NEW YORK – The National Institutes of Health's All of Us Research Program has started to return its first genetic results — a small number of non-clinical traits and ancestry information — to several thousand participants while it continues to generate data and build infrastructure to start returning health-related findings and make genetics data available to the research community in late 2021.
This summer, the population health study, which has been assembling a diverse longitudinal cohort of US residents, received a long-awaited investigational device exemption (IDE) from the US Food and Drug Administration, allowing it to return disease risk findings in the so-called ACMG-59 genes as well as certain pharmacogenetic results to participants.
Josh Denny, who took over as CEO for All of Us a year ago from Eric Dishman, said that the program released its first large batch of genetic results to about 7,000 participants last week, following a small trial run in November. The reports members have been receiving are all based on array-based genotyping data and are limited in scope, containing genetic ancestry results for broad geographic areas and just four traits: bitter taste perception, cilantro preference, earwax type, and lactose intolerance. The plan is to release more traits over time, he said, and more participants will receive their reports on a rolling basis as the project produces additional genotyping data.
Color, the program's partner, developed the algorithm and software for the return of the ancestry and trait results. They are built on one of the company's own genetic testing products, called Color Discovery, that is available to customers who have previously ordered a clinical genetic test from the firm. "We worked closely with the All of Us Research Program's staff and stakeholders to produce a version of the Discovery product that could meet the program's need," Color CSO Alicia Zhou said in an email.
Participants will need to wait about another year, though, to obtain health-related results, which will be based on whole-genome sequencing data, Denny said. Those data have yet to be generated by the three designated All of Us genome centers. "We're still scaling that up but we have that pipeline running pretty well right now," he said.
This summer, the FDA granted the program an IDE that allows it to return certain medically relevant results. "At this point, it's just about building out a pipeline and all the return infrastructure," Denny said. "Our goals are to have both of those components probably towards the end of 2021. We're looking at about a year from now that we would have the health-related and pharmacogenetic traits released."
The whole-genome sequencing data will be produced by Baylor College of Medicine, the Broad Institute, and the Northwest Genomics Center at the University of Washington in a CLIA environment using short-read technologies, he said. A pilot project at the HudsonAlpha Institute to test long-read sequencing is still ongoing.
Any risk variants found in the ACMG-59 genes will then be validated in a clinical validation lab before they are returned to participants. "We are working on all the language around what those [reports] look like, making sure those words are understandable, and setting up our genetic counseling resource to support that," Denny said. Color is also involved in this process, developing the back-end infrastructure and genetic counseling capabilities.
In terms of what types of results the program will be able to return, the FDA was "very comfortable with going with the ACMG-59 genes," Denny said, while "the pharmacogenetics side was maybe a little less straightforward."
Based on the FDA's review, the plan is to report only Clinical Pharmacogenetics Implementation Consortium (CPIC) level A variant-drug interactions, he said, which "we feel confident calling on a whole-genome sequencing platform." Also, the PGx report will merely state the name of drugs and variants that affect their metabolism, but not give any medical advice. Instead, it will instruct participant to talk to their healthcare professional about the results.
Researchers will also need to wait another year or so before they can get their hands on the first batch of genetic data from All of Us. "The goal is to have the first release towards the end of 2021," Denny said. Next spring, the program plans to start internal alpha testing, he explained, using on the order of tens of thousands of genotypes, to develop the necessary tools that will allow its cloud-based analysis platform, called Researcher Workbench, to handle information from hundreds of thousands of whole-genome sequences. "Hopefully towards the end of the year, we'll have a dataset big enough, with quality controls, with the infrastructure we tested, scaled appropriately, to be able to release some genotype and sequence data," he said. "That first dataset needs to be of sufficient size," he added, though he declined to mention a target size.
The first release will probably contain more array-based than sequencing data, he said, since the latter has bigger files and takes longer to generate. After that initial batch, the plan is to release two genetic datasets per year.
Meanwhile, the program has made other types of data from All of Us participants available to the research community. In May, it released a beta version of Researcher Workbench with data from 225,000 participants. So far, more than 130 organizations have signed the program's data use agreement so their researchers can access the resource, which contains survey results, physical measurements, and electronic health record data. An initial evaluation of the first data release and platform, led by researchers at Vanderbilt University Medical Center, Denny's previous affiliation, appeared as a MedRxiv preprint in June. Several research projects using the data are ongoing, Denny said, including a serology project looking at the spread of the SARS-CoV-2 virus in early 2020.
This week, an updated version of Researcher Workbench with data from more than 315,000 participants was released that contains new health data, tools, and results from a survey of participants' experience with the COVID-19 pandemic. In addition, it includes a first set of Fitbit data from a subset of participants.
Recruitment for All of Us has taken a bit of a hit this year as a result of the pandemic but has started to resume. "We paused our in-person activities in March," Denny said, including advertisement, recruitment, and consenting. Since then, a few of the more than 300 recruitment clinics have reopened and enrollment has started again, "but certainly not at the pace that we did before," he added.
As of now, the program has consented 362,000 participants, of whom 271,000 have completed all basic protocols. Also, 233,000 participants have electronic health records connected to their data. The cohort continues to be diverse: more than 50 percent of participants who have contributed biospecimens are non-white, and 80 percent are from communities underrepresented in biomedical research, meeting at least one of the program's diversity metrics, which include age, socio-economic status, gender identity, sexual orientation, disability, and education.
In the meantime, the program has been working on new technologies to facilitate enrollment, such as saliva kits to allow participants to collect their own sample at home, and the ability to collect small amounts of blood for the project from blood bank donors.
While the All of Us program is slowly building up steam, other population health resources have continued to expand and provide geneticists with large-scale data for their analyses. The UK Biobank, for example, which has about half a million participants, recently increased the number of exomes in its database to 200,000. In addition, it has genotyping data available for all its members.
However, All of Us is not aiming to play catch-up with the UK Biobank or similar projects. "We've certainly learned a ton from them," Denny said. "We're not competing with them but [ours] will be a nice complementary dataset to what they have."