American Association for Clinical Chemistry
Better health through laboratory medicine
February 2008 Clinical Laboratory News: Warfarin Pharmacogenetics


February 2008: Volume 34, Number 2 

Warfarin Pharmacogenetics
The Challenges of Laboratory Testing

By Linnea M. Baudhuin, PhD



Each year, about 2 million people begin warfarin therapy to prevent or treat blood clots (1). However, as many as 43,000 patients experience the drug’s life-threatening bleeding complications that require emergency treatment. In recent years, underlying genetic factors have been shown to account for approximately 35–40% of the variation observed in how patients respond to warfarin (2). In light of the potential dire consequences of incorrect warfarin dosing, the FDA recently approved updated labeling for Coumadin (Bristol-Myers Squibb, New York, N.Y.), the brand name version of warfarin, to highlight the importance of warfarin pharmacogenetic (PGx) testing and to suggest that physicians incorporate the information obtained from such testing into warfarin dosing decisions. Manufacturers of generic warfarin are also expected to add similar information to their products’ labeling. Shortly after the FDA approved the updated labeling, it cleared the first genetic test for warfarin sensitivity, and other manufacturers have announced plans to submit similar tests to the FDA.


These recent developments place the clinical laboratory in a unique position to help usher in the era of personalized medicine in which a patient’s genotype is used to select the appropriate drug and dosage. However, this field is not without significant challenges for the laboratory.


A Warfarin Primer

One of the most commonly prescribed drugs, warfarin is a coumarin-based anticoagulant that is often used for the short- and long-term management of thromboembolic and hemostatic disorders, such as deep-vein thrombosis, pulmonary embolism, inherited thrombophilia, and antiphospholipid syndrome. Physicians also prescribe warfarin for a number of other medical conditions and treatments, including the prevention of myocardial infarction and stroke, atrial fibrillation, and orthopedic surgery.


Because warfarin has a narrow therapeutic window, achieving a stable international normalized ratio (INR) through dosing and titration has been a significant challenge for physicians, largely due to multiple factors affecting an individual’s warfarin sensitivity and resistance. Dosing is monitored by coagulation testing, generally 4–5 days after initial loading dose and at regular intervals thereafter, to maintain the INR within specific limits. Physicians typically give patients an initial warfarin dose of 2–10 mg, but depending on INR results, the weekly maintenance dose can range from 4–80 mg (3).


In addition to the genetic factors related to warfarin’s metabolism and action, an individual’s response to the drug also depends on other factors, including sex, body size, age, drug-drug interactions, nutritional status, and comorbidities such as liver disease. If the warfarin dose is too high, the patient is at risk for major bleeding complications, such as intracranial hemorrhage, especially during the first weeks of therapy. On the other hand, if the warfarin dose is too low, the patient is at risk for dangerous blood clots that could potentially cause a thromboembolic stroke or other serious complication.


Although warfarin can be a very effective drug, its narrow therapeutic index also leads to a high rate of adverse drug events (ADEs). In fact, it ranks among the top ten drugs associated with serious ADEs—including a major bleeding frequency rate as high as 10–16% (4) —making it one of the drugs most frequently associated with hospital emergency department visits due to ADEs. Overall, anticoagulants were ranked as the primary cause of ADE-related deaths in 2003 and 2004 (3). Consequently, in October 2006, the FDA instituted a black-box labeling change to warfarin emphasizing the serious bleeding risks associated with the drug and the importance of regular INR monitoring.


While physicians have traditionally used nongenetic factors to individualize warfarin therapy, clearly, the ADE data show that nongenetic factors alone are insufficient for predicting warfarin dose variability in many, if not the majority, of cases. In fact, a combined approach incorporating both nongenetic and known genetic factors can account for 50–60% of warfarin dose variability (1, 5).


As mentioned above, the FDA updated the warfarin package label in August 2007 to provide information about genetic testing for warfarin sensitivity and encourage physicians to use the data to reduce ADEs. While the FDA stopped short of requiring genetic testing prior to a patient taking the drug, labs are likely to see more requests for testing as physicians become aware of how to use this information to better define dosing strategies on an individualized basis.


Warfarin Pharmacogenetics


Pharmacogenetic analysis of clinically relevant genetic markers is emerging as a useful tool to predict an individual’s response to certain drugs, as well as a means to avoid potential ADEs. The science of warfarin PGx is unique, however, in comparison to other PGx applications, since it is based on both the pharmacokinetics and pharmacodynamics of the drug  (Figure 1 below), rather than solely on one or the other. The major genes implicated in pharmacokinetic metabolism and pharmacodynamic targeting of warfarin are cytochrome P450 2C9 (CYP2C9) and vitamin K epoxide reductase complex subunit 1 (VKORC1).



