CHICAGO – With a new subsidiary called the Unified Patient Network (UPN), Seven Bridges Genomics seeks to allow multiple healthcare systems to combine their deidentified clinical patient data with pharmaceutical industry-sponsored whole-genome sequencing to advance research and return patient-specific test results.
Seven Bridges unveiled the UPN last week, announcing Washington University School of Medicine and BJC HealthCare, its affiliated health system, as founding partners of the network. Also participating is telegenetics firm Genome Medical. Going forward, Seven Bridges expects to add more health systems as partners, as well as biopharmaceutical companies to fund genome sequencing and research projects.
Seven Bridges CEO William Moss is serving as interim CEO of the UPN. The firm also appointed David Ledbetter, formerly of Geisinger Health System, to serve as the network's chief clinical and research officer.
Ledbetter said there are numerous private and academic networks for sharing data from electronic health records. "All of them, until UPN, have postponed the inclusion of genomics data because nobody can afford to do the sequencing," he said.
For example, the UK Biobank is a "tremendous research resource," he said, but is unable to return medically actionable results because consent forms were not written to allow that. The US National Institutes of Health's All of Us program, on the other hand, does permit return of results, but that program has been moving too slowly for Ledbetter's taste.
UPN addresses the affordability problem by matching health systems to pharma companies willing to sponsor sequencing as part of their research.
Some earlier, private-sector networks that UPN leaders have direct experience with have been closed to outside collaborators.
Ledbetter, for example, spent 10 years at Danville, Pennsylvania-based Geisinger Health System as founding CSO. He also ran the MyCode biobank there, which included a collaboration with Regeneron Pharmaceuticals to sequence patient exomes.
When he moved to Seven Bridges in January, Regeneron had completed exome sequencing on more than 175,000 Geisinger patients, he said.
MyCode was among the first biobank genomics programs in the world to return "medically actionable" sequencing results to patients and their clinicians, according to Ledbetter. "We got a lot of criticism for that when we announced it in 2013," he added, but that model has become the standard for All of Us.
About two years ago, Moss contacted Ledbetter to ask about the Geisinger-Regeneron collaboration because many of Seven Bridges' pharma customers were asking for help identifying patient cohorts that included both longitudinal clinical data as well as genomics information.
The Charlestown, Massachusetts-based bioinformatics firm quickly learned that "pharma would pay quite a bit of money to get access to patients that had rich clinical data and whole-genome sequence data, and the amount they're willing to pay is more than it costs to do genome sequencing," Ledbetter said.
The Geisinger-Regeneron collaboration had limitations, though. Regeneron operates a research laboratory without CLIA/CAP certification, so the Geisinger program cannot return results to patients without confirmation in a clinical lab. "We had to independently verify every BRCA pathogenic variant at our expense, and that was clunky, time-consuming, and expensive," he said.
Another issue was that Regeneron created a proprietary database for this collaboration. When Geisinger wanted to bring on other healthcare systems, the pharma firm balked at the idea of setting up a common data warehouse for sharing deidentified clinical and genomics data among multiple health systems, according to Ledbetter.
Also, when Regeneron entered into its own collaborations with other care providers, Geisinger researchers were unable to access data from those partners because the information was locked in silos, he said.
"I come from an open-science and broad data-sharing background," said Ledbetter, who was a founder and principal investigator of ClinGen and the ClinVar database. He was drawn to the UPN idea because Moss wanted to encourage data sharing across all participating health systems.
Geisinger actually withdrew from All of Us and returned a grant in 2017 because the NIH program was not going to use the health system's EHR data, nor would Geisinger investigators be able to see clinical, patient survey, or genomics data until the information was released to the public. "It was actually a step backwards from what we were already doing with MyCode and Regeneron," Ledbetter said.
All of Us still has not released any medically actionable genomics data to researchers and will not do so until sometime next year. UPN wants to move more quickly. "We're eager to make the genomics data available to the member health systems essentially immediately, and develop a model to make it available to the broader academic research community," Ledbetter said.
