7

Considerations in the Return of Genetic Results: Lessons from Other Research Studies (Session 6)

This chapter summarizes the presentations and the discussion in Session 6, which took place on December 8, 2022. For this session, the planning committee sought out studies that have grappled with the decision of returning genomic results to participants. Session speakers were asked to present on the lessons learned, with a particular focus on the decision-making process and rationales for the final decisions. Leslie G. Biesecker, National Human Genome Research Institute at the National Institutes of Health, moderated the session. Questions and discussion followed at the end of the session.¹ The planning committee chair’s recap of the workshop thus far, which took place just before Session 6, is included separately in Chapter 8.

GENOMICS ENGLAND

David Bick, principal clinician for the Newborn Genomes program that is part of Genomics England,² outlined the two core functions of Genomics England: to support the use of genomics in health care and

___________________

¹ Video recordings of the presentations and discussions, along with copies of the presenters’ slides for Session 6, are available online at https://www.nationalacademies.org/event/12-08-2022/workshop-on-considerations-for-returning-individual-genomic-results-from-population-based-surveys-focus-on-the-national-health-and-nutrition-examination-survey-day-3-virtual

² Bick explained that Genomics England is a company that is “wholly owned” by the National Health Service (NHS), which is responsible for providing health care to everyone in England.

provide genomic data to researchers to accelerate genomic discovery. He explained that in 2019 a workgroup was compiled to determine if whole-genome sequencing could be used as an adjunct to the newborn screening program. The public were engaged in 2021 and felt “very strongly” that it was appropriate to look for treatable genetic conditions that probably need to be treated in the first few years, which led to the government funding the program. Bick noted that the current National Health Service (NHS) Newborn Blood Spot Screening Programme screens for nine conditions using a bloodspot test (compared with 50 conditions in the United States), and Genomics England offered an opportunity to explore the merits of taking a broader approach in the United Kingdom (UK) by looking at the baby’s genome. He shared that an estimated nine children a day are born with a treatable condition. Bick said, “newborn screening, as it’s currently envisioned and currently undertaken in the UK, is very widely accepted by patients in this country.” Thus, a pilot research study comprised of the following three parts is currently under way:

1. Evaluating the utility and feasibility of screening newborns for a larger number of childhood-onset rare genetic conditions in the NHS using whole-genome sequencing
2. Understanding how babies’ genomic data could be used for discovery research, focusing on developing new treatments and diagnostics for NHS patients
3. Exploring the potential risks, benefits, and broader implications of storing a baby’s genome over their lifetime

Bick outlined five reasons Genomics England has decided to use whole-genome sequencing. First, it allows the program to detect all of the disease-causing variations (e.g., single nucleotide variants [SNVs], copy number changes, translocations) using a single technology. Another benefit of the approach is that it is flexible and conditions could be added over time. Bick went on to explain that whole-genome sequencing is “future-proofed” in a way that whole-exome sequencing is not, such that “children recruited from the beginning of the program will have just the same advantage as children later on.” Whole-genome sequencing is better suited to support research discovery activities, including research into pharmacogenetic variants, which are common but not present in the coding part of genes. Finally, whole-genome sequencing is sensible because it allows for rapid access to inform clinical care in the future.

Bick next shared that Genomics England is funded to do whole-genome sequencing of 100,000 newborns and will start toward the end of 2023. A conditions framework workgroup composed of physicians, scientists, researchers, the public, individuals from genetic support groups, and NHS

staff was created to decide how specific genes would be selected. Going beyond the Wilson and Junger criteria, Bick explained that the workgroup developed four core principles as follows, each of which must be met in order to consider adding that gene to the pilot program:³

1. There is strong evidence that the genetic variant(s) causes the condition and can be reliably detected. Where appropriate, there may be a confirmatory test that can establish whether the child has the condition.
2. A high proportion of individuals who have the genetic variant(s) would be expected to have symptoms that would have a debilitating impact on quality of life if left undiagnosed.
3. Early or presymptomatic intervention for the condition has been shown to lead to substantially improved outcomes in children, compared with intervention after the onset of symptoms.
4. Conditions screened for are only those for which the interventions are equitably accessible for all.

Bick explained that in addition to looking at particular genes and their associated conditions, Genomics England will only include variants with high positive predictive values (PPVs) and only pathogenic or likely pathogenic findings will be reported. He shared that the website rx-genes.com is one it is using to look for or choose genes; it currently lists more than 700 conditions that are treatable in childhood. Bick and colleagues expect that 15 percent of the 4,684 genes associated with phenotype-causing variants that are listed in the Online Inheritance in Man will have a treatment directed against the disease mechanism.⁴

Bick used Figure 7-1 to outline the four components of the consent model (left-hand side of figure) that were ultimately established, with the right-hand side of the figure showing the inputs and considerations that were used to derive the consent model.

