March 2018

Precise, Purposeful Health Care
By Dava Stewart
For The Record
Vol. 30 No. 3 P. 16

A health system's ambitious project dives into the genetics of healthy patients.

The promise of precision medicine depends in large part on technology, and nowhere is that more apparent than when it comes to genetics. In that vein, the MyCode Community Health Initiative, a project at Geisinger locations in Pennsylvania and New Jersey, explores how genetic testing can be applied to primary care.

Although various types of genetic testing have been available for some time, the MyCode initiative is one of the first efforts to sequence the DNA of a large number of mostly healthy people. It's also a pioneer of integrating genomic screening and counseling into routine health care.

The developing technology to actually perform the sequencing, methods for storing results, decision support tools for clinically actionable findings, and many other factors are being studied as a result of this wide-ranging initiative.

Thus far, more than 180,000 patients have consented to participate in the program. The goal is to enroll all of the more than 3 million people in Geisinger's system. When patients visit their primary care providers, they are asked to consent to the research project with the goal of sequencing becoming part of routine care.

The ambitious project spotlights opportunities for improving the health of individual Geisinger patients and population-level health as a whole while revealing gaps in HIT infrastructure.

Storing the Results
Once patients consent to participate, a blood sample is taken and whole exome sequencing is performed. The American College of Medical Genetics and Genomics classifies 59 genes as clinically actionable, with an additional 21 others recommended by Geisinger. If a pathogenic or likely pathogenic variant is found in one of these 80 genes, the patient and the primary care provider are notified.

The vast majority of human DNA is identical—there is only a 0.1% to 0.3% difference between individuals. Of that percentage, scientists understand only a few of many possible variations. These are classified as clinically actionable within the parameters of the MyCode Initiative.

At Geisinger, the results of the whole exome sequencing are separated into three files. "You have the file of the raw results, then the file of differences, then the file of differences we understand," says Andy Faucett, LGC, director of policy and education in the Office of the Chief Scientific Officer at Geisinger. The sequencing generates a massive amount of data, of which scientists understand only a small fraction and primary care physicians even less.

The raw, uninterpreted data are stored in the cloud. The file containing genetic variations is interpreted by experts. If there is a clinically actionable variant present, the primary care provider and the patient are informed. Approximately 3.5% of the samples have shown predicted deleterious genetic variants.

In those cases, a notation is added to the problem list in the patient's EHR. The lab report containing the genetic variant and its associated condition is scanned into a PDF, which is also stored in the EHR. Geisinger is currently working with Epic and Cerner to develop a better, more usable system for incorporating the genetic information into patients' EHRs.

Faucett says the sequencing results are treated much like any other problem or condition. For example, both a heart attack and a BRCA variant—a breast cancer susceptibility gene—are noted in the EHR's problem list.

Geisinger does not depend on only the PDF because "it is unlikely that people will understand the results and act appropriately, so we have a whole structure with messaging to physicians and to patients," says Marc S. Williams, MD, director of the Geisinger Genomic Medicine Institute.

It can be argued that the MyCode Initiative spotlights the industry's struggles with interoperability. For example, although the population in the geographical area served by Geisinger is more stable than the populations in many other places, patients who have been sequenced face major roadblocks when attempting to transfer their genetic data to another health system.

A lack of decision support is also a hurdle. According to Nephi Walton, MD, a pediatric geneticist and an assistant professor at the Geisinger Genomic Medicine Institute, although the project provides primary care providers with more detailed patient data, it would be ideal if certain variants triggered automatic decision support. For example, certain genetic variations can change the way particular drugs metabolize. When a physician prescribes one of these drugs to patients with those variations, it would be helpful if decision support was triggered automatically.

As it stands now, the information resides in a PDF in the patient's EHR, where physicians must locate it manually. Walton says such data can't be buried in the EHR. "It has to be front and center," he says, "It's critical for it to be integrated into the EHR, and it will be even more critical in five years."

Aaron Neinstein, MD, an assistant professor in the division of endocrinology and director of clinical informatics at the University of California, San Francisco Center for Digital Health Innovation, says the situation is similar for patients who have diabetes and other diseases. "The information [from the genetic testing] needs to be indexed and layered seamlessly," he says, adding that clinicians must be able to simply "click" to learn more.

Returning the Results to Patients and Providers
When a variant in one of the 80 clinically actionable genes is identified, the patient is contacted via Geisinger's patient portal, telephone, or in person and then given the opportunity to speak to a geneticist or genetic counselor to learn more about the test results. The primary care provider also receives the results as well as additional information about the findings.