UPN improves on the Regeneron-Geisinger model with whole-genome sequencing instead of exome sequencing, access to CAP/CLIA-certified labs, immediate return of clinically actionable test results, and open data sharing, he said.
The network will be using the Health Level Seven International Fast Healthcare Interoperability Resources standard to ingest data for early-stage cohort stratification and the Observational Medical Outcomes Partnership framework for the resulting longitudinal dataset.
Further details for data sharing will not be worked out until UPN members set up a governance structure, but Ledbetter expects that to start early next year.
Each member healthcare system will have a seat on UPN's executive committee that will make the rules and regulations for data sharing. Ledbetter said there will be discussions on how to share aggregated datasets with any qualified academic researcher, whether or not they are affiliated with a participating health system.
UPN will include data on multiple disease states. While pharma companies will define many of UPN's cohorts, Ledbetter said that he is interested in broader studies in population genetics and prevention.
Initially, the network will be looking for adult patients, but it wants to involve children's hospitals in the near future to introduce a pediatric component. BJC HealthCare does include St. Louis Children's Hospital.
BJC and WashU's participation is led by Philip Payne, chief data scientist at Washington University School of Medicine and director of the school's Institute for Informatics.
As an informatician, Payne said, he thinks about avoiding data silos in every project he works on, and UPN's choice of open standards for acquiring, organizing, storing, and sharing data helped assuage any such concerns.
Payne said that the UPN data platform is not terribly different from those of All of Us and the NIH's National COVID Cohort Collaborative, for example. "That gives me confidence that we're not reinventing the wheel," he noted.
While WashU has not settled on any specific research programs for UPN yet, Payne said that the institution does have several priority areas, namely solid-tumor cancers that are highly prevalent in Missouri and neurodegenerative diseases, a research strength of the medical school. He also mentioned cardiovascular disease and rare pediatric conditions as potential focus areas.
Payne said he had been approached by various other organizations that were interested in establishing networks around sequencing and phenotyping large cohorts of patients for research purposes, but UPN was the first he saw as a true partnership for WashU and BJC. To him, that means that investigators have joint scientific oversight of the projects, along with the research network.
Payne was also attracted by the opportunity to work across health systems, as well as by the ability to offer clinical-grade whole-genome sequencing to more patients. "That would help us to address an access and affordability issue for precision medicine," he said.
WashU, BJC, and future participants will send samples to UPN for sequencing, and the network will send the results to a shared data repository. The health systems will obtain results that meet criteria for being clinically actionable for sharing with patients and will be able to offer patients remote counseling services via Genome Medical.
South San Francisco, California-based Genome Medical also has ties to UPN leadership. When Ledbetter was still at Geisinger, Moss asked him if that health system could provide support for return of results, including with genetic counseling, but MyCode only had enough personnel to accommodate Geisinger's own needs. However, Geisinger had trained a group of people capable of providing support to other healthcare organizations, but they were hired away by Genome Medical in early 2020. That group included Genome Medical CSO and senior VP of medical affairs, Huntington Willard; VP of population genomics, Erica Ramos; and Beth Denne, director of education and engagement.
Seven Bridges' decision to work with Genome Medical made UPN more attractive to Ledbetter, he said. The telehealth component will help with genetic counseling for patients, their primary care physicians, and any specialists people might be referred to based on their genetic risk.
Ledbetter said that UPN is in "late-stage but not final discussions" with two other major health systems, plus in earlier-stage negotiations with other provider organizations. No pharma companies have joined the network yet. UPN will first need to sign up a few health systems and demonstrate those partners' ability to develop patient cohorts that would be useful in biopharma research, he said.
In the meantime, UPN is setting up an in-person meeting with Payne and his team in St. Louis soon and expects to launch the program at WashU officially by early spring. Payne said that the UPN would like to choose its initial studies by the first quarter of 2022, so WashU and BJC can begin early recruitment in Q2 and Q3.