Skip to main content

Future of Liquid Biopsy Weighed By Dx Stakeholders at FDA Workshop

Premium

NEW YORK (GenomeWeb) – In the wake of the first US Food and Drug Administration approval of a liquid biopsy test in June, stakeholders are grappling with the next steps in advancing additional clinical blood-based cancer tests under a changing landscape of regulatory oversight.

Representatives from liquid biopsy diagnostic firms, clinical researchers, FDA officials, payors, and drugmakers gathered on Tuesday for a joint FDA- and American Association for Cancer Research-sponsored workshop on the subject.

During the full-day conference, which was webcast, workshop members used Roche's blood-based cobas EGFR Mutation Test — approved this June as a companion diagnostic for Genentech's Tarceva (erlotinib) — as a case study and jumping off point to debate the practical challenges and discuss potential strategies for the continued advancement of existing and future tests.

One issue raised about this pioneering assay approval was the challenge of performing a technical validation with the right kind of samples that can show that a liquid biopsy technology can pick up mutations in a vast background of circulating wild-type DNA.

Collecting enough clinical samples with low-enough levels of mutant DNA is far from practical, or even impossible in some cases, according to attendees.

"If you can find [the] patients, you want to have a mix of contrived and clinical samples. But we are starting to push into a realm where it will be difficult if not impossible to identify even one or two patients, therefore there may be situations coming up where contrived specimens alone will be required," said Walter Koch, vice president and head of global research at Roche Molecular Systems.

In Roche's case, the FDA worked with the company to develop a strategy using contrived samples and spike-in experiments, along with commutability studies using clinical samples, to demonstrate the analytical validity of its PCR-based test.

Developing this strategy wasn't simple though. It was initially unclear whether sheared DNA or intact DNA would be a more appropriate sample type to analyze, the FDA's Reena Philip said. After some initial experiments, Roche and the regulator settled on a plan using sheared DNA.

However, later in the meeting, discussing the question of whether and how best to use contrived samples, Guardant Health President and COO AmirAli Talasaz said that based on Guardant's own validation experience, sheared DNA and synthetic materials are both inferior to cancer cell-line derived DNA with known mutation status spiked into cell-free DNA samples from young healthy donors.

Another challenge discussed by attendees, using the Roche test as a case study, is how to develop or choose a reference standard against which to compare a liquid biopsy test in order to show that it works.

Working with Roche on the approval of the Cobas test, the FDA recommended avoiding comparing its PCR method to another PCR method, Roche's Walter Koch said. So instead, the company developed its own orthogonal method using deep next-generation sequencing.

In a question and answer session, attendees raised the question of how test developers should best negotiate this moving forward. With the Roche Cobas test now approved, does that platform need to be the reference standard against which other assays are validated if they are seeking FDA approval?

According to Phillips, the FDA anticipates flexibility in that area, and had a history of working with companies advancing tests with no existing reference comparator to develop an appropriate one.

Most meeting participants agreed that even with the right sample for analytical validation an appropriate reference method, and a way to link test results to patients' clinical outcomes either through prospective or retrospective studies, an important hurdle remains: how to advance and educate physicians about the utility and limitations particular to liquid biopsy tests, especially given the rapidly changing technological landscape.

As it has before, Foundation Medicine took a position of caution during the discussion regarding liquid biopsy, with CSO Phil Stephens imploring his fellow participants to be wary of their own limitations.

"Doing broad-based genomics from the blood is orders of magnitude more difficult than tissue-based testing and I think that physicians, and academics, and we in the industry really do have to think about what is appropriate and how to use these tests," he said.

For example, he said, data presented at recent scientific meetings has begun to accumulate in which liquid biopsy testing is picking up particular mutations at much different frequencies than the consensus of prior studies of tumor tissue.

In some cases, these discrepancies may represent the ability of liquid biopsy to overcome tumor heterogeneity, Guardan't Health's Talasaz said. As it has built its commercial testing business, Guardant has seen cases of clinical benefit in patients who had positive liquid biopsy results but negative tissue testing.

