NEW YORK (GenomeWeb) – "The proposal we issued a few months ago, is just that. It's a proposal," Jeff Shuren, director of the US Food and Drug Administration's Center for Devices and Radiological Health, said yesterday at the start of a two-day workshop to discuss the agency's draft guidance on regulation of laboratory-developed tests (LDTs). "We do not claim we have gotten everything right."
In response, for the next eight hours, lab professionals, pathologists, and even a few diagnostic firms that have taken tests through the FDA regulatory process, pointed out everything that they believe the agency has gotten wrong in the risk-based, draft framework it issued in October.
The main failing, according to those that object to being regulated, is that the proposed regulations show the FDA still doesn't get lab professionals. These certified medical experts develop the tests, track patients' samples to make sure they're analyzed correctly, and they work closely with doctors to interpret what the results mean for a patient's disease or condition. And all the while, they are tweaking and making adjustments to ensure the tests are up to date and accurate.
The FDA has said it has the authority to regulate LDTs under the Medical Device Amendments of 1976 because LDTs are devices. The labs have countered that while the tests they provide use FDA-regulated device components, such as reagents, the critical role of lab professionals in furnishing those tests makes the whole operation a service, and not the type of kit or physical article that the FDA traditionally regulates. Moreover, LDTs are already regulated by CMS under the Clinical Laboratory Improvement Amendments (CLIA) and other bodies such as the College of American Pathologists, they argued.
FDA's limited understanding of lab professionals' involvement, stakeholders said, is reflected in the draft guidance, in discordant labeling requirements, burdensome test registration expectations, and too stringent clinical validity definitions. Forcing LDTs to fit the rubric of FDA regulations would hinder labs' ability to offer new tests for markers not gauged by FDA-approved kits, to quickly provide tests during infectious disease outbreaks such as the Ebola crisis, and to cost-effectively add new tests to state-mandated newborn screening panels, or furnish timely results for cancer patients awaiting personalized treatment strategies.
"From the lab perspective … you have to be realistic that most clinical laboratories are not set up, as you've been told over and over again, as reagent and device manufacturers," Girish Putcha, director of laboratory science at Medicare contractor Palmetto GBA and founding medical director of Orion Genomics, told FDA during the workshop, which was webcast.
"In the lab world we have very detailed and intricate standard operating procedures that we are required by CLIA, by CAP … that starts taking in … beyond what's there on the [test] label of the box … of the kit or the reagent, and adds the personnel component, adds the interpretive component, adds the clinical validity or use," said Catherine Hammett, who at the meeting was representing the American Association for Clinical Chemistry and who also directs the University of North Carolina HealthCare's core Lab at McLendon Clinical Laboratories. "All of these are present within our laboratories, so from a lab perspective, we become unclear [about] what more are we going to be expected to do."
Perhaps the most worrying of FDA requirements for labs is that they have to prove to the agency that their tests are clinically valid — that the test accurately detects a disease or the intended phenotype. The FDA has stated that under CLIA, tests are evaluated only for analytical validity, whether the test measures a specific analyte or marker.
But laboratory professionals speaking at the meeting took exception to that, maintaining that they deal with the clinical validity of tests all the time, and asked the FDA to be more flexible in how it considers the concept. "Clinical validity is an ever-moving target … [and] knowing and applying levels of evidence to LDTs is part of the practice of medicine," Dara Aisner, pathologist and co-director of the Molecular Correlates Laboratory at the University of Colorado, said at the meeting.
As an example, she cited a study that found 34 percent of patients with BRAF V600K mutations are deemed mutation negative by an FDA-approved BRAF kit. But a test developed in her lab can more accurately identify these mutations, she said. Roche, which markets the FDA-approved Cobas BRAFV600 mutation test, has done concordance analysis against Sanger sequencing and has reported that the Cobas test had a lower failure rate than Sanger, was more sensitive in detecting V600E mutations than Sanger, "and detected a majority of V600K mutations" in the study cohort. But concordance studies of the Cobas tests by other labs have yielded varying results.
Citing another example, Aisner said that an FDA-approved test for assessing ALK mutations yielded a negative result for a patient, but her lab "noted an unusual pattern," tested the patient with alternative LDTs, and found the patient was mutation positive, thereby enabling treatment with the lung cancer drug Xalkori (crizotinib). "Clinical validity is information that is accumulated over time, requires adoption into medical practice by trained, thinking professionals, and comes in many forms, and requires flexibility and nimbleness," Aisner said.
During a panel discussion, other speakers reflected on the clinical validity requirement. Most seemed reassured that FDA in its draft guidance said it would accept data from published literature on markers gauged by high and moderate risk LDTs, which would require premarket review. This would limit the burden on LDT developers to conduct new studies, noted Andrew Fish, executive director of the diagnostic manufacturers' industry group AdvaMedDx. Fish was among the minority presenters at the meeting who supported FDA's LDT guidance.
Cecile Janssens, professor of translational epidemiology at Emory University, noted that when establishing clinical validity of a test, labs need to define who is being tested and for what purpose.
