NEW YORK (GenomeWeb) – Any person living in America, who is willing to share their health data, give a blood sample, and be recontacted for research purposes should be allowed to join the government's Precision Medicine Initiative (PMI), an expert working group has advised the National Institutes of Health.
As part of the initiative announced by President Obama in January, the NIH is slated to receive $130 million to put together a cohort of a million or more people whose data —medical records, genetic and metabolomic profiles, microbiomes, environmental exposures, and lifestyle habits — will fuel research and hopefully advance understanding of diseases. This week a working group of experts from academia, government, and industry unveiled a 100-page report outlining how NIH might assemble this research cohort.
"With robust implementation, the working group expects the PMI Cohort Program to be able to recruit at least one million participants over about four years," according to the report. NIH Director Francis Collins is hopeful the PMI, which is still awaiting congressional appropriations, can begin enrolling participants as early as next year.
A national research pool of 1 million people could be used to gain insights into a range of healthcare questions — from figuring out how genetic and environmental factors conspire in causing diseases to collating data from mobile health gadgets to discover associations between physiologic measures, the environment, and health outcomes.
For example, in a population of 1 million Americans, researchers would be able to study 5,000 cases with 261 diseases, 10,000 cases with 136 different diseases, and 20,000 cases with 59 illnesses, the workgroup estimated. If the cohort continues to enroll volunteers, in a decade researchers will have access to 20,000 cases for many of the common causes of death in America. Overall, the working group calculated that a PMI cohort of this size will have sufficient power to elucidate how different genetic and environmental factors are linked to hundreds of health outcomes.
But in order to conduct such queries using data from so many people, the PMI needs this data in electronic health records (EHR), needs to follow people longitudinally, and must use up-to-date technologies to analyze the data, all while protecting participants' privacy and containing costs.
Recruitment strategies
To start, the PMI working group recommended NIH allow "any individual living in America" to partake in the national project who are, among other criteria, willing to share their data either by directing their EHR to the PMI or by undergoing a health exam with a healthcare provider. This is perhaps the most exciting part of the report for Russ Altman, a professor of bioengineering, genetics, and medicine at Stanford University.
"We need to break out of the same old, same old," Altman told GenomeWeb. "Direct participation is clearly the alternative in this age of consumer empowerment." This direct enrollment option, while it has its own pros and cons, could make PMI a higher profile project, he said. "You could imagine that the direct participants would make decisions about who they want to give access to their data that are independent of the PMI effort."
Altman is also a member of the Advisory Committee to the [NIH] Director, which unanimously voted on Thursday to accept the recommendations in the report. "It's a really good read," Collins said during a meeting in which the working group presented the report to the committee. While NIH searches for a director for the PMI, Collins appointed as interim head Josephine Briggs, current director of the National Center for Complementary and Integrative Health.
Although the recommendations of the working group need further investigation and flexible interpretation in some cases, Collins noted that NIH is ready to pursue some new recruitment and research approaches not typically undertaken by the institute.
In addition to allowing individual Americans to directly enroll in PMI, the working group also suggested NIH collaborate with healthcare provider organizations (HPOs), particularly those entities that serve diverse populations, and have EMRs and capabilities to follow their patients long term. These HPOs, would recruit participants, consent them, conduct a baseline exam, collect blood samples and share EHR information with the PMI in a common format.
This method, the workgroup said, will initially be the fastest and most cost-efficient method of enrollment. If 10 HPO sites enroll 25,000 participants per year, then the PMI will reach 1 million volunteers in four years. Some HPOs that fit the bill include the Veterans Affairs' Million Veterans project with 6 million people in the HPO system and 400,000 biobanked samples; Kaiser Permanente with 10.1 million people and 245,000 stored samples; and Geisinger MyCode, which has had 1.3 million people's EHR information in the last decade and more than 86,000 banked samples.
After recent legislative mandates, 95 percent of US hospitals have implemented EHRs, and the working group considers the broad availability of EHRs critical to recruiting and following patients cost-effectively within PMI. "The success and longevity of the PMI Cohort Program will be heavily influenced by the laws, regulations, and policies surrounding research, data security and privacy, and access and interoperability of EHRs," the working group noted.
However, there is widespread concern currently in the field about the limited or restricted ability to share data between different EHR systems. "If you've seen one installation of Epic, you've seen one installation of Epic," Altman joked. "They do diverge very quickly."
As such, Altman anticipates that managing interoperability issues when drawing on data from HPOs may be a costly aspect for PMI. "If you don't do that right, you're in big trouble in terms of your ability to say you have [a cohort] of a million," he said. "The HPOs are going to have to promise to share certain types of data, and they can only do that when they confirm they have the right to pull that data … and then send that curated data to a central [PMI] site that will do some or a lot of additional curation."
One of my worries was that current players would simply pool their [existing cohorts] to get it up to a million, but there would be no statistically valid sense in which that's a sample.
Sequencing a million samples
The collection of a blood sample from every participant should be the highest priority for the PMI in terms of biospecimen collection, according to the working group. Other types of samples that the project could also ask for from participants include microbiome specimens or a clip of hair. The working group advised NIH to store the collected samples in a central biorepository that should be up and running by the time patient recruitment begins.
When the president first announced the PMI, it seemed that that government's plans weren't to try to sequence a million people, but enroll cohorts that already had some genetic testing or even had their genomes sequenced. The project now seems to be moving away from that in favor of conducting its own, standardized analysis on the samples submitted by PMI participants. "We don’t need a national initiative to glue together existing studies," Altman said.
