Regulatory controls for direct-to-consumer genetic tests: a case study on how the FDA exercised its authority

Abstract

In February 2015, 23andMe received clearance from the United States Food and Drug Administration (FDA) for a carrier status test for a gene linked to Bloom syndrome. This was the first FDA authorization to market a direct-to-consumer (DTC) genetic test. Then, in April 2017, 23andMe obtained marketing clearance for an additional 10 tests that communicate information about individual disease risk. For roughly a decade it had been unclear whether FDA would regulate health-related DTC genetic tests. The recent approvals now provide an opportunity to examine how the Agency, in practice, has exercised its regulatory authority. This paper is the first case study to examine in detail how FDA has set standards for the marketing approval of DTC genetic tests.

Keywords:

• direct-to-consumer
• genetic testing
• FDA
• regulatory controls
• disruptive technology

Introduction

On 19 February 2015, the United States Food and Drug Administration authorized for marketing 23andMe’s Personal Genome Service^® for Bloom Syndrome Carrier Status Test (23andMe 2015b). The test determines whether a healthy person has a variant in a gene that can lead to future children inheriting the disorder that causes short stature and skin rashes. Bloom syndrome is an autosomal recessive disorder, meaning both parents would need to pass on the variant to have an affected child. This was the first FDA authorization to market a direct-to-consumer (DTC) genetic test. The FDA classified the autosomal recessive carrier screening test as a class II, or medium-risk, medical device, which has a less onerous regulatory pathway than class III high-risk devices. The Agency also clarified that similar tests can enter the market without FDA review if they comply with special controls (Rubin 2015). Since the February 2015 approval, 23andme obtained FDA marketing approval in April 2017 for an additional 10 genetic health risk reports for diseases including, late-onset Alzheimer’s disease, Parkinson’s disease, celiac disease, and factor XI deficiency, or hemophilia type C (23adMe 2017). Similar to the approval for its Bloom Syndrome Carrier Status Test, 23andMe conducted accuracy and user comprehension studies for the additional 10 tests to meet FDA standards (23andMe 2017). The FDA approval of these tests set a precedent and allowed the FDA to develop regulatory guidance that establishes standards for carrier and disease-related health risk reports.

For roughly a decade it had been unclear whether the FDA would regulate health-related DTC genetic tests. Policy-makers, academics, and industry representatives have discussed whether the FDA should regulate and how it might do so. The recent approvals now provide an opportunity to examine how the Agency, in practice, has exercised its regulatory authority. This paper is the first case study to examine in detail how the FDA has set standards for the marketing approval of DTC genetic tests, including carrier screening tests and tests that report genetic alterations that may increase one’s chances of developing certain diseases. The paper focuses on 23andMe, as it is the only DTC provider to date to have received authorization from the FDA to market its genetic tests in the United States. I apply the concept of governance to the case study to understand the systems in place to protect subjects undergoing genetic testing. Until recently, the FDA had been exercising enforcement discretion over most genetic tests, including DTC genetic tests. This oversight stance was backed by the assumption that clinical genetic testing provided sufficient governance through the norms of medical practice. In this model of clinical governance, the health care professional is the expert who helps the subject interpret and contextualize the meaning of genetic test results.

To understand how 23andMe challenged the boundaries of clinical governance model, I examine the DTC firm as a disruptive technology. Beginning in 2005 DTC firms including 23andMe offered a new form of health-related genetic testing and sought to disrupt the boundaries of clinical genetic testing by carving out a new domain for non-medical, or information genomics. In response, vested stakeholders have sought to reestablish the boundaries of clinical genetics by defining what constitutes a scientifically sound and clinically useful genetic test. 23andMe continued to push the boundaries of clinical genetics by offering testing that might direct therapeutic choices. This disruption to the model of clinical governance forced the FDA to take repeated action against 23andMe and establish a new model of governance for certain categories of DTC genetic tests within its existing regulatory structure. The paper asks how the FDA accommodates a disruptive technology within an existing regulatory regime and takes the recent marketing authorizations to explore how the FDA governs DTC firms and the test-taking subject in relation to and in the absence of clinical governance.

