Digital pathology recently took center stage in an FDA panel hearing on how best to regulate whole-slide digital imaging systems used for primary pathologic diagnosis. The Hematology and Pathology Devices Panel hearing on Oct. 22 and 23, in fact, tackled a number of issues the lab industry has been grappling with as digital imaging moves into the clinical realm.
The impetus for the panel meeting, surmises Yale pathologist and panel member John Sinard, MD, PhD, is that the FDA realized that the number of whole-slide digital imaging vendors, as well as pathology groups using the technology or looking into it, had reached a “critical mass.”
The meeting, says Ole Eichhorn, chief technology officer for Aperio, started out with “fairly technical presentations” on a variety of topics. The rubber later hit the road, though, when panelists and public speakers got to the heart of the matter: whether whole-slide digital imaging, or WSI, is as safe and effective as conventional light microscopy in routine surgical pathology—and, if so, how WSI manufacturers might prove that’s the case.
“People didn’t always agree,” Eichhorn says, “so it was a pretty spirited discussion at times.”
A major issue for manufacturers is how steep of a regulatory pathway they will have to take to be able to market their WSI systems for primary diagnosis. “There’s a point of debate,” Eichhorn says, about whether the systems will require “a premarket approval as an entirely new technology, or a 510(k) clearance comparison to the microscope.” His sense is that the FDA panel discussion seemed to be “kind of steering” toward the latter. Even though the microscope, which predates the FDA, is exempt from regulation, it’s considered to be the standard of care in pathology, he notes.
As for the economic impact of the FDA’s decision about whether to require vendors to go the approval or clearance route, Gary Syring, PE, RAC, principal consultant for Quality and Regulatory Associates LLC in Madison, Wis., says the FDA charges a company with more than $100 million in sales $4,000 to review a 510(k) application compared with $217,787 to review a premarket approval, or PMA. And a PMA almost always has some sort of clinical study of significant magnitude, he says. “The FDA may also require clinical or animal study data from a vendor submitting a 510(k), however,” he adds.
A number of vendors, including Aperio and BioImagene, have already obtained 510(k) clearances for immunohistochemistry image analysis software for breast cancer. There never was a PMA, Eichhorn says, as the products were all based on a predicate device.
Mayo Clinic pathologist Keith Kaplan, MD, who was asked to be on the FDA panel but had to decline because of a scheduling conflict, says that “when the pathologist uses image analysis software for ER or HER2, he or she is essentially writing the patient’s prescription for tamoxifen or Herceptin.” And “now the FDA is going back and saying, ‘We may need to do a PMA for the system when it’s used for primary diagnosis.’”
Eichhorn agrees the line blurs between using IHC software analysis, which gives the pathologist a score, and WSI systems used to make primary diagnoses. That’s “why there’s debate about it and why the FDA isn’t clear yet what it plans to do.”
The panel discussion, in Eichhorn’s view, appeared to reflect a consensus that ultimately it is the pathologist who is making the diagnoses using the WSI device.
The WSI devices aren’t “spitting out on a report—cancer or no cancer,” he says. If a device were to provide a diagnosis, that would “pretty clearly” require a PMA. “But digital pathology systems are really an aid to the pathologist to help them see what they need to see to make a diagnosis. That’s why the device is more like a microscope—a tool … .”
“It’s possible,” he adds, “that in the future there will be image analysis applications that actually do provide a diagnosis, but that’s not under discussion now.”
Another major question on the table is what types of studies the FDA might require vendors to conduct to vet their WSI products for primary diagnosis.
The study design issue is complex, says Robert Monroe, MD, PhD, chief medical officer for BioImagene. “Should the studies be prospective or retrospective? Should [they] look at a broad range of specimens or individually address specific types of specimens? What should a diagnosis rendered on a [WSI] system be compared to—a consensus diagnosis derived from a panel of experts, or the same pathologist’s diagnosis rendered with a microscope and glass slides?” BioImagene, he says, is taking a collaborative approach with the FDA in answering these questions and in study design.