Some patients do not wish to pursue information regarding their results. In those cases, Faucett says primary care providers treat these individuals just as they would any other patient. For example, take smoking. "If you see that a patient is a smoker, you are going to talk to the patient about it every time you see them—even if they tell you they are not ready to quit," Faucett says, adding that clinicians may recommend different tests at various stages for smokers compared with nonsmokers.

Similarly, patients with a BRCA variant that puts them at a higher risk of developing breast cancer are encouraged to undergo mammograms at an earlier age. "The things you would recommend would change over the patient's lifetime but it would be one of those things the physician would mention," Faucett says.

To provide information about the genomic results more easily, Geisinger has developed GenomeCOMPASS, a reporting system featuring separate interactive interfaces to help both patients and providers decipher the results.

Few people have the skills, knowledge, and training to interpret the results from genetic testing. However, as Faucett points out, "If a patient gets an MRI, the primary care physician doesn't interpret it—the radiologist does."

Primary care providers can be overwhelmed with information—most prefer it to be "just in time" rather than receive a massive amount disconnected from a clinical need. To help primary care providers get only the data they need, GenomeCOMPASS offers fact sheets, video courses, and access to Geisinger's genetic screening and counseling team. "Doctors want to help patients follow the recommendations of the experts," Faucett says.

The patient-facing side of GenomeCOMPASS, which is accessed through the Geisinger patient portal, has generated enthusiasm among users. "We are pioneering a model for sequencing healthy people and what to do with the information that it generates," Walton says.

The MyCode initiative is helping to answer two important questions about genetic testing: Can primary care providers use genetic testing results to improve care, and are outcomes improved when patients are aware of genetic variants?

Because the initiative is the first of its kind, Geisinger is building a model of how incorporating genetic tests into primary care might work. "It's not an easy process to figure out how to take all the data, process, decide what to return, how to return it, and how to do it with limited resources," Walton says. "One of the things we are discovering, and not just through Geisinger, is that we don't know as much as we thought we knew about genetic changes."

For example, patients with variants that would lead experts to believe would cause disease have not all followed that course, which raises questions regarding genetic testing on a larger scale, an issue Geisinger is taking to heart.

In a paper published in December 2017 in the Journal of Genetic Counseling, Geisinger researchers examined the effectiveness of GenomeCOMPASS for families with children with undiagnosed intellectual disability, autism spectrum disorder, and/or anomalies. The study compared families that received the GenomeCOMPASS report (the intervention arm) with those who received a summary letter from the medical geneticist (the usual care arm).

Although the sample size was small, there were two interesting findings. First, the GenomeCOMPASS report led to increased engagement and satisfaction among the patients' families. Second, the researchers found that reporting negative findings, or uninformative results, was not useful.

Learning More About What Is Not Known
Walton prefers the phrase "precision health" as opposed to "precision medicine" because, "We are taking people who are not sick and using sequencing data to improve their long-term health." With large numbers of healthy people being sequenced, the MyCode project is bringing previously unasked questions to the fore.

For example, the initiative will help show whether such large-scale testing can be cost-effective and lead to improved outcomes for patients. "We need to know the implications of the changes we are finding," Walton says, adding, "We don't want people to take unnecessary measures or get overly anxious about them when we don't know the impact they are going to have clinically."

One of the most important aspects of the MyCode Initiative is the research platform being built by Geisinger. Neinstein says by having collected genetic profiles for a decade, the project will give clinicians and researchers an opportunity to match phenotype with genotype. "[MyCode] creates a lot of power for following the development of the disease," he says.

The research is important for numerous reasons. For example, Walton says genetic testing could make an enormous impact and change the way specific patient populations, such as those in newborn ICUs and those with neurological conditions, are managed. In some cases, payers are willing to cover numerous expensive tests but will not pay for sequencing. "We could really eliminate thousands of dollars' worth of unnecessary tests," Walton says, adding that the MyCode Initiative is building a body of evidence that may help demonstrate the value of genetic testing.

Currently, the Centers for Disease Control and Prevention recommends genomics to improve public health for three Tier 1 conditions: hereditary breast/ovarian cancer syndrome, Lynch syndrome, and familial hypercholesterolemia. Williams says MyCode Initiative researchers have already learned that about one-half of the population of Geisinger patients who likely have the pathogenic criteria for hereditary breast/ovarian cancer syndrome fail to meet the testing guidelines. In other words, many who could benefit from screenings are not getting referred. "We estimate that about 1% of the population will have a pathogenic variant in one of these three Tier 1 conditions," Williams says.

While the MyCode initiative is defining definite gaps in HIT, its research platform is helping to shape the view of genetic testing. Clinical trials and studies conducted using data from the initiative are demonstrating the impact of sequencing and providing scientific evidence to back up the claim that genetic testing can save money and improve patient outcomes.

— Dava Stewart is a freelance writer based in Tennessee.