Mixed Results With Pharmacogenetics to Guide Warfarin Dosing

Susan Jeffrey

November 20, 2013

DALLAS, Texas — Two new trials examining the utility of genetic testing to guide warfarin dosing have come to divergent conclusions.

One, the EU Pharmacogenetics of Anticoagulant Therapy (EU-PACT) Warfarin Study, conducted in Sweden and the United Kingdom (UK), showed that pharmacogenetic-based dosing was associated with a higher percentage of time that the international normalized ratio (INR) was in therapeutic range during the first 12 weeks of treatment in treatment-naive patients with atrial fibrillation (AF) or venous thromboembolism (VTE).

Conversely, results of the Clarification of Optimal Anticoagulation through Genetics (COAG) trial, conducted in the United States, showed no advantage of pharmacogenetic testing in addition to clinical variables on improving time in therapeutic range up to 4 weeks after the start of therapy, also in treatment-naive patients. African-American patients actually did better without the genetic testing, the researchers reported.

Dr. Stephen Kimmel

"My feeling on the COAG trial, based on the overall results and in the African American group, is that the value added of genetics above and beyond clinical information, doesn't support the use of genotyping at all in patients prior to starting the drug," COAG lead investigator Stephen Kimmel, MD, professor of medicine and epidemiology, University of Pennsylvania School of Medicine, Philadelphia, told reporters during a press conference here.

In response, EU-PACT lead investigator Munir Pirmohamed, MD, PhD, NHS chair of pharmacogenetics, University of Liverpool, United Kingdom, pointed out that their trial "shows something different." Clinics in the UK and Sweden use clinical software programs to help maintain control of anticoagulation. The time in therapeutic range in the standard-care group in EU-PACT was very good, he noted, "but despite that, we were able to show that there was an improvement with genotyping."

Dr. Munir Pirmohamed

Dr. Pirmohamed attributed the difference in these findings to the specific algorithms used in conjunction with the genetic testing. "The algorithm that we used has been tested in a European population, and therefore we utilized it in a European population," he said in an interview. "And so if we want to institute that in all ethnic groups, we would need to do the work," creating ethnicity-specific algorithms.

Results of both trials were published online November 19 in the New England Journal of Medicine to coincide with their presentation here at the American Heart Association (AHA) Scientific Sessions 2013.

A third report, also from the EU-PACT group, showed that genotype-guided dosing of 2 other anticoagulants (acenocoumarol and phenprocoumon) did not improve the time in therapeutic range among in patients with AF or VTE.


EU-PACT was a single-blind, multicenter, randomized controlled trial that enrolled 455 patients with AF or VTE receiving warfarin for the first time.

For patients in the usual care group, warfarin doses were prescribed according to pharmacogenetic-based algorithms for the first 5 days. On day 4, the INR was checked and the next dose was given by using standardized software, as well as subsequent doses from day 6 onwards.

"Anticoagulation practices in the UK and Sweden are highly developed in terms of using these computerized software packages without incorporating any genetics in there at all," Dr. Pirmohamed noted.

Genotyping for CYP2C9*2, CYP2C9*3, and VKORC1 was done by using a point-of-care test that provides results in about 2 hours, Dr. Pirmohamed said. To incorporate the genetic information, the investigators used 2 different algorithms to define the individualized loading dose and then to revise the dose after the first INR check at day 4.

The primary outcome measure was the percentage of time in the therapeutic range for warfarin of an INR of 2.0 to 3.0 during the first 12 weeks of treatment.

The authors report that the time in therapeutic range was significantly higher for the genotype-guided group vs standard care.

Table 1. EU-PACT: Primary Outcome

Endpoint Genotype-Guided Group (%) Control Group (%) Adjusted Difference (95% Confidence Interval) (percentage points) P Value
Time INR in therapeutic range at 12 weeks 67.4 60.3 7 (3.3 - 10.6) <.001


There were significantly fewer instances of excessive anticoagulation, defined as an INR of 4 or higher, in the genotype-guided group, and the median time to reach a therapeutic INR was 21 days vs 29 days for the standard care group (P < .001).

"The first thing we need to do next is to look at the cost-effectiveness at what we did, incorporating the cost of the point-of-care genotyping," Dr. Pirmohamed said, which he estimated to be about €50 per test.

"My feeling is at the moment that we would need to undertake further implementation studies to determine whether this really had an impact over long term in terms of clinical outcome measures," he added during panel discussion after the presentation. "As I said, the limitation of our trial is we didn't look at clinical outcome measures."


The COAG trial included 1015 patients initiating warfarin therapy who were randomly assigned to receive warfarin dosing during their first 5 days of therapy based on a dosing algorithm using both clinical variables and genotype information for the CYP2C9*2, CYP2C9*3, and VKORC1, or clinical data alone.

This trial was double blind, with neither patients nor clinicians aware of the warfarin dose during the first 4 weeks of treatment. The primary endpoint was again the percentage of time the INR was within the therapeutic range, but the time point was different, measured from day 4 or 5 to day 28.

At 4 weeks, there was no significant difference between the groups on this measure, Dr. Kimmel said.

