ELIZABETH R. PIKE AND KAYTE SPECTOR-BAGDADY
THE 2007 launches of 23andMe and deCODEme heralded the beginning of direct-to-consumer (DTC) access to personalized genetic medical information (Kalf et al. 2014). Although DTC companies originally provided targeted testing for discrete disorders such as Huntington’s disease, DTC companies now offer large-scale and whole genome (genomic) sequencing that provide information about genetic predispositions to disease and pharmacogenomic responses to particular drugs. This broader inquiry raises novel concerns about how best to regulate the collected information.
The U.S. Food and Drug Administration (FDA) began direct engagement with the DTC genetic testing industry in 2010, issuing 23 “Untitled Letters” to companies that suggested that DTC offerings were medical devices that needed to be registered, cleared, or approved by FDA (2014b). In response, many companies began requiring a physician intermediary while some exited the market entirely. In 2013, FDA sent a more formal “Warning Letter” to industry leader 23andMe detailing its violations of the Federal Food, Drug, and Cosmetic Act and demanding that 23andMe cease marketing its health-related genomic information (Gutierrez to Wojcicki 2013).
In the wake of FDA’s enforcement actions, some DTC companies bifurcated into entities that produce a data-only product—a file of As, Ts, Cs, and Gs without interpretation—and entities that produce an information-only product, interpreting data to produce analysis regarding genetic predispositions such as to particular diseases. Because data-only offerings fall outside FDA’s definition of a device, they are unlikely to be regulated by FDA. Producing genomic information, however, falls squarely within FDA’s definition of a device and raises novel questions about how this product should be regulated.
This chapter describes the challenges that FDA faces in regulating genomic information as a medical device, including classifying these products according to levels of risk and implementing standard risk-mitigation strategies. Ensuring the safety and effectiveness of genomic information requires ensuring analytic and clinical validity, as well as minimizing risks of disclosure. This chapter proposes a path forward for satisfying these requirements.
I. EVOLUTION OF DIRECT-TO-CONSUMER GENETIC TESTING
Since the start of the Human Genome Project in 1990, knowledge of the relationship between the human genome and physical traits and predispositions to diseases has greatly expanded. In 1990, genetic tests were available for only 100 genes specifically associated with disease (Brower 2010). By April 2012, there were tests for 2,600 disease-causing variants (Palmer 2012).
Although once the exclusive province of clinicians and researchers, companies increasingly offer genetic testing directly to consumers without clinician involvement (Hudson et al. 2007). Comparatively inexpensive DTC tests predict carrier status for diseases like cystic fibrosis; the probability of developing disease in the future, such as breast cancer or Alzheimer’s disease; the ability to metabolize certain drugs, such as warfarin (a blood thinner); or traits as mundane as excessive earwax (Bair 2012; Beaudet and Javitt 2010; McGuire et al. 2010). Increasingly, however, DTC companies are moving from discrete genetic tests like the examples above to larger-scale genomic sequencing. In 2008, for example, Knome first offered DTC genomic sequencing for $350,000 per person (Singer 2008).1 The cost of genomic sequencing has since come down considerably with Illumina announcing the first “$1,000 genome” in January 2014 (Winnick 2014).
Although the expanded scope of DTC services offers benefits to consumers—potentially including greater autonomy and empowerment (Brower 2010; Frueh et al. 2011; Hudson et al. 2007)—expanded DTC access also raises concerns. These concerns include that consumers might get tested “without adequate context or counseling, will receive tests from laboratories of dubious quality, and will be misled by unproven claims of benefit” (Hudson et al. 2007:1392). Moreover, consumers might make “unwarranted, and even irrevocable, decisions on the basis of test results and associated information, such as the decision to terminate a pregnancy, to forgo needed treatment, or to pursue unproven therapies” (Hudson et al. 2007:1393–4).
Some of these concerns have merit. Reviews conducted by the U.S. Government and Accountability Office (GAO) in 2006 (GAO 2006) and 2010 (GAO 2010) raised serious questions about the accuracy and reliability of information being returned to consumers. A 2008 report by the Secretary’s Advisory Committee on Genetics, Health, and Society (SACGHS) also noted gaps in the regulatory oversight of these products (SACGHS 2008). Addressing some of these concerns requires additional regulatory oversight by FDA.