"One of my worries was that current players would simply pool their [existing cohorts] to get it up to a million, but there would be no statistically valid sense in which that's a sample," he added. "In other words, you have one study that's doing cardiovascular conditions, another that's doing cancer. They each have 100,000 people, and you mush them together and say you have 200,000. But really you have a biased cardiovascular disease and cancer sample, and it's not representative of anything."
The working group said that PMI leaders would need to figure out the "tipping point" for when lab capabilities and costs enable genomic sequencing and other types of molecular analysis on a million people or more. Initially, high-density arrays might be a cheap way (estimated to cost less than $50 per sample) to gather genotyping data. These arrays should test for known pharmacogenomic variants, the report noted, so that research on these markers could be useful in personalizing treatments for study subjects in the future.
Collins acknowledged that although there was a sense initially that the PMI would be a genomics-focused effort, the scope of the national project is much broader and the aim is to collect all types of data. He noted that the PMI will collect genomic information from individuals if they have it, but there is also interest in doing a variety of analysis within the PMI, including genome sequencing, as costs come down.
But given the highly competitive nature of the sequencing market in terms of capabilities and pricing, Altman guessed NIH would likely "play chicken" with platform vendors and choose the company that would analyze the cohort "at the absolute last minute."
Standards and privacy
Consistency of standards will also be critical in a study of this magnitude. The working group suggested that PMI develop a set of principles that apply to all PMI collaborators; use a standard consent protocol; collect a core set of data from EHRs, health insurers, research studies, and mHealth technologies; and establish a data coordination center. PMI participants should be able get their own results, as well as aggregated results from the full cohort, and they should be able to set and change their preferences to dictate how much information they get back from researchers.
However, they recognized that the return of annotated genetic data to participants might be particularly challenging and costly due to regulatory requirements for lab tests and the US Food and Drug Administration's evolving oversight in this area. Additionally, family history information and genetic counseling will be critical for interpreting genetic test results for study participants. "For these reasons, it is important that the PMI Cohort Program properly budget for specimen collection, genetic testing, and return of results," the working group said in the report.
One alternative, they said, was to tell participants who wish to know that an actionable result has been found that they can be retested in a CLIA-certified lab for specific results. This is the approach that the NorthShore University HealthSystem in Chicago is taking in implementing its Genomic Health Initiative, a study that has so far enrolled 8,000 subjects. The working group advised that a subcommittee should explore issues related to the return of results, for example, how to inform study subjects when they have markers predictive of adverse outcomes (i.e. BRCA mutations).
In order to ensure privacy, anonymized information should be accessible to researchers with the data accessible via "secure computing environments," the working group said. They also suggested that the PMI provide open-access summary data for research but establish controls to limit who can access volunteers' individual information.
Additionally, the working group said NIH should establish an internal framework of policies for the program to address privacy issues, security, potential misuse of information, and institutional review board issues. For example, there must be safeguards against "unintended releases of data" and penalties for revealing the identity of a participant without consent.
The NIH should ask all users of the PMI data to sign a certificate of confidentiality that restricts them from disclosing participants' information in civil, criminal, or legislative proceedings. "Further, NIH should seek legislation to strengthen Certificates of Confidentiality to ensure that disclosure by researchers is not optional, other than with consent or for certain public health exceptions," the working group said. "This will be critical to ensure that data on PMI cohort participants is not used for any purpose other than research."
When you open the door to the nation, you don't quite know what's going to happen.
Diverse, like America
Throughout the report, the working group emphasized that the 1 million-participant cohort should reflect the diversity of the US. According to the report, 96 percent of participants in recent genome-wide association studies are of European descent. And while African‐Americans comprise 12 percent of the US population, only 5 percent partake in clinical trials, as do only 1 percent of Hispanics, who are 16 percent of the population.
The PMI must be designed from the start to "ensure that people historically underrepresented in biomedical research are included in sufficient numbers to allow robust inferences in these groups," the experts in the working group wrote in the report.
"The US population includes people with extreme wealth and people living in abject poverty and with varying access to social, educational, and other resources. As a nation that historically includes indigenous people, people who originally migrated or were brought to this country from different parts of the world, and more recent and new migrants from numerous countries, the US benefits from the richness of ancestral diversity," they said. "The PMI cohort is intended to be a national resource that will, over time, benefit the entire US population."
For special populations, such as children, decisionally impaired individuals, and cohort members who are jailed after enrollment, the NIH should develop ways to make sure they're included and retained in the study, the report advised. Also, the outlined enrollment methods would accrue patients with a number of very rare diseases that wouldn't be represented in enough numbers to allow their study. The working group said that disease specific repositories and recruitment methods independent of the PMI Cohort program will be needed for studying rare conditions.
Also, experts noted that the PMI should enroll any American living in the US, and participants must be in the country in order to remain in the study. "Scientifically, individuals would not necessarily have to be a US citizen or permanent resident, and as such enforcing strict verification of a participant's citizenship or residence was not considered a priority" for the cohort, the working group wrote.
Ultimately, it remains to be seen after NIH builds the infrastructure for the PMI if the American public will join. According to a survey of 2,600 people by the Foundation for the NIH, 80 percent supported the development of the PMI cohort, more than half said they'd "definitely or probably participate in a study if asked," and a similar percentage said they wanted to be involved in deciding things like which research to pursue and how results should be shared.
There are more than 300 million people in the US, and not all need to sign up for the PMI for it to be a success, Altman noted. But would even one out of 300 people in the US "get psyched" about being part of a historic cohort for health? "I think those people are out there," Altman said. "We just have to make sure they're not the same type of people."
NIH leaders haven't quite figured out how they would manage too much interest in the PMI and ensure that certain demographics aren't overly represented. "When you open the door to the nation, you don't quite know what's going to happen," Collins said.