Journal of the National Cancer Institute Advance Access originally published online on March 11, 2008
JNCI Journal of the National Cancer Institute 2008 100(6):382-386; doi:10.1093/jnci/djn071
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© Oxford University Press 2008.
NEWS |
Personal Genetic Tests Facing Scrutiny
Seven years ago, the decoding of the human genome was followed by a publicity blitz that triggered high expectations for personalized medicine. Now companies have begun to cash in on those expectations by offering genetic tests for everything from cancer risk to family ancestry—even athletic prowess. A recent survey by Stuart Hogarth, a research epidemiologist with the University of Cambridge in the United Kingdom, finds that 27 companies—most of them based in the United States—are selling genetic tests directly to consumers through the Internet for costs ranging from hundreds to thousands of dollars.
But the rush to market genetic tests has been met with skepticism from within the scientific community. Experts increasingly warn that the science behind these tests may not be ready for prime time: The tests’ appearance now poses worrisome consequences, not just for consumer safety but also for the public's trust in personalized medicine. "Consumers may be getting test results that are at best noninformative and at worst misleading or false," said Gail H. Javitt, J.D., director of the law and policy center at Johns Hopkins University. "The risk is that they might make ill-informed health care choices that, over time, could implode the dream of using personalized medicine to improve patient care."
Adapting to Genomic Realities
A key issue is whether the tests offer less predictive power than the public may expect. Earlier beliefs held that single-gene mutations drove most disease risk. But scientists now know that most diseases arise from interactions involving environmental factors and many variations in the genome, including both genes and noncoding regions of DNA. Most tests on the market today base their predictions on a few genes—or even just one—and therefore may miss important parts of the story.
Take diabetes, for example. The Icelandic company deCODE genetics sells a $300 clinical diagnostic test for type 2 diabetes that is based on one gene variant, known as TCF7L2. This is the most powerful risk gene known for type 2 diabetes. However, it changes the odds of getting the disease from a baseline of 7% in the general population to 14%. At least 10 other variants have also been linked to type 2 diabetes, and none of them are screened by deCODE's test. Studies by David Melzer, M.D., Ph.D., a professor of epidemiology at the Peninsula Medical School in the United Kingdom, showed that 80% of a population who tested positive for TCF7L2 never got the disease in old age, while 40% of those who had the disease tested negative for the variant.
Consumers who test positive for TCF7L2 might adopt a healthier lifestyle. That would be a beneficial outcome, but there is no proof that it would happen. A negative result, on the other hand, could foster a sense of mistaken immunity from type 2 diabetes. And that could invite high-fat diets, poor exercise habits, and other risky behaviors (again, there is no proof that such a result would cause this outcome). Meanwhile, doctors are often ill-equipped to guide patients for genetic testing, said Molly Brewer, M.D., an associate professor of gynecologic oncology at the Carole and Ray Neag Comprehensive Cancer Center in Farmington, Conn. "When it comes to genetic risk, it seems the more we know, the less we know," she said. "We find that maybe a third of the patients who actually are at high risk for a given disease show up as positive on the state-of-the-art tests for cancer predisposition, so this is all much more complex than most companies will concede."
Quality Concerns Grow
Concerns over direct-to-consumer genetic tests have grown since a federal General Accounting Office (GAO) investigation in 2006. The GAO submitted cheek swabs from its own staff to four Web-based companies and found that each one returned predictions that were "medically unproven" and "so ambiguous they do not provide meaningful information to consumers." The GAO's conclusion was that the field's growth was outpacing its regulatory oversight.
That view prompted the U.S. Department of Health and Human Services (HHS) to launch a full-scale review of how genetic testing services are regulated in the United States. The HHS convened an expert panel, the Secretary's Advisory Committee on Genetics, Health, and Society, which issued a draft report in November 2007. The final report is expected in April.