Outlining the overall timeline for recruitment and consent, Bick explained that they will begin making prospective parents aware of the program around week 20 of gestation, with the goal of having consent by the 36-week mark. During the intervening weeks, parents will have “a variety of means” for making their decision, from visiting the website to calling the program directly. Then, at birth the consented parents are reminded of their participation in the study. The sample is collected at birth with results

___________________

³ Genomics England. (2022). Choosing conditions, genes, and variants: Four guiding principles. https://files.genomicsengland.co.uk/documents/Newborns/Four-principles-one-page-summary.pdf. Also see https://www.genomicsengland.co.uk/initiatives/newborns/choosing-conditions

⁴ See https://www.omim.org/

returned within 2 weeks of testing. Bick noted that this timeline and process were developed using input from women and families. He explained that each baby would have a structured care and treatment pathway as part of the program, and that a group has been convened “to ensure that any condition that is identified has a treatment that is available throughout the country.”

Bick spoke next about the program’s commitment to understanding both the services and resources required to support children and families, and the education and training needs for the workforce to provide high-quality care. “Understanding needs of the NHS workforce is crucial to ensure we can adopt an optimal approach and take its concerns into account,” said Bick. He outlined six workforce input activities the program is undertaking as follows:

1. Codesigning with midwives with a focus on recruitment, consent, and samples.
2. Working with clinicians across a range of specialisms to establish how the study would affect clinical care.
3. Discussing with nurses to understand approaches to care and treatment pathways.
4. Engaging with genetic counselors and genetic clinical and laboratory services.
5. Education and training working group with representation from various NHS workforce groups to establish learning and resource needs.
6. Liaising with Royal Colleges to understand workforce needs and capacity.

Bick noted that, unlike the typical newborn screening (which needs to be done a few days after birth so that metabolites can build up in the blood), the DNA sample collection can happen on day 0 (zero). He shared that a study is under way to determine what type of sample should be collected: dried blood spots, cord blood, heel prick into capillary tube, or saliva via sponge stick.

Wrapping up, Bick emphasized that Genomics England is a “very large program” that includes many individuals working on various aspects of the program. He also noted that the whole-genome sequencing of newborns is becoming a worldwide movement, and provided a complete listing of the many groups worldwide that are conducing or starting to conduct newborn genomic screening programs. Bick concluded his talk by displaying a long list of issues that still need to be resolved.

MILLION VETERAN PROGRAM

Jason Vassy, associate professor of medicine at the Harvard Medical School and a clinician investigator in the Veterans Affairs (VA) Boston system, presented next on the Million Veteran Program (MVP).⁵ Vassy started by clarifying to the audience that MVP “writ large is not a wholesale return-of-results program.” He explained that he would be discussing some of the deliberations and pilot studies around returning genomic results, but that in general MVP does not return results. Vassy next oriented the audience to the Veterans Health Administration (VHA), which is one of three large administrations that sit within the larger Department of Veterans Affairs.^6,7 He explained that the VHA is the largest integrated health care system in the United States, actively caring for more than 9 million veterans across every U.S. state and territory, including overseas. The VHA has more than 20 years of electronic health record data on more than 21 million individuals, all nested within a health care system.

Vassy shared that the VHA is complex in structure and policy. It is divided into about 18 geographically defined regions, which are composed of more than 1,200 health care facilities including more than 171 medical centers or hospitals and more than 1,000 outpatient centers. The VHA is a federated system, where many policy decisions are made at the very highest levels—for example, this year Congress passed the PACT Act⁸ to help revise the way the VHA addresses veterans’ concerns about their toxic exposures during military service. The VHA is now tasked with implementing the mandates within that law—at national, regional, and local levels. This means that while the VHA strives to use best practices across the entire organization, health care providers in one state may do things differently from those in another state.

The VHA is a health care system that provides care to patients, but it also has funding for the VHA Office of Research and Development, an intramural research program for VHA investigators, explained Vassy. It is organized into four major research and development service areas (Biomedical Laboratory, Clinical Science, Health Services, and Rehabilitation) with two major research programs (Cooperative Studies Program and MVP).

___________________

⁵ Suma Muralidhar, director of the MVP program, contributed to some of the presentation’s background slides.

⁶ The other two administrations are Benefits and Cemetery.

⁷ Vassy explained that people in health care and biomedical research will often use “VA” when speaking specifically about the VHA. However, in these proceedings, in order to be as clear as possible, VHA is used at all times when referring to the Veterans Health Administration.

