The workshop’s third session, moderated by Bruce Blumberg, a professor of clinical science and a planning co-lead of faculty development at the Kaiser Permanente School of Medicine, discussed factors that may affect how consumer genomics data are integrated into clinical care. The session’s four panelists also examined the challenges of and opportunities for using consumer genomics data for research and explored emerging cross-sector collaborations and potential lessons to learn. Andrew Singleton, a senior investigator in the Laboratory of Neurogenetics at the National Institute on Aging, discussed how data from direct-to-consumer (DTC) genomic tests can help inform research on the genetics of complex diseases. Siobhan Dolan, a professor and the vice chair for research in the Department of Obstetrics and Gynecology and Women’s Health at the Albert Einstein College of Medicine, spoke about work to integrate genetics and genomics into clinical care. Danielle Bonadies, the director of genetics at My Gene Counsel, addressed different paths to integrating DTC test result data into the medical model of disease. And Matthew Ferber, an associate professor of laboratory medicine and pathology and a consultant to the Division of Laboratory Genetics and Genomics in the Department of Laboratory Medicine and Pathology at the Mayo Clinic, discussed the lessons his institution has learned from its work on what he called a “near-consumer” testing experience.
INCORPORATING DIRECT-TO-CONSUMER DATA INTO COMPLEX DISEASE RESEARCH
Singleton reviewed some recent work in which he and his colleagues, in collaboration with 23andMe, examined genome-wide associations to identify genetic variants associated with Parkinson’s disease (Nalls et al., 2019). His group started investigating common genetic variability in Parkinson’s disease in 2005 and have identified 90 individual risk factors that increase the odds of developing this disease. This work required large sample sets, such as the population-scale cohorts of the UK Biobank and 23andMe’s growing cohort of individuals with Parkinson’s disease and controls. The group began working with 23andMe around 2011, and in a meta-analysis that year 23andMe participants accounted for approximately 3,400 out of the study’s 15,000 patients with Parkinson’s disease and approximately 29,600 out of 50,000 total controls, Singleton said (Do et al., 2011; IPDGC and WTCCC2, 2011). By 2019, 23andMe participants accounted for approximately 13,000
out of 37,700 patients with the disease and approximately 935,000 out of 1.4 million total controls, he added (Nalls et al., 2019). 23andMe participants currently account for one-third of the cases that Singleton’s group studies and 70 percent of its control group. Without those cases, Singleton said, he and his group would most likely not have learned of about 30 to 40 important genetic loci in Parkinson’s disease. Driven by a common interest, 23andMe and Singleton’s team are now looking for genetic associations between Parkinson’s disease and other traits and also seeking to determine whether the genetic basis for this disease differs between men and women.
Though some have questioned the quality of the data from DTC genomic testing services, Singleton said that, in his experience, the data are of the same quality as the data his group generates at the National Institutes of Health. In addition, he said, 23andMe has been quite responsive when it comes to sharing or analyzing data, perhaps because of the involvement of foundations and other philanthropic organizations. One issue that does arise has to do with not contravening any work the company is engaging in with corporate clients. The work that Singleton’s group does with the company has to be non-competitive with the work of 23andMe’s corporate partners. Another issue is that some types of data are not easily accessible, and current policy limits broad data-sharing agreements. For example, Singleton’s group never sees 23andMe’s raw data, only summary statistics, and anyone who wants those data after his group publishes its work has to engage in a separate data agreement with 23andMe. However, the company does have a clear and effective process for requesting data, he added. In terms of scale, some 7 million to 8 million people have been genotyped in the cohort being studied, with about 20,000 Parkinson’s disease patients in that cohort.
There are other ways that these data are being used in research, Singleton said, such as a project being conducted with The Michael J. Fox Foundation for Parkinson’s Research and 23andMe. In this project, individuals with Parkinson’s disease can upload reams of information about themselves to the Fox Insight web portal, including their electronic health records, diaries of daily activities, and genomic test results from 23andMe.1 The patients’ data belong to the individuals, but they are also available for researchers to use. “This is a neat research idea, and one that I am surprised has not been used more often,” Singleton said.