Ajit Singh, PhD, CEO of BioImagene, who has decades of experience in the transformation of radiology from an analog to a digital world, says he left the meeting with the impression that there are two schools of thought on how to approach study design: disease-specific versus broad-based. The broad-based approach mirrors what happens in radiology where manufacturers get approval for, say, a CT machine—but “not necessarily for how to use it for the lungs or heart or abdomen, etc.” He believes the FDA will use both approaches, depending on the specific application of digital pathology.
The CAP weighed in on study design with a written consensus position, which CAP governor Paul Valenstein, MD, presented during the hearing. “The CAP believes that WSI systems must be evaluated separately for certain specimen types,” Dr. Valenstein says. He added, “Almost everyone agrees that hematology specimens and cytology specimens should be evaluated separately from regular surgical pathology specimens.” And: “Some individuals believe that small biopsies that can fit on one slide should be evaluated separately from large sections involving multiple slides.”
Dr. Valenstein says, however, that “the CAP does not believe it is practical to validate WSIs separately for each individual diagnosis a pathologist might assign.”
Dr. Kaplan predicts it will be “very difficult for the FDA to come up with a study that is agreeable to everyone in terms of validating use of whole-slide imaging systems for primary diagnosis.”
Eichhorn says Aperio’s take on the subject is that the FDA will probably begin by clearing or approving more narrow applications, such as tissue- or disease-specific, which won’t require as broad of studies.
The College also advocates evaluating WSI systems by examining intra-pathologist diagnostic reproducibility instead of examining diagnostic accuracy, Dr. Valenstein says. “Going the accuracy route is problematic for several reasons,” he explains. For one, you’d have to identify experts, “which may be controversial.” And studies focused on diagnostic accuracy typically exclude the cases where experts don’t reach consensus on a diagnosis. Yet those are the types of cases the pathologist will see in clinical practice, he points out.
And “in the real world,” Dr. Valenstein continues, “diagnoses are influenced by context as well as morphology.” Having experts come up with a “’correct’ diagnosis outside of a clinical context doesn’t reflect the way pathology is really practiced.”
The College also stressed the need for manufacturers to power their studies sufficiently to pick up small differences in precision when using WSI. For example, as the CAP consensus statement points out, if use of WSI bumped up diagnostic imprecision rates for breast cancer by only one percent, an additional 6,000 to 10,000 patients would be misdiagnosed each year. “Manufacturers may resist a requirement to perform adequately powered studies because they are expensive to conduct,” Dr. Valenstein told CAP TODAY. “We are probably talking about a trial that is sized more like a drug study than a typical in vivo device trial.” But the patient safety stakes are high, he adds.
While the FDA examines what study design to require, manufacturers are left with the fiscal dilemma of whether to invest in studies that might turn out to be the wrong type.
An invited speaker at the hearing, Michael Becich, MD, PhD, is a pathologist at the University of Pittsburgh Medical Center and a member of the scientific board of Omnyx, a GE-UPMC joint venture working on a whole-slide digital imaging system for primary diagnosis. He says the FDA hearing “won’t chill” Omnyx’s efforts to conduct a clinical study. “There’s nothing to stop us from approaching the FDA about doing an equivalency study to show digital microscopy is equivalent to the light microscope.” Omnyx has plans to do that and is “positioning” its work toward that goal.
Eichhorn says while Aperio is working closely with the FDA, it is not waiting on FDA guidance to do clinical studies for primary diagnosis of breast tissue for its “Project Pink,” which the company has publicly announced it’s working on.
“We have four study sites for Project Pink,” Eichhorn says. “At some sites, we are taking cases out of the labs’ daily workload... and in other sites, we are using cases selected from their slide libraries with the goal of having a good representation of unusual cases.”
Dr. Monroe of BioImagene describes the studies as “tough and potentially expensive” and says: “We’re actively seeking FDA guidance during the study design phase.”
