Direct-to-consumer genetic testing

We use the term DTCGT to describe commercial genetic testing offers that are entirely organized outside the healthcare setting. In this approach, commercial providers of genetic tests advertise their products through the Internet directly to consumers, who can order the test online and have a sample collection kit delivered at home . Traditionally, proponents of this model have pointed out that DTCGT empowers individuals by increasing their access to personal genomic information . However, DTCGT has been received with vigorous criticism from both academics and professional medical organizations. It has been emphasized that although most DTCGT services lacked demonstrable analytic and/or clinical validity, ¹

1 Analytic validity refers to the accuracy with which the test can identify a particular genetic variant. Clinical validity is defined as the accuracy with which the test can predict clinical outcomes .

Physician-mediated genetic testing

Owing partly to legal challenges associated with the marketing of DTCGT services, many commercial providers have transitioned to a new model of commercial genetic testing. In this model, although genetic tests are still advertised through the Internet, they cannot be ordered directly by the consumer and must instead be requisitioned by a licensed healthcare professional . We refer to this model as physician-mediated genetic testing (PMGT). In principle, PMGT has the potential to resolve some of the ethical issues inherent to DTCGT by involving a medical professional who can guide the consumers through the testing process and assist them in interpreting test results. However, in practice, many medical professionals may lack adequate training in genetics and may be ill prepared to support patients undergoing genetic testing . Moreover, the PMGT model could give rise to conflicts of interest as commercial companies may approach medical professionals and offer them gratuities, for example, in the form of specialized training in genetics or commissions for patient referrals . Medical professionals who receive such benefits are more likely to prescribe commercial genetic tests when confronted with a request without critically assessing its adequacy or may actively recommend a genetic test to their patients, even where limited benefits are anticipated. In the most ethically problematic scenario, a healthcare professional may be hired by a commercial company to simply “sign off” on orders without ever seeing the patient . Finally, intermediation by a medical professional, even when they are comfortable with interpreting genetic information and have no conflict of interest, does not address the issue of analytical validity and clinical validity of the test . However, commercial companies typically do not disclose this information on their websites, making analytic and clinical validity difficult to assess prior to testing .

Clinic-based genetic testing

Some of the practical issues inherent to the PMGT model can be partly mitigated by integrating genetic testing services into clinical practice. In this clinic-based genetic testing (CBGT) approach, providers of genetic testing services may collaborate with medical institutions that will routinely offer testing in the clinical context. The principal advantage of CBGT, compared to the other models of genetic testing, is that in CBGT, testing is embedded in the context of clinical care and can be routinely supported by ancillary services, such as genetic counseling and, where necessary, specialized follow-up care. However, adoption of genetic testing services by a large number of medical clinics may only be possible where a genetic test has already gained substantial support among the professional medical community. An example of such a process is noninvasive prenatal testing (NIPT) for chromosomal abnormalities. The first commercial NIPT was made available in 2011 and, within the next 3 years, was adopted by a growing number of medical clinics in most developed countries . NIPT is in many ways superior to invasive chromosomal aneuploidy screening, and it is exclusively provided in the CBGT model where pregnant women have access to individualized medical supervision. However, in the absence of a standardized approach, the quality and clinical characteristics of NIPT vary across laboratories. Because most laboratories performing NIPT are operated by commercial companies, there are reports of healthcare professionals being rewarded by test providers for patient referrals . This suggests that CBGT may not be completely immune to some of the flaws of the PMGT model.

One type of health-related genetic testing that is currently available in all three models is carrier screening for recessive disorders. Therefore, in the remainder of the paper, we focus on reviewing the present landscape of carrier screening offers to better illustrate the distinct ethical and legal challenges of providing particular tests through the three models.