Conceptual framework: governing disruptive technology

Foucault’s notion of governance provides a framework to understand how modern nation states use techniques of governance to form the very subjects, or citizens, they aim to protect (Foucault 2001). Foucault argues that modern nation-states govern their citizens through networks of power that are distributed across different institutions. The concept supports the idea that forms of governance inscribe ideas about the subjects of governance, which in turn, further define the status of those very subjects. I argue that the FDA has regulated DTC genetic testing in a way that creates a form of conditional “empowerment” for subjects who undergo testing. The concept also allows for analysis of a range of institutions, not only formal regulatory bodies, but professional societies as well. This account of governance resonates with how Hogarth, Melzer, and Zimmern (2005, 3) frame regulation with respect to genetic testing:

Historically much discussion of regulation has focused on the regulation of business activities through what is termed “command and control” regulation: that is “regulation by the state through the use of legal rules backed by (often criminal) sanctions.” But there is now a greater focus on what is termed de-centred regulation – the use of (often non-statutory) instruments by a range of public and private actors from trade associations to public interest groups.

DTC genetic testing has generated a huge debate among medical professionals, bioethicists, government bodies, and oversight committees (Hudson et al. 2007; Borry, Cornel, and Howard 2010; Hogarth 2010; Skirton et al. 2012; Edwards and Huang 2014), which historically have employed and currently exercise forms of governance over genetic testing technologies and surrounding practices. These stakeholders have employed forms of de-centered regulation over genetic testing by establishing norms and professional guidelines. FDA has governed DTC genetic testing through a command and control form of regulation by classifying the technology as a medical device and now requiring marketing authorization for certain categories of DTC genetic tests. The opinions and concerns of vested stakeholders continue to provide a form of de-centered regulation that aligns with the FDA’s regulatory stance.

To explore how DTC genetic testing has disrupted the model of clinical governance, I use the concept of a disruptive technology. Scholars have described disruptive technologies as having the potential to “transform the way we live and work, enable new business models, and provide an opening for new players to upset the established order” (McKinsey 2013, iii). The theory of disruptive innovation was first developed by business scholar, Christensen (2013), who argued that disruptive innovation can transform an existing market or sector by, for example, reducing complexity and cost to eventually displace established competitors. Christensen (2013) describes disruptive technology as a new emerging technology that unexpectedly displaces an established technology. The concept has more commonly been used in business and finance contexts (Brummer 2015), and more recently scholars have applied the concept to examples beyond finance, to include whole-exome sequencing (McCullough et al. 2016), health care delivery (Kain, Hwang, and Warner 2015), and CRISPR gene editing (Lefferts 2016). This paper applies the concept of disruptive technology to DTC genetic testing to analyze how DTC firms have enabled a new business model for genetic testing, disrupted the established order for clinician-mediated genetic testing, and opened a new, relatively affordable, channel of access to genetic information. While the application of the concept to business case studies allows for analysis of business strategy and paths to market entry, in this paper I apply the concept to explore the causes and objects of disruption, specifically how DTC genetic testing technology has challenged existing governance structures and norms for accessing genetic information. DTC genetic testing firms that began delivering health-related information in 2005 sought to disrupt the genetic testing market place by challenging the assumption that health-related genetic information should only be provided within a clinical context. These firms also launched their services without seeking any regulatory oversight by the FDA. The paper by Hogarth (forthcoming) in this special issue provides further discussion of how the logic of disruption is often bound up with the idea that existing regulations are not appropriate for new technologies and disruptive innovation.

Debating the harms and benefits of consumer genetics

When genetic testing became available, the tests were developed within hospital laboratories and provided to patients usually with the coordination of a pathologist and clinician. The tests were developed “in-house” and viewed as low-risk due to professional oversight, and were therefore not regulated by the FDA. As demand increased for genetic testing, commercially distributed genetic testing kits became available to clinical laboratories. The FDA regulates these genetic test kits. Most of the policy and ethical concerns over genetic testing, such as consent, the right to know (or not know) one’s genetic information, and disclosure to family members, for example, have been addressed within the medical and bioethical professions. Advisory bodies in the United States, such as the Task Force on Genetic Testing (1997) and the Secretary’s Advisory Committee on Genetic Testing (2000), however, have called for regulatory reform and increased oversight of genetic tests. DTC genetic testing raised additional concerns about sufficient oversight as it disrupted the traditional model for clinical genetic testing. As Hughes (2013, 3) questions, “Is there something problematic about personalized genetic knowledge given to individuals, without the mediation of doctors and clinical geneticists, the usual gatekeepers of medical information?” Outside the context of clinical care, there have been two major concerns. One is whether the genetic information is reliable and the other is whether consumers might make rash decisions in the absence of medical advice and support (Lippi, Favaloro, and Plebani 2011; Leighton, Valverde, and Bernhardt 2011). Bioethicists have argued that misunderstanding one’s individual risk may lead to unnecessary anxiety (Hughes 2013; Delaney and Christman 2016).