A relatively new company, LunaDNA, also has a web portal for individuals to contribute their genomic test data and health care information, Singleton said.2 In return, everyone who contributes data receives a
___________________
1 For more information about Fox Insight, see https://foxinsight.michaeljfox.org (accessed December 11, 2019).
2 For more information about LunaDNA, see https://www.lunadna.com (accessed December 11, 2019).
small ownership interest in the company. When researchers pay to conduct research on the de-identified and aggregated data, the proceeds are passed along to those who shared their data. “I do not know if this will work,” Singleton said, “but I think it is an interesting concept in terms of thinking about ownership of data and patients or individuals taking control and making use of their own genetic data.” One participant asked how researchers can account for DTC data coming from a database like LunaDNA if they are also collecting data from a DTC company. In his laboratory, Singleton said, a check sum is created, which provides information about genetic identity that is cross-referenced with the check sum number coming from the DTC company. In that way, any duplicate individuals can be removed.
INTEGRATING GENETICS AND GENOMICS DATA
As a physician in the Bronx, Siobhan Dolan said, she provides care to a diverse and vibrant community of more than 1.4 million citizens. Almost 60 percent of this population speaks a language other than English at home, more than 30 percent live in poverty, and 10 percent do not have health insurance, yet more than 80 percent of households in the Bronx have a computer at home. The Department of Obstetrics and Gynecology and Women’s Health at the Montefiore Medical Center has a team of 7 genetic counselors and 3 clinical geneticists who together see between 30 and 40 patients per day. The patients include those undergoing prenatal genetic testing, cancer patients, and those seeking care for other reasons, such as multiple miscarriages and infertility. In her experience, Dolan said, she does not have patients coming to see her with their DTC genomic test results, although there are other patient-driven, non-medical concerns related to genetic testing that she sees as a clinician.
Prenatal genetic testing can provide expectant parents with a great deal of information, including aneuploidy diagnosis and carrier status, Dolan noted, but she added that what many patients are really interested in is the sex of their child. The gender reveal phenomenon,3 she said, creates a complicated paradigm for clinical care because the person taking the test often does not want to receive the results personally and because expecting parents want the results early in order to plan their gender reveal party. The problem with this, she said, is that people sometimes enter into prenatal genetic testing with little consideration that they might also receive the results of high-risk genetic screening tests, despite having gone through the informed consent process prior to testing.
___________________
3 For a brief background on “gender reveal parties,” see https://www.washingtonpost.com/news/parenting/wp/2018/05/13/how-do-parents-find-out-the-sex-of-their-baby-today-exploring-the-new-trend-of-gender-reveal-parties (accessed December 13, 2019).
Another challenge is the lack of integration of genomic test results into an individual’s electronic health record, Dolan said. As a result, test results are not easily accessible by the patient through a web portal, and patients cannot get the information they want. Given the challenges with the patient interface, patients can sometimes misinterpret a fragile X screening result for the gender as well. Paternity testing is another element that patients are interested in, although Dolan said that this is not something her clinic does because it falls under a legal, not medical, model. Nonetheless, many patients inquire about paternity testing and wonder whether DTC genetic testing can be used to establish paternity. From her perspective, Dolan said, there can be a lot of consequential family structure information to consider. Sometimes patients will also request amniocentesis in order to determine paternity, which can be ethically challenging for providers.
A question that arises with cancer genomics is how to connect the high-risk patients, those who need genetics or genomics services, with their department. On occasion, for example, individuals will come in with a list of relatives, each of whom had a different cancer when they were in their 80s, and they are worried about their risk of developing cancer. Dolan said that in such a situation she will explain that the pattern does not fit a pattern of genetic risk. In other cases, she sees individuals diagnosed with breast or ovarian cancer who have a strong family history of those cancers and regrets the missed opportunity to find those individuals early, before they developed cancer. Obtaining important information from family members can be further complicated by the challenges in finding and accessing old test results and family health history information. Family members may not remember the results and therefore may inadvertently convey incorrect information.