⁸ The Sergeant First Class Heath Robinson Honoring our Promise to Address Comprehensive Toxics (PACT) Act, passed the Senate in June 2022. For more information, see https://www.congress.gov/bill/117th-congress/house-bill/3967/text

Vassy shared that the MVP was launched in 2011 as a national VHA research program and is now one of the world’s largest programs on genetics and health. It was designed specifically to advance precision health care by learning how genes, lifestyle, and specifically military experiences and exposures might affect health and illness. The MVP’s mandate within the VHA is to focus on veterans’ health.

Vassy noted that veterans, in addition to military exposures, traumatic brain injury, higher rates of substance use disorder, and mental health concerns, are Americans who have the same diseases that other Americans have (e.g., cardiovascular disease, cancer, hypertension, diabetes). He recognized that “as a health care system that cares for those who have had military experience, the VHA has a unique opportunity to contribute to the science on understanding many military exposure–related conditions, and how that impacts not only military conditions, but then also just other chronic diseases.”

The MVP’s strengths include use of curated standardized datasets in a secure centralized computing environment with more than 20 years of curated, standardized electronic medical record data and other administrative data, said Vassy. The program strives to make the discoveries that will ultimately translate into improving veterans’ health, and in doing so also improve the health of other global citizens who could benefit from the science the MVP is producing. Vassy explained that at present the MVP has enrolled more than 900,000 of the planned 1 million and is currently the world’s largest genomic database. The program strives for racial and ethnic diversity, and as of a couple months ago, included more than 150,000 veterans who identify as African American.

The MVP has more than 650,000 genotype samples, and much of those data are available to the research community for analysis. MVP has made a name for itself in genome-wide association studies, stated Vassy. “It is looked to not only as a discovery cohort but as a validation cohort for other investigators who have done a genome-wide association study of a particular phenotype or disease,” he said.

Like NHANES and other studies, the MVP has grappled with what to do when it finds genetic results, and Vassy’s presentation covered how the MVP defines what is actionable, what is reportable, and how it would begin to think about returning results. MVP and NHANES share many other traits. Both are national in scope, which offers great opportunities in terms of generalizability, yet also makes the programs administratively complex when one begins to think about returning results. Both were initially designed as research programs, not clinical programs. “There was never an expectation among participants that they would get results back, that this was somehow their medical care, and that is true of NHANES too,” said Vassy. NHANES and the MVP are similar in the way that participants

were consented when they initially signed up to be a part of the research program without any intention of returning results. A key difference between the MVP and NHANES is that the MVP is linked to a national health care system and NHANES is not. However, Vassy noted that NHANES’s established processes for returning other clinical results, such as cholesterol levels, “could be a significant strength for NHANES depending on what direction they wanted to take.”

Vassy reflected that prior workshop speakers presented on protocols for how to return results back to participants, and ideally equip participants to share that information with their health care providers for appropriate use in clinical care. In the context of the MVP, the VHA is the clinical care provider, which Vassy stated “brings opportunities and responsibilities.” He outlined a few considerations pertaining to returning results the MVP takes into account, noting that there are many more.

First, the MVP’s initial consent, which was used to enroll more than 500,000 participants, did not include any expectation of return of results, but it did leave participants open to be recontacted for future research. “That ended up being an important distinction for how we were able to move forward with return of results for that first several-hundred-thousand people,” noted Vassy. Analytic validity is another important consideration, said Vassy. MVP uses an Affymetrix chip that was not designed to detect all pathogenic or likely pathogenic variants in the genes listed as actionable by the American College of Medical Genetics and Genomics (ACMG). A program deciding after the fact whether to return results “certainly needs to consider that bare minimum analytic validity before they can begin to think about the actionability,” offered Vassy. The next consideration Vassy outlined relates to being a research program within a health care system. He shared that the research and clinical “sides of the house” are very different. For example, the clinical side is burdened with high rates of burnout and new mandates coming down all the time. “In that context, the busy primary care provider doesn’t necessarily want to take care of something that a researcher found about their patient,” explained Vassy.

The MVP started to consider returning results around 2019 in two specific contexts: familial hypercholesterolemia (FH) and metastatic prostate cancer (mPC). Vassy explained that because the MVP is not doing gene sequencing, it could not claim to be detecting all possible FH variants. However, for those it could detect, the MVP wanted to explore what the return of results would look like and what the outcomes would be. Vassy continued by explaining that for men with mPC, having a known BRCA variant can inform therapy and change the clinical management.

