Skip to main content
Premium Trial:

Request an Annual Quote

Flatiron Analysis Touches Nerve in Debate Over Value of Broad Sequencing, Precision Medicine

NEW YORK (GenomeWeb) – A retrospective analysis of health records in the Flatiron Health database has painted a grim picture of the benefit comprehensive genomic sequencing currently offers to non-small cell lung cancer patients.

The study, published yesterday in the Journal of the American Medical Association, concluded that despite growing enthusiasm for more exhaustive assays, individuals whose tumors are analyzed for dozens of mutations don't seem to be surviving longer or better compared to those who get older, gene-specific tests — at least not in a community setting.

Conventional wisdom regarding the value of comprehensive genomics and precision medicine tends to differ depending on who you talk to, but one area in which clinicians have argued that broader sequencing has clear value is in the treatment of patients with advanced NSCLC, especially young, non-smokers, who are more likely to have tumors driven by targetable mutations.

In that light, the new JAMA study was met with some strong reactions, reflecting anxiety among clinicians that the findings might be interpreted by community oncologists as a reason not to use comprehensive, or broad-based genetic testing in NSCLC patients.

Carolyn Presley, the study's first author and a thoracic oncologist at Ohio State University's comprehensive cancer center said that she and her coauthors didn't intend that to be the takeaway from the study findings.

Paul Bunn and Dara Aisner, lung cancer experts at the University of Colorado, Denver, echoed this in an editorial accompanying the JAMA paper, writing that while studying the impact of precision medicine strategies is valuable, results like these "should not lead to the conclusion that broad-based genomic sequencing should be avoided in nonsquamous NSCLC." 

According to Presley, she and her colleagues' analysis speaks more to the challenge of realizing the promise of this type of testing than it does to its inherent value.

"When you drill down, I think the problem we found isn't that the mutations aren't there, it's that patients aren't getting the drug," she said, whether because the list of targeted therapies at the time of the study was still small, or because of a lack of access to clinical trials in the community setting, or because of the cost of treatment with available medicines.

Questions about how and when broad sequencing should be performed in clinical practice were accelerated by the Centers for Medicare & Medicaid Services draft national coverage determination for next-generation sequencing cancer panels last year, which sparked fear among some clinicians that use of broad NGS panels across cancer types could increase off-label drug use and either harm, or at the very least fail to help patients.

Enthusiasm for broader sequencing, especially in the lung cancer space, reflects growing observational and case control evidence showing that patients who receive molecularly informed targeted treatment fare better than those who don't. However, the field is still shadowed by studies, like the randomized controlled SHIVA trial, which found that patients matched to a molecularly targeted drug did no better than those who were not.

Newer efforts like the NCI's MATCH trial are poised to resolve some of these questions, and recent positive result have been reported, for example from MD Anderson's "Initiative for Molecular Profiling in Advanced Cancer Therapy" (IMPACT) study. But ongoing research programs still face the challenge of translating the identification of genomic targets into patient benefit.

For broad sequencing to show itself superior to more targeted approaches in NSCLC, individuals with targetable alterations other than EGFR and ALK need to actually be treated with a targeted drug. But in the JAMA study this week, it looked like this was not working in the community setting, at least during the time period for which data was collected.

A patient who has a MET amplification who gets single-gene testing obviously misses out on the opportunity to be treated with a MET-targeting drug. But so does a patient who gets broad sequencing and finds out that they have that target. Based on the study data, they are not getting the drug either, Presley said.

In their study, Presley and colleagues — mostly from Yale University — analyzed data from Flatiron Health's database representing more than 5,000 patients with advanced non-small cell lung cancer who were treated in a community oncology clinic. Researchers divided the data into two groups — patients who received broad genomic tests versus those who got testing for alterations in just EGFR or ALK — and compared both molecular findings and outcomes.

According to the authors, patients who received broad-based testing had a bit of a higher chance of receiving genetically targeted treatments, but the number who got targeted drugs was still very low, and they didn't show improved survival over those who got single-gene tests.

Presley called the results evidence that the ability to sequence has outpaced doctors' ability to get targeted therapies to patients. But that doesn't mean that it shouldn't be done.

Dana Farber medical oncologist Geoffrey Oxnard, who was not involved in the study, echoed this, saying that the results from the JAMA study aren't necessarily surprising, but they are also not damning for broad genomics in NSCLC.

Important in understanding what the analysis does and doesn't imply, he said, is recognizing the different steps that are required for comprehensive sequencing to be valuable.

"Getting the testing, that's one, and then interpreting and understanding results … and [the third] is having drugs to put patients on based on those results," he said.

"If you do NGS on a bunch of patients with a low pre-test probability — a small chance of finding something — and then have no drugs for what you do find, it's hard to have an impact," he added.

According to Oxnard, the takeaways from an analysis like this aren't black and white. "The utility of NGS does interrelate to the likelihood of having targets," he said. "So if you performed this analysis in breast cancer, the result would have probably been the same … but if you did it just in young never-smokers, NGS is totally going to be better."

In other words, the value that broad sequencing has in community settings is fundamentally connected to how it is used, which is something that is still evolving as physicians become more aware and more exposed to precision medicine.

"I don't want providers to think that they are giving standard of care by doing local testing instead of NGS – that's not necessarily true," Oxnard added. "But the idea that a heavy smoker needs to have NGS to be eligible for hospice, that may not be true either."

According to Oxnard, another important thing to recognize in contextualizing the JAMA findings is how fast the field is evolving and how much has changed since the period of the study, first for patients who are seen at centers like Dana Farber where there is more direct access to clinical trials, and then hopefully also for patients in the community setting as drugs are approved and, ideally, covered by insurers.

"My patients who get testing provably do better than their patients who get testing, because I have trials, and I get them on them," Oxnard said. "But we've got two great RET inhibitors now … two great TRK inhibitors, I have three or four HER2 trials. There's an FDA-approved ROS inhibitor, and better ones in development. I have patients in three trials for KRAS ... and this [progress] is going to be a big part of proving that NGS is better than more focal testing."

One caveat, Oxnard said, is that as more drugs become more accessible, more attention will have to be paid to making broad sequencing test results easy enough to interpret and implement.

In the Flatiron database, patients tested using broad sequencing received various tests with potentially wide-ranging quality or ease of interpretation, which could also have impacted the analysis.

In one interesting example, Guardant Health's Guardant360 liquid biopsy test was used for almost 15 percent of the broad genomics section of the cohort, despite the fact that research has shown that blood-based sequencing does not always detect targetable mutations present in tumor tissue, and that professional guidelines societies have cautioned that broad ctDNA assays like Guardant's aren't yet proven.

Flatiron, whose database was used in the Yale study this week, was acquired this year by Roche — as was comprehensive genomic testing firm Foundation Medicine this June.

Foundation Medicine was responsible for testing in most of the broad genomics section of the Yale study cohort. Caris Life Sciences represented about 13 percent, Guardant close to 15 percent, and the rest was made up of testing by in-house pathology labs and other companies.