In FDA's terms, the patients that a lab tests with an LDT and the claims the labs make about the test would be reflected in the device labeling. However, typical device labeling requirements also didn't stick well with lab industry representatives. Once a test's intended use is printed on an FDA green-lighted label, any other uses outside the scope of the label is considered off-label. As such, stakeholders worried at the meeting that FDA would hold labs accountable for off-label promotion if doctors ordered LDTs for non-FDA cleared or approved indications.
Putcha noted that in the lab community off-label testing happens frequently. UNC's Hammett recalled that an FDA official had called her a year ago and asked how her lab regulated physician ordering. "There are very few mechanisms we have in place that allow us to prohibit off-label use," she said. "I get in over 5,000 samples in the core laboratory. The testing that's ordered on each of those 5,000 samples is done automatically through our electronic medical records system."
Hammett further reflected that if labeling is intended to inform doctors about the intended use of a test, they don't generally use them and labeling won't affect their interactions with the lab. "They call us when they have questions with respect to the results and the context in which they are looking at the results," Hammett said.
Meanwhile, Aisner offered the example of a patient suffering from Erdheim-Chester Disease (ECD), a condition where the body overproduces infection fighting cells and causes organ failure in the absence of treatment. Her lab at CU decided to test for BRAF V600E mutations, which approximately 50 percent of ECD patients harbor. Based on the recent research, some researchers have hypothesized that the rare disease may be more akin to cancer instead of an autoimmune disorder. This LDT found the patient was positive for the BRAF mutation, and following a month of treatment with a melanoma drug, the BRAF inhibitor Zelboraf (vemurafenib), she had a near complete response.
"If the FDA guidance were implemented, I likely would not be able to employ my LDT," Aisner said. "Use of an FDA-approved test for this indication would be considered off label. So, there would essentially be no test for patients like this."
Given lab professionals' proclivity towards tweaking their in-house tests, industry representatives were particularly concerned with language in the LDT guidance stipulating that certain changes made to approved or cleared tests would need to be resubmitted for regulatory review. The FDA states in the draft guidance that when a clinical lab alters a test in a way that can change its intended use or its performance, for example, by altering the type of specimen analyzed or the target population it is meant for, then the lab is subject to premarket requirements.
Industry representatives advised FDA to convene an advisory committee of lab professionals who could provide the agency further input on what constitutes a substantive change to an LDT and would require fresh premarket review. "Playing" with the different components of an LDT is what laboratory scientists and medical pathologists do, said Elissa Passiment, executive VP of the American Society for Clinical Laboratory Science. "We can tell when we have messed up the test," she said. "We also know when we've tweaked a pH, or we've changed the buffer, and we've got something and when we haven't gotten something."
Passiment echoed the sentiments of several other speakers at the meeting who advised FDA to allow lab professionals to continue to develop their LDTs as they have been under CLIA for decades, and focus its regulatory efforts on multi-marker, high-complexity, algorithm-based tests. Of course, the FDA tried to regulate this subset of LDTs and issued a draft guidance in 2007. Back then, such tests were called in vitro diagnostic multivariate index assays. But according to the FDA, it abandoned that plan, and at the lab industry's request chose to promulgate more comprehensive LDT regulation.
"I think what you're here for is to worry about the stuff that we've never seen before," Passiment told FDA. The tests "that folks out here can't explain to you because it's so dense and buried in software and mojo that no one has any idea what's going on except the person that developed it."
The risk that faulty or unproven LDTs pose to the public's health is one of FDA's top reasons for regulating LDTs. One of the primary examples in this regard involved an algorithm-based test called Ovasure. The FDA sent a warning letter to LabCorp in 2008, noting that the ovarian cancer risk detection test wasn't appropriately validated for women at high risk for the disease. The test had a high false-positive rate and based on the results, some women with benign masses got their ovaries removed. "The Ovasure example may very well be why we're sitting here," said Emily Volk, a member of CAP's board of governors.
While Ovasure would have been categorized as a test that served an unmet medical need, it still posed a significant risk to patients, speakers pointed out, and it illustrates the need for a way to track LDT-related adverse events. Under FDA's draft guidance, traditional LDTs (tests developed and performed at a single lab), tests deemed low risk or Class I devices, tests for unmet medical needs, and for rare diseases would be exempt from having to meet premarket regulations and quality system requirements. However, they would still have to register and list their tests with the FDA and report adverse events.
Some in the lab community said at the meeting that this category of enforcement discretion should be broadened. Moreover, several speakers found too restrictive FDA's stipulation that tests must be performed for less than 4,000 patients annually in order to be deemed an infrequently performed test that qualifies for premarket review exemptions. They asked FDA to define rare disease tests according to disease prevalence instead.
To that, Rasika Kalamegham of the American Association for Cancer Research said that stakeholders were missing the spirit in which the FDA proposed these exemption categories. A rarely performed test can forgo premarket review, Kalamegham suggested, because the lab would have the expertise to do it and would discuss the results with the ordering physician. A rarely performed test is very different from a rare disease, however, she posited.
"We no longer talk about lung cancer or breast cancer; we talk about ALK-positive lung cancer or triple-negative breast cancer. Everything has become so compartmentalized and so specialized that you could argue that every cancer is becoming a rare disease," she said. "That doesn't mean the tests that are used to determine treatment options are not used quite liberally."