NEW YORK – As part of an academic-industry partnership with Thermo Fisher Scientific, investigators and clinicians at the University of Pittsburgh Medical Center are currently pushing forward with an effort to definitively demonstrate both utility and health economic value for preemptive pharmacogenomic testing — an area of genomics that has been greeted with much enthusiasm over the years, but which has also faced significant hurdles.
The effort, formalized with the creation of a PGx Center of Excellence by Pitt and Thermo in 2018, aims to analyze samples and return results to what will eventually be up to 150,000 individuals, measuring the impact of testing on healthcare utilization/spending to, hopefully, prove clinical utility and economic value, develop new standards for return of results, and inform best practices for patient education.
Because of its genomic scope — the program is utilizing Thermo Fisher microarrays that cover more than 4,000 variants in over 1,000 genes — there is also hope that the effort can help advance the science behind more uncertain gene-drug pairs to inform future generations of clinical PGx test panels.
Philip Empey, associate director for pharmacogenomics in Pitt's Institute for Precision Medicine and faculty in the School of Pharmacy and the Clinical and Translational Science Institute, is leading the center, which is recruiting patients through a larger through a larger CTSI biobank initiative called Pitt+Me Discovery.
He said in an interview this week that the program has so far enrolled about 5,400 individual, but investigators have not yet begun returning results to physicians.
"We officially are right at the point of finishing the lab validation and then we'll finalize the decision support and begin returning," he said. The group won't disclose which genes will make up the panel of 14 until their final clinical validation is complete, he added in an email.
The program is structured as a research trial, in which participants go through a full consent procedure. But in practice it will operate akin to the launch of a clinical PGx test within the UMPC health system. Individuals who enter the study will receive, through their physicians, potentially actionable results and the study will track how those results, when implemented, affect their care and outcomes.
Empey agreed that the effort comes at a somewhat fraught moment for pharmacogenomics, with actions by the US Food and Drug Administration over the last year calling into question the agency's intentions regarding laboratory-developed PGx tests. The following months have seen a backlash, but also the genesis of new efforts by laboratory associations and consortia to create guidelines and best practices that might curb the overreach and overpromises that FDA said prompted its crackdown.
"There are situations where I was happy [FDA] stepped in because I think there were times where it was sort of an arms race [of] whoever can get more genes on the test or more drugs listed," Empey said. "In that situation, folks like me who were implementing this had to run just as much reeducation [as] original education."
While there may be frustration in the field around FDA's recent actions, Empey added that he's optimistic about there being more resolution around certain open questions like whether and how the agency might recognize sources of clinical utility data beyond just its own drug labelling.
"I do think this next year, hopefully we'll see more clarity around either projects launching and showing things that are feasible, or the best case would be more of a blessing of data coming from multiple sources for clinical utility arguments," he said.
"At this point they've reached out via letters to certain laboratories who they thought were doing things inappropriately ... [but] they are also simultaneously working with groups, whether it's trying to start PGx communities with industry, [or working with] large programs who are proposing to do this, like the All of Us program. … It seems like they're trying to get it right."
In the meantime, he said, "I think it's smart to develop the right model." Hence UPMC's decision to build their program as a fully consented research study, as well as one in which return of results is fully through providers. "It's a well anchored … medical practice model, as opposed to anything that could be viewed as being outside of the practice of medicine or potentially risky … something entirely outside the health system or ... directly to patients or anything like that.
"I'm not against those things," Empey added, "but they inherently potentially have more risk and there's more to do research on in that area, so you can understand why [the FDA] may want to step in there."
The UPMC preemptive program is starting out using a relatively narrow panel of genes — just 14 — compared to some of the commercial tests that have come under recent scrutiny.
Inova, for example, which made waves last year when it shut down its entire PGx testing service after being served an FDA warning letter, had been marketing a suite of tests covering a broad range of genes and drugs, as well as allowing in-lab physicians to order tests for consumers and providing results directly to these patients.
A trend toward more judicious test design doesn't seem to be unique to the Pittsburgh research program, though. Apart from regulatory pressure, payor pressure also appears to be playing a role in reducing the breadth of existing PGx gene panels. Genomind (which was also one of several labs that were directed by FDA to change their reporting practices during the last year) has marketed for several years what evolved to be 24-gene test. Recently, though, it has added a more focused panel it calls the "Core Anxiety and Depression Report" to its menu, cutting that list to 15 genes: the maximum number that private insurer UnitedHealthcare outlined in a coverage policy that took effect last October.