Researchers have conducted empirical studies with DTC genetic test takers and concluded that these fears are unsubstantiated (Roberts and Ostergren 2013). They found no evidence of improved health outcomes through modified behavior nor concerns about psychological distress due to misunderstanding test results (Roberts and Ostergren 2013, 182). Much of the debate has centered on whether DTC consumers are really empowered by “directly” accessing their genetic information through private companies without the mediation of a physician or genetic counselor. In the words of Annas and Elias (2014, 986), “To oversimplify, the debate has been framed as a struggle between medical (or government) paternalism and individuals’ right to information about ourselves.” This debate is central to this case study and understanding the FDA’s regulatory approach towards health-related DTC genetic tests.

Advocates for DTC genetic testing promote the idea of patient empowerment. They argue that patients are empowered by having a more active role in the management of their own health care through direct access to their genetic information (Chung and Ng 2016). As part of its mission, 23andMe claims to empower its consumers and has openly sought to disrupt the traditional boundaries created by the medical and scientific establishments. In an interview just after launching the 23andMe Personal Genome Service^® in the United Kingdom, 23andMe co-founder and chief executive officer, Anne Wojcicki said:

There is a division between the people who wear the white coats and those who don’t … Science has a reputation of being inaccessible, for the super smart or the elite, but I want to make people realise that science is accessible to everyone and anyone can understand their genome and it is fascinating. (The Guardian, 1 December 2014)

Scholars have argued that the FDA should not presume that consumers are unable to understand their own genomic information and therefore deny access to these data (Baudhuin 2014). Some contend that DTC genetic testing has helped to create greater awareness of human genetics, but not necessarily correct understanding (Hughes 2013). Ethicists, scientists, and policy makers have questioned whether the average citizen can understand predictive, probabilistic genetic information, and, for example, distinguish diagnostic from predictive genetic testing. Some argue that testing should only be available with the mediation of medical professions.

User understanding is also connected to informed consent. Informed consent within the model of clinical genetic testing seeks to ensure sufficient patient understanding so that patient can make autonomous decisions. The patient, however, is dependent on the health care provider to explain medical information and choices to make autonomous decisions. This dynamic creates tension in the clinical model, as the patient cannot act autonomously without receiving sufficient, reliable information from the health care provider. The tension in this model combined with calls for patient empowerment beginning in the 1960s opened the door for DTC providers to disrupt this model. The debates over professional mediation, public understanding of genetics, and consumer empowerment are interwoven. The traditional clinical model operates on the assumption that subjects undergoing genetic testing cannot sufficiently understand genetic information without professional mediation, and therefore would not be able to make well-informed autonomous decisions. The logic follows, though, that if consumers can understand their personal genetic information, then professional mediation may not be needed, and consumers in fact, may be empowered by purchasing such information outside a clinical context.

Central to discussions about DTC genetic testing are attempts to distinguish and parse out the range of genetic test offerings and corresponding genetic information. The logic often follows that the risks to consumers may not be the same for all genetic tests. Currently, a wide range of testing is available, from diagnostic testing of single-gene disorders, to predictive or pre-symptomatic testing that identifies alterations in multiple disease-associated genes that contribute towards a polygenic risk score. The range of applications for new sequencing technologies continues to foster discussion about how to develop effective policy that addresses distinct applications, for example, prenatal and diagnostic genetic testing (Curnutte et al. 2016). Susceptibility testing based on genome-wide association studies (GWAS), large population studies that look for statistical correlations between genetic variations and phenotypes, can provide information about one’s likelihood of presenting a certain condition or disease relative to the studied population (MacArthur et al. 2016). Data from these studies contribute towards polygenic risk assessments that have served as the basis for both DTC nutrigenetic tests and DTC disease-related risk assessments. Scientists, however, have questioned the soundness and value of the information provided by polygenic risk assessments, as it might only infer minimal risk (Goldsmith et al. 2012). GWAS are the foundation of the polygenic risk assessment tests provided by 23andMe as part of its Personal Genome Service^®. There are limitations with these tests, given the composition and size of the studied population and combination of GWAS used to make a risk assessment (Ng et al. 2009; Klein, Lohmann, and Ziegler 2012). This is in part why it has been a challenge for 23andMe and other test developers to demonstrate the clinical validity of these tests (Ng et al. 2009), which is central to FDA’s evaluation of such tests.

