Beyond 23andMe’s Shutdown: The Role of the FDA in the Future of Direct-to-Consumer Genetic Testing
Kyle Edwards, Uehiro Centre for Practical Ethics and The Ethox Centre, University of Oxford
Caroline Huang, The Ethox Centre, University of Oxford
On November 22, in a harshly worded “warning letter,” the US Food and Drug Administration (FDA) informed the direct-to-consumer genetic testing (DTC-GT) company 23andMe that it had 15 working days to discontinue marketing of its services. By December 5, 23andMe had canceled television, radio, and online advertising and stopped selling its $99 ‘spit kit’ DNA test online.
To put it mildly, the FDA and 23andMe have some communication issues to resolve. A working relationship dating back to 2008 appears to have soured after a six-month period of silence from 23andMe, prompting the warning letter and causing many observers to comment on the apparent stupidity and mystifying nature of 23andMe’s communication ”strategy.” While the FDA’s letter is quite clear that 23andMe must communicate better, particularly in reporting the accuracy of its tests, it is not at all clear on how the FDA plans to regulate companies like 23andMe after these accuracy results are in. Moreover, it hints strongly that some tests may be banned even if they are as accurate as the tests you could receive through a physician.
Assuming 23andMe follows through on its promise to cooperate with the FDA, how exactly should these DTC-GT services be regulated to best serve the public?
The Nature of 23andMe’s Genetic Testing
Before tackling this question, let’s be clear about the nature of these tests. Using a non-invasive saliva sample, 23andMe offers its customers a vast range of genetic information, including information about ancestry, traits, drug response, and risk predictions for diseases and conditions. Although 23andMe technically states that the test results should not be used as a diagnosis, and that any decisions about health and health-related behavior should be made in consultation with a physician, the FDA notes that in practice many of the tests seem to offer both a diagnosis and treatment advice. For example, on the “Key Health Recommendations” webpage, a 23andMe client might read a recommendation to “Eat more leafy greens and fish” based on the finding that his “DNA puts [him] at risk for an eye condition called age-related macular degeneration (AMD),” a common cause of irreversible vision loss in people over 60.
The more serious example given by the FDA in its letter concerns the anticoagulant drug warfarin. A client with a prediction of slightly increased warfarin sensitivity due to specific genetic variations might read, “Your DNA suggests that if you take a blood thinner called warfarin (Coumadin) you may require a lower dose.” The FDA worries that patients “may begin to self-manage their treatments through dose changes”: “For example, false genotype results for [23andMe’s] warfarin drug response test could have significant unreasonable risk of illness, injury, or death to the patient due to thrombosis or bleeding events that occur from treatment with a drug at a dose that does not provide the appropriately calibrated anticoagulant effect.” This suggests that the FDA is not concerned with self-management of drugs per se, but rather with the risk of harm if the information that this self-management is based on is false.
When Are Genetic Test Results “False”?
So when exactly are genotype results “true” or “false”? In the context of genetic testing, there are three kinds of evaluations for accuracy. The first, and most basic, is analytic validity: a test is analytically valid to the extent that it turns out the right “raw data.” This includes the ability to accurately and reliably report specific mutations, single nucleotide polymorphisms (SNPs), and more complex genotypes. So analytic validity includes, for example, the ability to say whether a person has the nucleotide cytosine (C) or the nucleotide thymine (T) at a given position in a gene on a specific chromosome. The second kind of accuracy is clinical validity: a test is clinically valid to the extent that the interpretation of the raw data for presence or risk of a specific disease, condition, or trait correlates well with the actual presence/risk. It’s the ability to say that because a person has a C instead of a T at this location, he is more likely to develop a given disease. The final type of accuracy is clinical utility: a test has clinical utility to the extent that its use can help improve patient outcomes. Using our example, this is the capacity to say that although a person is more likely to develop a given disease due to the presence of a C rather than a T, research shows that regular exercise nearly eradicates this increased risk and is thus advisable.