The advisory committee's investigation yielded several key findings. Like the GAO, the committee determined that regulatory oversight of genetic testing is fragmented, poorly coordinated, and beset with holes that could "lead to harms." The committee also stressed the need for better interagency coordination, with an eye toward closing regulatory gaps.
Inadequate Regulations?
Under the current scheme, genetic tests are split in two categories, each regulated by different federal agencies. So-called in vitro diagnostics (IVDs) are regulated by the U.S. Food and Drug Administration. According to Steve Gutman, M.D., who directs the FDA's office of in vitro diagnostics, IVDs include test kits or medical devices that can be purchased by the public and sent to different laboratories. IVDs must go through a premarket FDA review.
The other category includes laboratory-developed tests (LDTs), which are regulated by the Centers for Medicare and Medicaid Services through the Clinical Laboratory Improvement Act (CLIA). A key difference between the types of tests is that whereas IVDs get shipped to customers, LDTs remain with their creators. Because they’re considered medical services and not devices, LDTs don’t have to undergo premarket review, which makes them much easier to commercialize.
Numbering roughly 1,400, almost all direct-to-consumer genetic tests on the market today are LDTs. CLIA has one overarching mandate in this area: to ensure that genetic tests achieve analytical validity, or accurate measurement results (i.e., the ability to correctly identify a gene). However, CLIA is silent with respect to clinical validity, or the tests’ ability to serve purported medical aims. For clinical validity, the FDA—charged with assessing a test's effectiveness—is the responsible party. But the agency has a policy of enforcement discretion.
This policy provides a potential loophole in the oversight system. According to Gutman, the FDA invoked the policy because it doesn’t want to impede technology development in such a rapidly moving field. But Javitt worries that the loophole could allow bogus genetic tests to reach the public just because they’ve been cleared for analytical validity under CLIA. The question is how to balance those concerns. "In some cases, we’d want to see premarket evidence that supports clinical validity," Javitt said. "But in other cases, premarket reviews wouldn’t be necessary. We don’t want to reinvent the wheel."
The genetic testing industry, meanwhile, insists that the current system is self-correcting. "The bad apples tend to get shut down quickly," said Trish Brown, vice president for clinical affairs with DNAdirect, a Web-based provider of genetic testing services. "My concern is that with more regulation, costs for genetic testing will rise, so patients will ultimately end up paying more for services that are already expensive on the basis of the technology alone."
Dealing With Clinical Validity
Underscoring the issue is the fact that no one is sure how to regulate clinical validity in the first place. Sources with the HHS advisory committee say new data emerge so fast that a test might be obsolete by the time a premarket review is completed. "We’re calling for more data, more research, and more postmarket surveillance," one source said.
Still, the FDA has signaled its intent to regulate one type of LDT, known as the in vitro diagnostic multivariate index assay (IVD-MIA). These particular tests quantify gene-specific risks for a given disease by using data from the scientific literature. Algorithms are then used to calculate a consumer's overall risk, depending on his or her gene expression profile. These algorithms are proprietary, however, which puts them beyond scientific scrutiny. And that makes regulators nervous. "If done right, IVD-MIAs could be terrific for personalized medicine, but if done poorly, they can report false biological information that could actually harm patients," FDA's Gutman said.
So far, just one IVD-MIA has undergone the FDA's premarket review: a diagnostic test for breast cancer recurrence, marketed by the Dutch company Agendia, which voluntarily cleared the process in 2007. The FDA issued draft guidelines for IVD-MIA premarket review last July. Gutman can’t predict when the guidelines will be finalized. But he's careful to note that the agency won’t impose premarket review for LDTs as a whole. "This is not a signal that FDA intends to regulate all LDTs," he said. "The guidance covers a small niche of tests that are subject to error in design, and error in performance, and that can lead to high-risk decision making in patient care."
In recent months, media reports suggesting that genetic tests have serious problems have appeared throughout the world. Responding to growing concerns, the Human Genetics Commission, a government advisory body in the United Kingdom, has called for more premarket evaluation of genetic tests and an end to consumer advertising for tests "with significant health implications."