It is not an easy picture to tease apart when considering that genetic tests can diagnose a disease, provide the probability of a future child having a health condition, and provide some information that contributes to one’s risk of developing a disease such as colon cancer over a lifetime. Genetic tests can also provide information that some might not consider medical, such as eye color, freckling, earwax type, and athletic abilities (Camporesi 2013; Webborn et al. 2015). Some scholars have termed this sort of testing “recreational genomics” and have argued that this sort of testing is only of entertainment value, or marginally scientific (van Ommen and Cornel 2008; Caulfield et al. 2015; Chung and Ng 2016). Others have looked more critically at how consumer genomics has led to the “lifestylisation” of health care (Lucivero and Prainsack 2015). Saukko et al. (2010) studied DTC nutrigenetic tests and found that they mapped onto four frames or models for genetic testing: clinical genetics, medicine, intermediate, and lifestyle. In this work, the authors argued that the genomics researchers and policy makers appeared to perform boundary-work. Building on the scholarship of Gieryn (1983), which shows how sociological processes lead to the demarcation of science from non-science, the authors argued that this sort of boundary-work delegitimized the nutrigenetic tests as outside proper medicine and science (Saukko et al. 2010). DTC stakeholders have continued this same boundary-work. The DTC industry, regulators, and professional societies have struggled over where these boundaries should be drawn (Becker et al. 2011). The recent FDA approvals for 23andMe’s tests show where the Agency has drawn some lines for permissible DTC genetic testing – carrier and disease risk assessments can be marketed directly to consumers.

The multiple framings of genetic information, in part, have enabled 23andMe to tread a fine line between offering a testing service that is not medical, yet “health related” (Curnutte and Testa 2012). Predictive presymptomatic tests are not diagnostic, and therefore, providers of such tests like 23andMe, Navigenics, deCODE genetics, and Pathway Genomics, were able to minimize some concern about whether professional mediation was needed and what consumers might do with the genetic information provided by the companies. None of these companies, including those that are still operational, either provide or provided single-gene and/or diagnostic genetic tests. Most of the information that has been provided through DTC genetic testing companies is not medically actionable in the sense of an immediate medical intervention. Most test results that have been provided by these companies would suggest following well-known lifestyle recommendations – eating well, exercising, getting good sleep, and not smoking. It is harder to make this argument, though, for pharmacogenomic tests, which aim to use gene alterations to predict whether a medication will be effective or cause an adverse reaction in an individual. Currently, there is a range of genetic information available through genetic testing, and both the science backing the test as well as the sort of information it provides are central to the debate over whether people should be able to access genetic information over the Internet without professional mediation.

Key policy developments for DTC genetic testing: 2006–2017

DTC advertising of genetic tests and other DTC genetic testing platforms pre-date the launch of 23andMe in 2007. A handful of nutrigenetic companies that tended to offer polygenic risk assessment tests emerged in the 1990s and early 2000s claiming to shed light on the genetic basis of individual metabolism and nutrition (Saukko et al. 2010). The United States Congress (GAO 2006), FTC, CDC, and FDA (2006) cautioned against the veracity of the companies’ claims. This, combined with the DTC marketing of genetic testing for the Breast Cancer genes 1 and 2 (BRCA1/2), the best-known genes linked to breast cancer risk (Gollust, Hull, and Wilfond 2002; Matloff and Caplan 2008), increased awareness, concern, and discussion about new channels for the delivery of genetic information.

A new wave of health-related DTC genetic testing began with the launch of Knome in 2005. Harvard geneticist, George Church, founded the company and began offering whole genome sequencing for $350,000 (Harris, Kelly, and Wyatt 2016). 23andMe and similar companies like Navigenics, founded in 2006, and Pathway Genomics, which now offers its testing service through physicians, and Iceland-based deCODE genetics also began to provide health-related genetic information to consumers directly over the Internet (Harris, Kelly, and Wyatt 2016, 7–8). In 2008, The European Technology Assessment Group conducted a survey of DTC companies and found 38 active companies (ETAG 2008). Of these companies, 23andMe has had the greatest staying power, most likely due its venture capital backing and evolving business model. The company has created a valuable database of more than 1,000,000 genetic profiles (23andMe 2015a) that has led to profitable pharmaceutical partnerships (Alba 2015).