An overview of the genetic testing market for expanded carrier screening

In 2010, the health technology company Counsyl reported the development of a “universal carrier test” that screened for more than 100 recessive disorders , effectively launching the market for ECS. Counsyl ‘s test was first made available to consumers through the company’s website in February 2010, primarily as a DTCGT service. At the time, Counsyl ‘s website read, “You can order the test directly from our website to receive your kit immediately. Everyone has a prescription: the American College of Medical Genetics recommends that adults of reproductive age be offered carrier testing for cystic fibrosis and spinal muscular atrophy, two of the many conditions assayed by the [Test]. Alternatively, you may get the test through your doctor.” However, in May 2010, Counsyl discontinued its DTCGT offer, retaining PMGT as the primary model of provision . At the same time, Counsyl also adopted the CBGT model by beginning to collaborate with specialized medical clinics, such as fertility centers, that would routinely offer the company’s ECS test to their patients . Subsequently, several other companies started advertising their own ECS tests, almost all of which have been sold through PMGT and CBGT models. As of late 2016, PMGT and CBGT remain the dominant models of ECS test offers. In reviewing the global landscape of Internet-based ECS offers, we have identified more than 20 providers, including both commercial companies and academic laboratories, the vast majority of which provided their tests through healthcare professionals and/or medical clinics (unpublished data). In contrast, only two companies ( 23andMe and Macrogen ) used the DTCGT model. Of note, Macrogen is a South Korean company whose ECS test cannot be purchased by consumers located in most Western countries. This paucity of DTCGT offers can be explained by adverse regulatory frameworks in Europe and the US. As discussed earlier, several European countries explicitly prohibit genetic testing without medical supervision, which also applies to carrier screening. In the US, carrier screening tests for autosomal recessive disorders (i.e., excluding X-linked disorders) can be marketed directly to consumers. However, any company willing to do so must comply with special additional requirements of the FDA, such as conducting “pre- and post-test user comprehension studies to assess user ability to understand the possible results of a carrier test and their clinical meaning” . These additional requirements, along with the FDA’s traditionally active supervision of the DTCGT market, may discourage commercial providers to opt for the direct-to-consumer model. From a provider’s point of view, another disadvantage of the DTCGT model is the fact that the cost of testing in this approach is typically borne by the consumer alone, whereas genetic tests prescribed by a healthcare professional can be covered by insurance. For example, 23andMe ‘s “health + ancestry service,” which includes ECS, costs $199.00, the lowest price among all US-based ECS providers . As a comparison, Counsyl ‘s ECS test, called the “Family Prep Screen,” costs $349.00. However, because all of Counsyl ‘s tests are ordered by a physician, many consumers are eligible for partial coverage by their health insurance. Because of this, Counsyl states on its website that “The majority of our … patients pay less than $199.00 for the Family Prep Screen” . Given the regulatory challenges to obtaining the FDA’s approval and the absence of insurance coverage, it is not surprising that DTCGT offers of ECS are rare. To date, 23andMe is the only provider that has been authorized by the FDA to market its ECS test directly to consumers.

In addition to ECS services offered by commercial genetic testing companies, a growing number of medical institutions are also offering ECS tests using the CBGT model. These are primarily clinics specializing in reproductive healthcare services, such as in vitro fertilization and preimplantation genetic diagnosis. In addition, CBGT providers also include general hospitals and nonprofit community public health initiatives, particularly those organized within the Ashkenazi Jewish community. In general, CBGT providers of ECS do not have proprietary ECS tests and rely on a third party for laboratory services. However, two CBGT providers, Mount Sinai Health System (New York, US ) and Academic Medical Center , in collaboration with VU University Medical Center Amsterdam (Amsterdam, the Netherlands ), use internally developed bespoke ECS tests in clinical practice.

Clinical characteristics of ECS tests marketed to consumers

Characteristics of consumer ECS tests vary immensely across providers. For example, 23andMe currently screens for 41 autosomal recessive disorders by using targeted genotyping of known pathogenic variants. In contrast, Macrogen screens for more than 1700 recessive conditions, both autosomal and X-linked, using a whole-exome sequencing approach. Other providers fall between these two extreme examples, with most screening for between 100 and 400 autosomal and X-linked recessive disorders. These differences in the characteristics of ECS tests can be explained by the combination of the providers’ operating models and the applicable regulatory framework. For example, being an FDA-compliant DTCGT company, 23andMe must follow the requirements outlined in the Code of Federal Regulations 21CFR866.5940. The scope of this legal document is limited to autosomal recessive disorders, which prevents the inclusion of X-linked conditions in FDA-authorized ECS tests. Furthermore, the document lays out rigid criteria for the clinical validity of carrier screening tests. In particular, it states that “[C]linical validity of each variant detected and reported by the test … must be well-established in peer-reviewed journal articles, authoritative summaries of the literature … or similar summaries of valid scientific evidence, and/or professional society recommendations” . There are two implications of this requirement. First, it effectively forestalls the use of nontargeted sequencing, which may identify mutations previously unreported in the literature. Second, it discourages the inclusion of extremely rare disorders where genetic research is scarce and genotype–phenotype correlations may not be well elucidated. While these restrictions may translate into relatively low carrier detection rates, they also ensure that FDA-approved ECS tests have a very high positive predictive value. In other words, approximately 100% of couples identified as being at risk by these tests will have a one-in-four chance of conceiving an affected child in each pregnancy. In contrast, the use of nontargeted sequencing, particularly of lesser-studied recessive disorder genes, translates into low positive predictive value, meaning that the test will incorrectly identify many couples to be at risk . This is highly problematic as some of these couples will experience undue anxiety or may undergo prenatal diagnosis and selective termination of pregnancy on the basis of false information about the fetus’ affected status .