Table 7-1 provides a summary of both projects. Returning mPC results meant trying to recontact individuals as long as 5 years postdonation to reconsent them to get clinical confirmation and get the results in the hands

TABLE 7-1 Key Features of Two Pilot Studies for Returning Genetic Results in the Million Veteran Program (MVP)

SOURCE: Adapted from workshop presentation by Jason Vassy on December 8, 2022 (slide 10).

of their clinicians in a timely way. However, Vassy shared, “It didn’t work in a timely way.” The protocol was reconfigured to say that a care gap had been identified. The investigators decided to identify the men across the VHA system, MVP or not, who have metastatic prostate cancer and have not had genetic testing. “I bring this up as a challenge of return of results, but also a way that thinking through that process helped us improve, potentially, other processes of care that should have been happening anyway,” explained Vassy.

Vassy spoke next about the FH study that he leads. That study reaches out to participants, lets them know that something may have been found in the initial research data they provided, and invites them to participate in a new research protocol. For participants who agree to participate in the new protocol, the MVP sends a second specimen for clinical confirmation in a commercial clinical lab. Regardless of whether that result is confirmed, a genetic counselor uses a telegenetic counseling intervention for both pre- and post-test counseling.

Vassy concluded by sharing that the MVP is actively considering what MVP 2.0 will look like. “We hope to enroll a million individuals by Veterans Day of 2023, but we’re not stopping,” said Vassy. The program is beginning to consider what the consents and biospecimens will be for the next million, with the benefit of being able to consider beforehand what results to return to participants. Vassy shared that there is potential interest in expanding the list of genes or conditions, as well as possibly using pharmacogenomics as a way of returning value to the MVP participants. “I think clearly what a health care system like ours needs to consider is what kind of new service delivery models can help connect the research to the clinical care, because asking primary care, for example, to take this on, is I think is too big an ask,” said Vassy. For example, there may be some new service delivery models such as telegenetic consultation services that allow regional or national-level experts to connect with local providers.

Vassy ended by thanking the veterans and MVP participants who made the research possible and acknowledged that, while the MVP has helped find new ways to improve veterans’ health, there is still more work to do.

ALABAMA GENOMIC HEALTH INITIATIVE

Bruce Korf, associate dean for genomic medicine and the chief genomics officer at the University of Alabama at Birmingham (UAB), presented on the Alabama Genomic Health Initiative, a collaboration of UAB and HudsonAlpha Institute for Biotechnology in Huntsville that is in its sixth year. Korf described the program as “an IRB [Institutional Review Board]–approved research program with a clinical workflow,” and informed the audience that the program has gone through two iterations. The initial phase, from 2017 to 2020, included a population cohort of adults from all over Alabama who were interested in participating.⁹ Those individuals had genotyping done with what is now the Illumina Global Diversity Array,¹⁰ which included an analysis of actionable variants and a return of results of these actionable variants with genetic counseling “and then plugging into supportive care.” The study was put on hiatus because they were unable to enroll new participants during the COVID-19 pandemic; when it was reconstituted in 2021, it was modified to enroll participants through three family medicine practices in Birmingham, Hoover, and Selma, Alabama. In the modified program, physicians order the testing, which still consists of a genotyping array, to look at both actionable variants and pharmacogenetic variants. Actionable variant and pharmacogenetic variant results are returned to the participant through the physician. All participants are offered the opportunity to opt in to having their DNA kept in a tissue bank for future research purposes and their data shared in a genomic database and merged with electronic medical records through the Informatics for Integrating Biology and the Bedside (i2b2) database.¹¹ The actionable variants and pharmacogenetic analysis are on one consent; whole-genome sequencing for diagnosis of rare disorders is on a separate consent form.

Korf summarized the workflow, which begins with recruiting and enrolling participants from three family medicine practices. Once obtained, the samples go to HudsonAlpha, where DNA is obtained in a laboratory

___________________

⁹ The Alabama Genomic Health Initiative has had two cohorts: the population cohort (discussed here) and the affected or rare disease cohort, which includes people with evidence of a genetic disorder, in which case whole-genome sequencing is offered in an effort to make a diagnosis.

¹⁰ Known as the “Global Screening Array” in the earlier phases of the initiative.

¹¹ The i2b2 software was developed at the National Institutes of Health–funded National Center for Biomedical Computing based at Partners HealthCare System. For more information see https://www.i2b2.org/index.html

certified under Clinical Laboratory Improvement Amendments of 1988 (CLIA) regulations with a Sanger confirmation of any pathogenic or likely pathogenic variant that is identified. Individuals with a positive finding receive a formal laboratory report, while those with no “informative findings” get a letter. The expected rate of return was between 1 and 3 percent, but to date it has been closer to 1 percent.

