Regulating Direct-to-Consumer Genetic Testing vs. The Right to Know
For about $100, customers can have DNA from their saliva analyzed by the genetic testing company 23andMe for hundreds of gene variants thought to be linked to traits ranging from ancestry and earwax type to warfarin sensitivity and Alzheimer’s disease.
While 23andMe gives customers access to their raw genetic data, some would say the real value of their product lies in the interpretation packaged with these data. For gene variants linked to health and disease traits, however, the U.S. Food and Drug Administration has taken notice as 23andMe has yet to provide them with premarket assurance that these interpretations are clinically validated and accurate. The FDA argues that using 23andMe’s service this way qualifies it as a medical device, and ordered the company to cease marketing their kit as a health management tool until these assurances were met.
How should the direct-to-consumer (DTC) genetic testing industry answer to the FDA?
Last month, the San Francisco Bay Area chapter of Oxbridge Biotech Roundtable (OBR) convened experts at the University of California, San Francisco to formally debate the FDA’s recent decision to bring DTC genetic testing under its regulatory authority. This decision marks the most recent chapter in five years of largely unproductive communications between the federal agency and 23andMe, culminating in December, when the FDA ordered the company to “immediately discontinue marketing the Personal Genome Service (PGS) kits until such time as it receives FDA marketing authorization for the device.” (http://www.fda.gov/iceci/enforcementactions/warningletters/2013/ucm376296.htm)
Arguing in favor of the FDA’s actions were Gianfranco de Feo, VP of Marketing at the Redwood City-based computational genetics company Bina Technologies, and Patricia Zettler, a fellow at the Stanford University Law School Center for Law and the Biosciences. In opposition were Andro Hsu, Director of Product Marketing at Syapse, Inc., a Palo Alto-based data management software company, and Paul Billings, Chief Medical Officer for Life Technologies. They were joined by Glenda Anderson, CEO and co-founder of the healthcare IT company Farsight Genome Systems.
A medical device?
If a medical device calls to mind images of heart stents and CT scanners, it might seem surprising that the FDA wants to regulate 23andMe’s DNA testing service under this classification. The distinction is worth thinking about, however, because it relates to the type of service 23andMe claims to be offering. Here’s an excerpt from the official FDA definition for a medical device:
"An instrument, apparatus, implement, machine, contrivance, implant, in vitro reagent, or other similar or related article, including a component part, or accessory which is intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease.” (http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/Overview/ClassifyYourDevice/ucm051512.htm)
Patricia Zettler argued that 23andMe’s test unambiguously satisfies the FDA definition because it was marketed as a way to prevent and diagnose disease. Last year, 23andMe ran a television spot which promised—or at least strongly insinuated—that they could tell customers “hundreds of things” about their health, including their risk for heart disease, arthritis, gallstones and hemochromatosis .(http://www.ispot.tv/ad/7qoF/23-and-me)
“Know more about your health,” the sanguine narration goes. “Change what you can, [and] manage what you can’t.” While the FDA doesn’t object to providing customers access to their raw genetic data, they drew the line when 23andMe started marketing the test as a way to manage disease risk.
Andro Hsu saw things differently. Despite 23andMe’s agressive marketing, he said that their product can’t be classified as a medical device because the interpretation provided with raw test results gives only a statistical risk for health-related conditions, not a strict diagnosis. As a former writer and early employee of 23andMe, he recalled obsessing over the wording of their online content as they tried to communicate this distinction to customers. They tried, he said, to relay that genetic risk accounts for only a portion—usually a small portion—of the overall risk of developing a certain disease. Other non-genetic factors, like environment and lifestyle, are important too.
Even conceding that people may effectively use 23andMe’s test to manage disease, Hsu said that this definition of a medical device “is too broad to be useful.” He argued that gyms and bathroom scales can also be used to prevent and manage disease, but are obviously unregulated by the FDA. Fitbits, and other wearable, fitness-oriented technology are also unregulated. Other services, such as WebMD.com’s symptom checker—a tool which allows users to input symptoms to see a list of possible medical conditions—carry a familiar “for informational use only” disclaimer, and are unregulated by the FDA.
Paul Billings said that calling 23andMe’s DNA test a medical device promotes a “vague definition of [the FDA’s] authority and of what a device is.” Given the lack of regulation over other similar entities, he called its action against 23andMe “unusual and capricious.”
Medical maternalism or consumer advocacy?