The health-related DTC genetic testing firms operated for many years with relatively few regulatory restrictions. The California and New York Departments of health insisted that laboratories operating in each state demonstrate Clinical Laboratory Improvement Amendments certification (State of California – Health and Human Services Agency 2008). The publicity received by these companies, however, led to intense scrutiny of the test results provided by these companies. In 2010 the Government Accountability Office (GAO), at the request of the United States Congress, published a report that examined the practices of four companies to evaluate the reliability of the tests and company privacy policies. The investigative arm of Congress reported that companies misled consumers and used deceptive marketing. The report documented inconsistent test results between and within companies, raised questions as to whether test results were accurate for ethnic and racial minorities, and cited misleading company interpretations of genetic information in post-testing interviews (GAO 2010). The GAO report was featured during a 2010 Congressional hearing, where medical and scientific experts testified that the information provided by the companies was only of entertainment and not medical value. Throughout the hearing, company representatives, FDA officials, and Congressional representatives debated the status of the genetic information provided by the companies (U.S. Congress 2010). Their questions centered on whether the information was medically significant and scientifically sound (U.S. Congress 2010).

Following the Congressional hearing, in 2010, the FDA sent warning letters to 15 DTC firms clarifying that the testing platforms were medical devices, and therefore under the jurisdiction of the FDA (FDA 2010b). The Agency reasoned that harm and unreasonable risk could result from both incorrect and unreliable data, which could lead to bad or harmful health care decisions. The Federal Food, Drug, and Cosmetic Act of 1938 defines a medical device as “an instrument, machine, contrivance, implant, in vitro reagent, or other similar article that is intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment or prevention of disease” (PL 95–295, 90 Stat 539). Genetic tests can either be kits sold to laboratories or laboratory developed tests (LDTs), a type of in vitro diagnostic that are classified as medical devices. Historically, the FDA has used enforcement discretion over LDTs, allowing them to enter the market without prior approval. It has been unclear whether the FDA would increase its oversight of LDTs, although it had indicated its intention to do so and had consulted twice on proposals for new regulatory frameworks that would clarify the scope of its activity (FDA 2014; Ray 2016).

The FDA has been primarily concerned with establishing that medical devices and the information conveyed to consumers are reliable and accurate. In 2009, Alberto Gutierrez, head of the FDA’s Office of In Vitro Diagnostics said: “We really don’t have any issues with denying people information. We just want to make sure that the information they are given is correct” (New York Times on 11 March 2009). The regulatory emphasis has been on three evaluative criteria: analytic validity, the reliability of the actual genetic test and laboratory performance; clinical validity, whether the genetic variant corresponds to the condition or trait; and consumer comprehension.

After 23andMe received its 2010 letter from the FDA, the company began working with the FDA to undergo premarket review of its tests. In November 2013, however, the FDA sent 23andMe a cease and desist letter informing the company that it did not have approval to offer interpretation of its health-related genetic tests to new consumers. Genetic consumers could still receive their uninterpreted raw genetic data and ancestry results, and existing consumers could continue to access their health-related results. The FDA was not convinced of the tests’ analytical validation, clinical validity, and risk-mitigation through appropriate labeling. The company was working towards regulatory approval since 2010, but stopped communicating with the FDA (FDA 2013). The timing of the 2013 letter also coincided with increased national visibility for the 23andMe testing platform, as described on the company’s blog, the firm had recently launched a $5 million advertising campaign (5 August 2013, 23andMe As Seen on TV).

The evolution of 23andMe and its regulatory comeback

When 23andMe first launched its testing platform in 2007, in addition to its ancestry testing, there was one alphabetical list of conditions and traits, mixing tests for one’s earwax type and asparagus metabolite detection with one’s predisposition to lactose intolerance, Crohn’s disease, and prostate cancer, as well as information about how one might metabolize a particular drug. When the DTC tests were first offered, there were only two categories of tests, “Clinical Reports” and “Research Reports.” Through the company’s own methods of scientific curation, the “Clinical Reports” were deemed more scientifically sound than the “Research Reports” based on their assessment of the relevant literature. The company has retained its research confidence ratings, which are communicated on a scale of one to four stars. Over time, the company began to break down its tests into four categories, “Carrier Status,” “Drug Response,” “Traits,” and “Disease Risk.” At the beginning of 2010, the company had gone from something that many termed “recreational” to clearly demarcating test offerings that provide information about how one might metabolize a particular drug. This was an important moment for the company, as it continued to market and draw boundaries around its genetic test offering as “health related” but not medical.