It seems uncontroversial that 23andMe should be required to prove a reasonable level of analytic validity for any kind of testing, simply as a matter of integrity in advertising. If 23andMe claims (as it did) that it will sequence your DNA, it would seem to constitute false advertising if a significantly large proportion of results were inaccurate due to mislabeling of samples, faulty equipment, or lab technician mistakes (typical causes of analytically inaccurate results). Beyond analytic validity, the clinical validity and clinical utility of many genetic tests is admittedly low. We either 1) don’t know enough about the gene in question, 2) don’t know how the gene interacts with other genes or with environmental factors, or 3) in the case of a test for a devastating disease like Huntington’s, know we can predict the development of the disease with certainty but cannot treat it.
In thinking about what levels of clinical validity and utility the FDA should require for these tests, it seems right, as the director of the FDA’s Centre for Devices and Radiological Health noted in 2010, that “the level of regulation applied to [in vitro diagnostic devices] is based primarily upon risk to the patient of an undetected incorrect test result.” However, we think the “risk” of an incorrect test result negatively affecting individual health in the context of DTC-GT is much less significant than the FDA purports it to be.
How Serious Are the Risks of Incorrect DTC-GT Results?
Let’s take the example that the FDA uses in its letter to 23andMe about BRCA (breast cancer susceptibility genes 1 and 2) test results: “If the BRCA-related risk assessment for breast or ovarian cancer reports a false positive, it could lead a patient to undergo prophylactic surgery, chemoprevention, intensive screening, or other morbidity-inducing actions.” This passive construal of the problem ignores a crucial character in situations in which a test “lead[s] a patient to undergo prophylactic surgery”: her physician. A 23andMe client who tests positive for a harmful BRCA mutation doesn’t run out to the garage and fire up the chainsaw. She brings this information to her doctor. She might ask for, and even demand, a mastectomy, but an unnecessary mastectomy is only a “risk to the patient” if a doctor agrees to perform one.
In this situation, there should be at least two safeguards in place to ensure that the doctor does not perform an inappropriate mastectomy. First, the doctor must be assured of the accuracy of a test result; if the patient has tested positive for a harmful mutation, the doctor should order a follow-up test for confirmation, as in this 23andMe “patient success” story. Second, the doctor and patient must work together to integrate any genetic test results with information about “age, medical history, prior treatments, past surgeries and personal preferences,” as well as the doctor’s own comfort with performing a mastectomy on that particular patient. If either of these conditions is not met, then a responsible doctor will not proceed with the surgery. The same principle should apply to the less radical options, including chemoprevention and intensive screening.
If anything, the more dangerous case is that of a false negative. As 23andMe notes in the patient case referenced above, Ashkenazi Jewish women who have harmful genetic mutations usually test positive for one of three common mutations. As such, 23andMe tests only for those three mutations and could miss a less common harmful mutation (there are hundreds of BRCA1 and BRCA2 mutations that have been found to be associated with cancer). Even if 23andMe’s test battery did look at less common mutations, the absence of a robust data repository on the effects of such mutations would likely result in a higher percentage of uninformative results – a problem particularly challenging for minorities. But this scenario is not enough to discourage information-gathering; as mentioned above, genetic information alone is not enough to make therapeutic decisions. A woman with a family history of breast cancer, for instance, can still seek her doctor’s guidance about additional genetic testing or risk evaluation in the absence of positive DTC-GT results. And a doctor should not summarily dismiss cancer risk because of an inconclusive genetic test if there are other reasons to suspect the patient may be at high risk.
These examples highlight a more general and crucial principle in the context of DTC-GT: information provided by DTC-GT companies, even if it has medical implications, is still just information. The risk of harm based on an inaccurate test only becomes exceptionally high within the medical setting, where the presence of a health care professional should ameliorate most if not all of the risk of harm through the safeguards described above. The type of behavioral changes that a 23andMe client can make on his own, such as exercising more or eating more fish, do not entail the same high risk of harm in the case of false positives or false negatives. The risk of harm from self-management of drugs based on inaccurate test results, while usually more serious than exercise or dietary changes, is similar to the risk of harm that we accept every day in trusting patients to take prescription drugs in the amount and duration prescribed, or even over-the-counter drugs purchased and consumed without any direct physician involvement. Patients could obtain and act upon inappropriate drug management advice, and indeed many types of health-related behavior advice, from a slew of sources, including family members, friends, the Internet, horoscopes, and self-help books.
