gatchel SK, Gage BF. Prospective dosing of warfarin based on cytochrome P-450 2C9 genotype. Thromb Haemost. 2005;93:700–705. doi: 10.1160/TH04-08-0542. [DOI] [PubMed] [Google Scholar]
- 13.Gage BF, Eby C, Johnson JA, Deych E, Rieder MJ, Ridker PM, Milligan PE, Grice G, Lenzini P, Rettie AE, Aquilante CL, Grosso L, Marsh S, Langaee T, Farnett LE, Voora D, Veenstra DL, Glynn RJ, Barrett A, McLeod HL. Use of pharmacogenetic and clinical factors to predict the therapeutic dose of warfarin. Clin Pharmacol Ther. 2008;84:326–331. doi: 10.1038/clpt.2008.10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wu AH, Wang P, Smith A, Haller C, Drake K, Linder M, Valdes R., Jr Dosing algorithm for warfarin using CYP2C9 and VKORC1 genotyping from a multi-ethnic population: comparison with other equations. Pharmacogenomics. 2008;9:169–178. doi: 10.2217/14622416.9.2.169. [DOI] [PubMed] [Google Scholar]
- 15.Zhu Y, Shennan M, Reynolds KK, Johnson NA, Herrnberger MR, Valdes R, Jr, Linder MW. Estimation of warfarin maintenance dose based on VKORC1 (−1639 G>A) and CYP2C9 genotypes. Clin Chem. 2007;53:1199–1205. doi: 10.1373/clinchem.2006.078139. [DOI] [PubMed] [Google Scholar]
- 16.Sconce EA, Khan TI, Wynne HA, Avery P, Monkhouse L, King BP, Wood P, Kesteven P, Daly AK, Kamali F. The impact of CYP2C9 and VKORC1 genetic polymorphism and patient characteristics upon warfarin dose requirements: proposal for a new dosing regimen. Blood. 2005;106:2329–2333. doi: 10.1182/blood-2005-03-1108. [DOI] [PubMed] [Google Scholar]
- 17.King CR, Porche-Sorbet RM, Gage BF, Ridker PM, Renaud Y, Phillips MS, Eby C. Performance of commercial platforms for rapid genotyping of polymorphisms affecting warfarin dose. Am J Clin Pathol. 2008;129:876–883. doi: 10.1309/1E34UAPR06PJ6HML. [DOI] [PubMed] [Google Scholar]
- 18.Lenzini PA, Grice GR, Milligan PE, Dowd MB, Subherwal S, Deych E, Eby CS, King CR, Porche-Sorbet RM, Murphy CV, Marchand R, Millican EA, Barrack RL, Clohisy JC, Kronquist K, Gatchel SK, Gage BF. Laboratory and clinical outcomes of pharmacogenetic vs. clinical protocols for warfarin initiation in orthopedic patients. J Thromb Haemost. 2008;6:1655–1662. doi: 10.1111/j.1538-7836.2008.03095.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Gage BF, Eby C, Milligan PE, Banet GA, Duncan JR, McLeod HL. Use of pharmacogenetics and clinical factors to predict the maintenance dose of warfarin. Thromb Haemost. 2004;91:87–94. doi: 10.1160/TH03-06-0379. [DOI] [PubMed] [Google Scholar]
- 20.Borgiani P, Ciccacci C, Forte V, Romano S, Federici G, Novelli G. Allelic variants in the CYP2C9 and VKORC1 loci and interindividual variability in the anticoagulant dose effect of warfarin in Italians. Pharmacogenomics. 2007;8:1545–1550. doi: 10.2217/14622416.8.11.1545. [DOI] [PubMed] [Google Scholar]
- 21.Cooper GM, Johnson JA, Langaee TY, Feng H, Stanaway IB, Schwarz UI, Ritchie MD, Stein CM, Roden DM, Smith JD, Veenstra DL, Rettie AE, Rieder MJ. A genome-wide scan for common genetic variants with a large influence on warfarin maintenance dose. Blood. 2008;112:1022–1027. doi: 10.1182/blood-2008-01-134247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Takahashi H, Wilkinson GR, Nutescu EA, Morita T, Ritchie MD, Scordo MG, Pengo V, Barban M, Padrini R, Ieiri I, Otsubo K, Kashima T, Kimura S, Kijima S, Echizen H. Different contributions of polymorphisms in VKORC1 and CYP2C9 to intra- and inter-population differences in maintenance dose of warfarin in Japanese, Caucasians and African-Americans. Pharmacogenet Genomics. 2006;16:101–110. doi: 10.1097/01.fpc.0000184955.08453.a8. [DOI] [PubMed] [Google Scholar]
- 23.Millican EA, Lenzini PA, Milligan PE, Grosso L, Eby C, Deych E, Grice G, Clohisy JC, Barrack RL, Burnett RS, Voora D, Gatchel S, Tiemeier A, Gage BF. Genetic-based dosing in orthopedic patients beginning warfarin therapy. Blood. 2007;110:1511–1515. doi: 10.1182/blood-2007-01-069609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Limdi NA, Arnett DK, Goldstein JA, Beasley TM, McGwin G, Adler BK, Acton RT. Influence of CYP2C9 and VKORC1 on warfarin dose, anticoagulation attainment and maintenance among European-Americans and African-Americans. Pharmacogenomics. 2008;9:511–526. doi: 10.2217/14622416.9.5.511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Bodin L, Horellou MH, Flaujac C, Loriot MA, Samama MM. A vitamin K epoxide reductase complex subunit-1 (VKORC1) mutation in a patient with vitamin K antagonist resistance. J Thromb Haemost. 2005;3:1533–1535. doi: 10.1111/j.1538-7836.2005.01449.x. [DOI] [PubMed] [Google Scholar]