Empey said that in planning the Center of Excellence effort it was clear that there wouldn't be much progress made in trying to implement and validate one gene at a time, as Pitt had previously done with a successful clopidogrel-focused pilot, called PreCISE-Rx.
Other lessons learned from PreCISE-Rx, and from reviewing other PGx efforts, include the need to have advocates in place to implement and address logistical challenges. Additionally, he said, it's clear that education, especially in scaling to a population-scale effort, will be a huge factor.
"It's relatively easy to train a small focused group, and that worked really well [with clopidogrel]. But as we try to expand across our whole system, there are a lot more providers to be involved, and so we're focusing on education considerably," Empey said.
"Another thing we really learned a lot from … was how to get all the informatics part right," he added. For example, the UPMC system is, maddeningly, split between two different electronic health record systems: EPIC and Cerner. "When we think of things like that, [it's clear that] the data has to come in as discrete results because we have to deliver it from more than one lab to more than one electronic health record with decision support. So, if people were being forced to read PDFs ... it just wouldn't be as efficient."
Although the study will only be returning results for 14 genes, at least at first, Thermo Fisher arrays being used on patient samples as they are recruited into the Pitt+Me database cover approximately 1,200 genes. According to Empey, the genotype-phenotype research that this will enable probably won't call into question any of the established drug-gene pairs that the field already views as having significant validity data behind them, but hopefully it can help build evidence for some that don't yet carry the same weight.
"If you look PharmGKB, there's thousands of … associations that are below the evidence threshold and you either need a lot of patients or focused data to be able to answer some of those questions. So, part of the research project is also to validate some of these associations," he said.
That said, the study isn't randomized. "We're going to be targeted in our recruitment, so it's a pragmatic type of trial, observational. But, it should generate some data," Empey added. For example, his own primary research is in drug transport.
"There's not a lot of transporters besides SLCO1B1 that have evidence grade that's high enough to implement. I'm not saying there will be, but I'd love to generate data to say whether there is."
According to Empey, the choice to launch this project in partnership with Thermo Fisher reflects to some extent the fact that the firm has made a point in recent years to publicly emphasize its dedication to this area, centering PGx in its workshops at scientific and industry meetings.
Users from Coriell Life Sciences and from Genomind, for example, have headlined corporate workshops at previous AMP meetings.
"We were looking for a pharmacogenomics product that could help us generate the research data we're looking for… as well as have the ability to produce clinically actionable, meaningful data," Empey said, "so, for us, it made sense to partner.
"There's value in us having the data in the long term, and being able to manage our turnaround times and the logistics associated with sending it out, transferring data back and forth didn't make as much sense to us as to partner and bring a test in-house," he added.
While the information gleaned through the biobanking aspect of the effort stands to be valuable, as Empey said, the more immediate thrust of the effort is to assess the health and economic impact of the 14-gene clinically actionable sub-panel.
"The data for pharmacogenomics remains strong, but there's always going to be a need to demonstrate clinical utility," Empey said. Thus far there have been smaller efforts, like the center's clopidogrel project and similar efforts elsewhere that "are working really well with strong reimbursement."
"We are also seeing signals like the United Health Plan reimbursement announcement [to cover PGx panels up to 15 genes] in the fall and the MolDx LCD [for limited PGx panels] that was presented [also last fall] as being signals of the payer space starting to warm up to the idea," he added.
But for larger-scale preemptive PGx, "It's kind of been a chicken or the egg thing, in terms of, trying to prove value to do a large study, but needing a large study to prove value."
One aspect of Pitt's success in getting its new effort off the ground has been the fact that UPMC has an integrated financial delivery system with a payor-provider model, Empey explained. "We have a health plan and they care about our endpoints, so if we could do a program that's large and can answer some of these big questions … at the same time of delivering value to researchers and to participants, it just makes a lot of sense when it comes together."
"If you take a step back and look at the field, we're not the only ones," he added, citing the announcement last year by the US Department of Veterans Affairs and Sioux Falls, South Dakota-headquartered healthcare system Sanford Health that they were setting out to provide pharmacogenetic testing for up to 250,000 veterans.
"Maybe what got that [program] across the finish line was more of a philanthropy play as opposed to a strategic play for clinical utility by a forward-thinking health system, which is the bin I put us in … but [I think people are recognizing] that this is one of the easiest use cases to deploy [genomics] where there's short-term value," Empey said.
"If you think about it, we're exploring doing whole-genome sequencing for big projects … and if folks are willing get into the space for large, complex diseases ... certainly the return on some of the drug-related stuff has a much shorter horizon," he added.