Clinical Informatics News: Risk to Patients Balanced Against Risk to Labs as FDA Moves on Lab Developed Tests

Published on: Tuesday, September 23rd, 2014 View all Articles

By Aaron Krol

September 23, 2014 | Nearly two months have passed since the FDA announced its intention to start regulating laboratory developed tests, or LDTs, by mid-2015. These tests, which are designed, made and performed entirely in a single laboratory, have a huge footprint in the American healthcare system, and have often been on the front lines of innovation in areas like personalized medicine and cancer diagnostics. They are also the only in vitro diagnostic tests, or IVDs, that aren’t required to submit data to the FDA proving their safety and accuracy, although the labs that perform them do need to be certified under the Clinical Laboratory Improvement Amendments (CLIA).

In the nearly forty years that LDTs have enjoyed this policy of “enforcement discretion,” they have grown from a small but vital niche used mainly by hospital labs to diagnose rare diseases, to an industry where big players like Myriad Genetics and Genomic Health can post profits in the hundreds of millions of dollars. Conventional IVD manufacturers have been crying foul, and the FDA agrees that its LDT policy has led to some perplexing outcomes. “It certainly calls into question why a company would go through the entire regulatory process, if they can just open their own clinical lab,” says Katie Serrano, from the FDA’s Office of In Vitro Diagnostics. “And we’ve seen cases where they have done exactly that. The larger manufacturers have said, to heck with this PMA [premarket approval] process, we’re just going to open our own CLIA lab.”

Sample draw
Tests performed on patient samples, like blood, spit, or tissue biopsies, must be cleared by the FDA — unless developed and performed by a single laboratory. ©

At the same time the FDA declared that enforcement discretion was coming to a close, it also released an oversight framework explaining what it will require of labs that run LDTs, and the timeline for implementation. (For details of the proposed regulations, see “FDA Announces Plans to Regulate Laboratory Developed Tests.”) As mandated by Congress, the FDA has extended 60 days’ notice before officially issuing this framework as a draft guidance, but that period ends on September 29, and so far, the document appears likely to be issued more or less as written.

“It’s about time,” says Tadd Lazarus, CMO and Head of Global Molecular Diagnostics at QIAGEN, a large IVD manufacturer whose tests have sometimes competed with LDTs. “This is the first major push with a tight timeline, and I think it’s the right rulemaking at the right time.”

“We’ve all been bracing for it to come,” adds Judi Smith, who as Vice President for In Vitro Diagnostics at the consulting firm Precision for Medicine has advised both IVD manufacturers and CLIA labs on regulatory issues. “The IVD industry has been saying, have a level playing field. When we develop kits, we have to go through an approval or a clearance process, and [the FDA requires] very stringent analytical and clinical studies to be performed.”

The three main provisions of the oversight framework will all be familiar to IVD makers. Labs that perform LDTs will have to compile a list of the tests they offer; have a new mandate to report adverse events, including deaths or serious injuries to patients; and, most importantly, must submit to premarket review of any tests that the FDA designates as high- or medium-risk. This review could take the form of a PMA submission, which would typically include a clinical trial — the largest expense IVD manufacturers face when bringing a new test to market.

But most labs that develop LDTs aren’t commercial giants. The large majority are affiliated with hospital systems and academic centers, and they are raising alarms about what LDT regulation will mean for clinical care at their organizations. Uncertainties still surround almost every aspect of the proposed regulations, including just how many LDTs are out there, and whether the FDA even has the authority to regulate them. But for lab directors around the country, the biggest question is whether hospital and academic labs can continue offering their in-house tests at all, or whether the high cost of clinical trials will sink their operations.

Pain Points and Clinical Trials

Outsiders might be surprised to hear that the FDA prides itself on flexibility and openness to change. Members of the agency sincerely believe that their job is to help new products win approval, and they try hard to get this message across whenever they get the chance.

“We work with people who have great ideas all the time to help them through this regulatory process,” says Serrano. It’s a point that comes up often, as a burst of new technologies has challenged some of the traditional ways the FDA validates products. Next generation sequencers, for instance, can capture countless types of genetic variation, far too many to verify that every possible test result is accurate. And genetic diagnostics for rare diseases may need to consider dozens of related mutations, making it impractical to find a relevant patient population for a clinical trial. “The applications, and getting the appropriate studies to support claims that are made, are the trick,” Serrano adds. “We are trying to be flexible and creative, but still maintain our standards for these new technologies.”