Patients are often interested in multigene panels and in learning as much as they can about their genetic risk factors, Dolan said. She talks to patients about variants of uncertain significance before ordering the test so that they are not blindsided when they receive their test results. As a provider trying to communicate uncertainty, she said, it can be challenging to explain the complexities of variants of uncertain significance in a way that patients can really understand. Many patients are very interested in how to use the information they receive, but when Dolan explains that uncertainty is complex and that there are not necessarily clear guidelines for how to interpret that information, her patients will often try to find utility in in spite of that, and say, for example, that they will eat better and exercise more.
While much of the focus in consumer genomics is on helping increase consumers’ understanding, Dolan said that there are various challenges relating to providers as well. For example, she mentioned a surgeon wanting to operate based on a variant of uncertain significance. Risk is a con-
tinuum where the evidence is constantly evolving, and really understanding the intricacies of genetic test results is quite complicated, Dolan said.
INTEGRATING DIRECT-TO-CONSUMER DATA INTO THE MEDICAL MODEL OF DISEASE
Given the current estimates of DTC genomic testing services, an average of about 42,000 people undergo testing every day, Danielle Bonadies said, and by 2021 about 100 million Americans—one-third of the U.S. population—are expected to have had their genomes analyzed by one of these services. What that means, she said, is that whether a given individual had a DTC test, at least one relative of that individual will likely have had a DTC test (Khan and Mittelman, 2018). “Therefore,” she said, “some of your data, by being genetically related, is in that database.”
Bonadies said that she has seen in her clinic that patients have many residual questions after getting the results of their tests, such as what the tests did and did not look for, what they need to do next with the information from the tests, and how the information will affect the management of their health. On social media, she said, there are discussions about the accuracy of these tests and attempts to correct misinformation about positive and negative results. In some cases, the DTC model is merely transactional—send in a sample, get results—with little discussion about family history, verification testing, risk, medical management, or long-term care and surveillance.
A major question is how to integrate the data from DTC testing into the medical model so that an individual’s providers can access the data and help the individual understand what the data mean. One path, Bonadies said, could be through verification testing, where the verified results are integrated into the electronic health record. In one study of just under 50 patients, verification testing found that 40 percent of the results were false positives and that 19 percent were confirmed but classified inaccurately as pathogenic by the DTC service (Tandy-Connor et al., 2018). What was not examined in this study but that has been looked at by other investigators is how many positives DTC testing would have identified in individuals with a strong personal or family history of breast, ovarian, or colon cancer. The results, which have not yet been published, Bonadies said, showed that between 80 and 90 percent of the associated genetic markers would not have been picked up on a DTC genomics test, indicating that some consumers may be missing potentially important health information.
Despite the limitations, there are still routes for incorporating the significant and valid information that DTC genomic tests can produce, Bonadies said. Her company, My Gene Counsel, is building a bridge from DTC testing into medical-grade testing and the medical system by offering
genetic counseling and verification testing to individuals who have received a DTC finding or individuals who have concerns about their personal or family medical history. If those individuals are candidates for genetic testing and counseling, the company will help them navigate those routes. Several insurers, including Blue Shield of California, Anthem, and Aetna, now cover verification testing, particularly for the BRCA genes, Bonadies said. There are also research-grade testing options available through academic medical centers. Part of the challenge with this approach is that there are fewer than 5,000 genetic counselors in the United States who can provide the proper guidance based on these results (ABGC, 2019), and this challenge is likely to increase as gene panels become more complex and require more expert interpretation, Bonadies said.
Another avenue for integration could be through DTC medical-grade testing since there are now several laboratories offering this type of testing directly to consumers. On the other hand, some laboratories are pivoting away from a one-on-one interaction with a customer to large population-based studies, which could make it harder for consumers to access the services directly, Bonadies said.
An ongoing challenge will be getting back to patients with new information as research identifies more genetic associations with disease. “How do we reach back to our patients and notify them of those updates in the field?” Bonadies asked. Her group, for example, found that over the past 5 years there were more than 600 changes in medical management recommendations associated with the 59 genes that the American College of Medical Genetics and Genomics has identified as important for individuals to know if they have a pathogenic variant in those genes. “The medical system is not well set up to re-contact those patients and to keep them in the loop about their ever-evolving medical management and how they need to be followed,” Bonadies said. The system that she and her colleagues have developed, however, does include the ability to reach out to patients and offer them the opportunity to participate in relevant clinical trials.