- 26.Harrington DJ, Underwood S, Morse C, Shearer MJ, Tuddenham EG, Mumford AD. Pharmacodynamic resistance to warfarin associated with a Val66Met substitution in vitamin K epoxide reductase complex subunit 1. Thromb Haemost. 2005;93:23–26. doi: 10.1160/TH04-08-0540. [DOI] [PubMed] [Google Scholar]
- 27.Rost S, Fregin A, Ivaskevicius V, Conzelmann E, Hortnagel K, Pelz HJ, Lappegard K, Seifried E, Scharrer I, Tuddenham EG, Muller CR, Strom TM, Oldenburg J. Mutations in VKORC1 cause warfarin resistance and multiple coagulation factor deficiency type 2. Nature. 2004;427:537–541. doi: 10.1038/nature02214. [DOI] [PubMed] [Google Scholar]
- 28.D'Ambrosio RL, D'Andrea G, Cafolla A, Faillace F, Margaglione M. A new vitamin K epoxide reductase complex subunit-1 (VKORC1) mutation in a patient with decreased stability of CYP2C9 enzyme. J Thromb Haemost. 2007;5:191–193. doi: 10.1111/j.1538-7836.2006.02261.x. [DOI] [PubMed] [Google Scholar]
- 29.Loebstein R, Dvoskin I, Halkin H, Vecsler M, Lubetsky A, Rechavi G, Amariglio N, Cohen Y, Ken-Dror G, Almog S, Gak E. A coding VKORC1 Asp36Tyr polymorphism predisposes to warfarin resistance. Blood. 2007;109:2477–2480. doi: 10.1182/blood-2006-08-038984. [DOI] [PubMed] [Google Scholar]
- 30.Wilms EB, Touw DJ, Conemans JM, Veldkamp R, Hermans M. A new VKORC1 allelic variant (p.Trp59Arg) in a patient with partial resistance to acenocoumarol and phenprocoumon. J Thromb Haemost. 2008;6:1224–1226. doi: 10.1111/j.1538-7836.2008.02975.x. [DOI] [PubMed] [Google Scholar]
- 31.Bodin L, Perdu J, Diry M, Horellou MH, Loriot MA. Multiple genetic alterations in vitamin K epoxide reductase complex subunit 1 gene (VKORC1) can explain the high dose requirement during oral anticoagulation in humans. J Thromb Haemost. 2008;6:1436–1439. doi: 10.1111/j.1538-7836.2008.03049.x. [DOI] [PubMed] [Google Scholar]
- 32.Harrington DJ, Gorska R, Wheeler R, Davidson S, Murden S, Morse C, Shearer MJ, Mumford AD. Pharmacodynamic resistance to warfarin is associated with nucleotide substitutions in VKORC1. J Thromb Haemost. 2008;6:1663–1670. doi: 10.1111/j.1538-7836.2008.03116.x. [DOI] [PubMed] [Google Scholar]
- 33.Scott SA, Edelmann L, Kornreich R, Desnick RJ. Warfarin pharmacogenetics: CYP2C9 and VKORC1 genotypes predict different sensitivity and resistance frequencies in the Ashkenazi and Sephardi Jewish populations. Am J Hum Genet. 2008;82:495–500. doi: 10.1016/j.ajhg.2007.10.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Anderson JL, Horne BD, Stevens SM, Grove AS, Barton S, Nicholas ZP, Kahn SF, May HT, Samuelson KM, Muhlestein JB, Carlquist JF. Randomized trial of genotype-guided versus standard warfarin dosing in patients initiating oral anticoagulation. Circulation. 2007;116:2563–2570. doi: 10.1161/CIRCULATIONAHA.107.737312. [DOI] [PubMed] [Google Scholar]
- 35.Caraco Y, Blotnick S, Muszkat M. CYP2C9 genotype-guided warfarin prescribing enhances the efficacy and safety of anticoagulation: a prospective randomized controlled study. Clin Pharmacol Ther. 2008;83:460–470. doi: 10.1038/sj.clpt.6100316. [DOI] [PubMed] [Google Scholar]
- 36.Hillman MA, Wilke RA, Yale SH, Vidaillet HJ, Caldwell MD, Glurich I, Berg RL, Schmelzer J, Burmester JK. A prospective, randomized pilot trial of model-based warfarin dose initiation using CYP2C9 genotype and clinical data. Clin Med Res. 2005;3:137–145. doi: 10.3121/cmr.3.3.137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Wen MS, Lee M, Chen JJ, Chuang HP, Lu LS, Chen CH, Lee TH, Kuo CT, Sun FM, Chang YJ, Kuan PL, Chen YF, Charng MJ, Ray CY, Wu JY, Chen YT. Prospective study of warfarin dosage requirements based on CYP2C9 and VKORC1 genotypes. Clin Pharmacol Ther. 2008;84:83–89. doi: 10.1038/sj.clpt.6100453. [DOI] [PubMed] [Google Scholar]
- 38.Schwarz UI, Ritchie MD, Bradford Y, Li C, Dudek SM, Frye-Anderson A, Kim RB, Roden DM, Stein CM. Genetic determinants of response to warfarin during initial anticoagulation. N Engl J Med. 2008;358:999–1008. doi: 10.1056/NEJMoa0708078. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.