To understand the effect of these genes, it is helpful to look at the chemical

structure of warfarin, which is comprised of a racemic mixture of stereoisomers, known as R- and S-enantiomers. As the principal active form of warfarin, S-warfarin is approximately three to five times as potent as the R-enantiomer in inhibiting the enzyme vitamin K epoxide reductase (VKOR). The hepatic CYP2C9 enzyme is primarily responsible for the metabolism of S-warfarin (Fir).


Genetic polymorphisms affecting CYP2C9 activity occur at a fairly high frequency in some populations and may result in decreased clearance of S-warfarin (Table 1). These individuals are more sensitive to the drug, and normal doses tend to result in more frequent overmedication with warfarin, increased INRs above the therapeutic target level, and the potential for serious bleeding events. One study documented that individuals carrying the two most commonly studied CYP2C9 polymorphisms, *2 (2608C>T) and *3 (42614A>C), have a two- to threefold greater chance of bleeding during initial warfarin dosing (3).


The most common (or wild-type) allele for CYP2C9 is CYP2C9*1, which is associated with normal or extensive enzyme activity. The allele frequency of CYP2C9*1 in Caucasian, African American, and Asian populations is approximately 80–82%, 83–98%, and 97–98%, respectively (1). CYP2C9 alleles that are known to lead to inactive enzyme or reduced enzymatic activity include CYP2C9 *2, *3, *4, *5, *6, and *11 (Table 1, below). Of these variants, CYP2C9 *2 and *3 are more frequently present in Caucasians, 8–13% and 6–10%, respectively, but comprise <1–4% of alleles in African Americans and Asians. In the Asian population, CYP2C9 alleles other than *1 and *3, which occur with 1–2% frequency, are not observed at an appreciable frequency. In addition to the CYP2C9 *2 allele, *5 and *11 each occur with a frequency of approximately 3% in African Americans. An additional allele, CYP2C9 *8 has been shown to occur in approximately 6% of African Americans, but its effect on enzyme activity in the presence of warfarin has not been determined.



Table 1
CYP2C9 Alleles and Effect on Enzyme Metabolism
CYP2C9 Allele
Nucleotide Change
Effect on Enzyme Metabolism
None (wild type)
Normal (extensive) activity
Reduced activity
Minimal activity
Reduced activity
Reduced activity
No activity
Reduced activity

Pharmacodynamically, warfarin acts by inhibiting VKOR, which interferes with recycling of vitamin K and decreases its availability (Figure 1). Since vitamin K is necessary for the activation of key coagulation factors—factors II, VII, IX, and X, and proteins C, S, and Z—a decreased pool of vitamin K leads to thrombin formation. VKOR is encoded by the VKORC1 gene, which has multiple polymorphisms that affect its expression. In particular, a polymorphism within the promoter of VKORC1 (-1639G>A) decreases expression of the gene: a heterozygous or homozygous adenine (A) at position -1639 in the VKORC1 promoter significantly reduces VKOR expression compared with individuals who are homozygous for a guanine (G) at that position. Studies have demonstrated that the wild type -1639G VKORC1 promoter has 44% higher activity compared to the variant -1639A promoter (6). As expected, the -1639A genotype corresponds to warfarin sensitivity, so that a patient with this genotype will require a lower dose of warfarin compared to the -1639G genotype. The frequency of the lower dose, warfarin-sensitive homozygous VKORC1 -1639A genotype is approximately 14–17% in Caucasians, 72–78% in Asians, and 4–5% in African Americans (4, 7). The high frequency of this VKORC1 promoter polymorphism in Asians is thought to explain the majority, but not all, of the warfarin-sensitive phenotypes in this population.

Laboratory Testing for Warfarin PGx

Incorporating results from genetic testing of both CYP2C9 (*2 and *3) and VKORC1 (-1639G>A) into warfarin dosing prediction models is expected to account for approximately 50–60% of the variability in warfarin response (1, 5). With the increased emphasis on patient safety and preventing ADEs, as well as the widespread use of molecular diagnostic technologies in clinical labs of all sizes, many labs have initiated or are considering warfarin PGx testing.

But significant challenges for performing this test exist, including: choosing an appropriate analytical platform; meeting the demand of rapid turnaround time; providing meaningful interpretation of results, such as dosing guidance based on genotype and potential drug-drug and drug-CYP2C9 interactions; and keeping abreast of reimbursement and regulatory issues.