II. CURRENT REGULATION OF DIRECT-TO-CONSUMER GENETIC TESTING
FDA protects public health by assuring the “safety, effectiveness, quality, and security” of medical interventions such as drugs and medical devices (FDA 2014). FDA defines medical devices as any “instrument, apparatus, implement, [or] machine” that is “intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease…” (Federal Food, Drug, and Cosmetic Act (FDCA), 21 USC § 321h). Medical devices that pose risk to patients or consumers generally must be cleared or approved by FDA before they can be marketed or distributed (FDCA, 21 USC § 360e; 21 USC § 807.81). For potential violations, FDA often begins enforcement with either a Warning Letter, which highlights violations that may lead to enforcement action such as a recall or seizure of products, or an Untitled Letter for less significant violations (FDA 2014c).
In 2010, FDA sent 23 Untitled Letters to companies in the DTC genetic-testing industry, notifying them that their products were considered medical devices and therefore needed to be classified and registered, cleared, or approved, as appropriate (Spector-Bagdady and Pike 2014). For example, the letter to deCODE Genetics argued that despite deCODE never having submitted any analytic or clinical validity data for its tests, “your website states that the deCODEme Complete Scan identified twelve common genetic variants and provides interpretation of their associated risk for developing breast cancer…[and that] consumers may make medical decisions in reliance on this information” (Gutierrez to Collier 2010). These Untitled Letters resulted in many recipients exiting the DTC market. One exception, 23andMe, responded by beginning the FDA clearance process for seven of its then-254 tests (Perrone 2012).
But on November 22, 2013, FDA issued a Warning Letter to 23andMe noting that it “failed to provide adequate information” in support of its devices and was therefore ordered to “immediately discontinue marketing” its product. In addition, FDA stated that the entire personal genome service would be considered a Class III device—warranting the highest level of regulatory oversight (Gutierrez to Wojcicki 2013). As a result, 23andMe began selling individuals their genetic data (the file of As, Ts, Cs, and Gs) and ancestry information but stopped providing health-related genetic information (“Status of our health-related genetic reports”). In February 2015, FDA authorized 23andMe to market a DTC test to determine whether a consumer is a carrier for Bloom Syndrome, a rare autosomal recessive disorder. In its press release, FDA also announced that it intends to exempt similar carrier screening tests from premarket review in the future (FDA 2015). In response, 23andMe noted that it would not immediately begin returning Bloom carrier status testing results or other health results, instead preferring to wait until it could offer a more comprehensive product under the new FDA guidance (“23andMe Granted Authorization by FDA to Market First Direct-to-Consumer Genetic Test Under Regulatory Pathway for Novel Devices.”).
III. BIFURCATED DIRECT-TO-CONSUMER GENETIC TESTING ENTITIES
In response to the changing regulatory landscape, some DTC companies bifurcated into entities that produce either a data-only or information-only product (Vorhaus 2012). For example, in November 2012, Gene By Gene announced “DNA DTC” (“23andMe, Inc. Provides Update on FDA Regulatory Review”; “Gene By Gene Launches DNA DTC”), a DTC sequencing service that provides raw genomic data without interpretation. DNA DTC marked the first “truly direct-to-consumer” offering of such a data-only product (Vorhaus 2012).
Other companies have made the parallel offering of information-only services. For example, openSNP, an open-source project, allows consumers to input raw genomic data and receive medical information; openSNP is compatible with raw data from 23andMe, deCODEme, or FamilyTreeDNA (“Welcome to openSNP”). Promethease offers another free basic report and an enhanced report for five dollars (“Promethease”; “Promethease/Features”).
FDA’s approach to regulating these bifurcated entities is still somewhat unsettled. For data-only products—the files of As, Ts, Cs, and Gs—FDA’s approach is straightforward: because genomic data does not itself diagnose, cure, mitigate, treat, or prevent disease, to the extent it is marketed and sold as a stand-alone product, it falls outside FDA’s definition of a medical device and therefore outside its regulatory jurisdiction (Carmichael 2010). Officials at FDA have acknowledged that for companies that produce a data-only product, “[i]f they don’t make any medical claims about that data, then they’re free to provide information [without approval] as far as we’re concerned” (Carmichael 2010).
