Genetics lands in primary care inboxes

He points to several broader highlights as well. “For many Mendelian genetic diseases, when you screen a population by genetics, by sequencing the genes—BRCA1 and BRCA2, for example—the prevalence is double that reported in the literature through clinical ascertainment. And fully 50 percent of individuals do not meet the current guidelines, based largely on family history, for ordering a BRCA genetic test.”

When they started returning results five years ago, many in the genetics and biomedical ethics communities expressed concern about the possibility of psychological harm to patients who were unaware of their risks. Others feared it could lead to unnecessary testing. But, Dr. Ledbetter says, patient response has been almost universally positive, and it was matched by enthusiastic primary care physicians—to the point that in May 2018, Geisinger’s CEO suggested expanding genetic testing beyond the subset of MyCode volunteers to offer it to all Geisinger patients.

In July 2018, two primary care clinics began offering clinical exome sequencing to patients. Some 500 have undergone testing so far. And similar to the MyCode experience, patient and PCP response has been positive. “We’re pretty enthusiastic about continuing to expand,” Dr. Ledbetter says, although cost remains an obstacle to offering it to all patients right away. (Patients are not charged for the testing.)

One irony of precision medicine is that while it ultimately ends with the patient, it doesn’t start there. As Geisinger’s Melissa Kelly, MS, LGC, explains it, “We’re saying, almost irrespective of what your health status is, we’re just going to look at what variants you have and see which ones put you at risk. Or, explain a disease, but not using health status to inform what we’re looking at,” says Kelly, variant scientist, Geisinger Genomic Medicine Institute.

Making variants the focus helps tame the enormous scope of projects like MyCode. And by targeting variants that are “beyond reproach,” Kelly says, “it allows us to quickly filter down these massive numbers of variants and patients to a much more manageable number. When we find a variant, we know who’s at risk, we know what to recommend. That’s where we are right now.”

But later? “It’s an interesting question as we start to expand down the road. What other types of results would have utility for patients?” Kelly asks. Should standard carrier screening be included during reproductive years? “It’s less likely to change medical management, but it could change reproductive decisions.” What about risks for which there is no intervention? “We’ve decided not to approach those yet, but they’re definitely things we’re continually discussing.”

When Geisinger adds new genes, Kelly and her colleagues consider the CDC tier 1 genomic applications and the definition of actionability proposed by the ACMG’s secondary findings recommendations. But they also listen to their internal clinicians and researchers. “We basically allow clinicians and researchers to propose genes or conditions that they think would be valuable and meet some of those same criteria.” These genes are then thoroughly vetted.

And while a finding may be static, the knowledge behind the finding—or lack of findings—isn’t. Large screening projects will require regular queries to see what changes might be relevant to the patient population. If someone has a variant of uncertain significance, says NorthShore’s Dr. Hulick, “typically we have to be more passive in waiting for whatever lab issued the report to issue an amended report.” But “once we know what variants are in our system, we can query that against something like ClinVar [www.clinvar.com], so if some other lab has changed the classification, that may be a time when we instigate our own review.”

MyCode uses a very conservative bar to determine what pathogenic or likely pathogenic variants should be returned, Kelly says. Once the variant has been identified, it’s sent out for confirmation to a CLIA-certified clinical lab, which issues a clinical report. “That’s what we return to the patient, assuming we agree with their final classification,” says Kelly, who manually reviews every result.

Population screening versus diagnostic testing presents one possible classification hitch, Kelly says. “At least we think about it a little bit differently. If a patient has a disease and you find a variant that you think is causative, it almost starts you off at a higher a priori likelihood. But finding something incidentally in someone, particularly someone who doesn’t have any features, we feel you need a slightly higher evidence level to be able to return those variants to patients.”

Variant classifications encompass a spectrum of evidence rather than fall into discrete buckets. “A clinical lab might technically report a variant as likely pathogenic—but the evidence can range from just over the threshold from uncertain significance to just shy of pathogenic,” Kelly continues. “But we don’t want the variants that are straddling the border with uncertain significance. We want the variants that are in the middle or on the verge of being pathogenic. Because we only want to return variants that we have high confidence are putting someone at risk.”

She adds: “Collectively as a field, it will be interesting to see how this approach to variant classification evolves as more of this type of population screening happens, how much that approach is incorporated, and if there is confidence in that result not changing down the road. Our goal is to capture the variants that are so well vetted that no matter what we learn in the future, they will always stay pathogenic.”

Results are routed to patients through the genomic screening and counseling program that Sturm co-directs. “We upload the results into the electronic medical record and give them to patients’ primary care physicians, both internal and external to Geisinger,” Sturm says. A couple days after that, the genetic counseling team begins calling the patient-participants to provide them with their result.

“We send the report to their primary care physician in Epic, but we don’t release it to the patient in their MyChart portal,” Sturm says. “We make sure we call the patient to have a conversation with them first.” If, after three attempts, they fail to reach patients by phone, “we send them a packet in the mail and still ask them to call us.” Sturm and her team also refer them to specialty care and recommend family communication and uptake of cascade testing for all at-risk relatives.

Supporting the primary care physicians has also been crucial, and Sturm notes that these clinicians have been instrumental in deciding what they need; in fact, she leads a clinician advisory committee that meets every other month. Geisinger provides an on-call genomic counseling resource center, and physicians can ask questions through a number of portals.