The pharmacogenetic results are drawn from the same array and include a separate assay for CYPD6 copy number changes, said Korf.¹² The laboratory issues a formal report, and then the pharmacogenetic pharmacists review all of the pharmacogenetic data, as well as the individual’s medication lists, and make recommendations to the physician if there are any instances in which a change of medication could be beneficial. The participant gets a more general report of the pharmacogenetic findings. The expected return rate for the pharmacogenetic results is essentially 100 percent because everybody has something that potentially could be actionable pharmacogenetically, “although not everybody is currently on a medication where that would be relevant,” explained Korf.

In terms of resources provided to participants and their physicians, Korf said HudsonAlpha provides a clinician resource hub that includes information about the program in general and more detailed information about the various genes being tested. Separate patient-facing and physician-facing informational materials are available for both the pathogenic and pharmacogenetic results.

Enrollment Demographics

Korf reported that the original population cohort was 75 percent female and was not representative of the racial and ethnic diversity in the state population. The new family medicine or clinical cohort has more racial and ethnic diversity, with two-thirds of participants identified or identifying as Black or African American. “Still about three-quarters are female, which we’re told actually mirrors the demographics of the family medicine practice itself,” noted Korf. Korf shared that the clinical cohort also skews toward middle-aged or older individuals—a phenomenon that was explored using surveys to understand individuals’ motivations for participating.

Korf reported that cumulatively, the Alabama Genetic Health Initiative has enrolled more than 7,000 people from all counties in the state, and 105 actionable results have been returned to participants in either the population or clinical cohorts, with about a 1 percent return rate.

___________________

¹² For more information on the CYP2D6 gene, see https://www.ncbi.nlm.nih.gov/books/NBK574601/

Korf acknowledged that, by using the global diversity array, the program is not designed to pick up all possible pathogenic or likely pathogenic variants in the genes in question because many are not represented on the array. He noted that this is an important part of both the informed consent and the return-of-results processes, because “we want to make sure that people understand that this should not substitute for clinical testing if there actually are clinical indications.” Korf added that the program is also “very cognizant of the risks of false reassurance.”

Results from Screening

Korf said that the ACMG secondary findings list, which despite not being designed specifically for population screening was “a good starting point for a list of genes that would be actionable, would not necessarily be obvious clinically, and where there was at least some reasonable frequency that we would expect to see individuals with positive results.” In Table 7-2, Korf provided a full list of the genes that are currently defined as “actionable” for the clinical and population cohorts of the study, with asterisks (*) denoting where positive results have been returned to study participants.

“From a phenotypic point of view, things are all over the map,” summarized Korf, and highlighted some specific examples from the program as follows:

• Some people were aware that they or a member of their family had tested positive for a relevant gene.
• Some individuals with various of the cardiovascular pathogenic variants who did not have signs or symptoms of cardiovascular disease are now being followed but are “at this point in time” nonpenetrant.
• Genetic results helped an individual who had received a heart transplant because of cardiomyopathy without a diagnosis better understand the origin and nature of their condition.
• Each individual who had a positive hypercholesterolemia result was aware they had hypercholesterolemia, but many of them were not aware that it could have a genetic basis and that it could have relevance to other members of their family.
• BRCA2, MYBPC3, BRCA1, and RYR1 have had the highest absolute number of results being returned.

Upon enrollment into the program, participants are asked to complete a questionnaire that collects family history information of genes that are relevant to the conditions the program is screening for. Those family histories are reviewed by a team of genetic counselors and divided into one of three

TABLE 7-2 Genes Currently Defined as Actionable for the Clinical and Population Cohorts of the Alabama Genomic Health Initiative, with Disease Indication, Grouped by Disease Type

NOTE: Asterisks (*) indicate genes that have been returned to study participants.

SOURCE: Adapted from workshop presentation by Bruce Korf workshop presentation, December 8, 2022 (slide 10).

color-coded categories. Korf summarized the results of the review of genetic results alongside family histories, where green are instances in which nothing was found in the family history that raised significant concern (49%), yellow are instances in which there were things that could potentially benefit from genetic counseling (5%), and red were things where there were significant red flags in the family history (46%). Program participants are provided a written report and advised of the possibility of obtaining clinical genetic counseling, “although that’s not a direct part of the study,” Korf noted. “Interestingly, in those instances where somebody had an actionable pathogenic or likely pathogenic variant and a family history, the family history does not always align with what you would predict from that particular variant,” shared Korf.

He next shared the results from a “behind the scenes” pharmacogenetic analysis of the less representative original population cohort, which shows the frequency with which actionable variants were found for individuals of European versus those from African American or Black ancestry (see Figure 7-2).