Bonadies proposed a vision of the future focusing on long-term engagement with consumers that offers verification programs, tools for health care providers, accessible and scalable genetic counseling, gene- and variant-specific reports, and updating reports with notifications that are important for medical management. Such a future would also include comprehensive, searchable resources for both patients and providers and focus on the engagement and retention of those patients and their providers.
LESSONS FROM A NEAR-CONSUMER TESTING EXPERIENCE
Describing the various types of genetic testing available today, Matthew Ferber discussed the spectrum ranging from purely diagnostic applications
to what he called “edutainment”—how closely someone is related to Neanderthals, for example—and how the tests are ordered (i.e., via a provider or via more consumer-facing routes). DTC companies are now moving into the diagnostic space, he said, while the diagnostic companies are considering medically actionable mutations beyond BRCA.
Ferber said that when he first became a clinical laboratory director, he felt that genomics fell strictly in the medical sphere and that consumers might not understand the information received directly or that it might not be beneficial for them. Over the years, he said, his views have changed, and he now believes that DTC genomic testing creates an opportunity for people to talk about things that are important for their health, engage with their providers, and have meaningful conversations about genetics and genomics.
When the Mayo Clinic was deciding to launch its GeneGuide™ product, he said, one of the biggest concerns was not about the ability to interpret test results, but about keeping the cost of sequencing at a price affordable to the general consumer.4 The breakthrough, Ferber said, came when Helix was able to provide low-cost, high-quality sequencing data with no interpretation. Combining Helix’s expertise with the Mayo Clinic’s expertise made for a good match, he added. The goal in creating a product like GeneGuide™ was to help people better understand genetics and genomics before they were faced with a critical result. It is not diagnostic testing, Ferber said, so people who suspect they may have hereditary breast or ovarian cancer in their family should not use this test. If an individual within the health system was determined to have an indication for clinical testing, Mayo’s clinical partner, PWNHealth, would flag an individual’s test order, and it would not go through the system. Instead, Ferber said, PWNHealth would help the individual find a genetic counselor in the area that he or she could work with to navigate next steps.
The result of this partnership was an entry-level product that sought to not overwhelm consumers yet at the same time provided an avenue for educating consumers and helping them understand how genetics can affect their health. The idea, Ferber said, is to use people’s own genetic information to engage them in the process of learning about genetics. What he and his colleagues have learned, Ferber said, is that consumers enjoy learning about genomics but want more information about medically actionable variants, carrier screening, and pharmacogenomics information related to drugs they take.
An important difference between DTC tests and near-consumer tests, Ferber said, is that in the near-consumer environment consumers initiate the ordering process, but their physicians need to approve the orders and
___________________
4 For more information about the Mayo Clinic GeneGuide™, see https://www.mayoclinic.org/mayoclinic-geneguide (accessed December 13, 2019).
receive copies of the results to review before they are sent to the consumers. Partnering with a health care provider can achieve the right balance, he added. This can be important in certain instances where there are additional factors at play, such as in the case of a consumer who has had a liver transplant, in which case the individual would metabolize medications like the liver donor and DTC results would not be accurate. In cases where consumers may receive critical results from a GeneGuide™ test (e.g., malignant hypothermia), the result would be held back for a genetic counselor to deliver. There have been many questions over the years about whether consumer-focused genomic testing should be done, Ferber said, but it is time to move past that conversation because such testing is here, and it is up to the field to try and figure out how to address related issues in the most appropriate manner.
DISCUSSION
Facilitators and Barriers to Integration
To start the discussion, panelists discussed impediments and facilitators of consumer genomics integration, ranging from including genomic sequencing results in electronic health records to the cost of and reimbursement for genomic sequencing. It is important, Singleton said, to consider the day in the near future when whole-genome sequencing will be inexpensive enough that everyone would be sequenced at birth and that information becomes part of their medical records. In his opinion, he said, this becomes a facilitator because it starts to turn health care systems into learning systems, where research and health care become one and the same. Ferber agreed that newborn screening with whole-genome sequencing could one day become routine, but added that DTC screening tests should not be done in otherwise healthy minors who do not have the ability to fully understand the gravity of an outcome or have complete say in what course of action to take based on an outcome.