The agency’s recent track record bears this out. Last November, the FDA gave its first nod to a next generation sequencer, the MiSeqDx. The review involved a limited study in which just 13 clinical samples were sequenced, and a representative set of calls was analyzed for accuracy. The FDA also cleared two cystic fibrosis tests for the instrument, and instead of insisting on a clinical trial, allowed the manufacturer to test cell lines and refer to a well-curated online database of cystic fibrosis mutations.

Lazarus says that QIAGEN, too, has benefited from discussions with the FDA about its regulatory submissions. “The FDA creates a pre-submission dialogue,” he says, “a space where everyone has the ear of the agency, and where questions and concerns can be addressed fully. By the time we as manufacturers submit, we can get a lot of the pain points under control.”

So the FDA would like labs to believe that there will be many routes to LDT approval, and that the agency is prepared to help tailor submissions that are appropriate to each test. In particular, it wants to soothe some fears about clinical validation, the standard that a premarket review should show not only that a test is accurate, but also that it offers a health benefit to patients.


Labs increasingly rely on complex computer pipelines to analyze the results of testing, opening new questions about validation. Photo by Michael J. Ermarth for FDA.

This will be a difficult sell. The gold standard for clinical validation will always be a clinical trial, tracking patients long enough to collect data on their health outcomes, and clinical trials are hugely expensive. Even on the low end, they cost hundreds of thousands of dollars to conduct and take months to complete. For budget-constrained academic labs, alternative sources of validation, like databases, peer-reviewed literature, and studies on existing tissue samples, may be the only feasible options, and it’s not always clear when the FDA will find these acceptable.

Serrano would like labs to focus less on clinical trials, and more on clinical practice. “The question we’re trying to get at is, what does your measurement mean in context?” she says. “How would a physician make a decision based upon that test result?” But she also acknowledges that the FDA is wary of any submission that can’t reference known patient outcomes, and that clinical trials remain the best way to demonstrate safety and efficacy.

“With something that’s very novel, sometimes we have to follow it through, because the community doesn’t know how to interpret that value or that analyte,” says Serrano. “Whereas with, for example, glucose or vitamin D, those are pretty well established, so we wouldn’t be looking for anything new to establish clinical utility.”

Of course, laboratory pathologists will see a lot of gray area between a glucose test and a completely novel analyte. As LDT regulations move forward, a great deal will hinge on just where the requirements for clinical trials fall.

The Question of Harm

The FDA is convinced its new stance on LDTs is necessary to protect patients in a changing healthcare system. Representatives of the agency have cited cases like OvaSure, a test offered by the company LabCorp in 2008 to detect the early stages of ovarian cancer. Independent studies showed the test was less effective than the standard of care, with one study reporting that OvaSure’s biomarkers had just 34% sensitivity. Nevertheless, the FDA was only able to shut the test down by claiming that, because it had been developed at Yale and not in-house at LabCorp, OvaSure didn’t qualify as an LDT.

Serrano says there may be many more examples, but they’re hard to track down. “Because these tests aren’t required to make any systematic reporting of adverse events, it’s difficult to get this information,” she says. The FDA has to rely on third-party investigations to flag problems, and may lack enforcement options if they arise.

On the other hand, it’s unclear how much risk is posed to patients even by unreliable LDTs. Results from the tests are always communicated to physicians, not directly to patients. Particularly with speculative tests like OvaSure, physicians tend to shy away from recommending drastic medical interventions without some supporting evidence, like a family history of disease. In a webinar addressing the proposed LDT regulations on September 16, hosted by the Association for Molecular Pathology, Roger Klein of the Cleveland Clinic told attendees he had a hard time imagining any LDT result leading to the kinds of extreme patient harm that would trigger an adverse event report to the FDA.

The agency is not so relaxed. In its oversight framework for LDTs, it names several categories of tests it believes could lead to disastrous patient outcomes, including blood tests for infections in immune-compromised patients, and tests used on healthy patients to predict the risk of serious conditions like heart disease.

