Geisinger Experience Implementing Precision Medicine Spurs New Effort to Help Others Do the Same

BOSTON – Geisinger launched an initiative this week to help health systems and organizations around the country integrate genomics to advance research and patient care.

The Geisinger National Precision Health Initiative will operate out of Washington, DC next year, and will be led by geneticist Huntington Willard, who was the founding director of the Duke Institute for Genome Sciences & Policy and is a former president of the American Society of Human Genetics. Through this initiative, Geisinger is hoping to apply what it learned from its MyCode Community Health Initiative on a national scale. 

In 2014, the integrated health system, which serves a largely rural population in Pennsylvania and New Jersey, launched the biobanking and genomic sequencing project, called the MyCode Community Health Initiative, in partnership with Regeneron Pharmaceuticals. The original goal was to enroll and sequence the exomes of 100,000 people to advance research into the genetic underpinnings of disease and  to spur new drug development. After quickly enrolling that many people, Geisinger announced this summer that it was setting a new goal of enrolling 250,000 participants.

Geisinger wasn't initially returning genomics results to patients but more recently decided to do so for around 76 genes, including those that its own experts have deemed actionable  and those that the American College of Medical Genetics and Genomics says should be returned to individuals. So far, the project has enrolled more than 160,000 participants, sequenced more than 92,000 exomes, and 500 individuals have received results. Regeneron in a couple of months will be sequencing around another 30,000 exomes, Geisinger Chief Scientific Officer David Ledbetter said.

Currently, when sequencing identifies actionable variants in the research setting, those findings have to be confirmed in a clinical lab before they can be reported back to patients, and Geisinger is working with the Laboratory for Molecular Medicine at Partners Healthcare Personalized Medicine in this regard. Approximately 3.5 percent of patients receive actionable results.

In a Genetics in Medicine paper published online last month, Ledbetter and others involved in MyCode described the impact of identifying pathogenic or likely pathogenic BRCA1 and BRCA2 genetic variants in 33 Geisinger patients who previously didn't have cancer and were at an age to receive risk mitigating strategies. Out of 26 people who underwent a screening procedure over a 12-month follow-up period, three patients were diagnosed with an early-stage BRCA1/2-associated cancer.

"Through this learning healthcare system approach, we can learn whether returning BRCA1/2 findings to unselected individuals offers clinical utility," the study authors wrote in the paper.

"The pattern that's emerging in our data and other people's data when you do genetic analysis is, the prevalence of disease-causing mutations putting people at high risk turns out to be double what most of the published prevalence is for most of the Mendelian diseases," Ledbetter said.

Currently, insurers use a patient's personal and family history of cancer to determine when to reimburse for cancer predisposition testing. However, research led by the University of Washington's Mary-Claire King has shown that 50 percent of women with deleterious mutations in BRCA1/2 have no family history of the disease. Based on this finding, King has been advocating for population screening for women starting at age 30 for pathogenic variants in BRCA1/2, in an effort to identify women at high risk for breast and ovarian cancer who are not being identified in the current healthcare system but who would benefit from risk-mitigating strategies. 

Ledbetter noted that the data emerging from MyCode not only supports this for BRCA1/2 screening, but also for other diseases, like familial hypercholesterolemia. Still, Geisinger has proceeded carefully.

"It's a conservative starting point in terms of the genes and conditions we're looking at, and we're very conservative in the clinical interpretation in terms of what we consider disease-causing variants," Ledbetter said. "It's because it's still early and we're learning how to do this."

Implementing precision medicine requires significant investment in genomics expertise, as well as infrastructural and operational changes, and so, most health systems around the country haven't gotten as far as Geisinger has. Around the country, when genomics approaches are being implemented, it's happening at the major cancer centers or at organizations and academic centers focused on investigating rare, unknown diseases.

"Going national means, can we take that approach that works and apply it to other populations and teach other health systems and doctors to do this?" Willard said. "It's sort of a turnkey operation. Think of all the hospitals that don't have even one geneticist on staff, and they don't have genetic counselors."

Although the specific solutions the initiative will provide clients haven't been fully ironed out, Willard suggested that one possible model could be that Geisinger partners with other groups to provide the genetics expertise they're lacking. Under another model, health systems would send patient samples directly to Geisinger, and after sequencing and data interpretation, Geisinger experts would engage with patients via teleconference.

In considering all the factors that have allowed Geisinger to successfully run a project like MyCode, it can't be ignored that the integrated healthcare system serves a highly engaged patient community, with around 85 percent agreeing to partake in the effort. The people Geisinger services are mostly Caucasian, live in rural regions, and don't move around much. The organization also has an electronic health records system and operates its own health plan, both of which facilitate more efficient integration of genomic medicine.

Less integrated systems, or health systems serving diverse, transient, and urban populations, may face unique challenges in trying to implement a MyCode-like project. "We'll learn as we go," Willard acknowledged. "We're not going to know until we try it."

The next "big step" for Geisinger will be to move all its sequencing out of research mode and into a clinical care setting, since the experience so far shows there is a clinical benefit for some patients. To do so, Geisinger will need partners other than Regeneron, with expertise in clinical sequencing, according to Ledbetter.

"There are some employers and healthcare systems that want to offer this as a clinical screening test," he said. "So, we're looking for ways to move it out of research sequencing and into the clinical care environment, [though] it won't be routine clinical care until we have better cost-benefit data and insurance will pay for it."

In an effort to address the difficult reimbursement environment for molecular diagnostics, Geisinger plans to employ policy experts who can engage the payor community in identifying new reimbursement models to evaluate the value of implementing genomic medicine on a population level, according to Willard. Ledbetter added that Geisinger is working on developing a framework for generating good cost/benefit data that will sway payors, though this has proven more difficult than he anticipated.

"There's a disconnect between what health economists are interested in and what insurance companies are interested in," Ledbetter said. "The health economists do a lot of modeling without coming up with a lot of useful practical information for the insurers. And insurers do their own actuarial analysis for coverage."

Geisinger's Washington, DC-based initiative will also operate a Precision Health Innovation Lab, where computational scientists and experts in artificial intelligence and machine learning will try to derive more research and clinical value out of all the data being generated within MyCode. "The key for precision health is the combination of comprehensive electronic health data and genomics data and, later, other omics data," Ledbetter said. "So, it is a big data challenge."

Geisinger's decision to launch its National Precision Health Initiative follows the health system's decision to drop out of the All of Us Research Program ― the NIH-backed effort to collect a variety of data, including genomics, from a million participants, and use it to fuel research and advance precision medicine. All of Us is currently in beta phase, testing out its interface for electronically collecting data from volunteers and mechanisms for importing electronic health records. 

Meanwhile, the project's ambitions to collect genomics data will likely take some time to implement, given the project's scale and the cost of sequencing, and because the initiative's organizers have decided to give further consideration to difficult ethical issues around returning results. Although All of Us has formed a genomics working group, that expert body has yet to release a plan for how it plans to collect molecular data from participants. 

Ledbetter previously said that Geisinger decided to step back from the government project because it became too difficult to participate in that and focus on MyCode. Moreover, Geisinger's precision medicine focus is very much on genomics, while All of Us seems, for the time being, to be more focused on collecting lower-hanging fruit, such as data from electronic health records.

"Our goal is to sequence every single patient who comes through the door at Geisinger," Willard said, acknowledging that multi-institutional government projects like All of Us tend to be slow to get up and running.

"[All of Us] will get there someday," he said. "Geisinger didn't want to wait for someday. Someday is today."