However, an official with the Centers for Medicare and Medicaid Services stressed that regulating clinical validity is unnecessary because clinicians neither administer nor interpret genetic tests in a vacuum. "It's not like we’re in some total dead end because CLIA is limited to analytic validity," the official said. "Clinicians can rely on their own judgment and supplemental knowledge from colleagues and from the literature. You have to incorporate genetic tests together with other clinical symptoms and patient history before you can make treatment decisions."
Where the issue goes from here remains to be seen. The genetic testing industry has some growing pains to contend with. The regulatory framework that watches over it, meanwhile, must adapt to new realities for which it may be poorly suited, Javitt said. And sustained by the promise of a genomic transformation in medicine, the public is hungry for tangible products, for which it is apparently willing to pay. Unfortunately, what they’re getting for their money still isn't clear.
Prostate Cancer Test Combines Gene Variants To Predict Risk
One of the more promising upcoming gene tests will assess prostate cancer risk. Under development at Proactive Genomics, a Web-based company, the test's predictions are built on five gene variants that collectively are associated with an increase in prostate cancer risk of more than 400%. If all five variants are combined with a family history of the disease, then the risk of prostate cancer goes up ninefold compared with those with none of these risk factors, said Jianfeng Xu, M.D., Dr.P.H., a professor at Wake Forest University in Winston-Salem, N.C. A cofounder of Proactive Genomics, Xu led the team that produced these findings, which were described in the January 16 issue of the New England Journal of Medicine.
The test is compelling in part because it combines risk genes that individually have small effects, said Edward Gelmann, M.D., of the Herbert Irving Comprehensive Cancer Center at New York's Columbia University and author of an accompanying editorial about the Wake Forest findings. "The study shows that, together, the genes are more predictive than they are alone," he said. Indeed, most tests on the market base predictions on just one or a few genes, whose purported links with disease are often minor and poorly validated. The association of the five genes with prostate cancer was confirmed in two independent study populations of 4,065 case patients and 2,938 control subjects combined, Xu said. According to the results of the validation study, which has recently been submitted for publication, men with all five risk genes and a family history have an 11-fold greater risk of prostate cancer.
Identified in blood samples from 4,600 Swedish men, including 2,900 prostate cancer patients, the variants reside on three DNA regions: 8q24, 17q12, and 17q24.3. Four of the five variants are located in noncoding regions of the genome, which means that they don’t make proteins. Scientists used to call these regions "junk DNA." More recently, evidence suggests that noncoding regions are functional, although their role in the cell remains largely a mystery. Xu conceded that he doesn’t know how the variants contribute to prostate cancer.
William Catalona, M.D., a professor of urology at Northwestern University in Chicago, said that the findings challenge the standard view of how prostate cancer risk is inherited. Previous thinking held that risk genes for the disease are passed from father to son. However, the genes described by the Wake Forest researchers can also be inherited from the mother, he explained. "This shows the risk can come down through both parents," Catalona said. "What's more, one of the variants on 8q24 is also associated with colon cancer. The implication is that this region may be associated with cancer in a broader sense."
It's unclear who will benefit most from the test. Positive findings in just one or two of the five genes wouldn’t point to any clear recommendations, Xu said. "The test is most useful when it generates findings at one extreme or another, meaning no positive hits, or positive hits for all the genes," he explained. "For those in the middle, there's more uncertainty." Xu's investigations show that roughly 10% of men have none of the risk variants; 88% have between one and three variants; and roughly 2% have four variants or more.
Gelmann, however, questions the test's relevance for public health. "The issue with prostate cancer is that we haven’t yet proven that screening decreases mortality," he said. "Until that's been demonstrated, it's not clear how much a test like this can affect long-term outcomes."
Meanwhile, Xu predicts that the test will be on the market within a year. However, no one knows whether it will be made available directly to consumers or only to clinicians, he said.
—Charlie Schmidt
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||