The FDA, of course, regulates manufacturers in terms of the claims they can make about their devices, whereas labs and pathologists answer to CLIA. And, as Dr. Valenstein points out, some pathologists have argued that use of WSI systems to make primary diagnoses represents a legitimate off-label use of the systems. And “some laboratories that use WSI for primary diagnosis consider the systems to be part of a laboratory-developed test that can be implemented under the CLIA framework without FDA approval.”
And therein lies another source of controversy: “We don’t know much about the sort of validation studies pathologists and laboratories are conducting before implementing these systems for primary diagnosis,” says Dr. Valenstein. Nor is much known “about the sorts of ongoing controls that are being applied.”
Others view digital pathology as a train that left the station some time ago. “To me,” says Jared Schwartz, MD, PhD, “the reality of this technology is that it’s widely used around the world for a variety of purposes, and it’s been used for diagnostic purposes for a long time.” Dr. Schwartz, who is set Jan. 4 to become Aperio’s chief medical officer, sees a “disconnect” in how the FDA is conceptualizing primary diagnosis based on who views the specimen first.
He said in a letter to the FDA panel, “The primary diagnosis is the one which results in action being taken by the physician to treat or manage the patient.” This action can result when the first pathologist examines the slide, or when ‘the consultant’ or secondary examiner takes a look. “The whole reason to get a second opinion is because one is unsure of the diagnosis.” Thus, the letter continues, to “say the first person to examine a slide is limited to a traditional microscope but the second or consultant who actually makes the diagnosis can use either a traditional microscope or digitized images seems hard to justify if one’s assumptions are that the traditional microscope is superior.”
UPMC’s Dr. Becich points out that digital is being used routinely for frozen section diagnosis, which, as he puts it, is “primary diagnosis on steroids,” because the diagnosis determines how the surgeon decides to proceed.
FDA clearance or approval of WSI systems for primary diagnosis may eventually allay many current concerns about digital diagnoses, experts predict. If FDA-regulated WSI products become the standard in the industry, Eichhorn says, labs will “naturally” use those products. “Labs and pathologists will have a sense of comfort knowing what the company went through to get it cleared or approved.” In addition, “device-makers also track customer feedback after the product is on the market.”
In fact, Eichhorn believes that the FDA fears that if it doesn’t “get out in front” of digital pathology, there could be a proliferation of non-cleared/non-approved WSI systems “because the devices are useful and valuable. By the way, that’s what we have with the microscope,” which is not regulated by the FDA, he says. “Pathologists can sit down at the least expensive microscope or at the highest-end research microscope—it’s up to them to determine whether the instrument is giving them what they need to make an appropriate diagnosis.”
Consultant Syring says it’s important for the Department of Health and Human Services, which manages the FDA and CLIA, “to be comfortable that [digital imaging] devices are being handled appropriately.” If the Department of HHS isn’t on top of the issue, someone will be more likely at some point to petition Congress to take action, claiming they or a family member was misdiagnosed because of a device that wasn’t properly regulated, Syring says.
As for the next step on the regulatory front, many in the industry expect the FDA to provide a guidance document for manufacturers planning to seek clearance or approval for WSI systems marketed for making primary diagnoses.
Eichhorn predicts that before that happens, one or more companies may file for a 510(k) clearance for a WSI system to be used for primary diagnosis, Aperio being one of them. At that point, the FDA may announce its decision about whether to go the clearance or the PMA route, he predicts.
There’s still room for digital imaging to get better, but it’s come a long way and it’s here to stay.
Dr. Schwartz, for one, admits digital imaging isn’t perfect but says it’s certain to improve. “When you look at a cell phone from five or 10 years ago, you laugh. But people used them,” he says.
“Technology always wins,” says Dr. Kaplan, who, despite the potential difficulties in coming up with study designs to prove WSI is equivalent to the microscope, is sure the widespread use of digital pathology “is not a matter of if but when.”
And Dr. Becich says that as much as regulations and oversight may follow the hearing, he left the meeting assured the right thoughts are being expressed and the right questions are being asked. “I’m quite excited about the fact that this is receiving attention,” he says.
Karen Lusky is a writer in Brentwood, Tenn.