The 2013 letter that FDA sent to 23andMe specifically addressed the Agency’s concerns over what consumers might do with information provided through pharmacogenomics tests. While the FDA did not prohibit 23andMe from continuing to offer testing for its “Traits,” the following three categories of tests were: “Carrier Status,” “Drug Response,” and “Disease Risk.” In its own act of boundary-work, the Agency separated out which types of tests counted as a medical device raising a level of risk requiring FDA oversight. In response to the 2013 letter, Wojcicki put together a stronger regulatory team and launched a successful regulatory strategy for the submission of its Bloom syndrome carrier screening test (23andMe 2014). The FDA’s marketing authorization for 23andMe Bloom’s syndrome test was a milestone for the DTC industry, not only for the company’s demonstration of analytical validity and clinical validity to the FDA, but for the company’s successful navigation of the ethical and policy concerns that have beset DTC genetic testing. The test was a good place to reestablish trust and facilitate a regulatory pathway with the FDA, because the intended use of the test is quite narrow, does not diagnose a disease, and is indicated for an adult population. This type of genetic test skirts ethically sensitive issues, such as providing actionable diagnostic genetic information outside a medical context and directly to minors (see e.g. Borry et al. [2010] and Botkin et al. [2015] on this subject).

23andMe’s Bloom’s syndrome test is a carrier screening test, which means the company is not providing consumers with a disease diagnosis or information about how a drug might affect them. As for immediate medical actions, the stakes are relatively low. The test is for carrier status of an autosomal recessive disorder, which means the associated risks of a false positive or negative are relatively low, as the carrier status of one’s partner would also need to be known to assess the probability of having an affected child. Without this additional information, one’s carrier status alone is not actionable.

In the press release for the Agency’s marketing authorization of 23andMe Bloom’s syndrome test, Gutierrez said:

The FDA believes that in many circumstances it is not necessary for consumers to go through a licensed practitioner to have direct access to their personal genetic information. Today’s authorization and accompanying classification, along with FDA’s intent to exempt these devices from FDA premarket review, supports innovation and will ultimately benefit consumers. These tests have the potential to provide people with information about possible mutations in their genes that could be passed on to their children. (FDA 2015b)

At the time of the marketing approval for the Bloom’s syndrome test, it was still unclear whether a licensed practitioner would be needed for the provision of pharmacogenomics tests and disease-related susceptibility testing. The April 2017 approval of 10 additional 23andMe tests provides clarity for disease-related susceptibility testing. The 10 approved tests communicate information about one’s likelihood of developing certain diseases. With this class of tests, FDA’s concern has been the validation and clinical performance of the tests. On its blog (6 April 2017, Good News About Health Reports) 23andMe communicated that for this recent authorization, “23andMe conducted extensive validation studies for accuracy and user comprehension that met FDA standards.” It remains to be seen, though, whether FDA considers pharmacogenomics testing too disruptive and requires that it be provided through a licensed practitioner. For the approved carrier screening and disease-related susceptibility tests, the FDA has required a demonstration of genetic understanding before consumers can be “empowered” by accessing their genetic information. This next section focuses on the special controls issued by the FDA for DTC genetic tests.

Special controls for approval

The FDA decision summary for 23andMe’s submission for its Bloom Syndrome carrier screening test provides clarity about the special and general controls needed to provide adequate assurances of safety and efficacy for over-the-counter (OTC) use and DTC marketing of genetic tests (FDA 2015a). To satisfy the general controls, 23andMe provided information about the collection device (which must be FDA-cleared), instrumentation, software, as well as details about the test’s analytical performance. FDA cites specific pre-existing performance standards for the accuracy and reproducibility of test results in the decision summary. The clinical validity for these tests can be demonstrated, for example, by simply citing scientific literature and clinical guidelines. In addition to demonstrating the analytical and clinical validity of the test, to meet special controls 23andMe tested user comprehension pre- and post-testing with an educational intervention. In section 3(iii) of the summary FDA (2015a) clarifies:

If the device is offered over-the-counter, including cases in which the test results are provided direct-to-consumer, the manufacturer must conduct a study that assesses user comprehension of the device’s labeling and test process and provide a concise summary of test results of the study.

