A federal advisory panel stressed the need for further research about potential expanded use of devices in detection of skin cancer, with US regulators readying for challenges in determining how well these tools need to work and how to address long-standing issues of racial equity in this field of medicine.
The Food and Drug Administration (FDA) has scheduled two meetings to gather expert feedback about managing an expected expansion in the use of skin lesion apps and devices. Outside of the US, there are apps promoted as being able to help spot skin lesions that should trigger a medical visit.
The general and plastic surgery devices panel of the FDA’s Medical Devices Advisory Committee began work on this topic Thursday, July 28, with a wide-ranging discussion about potential expanded use of computer-aided, skin lesion analyzer (SLA) devices. On Friday, the panel is considering an FDA proposal to shift the designation for an approved device for aiding dermatologists in skin cancer diagnoses from the most stringent regulatory category, class III, to the less restrictive class II.
The FDA called this meeting amid growing interest in using technology to aid in finding cancers, with some of these products already marketed to consumers outside of the US. There are presently no legally marketed, FDA-cleared or FDA-approved SLA devices indicated for use by clinicians other than dermatologists or the lay public, the agency said in a briefing memo for the Thursday meeting. There are two devices with FDA approval, though, for aiding dermatologists. The FDA approved SciBase’s Nevisense in 2017 and Mela Sciences’ MelaFind, which has fallen out of use, in 2012. Both are class III devices.
But some companies intend to offer products for consumers in the US. The company SkinVision, for example, has developed an app of the same name, which is intended to detect suspicious-looking skin spots via smartphone photos. SkinVision’s website says the product has been offered to consumers in Australia for remote skin checks since 2015. People in the Netherlands and UK also can use SkinVision, according to the company’s website. SkinVision says the company is working on providing the app for US customers, “but we are not quite there yet.”
FDA panelists Thursday repeatedly emphasized the potential risks of these devices in terms of sensitivity (how often a test correctly generates a positive result) and of specificity (how often a test correctly generates a negative result).
New tools intended to aid in detection of skin cancer might produce too many false positives and thus trigger floods of worried patients seeking care and often facing unnecessary biopsies, the FDA panelists said. But more worrisome would be FDA clearance of tools that delivered too many false negative results, leaving people unaware of their cancers.
The standards would have to be set very high for new products, especially those intended for consumers, said FDA panelist Murad Alam, MD, a dermatologist from Northwestern University, Chicago, Illinois. Current technologies for analyzing skin lesions are not yet up to that task, he said. Alam likened the situation to the hopes for self-driving cars.
“It sounds great in principle. If you read the predictions from 20 years ago, it should already have happened,” Alam said. “But we’re still struggling with that because there are serious points of failure.”
FDA panelist Veronica Rotemberg, MD, PhD, a dermatologist at Memorial Sloan Kettering Cancer Center in New York City, also argued for well-designed studies to understand how consumers and clinicians would react to new tools.
“We have to define what prospective information, in the intended use setting, we need to feel comfortable saying that these tools could be in a layperson’s hand or a primary care person’s hand,” Rotemberg said.
The studies would not need to be large, especially in the case of nonmelanoma skin cancer, which is common, she added.
“There’s too much nuance here for us to be able to say, ‘This is what would happen,’ without testing it,” Rotemberg said. “I do not think these prospective studies would be very burdensome, but they would help us understand what the burden would be and what the costs would be and what the potential harms would be.”
Because of rules against disclosing corporate information, the FDA cannot tell the public about the kinds of inquiries it already may have fielded from companies interested in selling skin cancer detection tools.
But in response to a question during the Thursday meeting, Binita Ashar, MD, a top official in the FDA’s Center for Devices and Radiological Health, said there is interest in having these kinds of products sold in the US as well.
“I can tell you that this a very timely discussion and questions that we’re posing to you are the questions that we’re encountering or that we have been grappling with,” Ashar said.
FDA panelists noted that many patients cannot get access easily to dermatology visits.
Companies seeking to develop SLA devices likely will market their tools as attempts to fill a gap that now exists in medical care.
But there will be challenges ahead in explaining to patients how to interpret readings from these tools, the FDA panelists said. Consumers should know these tools are meant to assist in diagnosis, and not to make it, they said.
“I’m not sure the layperson will hear that,” said FDA panelist Paula E. Bourelly, MD, a dermatologist from Olney, Maryland.
As a result, use of SLA tools could create tension between physicians and patients, with consumers demanding biopsies after seeing readings they don’t understand.
“I do have great concerns about the layperson feeling overly confident and reducing the provider to a technician,” she said.
The FDA panelists were not asked to cast formal votes on any issues discussed Thursday. They instead engaged in broad discussions around questions posed by the FDA in three key areas:
What standards should be used to confirm lesion diagnosis in clinical testing of the accuracy of SLA devices?
What would be acceptable true false positive and false negative results (sensitivity and specificity) for different diagnoses and users?
How can the FDA address health equity considerations based on variable incidence of skin lesions?