Perhaps most importantly, it also singles out companion diagnostics, which recommend the use or disuse of specific drugs, as falling in the highest risk category. That means these tests would be the first to go through premarket review under the proposed framework. If the FDA finds companion diagnostics unsafe or inaccurate — or if a lab decides it can’t afford to file for approval — they could be pulled off the market as early as next fall.

The FDA prefers companion diagnostics to be developed, and used in a clinical trial, alongside the drugs they indicate. But especially in cancer care, where companion testing is increasingly routine, labs see good reasons to develop tests on their own, and worry that their ability to do so will soon be sharply limited.

A High Risk Area

Cancer Genetics, Inc. (CGI), a midsized commercial test provider based in Rutherford, NJ, launched its first test in 2011, a genetic probe for patients with chronic lymphocytic leukemia. Since then, the company has built up a menu of dozens of molecular diagnostics, and now does about one million dollars of business a month, half of which is testing for hospitals and other care centers. Most of CGI’s tests are prognostic, trying to gauge the severity of cancer cases based on tumors’ genetic makeup. A few, however, like solid tumor testing for KRASBRAF andEGFR mutations, could be used to recommend specific drug regimens.

Like other LDT providers, CGI receives patient samples in the mail — its tests cannot legally be performed outside its CLIA lab — and consults with clinicians over the results, helping them interpret the often complex matrix of mutations found in a given tumor. Oncologists who want an even broader look at a tumor’s unique genome can work with a company like Foundation Medicine, whose Foundation One test sequences the whole coding regions of over 300 cancer-related genes.

This thirst for genetic information is a result of a rapidly evolving view of cancer itself. “You have to have a real appreciation for the biology of these diseases,” says Panna Sharma, CGI’s President and CEO. “There are over 90 different subtypes of leukemias and lymphomas, all with different outcomes and different genomic profiles. Seven or eight years ago, we certainly didn’t have that depth of knowledge… It’s already making a difference in treatment.” For some types of cancer, like non-small cell lung cancer and certain melanomas, that difference is profound, and thanks to the abundance of LDTs in this arena, it has largely bypassed the FDA.

There are compelling reasons clinicians have leaned on LDTs to diagnose and treat cancer. One is the pace of technological change: when the FDA first started approving companion diagnostics, genetic testing was limited to narrow methods like polymerase chain reaction. As next generation sequencers expanded what was possible in genetic medicine, molecular pathology labs rushed to take advantage of them, designing new tests to run on these powerful but unapproved instruments. Even today, only a small minority of labs own an FDA-cleared MiSeqDx, and no cancer panels have been approved for the device.

Then there’s the state of genetic knowledge itself, which is changing at a rapid clip. Take mutations to the KRASgene, which are known to affect patients’ response to certain colon cancer drugs. In 2012, the FDA approved QIAGEN’s therascreen test for seven KRAS mutations, and many labs replaced their LDTs with the new, FDA-validated test. But additional KRAS mutations are speculated to have the same effects, and some oncologists want tests to detect these, too. LDTs can swiftly fill the gap, while QIAGEN would need to compile new data on safety and efficacy before it could make any changes to the therascreen test.

“If these labs didn’t offer these tests, and attempt to make available new technology, I think we would have stymied the pace of innovation,” says Sharma, whose own company is planning to introduce its first next generation sequencing tests by early 2015.

Pill Doctor
Companion diagnostics are used to recommend specific courses of therapy, heightening the FDA’s concerns for patient safety. ©Forestpath

Sharma isn’t a pessimist about new FDA regulations. “Adverse event reporting, quality maintenance, reporting of when tests work and don’t work — I think all those are well overdue,” he says. But he does believe LDTs have offered labs the freedom to pursue new testing options for patients with difficult diseases.

“[Premarket reviews] will add cost, and they could delay innovation,” says Sharma. “A lot of companies don’t have the infrastructure or capital to do this kind of work, so a lot of things you see in the pipeline… will face some challenges. I think there will definitely be companies that fall out.”

Still, there are steps companies like CGI can take to minimize the disruption, even while working in higher-risk testing areas. Sharma says that the same adverse event reporting procedures his company uses for overseas testing will meet the FDA’s standards back home. Most importantly, he has worked with hospital partners, including the Cleveland Clinic and Memorial Sloan-Kettering, to collect and publish data on CGI’s test results on an assortment of patient samples. That wealth of data should give the company a leg up when discussing clinical validation with the FDA.

