To Test or Not to Test: Is Pharmacogenomic Testing for Warfarin Valuable?

Associate professor and director of the Center for Pharmacogenomics at the University of Florida College of Pharmacy in Gainesville
Chief of Hemostasis-Thrombosis and director of the Blood Coagulation and Vascular Biology Training Program at Beth Israel Deaconess Medical Center in Boston, Massachusetts

Warfarin has been standard anticoagulation for more than 60 years, but the significant variability in both warfarin pharmacokinetics and pharmacodynamics makes achieving therapeutic drug levels a challenge. Researchers are now investigating whether adding genomic information to dosing algorithms could better identify the optimal starting dose for patients new to warfarin, minimize variability in drug response, and, ultimately, improve clinical outcomes.

ASH Clinical News invited Larisa H. Cavallari, PharmD, and Bruce Furie, MD, to debate the question: “Is there a role for pharmacogenomic testing in guiding initial warfarin dosing?” Dr. Cavallari will be arguing on the “pro” side, and Dr. Furie will be arguing on the “con” side.

Larisa H. Cavallari, PharmD: The starting dose of warfarin depends mainly on clinical factors, including age (with older individuals generally requiring a lower dosage), followed by body mass index (with larger individuals generally requiring a higher dosage).

However, data from genome-wide association studies consistently show that certain genetic variants (including CYP2C9 and VKORC1) affect the pharmacokinetics and pharmacodynamics of warfarin and can influence warfarin dose requirements. It’s hard to argue against using these data, as long as you use the right information in the right population.

Bruce Furie, MD: Yes, genotyping does provide great information, but I’m not sure where I could use that information, and I know I certainly can’t use it alone. Warfarin dose is titrated against a patient’s vitamin K intake (which is affected by diet, certain medications, changes in vitamin K–dependent clotting factors) and other factors that influence warfarin absorption or metabolism. Medications also can greatly influence the efficacy of warfarin – making it too active or not active enough – because they share a binding site on albumin with warfarin.

Given this variability, we need to closely monitor warfarin’s efficacy through clotting tests like prothrombin time (PT) and international normalized ratio (INR). There is room for improvement with monitoring because warfarin monitoring is complicated and dangerous. However, testing a patient’s genotype doesn’t mean we still don’t have to perform these tests.

Dr. Cavallari: I don’t expect pharmacogenomics to take the place of INR monitoring. That assumption might be perceived as a potential danger of pharmacogenomics: Clinicians could become so comfortable using the genomic information that they wouldn’t perform any INR monitoring.

Instead, the value of pharmacogenomics is its potential to reduce the frequency of INR monitoring. If a genotype-guided dosing approach can get patients to a therapeutic INR or identify a stable warfarin dose more efficiently, it is possible that patients will need less frequent INR monitoring and the patient may not need to come in to the clinic as often. However, genotyping certainly wouldn’t take the place of monitoring.

The value of pharmacogenomic-guided warfarin dosing has been studied in two clinical trials ­– both of which reached different conclusions. First, in the multicenter, randomized, controlled EU-PACT trial, investigators compared time in therapeutic INR range (defined as INR 2.0-3.0) during the first 12 weeks after warfarin initiation between 455 patients with atrial fibrillation or venous thromboembolism who were treated with either a genotype-guided approach or a standard dosing approach (which included a 3-day loading-dose regimen).1 Patients in the genotype-guided group had significantly longer time in therapeutic range (67.4% vs. 60.3%; p<0.001), and fewer incidences of excessive anticoagulation (defined as supra-therapeutic INR ≥4.0: 27.0% vs. 36.6%; p=0.03) and a shorter time to reach a therapeutic INR (21 days vs. 29 days; p<0.001).

Second, in the U.S.-based COAG trial, which included a larger, more diverse patient population than the EU-PACT trial, investigators assessed time in therapeutic INR range in 1,015 patients (275 of whom were African American) treated according to one of two dosing algorithms: one that included both clinical variables and genotype data or one that included only clinical variables.2 At the end of four weeks, there was no significant benefit with the genotype-guided dosing versus the clinical-guided approach (mean percentage of time in therapeutic INR range = 45.2% vs. 45.4%, respectively; p=0.91).

The rates of adverse events (any INR of ≥4, major bleeding, or thromboembolism) did not differ significantly according to dosing strategy, as well.

So, in EU-PACT, the genotype-guided dosing approach showed some benefit, but it showed no advantage in the COAG trial.

Dr. Furie: However, in EU-PACT, the improvement was minimal – only about 6 or 7 percent. So, considering this modest change, we have to ask, “Is pharmacogenomic testing worth it?” I am all for improving warfarin therapy technology, but I’m not sure that the pharmacogenomics is the way.

