COAG and EU-PACT: The Pharmacogenetics of Warfarin Dosing

Naveen Pereira, MD


December 05, 2013

Editorial Collaboration

Medscape &

This feature requires the newest version of Flash. You can download it here.

COAG and EU-PACT: The Pharmacogenetics of Warfarin Dosing

Naveen Pereira, MD: Hi. My name is Naveen Pereira. I am a consultant in cardiovascular diseases at Mayo Clinic in Rochester. I specialize in heart failure and transplant, but I have a specific research interest in pharmacogenetics. Today, I would like to talk to you about two very interesting warfarin pharmacogenetic trials that were presented at the recently concluded American Heart Association meetings and were published online in the New England Journal of Medicine.

The two trials are EU-PACT, from the United Kingdom and Europe, and the COAG trial, which really emanated from the United States.[1,2] We know that warfarin has a very narrow therapeutic index, and in fact one-third of emergency-room visits due to adverse drug reactions is due to warfarin; therefore, the FDA has emphasized that warfarin dosing should be carefully made based on INR, but specifically the dosing recommendations in the drug labeling information are based on the genotype of patients. The two most important genes that determine warfarin dose variability are CYP2C9 and VKORC1. CYP2C9 is essentially a metabolic enzyme that clears warfarin and VKORC1 is the molecular target of warfarin. Genetic variation in these two genes leads to increased sensitivity in the carriers of these genotypes, and doses typically need to be decreased.

Both of these trials have looked at common genetic variation in CYP2C9 and VKORC1. The first trial is the EU-PACT trial, from Europe, and it was a randomized controlled trial involving 455 patients, where one arm received standard dosing, so subjects who were less than 75 years got 10 mg on day 1, 5 mg on day 2, and 5 mg on day 3, and subjects who were greater than 75 years of age got 5 mg, 5 mg, and 5 mg on days 1 through 3. The other arm was a genotype-guided dosing arm, and the two arms were compared for the primary outcome, which was the percentage of time that these subjects were in therapeutic INR within a 12-week period. The results were that the percentage of time in therapeutic INR within 12 weeks was 67.4% in the genotyping arm. While using standard dosing, the percentage of time in therapeutic INR was 60.3%, and this difference was statistically significant, with the p value of <0.001. There were no differences in other outcomes, secondary outcomes assessed such as in bleeding or thromboembolism events. However, with regard to INR >4 in the genotyping arm there was a lower proportion of patients, 27% of patients, who had INRs >4, while in the standard-dosing group there were about 37% of patients who had INRs >4. As we know, INRs >4 could lead to adverse events. The time to achieving therapeutic INR in fact was also shorter in the genotyping arm and was 21 days as compared with 29 days in the standard-dosing arm, so the EU-PACT trial was a positive trial favoring genotyping and monitoring these patients.

Now the next trial was the COAG trial, and it was a randomized double-blinded trial and essentially examining genotyping strategy using clinical variables vs just using clinical variables in the other arm. Again, the primary outcome was assessing the percentage of time in therapeutic INR within 4 weeks. So the results were negative in the sense that the clinical arm, 45% of time was spent in therapeutic INR within 4 weeks, while in the genotyping-guided arm, 45% of time in therapeutic INR was present. There were also no differences in bleeding or thromboembolic events, and unlike EU-PACT, there weren't any greater proportion of patients with an INR >4 in one arm compared with the other.

What is interesting about the COAG trial was that approximately one-third of the patients in that trial were African Americans, and the percentage of time in therapeutic INR within the African American subgroup who received dosing based on clinical variables was 43.5%, while in the genotyping arm, the percentage of time in therapeutic INR was 35%, so actually favoring clinical-variable dosing vs genotyping dosing in African Americans; the p value in a subgroup analysis was significant (0.01). Now, it is important to know that 76% of African Americans (compared with 25% of non–African Americans), had absolutely no genetic variation in the two genes assessed for those common genetic variants. It is also important to know that genetic variants to predict dosing variability in African Americans such as *5, *6, *8 and *11 were not tested in the COAG trial.

