NEW YORK (GenomeWeb) – Before the National Institutes of Health can begin to genetically test participants within its precision medicine initiative, it will have to figure out what results to return, how to minimize reporting false positives, and how to provide counseling to help them navigate the often uncertain and evolving evidence on genetic information.
And the project will have to figure out how to do all this on an unprecedented scale, for a million participants that the All of Us Research Program hopes to enroll over the next four years.
At a two-day NIH workshop this week, genetics researchers, healthcare technology firms, labs, and healthcare providers discussed best practices for returning genetic testing results and the challenges the project might face in communicating this information to so many participants with different levels of comfort in sharing their genetic data and varying expectations as to what they should get back in return.
"The biggest problem facing us is scale," said David Ledbetter, chief scientific officer at Geisinger Health System, during the webcast meeting. In 2007, Geisinger launched the biobanking and research project MyCode which has enrolled approximately 140,000 people and sequenced the exomes of more than 60,000 individuals. Since the project decided to begin returning clinically actionable results in 2013, 270 participants have received them.
During the workshop it became clear that while All of Us can draw on the experiences of projects like MyCode, it will also have to advance new strategies for communicating results to 1 million people. "You have heterogeneity in everything," said Pearl O'Rourke, director of human research affairs at Partners HealthCare.
The goal of All of Us is to collect in a longitudinal fashion a variety of data, including genomic, medical, and environmental, and use that to fuel research and advance precision medicine. In order to participate, individuals will have to agree to share their data and biological samples for research, take a baseline physical exam, and answer questionnaires from time to time. In return, the program has promised to return their personal study results, as well as summarized results across studies.
NIH Director Francis Collins noted at the start of the workshop that genomics studies haven't historically had policies in place to return results to participants. But because the findings from genetic testing can sometimes have significant implications for participants and their families, researchers have at times gone against standard practices.
Back in 1992, when Collins was part of a research team hunting for the BRCA1 gene, the consent form that participating families had to sign said results wouldn't be shared. However, Collins noted, breast cancer researcher Mary-Claire King and colleagues decided to break with this policy when they learned that a woman in the study was considering getting a prophylactic mastectomy but hadn't inherited a risk allele. But informing her of her result meant that researchers had to share results with her entire family, and the knowledge had dramatic implications for a number of relatives.
The story is a reminder of the ethical tensions that exist when returning genetic test results within research. "You don't want to give people information that will do harm, if you don't know what the consequences are going to be, but then … [there is] the principle of autonomy," Collins said. "Individuals should have access to information about themselves, if that's what they want."
But All of Us will have to deal with participants who may not want to know this information. Wondros, a company that is leading communications and engagement efforts for the program, conducted 60 one-hour interviews in four US cities to model the types of "personas" that might participate in a research project like All of Us, and found a range of archetypes. On one end of the spectrum were those who were willing to share everything in an altruistic fashion, while the other end included those who had lots of fear about data sharing and government involvement. Some people wanted to get back everything, regardless of whether it is actionable or not, while others weren't sure if they wanted to learn about things they couldn't do much about.
Within the All of Us program, the NIH wants to eventually sequence the whole genomes of a million people, but this won't happen until costs are more manageable. In the near term, the program will launch genotyping pilots in order to figure out some of the operational, ethical, and legal issues of returning this data to subjects on such a large scale. "We don't have to get everything perfect out of the gate," said Eric Dishman, director of the All of Us Research Program. "We want to do no harm. We want to deliver value to the participants."
During the two-day meeting, experts who have been providing genetic test results to research participants through projects like MyCode in regions of Pennsylvania and New Jersey, and the MedSeq Project in Boston, advised the All of Us program to start conservatively, and only report those results that are clinically actionable.
Most agreed that the program report only pathogenic, or maybe even likely pathogenic variants, in the more than 50 genes that the American College of Medical Genetics and Genomics tells labs to report as incidental findings following whole-genome or exome sequencing. Some speakers suggested the program return pharmacogenetic results, since most people expect that genetics will help them personalize their pharmacotherapy. Within 77 community engagement meetings across the US, people seemed to value different types of genetic information, said Consuelo Wilkins of Vanderbilt University Medical Center, but PGx results came out on top.
All of Us will also have to develop procedures for sifting through the hundreds of variants identified through testing and figure out which to report. Heidi Rehm, director of the Laboratory for Molecular Medicine at Partners Healthcare Personalized Medicine and an investigator in MedSeq, noted whole-genome sequencing reveals between 200 to 300 variants, and it takes scientists around 90 minutes to evaluate a variant when there is published literature on it, and 26 minutes when a variant couldn't be found in the literature.
As we got bigger and bigger numbers, it became more and more expensive.
Gold standard variant interpretations of this kind, as the MyCode project found out, can become expensive quickly. "Every variant we sent through the interpretation process was costing more and more money," said Christa Martin, director of Geisinger's Autism & Developmental Institute, who estimated that doing analysis the traditional way for the 250,000 participants MyCode hopes to enroll would take 40 years. "As we got bigger and bigger numbers, it became more and more expensive."
MyCode, which sends variants for analysis to Rehm's lab, reports to participants likely pathogenic and pathogenic variants in genes the ACMG recommends plus 20 additional genes. The investigators in that effort have now implemented a "variant triage" system for quickly identifying variants that have sufficient evidence to be likely pathogenic or pathogenic and filter out those that don't, before sending variants to Rehm's lab.