Another facilitator, Dolan said, would be broadening the insurance coverage of genomics tests before receiving a diagnosis. Medicare, for example, does not pay for testing prior to diagnosis even when someone has a strong family history of cancer. “The opportunity to identify someone at risk and take steps is essentially precluded for many of my patients if Medicare will not pay because the out of pocket cost is too substantial,” Dolan said. Another insurance-related facilitator would be to have a system where genetic counselors could be licensed and paid independently in every state, Dolan said, which would likely increase the supply of genetic counselors. One workshop participant referred to the challenges related to reimbursing pathologists and genetic counselors that had been discussed
throughout the day and asked how Mayo developed its business plan for making its product available to patients. Activities like this are difficult in an academic medical setting, Ferber said, which is why partnerships have been crucial to the success of the GeneGuide™. In determining the price of the testing, Mayo negotiated with its partners based on the estimated number of people who might require additional genetic counseling, whether on the front end or following the test, and used that number to distribute costs. Insurance companies are not billed for the testing, which Ferber said helps spread costs. Reflecting on the comments in the session about partnerships, Blumberg added that one facilitator or barrier, depending on whether it is alignment or misalignment, could be how common the interests are between the various parties involved in the collaboration.
Overburdened clinicians asked to review and comment on a DTC genomics report are an impediment to integration, Bonadies said. Primary care physicians should not be expected to stay up to date with all of the various genetic conditions, she said. Along the same lines, Ferber said that the genomics field has not made it easy for clinicians to use the information they are getting now. In his opinion, he said, this issue could be addressed using clinical decision support tools integrated into the electronic health record. That said, getting data from DTC test services into the electronic health record will be a challenge and will require setting some standards to enable integration.
Data Quality and Data Sharing
One participant asked if there is a clear distinction between medical grade versus DTC genomic testing. Many DTC genomic tests are performed in Clinical Laboratory Improvement Amendments (CLIA)-approved laboratories, Bonadies replied, but she said that, as a genetic counselor, she would always repeat a DTC test in a medical grade laboratory to confirm a result. Drawing on her earlier presentation about variance between the data returned from the DTC genomic testing companies and medical grade laboratories, she said that the discrepancies occurred on both the level of variant interpretation and the interpretation of the raw sequencing data.
Research, clinical care, and drug development are likely going to intersect around DTC genomic tests, a workshop participant said, and it is not difficult to imagine needing to connect individuals with polygenic risk scores to clinical trials being done with pre-symptomatic patients in the future. There are current efforts to collect patients with GBA and LRRK2 mutations for therapeutic development in the Parkinson’s disease space, Singleton said, and the field is also moving toward engaging relatives who carry those mutations for clinical trials. Therapeutic development is still in its early days in this space, Singleton added, but there could be opportuni-
ties to better understand and interpret complex variant data. This could allow researchers to understand the genetic basis of diseases and to develop specific therapeutics for those individuals or even define subpopulations who could benefit from a therapeutic before becoming symptomatic. In terms of using polygenic risk scores, Singleton said, there is more caution because of the regulatory issues involved.
Moving from a paradigm in which the responsibility of re-contacting patients lies with the physician who ordered the test to a living lab report is a good concept, Ferber said. Communicating new information to patients about old genetic test results has always been challenging, Bonadies said. She described how as a genetic counselor she used to send a newsletter to patients that included a list of gene-related updates, which left patients to understand and determine which of those updates applied to them. My Gene Counsel has developed a living lab report, Bonadies added, which includes the ability to re-contact patients in order to connect them with new clinical trials as the knowledge about gene associations changes over time. The new living lab report model allows her company to connect with patients and their providers about specific genes and variants, which means that the patients do not have to sort through extra information that may not apply to them.
This page intentionally left blank.