Characteristics and Motivations of Participants

Korf next highlighted a somewhat counterintuitive finding from the program: 92 percent of the population cohort consented to having their DNA sample stored in a biobank and data stored in a database, whereas only 82 percent of the clinical cohort did so. This was different from what was expected because the clinical cohort consists of individuals who are expected to have some sort of relationship, and therefore greater trust, with their primary care physician. Korf speculated that the shifts in demographics may at least partially account for this finding but acknowledged that they had not done analyses to explore that theory.

A genetic counselor used a survey of population cohort participants who received a result and agreed to be recontacted to look into their motivations for consenting to the DNA portion of the study. The most frequently reported reasons participants provided were desire to contribute to research on genetics (67%) and concern about an individual’s future risk of a health problem (60%). More than half reported being motivated by “general curiosity” (57%) and around one-third wanted to know if they might have transmitted something to their offspring (34%). Smaller proportions reported already having signs or symptoms of a problem (28%) or seeking to fill a gap in knowing about their family history (19%). Some participants were looking to understand why they had a particular health problem in the past (13%) or wanted to determine if they currently have a health problem (28%). Finally, despite being told that the research testing was not equivalent to clinical testing,¹³ about 10 percent of participants admitted that they were interested in participating in the genetic testing arm of the research because it provided access to testing that their insurance would not pay for.

A small proportion in the overall population cohort discussed their results with their health care provider (22%), which Korf explained was not surprising because the majority did not get any actionable results. However, the majority reported sharing their results with family members (78%), and the majority of those who did not share the results with family

___________________

¹³ Korf clarified, “And I mean by that, I should say, that we don’t have nearly the coverage that, for example, a person would get if they presented with a family or personal history, let’s say, of breast cancer.”

members did so because they did not feel that the results were important enough to share.

Many people rated the results as being very or extremely valuable and most reported being satisfied with the experience. Few individuals made changes in insurance based on the results they received, but some did claim to have made changes in health and wellness. They were not asked to provide specifics. “The number that actually did additional tests or exams or procedures corresponds pretty well with the number who were getting actually pathogenic or likely pathogenic results back,” noted Korf.

Korf offered a few conclusions based on his experiences with the program. In general, the program has seen “strong community interest.” The rate of results returned was on the lower end of the expected range of 1–3 percent, which Korf thought made sense since the program was not designed to detect all possible pathogenic variants. The program remains cognizant of the need to avoid false assurance. Finally, he noted that “non-penetrance is common at [the] time of testing.”

ALL OF US

Stephanie Devaney, chief operating officer at the All of Us Research Program, gave the workshop’s final talk, starting by sharing the All of Us Research Program’s mission to accelerate health research and medical breakthroughs that enable individualized prevention, treatment, and care for “all of us.” Devaney highlighted that from the All of Us Research Program’s creation, returning genetic results was a key goal. For example, both President Obama, in his State of the Union address in 2015, and the Advisory Committee that outlined the All of Us Research Program blueprint specifically mentioned the importance of the responsible return of results. As a result, many of the All of Us Research Program’s core values were built around this goal, including “participant return of value,” “participants have access to their information,” and “participants are partners.” Although eventually, the All of Us Research Program plans to return several kinds of information to participants (e.g., comparative survey data, scientific findings, and opportunities to be contacted for other research opportunities), it is currently focused on returning genetic information.

All of Us’s Commitment to Diversity and Participant Engagement

Since the All of Us Research Program launched in May of 2018, about 556,000 participants have consented to the study and are in varying stages of the protocol. For example, about 338,000 are regularly contributing electronic health records, and about 400,000 have provided banked bio-samples, which are being analyzed using whole-genome sequencing. The All

of Us Research Program aims to enroll at 1 million participants who reflect the diversity of the United States and eventually deliver one of the “largest, richest biomedical datasets for broad and open access” across global research communities. Devaney emphasized that the All of Us Research Program is committed to increasing the diversity of the datasets used in science and that about 50 percent of the All of Us Research Program’s research participants come from racial and ethnic minority groups. In addition, Devaney highlighted that the All of Us Research Program thinks about diversity in many ways, including age, sexual and gender identity, educational attainment, income, geography, and disability.