In addition to variant triage, Rehm suggested All of Us use variants with inter-lab agreement and expert review within ClinVar, NIH's repository of genotype/phenotype associations. Within ClinVar, 17 percent of the variants that two labs have interpreted have conflicting interpretations. But by sharing the data publicly within ClinVar and working with labs to reach consensus classifications, the discrepancy came down to 15 percent last year. A minority of variants in the database have been reviewed and interpreted by expert groups, but even then, there are two examples where variant classifications have changed.
"Knowledge [on variants] definitely changes and we have to deal with it," Rehm said, suggesting this is something All of Us will also have to contend with if it wants to return genetic testing results to participants. She noted, for example, that within MedSeq a reanalysis of 100 participants' genomes found 21 percent received updates to their result, half due to newly reported findings and half due to reclassified variants.
Another technical challenge that the All of Us program will have to consider is that most labs currently do orthogonal confirmation for all or a subset of variants (i.e. substitutions in homologous regions or indels) detected through next-generation sequencing, but such confirmation isn't possible for a project this large.
"In thinking about how to do this at scale, I would argue you have to bifurcate data you feel has the technical validity to be directly returned off of next-gen sequencing, versus those things … that would require additional confirmation, and might be relegated to more of a research result," Rehm said.
Genetic testing at the scale the All of Us program aspires to do will undoubtedly yield false-positive results, cautioned Robert Green, director of the Genome2People Research Program within Brigham and Women’s Hospital and Harvard Medical School. Based on a rough, back-of-the-envelope calculation, he estimated that sequencing a million people for rare, genetically dominant conditions could yield around 8,000 true positive results and 49,500 false positive results. "We have to think very carefully about how we want to contextualize a mix of true and false positives" in the study, he advised.
"I think we should keep in mind that clinical utility is not established" for most of the variants detected by NGS, Green further added. "We're performing a significant experiment here in using genomic information for population screening."
Green has extensively researched how direct reporting of genetic test results impacts consumers, and has generally found that when results are communicated in the correct context and with appropriate education, it doesn't cause people to become anxious or overutilize healthcare resources. But research participants often conflate investigational data with clinically actionable data, said Green, a problem that workshop participants kept coming back to throughout the two-day meeting.
"Members of the cohort are going to think this is clinical information," even if the All of Us program informs participants through the consent process that what they're engaging in is research, said Sharon Plon, professor of molecular and human genetics and Baylor College of Medicine. "I've had the experience of giving people consent that says you won't get anything back," she said. And yet, "they expect to get stuff back and they expect it to be clinically meaningful."
[T]hey expect to get stuff back and they expect it to be clinically meaningful.
Research by Wondros suggests that participants indeed expect to get something back for partaking in All of Us, and they hope to have access to genetics experts to help them understand their test results. "Many people said, 'If I'm going to get results back I don't want an e-mail [or] a letter in the mail. I want a phone call. I want to have a face-to-face meeting. I want an opportunity to ask questions, find out more and feel a bit safer with this whole process,'" said Kelli Auerbach from Wondros.
This was especially true if research identifies something that can have a negative impact on their health. "They hope that the program is going to take care of that," Auerbach said. Some expected to be connected to a specialist or genetic counselor, while others hoped the program would cover costs of further analysis. "The root of these desires was that [they felt] this was a big deal," she said. "I am giving a lot to you. And so, if I'm going to give you that … don't abandon me."
Workshop attendees also suggested All of Us employ a variety of strategies in counseling participants about returned results, from educational videos, webinars, telephone discussions, and face-to-face interactions for specific types of results. The All of Us program may need to get particularly creative on this front since there are only around 4,000 genetic counselors in the US.
Though participants may make little distinction between the research data they receive through All of Us and clinicaldata, once they walk into doctors' offices with this information, time-strapped providers may not be too happy. During the workshop, presenters advised the All of Us Program to create educational resources, and maybe even provide continuing medical education credits to doctors who want to learn how to discuss genetic test results with patients. On the other hand, the program also must accommodate doctors who want nothing to do with this research data.
There are also privacy risks to consider in returning genetic test results. Duke University conducted 60 interviews with thought leaders about hypothetical research involving 1 million participants, and only around 17 percent identified return of genetic information as a potential benefit. But more than half of interviewees identified receiving genetic information as a potential confidentiality risk.
These concerns have implications for how All of Us informs people about the risks of participating in research. Laura Beskow of Duke University cited a survey where one-third of people were unable to answer comprehension questions after going through the consent process, meaning these people may not be informed enough about the study to partake. This raises questions about whether All of Us will still enroll such individuals into the study.
Similarly, Beskow cited preliminary research on how people perceived the Genetic Information Nondiscrimination Act when information about it was included in consent documents. GINA bars the use of genetic information in health insurance and employment decisions, but it doesn't restrict its use in other areas, such as life insurance.
People Beskow surveyed had concerns about these gaps in the law. They wanted to hear some scenarios in which their confidentiality might be breached, despite assurances in the consent form that the study would take every step to protect their privacy. "Giving people back their genetic information is one way that [could happen,] Beskow reflected. "The information would then be in the hands of participants. They'll do with it what they like. It also might be in their medical records and that's one way that some of these loopholes in GINA might come into play."
Ultimately, in aspiring to return genetic test results to participants, All of Us will have to tackle a lot of issues about which there is currently little consensus in the field, Green said. For example, at a very basic level, genetic testing labs have different policies on how often they review the evidence and reclassify variants for customers. "Maybe this is a good thing," he said. "This research would lead clinical practice [rather] than follow it."