Ultimately, Sharma believes that CGI will take regulations in stride, and his industry may even benefit from greater assurances that its products are effective. In terms of submitting new tests for premarket review, he says, “the broader picture is that something that would have taken months might take a year. I think that’s a reasonable tradeoff to improve safety and develop a model that investors and operators of businesses can feel comfortable with.”

Down in the Trenches

Sharma can afford to be sanguine about LDT regulations. CGI is a successful business, with a steady stream of clinical customers — and more sales on the research side, which can help support new products before they are cleared for clinical use.

Many academic and hospital labs see the proposed regulations in a whole different light. Nina Longtine is the President-Elect of the Association for Molecular Pathology (AMP), an advocacy organization for the field of molecular diagnostics and one of the fiercest opponents of FDA involvement in LDTs. She was also the founder of the molecular diagnostics lab at Brigham & Women’s Hospital, and now serves as Vice-Chair of Molecular Pathology and Genetics in the Mount Sinai hospital system.

Longtine, like other AMP members, believes that labs like hers already work within a robust regulatory environment, thanks to CLIA certification, which is administered by the Centers for Medicare and Medicaid Services (CMS) and must be renewed every two years. CLIA inspections review the reagents and instruments a lab uses, and the professional training of the technicians, to ensure that tests are analytically accurate.

IVD manufacturers are quick to point out that this still gives LDTs leeway other tests don’t enjoy. “LDTs can be used clinically on patient specimens for months or years before a laboratory seeks CLIA or CAP [College of American Pathologists] certification,” says Lazarus. “[And] these are certifications that the lab is able to perform the test. They are not vetting the genomic targets for clinical efficacy.”

But Longtine sees a huge gulf between a company like QIAGEN mass producing reagents, and a molecular pathology lab like hers operating closer to patients. “We’re not a manufacturer,” she says. “We don’t make an instrument, we don’t make reagents. We use our own professional expertise to design and develop and validate and interpret. It requires a different framework that recognizes that professional input.”

The new oversight of LDTs comes at a time when labs are already feeling under siege from regulators. Last year, CMS undertook a massive overhaul of the Clinical Laboratory Fee Schedule, the coding system by which labs are reimbursed for testing. The old system was well known to be broken, and labs generally welcomed a new set of codes developed by the trusted American Medical Association, but the transition was badly bungled. Unsure what rates to set for each new testing code, CMS let regional billers choose their own pricing in a “gap-fill” process.

As a result of the large gap-fill, in many regions of the country, months passed without labs being reimbursed at all. Even after the issue was resolved, many labs remained critical of CMS’ downward pressure on pricing, particularly in genetic testing. “There was a whole year where we almost got no reimbursement, because CMS hadn’t put any dollar values along with their new codes,” says Longtine. “We’re really being tested, to be able to deliver innovative testing at very low cost, because reimbursement is so poor.”

Sharma has been feeling the same pressures at CGI. “I think there’s a real disconnect in government, in terms of how they view the role of genomic medicine in our country,” he says. “It’s very frustrating. They want to add more layers of cost, yet there’s no thoughtfulness about how we make sure these companies continue wanting to innovate.”

The FDA doesn’t make reimbursement decisions, and sees its role strictly as protecting patients from unsafe or ineffective products. But the impression labs get of the two main regulators is often of the right hand not knowing what the left hand is doing. This has soured the idea of new LDT regulations from the start, despite broad agreement even at organizations like AMP that the FDA is well-intentioned and wants to get the best care to patients.

A Big Job for a Beleaguered Agency

AMP doesn’t even like the term “LDT.” Its members prefer “LDP,” for laboratory developed procedure, emphasizing that these aren’t products you can pick off the shelf, but often a complex set of lab practices that can involve numerous sample preparation steps, computer analysis, and expert interpretation.

“This is not so much a medical device,” says Longtine. “It’s a medical practice that I’ve devoted my entire career to.”

The term LDP isn’t just a semantic quibble. It’s aimed pointedly at the limits of FDA authority, which extends to medical devices but not to the practice of medicine. At first glance, this ploy seems unlikely to make much headway: the definition of “medical device” was explicitly written to cover testing of patient samples when the FDA was given the power to regulate devices in the 1970s. But the LDP theory does have high-level supporters, including Rep. Mike Burgess (R-TX), who has fought to keep LDT oversight firmly under CMS.