Next, Devaney described the All of Us Research Program’s commitment to engaging with their participants and shared several ways the All of Us Research Program builds trust. For example, one of the first things the All of Us Research Program did was to try to understand “what exactly do people want to get back? What would it mean to be involved in a program like this?” To answer this, in 2016, the All of Us Research Program funded a group of researchers led by Consuelo Wilkins at Vanderbilt University to conduct 77 engagement studios¹⁴ in 17 cities. Some key findings of this research included that people are particularly interested in receiving their genetic results and that while people are generally interested in advancing science, they also want to know precisely what discoveries are made due to their participation in research. Another way the All of Us Research Program actively engages with its participants is through communication with the participants themselves. The All of Us Research Program Participant Partners are active research participants who are also “active in all levels of the governance of the program.” Devaney acknowledged and thanked this group for being “a really important connection point for us” that has helped guide the All of Us Research Program through returning information and communicating with participants.

Building the Infrastructure to Return Genomic Results

Devaney showed a slide depicting the numerous All of Us Research Consortium member organizations. Then, she walked the audience through several sectors in which these organizations operate, such as communications and health care provider organizations. However, per the discussion at hand, Devaney emphasized the importance of the group of partners who facilitate the All of Us Research Program’s return-of-results workflow. This

___________________

¹⁴ During fact checking, Devaney explained that in the context of the All of Us Research Program, engagement studios are the same as focus groups, named “engagement studios” to reflect that they are part of the program’s core engagement work, to get insights and feedback from communities directly.

effort is advanced by the Participant Technology Systems Center (PTSC),¹⁵ the Mayo Clinic Biobank, the Data and Research Center, and various Genomics Partners.¹⁶

For the first year and a half, the All of Us Research Program only returned genetic ancestry and trait results to participants. Devaney shared that this was “a good opportunity for our program to start the return of results and generating the genomic data with our participants who had told us in a consent that they wanted to get information back. [It] also gave us an opportunity to help participants who might not be as familiar with genetic testing on what kind of information we can get from DNA.” The All of Us Research Program had only very recently begun returning health-related genetic results, which included a Hereditary Disease Risk (HDR) reporting on SNVs and Insertion and Deletions in 59 disease-related genes and a Medicine and Your DNA report (i.e., pharmacogenetic report), providing information about seven genes and commonly prescribed drugs. Devaney noted that they expect a small percentage of people will have an HDR positive report, estimating this might fall somewhere between 1 and 3 percent.¹⁷ However, Devaney predicted that over 90 percent of participants would receive a Medicine and Your DNA report in the future.

Devaney went on to briefly describe the current return-of-results process. If a participant elects to receive their results, their DNA is interpreted at a clinical validation laboratory, and then all results are “supported” by genetic counselors. Devaney explained that genetic counselor support means that all positive results are returned through genetic counselors. However, any participant who receives a report, even those with negative findings, can opt to call and speak to genetic counselors at no cost. Devaney also noted that since the All of Us Research Program has an Investigational Device Exemption approval from the Food and Drug Administration, it must clearly communicate with participants that initial results are “research” rather than “clinical” results. However, she added that participants who have a positive result on their HDR can also have clinical genetic confirmation of a positive result at no cost.

Finally, Devaney shared data regarding participants’ interest in and engagement with their genetic data. At the time of the workshop, the All of Us Research Program had sent notifications to over 180,000 of its participants asking them if they would like to receive genetic ancestry and traits information, and about 71 percent responded yes. It has since

___________________

¹⁵ The PTSC manages the participant portal.

¹⁶ Their Genomic Partners include centers that run their genomic analyses and Color Genomics, a company that supports the return of results and genetic counseling for participants who receive a positive result.

¹⁷ The All of Us Research Program did not yet have precise numbers since it had only recently begun returning health-related genetic results a few weeks before the workshop.

conducted surveys and found that 88 percent of those individuals who viewed any result viewed their trait results (e.g., bitter taste perception, cilantro preference), and 97 percent viewed their genetic ancestry results. Devaney commented that this level of interest in their genetic results was consistent with their preliminary research findings that individuals are interested in receiving genetic information about themselves. Unfortunately, engagement data were not yet available for health-related genetic results.

Lessons Learned

Devaney concluded by sharing some of the lessons learned at the All of Us Research Program. However, she highlighted the caveat that the lessons the All of Us Research Program has learned are mostly limited to establishing the infrastructure and the regulatory pieces around the return of results, given that it was still at the very beginning of returning health-related genetic results to participants at the time of the workshop.