The FDA asked the panel to consider several scenarios for SLA devices and to discuss how standards might vary depending on the user of the device, whether it would be dermatologists, other clinicians, or consumers.
The agency sought comments in particular about using histological diagnosis (core specimen processing with a consensus diagnosis from an expert dermatopathologist panel). In the briefing document for the meeting, the FDA argued that this approach provides the greatest certainty in the diagnosis.
“Device developers, however, cite concerns, both practical and ethical, in requiring biopsy of all lesions, particularly those that appear benign,” the FDA said. “They have proposed alternate means of defining ground truth, including consensus opinion of experts (of visual or dermoscopic examination of the lesion(s)), opinion of one expert (visual or dermoscopic examination), or other methods.”
In summarizing the discussion on this question, the FDA panel chairman, Hobart W. Harris, MD, MPH, a surgeon from the University of California, San Francisco, noted that there was broad support for histological data in clinical trials of SLA devices, with some allowance for cases where more hybrid approaches would be used.
There were also suggestions offered about designing trials and the need for biopsies of lesions that are clearly benign, as this would help gather data to help in developing algorithms.
FDA panelist Alam said care should be taken in explaining to study participants that they might have to undergo biopsies that they didn’t need, as part of the larger effort to gather data. This should be detailed in the consent form, he said.
“But I also think this is a relatively minor risk,” Alam said, comparing these biopsies to the blood samples that patients in many clinical studies routinely give.
“Are all of those blood draws necessary to track the change in whatever parameters that are being tracked? Probably not,” Alam said. “I think it would be possible to explain to a reasonable patient what this entails.”
Alam noted that companies might face extra hurdles in enrolling study participants and keeping them in the trials if the FDA seeks this kind of biopsy data. “But I don’t think inconvenience to the study sponsor is a good argument” for not seeking this kind of data, said Alam, who is vice chair of the department of dermatology at Northwestern University’s Feinberg School of Medicine.
Leaving a loophole where certain kinds of clearly benign lesions don’t require a biopsy would eventually erode the quality of the research done on these devices, Alam said.
“That bar will be moved to accommodate the convenience of the sponsor, to make the study feasible,” Alam said. “And pretty soon, you’ll be missing a lot of patients that really should have biopsies.”
Acceptable Rates of False Positives, False Negatives
The FDA panel chair noted that his colleagues had strongly urged review standards that would require that the devices improve on the rates of successful catches of suspicious lesions and lower false positives. But they did not endorse specific targets regarding the sensitivity and specificity rates.
“No one seems to be comfortable with providing or preordaining” these targets, Harris said.
Panelist Deneen Hesser, MSHSA, RN, urged a deep recognition of the power of a FDA clearance in the view of consumers.
“We need to be cognizant of what the term ‘FDA-approved’ means to the lay individual,” said Hesser, who served as the patient representative on the panel. “A patient who sees that those tools are FDA approved will assume that each of those is the gold standard” in terms of expectations for delivering accurate results.
Like many of the panelists, Rotemberg, of Memorial Sloan Kettering Cancer Center, urged the FDA to gather data about how patients would react to different messages encoded in consumer-oriented products.
“If the device says, ‘You should see a dermatologist for this’ and no other information, that’s very different from [saying,] ‘That lesion is suspicious for melanoma,'” Rotemberg said.
Despite the likely difficulties in conducting trials, the FDA needs to have the data to answer key questions about patient and physician reactions to readings from new tools, Rotemberg said.
“We don’t know how many additional biopsies we would cause with a specificity of 80%” for a new SLA tool, Rotemberg said, giving an example. “We don’t know how confident a dermatologist might be to say, ‘Actually, I’m not suspicious about that lesion and we can just fudge it or not biopsy it,'” she said. “We don’t know any of that until we study it in real life.”
The panelists also urged the FDA to seek to ensure that new tools used in analyzing skin lesions improve the quality of diagnosis.
The FDA also asked the panel to weigh in on whether the agency should clear SLA tools in cases where the existing study data is drawn heavily from people considered to be at higher risk for skin cancer.
“To ensure generalizability across the entire US population, should FDA require SLAs indicated for use beyond cancerous lesions be tested in a representative US population?” FDA asked.
The three most common skin cancers — melanoma, basal cell carcinoma, and squamous cell carcinoma — are more prevalent in people with Fitzpatrick I and II skin types, who tend to get sunburns, not tans. But people of color are more likely to develop melanoma in areas that are not sun exposed, such as the sole of the foot or under fingernails or toenails.
“Due in part to lower expected risk and screening, these melanomas are often detected late,” the FDA said in the briefing document.
There was broad consensus among panelists that the FDA should encourage companies to enroll people with all skin types and tones.
But they also looked for ways that the FDA could clear devices based on initial studies conducted largely with people considered to be at higher risk, with the agency then requiring follow-up trials to see how these products would work for the general US population.
Kerry Dooley Young is a freelance journalist based in Miami Beach, Florida. She is the core topic leader on patient safety issues for the Association of Health Care Journalists. Follow her on Twitter at @kdooleyyoung.