Although the majority of commercial providers are based in the US, they do not have FDA authorization, and their ECS tests do not conform to the Code of Federal Regulations 21CFR866.5940. These ECS tests typically use nontargeted sequencing, but they interrogate a relatively limited number of genes (between 100 and 400) associated with recessive disorders. In this way, commercial providers strive to achieve greater carrier detection rates without substantially lowering the positive predictive value of the ECS test. However, in the absence of clinical evidence, providers using nontargeted sequencing tend to rely on different criteria when deciding which genes to screen for and how to interpret the test results. Such differences give rise to cases where a particular mutation is reported as pathogenic by one provider, whereas others may regard it as a variant of unknown significance or a benign polymorphism . This highlights the need for better monitoring of the clinical validity of ECS consumer genetic tests in general, rather than focusing solely on the DTCGT model.

Provision of pretest information and genetic counseling

To ensure informed and voluntary participation in screening, it is essential that all consumers undergoing ECS tests receive adequate information about the purpose, potential benefits, and main limitations of the test . However, in practice, the amount and the quality of pretest information communicated to potential consumers vary greatly among ECS providers. Depending on the operating model adopted by the provider, they can choose different strategies to impart this information. For example, companies in the DTCGT may rely exclusively on using web-based communication, whereas PMGT and CBGT models may additionally incorporate pretest consultation with a medical professional or a geneticist.

In general, it is problematic where the decision to purchase a genetic test is based solely on the information presented on the website of a genetic testing company because such websites tend to emphasize the benefits of testing and downplay its disadvantages . However, in the case of 23andMe ‘s ECS service, it is encouraging that the company provides high-quality information to its consumers, such as an extensive description of its test and sample reports for the disorders being screened . Using sample reports may be particularly helpful, as shown by 23andMe ‘s study that found more than 90% user comprehension rates of these reports . Furthermore, the company recommends its potential consumers to discuss carrier screening with their healthcare provider and/or a genetic counselor and provides a link to the National Society of Genetic Counselors . However, it is unclear whether consumers purchasing 23andMe ‘s carrier screening service actually take time to familiarize themselves with this information or seek a medical consultation if in doubt. What is particularly problematic in the context of providing adequate pretest information is the fact that 23andMe offers its ECS service in combination with ancestry and health testing, rather than as a standalone test . Research with users of DTCGT services has revealed that some consumers find it difficult to grasp the difference between carrier status and being affected by a disorder . Therefore, it is likely that combining carrier screening with other types of genetic tests in a single service will increase the potential confusion of the customer over the purpose and expected outcomes of carrier screening.

