Medical Device Daily Washington Editor

WASHINGTON – Pharmacogenomic testing for sensitivity to the blood thinner warfarin (Coumadin) looks like a great idea to FDA (Medical Device Daily, Aug. 20, 2007), but that might not be the majority view among members of the American Association for Clinical Chemistry (AACC; Washington) if an informal and utterly unscientific poll conducted on Monday is any indication.

Whether the debate over testing for warfarin sensitivity will serve as a template for other such debates in the future is tough to predict, partly because this is perhaps the most conspicuous such debate on "personalized medicine" to date. Still, pharmacogenomic testing seems certain to generate a lot of controversy in the months and years ahead, especially at AACC annual meetings.

Michael Hallworth, the session moderator and a consultant clinical scientist at the Royal Shrewsbury Hospital (Shrewsbury, UK), said pharmacogenomic testing for warfarin use is, "in many ways, the poster child for large-scale pharmacogenomic testing," but the test has problems.

"It contains all the opportunities and all the challenges" needed to make it both controversial and important, Hallworth said, and "getting it right" will set the stage for other testing regimes in the future. Getting it wrong, on the other hand, will cost a bundle and set the field back substantially.

Hallsworth then conducted an informal poll of the audience, members of which held an orange card overhead if they felt warfarin sensitivity testing is ready for prime time, and a green one if they did not. The crowd seemed fairly evenly divided on the question at the commencement of the session.

Speaking first in favor of the warfarin pharmacogenomic test was Shiew-Mei Huang, PhD, deputy director of FDA's Office of Clinical Pharmacology. "Why warfarin?" she asked. "It has been presenting significant problems for humans," is prescribe 30 million times a year, and "ranks No. 1 in total mentions of deaths for drugs causing adverse events from death certificates," Huang said. She also noted that estimates of the frequency of major bleeding range from 2% to 16%.

Huang described warfarin dosing as a trial-and-error exercise. The approved dose range is between 16 mg and 35 mg, but the drug is sometimes prescribed at 80 mg a day. Controlling the variability in response is the trick, Huang said, with variables such as two genes, CYP2C9 and VKORC1, among the contenders for chief confounders.

Huang cited several studies that indicate that as much as 54% of the variability in a patient's reaction to warfarin can be chalked up to genetic response. She said FDA's position did not include a recommendation regarding specific dosage, but stated that a fair amount of data and algorithms are already in play to guide a clinician's interpretation of the test.

On the other hand, a decision to avoid testing would hamper efforts to establish a clear clinically effective dose of warfarin and "would result in 4,500 to 22,000 serious bleeding events" each year, Huang said. In summary, she said there is sufficient data to warrant the use of the test.

Speaking against the proposition was Charles Eby, MD, associate professor of pathology and immunology at the School of Medicine at Washington University (St. Louis). "The key issue here is whether this should be the standard of care," Eby opened, and making this a standard of care means that any physician writing a prescription for warfarin risks a lawsuit if he or she does not order a test.

"Warfarin is essential and prevalent, and is a very difficult drug," he said, describing the drug's safety as "clearly sup-optimal." Still, Eby said "the big elephant in the room is, 'Can pharmacogenomic testing be used appropriately?'" He said it cannot.

Eby asserted that warfarin, which works by cutting down on vitamin K production, is metabolized with sufficient variation that the effective dose can vary by as much as a factor of 30. He said the current crop of approved platforms return somewhat different numbers in real-world applications as compared with study data. In reference to the 54% predictive power cited by Huang, Eby said that "instead of being excited about being able to predict half the variability," he asked, "Where is the other half?"

He found fault with the existing algorithms which, when "developed in a limited population, do not necessarily transport" to larger populations, stating further that the data indicate that variability for race and ethnicity are "not accounted for adequately.

"I consider this a research tool, not a clinically accepted one," Eby said.

He made reference to the International Warfarin Pharmacogenomics Consortium, a joint effort between the National Institutes of Health and Japan's Center for Genomic Medicine, which is among several entities working to refine algorithms that would more clearly demonstrate an individual's sensitivity to warfarin, but he asserted that the work is not a finished product. "We look forward to seeing their work," Eby said.

Another speaker in favor of the idea was Mark Linder, PhD, associate professor of pathology at the University of Louisville's School of Medicine and a senior VP at pharmacogenomics specialist PGxL Labs (both Louisville, Kentucky). He said the data presented by Eby was "relatively antiquated" and that the mechanisms for the two most-often cited genes in relation to warfarin, VKORC1 and CYP2C9, are "very well-defined, and I have never seen papers that would add controversy to these positions."

Linder said the association of these two genes with warfarin sensitivity "has a sound fundamental principal." He said a report published in 2006 by the American Enterprise Institute/Brookings Joint Center (Washington) indicated that testing could prevent 85,000 adverse events a year at a savings of $1.2 billion. "We don't see that there is actually any controversy here," Linder said after citing other sources that backed the AEI/Brookings paper.

"These are mechanism-based causal effects," he asserted, and "technology for testing is available and the benefits have been prospectively studied." Linder said "the disclosure of risk is an obligation," but "ultimately the decision rests with the physician and patients."

The last speaker, an opponent of the idea, was Amir Jaffer, MD, associate professor of medicine at the University of Miami's Leonard Miller School of Medicine (Miami). Jaffer said that while "we all agree that warfarin is a tricky drug – it's important to recognize other factors that influence the dose," including the presence of other drugs.

"We need evidence from well-powered outcomes-based studies," Jaffer said, remarking that the test "may be easy to order," but using the test results "is another matter. "I think we are on our way, but we're not there yet."

One of Jaffer's main points was that the AEI/Brookings paper "assumed a bleeding risk of approximately 17.5%. That's huge," he said, adding that it also assumes that 5% of all transient ischemic strokes would be prevented. "There is no data thus far that supports" these assertions, he said.

In response to Jaffer's suggestion that a full-blown randomized trial is needed to assess the impact of genomic testing on the prevalence of bleeding events, one member of the audience stated that the genetic polymorphisms in question are sufficiently rare that the task of recruiting and enrolling subjects into a study large enough to statistically validate or invalidate the idea would be "enormous," and implicitly impractical.

To assess the effect of the discussions at the end of the session, Hallworth asked for a show of cards again, and he declared the attendees, who probably numbered nearly a thousand, voted a slight majority against the idea that pharmacogenomic testing for warfarin is ready for prime time. When Hallworth asked those who had changed their minds during the course of the debate to raise either card, only about 10 members of the audience indicated that the discussion had done so.

Oddly enough, it turns out that payers are already covering the test, although labs have had to go to some lengths to obtain coverage. Linder told Medical Device Daily that it is "very rare that we don't have a test paid for by private insurance, and we have never been rejected by Medicare." Linder said Medicare is paying $326 for the test.