Before it becomes a regular component of prescribing warfarin, I need to be convinced that adding genomics to dosing algorithms will improve patient safety. I can argue on theoretical grounds that that is not the case, but these two clinical trials have come to a similar conclusion.

These trials also studied the value of pharmacogenomics in guiding the initial dose of warfarin, not chronic warfarin therapy – an area in which we need more guidance. In my practice, I generally start patients at a low dose (5 mg/day); that amount does not cause problems, even in patients who do not metabolize warfarin rapidly. The EU-PACT and COAG trials attempted to identify patients who are susceptible to having a supratherapeutic INR on warfarin, which can be caused by variations in vitamin K intake, medications, body size, and so on. These are variables that require prothrombin monitoring to adjust dosing. Genetics, on the other hand, don’t change; they cannot be used to monitor and adjust warfarin dosing throughout the course of anticoagulation.

Dr. Cavallari: I completely agree that the true benefit of pharmacogenomic testing will be in helping us select an initial dose for patients starting warfarin. This approach may not be helpful in many patients because they will end up having a genotype that is associated with needing a normal warfarin dose; the value of pharmacogenomics is limited to those individuals who will need a particularly low or high dose based on their genotypes.

Dr. Furie: Also, to continue the discussion about pharmacogenomics and patient safety, the primary endpoints in these trials focused on anticoagulation-related outcomes, not bleeding or thrombotic complications. We all know that, even when patients are in the therapeutic INR range, they can have thrombotic or bleeding complications, and these can be clinically silent events.

Dr. Cavallari: The Genetics Informatics Trial (GIFT) hopefully will help answer questions about genotype-guided dosing’s effect on patient safety.3 The trial, which is nearly completed, is evaluating whether genotype-guided dosing can reduce the rate of adverse events in older patients who have undergone hip arthroplasty and are taking warfarin to prevent venous thromboembolism – a very different patient population than the EU-PACT and COAG trials. The primary endpoints in this very high-risk population include bleeding, thrombosis, and surrogate markers of excessive anticoagulation (INR ≥4.0), and investigators are also looking at time in the therapeutic INR range.

Dr. Furie: If GIFT or other trials do support the use of pharmacogenomic testing when initiating warfarin, the next challenge is making sure that everyone benefits equally from this approach. We work in a resource-poor environment, where patients who live in urban areas and have access to anticoagulation clinics fare better with warfarin than those who live in rural areas. The question is: How do we bring the quality of the urban areas to the rural populations?
We have more than 60 years of experience with warfarin and a standard – though cumbersome – system: A patient goes to the lab to have his or her PT measured; the lab then performs the test and faxes that information to the doctor’s office; the doctor or nurse then makes the decision about how the patient’s warfarin will be managed. I believe the infrastructure needs to be improved.

Dr. Cavallari: Turning around the genetic testing quickly could be quite labor-intensive for institutions, but we are slowly moving toward more widespread implementation of pharmacogenomics into practice. Also, as you said, a patient’s genetics don’t change: Genotyping only needs to be performed once in a patient’s lifetime. It is an expensive test, but unlike other tests such as creatinine, it needs to be done once and can have far-reaching implications.

Dr. Furie: The cost and reimbursement issues with genotype-guided warfarin dosing are a concern, but that issue is also tied up with clinical trials that may provide more answers in the future. For now, though, I think that there are ways of improving patient safety and monitoring techniques – and they don’t necessarily include pharmacogenomics.

Dr. Cavallari: Perhaps a combination of both approaches would work: Pharmacogenomics upfront to help with choosing the initial dose, then improved monitoring techniques to help with managing a patient’s chronic therapy. The two approaches might not be mutually exclusive.


  1. Pirmohamed M, Burnside G, Eriksson N, et al. A randomized trial of genotype-guided dosing of warfarin. N Engl J Med. 2013;369:2294-303.
  2. Kimmel SE, French B, Kasner SE, et al. A pharmacogenetic versus a clinical algorithm for warfarin dosing. N Engl J Med. 2013;369:2283-93.
  3. Do EJ, Lenzini P, Eby CS, et al. Genetics informatics trial (GIFT) of warfarin to prevent deep vein thrombosis (DVT): rationale and study design. Pharmacogenomics J. 2012;12;417-24.

Disclaimer: The following positions were assigned to the participants and do not necessarily reflect ASH’s opinion, the participants’ opinions, or what they do in daily practice.

Agree? Disagree? We want to hear from you! Send your thoughts and opinions on this controversial issue to [email protected].