Exploring the advances moving us forward in digital pathology
Digital pathology is playing an increasingly important role in modern pathology practice. This field is rapidly advancing, with the potential to improve lab workflows, increase collaboration, and ultimately, enhance patient care. To gain insight on developments in the field of digital pathology, ADVANCE spoke with Liron Pantanowitz, MD, a professor of Pathology and Biomedical Informatics at the University of Pittsburgh and the director of the Division of Pathology Informatics at the University of Pittsburgh Medical Center (UPMC); and Michael Montalto, PhD, the executive director and Head of Translational Pathology and Biomarker Technologies, Translational Medicine at Bristol-Myers Squibb and president-elect of the Digital Pathology Association (DPA) Board.
First of all, as digital pathology has evolved, there has been a shift in the market from the vendor’s perspective. When digital pathology first started, vendors focused on making hardware, such as scanners and image slides. They then moved on to software to manage, view, share and annotate images. “Now, vendors are focusing on the content—which are the apps, all the image algorithms,” Pantanowitz said. “Everyone is looking for that killer app.”
Digital pathology’s original focus was not just on improving workflow, the efficiency of sharing cases and getting them to the right experts; it was to have the ability to analyze pixels so that pathologists could get more out of images than a microscope can deliver. That, Pantanowitz said, is going to be the new era of computer-aided diagnosis. “As precision medicine evolved, and we moved into the era of personalized medicine, we’re being asked to make more and more accurate and reproducible diagnoses, which is why people are looking to computers, image algorithms and machine-learning to do just that.”
As a result of the vendor shift, the market has shifted accordingly, with new vendors coming into the market and some leaving the market. For example, GE Healthcare and UPMC’s Joint Venture in digital pathology, Omnyx LLC, was dissolved in 2016, but newer vendors are emerging, such as PathAI, which is looking to provide the content. “These vendors are looking to provide us with the software and apps; they’re not focusing anymore on the hardware,” Pantanowitz said.
Another key advancement concerns the regulatory environment, which is currently one of the biggest issues related to digital pathology for clinical use. The FDA currently classifies whole slide imaging systems for primary diagnostic use as Class III, which is the highest risk classification for medical devices. These devices therefore require the highest level of regulatory control. However, the regulatory environment has eased up in several key ways. One, the DPA has assembled a regulatory taskforce to engage the FDA to move digital pathology forward, and has been very successful at doing so, Pantanowitz said.
The FDA is working with the pathology and vendor communities to come to a solution to get an FDA product through, which is clear progress from when digital pathology regulatory issues started in 2009. “It was much more of a restrictive environment with no pathway created,” Pantanowitz explained. “Now we have a yellow brick road to follow—the FDA is helping to pave that—and there’s some light at the end of the tunnel.”
Montalto, who spear-headed efforts with the FDA on behalf of the DPA said, “The FDA has been very engaged in whole slide imaging; they have come to conferences and given talks, they’ve issued guidance documents, they’ve worked very closely with the DPA and individual digital pathology companies. This is an agency that is motivated to advance the field, provided that the evidence from the clinical trials shows that it’s safe for routine use.”
Philips recently announced the results of its pivotal U.S. clinical trial, submitted to the FDA. The trial, which compared diagnoses made by pathologists from digital whole slide imaging to making diagnoses with a microscope, met its pre-specified endpoint according to the release. Though the FDA hasn’t reviewed the application yet, Montalto feels that the Philips’ announcement indicates clear progress in getting approval for digital pathology devices for primary diagnosis applications.
“Hopefully we’ll get an FDA approval soon—and that may completely change the market,” Pantanowitz said. “We’ll move away from just having to use digital pathology for education and research only, which is really what’s been happening in the U.S., to actually using it for primary diagnosis.”
Meanwhile, several large institutions, such as UPMC and the Cleveland Clinic, are actually moving in the direction of wholeheartedly adopting digital pathology.
Quantitative Image Analysis
Moving forward, Montalto believes that quantitative image analysis or “apps,” which uses computers to extract information out of a pathology image that a pathologist would not otherwise be able to see, will be an increasingly important application for enhancing lab efficiency and potentially diagnostic accuracy.
In fact, ARUP Laboratories recently announced it is are going to partner with Applied Spectral Imaging to standardize image analysis for digital pathology workflows. According to Montalto, this indicates labs aiming for higher efficiency may start using automated image analysis to perform readings a pathologist would normally do.
Montalto also believes image analysis can play a role in the evolving world of companion diagnostics. An immunohistochemistry test for PD-L1, for example, is a companion diagnostic test for the latest immuno-oncology drugs on the market. According to Montalto, there’s been much debate and confusion in the community as to how to read these assays. This is largely due to the fact that several different assays are manufactured by different companies and each is interpreted differently and has different instructions for use.
Efforts have been made to try to harmonize these assays and make them more standard. It is likely that PD-L1 assays are only the beginning of more complex tests coming to market. “Image analysis is an obvious place where you could standardize more complex immunohistochemistry tests that will be coming out in the future,” Montalto said. Another classic companion diagnostic test is an immunohistochemistry test used to determine a HER2 status. Recently, the FDA approved image analysis tools to analyze HER2 slides in a more standardized way, although the test is not overly complex, Montalto said.
“With the increasing immuno-oncology drug pipelines in Pharma, measuring protein and cell interactions in the tumor micro-environment is critical to our understanding of how this class of drugs works and for which patients,” Montalto said. Montalto feels that image analysis could be required clinically for the use of some of these drugs. It’s already increasingly being used in pharmaceutical companies as exploratory tests to identify patients who may respond to certain drugs.
“It remains to be seen, but certainly the research and the technology is headed in that direction,” Montalto said. “The platform needs to come out on the market and needs to be approved by the FDA, and once the platform is on the market and used, companies will layer onto that platform image analysis in more advanced applications.”
More Labs Going Digital
Another major advancement in the digital pathology field is a movement away from just early adopters dappling with digital pathology to several labs around the world now going fully digital, such as in the Netherlands and Singapore. “That completely changes the picture, because now, finally someone has turned the switch on and we can actually see how this really should be used and can be used, and if there are problems, we can figure out what those problems are,” Pantanowitz said.
As more pathologists begin to use digital pathology, they will need guidance, and organizations like the College of American Pathologists have stepped up to provide guidelines on how to validate the technology. The CAP has already created a guideline for whole slide imaging, but now that more pathologists want to use image algorithms and machine learning, it is now developing a guideline for quantitative image analysis. “That’s all further evidence that the field is not just moving forward, but it’s advancing at the same time,” Pantanowitz said, who is deeply involved in the CAP Digital Pathology guideline working groups.
Overall, the future is looking bright for digital pathology. For the future, pathologists may look forward to further app development, additional guidelines and a possible increase in advanced image analysis applications to promote lab efficiency and diagnostics.