But Stephens raised another example, that of c-MET alterations, which some reports indicate are coming up more than 15 times more frequently in blood than one would expect from the wealth of data from tumor tissue sequencing studies.

What it looks like is happening, Stephens said, is that liquid biopsy assays are not distinguishing polysomic c-MET from focally amplified c-MET.

The former does not predispose patients to respond to current targeted drugs, while the latter does. "If this were broadly recapitulated, for every 15 patients recommended for c-MET therapy, 14 would not actually have a targetable alteration. Clearly this is an area of concern," he said.

As an illustration, he shared a case study of a Foundation Medicine patient in whom tissue testing identified a RET fusion, a P53 mutation, and a loss of P16. "The patient was fortunate to get on a RET inhibitor and had a blockbuster response," Stephens said.

However, blood-based testing by a company that was not Foundation Medicine (but which Stephens did not name) identified an EGFR mutation. The patient's physician, uncomfortable abandoning a treatment that was working, sought to do additional blood analysis before acting on the EGFR result, and it turned out after extensive deep sequencing of the patients' plasma samples that the EGFR mutation was a false positive.

According to Stephens, Foundation Medicine has now "heard dozens of examples in a similar vein."

As a result of open questions about the cases of discordance between tissue and blood-based testing, most at the workshop agreed that tumor tissue testing should still be considered standard of care.

In reflection of this, FDA approved Roche's assay with a reflex indication: If the results of the blood-based test are negative, physicians are recommended to seek out tissue testing, and in blood-negative/tissue-positive cases, the tissue should trump the liquid biopsy in terms of whether a patient is eligible for treatment.

Most at the meeting agreed on this point. However, there was some variation in how participants considered the question of a tissue-negative, blood-positive case.

Some questions from the audience reflected skepticism that blood-positive, tissue-negative cases reflect true positives, and a call for more data on patient outcomes in these cases to confirm the clinical implications.

Guardant's Talasaz said that based on its preliminary data, the company believes variants in ctDNA but not tissue should be considered equally actionable as concordant calls, especially recognizing that part of the power of liquid biopsy is that it can be used in patients for whom tissue testing is not possible or overly detrimental.

He said that the company has not published on this yet, but it has data from community practices that suggests only 60 to 65 percent of non-small cell lung cancer patients are getting tested for EGFR, and even lower percentages for other biomarkers.

"There are inherent barriers in the system. Patients are under-genotyped, and one advantage of liquid biopsy is that we can get more of them [tested,]" Talasaz said.

However, he said, in the case of negative liquid biopsy results, Guardant agrees that the field will have to work hard to make sure physicians are educated to pursue tissue testing before denying a patient a particular treatment based on a negative liquid biopsy.

This was echoed by many other meeting participants. "There is a very heavy educational component that has to be addressed. Practicing physicians in the absence of that education may inappropriately interpret a negative result and not reflex to tissue appropriately and deny patients access to care, so we have to figure out how we as test developers can address that educational need," said Mark Lee from Illumina spinout Grail.

Another area where transparency and education are greatly needed, participants said, is in accurately assessing and communicating liquid biopsy tests' limits of detection, and what those cutoff points mean. If a test detects mutant DNA, but it does not meet the pre-specified limit of detection cutoff point, doctors may still want to know about those results and consider acting upon them. In fact, at least one clinician at the meeting described weighing exactly that choice.

"This is where everyone in the FDA has to close their ears," one attendee said in a question and answer period. "But this is going to come up all the time. So if you are going to report something below your LOD, and a physician wants to act on that, a lab needs to make that abundantly clear what that means."

Absent anticipated FDA oversight over laboratory-developed tests, much of what the participating stakeholders discussed is somewhat moot, with companies able to launch into the field even without making public their analytical validation results, a practice Foundation Medicine's Stephens criticized firmly.

"If a company offering a CLIA-based test has not released an analytic validation it does not know what the accuracy of its test is and neither do the treating physician or the patients," he said.