Legal resistance to the new LDT oversight, either through congressional action or a lawsuit challenging the FDA’s authority over the tests, is still a possibility, and a September 9 congressional hearing about the proposed framework quickly turned contentious. At that hearing, Rep. Burgess asked whether the oversight framework was being issued as a draft guidance to avoid an economic review by the Office of Management and Budget, which would be triggered if the framework were issued as a formal rule.

Longtine shares this concern that the financial impact of regulation is not being seriously considered. “Before we move down this road, we should see how it’s really going to affect the economy of healthcare, and of laboratories,” she says. “That doesn’t seem to be on the agenda.”

FDA building
The FDA believes it has always had the power to regulate LDTs, but has been practicing “enforcement discretion.” Not everyone agrees. Image by FDA

Members of the hearing also questioned the FDA’s capacity to take on LDT regulation in addition to its other responsibilities. While Serrano says the agency plans to regulate these tests with its current resources, this could prove difficult given how large the LDT industry has grown.

It’s an open question how many different LDTs are out there. “It’s very difficult [to estimate],” says Serrano, “because there is no systematic way to collect information about these tests.” But she says the FDA’s best guess, based in large part on figures from New York State, which does track laboratory testing, is that there are around 11,000 labs offering LDTs around the country. The FDA believes around half of these labs’ tests are of minimal risk to patients and won’t need to go through a clearance process.

Critics of the oversight framework say this volume of LDT submissions would dwarf the FDA’s normal caseload of PMAs. Alan Mertz, president of the American Clinical Laboratory Association, told Congress that the agency approved just 23 original PMAs last year, as reported by Turna Ray at GenomeWeb. (Other sources repeated that figure to Clinical Informatics News; curiously, the FDA website lists only 21 original PMAs approved in 2013.) Even more dramatically, just four of those PMAs were for diagnostic tests, while the rest supported more conventional devices like heart valves.

Yet the focus on original PMAs is misleading. The FDA also processes a much larger number of 510(k)s, or submissions to clear a new device that is “substantially equivalent” to an already approved device. A test is eligible for a 510(k) if it is used for the same purpose as an existing low- or medium-risk test, and data can be produced showing it does not raise new concerns about safety and efficacy. That data may be hard to collect to the FDA’s satisfaction, and a 510(k) is not necessarily an easier route to approval than a PMA. But it is a much more common one: in 2013, the FDA processed over 3,000 510(k) submissions.

Nonetheless, it’s clear that LDTs will represent more than just a drop in the bucket of the agency’s review schedule. The oversight framework suggests a nine-year timeline for reviewing every LDT currently on the market, but even then, says Longtine, “I question the FDA’s ability to handle all the things that they’re suggesting. I think it’s huge.”

A Tight Path to Approval

Longtine is blunt about the prospects for labs like hers securing FDA approval for their in-house diagnostic tests. “It would be a huge challenge, and probably the end of academic laboratories, if we expected academic labs to go through some type of PMA process,” she says. “People spend millions of dollars on that, while we can make only a couple hundred dollars on a test.”

Judi Smith at Precision for Medicine agrees that a PMA is probably out of reach for most academic labs. From her experience with commercial IVD manufacturers, she estimates a typical PMA, including a clinical trial, would cost at least $1 million, a cost that would be prohibitive in niche areas like personalized medicine. “The reason personalized medicine is going through LDTs first is because these assays don’t have a huge demand,” she says. “The business model isn’t there to go through the FDA.”

From this perspective, it may matter less how cooperative the FDA is in helping design clinical trials, and more what mechanisms exist to get around them. Some tests, especially if they are similar to previously approved IVDs and don’t fall in the highest risk categories, could get by on support from the scientific literature, an established FDA practice that labs should find encouraging. “I do think that relying on published papers that have shown clinical validity should be a very helpful starting point, rather than having to collect your own patients and run a trial,” says Longtine.

The oversight framework also offers a few exemptions from premarket review, designed to protect vulnerable patient populations that are especially reliant on LDTs. These include continued enforcement discretion for rare disease tests, and for “unmet needs,” covering any LDT that has no FDA-approved equivalent.

