A controversial pharmacogenomic study that some say disproves the value of gene-based warfarin dosing may have important implications for how future clinical trials studying the subject are designed, according to the evaluation of a prominent researcher and a high-ranking US Food and Drug Administration official.
The Couma-Gen study — conducted by researchers at the University of Utah School of Medicine and LDS Hospital, Intermountain Healthcare — enrolled more than 200 patients initiated on warfarin, randomized to either PGx-guided dosing or standard dosing protocols using only clinical factors.
Although the study did not reach its primary endpoint of a reduction in bleeding outcomes, the researchers concluded that “an algorithm guided by pharmacogenetic and clinical factors improved the accuracy and efficiency of warfarin dose initiation.”
Furthermore, subset analyses indicated that “pharmacogenetic guidance showed promise for wildtype and multiple-variant genotypes.”
Shortly after the Couma-Gen study was published in the Nov. 7 issue of Circulation, Lawrence Lesko, director of FDA’s Office of Clinical Pharmacology, rebuked participants at a healthcare conference for focusing on the negative aspects of the study and overlooking the positive findings, particularly the discovery that PGx-guided dosing was more accurate than standard dosing methods [see PGx Reporter 12-07-2007].
“There was actually very positive news coming out in the article which people didn’t really focus on,” Lesko said at the time. “There is a sort of bias in the way we look at personalized medicine to kind of look for reasons why not, instead of the reasons why.”
Now, in an article published in the latest issue of Personalized Medicine, Lesko is attempting to more objectively evaluate the significance of the Couma-Gen study by analyzing its strengths and weaknesses.
Lesko authored the article, entitled “Couma-Gen: implications for future randomized trials of pharmacogenetic-based warfarin therapy,” with Brian Gage, an associate professor of medicine at Washington University.
Gage is the creator of www.warfarindosing.org, a site designed to help doctors apply an algorithm, using clinical factors and patients’ genotypes, to estimate an appropriate warfarin dose for a given patient. The free website is mentioned in a new brochure issued by the Critical Path Institute and the American Medical Association aimed at educating physicians about gene-based warfarin dosing (see related article, this issue).
“The results of [the Couma-Gen study] have been reported to deny the value of genotyping, but [in fact] they were intriguing,” Lesko and Gage wrote. “The trial has important implications … for the design, randomization, blinding, and end-point definition of future studies.”
Couma-Gen failed to show a significant mean difference in bleeding outcomes between patients who were given warfarin based on PGx-guided dosing and patients whose clinical factors determined their dose. According to the authors, Couma-Gen’s failure was due to the study’s over-ambition.
“The negative primary end-point could lead to the conclusion that pharmacogenetic dosing is no better than clinical dosing,” Lesko and Gage wrote. However, the randomized-controlled trial was powered “to detect an ambitious difference in out-of-range [international normalized ratios] between pharmacogenetic (20 percent) and clinical arms (40 percent).
“Thus, future studies should be powered to detect smaller differences,” the authors suggested, adding that the out-of-range INRs were significantly reduced for an “important subgroup”: patients who were either wildtype or had multiple variant genotypes.
Critical Path Institute President Ray Woosley noted that Couma-Gen was a rigorously conducted study, in which “everything was pointing in the right direction.
“There’s not single study that’s going to tell you all you need to know about warfarin and how to use genetic testing,” he told Pharmacogenomics Reporter. Anderson et al. “was positive in many, many ways. It didn’t reach their pre-conceived endpoint for several reasons, one of them being that [University of Utah] has a very efficient [dosing] system and they were trying to make it better. They had very few out-of-range INRs to start with because they have a very aggressive warfarin clinic.”
Lesko and Gage pointed out in their article that Couma-Gen was conducted in inpatients with aggressive starting doses and dosing adjustments.