Table 2. COAG: Primary Endpoint

Endpoint Genotype-Guided Group (%) Clinically Guided Group (%) Adjusted Mean Difference (95% Confidence Interval) (percentage points) P Value
Time INR in therapeutic range at 4 weeks 45.2 45.4 –0.2 (–3.4 to 3.1) .91


No significant difference was seen between groups among patients with a predicted dose difference between the 2 algorithms of 1 mg per day or more, the authors note.

There was a significant interaction between the dosing strategy used and race (P = .003), Dr. Kimmel pointed out. "As you can see, in the African American population, the clinical algorithm did significantly better on anticoagulation control than the genetic algorithm," he said. This group had a significantly higher number of patients with INRs over 3, and took a significantly longer amount of time to get to their first therapeutic INR vs African Americans in the clinical algorithm group, while there were no differences for non-African Americans on these endpoints between groups.

The groups did not differ in the rates of the combined safety outcome of any INR of 4 or more, major bleeding, or thromboembolism.

"These findings highlight the importance of developing and evaluating pharmacogenetic testing in patients from diverse racial and ethnic backgrounds," said Gary H. Gibbons, MD, director of the National Heart, Lung, and Blood Institute, Bethesda, Maryland, which funded the COAG trial. "We are optimistic about the prospects of personalized, precision medicine, but we must make sure that we put these approaches through the same type of rigorous testing as any other prognostic test or clinical treatment strategy."

Trial Results Similar

In an editorial published with these papers, Bruce Furie, MD, from the Harvard Medical School and the Division of Hemostasis and Thrombosis, Beth Israel Deaconess Medical Center, Boston, Massachusetts, writes that the results of all 3 of these trials are actually similar, showing no between-group difference for adding genetic information. The results of EU-PACT, Dr. Furie writes, are "significantly different albeit similar, indicating a modest improvement."

"What can we conclude from these trials? First, we must recall that these trials address the process of the initiation of anticoagulant therapy — during the very first week — and not an approach to intermediate or long-term anticoagulation," he notes. "Second, it would appear that, despite the variation in trial design, these trials indicate that this pharmacogenetic testing has either no usefulness in the initial dosing of vitamin K antagonists or, at best, marginal usefulness, given the cost and effort required to perform this testing."

Instead, Dr. Furie suggests more attention be paid to improvements in the infrastructure of INR testing, the use of formal algorithms for dosing, and increased diligence in following these patients, "given the high percentage of medical mismanagement associated with these anticoagulant agents."

Here at the AHA meeting, Patrick Ellinor, MD, PhD, director, Arrhythmia/Step Down Unit, Massachusetts General Hospital Heart Center, Boston, was the invited discussant for both the EU-PACT and the COAG trials.

He reviewed the differences between the trials and pointed out that the findings in the EU-PACT trial might be representative of "the standard of care that's commonly employed with a fixed dose for many of our patients vs the incorporation of additional clinical variables."

Although these genetic differences account for 35% to 40% of the variability, other variables, including include age, race, smoking, body surface area, and medication use, also account for a large amount of this variability, "so the incremental benefit of the genetic data on top of these other clinical data is limited."

He suggests that warfarin should not be used as a fixed dose, these 3 genetic variants should not be relied upon for African American patients, and physicians should be incorporating clinical algorithms when initiating warfarin therapy. "It's not being done in a systematic way in most healthcare systems," Dr. Ellinor noted.

But during a panel discussion after presentation of these studies, Elaine Hylek, MD, professor of medicine at Boston University School of Medicine, pointed out, "I think all of us, from a scientific stand point, clearly believe that there has to be an association when 40% of the variability [in anticoagulation control] is explained with genotyping. I think what's difficult, in trying to better understand and to test interventions in this area, has already been stated, is that we're relying on INR, relying on surrogate measures, which was the reason that Medicare did not want to reimburse for genotyping, because there were not hard endpoints associated with the genotype dosing."

The other issue that she finds "fascinating," she noted, is the good results seen in the comparator group with warfarin. "I'm hearing about the EU-PACT trial, that your comparison used software to guide [dosing], and I really think that probably isn't usual care. And the fact that you were so vigilant about measuring the INR on day 4, which actually helps you with the slope of where your INR is going, to improve the comparison arm, I think makes it more difficult to demonstrate the benefit of the genotype dosing."

At the end of the day, though, all of this discussion about genotyping for warfarin treatment may be moot, even Dr. Pirmohamed acknowledged. "I guess the elephant in the room is the novel anticoagulants, how they're taken up, and obviously there are data showing they are as effective, and may be safer, particularly with regard to brain hemorrhage. I completely agree with that."

The EU-PACT trial was funded by the EU-FP7 Program. COAG was funded by the National Heart, Lung, and Blood Institute. GenMark Diagnostics and AutoGenomics Inc loaned genotype platforms to the study, and Bristol-Myers Squibb donated the study drug.

N Engl J Med. Published online November 19, 2013. EU-PACT COAG Editorial

American Heart Association (AHA) Scientific Sessions 2013. Abstracts 19658, 19535. Presented November 19, 2013.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.