Specifically, the user comprehension study must obtain values of 90% or greater for each comprehension concept. It is not uncommon for the FDA to issue special controls for marketing authorization. Drug manufacturers, for example, provide results on user comprehension and readability of OTC medication labels with FDA submissions (Trivedi, Trivedi, and Hannan 2014). The FDA has also issued guidance for industry on how to conduct these studies (FDA 2010a). In the Agency’s press release (FDA 2015b) for the marketing authorization of 23andMe’s Bloom’s syndrome test, the FDA explained how the decision follows from regulation of other medical devices:

Like other home-use tests for medical purposes, the FDA requires the results to be conveyed in a way that consumers can understand and use. This is the same approach the FDA has taken with other over-the-counter consumer products such as pregnancy, cholesterol and HIV.

However, the level of comprehension required for DTC genetic tests, unlike other OTC tests, is not just a matter of whether users understand how to take the test, read the label, and interpret results. The special control for DTC genetic tests demands more of future users and is founded in a concern over whether consumers can more broadly understand genetic testing and genetics.

The FDA also requires that if the test is provided OTC, the manufacturer must provide information about how to find a qualified professional genetic counselor, as well as specific labeling about the limitations of the test based on its intended use. Lastly, for OTC tests, the following warnings must be included:

• This test is not a substitute for visits to a health care provider. It is recommended that you consult with a health care provider if you have any questions or concerns about your results.

• The test does not diagnose any health conditions. Results should be used along with other clinical information for any medical purposes.

• The laboratory may not be able to process your sample. The probability that the laboratory cannot process your saliva sample can be up to [actual probability] percentage.

• Your ethnicity may affect how your genetic health results are interpreted. (FDA 2015a)

These Class II medical devices are also exempted from premarket notification requirements. This means that companies seeking to market similar carrier screening tests can do so without filing documentation with the Agency before marketing, as long as all of the general controls for medical devices and special controls for carrier screening tests have been met. Following the launch of its Bloom’s syndrome tests, 23andMe brought back more than 35 carrier status reports that FDA had prohibited in 2013 (23andMe 2017). In the same way that the marketing approval for the Bloom’s syndrome test opened the door for the exemption of future carrier screening tests, the April 2017 marketing approval for 10 single-gene disease-related susceptibility tests has also opened the door for exemption of similar tests. As explained by 23andMe, “genetic health risk reports, by contrast [to carrier screening tests], convey personal health risk, necessitating a separate FDA review classification pathway” (23andMe 2017). The FDA granted marketing authorization through this new classification pathway for genetic health risks reports that include: alpha-1 antitrypsin deficiency – a condition that can lead to lung and liver disease; celiac disease; early-onset primary dystonia – a condition characterized by involuntary muscle contractions; factor XI deficiency, also known as hemophilia type C; G6PD deficiency, or anemia; Gaucher’s disease which can lead to an enlarged liver or spleen; hereditary hemochromatosis, which means the body can absorb too much iron; hereditary thrombophilia – a blood clotting disease; late-onset Alzheimer’s disease; and Parkinson’s disease (23andMe 2017). Notably, BRCA testing for breast and ovarian cancer susceptibility is not included in the list of authorized disease risk reports. 23andMe communicated to its customers that the company is continuing to work with the FDA for approval of additional health risk reports.

Discussion

In a recent presentation at the American Association for Clinical Chemistry (AACC) annual meeting, Gutierrez discussed the FDA’s regulatory history of DTC genetic testing and referenced the debate over the first marketed at-home pregnancy test in 1977. Drawing from a New York Times opinion piece covering the history of the test (Kennedy 2016), he used the following quotations in his presentation: “What if a senator’s daughter, unmarried, found she was pregnant and jumped off a bridge” and “Some regulators worried that ‘frightened 13-year-olds’ would be the main users of the test kits” (Gutierrez 2016). He encouraged the audience to conclude from these examples that similar concerns about how people will handle and act on genetic information will sound equally dated and unfounded, maybe even less than 40 years from now. Gutierrez concluded the presentation emphasizing the Agency’s support for DTC genetic testing when users can understand test results (Gutierrez 2016).