Couma-Gen used a dosing algorithm based on research by Kovacs et al. that uses a much higher initial warfarin dose (10 mg) than the FDA recommends (between 2 mg and 5 mg). The researchers developed an algorithm for the PGx-dosing arm that was based on a regression equation analogous to Kovacs et al. — beginning warfarin at twice the typical daily dose — based on clinical and genetic factors.
“The results of [the Couma-Gen study] have been reported to deny the value of genotyping, but they were intriguing.”
Another aspect in the study that may have impacted its outcome was that patients in both the clinical and pharmacogenetic arms received regular INR monitoring. “It is not uncommon for inpatients to receive daily INR monitoring,” Lesko and Gage noted in their analysis. “Thus, the quality of INR monitoring in the clinical arm in Couma-Gen would be better than typical outpatient settings.”
There was also a potential for bias in the study, since Couma-Gen was single-blinded. The research investigators and patients were blinded; the pharmacist was not.
“Empirical evidence suggests that lack of blinding my inflate effect estimates,” Lesko and Gage state. While ascertainment bias was minimized by blinding patients and investigators, performance bias may have crept in since “good compliance with an intensive monitoring protocol … could have allowed for earlier identification of warfarin-sensitive patients in the clinical arm.”
Lesko and Gage concluded that “the putative benefit of pharmacogenetic dosing may have been diluted by: a reliable INR-monitoring schedule in the clinical-based arm; inclusion of a larger percentage of patients with genotypes who did not benefit from pharmacogenetic dosing in the clinical-based arm; and the use of an aggressive initial dose in the clinical-based arm.”
Impact on Future Studies
According to Lesko and Gage, future PGx dosing studies for warfarin should be randomized, multi-centered, double-blinded, and powered to show a clinically significant improvement in outcomes with the use of genetic tests. Studies should also include a third arm looking at, for instance, patients whose warfarin dose is based on an algorithm developed from initial INR responses.
The authors suggest that researchers should enrich the study population before randomizing patients who have multiple gene variants of CYP2C9 and VKORC1 since they are at greater risk for bleeding and, thus, represent a “sensitive” population for demonstrating the clinical benefit of genetic testing.
There are several ongoing studies that can benefit from Anderson et al.’s experience with Couma-Gen. For instance, the National Heart, Lung, and Blood Institute this year plans to start a randomized controlled study for warfarin.
Also, the FDA is working with the principal investigators of the Harvard Creating an Optimal Warfarin Nomogram Trial to develop a PGx-based dosing algorithm for the drug. The FDA and CROWN researchers are currently using eight clinical endpoints to assess their algorithm’s performance compared to other algorithms.
Finally, 21 PGx research centers are working with PharmGKB to pool data from 6,000 patients on warfarin dosing using clinical and genetic factors. With this information, the Warfarin Pharmacogenomics Consortium intends to develop a PGx model to gauge the therapeutic dose for the drug.
World Without Warfarin?
Indeed, there may come a point when the warfarin-dosing bugaboo may be avoided altogether by using a new, less-variable anticoagulant.
With new anticoagulants in development, a study powered to quantify changes in adverse events “may not be feasible,” the authors noted.
Sanofi-Aventis markets the anticoagulant Lovenox, also called enoxaparin, which the FDA approved last May for the treatment of prophylaxis and deep-vein thrombosis, prophylaxis of post-op DVT, and unstable angina or non-Q-wave myocardial infarction.
Bayer currently has in Phase III studies the anticoagulant rivaroxaban. Bayer plans to seek FDA approval for the drug as a treatment for stroke prevention in atrial fibrillation and for long-term secondary prevention of venous thromboembolism.The company is also studying the drug’s effectiveness in the management of acute coronary syndrome.
Researchers are hoping rivaroxaban and enoxaparin can eventually replace warfarin since there have not been any reports of significant bleeding events in clinical trials. Studies have shown that bleeding rates with rivaroxaban and enoxaparin have been similar.
Last year, Bayer applied for approval of the anticoagulant in the European Union under the brand name Xarelto.