First, the All of Us Research Program has consistently found that participants want to receive their genetic information. Second, Devaney emphasized, returning results responsibly is time sensitive and costly. Devaney noted the commitment and effort required to build the necessary infrastructure and workflows that comply with all regulatory requirements. On these matters, Devaney offered that the All of Us Research Program would be happy to participate in follow-up conversations with NHANES if it would be helpful. Third, ensuring support is in place for participants is an essential element of the program, and it requires significant effort from the All of Us Research Program, particularly for participants who opt to receive their results. Creating a suitable support system within the All of Us Research Program required establishing a network of organizations and experts who run genomic analyses, building technology and communication pathways, ensuring appropriate educational materials and support are available, and providing genetic counseling for participants. Finally, Devaney expressed her belief that the ability to recontact participants is critical to providing participants with a suitable support system. She shared that the All of Us Research Program intends to continue to update the information provided to participants as the science advances, and it can more confidently return different types of results over time. Devaney suggested that recontacting should be emphasized during the discussion, since NHANES’s limited longitudinal relationship with participants hinders its ability to recontact. In closing, Devaney reiterated her commitment to keeping the conversation going with the NHANES team after the conclusion of the workshop, and she offered to be a resource as it continues to consider returning genomic results to its participants.

QUESTIONS AND REFLECTIONS FROM SESSION 6

Biesecker, the Session 6 discussion moderator, started by stating that the goal of this discussion is to ask questions, get clarifications, and highlight topics or issues that may be useful to explore further. He started by offering a clarification that to the best of his knowledge there is no secondary findings return policy, either in the clinical realm or the research realm, that obligates high sensitivity or completeness of coverage for genes as a criterion of secondary findings return. He asked for input on this point from the audience.

Bick explained that individuals enrolling in the Genomics England research are told this up front. He noted that this must be repeated many times, and he offered an example of a participant who says that they already had a screening test for breast cancer and may be confused if they receive results that indicate they are at risk of a genetic breast cancer. Korf agreed with Bick’s and Biesecker’s points, and amplified Bick’s point by saying “no matter how many times you tell people, you just always worry that they’re going to misinterpret the comprehensiveness, and we drive that home every opportunity we have.” Vassy noted that this might be an interesting question for NHANES. He explained that as a VA clinician, his patients will say they have been tested for particular things, only to find out that they are referring to their participation in the MVP. “Maybe the participants don’t feel that the NHANES exam was a part of their medical care, which might be different from other settings,” said Vassy.

Biesecker asked next if each of the speakers could talk about how they decided which platform and genomic technology to use. Bick spoke first, answering that Genomics England is using whole-genome sequencing because it is the only method for capturing all of the molecular mechanisms. “We wanted a technology that would have legs over several years, unlike exome where you have to update that … but cost is a significant issue,” explained Bick. Vassy explained that MVP’s intention was discovery on a large scale, specifically genome-wide association studies, and so a chip “chosen for significant ancestral diversity enriched for some clinical content was an appropriate choice at that time.” He indicated that because of cost considerations, MVP uses something it does not consider a clinical-grade test.

Korf shared that after looking at many platforms, the Alabama Genomic Health Initiative opted for an array because of cost considerations, a factor it is constantly reevaluating as the cost curves for whole-genome sequencing continue to come down. He pointed out that arrays are very difficult to validate when dealing with rare variants, with a very high rate of false positives when looking at the raw data. “We do a CLIA Sanger confirmation of anything that we return, so we know that we’re not returning false positives

when it comes to returning results, but of course that adds a layer of effort and cost that has to be factored in as well,” explained Korf.

Devaney explained that the All of Us Research Program set out to do whole-genome sequencing. For them, a big consideration was having consistent technology across Genome Centers running the analyses, which led them to select the Illumina platform. In addition to whole-genome sequencing, it also had an array specifically developed for the program. “To Jason [Vassy]’s point, it was really about ancestral diversity and making sure that we had a lot of coverage,” she stated.

Biesecker asked the speakers to share their experiences striving for PPV which is “not so great on chips.” Vassy shared that from release 3 of the MVP data to release 4, it updated the calling algorithm used for some of the rare heterozygotes in the sample, which significantly fixed the problem that the return-of-results team found: up to 40 percent of their calls from array data were not confirmed by clinical testing. “Since then it’s been one for one,” said Vassy, acknowledging that this was an unexpected result that they did not fully consider initially.

Biesecker asked if Bick could talk about how Genomics England is doing orthogonal confirmations and how that affects their participant interactions. The study explains up front before being consented that they will be provided with a preliminary result through their physician who can help these parents of newborns recall their participation in the study. He explained that for many conditions there is a non-DNA test that can be used to determine if the child has the condition. “That’s important because if you want to integrate a DNA test into the routine care of medicine, the doctor will want to do a test that he is used to,” Bick emphasized. He clarified that it is not general pediatricians calling the moms and dads who are likely to be upset upon hearing the news, but rather specialists who are quite specific. “The first person who speaks to them needs to be very knowledgeable.”

This page intentionally left blank.