Given that high-risk courses of action will be mediated and should be ameliorated by a physician, we conclude that the risk of harm from inaccurate DTC-GT results is low enough that the FDA should not ban or make DTC-GT more difficult for consumers to access so long as the tests meet standard levels of analytic validity. (Although it has yet to prove it to the FDA, 23andMe states that its “lab partner adheres to strict quality standards that are part of the Clinical Laboratory Improvement Amendments of 1988.”) In general, the risk of harm is not high enough to warrant the significantly extra time and money that it would cost individuals if the FDA required pre- and post-test individual counseling with a genetic counselor. Nor is a lack of conclusive research on or therapeutic interventions for a given condition, resulting in tests with low but still appreciable clinical validity and/or clinical utility, grounds for preventing an individual from accessing this test, provided the online consent process is sufficiently thick to impart the limitations of the test. In fact, clinical validity and utility for genetic information will only improve to the extent that researchers can access and analyze more genomes and personal/family history profiles, which 23andMe is well-placed to gather with its cheaper and more easily accessible genetic testing. Instead of blocking these advances for both individual and population health, we argue that the FDA should work with the medical community and DTC-GT companies to ensure a smooth transition and a proper “access point” out of the consumer setting and into the medical setting for situations in which individuals could receive troubling or inconclusive test results.
The Consumer-to-Patient Transition
As an example of this type of consumer-to-patient transition and access point, consider the case of a woman who purchases a home pregnancy test. These pregnancy tests, like certain DTC-GT results, reveal potentially life-changing information to the woman about herself, namely whether or not she is pregnant. Like DTC-GT, they allow the woman to access this information without consultation and in the privacy of her own home. If the test is positive, the Women’s Health Office of the Department of Health and Human Services advises, “You should call your doctor right away. Your doctor can use a more sensitive test along with a pelvic exam to tell for sure if you’re pregnant.” There’s an easily recognizable “access point” for the woman into the medical setting (making an appointment with her doctor or going to a women’s health clinic) and a clearly defined pathway for a doctor whose patient brings in her pregnancy test results (run a more sensitive blood test and a pelvic exam).
Admittedly, the results of DTC-GT tests will be more complex and usually far less conclusive than a pregnancy test, but the principles remain the same. The 23andMe client needs a recognizable “access point” if faced with confusing or serious results. Should he bring a printout of his results to his next doctor check-up? Should he spend money on a private counseling session with a clinical geneticist? On the other side, individual doctors and health care professionals need guidance on the proper pathways to take when a patient brings in such DTC-GT information. Should a doctor order a follow-up test based on these results, even if the test is invasive or expensive? The answer to this will most likely depend on how similar 23andMe’s tests are to those ordered by physicians, the general clinical validity of tests for the given risk prediction, and whether 23andMe’s reports take family history, age, ethnicity, and other factors into account alongside genetic information – all questions to which doctors, who are usually not experts in genetics and particularly in recent genetic testing developments and specific DTC-GT companies’ practices, generally don’t know the answer.
In the December 5th statement announcing 23andme’s intent to comply with FDA regulations on the company blog, CEO Anne Wojcicki wrote: “23andMe has been giving consumers access to health information for six years and is committed to finding the right regulatory path for our customers.” We think a strong option for the “right regulatory path” places the FDA in an educative, synergistic role in promoting better relationships between the doctor, patient/consumer, and DTC-GT company. Rather than focusing exclusively on protecting individuals from accessing their own information, the FDA has the opportunity to strengthen the safeguards and access points for appropriate utilization of genetic information.