That’s a far cry from early attempts, Sturm notes, which included lengthy CME modules to help PCPs learn about genetic conditions. No one was interested, she says. “What we heard is, Listen, I don’t need to become an expert in this condition. I just need to know the exact clinical next steps I need to take with my patient. And that may be referring them to a specialist.”

As Sturm noted earlier, anecdotal evidence has shown how the program can work for individuals. But Geisinger has also been doing health outcomes data collection. Those who’ve received results linked to familial hypercholesterolemia, Lynch syndrome, or hereditary breast and ovarian cancer syndrome appear to be using this information for positive health behavior changes, Sturm says, including cholesterol panels, colonoscopies, and mammograms. Those who opt for genetic counseling also show statistically significant correlation with higher positive behavior changes, she reports.

Her hope is that genetic screening will become the norm sooner in patients’ lives. She calls familial hypercholesterolemia “an example of what we are calling missed opportunity.” MyCode participants tend to be older and already have a significant degree of heart disease, stroke, and peripheral disease. Earlier identification and intervention would have prevented those diseases, she says. “I think this should be done at young ages,” she adds, noting that patients as young as eight can be put on statins.

Dr. Hulick thinks this type of testing will become a standard offering—though not a mandatory one—that’s dependent on a patient’s age and circumstances. While the entire genome might be sequenced, “you’re not going to access it all at once. But at key clinical time points in your life you’re going to tap into it.” This might include preconception counseling, or as patients near screening ages or anticipate a stage in life where they might begin taking medications.

While it remains important to find patients with single-gene conditions, those cases are a minority. “The real population impact,” says Dr. Hulick, will be combining polygenic risk factors with other traditional risk factors. “How do you start marrying all that information together?”

Like Geisinger, NorthShore has continually expanded its genetics testing. In April it launched the DNA-10K program, offering free genetic screening—using Color Genomics’ inherited cancer panel and cardiovascular panel—to up to 10,000 patients. Testing will also include several “fun” traits, or what Dr. Hulick calls “recreational genomics.” The last of the 13 participating primary care sites went live in early May. The endeavor also allows patients to sign up for NorthShore’s research arm, i.e. the Genomic Health Initiative.

Dr. Kaul reports there are currently about 20,000 samples in the GHI biorepository; about 30,000 patients have consented, which they do online through the patient portal. The order is automatically teed up, and the next time a patient has a blood sample taken, for whatever reason, an extra tube will be collected for the research study.

Says Dr. Kaul: “We are managing all the collections and DNA production and disbursement of samples. It’s like another lab test,” involving phlebotomists, bar codes, special biorepository software. “So the lab is deeply ingrained in all that.” (The lab is less involved in the DNA-10K study, she says, though it does perform collections.)

But that’s it for the lab, at least for now. Says Dr. Hulick: “We’ve worked quite a bit with Dr. Kaul in terms of prioritizing what we should do in-house.” Clearly the expertise was there, he says, but the capacity was not. “We do somatic testing in-house because turnaround time is more critical,” Dr. Hulick says, and NorthShore’s own lab can work with very small sample sizes and integrate genomic with other testing needed to optimally use the tissue or cells. For the population genomics testing, however, NorthShore has partnered with multiple outside companies, including Ambry Genetics, Invitae, and Helix, as well as Color.

Dr. Kaul says she was a bit disappointed when she first learned that the projects would be relying on outside labs. “But it makes sense,” she says, given the focus on population screening rather than diagnostics. “The throughput and price points are much different than we could accommodate, and much different than most hospital labs could accommodate.” Down the road, of course, it might be a different story.

At Geisinger, it likewise made sense to outsource the sequencing to Regeneron, says Dr. Ledbetter, but “We’re still evaluating what it will take to build a high-throughput sequencing capability within Geisinger’s pathology lab medicine department. They’re very interested,” he says, “but we are not yet at the point of doing this in-house.”

As the cost of sequencing drops and databases for interpreting pathogenic variants improve, will every hospital do its own sequencing? Or will this become centralized in a smaller number of national reference labs? Both approaches have their pluses and minuses, says Dr. Ledbetter. “I’m not sure yet where it’s going to end up.”

For now, Dr. Kaul says, labs need to understand a key point. “This is happening. We need to recognize there’s a lot going on outside our walls. Our colleagues in primary care do get patients coming in with a 23andMe report.” She says she’s thrilled to see her PCP colleagues embrace the programs that are bringing them into the center of this. “It’s great for our whole hospital system that they’re really engaged and are driving this.”

She remains struck by one novelty in particular. “Something that’s been interesting to watch here at NorthShore is that it’s not all about the lab,” says Dr. Kaul. Then again, in the years ahead it might be. “Someday, are we going to have whole genome sequencing on every bench? I don’t know. I used to think I would never utter the words ‘point-of-care PCR’ in the same sentence, and obviously that’s come to pass.”

“It may be that we have whole genome sequencers, with all the data analytics onboard, sitting in every lab in 10 years, and this will be commonplace and done in every hospital,” she says. Despite having spent 30-plus years in molecular medicine, Dr. Kaul says the field continues to amaze her. “It’s just mind-boggling. I have learned not to doubt any possibility.”

Karen Titus is CAP TODAY contributing editor and co-managing editor.