Today, labs have a choice of several molecular platforms to devise homebrew CYP2C9 and VKORC1 genotyping tests, including bead-based systems, microarrays, fluorescent probe-based assays, allele-specific PCR, and pyrosequencing. Labs may also choose to use a commercial assay recently cleared by the FDA. The Verigene Warfarin Metabolism Nucleic Acid Test (Nanosphere, Northbrook, Ill.) is a pharmacogenetic test to assess for warfarin sensitivity that uses a multiplexed platform for detecting nucleic acid targets for the CYP2C9 *2 and *3 polymorphisms, as well as the VKORC1 1173C>T polymorphism, which is in strong linkage disequilibrium with -1639G>A).

Laboratory-developed assays allow labs to define their own set of polymorphisms for warfarin sensitivity based on the ethnic populations served. For example, alleles other than CYP2C9 *2 and *3 that predominate in European Caucasian populations may be required for appropriate testing in other populations, such as CYP2C9 *5 and *11, which occur more frequently in individuals of African descent. Other factors that influence the lab’s choice of analytical platforms include cost, ease of use, and turnaround time.

In fact, turnaround time is a major issue in warfarin PGx testing. The importance of obtaining a patient’s genotype prior to initiating warfarin therapy has been promoted by many supporters of personalized medicine. While the highest risk for warfarin-associated ADEs is during the initial phase of therapy, the necessity of a single day turnaround time for genotyping CYP2C9 and VKORC1 is controversial. This is largely because variants in these genes do not affect warfarin volume of distribution but rather the time for plasma warfarin concentration to achieve therapeutic levels (VKORC1) and steady state (CYP2C9) (2).

Reynolds et al. (2) promote a standard initial dose of warfarin, 5 mg/day, for the first 3–4 days of therapy. In this scenario, if genotyping is initiated on day 1, then with a maximum 4-day turnaround time for genotyping, the results could be incorporated into maintenance dosing strategies. The “standard initial dose/genotype-guided maintenance dose” approach is especially appealing in the acute/emergent setting and when the test must be sent out to a reference lab, which precludes the possibility of a 1-day turnaround time.

On the other hand, the advantages to knowing a patient’s genotyping results prior to initiating therapy cannot be disregarded. If a patient’s genotype-guided maintenance dose is known prior to initiation of therapy, then ideally, the maintenance dose would essentially be the same as the initiation dose. Another advantage to upfront genotyping is related to the timing of the first INR measurement. As mentioned above, patients are usually tested 4–5 days after the initial warfarin dose is given, and this INR is used to titrate the subsequent dose until the patient achieves a target INR. However, in an individual with a warfarin-sensitive CYP2C9 genotype, the time to achieve warfarin steady-state plasma concentrations is delayed. Measuring the patient’s INR at the standard 4–5 days would be premature (2). In such cases, if the genotyping results were available prior to initiating treatment, the first INR measurement could be scheduled more appropriately. Overall, by having the genotyping results upfront, more comprehensive counseling on the recommended therapeutic management plan could be provided to the patient at an earlier stage. This approach could lessen the number of ADEs and may give both the clinician and patient an increased level of confidence when selecting a warfarin dosing schedule. For these reasons, labs may choose to implement rapid, 1-day genotyping.

Other factors that labs need to consider are the number of staff required to do the testing within the desired turnaround time, the level of training required, and how to review results and release reports. Results of CYP2C9 and VKORC1 genotyping can range from a report with minimal dosing guidance to one with more explicit dosing guidance depending on patient information available to the lab at time of testing. While several warfarin dosing algorithms exist, there is neither enough data nor a consensus that supports highly specific genotype-based warfarin dosing guidance for all patient populations (8). In addition, unless the client provides such information when ordering the test, it is difficult for the lab to obtain patient information such as age, body size, co-drugs, etc., in a timely fashion. Clearly, these factors influence how specific the dosing recommendations can be.

Nonetheless, the lab is in a position to provide some general warfarin dosing guidance based on the genotype analysis. At a minimum, the lab report should indicate which genotypes have a normal, mild, moderate, high, or very high sensitivity to warfarin (Table 2, below). In addition to sensitivity, the lab should also provide general guidance regarding frequency of INR monitoring and warfarin dosing. For example, an individual who is CYP2C9 *1/*3, or heterozygous for the *3 polymorphism, and VKORC1 -1639 G/A is predicted to have a moderate sensitivity to warfarin, so more frequent INR monitoring and warfarin dose reduction should be considered.