To be used in the diagnosis, cure, mitigation, treatment, or prevention of disease, data-only products require a parallel information offering—which is of concern to FDA. Indeed, FDA sent several 2010 Untitled Letters to “software program[s] that analyz[e] genetic test results created by an external laboratory” (FDA 2014b) because even though these companies were not producing genetic data from biological samples, they were producing the medical information marketed for the diagnosis, cure, mitigation, treatment, or prevention of disease. This reasoning would similarly apply to information-only products such as Promethease (Palmer 2012).
Attempting to regulate information-only services raises novel questions about FDA’s regulatory authority that are beyond the scope of this chapter but have been discussed elsewhere (Spector-Bagdady and Pike 2014).2 The remainder of this article therefore focuses on the challenges FDA faces in regulating genomic information as a medical device, in particular, the challenges of classifying genomic information on the basis of perceived risk and the limited applicability of standard risk-mitigation strategies.
IV. REGULATING GENOMIC INFORMATION AS A MEDICAL DEVICE
As discussed previously, a medical device is a product “intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease” (Federal Food, Drug, and Cosmetic Act (FDCA), 21 USC § 321h). Because data-only products that return the list of As, Ts, Cs, and Gs do not themselves cure, mitigate, treat, or prevent disease, they generally fall outside FDA’s definition of a device. Genomic information, by contrast, can be used in the diagnosis, mitigation, or perhaps even prevention of disease, and so the process of producing it falls squarely within FDA’s jurisdiction. Regulating genomic information as a medical device, however, poses serious challenges for FDA. In particular, FDA will have difficulty classifying genomic information on the basis of perceived risk and successfully implementing common risk mitigation strategies.
A. Risk Classification of Genomic Information
Risk classification is unwieldy as applied to bifurcated genomic information. FDA classifies medical devices into three categories—Class I, II, and III—on the basis of risk to the consumer. The greater the risk to the consumer, and the more control presumed necessary to assure the safety and effectiveness of the device, the higher the classification a device receives (FDA 2012).
Class I devices are assumed to pose low to moderate risk and are subject to “General Controls,” including registration and labeling requirements. Devices are classified as Class II if general controls alone are perceived to be “insufficient to provide reasonable assurance of…safety and effectiveness” (21 CFR § 860.3(c)). Class II devices are subject to both premarket notification and “Special Controls,” including stricter labeling requirements and/or postmarket surveillance (FDA 2012; FDA 2013). If there is insufficient evidence that Special Controls provide assurance of safety and effectiveness, FDA will consider a device Class III, thereby requiring rigorous “premarket approval,” including “sufficient valid scientific evidence to assure that the device is safe and effective for its intended use(s)” (21 CFR § 860.3(c); FDA 2012a). If FDA is particularly concerned about a device, the Secretary of Health and Human Services may restrict a Class III device to sale, distribution, or use “only upon written or oral authorization of a practitioner licensed by law to administer or use such device” (FDCA, 21 USC § 360j(e)(1)).
Assessing the risks posed by genomic information is challenging because there are a number of different types of risks to consider. In addition to the risk inherent to the potential underlying disease, risks also include that a sample might be inaccurately sequenced (thereby lacking analytic validity); the sequence might be inaccurately interpreted (thereby lacking clinical validity); the interpretation might be inaccurately communicated; and that recipients might take actions in response to the information that are detrimental to their health (Gniady 2008). In the 1980s, similar concerns arose regarding DTC HIV tests. FDA initially concluded that the risks of DTC HIV tests were so great that they would be approved only for use by professionals within a health care environment and with appropriate counseling (54 FR 7279). FDA later concluded that DTC HIV tests “may be approvable” provided that all technical aspects of the test were validated, that instructions and educational materials were comprehensible to a lay audience, that adequate pre- and post-test counseling were provided, and that consumers would be able to take appropriate follow-up action (60 FR10087).
Like DTC HIV tests, one risk of genomic information stems from actions a consumer might take in response to the information received (Gniady 2008). Controlling genomic information on the basis of potential actions is unusual, although not unprecedented. With DTC HIV tests, the concern was that positive results would trigger widespread suicide and panic, which could not have been mediated by clinical professionals who had previously been the bearers of such news (Gniady 2008). By contrast, the most concerning medical actions discussed in the context of DTC genomic information—such as a mastectomy upon learning about a predisposition to breast cancer—require clinician involvement (Frueh et al. 2011; Palmer 2012). Some responses, however, such as a decision to self-medicate or dose in response to pharmacogenomic predispositions, could fall outside the influence of clinicians thereby raising similar concerns (Gutierrez to Wojcicki 2013).
