"There is a very strong need for clinical validation," says Brian Spear, director of pharmacogenetics at Abbott Labs. Many SNPs have been linked to diseases, but few studies have demonstrated that genotyping makes a measurable difference in treatment outcome. The first examples of pharmacogenomics — TPMT and Herceptin — all involve cancer. It is easier to justify an expensive treatment that helps only a few patients if the disease (or side effects) are life-threatening.

But there are other attractive targets, such as the anticoagulant Warfarin, which Spear cites as "a good example because there is strong evidence that genetics can be used to predict the initial dose and later dose adjustment. It's also been well demonstrated CYP2D6 variations cause poor metabolism of tricyclic antidepressants." The dosage of Warfarin is critical; dosed improperly, patients can have serious complications. Tricyclic antidepressants, meanwhile, can cause cardiovascular problems in poor metabolizers.

At least two groups have tried to tackle the issue of outcomes in exactly those cases, although funding is hard to come by. Mark Linder at the University of Louisville in Kentucky has proposed a study of the influence of genotyping on the outcomes of patients taking Warfarin. Peter Wedlund at the University of Kentucky's College of Pharmacy and Eastern State Hospital Mental Health Research Center wants to determine if genotyping can improve outcomes for patients who are candidates for tricyclic antidepressants.

Both researchers have had grant applications to the National Institutes of Health (NIH) turned down. "The study section officers ... encouraged me and said Warfarin is a problem and this is a high priority," Linder says.

Wedlund and Linder acknowledge there could be other problems with their proposals, but they are concerned that the NIH may have set up a catch-22 for researchers seeking funding for outcomes studies. "Some of the reviewers said, 'This is old hat, we don't need to prove these associations,'" says Linder. "But you can't do a proper outcomes study unless the association is proven." Wedlund received similar criticisms. "One reviewer said, 'This is so obvious,' and that's true of course," he says. "We've known for 20 years that the CYP2D6 variations influence drug response. The question is, would a test influence outcomes in a clinical setting?"

"NIH funds good research," says Rochelle M. Long, chief of the pharmacological and physiological sciences branch of the NIH and head of the National Institute of General Medical Science's Pharmacogenetics Research Network (found online at www.nigms.nih.gov/pharmacogenetics), which funds numerous studies in this field. She attributes the researchers' complaints to "sour grapes," and insists that "anyone who submits an application to the NIH must defend the approach they are taking." The NIH is interested in outcomes studies, she says, but "our primary goals were to get the molecular commonalities nailed down." She is unaware of any outcomes studies being funded by the NIH.

The underlying question is whether there is a source for funding for these studies, particularly since drug manufacturers are not interested in them. Wedlund is resubmitting his grant to the Agency for Health Research and Quality and is pursuing other sources as well. But he's worried that the NIH is sending conflicting messages to those seeking to do outcomes research. "This has nothing to do with sour grapes," he insists. "It has to do with making sure there are appropriate avenues for advancing genomics into the clinical arena."

No matter who ends up doing these studies or when they get done, the impact will be enormous. As Wedlund says, "Everybody has been promoting the concept of genomics as a therapeutic tool, but if you could demonstrate that a genetic variation is influencing therapeutic costs, then you would have every HMO in the country saying, 'We have to start using this.'"

—Malorye Branca 

Back to The New, New Pharmacogenomics