Q&A: Forthcoming tool will use genetics to help eliminate ‘diagnostic odyssey’

Geisinger Health System is working to develop a tool that it said will dramatically lower the time it takes to diagnose rare conditions with a genetic cause.

The Pennsylvania-based health system said it received a $5 million grant from the NIH’s National Human Genome Research Institute for its work on the tool.

Some of the diseases the tool will initially help diagnose include heritable CVD, single-gene type 2 diabetes and congenital kidney disease, Marc Williams, MD, a professor at Geisinger’s Genomic Medicine Institute and principal investigator on the project, told Healio Primary Care. This list will likely grow over time, Williams said.

“Complex diseases frustrate patients and create a burden on health care systems through multiple hospitalizations and frequent testing,” he said in a press release. “Enabling physicians to access genetic information in real time could prevent much of this burden by eliminating the gap between onset of symptoms and genetic diagnosis.”=

Williams provided more information about the development of the tool in an interview with Healio Primary Care.

Healio Primary Care: What led you to seek this grant?

Williams: A vast majority of the conditions we treat in genetics are rare or extremely rare. Therefore, when an individual with one of these conditions presents to a health care delivery system, the condition is not the first thing that most clinicians think about when they’re trying to put together what’s going on with this individual patient.

This leads to a diagnostic odyssey where these patients see lots of different doctors with lots of different tests and empiric treatments. It can take a very long time for them to get a correct diagnosis, if they even get one at all. As we move into an age where genomic information is more available, and even the time when everyone has their genomic sequence available, the question became how we could use that information to create a tool that shortens or perhaps even eliminates this diagnostic odyssey.

Healio Primary Care: What exactly is the tool? How will it work?

Williams: There are three major pieces to the tool that we’re going to be exploring in this grant.

No. 1 is, can we use the electronic health record to identify a pattern of things, whether they be laboratory tasks, results from an imaging study or clinical notes, that would lead us to think that this individual might have a genetic condition, or at least has a higher risk for having a genetic condition?

The second piece is a genetic analysis. Assume an individual has a genomic sequence that increases their risk for a certain disease, such as pediatric epilepsy. We want a way to send information from the individual’s electronic health record to this genetic analysis tool that analyzes all the genes associated with pediatric epilepsy to see if there is a variant in one or more of those genes that might be causing the seizures.

The third piece is sending a suggestion about the possible diagnosis to the individual’s clinician based on the variants found. The piece will also recommend a confirmatory test regarding its suggestion and some medications that the clinician might want to try based on the genetic analysis.

Healio Primary Care: How could the tool revolutionize the way physicians diagnose certain diseases?

Williams: The tool will create an effective interface between a computer and a clinician to assist in diagnosis and create alerts to help with their workflow based on their preference.

However, we know that if we simply send an alert to doctors, that is usually not sufficient to result in a change in practice. So, we also want to understand and then build a tool that allows clinicians and computers with their respective, complementary expertise to be most productive.

There have been quite a few articles that talk about how a computer does better than a doctor at reading X-rays, pathology slides or something of that nature. What these articles never focus on, because it’s not as interesting, is that when the clinician and the computer work together, it outperforms the clinician or the computer alone. That to me is where this tool will revolutionize medicine, when we bring both the clinician and computer together to take advantage of the strengths of each.

Healio Primary Care: Research suggests that primary care clinicians are cautious about the clinical benefit of genomic medicine . What do you want them to know about the tool?

Williams: As part of developing the tool, we will ask clinicians about their experiences diagnosing genetic conditions. What about it has been good? What things can we do better? What are the tools you need to practice genomic medicine?

We want to be able to identify the current pain points that clinicians are experiencing in genomic medicine, what we can do to improve those and how we can be sensitive to the way clinicians do their work.

The reason for a lot of the reluctance that clinicians have for innovation is that we tend to come in with a pre-configured idea about how a new technology should work. But it really doesn’t address problems that impact clinicians or reflect how clinicians are doing their work. So, what a lot of innovation ends up doing is increase workload and does not result in that much improvement, because we’re not using the clinicians’ expertise as part of the design. We really want to bring a very robust user-centered design approach to this tool, because we know that if we engage the clinicians from the get-go, the likelihood they will come out with something at the end that is going to improve their life is much higher.

Healio Primary Care: How soon do you think the tool could be available for physicians outside of Geisinger to use?

Williams: Maybe if we’re lucky and the wind is with us and we get the right team assembled, maybe we’ll have a prototype that we can test before the end of the 5 years. But the real idea here is to do all the heavy lifting within those 5 years, so that the next project would be to create it. That’s sort of a very unsatisfying answer for you, and to some degree, it’s unsatisfying for us too. But we also knew that just given what we were proposing to do, it really would not be realistic to move this tool into an actual health care environment in that amount of time. Just getting the permissions to kind of play around in the electronic health record will take a very long time, because we cannot disrupt all the work the electronic health record is already intended to do at the same time.

We will also realize we probably will not have a one size fits all solution and that there will be differences by specialty, because a cardiology condition is very different than a neurologic condition or a rare condition that affects kidneys. We also have multiple institutions that are involved, and so there will be institutional differences.

Based on 20 years of my prior experience, knowing how hard each of the tool’s different components are and what we’re proposing to do, at the end of 5 years, we want to at least have a deep understanding of every part of the tool that I’ve described.

References:

Carroll JC, et al. Front Genet. 2019;doi:10.3389/fgene.2019.01189.

Geisinger awarded $5 million to develop diagnostic tool for genetic disorders. https://www.eurekalert.org/news-releases/929339. Published Sept. 23, 2021. Accessed Oct. 2021.