Assessing the risk of returning genomic information is also more complex than assessing the risk of DTC HIV testing because many different types of results can arise from one test. In 2010, an FDA official stated that a test for a benign trait, such as baldness, would be considered Class I—if considered a device at all (Rubin 2010). Meanwhile, 23andMe’s Warning Letter stated that “FDA has spent significant time evaluating…whether certain uses might be appropriately classified into [C]lass II” before ultimately concluding that, without additional information, FDA was going to consider 23andMe’s entire product Class III (Gutierrez to Wojcicki 2013).
For information-only products that return a range of assessments from breast cancer to earwax type, assessing the level of perceived risk is difficult. FDA could classify an entire information-only product on the basis of the riskiest piece of information returned—a “weakest-link” approach to classification. In this scenario, information-only products that return genomic information about both baldness and BRCA mutations could potentially be considered Class III. With this approach, if at least one Class III piece of information is returned, the entire information-only product would be considered Class III.
Alternatively, FDA could assign each type of genomic information being returned to a class. Bundled products would therefore require independent assessments of each variant being returned. For example, information about BRCA mutations could be Class III (with its attendant premarket approval requirements and risk mitigation strategies), while information provided by the same device about baldness could be considered Class I. But an approach requiring individualized assessments of each potential piece of genomic information being returned—including information about analytic and clinical validity, along with information about potential responses to this information—would be time consuming to implement (Beaudet and Javitt 2010) and might result in a stark disparity between the most advanced genetic information available clinically versus DTC.
B. Risk Mitigation Strategies for Genomic Information
Should genomic information be classified as a Class II or Class III device, as appears likely, FDA faces additional challenges in implementing risk mitigation strategies sufficient to ensure its safety and effectiveness.
For Class II devices, pertinent special controls include labeling and postmarket review. If genomic information is classified as Class III, FDA can implement “[s]uch other requirements as FDA determines are necessary…” (21 CFR § 814.82(a)(9)), which may include counseling or, if a product is a restricted device, mandated clinician involvement. But each of these controls or conditions has particular challenges when it comes to providing additional assurances of the safety and effectiveness of genomic information.
One control FDA has over regulated medical devices is restrictions regarding the device’s “labeling,” which cannot be “false or misleading in any particular” (FDA 2009). Moreover, the “labeling must bear adequate directions for use and any warnings needed to ensure the safe and effective [ ] use of the device” (FDA 2009). In vitro diagnostics—a category of devices into which DTC genetic testing falls—must include “any warnings…appropriate to the hazard presented by the product” (21 CFR § 809.10(a)(4)). This special control seems particularly challenging to implement for information-only products, in part, because it would be impossible to fully disclose the risks of all possible results being returned given the large number of potentially returnable results and the rapidity with which the science is advancing (when combined with the time-consuming nature of the clearance/approval process). The alternative—making a general disclosure about the range of risks that can arise from any of the tests—might be too general to be effective.
Should genomic information be classified as at least a Class II medical device, FDA can mandate postmarket review, generally used when premarket review cannot adequately assure FDA of safety and effectiveness (FDA 2012; 2013). However it would be difficult to craft postmarket review that provides additional assurances of safety and effectiveness. This is true in part because the effectiveness of genomic information is difficult to measure (Palmer 2012); recent research has, however, found that “neither the health benefits envisioned by DTC-[genetic testing] proponents…nor the worst fears expressed by its critics…have materialized to date” (Roberts and Ostergren 2013:182).
Should genomic information be classified as a Class III device, FDA could require counseling for tests that “‘have the potential to cause distress’” (Brower 2010:1612). FDA took this approach for DTC home HIV tests, which were finally approved as a Class III device requiring anonymous over-the-phone counseling (Palmer 2012). Some DTC genetic companies already offer counseling, but this service is generally available only after an individual purchases the service. Counseling therefore might be unavailable to assist in the decision of whether to purchase testing in the first place (Bair 2012). A requirement that DTC companies provide counseling would likely make DTC genetic testing more expensive and time consuming—both of which will limit its accessibility. And consumers might balk at mandated counseling, particularly if they choose DTC services specifically to learn medical information independently from clinical professionals.