The tutorial and test provided by 23andMe proved effective to meet the FDA’s threshold for user understanding and reassure the Agency that 23andMe can effectively communicate genetic information. The high bar for genetic comprehension is similar to that which Church, founder of the Personal Genome Project and coincidentally a board member for 23andMe, demands of his volunteers. To volunteer to have one’s entire genome sequenced and made publicly available for research, Church requires that participants score 100% on a lengthy exam that not only evaluates one’s genetic literacy but awareness of the social, legal and ethical implications of genetic testing (Ball et al. 2014). While Church asserts his own form of governance over participating subjects, he interestingly has opposed FDA regulation on the grounds that people should have access to their raw, uninterpreted, data, but separates access to raw data from ethical concerns about the return of genetic tests results (Evans 2014). The FDA has set the comprehension bar lower than Church, but has squarely taken a position on whether the general public can understand certain categories of genetic information and how they should be educated. It is still unclear, though, whether sufficient user understanding would be enough for the FDA to allow consumers direct access to pharmacogenomics information. Given the potential therapeutic implications of such information, these tests may be too disruptive if provided outside clinical care.

To accommodate the disruptive DTC genetic testing technology within an existing regulatory pathway the FDA required 23andMe to demonstrate that consumers can understand genetic information about carrier status, and most recently, disease-related health risk reports. Given the debate about the general public’s genetic literacy, the special control to demonstrate that users score 90% or greater on a range of comprehension concepts is striking when compared to the FDA’s approach to validating user comprehension with other OTC tests. Prior to health-related DTC genetic testing, the medical profession governed clinical genetic testing. 23andMe emerged as a disruptor with a business model meant to unsettle the norms of clinical genetic testing. In the process, the firm disrupted the existing form of clinical governance aimed to protect those undergoing genetic testing. Without the mediation of a medical professional, the FDA became concerned about both the providers and consumers DTC genetic information. In the clinical model the FDA has trusted clinicians as reliable providers of genetic information grounded in both scientific and clinical knowledge. For a range of reasons, including challenges from the medical profession and Congress, the FDA has been more skeptical of the accuracy and reliability of genetic information provided by DTC firms than clinicians. Outside the clinical context the FDA has also been concerned about whether consumer can understand their genetic test results.

The recent marketing approvals for 23andMe’s tests had to address both of these concerns. 23andMe demonstrated the analytical validity and clinical validity of its tests and through its user comprehension studies also demonstrated that the company could clearly communicate genetic information to the consumer and serve as a new intermediary for the test-taking subject understanding how customers might interpret test results and establishing baseline user understanding, however, was a more complicated problem to address. The governance approach taken by the FDA, requiring demonstration of some pre-defined measures of user understanding, effectively serves as proxy for professional mediation. Standards, or special controls for comprehension stand as a substitute for the form of patient autonomy sought through informed consent. In this act of exercising its regulatory authority, the FDA mitigates risk with the logic that consumers will not be harmed by genetic information if they can understand it. The consumers of this information have the potential to achieve a form of conditional “empowerment,” because they must first demonstrate a threshold of sufficient understanding to be empowered. Just as there is tension in the clinical model that seeks to achieve patient autonomy through informed consent, there is tension in the consumer model that seeks empowerment through demonstration of sufficient understanding.

Conclusion

This paper is the first comprehensive case study to examine in detail how the FDA has set standards for the marketing approval of DTC genetic tests, which now includes carrier screening and disease-related health risk tests. This paper takes FDA’s marketing authorizations for 23andMe’s health-related genetic tests to analyze how a disruptive technology can challenge a broader system of governance, and yet be accommodated within an existing regulatory structure. The primary concerns of the FDA, which fit within a larger debate over DTC genetic testing, have been about the accuracy and validity of the tests, as well as whether consumers can understand their test results. The regulatory approach taken by the FDA and articulated through the special controls follows the logic that consumers cannot be harmed by certain kinds of genetic information if the providers demonstrate they can clearly communicate the information the consumers demonstrate that they can understand its significance. What remains to be seen, though, is whether the FDA will allow direct access to pharmacogenomics tests that achieve sufficient test validity and demonstrate high user understanding. Some categories of genetic tests may still prove too disruptive to be accommodated outside the model of clinical governance.

Acknowledgements

The author would like to thank Stuart Hogarth, Silvia Camporesi, and Smita Sarkar for their helpful input on this manuscript.

Disclosure statement

No potential conflict of interest was reported by the author.