October 2008: Volume 34, Number 10
2008 Annual Meeting Highlights
Experts Debate Pharmacogenetic Testing for Warfarin Sensitivity
Tests Could Enable Better Dosing, But Some Question the Data
By John R. Bell
In this year of presidential debates, annual meeting attendees got a chance to participate in a hotly contested clinical issue during the symposium, “Warfarin Pharmacogenetic Testing: Now Ready for Prime Time?” Experts took sides and debated whether pharmacogenetic testing should be standard practice in prescribing warfarin (Coumadin). In August 2007, FDA revised the labeling for warfarin to make clinicians aware that patients with certain variants of the CYP2C9 and VKORC1 genes probably need lower initial doses of the drug. But not everyone is convinced that such tests are necessary.
Shiew-Mei Huang, PhD, deputy director of the Office of Clinical Pharmacology in FDA’s CDER, presented the viewpoint that the benefit of such testing is now so clear that it should be a routine step in administering the drug. Warfarin is prescribed more than 30 million times per year, with about 2–3 million new patients per year, Huang said. Among drugs mentioned in death certificates as causes of death-related adverse events, warfarin ranks first, she noted. It ranks second among drugs associated with ED visits and is among the top for visits related to bleeding. In the current climate of heightened patient safety, FDA has been struggling with serious adverse reactions and is looking to the medical community to embrace new ways to improve outcomes.
“It’s a very difficult drug to take,” Huang said. She noted that now that FDA has cleared assays for response to the drug, the barriers to testing are becoming less problematic. To date, more than 12 published studies have shown a benefit for genetic testing for Warfarin sensitivity, she said.
Symposium attendees cast their votes regarding warfarin genetic testing. After presentations by Huang and Eby, as well as Mark Linder, PhD, a supporter of testing and Amir Jaffer, MD, a critic of testing, a slight majority thought that testing was “not ready for prime time.”
“Up until now, finding doses to maintain therapeutic anticoagulation levels of warfarin has been largely trial and error,” Huang said. The European Atrial Fibrillation Trial showed a steady-state distribution of the stable dose ranged from 5–10 mg to 75–80 mg, she noted. And 16% of patients did not need dose adjustment. But more than 50% needed a lower dose than 5 mg.
The FDA-approved dose is 2mg-5mg, and patients are usually given a 5-mg dose to start. Dose adjustment based on international normalized ratio (INR), shows that when INR is less than 2, there is increased risk of stroke. But an INR of more than 3 brings an increased risk of bleeding. That risk goes up approximately 26% when INR is greater than 4 and up to 60% when INR is more than 5, Huang explained. But INR is difficult to maintain, she cautioned. And for the 70–75% of patients not being treated in anticoagulation centers, INR is not feasible due to costs and complexity.
Moreover, many studies show that age, sex, race, body weight, and other clinical factors explain only 15% of variability in warfarin absorption, Huang said. But if CYP2C9 and VKORC1 status are also considered, more than 50% of variability is explained. Although this leaves 45% of variability unexplained, this amounts to better control than is available for other drugs, she noted.
Huang acknowledged that the American College of Chest Physicians published guidelines in June on the use of vitamin K agonists such as warfarin in which the organization recommended against pharmacogenetic testing until randomized controlled trials show a clear benefit (Chest 2008;133:60-198). However, she pointed to several papers finding a link between genetic status for these two markers and warfarin sensitivity—one of which was coauthored by one of her opponent in the debate, Charles S. Eby, MD, associate professor of pathology and immunology at the Washington University School of Medicine in St. Louis (Clin Pharmacol Ther. 2008 Sep;84:430). That study of more than 1,015 patients used a dosing algorithm prospectively to determine that 17%–22% of variability in warfarin absorption could be explained by clinical factors, but 54% of variability could be explained if genetics were also considered. Other, forthcoming studies that are part of the International Warfarin Pharmacogenetics Consortium also show a benefit for pharmacogenetic testing, she said.
Eby presented the viewpoint that the evidence in favor of genetic testing for warfarin dosing is not sufficient to make testing the standard of care. In his view, the question is not whether testing can be of benefit, but whether the published data are sufficient to indicate that a physician who does not use such testing in administering warfarin is guilty of malpractice if a patient experiences a bleeding complication.
“The big elephant in the room is, can pharmacogenetic dosing benefit patients? My answer is no. The data is insufficient at this time to say that patients will benefit in a way that is cost-effective.”
He acknowledged that the therapeutic window for warfarin dosing is narrow, that doses vary 30-fold, and that there is a plausible biological mechanism for warfarin sensitivity based on certain CYP2C9 and VKORC1 SNPs or genetic variants. Method validation of the genetic tests approved by the FDA so far, from Nanosphere, AutoGenomics, Osmetech, and ParagonDX, has shown concordance of 99%-100% across all platforms, he said. However, he noted that although a recent proficiency survey of pathologists showed good predictive value of approximately 96%-98% for the CYP2C9 genotype, the value for VKOR SNPs was less, at roughly 83%-85%. This is an important discrepancy, Eby said, given that the common assumption is that genetic tests are 100% accurate. “Maybe [a genetic test is] not going to be 100% accurate. And we must consider this when we consider making this a standard of care.”
In addition, the fact that clinical variables such as age, race, smoking status, concomitant drugs, and weight account for such a great degree of variability in warfarin response is problematic, he said. “Where is the other 50% of the variability?” Furthermore, published algorithms that predict response in some populations, such as African Americans, do not translate to other populations, he concluded, making warfarin pharmacogenetic testing not ready for prime time.
An informal audience vote taken at the end of the debate showed a narrow victory for Eby’s side of the argument, but laboratorians can no doubt expect the debate to continue.
Eby disclosed relationships with Luminex, AutoGenomics, Third Wave, Osmetech, and Roche Diagnostics.