The unmet needs exemption is a generous gesture, showing the FDA doesn’t want to disrupt the practice of healthcare, but it only applies to labs testing patients within their own hospital systems, leaving some patients without testing options. Serrano says the close communication between labs and physicians in the same care center will help minimize any risks to patients. “Ultimately, the lab and the physician need to be working together, especially in these cases where they don’t have as much information about the performance of the test, to make sure they’re making the right decision for the patient.”

But labs worry that their in-house tests will be repeatedly pushed out as FDA-approved tests fill previously “unmet needs,” even when labs invest heavily in their own, preferred solutions. The definition of an unmet need isn’t crystal clear, and the FDA would most likely, for instance, see the KRAS test niche as filled while a lab feels its broader look at the gene is indispensable. Then there’s the issue of platform lock. The FDA-approved therascreen KRAStest is tied to a specific instrument, QIAGEN’s Rotor-Gene Q, while LDTs for the same purpose could use a lab’s existing instruments.

For both the FDA and companies like QIAGEN, specifying platforms is a feature, not a bug, offering better quality control and more consistent results. “Our genomic targets have been clinically vetted by FDA, and the onus of quality is on us,” says Lazarus. “Our manufacturing processes, our sourcing of components, our continuous quality improvement and quality assurance, providing them with a fully formed kit on a specific platform — everything is designed to go together seamlessly.”

Getting to Yes

Whatever labs’ fears about these highly controlled pipelines, however, the very existence of FDA-approved tests could lighten the load for labs who want to use their own tools and processes for the same patients. “Most academic laboratories aren’t trying to develop a brand-new test that has never shown clinical validity,” says Longtine. “What we’re doing is trying to provide an efficient test that has documented clinical validity in some clinical trials, which our doctors can then use to evaluate their own patients.”

If that’s the case, academic labs may be closer than they think to meeting premarket review requirements. As the FDA has lately demonstrated with its high-profile clearances of sequencing tests, if clinical utility can be convincingly shown without a new clinical trial, the agency is open to looking at new testing methods that measure the same analytes. CLIA regulations already monitor the analytical accuracy of LDTs, which should help assure the FDA that labs can meet the same standards as approved IVDs using their own procedures.

As a regulatory counsel, Smith’s advice for labs is to be consistent with their testing pipelines, always validate any changes to the reagents or instruments used, and keep supporting data about their test results, even if it doesn’t reach the level of a clinical trial. “Make sure you have a report of some type with your supporting data saved somewhere, so you can pull that off the shelf if you need it for an FDA submission,” she says.

If a test is truly novel, she adds, the barriers to approval will be much greater, but reaching out to the FDA early can go a long way. “It is very difficult, when you’re developing a test that is far ahead of the currently accepted gold standards,” she says. “Think outside the box of traditional validations. Think about other tests or methods that are out there to help support your claim. And then go to the FDA and negotiate with them.”

As for the draft guidance due to be issued next week, it won’t be set in stone. Advisory committees are slated to advise the FDA on which LDTs should fall in high- and medium-risk categories. Even the highest risk category, which the agency has already outlined, will be subject to revision. Serrano says that laboratory pathologists, clinicians, and patient advocates will all have opportunities to weigh in, and that the rare disease and unmet need exemptions may also be broadened if the FDA sees this as in the interest of patients.

“We wanted to put something in the framework to give people a sense of what’s coming, because people want to make plans and start to get their documentation in order,” says Serrano. “We see this as starting the discussion with the community, not necessarily where we’ll end up.”

Academic labs don’t have regulatory affairs departments, and there’s no doubt that any new regulations will add costs that may be a strain on hospital resources. But they don’t have to be the end of innovation in the lab, if both the FDA and the labs it regulates keep an eye on alternative methods of validation, and accept that cutting-edge areas like personalized medicine will move only gradually to a perfect clinical understanding of the risks and benefits of testing.

In the meantime, LDT providers want to head off the idea that, because they have operated independently from the FDA, they are somehow unscrupulous or unconcerned with how their tests affect patient care.

“One of the things that seems to be lost in these discussions is that, while regulations are important, so is the professionalism of the people who are running the laboratories,” says Longtine. “You don’t want people to have a misperception that because it’s a lab developed test, it’s not a good test. It’s up to us to make people trust and understand that these are well run, well executed tests, and the patient on the other end is of the utmost importance to us.”

Originally published online by Clinical Informatics News.