Table 2
Sensitivity to Warfarin Based on Combined CYP2C9 and VKORC1 Genotypes
Warfarin sensitivity
CYP2C9 genotype
VKORC1 promoter genotype
Less than normal
Very high

The value of knowing a patient’s genotype for CYP2C9 extends beyond warfarin. Because many other drugs are influenced by this gene, it will also be important for labs to consider including other information in the patient genotyping results, such as any patient-specific potential drug-CYP2C9 interactions, a list of more common drug-CYP2C9 and drug-warfarin interactions, and a reference to the warfarin package insert. For example, co-administration of warfarin with drugs known to decrease the activity of CYP2C9, including amiodarone, fluconazole, fluvastatin, fluvoxamine, isoniazid, lovastatin, and ticlopdine (9), will lead to an increased possibility of toxicity, particularly in individuals with CYP2C9 variant alleles.

While drug-CYP2C9 and drug-warfarin interactions are important for the laboratory to recognize and report, oftentimes the list of drugs the patient is taking is not provided to the laboratory. In this case, it can become a major undertaking that requires additional staff time to track down this information in a timely fashion. In these situations, the responsibility should fall on the client to consult with the lab and/or a clinical pharmacologist if needed. In addition, because of the paucity of information regarding specific warfarin dosing guidance for most drugs that interact with CYP2C9 or warfarin, it is very difficult for the laboratory to provide that information when reporting testing results. An exception to this is when the patient is taking amiodarone, atorvastatin, or azole or sulfa antibiotics, in which case, an approximate 15–25% warfarin dose reduction is recommended.

The Wave of the Future

Although the field is just emerging, personalized medicine clearly represents the wave of the future. As experience with warfarin PGx grows, laboratorians will want to close-ly monitor physicians’ attitudes toward this new practice for patient dosing. In addition, forthcoming resources and guidelines will help laboratorians add warfarin PGx to their labs’ menus. The American College of Medical Genetics (ACMG) is currently drafting a warfarin PGx laboratory testing standards and guidelines document that is expected to contain their recommendations for laboratory reporting of results. A draft of general guidelines regarding PGx genotype reporting is also available from the National Academy of Clinical Biochemistry on its Web site (10).

Overall, while the promise of warfarin PGx is apparent, there are many hurdles to overcome in providing optimal individual patient management. The forthcoming laboratory practice guidelines from ACMG and NACB will likely have a significant impact on improving and standardizing clinical warfarin PGx testing. Furthermore, many centers are currently conducting studies to investigate and better establish warfarin dosing algorithms.

Eventually, clearer and more widely accepted warfarin dosing guidance based on a combination of genetic and nongenetic factors will likely become available. Such guidance will greatly enhance the clinical practice of warfarin PGx as well as promote more widespread adoption of its use.


  1. McWilliam A, Lutter R, Nardinelli C. Health Care Savings from Personalized Medicine Using Genetic Testing: The Case of Warfarin; Working Paper 06-23. AEI-Brookings Joint Center for Regulatory Studies, November 2006.
  2. Reynolds KR, Valdes R Jr, Hartung BR, Linder MW. Individualizing warfarin therapy. Personal Med 2007;4:11-31.
  3. Gage BF, Lesko LJ. Pharmacogenetics of warfarin: regulatory, scientific, and clinical issues. J Thromb Thrombolysis 2007. DOI# 10.1007/s11239-007-0104-y. (Accessed December 2007)
  4. Wysowski DK, Nourjah P, Swartz L. Bleeding complications with warfarin use: a prevalent adverse effect resulting in regulatory action. Arch Intern Med 2007; 167:1414-1419.
  5. 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. Pharmacogenetics J 2005;5:262-270.
  6. Yuan HY, Chen JJ, Lee MT, Wung JC, Chen YF, Chang 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 2006;14:1745-1751.
  7. Marsh S, King CR, Porche-Sorbet RM, Scott-Horton TJ, Eby CS. Population variation in VKORC1 haplotype structure. J Thromb Haem 2006;4:473-474.
  8. Baudhuin LM, Langman LJ, O'Kane DJ. Translation of pharmacogenetics into clinically relevant testing modalities. Clin Pharmacol Ther. 2007;82:373-376.
  9. FDA Website
  10. Valdes R, Payne DP, Linder MW, et al. The National Academy of Clinical Biochemstry. Laboratory medicine practice guidelines: guidelines and recommendations for laboratory analysis and application of pharmacogenetics to clinical practice. Draft Guidelines, Version 0606. Draft Guidelines

Linnea M. Baudhuin, PhD, DABMG, is Co-Director of the Nucleotide Polymorphism Laboratory, Co-Director of Cardiovascular Laboratory Medicine, and Assistant Professor of Laboratory Medicine in the Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, Minn.