Digitally scanned slides may suffice for diagnosing some skin conditions

By Marilynn Larkin

NEW YORK (Reuters Health) – In most cases, whole-slide imaging (so-called “virtual slides”) is not inferior to traditional microscopy for diagnosing various skin conditions, researchers say.

Whole-slide images (WSIs) are produced by digitally scanning conventional glass slides. Advantages over traditional microscopy have been shown in training and education, tumor boards, teleconsultation and other dermatology-related areas, according to Dr. Thomas Olsen of Wright University in Dayton, Ohio, and colleagues.

In April 2017, the U.S. Food and Drug Administration approved the first WSI system for primary diagnosis (https://bit.ly/2r4LKYh), which the authors call a “major step to facilitate widespread adoption” of the technology.

To assess whether using WSI is noninferior to traditional microscopy for diagnosing skin conditions, the team retrospectively studied 499 dermatopathology cases representing the 30 most frequent diagnoses, including 15 melanomas, at their high-volume laboratory.

Three dermatologists assessed the samples to establish interobserver ground truth consensus (GTC) diagnosis. Three different dermatologists then diagnosed the samples using both WSI and traditional microscopy, with the two methodologies separated by at least a month.

Noninferiority was determined by comparing discrepancies of paired samples by each readout method to GTC, allowing a 4% noninferiority margin between methods.

As reported in JAMA Dermatology, online October 11, the mean intraobserver concordance between WSI and traditional microscopy was 94%. Similarly, mean interobserver concordance was 94% for WSI and GTC, and 94% for traditional microscopy and GTC; it was 91% among the three methods.

While diagnoses from WSI were noninferior to those from traditional microscopy overall, WSI approached, but did not achieve, noninferiority for the subset of melanocytic lesions, for which “major discrepancies” were found.

“Discordance in (traditional microscopic) diagnosis of this challenging group is broadly recognized and further investigation of melanocytic neoplasms is recommended,” the authors state.

Dr. Jonhan Ho, director of UPMC Dermapathology in Pennsylvania, told Reuters Health, “Whole-slide images . . . have the ability to transform the way pathology practices in much the same way that digitalization of radiology studies has changed radiologists’ workflow.”

“Digitization not only allows for a more distributed workforce, it also allows for artificial intelligence to assist pathologists in making diagnoses,” he explained by email.

“The technology has undergone extensive clinical validation over the past 10-15 years, and systems are now seeking FDA validation for primary diagnosis,” he continued.

“The FDA has suggested that the industry structure their validation studies using a non-inferiority method, as seen in this publication,” he observed. “This shows that whole-slide images are not inferior to traditional microscopy, a conclusion that other large studies have also come to.”

“We will see more and more of these non-inferiority studies as more and more of these medical devices apply for and obtain FDA approval, signaling a clinical turning point in the growing adoption of digital pathology,” Dr. Ho concluded.

Dermatologist Dr. Brent Wainwright of CareMount Medical in Katonah, New York, told Reuters Health, “There are a number of advantages of the use of digital images, which include, but are not limited to, benefiting underserved or remote areas without a local pathologist, the ability to share images with consulting pathologists worldwide, and eventually the development of algorithms to help with diagnosis.”

“The ability to manipulate digital images may allow for improved diagnostic criteria,” he added, “which in the end (also) may improve our diagnostic accuracy.”

Dr. Larisa Geskin, associate professor of dermatology at Columbia University Medical Center in New York City, commented, “I agree that we will be utilizing digital technology in the future. However, I’m concerned that the major discrepancies included melanoma and dysplastic lesions, which is a huge problem.”

“In addition, the (non-inferiority) studies are not ‘real life’ scenarios,” she said in an email to Reuters Health. “They usually use academic dermatopathologists for lesions evaluation. This is another huge problem, as (academics’) diagnostic acumen is usually higher than that in private practice.”

Dr. Olsen, an owner of the laboratory information system used to view and report the scanned whole-slide images, did not respond to requests for comment.

SOURCE: https://bit.ly/2zsPUi0

JAMA Dermatol 2017.