Achieving informed participation in carrier screening also remains a challenge where the test is ordered through a healthcare professional, as in the PMGT model. In particular, where the physician’s primary function is to approve consumer orders, involvement of a physician may be little more than a formality. An example of this is Good Start Genetics’ carrier screening test VeriYou, launched in October 2016 . Although by screening for only two disorders (cystic fibrosis and spinal muscular atrophy) VeriYou should not be considered an ECS test as such, we choose to discuss the product here as VeriYou is one of the few genetic tests to be offered through amazon.com and is accessible to potential consumers through the company’s website . This, for all practical purposes, makes VeriYou a DTCGT service. However, consumers do not directly order the test. Instead, they request it online by providing information on the company’s website. Consumer requests are then reviewed and approved by Good Start Genetics’ licensed physician, who orders the test on behalf of the consumer. Although the role of the licensed physician is largely ceremonial, technically this makes VeriYou a PMGT test. Although the case of VeriYou is exceptional in that it has clear characteristics of a DTCGT offer, employing licensed healthcare professionals to validate test orders is not unique to this product, and several other commercial providers of carrier screening services facilitate consumers’ access to the test through a healthcare professional employed by the company. One such provider, Baby Genes , which offers ECS for 71 autosomal recessive disorders, displays the following information on their website: “All Baby Genes tests are ordered by a physician—either your own or one of ours who will review your information and can order testing on your behalf.” The possibility of accessing the test through a company-affiliated physician raises the concern that, in practice, the primary function of such a physician may be approving test orders, rather than addressing the pretest information needs of potential consumers. Involvement of a healthcare professional could be viewed as the company’s strategy to avoid being subjected to rigorous regulatory supervision while, for all practical purposes, selling testing services through the DTCGT model . In addition, in PMGT, some potential consumers may choose to address their own healthcare provider for ordering a commercial ECS test. Although these healthcare providers are less likely to have a conflict of interest, many of them may not be prepared to offer qualified advice regarding ECS to their patients. For example, one study in the US found that more than one-third of reproductive healthcare professionals did not know the probability of passing down a recessive disease allele to one’s offspring and more than 40% could not accurately estimate the likelihood of having an affected child when both parents are carriers . Advising patients on the appropriateness of ECS will require an in-depth understanding of medical genetics on the part of healthcare providers. Where this knowledge is lacking, ordering an ECS test through a healthcare professional is unlikely to improve informed decision-making among consumers.

Embedding ECS in the clinical setting, in accordance with the CBGT model, may be the most effective strategy to enhance potential consumers’ understanding of the testing process. This is primarily because CBGT typically entails a pretest consultation with a geneticist or a medical professional with expertise in clinical genetics. However, the extent to which such a consultation is unbiased will depend on whether the medical professional has an incentive to encourage their patients to take the test. As suggested by the experience with NIPT, conflicts of interest among practicing clinicians are not uncommon , which has to be considered when implementing ECS through the CBGT model.

Reporting of results and post-test counseling

Similar to other aspects of ECS, practices regarding the reporting of results and/or post-test genetic counseling vary considerably across ECS providers. For example, 23andMe sends individual written reports to its consumers without post-test genetic counseling. These reports are limited to describing heterozygosity for autosomal recessive disorders screened for by the 23andMe ‘s panel. Given the >90% comprehension rate of 23andMe ‘s test reports among consumers , their strategy of issuing written reports, without face-to-face genetic counseling, may be sufficient for communicating results in most cases. However, this may prove inadequate for informing at-risk couples, where post-test genetic counseling is strongly recommended to assist them in reproductive decision-making . As 23andMe offers its ECS test to individuals, it may be challenging to ensure that all carrier couples are appropriately followed up and receive post-test genetic counseling. This concern is also present in the PMGT model, where all companies make their tests available to individual consumers. However, some commercial providers, such as Counsyl and Natera , offer complimentary post-test genetic counseling services to all individual consumers (unpublished data). Although these services are discretionary and, typically, should be explicitly requested by the consumer, their availability nevertheless makes it more likely that at-risk couples receive adequate counseling and follow-up. In the context of CBGT, because post-test genetic counseling is fully integrated into the clinical practice of medical clinics offering ECS, at-risk couples routinely receive in-person counseling and follow-up (unpublished data).

In addition to different approaches to post-test counseling, ECS providers also differ in terms of reporting information to their consumers. For example, 23andMe ‘s carrier screening reports exclude information relevant to the screened individual’s own health that may be incidentally identified in the process of screening. The test does not report if the individual has two pathogenic variants associated with the same disorder and are, therefore, likely to be affected themselves . This is typically not the case with other ECS providers that are not subject to the same regulation as 23andMe, and they typically communicate incidentally identified health-related information to consumers. However, such providers usually allow consumers to opt out of receiving health-related data (unpublished data). Those ECS providers that use nontargeted sequencing and in-house variant interpretation approaches typically limit the reporting of results to the variants categorized as clearly or likely pathogenic. However, some providers may also report variants of unknown significance (VUS) under certain circumstances. For example, Baby Genes communicates the findings of “VUS that result in nonsynonymous protein changes with no known clinical association” . These discrepancies across providers indicate the need for using harmonized criteria for interpreting and reporting variants.