FDA could also determine that some information—for example, mutations associated with risk for breast cancer or Huntington’s disease status—has such potential to cause distress that it should be a restricted Class III device, which must be prescribed through a clinician intermediary. For example, in its Warning Letter to 23andMe, FDA pointed out that some of its concerns were “typically mitigated by…management under a physician’s care” (Gutierrez to Wojcicki 2013). A requirement that clinicians be involved in the communication of genomic information would effectively end the DTC genetic testing industry. It is not clear, however, that clinician intermediaries make the communication of genomic information more effective or resulting patient response more beneficial to their health. Genomic interpretation is complicated, and most clinicians lack the training to feel comfortable returning genomic information (Beaudet and Javitt 2010; Mardis 2010; Palmer 2012). Moreover, a clinician’s involvement is often already required for the most concerning types of therapies resulting from genetic test results. For example, while FDA has cited unfounded prophylactic mastectomies as a driving concern for regulating the DTC genetic testing industry (Gutierrez to Wojcicki 2013), such surgical intervention would require physician involvement (Palmer 2012).
C. A Path Forward for Regulating Genomic Information
Three steps are necessary to ensure the safety and effectiveness of DTC genomic information. First, the underlying data must be analytically valid—that is, the genomic data sequence must be accurate and precise. Second, the information must be clinically valid—the findings must be causally associated with clinical outcomes. And third, the risks of disclosing the genomic information must be minimized. FDA’s ability to effectively regulate genomic information hinges upon the approach taken to each of these challenges.
The first step, ensuring the analytic validity of the underlying data, applies primarily to data-producing products. Although not directly applicable to information-only products, analytic validity is a necessary predicate to ensuring that information-only products are safe and effective. FDA could rely on the Clinical Laboratory Improvements Amendments (CLIA) of 1988, which help “ensure the accuracy, reliability, and timeliness of patient test results…” (CLIA). Some DTC genetic testing companies, such as 23andMe, already produce data in a CLIA-certified laboratory. If genomic information is classified as a Class III device, FDA could require as a condition of premarket approval that information-only products rely on data sequenced in a CLIA-compliant manner.
The second step, ensuring clinical validity, requires regular review of evolving scientific literature. Reviewing the scientific literature and assessing the clinical validity of hundreds of variants being returned for each submission is incredibly labor intensive and, in a fast-moving scientific field, is likely to evolve as rapidly as the review can progress. Companies seeking approval can shoulder the burden of providing assurance of clinical validity, but FDA would still ultimately have to clear or approve these claims. FDA could therefore focus on classifying and clearing tests of particular concern—including those for breast and ovarian cancer, Huntington’s disease, Alzheimer’s disease, and critical drug responses—while leaving less-risky interpretation accessible DTC in the interim—ensuring continued access to this more innocuous information.3
The third step, ensuring the safety and effectiveness of genomic information by minimizing the risks of disclosure, can be accomplished through the use of special controls and conditions of premarket approval. Although several possible controls become conceptually unwieldy when applied to DTC genomic information (e.g., seizure of product), effective labeling requirements could help ensure the safety and effectiveness of information-only products. Even if labeling requirements fall short in ensuring safety and effectiveness, there is some reassurance in the fact that many of the most concerning actions taken in response to genomic information require clinician involvement, providing an additional safety mechanism.
V. CONCLUSION
FDA’s recent engagement with DTC genetic testing companies has brought major changes to this nascent industry. A result of this uncertain regulatory environment is bifurcated DTC genetic testing entities: entities that provide either a data-only product—a file of As, Ts, Cs, and Gs without any interpretation—or an information-only product, interpreting this genomic data and providing information about genetic risks and pharmacogenomic interactions. Although data-only products generally fall outside FDA’s regulatory purview, regulating genomic information as a medical device will pose serious challenges.
FDA’s device-regulation structure is based on perceived risk, an assessment that is particularly difficult for the wide array of genomic information that can be returned. In addition, typical risk mitigation strategies—including labeling, postmarket review, counseling, and mandated clinician involvement—will be more complicated to implement for information-only products.