So why do we have these different results? We have the EU-PACT trial suggesting that genotyping is important and makes a difference, and we have the COAG trial that shows no difference in terms of measuring percentage of time in therapeutic INR with genotyping. There are important differences between the two trials. For example, the control arm in the EU PACT trial used fixed dosing, while the control arm in the COAG trial used clinical variables such as age, sex, African American race, concomitant drug use, etc. This raises the question of whether clinical-variable dosing perhaps is more precise than fixed dosing; however, at 4 weeks, if you look at the percentage of time spent in therapeutic INR it was equivalent either with fixed dosing or clinical-variable dosing. The other issue is that genotyping results were made available within two hours for subjects in the EU-PACT trial; hence, day 1 dosing was determined based on genotype. In the COAG trial, just 45% of subjects had genetic information available to guide dosing [on day 1]; however, by day 3, 99% of subjects did have genetic information to guide dosing. So does the availability of genotyping results alters the ultimate outcome? We are not sure. Now in the COAG trial, the African American patients actually had results opposite to the predicted effects of genotyping in COAG, and this may have made it a negative trial since African American patients made up almost one-third of that population, unlike the EU-PACT trial.

One of the most important considerations, since this is comparing a genotyping strategy, is to understand the prevalence of the genetic variants. You have got to have adequate subjects at risk in the genotyping arm and in the standard-dosing or clinical-variable arm to show a difference. If you take the extreme genotype, those homozygous for VKORC1, the prevalence of those subjects was 11% in the COAG trial but was much higher in the EU-PACT trial—17% of patients in the genotyping arm were homozygous for VKORC1. Also, if you look at genetic variation in CYP2C9, just 1% were homozygous carriers in the COAG trial compared with 3.4% of subjects in EU-PACT trial, and this is important because according to the FDA dosing recommendations for the wild types (those who don't carry these genetic variants), the dosing is typically 5 to 7 mg a day, while for those homozygous for VKORC1 and CYP2C9*2, the dosing is almost 40% lower—3 to 4 mg as the recommended dosing, while in CYP2C9*3 allele, the dosing is even lower and is recommended to be 0.5 to 2 mg of warfarin.

So did the different prevalence of genetic variance make a difference in the primary outcome with COAG? With the negative results from COAG and the positive results from EU-PACT, have we decided that perhaps warfarin pharmacogenetics is not useful and save the cost of doing genetic testing and not bother with it at all? Many medical centers now are moving to providing genotyping information within the electronic medical record, and this is because the cost of genotyping has dramatically decreased, so we may have this genetic-variant information within the medical record. It may be like using a creatinine value to adjust drug dosing. Second, we ask ourselves, would the trial results have been different if there were a greater proportion of subjects with warfarin sensitivity identified by genotype, especially in the COAG trial—therefore, greater at-risk subjects and hence different outcomes? The other question that comes up is, would just regular frequent INR monitoring be adequate without using genetics? INR monitoring was performed at least 6 times in the EU-PACT trial, and the EU-PACT trial showed that genotyping was superior to just clinical dosing with warfarin, and hence that question remains unanswered; however, in the COAG trial, the median INR testing within the 4 weeks was also 6 times, which showed no difference, so this question remains unanswered, but it is important to know that the value of pharmacogenetic testing diminishes with time and more frequent INR monitoring, and pharmacogenetics perhaps plays the greatest role within the first week or two weeks of dose prescription for warfarin.

So the last issue is why bother with warfarin? Why not just prescribe the newer oral anticoagulants like the factor Xa inhibitors, which don't depend on genetic variation with CYP2C9 or VKORC1? Well, there are several reasons we will continue to use warfarin. One is these oral anticoagulants are not recommended and are in fact contraindicated in patients with mechanical valves. Dosing with these newer oral anticoagulants becomes very tricky, especially in patients with renal insufficiency. The concomitant use of P glycoprotein inhibitors complicates the dosing of these novel anticoagulation agents, and we don't have clear guidelines on what to do if you have to interrupt anticoagulation, and reversing these newer oral anticoagulants has become far more complicated; and finally, the costs need to be taken into consideration. So overall, we have two pharmacogenetic trials, which guide us as to whether to use pharmacogenetic testing. I think the jury is still out, especially on the most important question, whether pharmacogenetic testing affects clinical outcomes in terms of thromboembolic events or bleeding events has still not been assessed, and that is why Medicare still doesn't pay for pharmacogenetic testing for warfarin.

Thank you very much for your attention.


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.
Post as: