THE BLOG

How FDA and 23andMe Dance Around Evidence That Is Not There

01/27/2014 04:25 pm ET | Updated Mar 29, 2014
  • Cecile Janssens Professor of translational epidemiology at Emory University

On July 20, 2010, Andrew Alexander, founder and director of easyDNA, received a letter from the US Food and Drug Administration. The FDA had noticed that easyDNA was marketing the "Genetic Predisposition Health Test, a home-use device that is intended to allow individuals to discover whether they are genetically predisposed towards developing a number of diseases and medical conditions, including cardiovascular conditions, different types of cancers, disorders of the immune system, diabetes as well as medical conditions related to ageing," a test that the FDA had not approved.

The FDA wanted to protect the public from tests that had no scientific basis, and which were unlikely to accurately predict disease risk. And it had a good point: most common diseases are caused by such a complex interaction between numerous genetic and lifestyle risk factors that tests based on a few genes cannot meaningfully predict diseases.

EasyDNA was among the 17 companies that received a letter from the FDA during that summer. The FDA argued that their personal genome services are medical devices that require review and approval before they can be offered on the market. The companies were invited to discuss with the FDA whether their product indeed is a medical device and what evidence they need to provide to obtain approval.

Less than four years later, and without much discussion in public, almost all of the 17 companies have stopped selling personal genome tests: some have completely exited the business, others are selling other tests or have matured into successful biotech companies. But two of them are still in business.

EasyDNA is still offering the Genetic Predisposition Test. Its website now notes that "due to the nature of this testing, the FDA prohibits the in-home collection of your sample as defined in 201 (h) of the Federal Food, Drug and Cosmetic Act. You can therefore choose to have your sample collected by your physician or a professional collection agency." The company has found a way around the developing regulation of this new market - for its totally outdated test, the same it had on the market in 2009.

And then there is 23andMe. On December 6, 2013, "the leading health and ancestry DNA service" rebranded itself as "the largest DNA ancestry service in the world," a change it made after the FDA urged it to halt the marketing of its personal genome service until the company provided evidence that its tests actually work.

The FDA once again had good reason to question 23andMe's claims: there is still no scientific evidence to support them, and the tests are unlikely to accurately predict disease risk. Yet, within a few days, the FDA's request was reported all over the web. This time, surprisingly, 23andMe was not blamed for offering a bogus test - the FDA was blamed for preventing people to access their DNA. Is the FDA taking its job too seriously and focusing on bureaucratic details? To look at people's responses, you would think so. But all of us - except perhaps those who stand to profit from 23andMe - should be thankful for the FDA's insistence that the company provide scientific evidence to support its claims.

As expected, 23andMe did not stop the marketing - it is a for-profit business, after all. The company used to offer DNA analyses from which it predicted risks of multiple diseases, informed customers about carrier status of mutations for genetic disorders, and provided ancestry information. At the FDA's urging, it removed all health-related results from the consumer reports and its website, and continues offering the same laboratory testing for the analysis of ancestry information only--for the same USD99 price tag.

The company promises new customers that "in the future, [they] may be able to receive health-related results, dependent upon FDA marketing authorization." But how likely is it to receive that authorization? Not terribly, it seems.

23andMe has to demonstrate that their test predicts all of the 120 common diseases and 25 tests for drug response about which it makes claims - for customers of all ages and ethnicities. And the FDA is right, there are no scientific studies that have investigated the predictive ability of 23andMe's testing service even for one disease in one subpopulation. The company still has to do all the research - but it does not have the right data to do it.

In the absence of the right data, my students and I sought to investigate the predictive ability of the personal genome tests using simulated data. We showed that 23andMe, deCODEme and Navigenics (by the time we published our findings, the latter two were already out of business) predicted some diseases relatively accurately but others not at all. Moreover, we found that the risk estimates from the three companies differed enormously for individual customers: their risk of disease was largely determined by which company had done the test. In other words, the test results cannot be relied upon, there are too many issues and uncertainties about the reliability and the predictive ability.

The stakes here are high, and mistakes will be costly for customers, in more ways than one. When the risk estimates of, say, cancer risk are followed by recommendations to change health behavior or referral to the doctor, it is essential that these estimates better be fairly correct and up to date so as not to mislead customers. And that is why the FDA wants to know how predictive and accurate the risk estimates are. Its request is very reasonable, because the risk calculations of the company rely heavily on unproven assumptions.

It is unfortunate that the FDA formally has to ask for evidence while it knows this is not available and while there is so much indirect evidence that genetic tests for common diseases have such limited predictive ability. The directive gives the impression to the public that the FDA is nitpicking on a formality to halt the commercial offer of DNA tests, while it is protecting the public from misleading health information.

When individuals 'at increased risk' of disease are recommended to visit their doctor, many will go unnecessarily. Do we want to pay for the health care costs of people who have no symptoms, who do not need to see a doctor, and who may not even benefit from it? Better not. If people want to have the right to buy tests of unproven value, they should also take the financial consequences of follow-up doctor visits and examinations. Solidarity has its limits.