Ensuring the safety and effectiveness of genomic information as a medical device requires ensuring its analytic and clinical validity and minimizing risks of disclosure. To do so, FDA could rely on CLIA to ensure that genomic information is based on analytically valid genomic data. FDA could also focus on validating tests for the riskiest types of information. FDA could minimize the risks of disclosure through labeling requirements while also being reassured that the most concerning therapeutic actions taken in response to genomic information often require clinician involvement. This approach allows continued access to important medical information while enabling FDA to meet its public health mission of ensuring the safety and effectiveness of this new technology.
The findings and conclusions in this chapter are those of the authors and do not necessarily represent the official position of the Presidential Commission for the Study of Bioethical Issues or the Department of Health and Human Services. The use of trade names and commercial sources in this report is for identification only and does not imply endorsement by the Presidential Commission for the Study of Bioethical Issues or the Department of Health and Human Services. The authors would like to thank Lisa M. Lee and Paul Lombardo for their thoughtful review of this article and Tenny R. Zhang for her editorial assistance.
1. Knome no longer offers this service. In 2010, Knome changed its model and currently markets to medical institutions, researchers, and pharmaceutical companies (Palmer 2012).
2. For example, FDA’s ability to regulate the labeling of medical devices derives from the government’s ability to regulate commercial speech, but some of the newest genomic interpretation platforms neither sell their services nor advertise their websites. Without any commercial transaction, it is not clear that FDA has the authority to regulate these services. Moreover, unlike other more typical devices, the device here is entirely speech—or information. These services therefore touch upon FDA’s ability to regulate true speech, an issue that has recently made its way through the courts (Spector-Bagdady and Pike 2014).
3. This argument was originally made in Spector-Bagdady, K. and E. R. Pike. 2014. “Consuming Genomics: Regulating Direct-to-Consumer Genetic and Genomic Information.” Nebraska Law Review 92:677–745, but this article offers an expanded analysis and assessment of risk classification and risk mitigation for genomic information.
REFERENCES
Bair, S. 2012. “Direct-to-Consumer Genetic Testing: Learning from the Past and Looking Toward the Future.” Food and Drug Law Journal 67:413–33.
Beaudet, A. L. and G. Javitt. 2010. “Which Way for Genetic-Test Regulation?” Nature 466:816–8.
Brower, V. 2010. “FDA to Regulate Direct-to-Consumer Genetic Tests.” Journal of the National Cancer Institute 102:1610–12, 1617.
Frueh, F. W. et al. 2011. “The Future of Direct-to-Consumer Clinical Genetic Tests.” Nature Review Genetics 12:511–15.
Gniady, J.A. 2008. “Regulating Direct-to-Consumer Genetic Testing: Protecting the Consumer Without Quashing a Medical Revolution.” Fordham Law Review 76:2429–75.
Government Accountability Office (GAO). 2010. Direct-to-Consumer Genetic Tests: Misleading Test Results Are Further Complicated by Deceptive Marketing and Other Questionable Practices. Washington, DC: GAO.
——. 2006. Nutrigenetic Testing: Tests Purchased from Four Web Sites Mislead Consumers. Washington, DC: GAO.
Hudson, K. et al. 2007. “ASHG Statement on Direct-to-Consumer Genetic Testing in the United States.” Obstetrics and Gynecology 110:1392–5.
Kalf, R. R. J. et al. 2014. “Variations in Predicted Risks in Personal Genome Testing for Common Complex Diseases.” Genetics in Medicine 16:85–91.
Mardis, E. R. 2010. “The $1,000 Genome, the $100,000 Analysis?” Genome Medicine 2:84.
McGuire, A. L. et al. 2010. “Regulating Direct-to-Consumer Personal Genome Testing.” Science 330:181–2.
Palmer, J. E. 2012. “Genetic Gatekeepers: Regulating Direct-to-Consumer Genomic Services in an Era of Participatory Medicine.” Food and Drug Law Journal 67:475–522.
Roberts, J. S. and J. Ostergren. 2013. “Direct-to-Consumer Genetic Testing and Personal Genomics Services: A Review of Recent Empirical Studies.” Current Genetic Medicine Reports 1:182–200.
Spector-Bagdady, K. and E. R. Pike. 2014. “Consuming Genomics: Regulating Direct-to-Consumer Genetic and Genomic Information.” Nebraska Law Review 92:677–745.