The FDA’s decision to regulate 23andMe’s DNA testing kit could be interpreted as an extension of its mandate to “protect and promote the public health.” Specifically, it worries that patients might make hasty medical decisions without consulting their primary care doctor, and of the “potential health consequences that could result from false positive or false negative assessments for high-risk indications.” (http://www.fda.gov/iceci/enforcementactions/warningletters/2013/ucm37629...) Examples might include a patient inappropriately adjusting his or her medication dose or seeking unnecessary prophylactic surgery.
Billings said that this concern is overdrawn, and cited published studies which failed to show significant harm in providing customers with their genetic test results. In one study, Stanford geneticist Dr. Uta Francke surveyed women who had tested positive for variants of the gene BRCA1, which is associated with increased risk for breast cancer. She found that all the women surveyed acted appropriately with the information, seeking more information and re-testing. Furthermore, family members who were unaware of their risk were also informed (https://peerj.com/articles/8.pdf). In another study from Boston University, researchers found that disclosure of APOE genotype (a risk factor for Alzheimer’s disease) to adult children of patients with the disease produced no significant short-term psychological risks.(http://www.bu.edu/alzresearch/files/pdf/Green_NEJMREVEALIpaper_final6.pdf)
Gianfranco de Feo argued that just because knowing test results doesn’t seem to hurt people, it doesn’t obviate the need for these results to be accurate and clinically meaningful. He alluded to studies conducted by researchers at the University of North Carolina at Chapel Hill (http://www.ncmedicaljournal.com/wp-content/uploads/2013/11/74608.pdf) as well as by the U.S. Government Accountability Office (http://www.gao.gov/assets/130/125079.pdf), which reported discrepancies between test results for the same samples from different leading personal genomics companies, including 23andMe. Furthermore, some of these companies were unable to provide good clinical interpretations for African American and Asian customers, a limitation which they did not communicate well when the kits were sold.
Consumer misdirection aside, de Feo emphasized that the science behind genetic testing is not easy. Interpreting raw test results depends on many factors, such as which markers are assayed, which algorithms are used to process the data, and how the clinical trials were designed in which the gene variants were first identified. Many of these original trials were conducted in very small cohorts, many years ago.
Since the current model of conducting large clinical trials or genome-wide association studies (GWAS) for each genetic variant might be unfeasible given the quantity of targets, de Feo advocated for using the technology provided by the personal genomics industry to create large translational databases that could link phenotypes to specific gene variants. With enough participation, this approach could streamline better and more clinically relevant test results.
Zettler reminded us that the FDA’s regulatory powers have historically been added after public health disasters. Since the agency was formed in the early 1900s, lawmakers added premarket requirements that drugs be shown safe (1938) and effective (1962), following, respectively, hundreds of deaths in the United States due to diethylene glycol-contaminated sulphanilamide and thousands of birth defects in Europe caused by thalidomide.
Medical devices were brought under the same umbrella in 1976 after defective intrauterine devices (IUDs) injured thousands of women in the United States. de Feo cautioned that only one salient public health disaster, even among a small group of individuals, could set back the consumer genetics industry decades, along with any potential benefits for genomics research. In that light, he thought it was worth it to preventatively regulate the industry and ensure that tests are high quality and clinically validated.
The right to know, the right not to know
Gianfranco de Feo recognized that the issue of genome ownership evokes a strong emotional reaction, and many people feel that access to their genetic information is a basic right. California and Oregon both extend civil or property rights guaranteeing residents access to their available genetic information.
Even so, some patients may not actually want to know their genetic status, especially for untreatable diseases. Economist Emily Oster studied the behavior of at-risk individuals for serious diseases. Among people at high risk for Huntington’s Disease (i.e. one parent has or had the disease), she found that only five percent sought genetic testing, despite the near 100 percent predictive power of the test (http://faculty.chicagobooth.edu/emily.oster/papers/hdbeliefs.pdf).
Similarly low rates of voluntary testing are reported for breast, ovarian and colon cancer and Alzheimer’s disease, for which predictive genetic tests also exist, albeit with less predictive certainty. When the course of medical treatment is the same in either case, some argue that knowing one’s genetic status adds no value. On the contrary, it could diminish the quality of life for these individuals as well as that of close family members, who may share the same gene variants but may not have sought testing themselves.
For most customers, however, the central issue may be one of packaging how the information is presented. Glenda Anderson called 23andMe “recreational genomics,” highlighting the fact that the technology’s ability to predict and manage disease remains in its infancy.
While 23andMe encourages customers to share their test results with their doctor before making any medical decisions, most doctors are not trained to interpret this information. Building the requisite databases to improve the technology will take time. Meanwhile, companies like 23andMe must help consumers and their doctors understand how well the science supports the clinical interpretation.