Digital pathology now, and where to from here

Create PDF

Your department at the University of Michigan has a huge business in consultations and second opinions. How has digital pathology been influencing the way that’s worked in the past year or so?
Dr. Pantanowitz: Unfortunately we haven’t capitalized on that. Digitizing our consultation business is one of the things I’ll be doing. We have close to 900 labs sending us their consult cases, and they’re still doing so by mail. We have had one or two clients who have purchased a scanner and asked if we’d be willing to accept their whole slide images instead of glass slides. So I’m in the process of converting this department to fully digital, hopefully by next year. Part of that plan will be not just to address our in-house primary diagnosis needs but also to receive digital consult slides.

For the reference lab business it works both ways. A client sends you a case digitally, because they have a scanner, and it expedites the process for them, saves on mailing costs, and they don’t worry about the slide getting lost or broken in the mail. From a reference lab point of view, MLabs has clients who send us their tissue blocks to be stained. They want us to do the technical component and then they want the slide back so they can interpret it themselves and bill for the professional component. Once the slide gets stained, we’d like to offer to scan the slide and make it immediately available digitally so they can sign it out right away and bill for their professional component, rather than have to wait for the slides to be mailed back.

What few things will be at the top of your agenda once your term as DPA president starts?
Dr. Pantanowitz: Number one, it’s time for the DPA to extend its global reach, membership, activities, and so forth. We’ve done an excellent job within the United States, but I don’t want people to view the DPA as an American or U.S.-centric organization because it’s not. Under my presidency we’ll assemble a global task group to do that, perform a gap analysis, partner with other organizations—Japanese and European societies, et cetera—and then collaboratively promote digital pathology that way.

Number two, the digital pathology community has always been about education, and I’d like to extend the education about pathology itself using digital pathology. Coming out of COVID we realized that virtual education is feasible, and that’s exactly how it’s being used in the U.S. and abroad. But there are countries with major shortages of pathologists, like Vietnam and elsewhere, that are in dire need of being educated about pathology, not just digital pathology. One way to teach them is with digital pathology tools. People have been creating nonprofit organizations—there’s one called OPEN [Open Pathology Education Network] that is assembling training courses and modules and using digital pathology to train people in Vietnam and elsewhere. And the DPA has DAPA, the Digital Anatomic Pathology Academy, which has a large repository of educational slides in partnership with PathPresenter. We’ve been building and creating content. It’s now time to deliver it broadly.

Number three, I’d like to see artificial intelligence become more mainstream and have pathologists adjust their mindsets and not feel threatened by AI—help the industry and pathology community better understand AI because we need to get to the point where people have trust in it. Yes, there’s a lot of hype around it, but at the end of the day, will a pathologist accept a diagnosis from a computer algorithm? I think we have a way to go around that. All of that requires the might of the DPA, with the regulatory groups and so forth, to push the agenda of AI.

Esther Abels, are we coming to a greater maturity and adoption cycle for digital pathology in your estimation?

Abels

Esther Abels, MSc, precision medicine and biomedical regulatory health science expert: Yes. Not only was the number of meeting registrants high but the distribution is picking up from different areas—AI companies, pathologists, histotechnologists, students, biotech companies, and pharma.

Leica started Pathology Visions and had a lot of pharma attendees. Then pathologists and different industry took it over and it became the Digital Pathology Association, and the number of pharma registrants dropped. Now we’re seeing pharma pick up again. They still use it in research, but now they’re seeing the potential of bringing it into their product development pipeline. Digital pathology can be used for, among others, quality purposes and objective quantification to enrich a clinical trial population, which could result in a reduction of number needed to treat in those trials. If you use quantitative algorithms in digital pathology, you can become more accurate—that’s the hypothesis. You can better predict who responds and who doesn’t, and with that you can have a higher success rate in your trials and accelerate your trials, for example. That is what they envision now. They also envision multiplexing more easily with digital pathology, something that pathologists cannot do, or spatial biology, looking more into the tumor microenvironment.

It’s also picking up within health care providers and pathologists. Look at what Mayo Clinic is doing with regard to its Mayo Clinic Platform. They’re seeing the added value as well.

I asked Liron if he thought the shortage of pathologists in the U.S. was increasing interest in digital pathology. What is your view? Do you think it’s fueling the uptake of digital pathology in the U.S.?
Esther Abels: It’s difficult to say yes or no. We hear there might be a shortage of pathologists, but do we really use digital pathology to the extent it can be used and with the support of pathologists? David Rimm [MD, PhD, Department of Pathology, Yale University School of Medicine] said something interesting: If we’re not going to use it as pathologists, then someone else will, so let’s start using it. If there is a shortage, we need to make sure we embrace digital pathology and use it to support us—for example, to measure—and then pathologists can take care of the other complex things, such as reading, which the machine cannot do, and focus on being the medical expert.

A prominent theme at the American Society of Clinical Oncology annual meeting and many other cancer and pathology meetings was the disparity in cancer care between what’s provided to patients in academic and tertiary care centers versus in community practice. In particular, there’s a smaller percentage of patients getting adequate biomarker testing for their initial cancer diagnosis. Might digital pathology be seen as a way to help bridge the gap between the academic medical centers and community centers? Do you see a value in that idea?
Esther Abels: Absolutely. It’s also an ethical obligation of medical practitioners and of all mankind. It will contribute to adoption and help underserved and rural areas, but also level the difference between academic centers and reference labs and community centers. It might be possible now for a patient in an underserved area or a community center to get access to reference labs. Their slides or images could easily be reviewed by someone else.

I would like to refer again to what we see at Mayo Clinic. They have included products in their platform for this type of diversity [https://bit.ly/MayoPlatform]. For example, their partnership with Mercy will ensure that the data set for algorithms being developed and validated represents the targeted population. In other words, when AI is used to develop algorithms, you also need to ensure that when you have applications run they fit the intended purpose.

You saw that at the beginning of digital pathology. We knew in certain geographic areas, not only between academic and community centers, that they were having difficulty seeing all the patients. That’s how teleconsulting started. And with the pandemic we have seen this become more efficient and effective.

We know this can be done, so it’s up to us to ensure that patients are informed, know it’s an option, and request it. And then it will level out the differences.

Would you say the understanding and application of artificial intelligence is increasing? And is its definition becoming clearer in the minds of people who observe the field?
Esther Abels: People are accepting it more and seeing its benefits and what it can do. I don’t know if they understand it more. I’m still learning a lot about artificial intelligence, how it can and will be used and what you should consider. There are pitfalls in using it but also a lot of opportunities. I don’t think we have discovered even half of it.

You can use artificial intelligence for more than digital pathology. You can use it in pathology, to link it to other data, patient reports, outcome data, which I’m a strong believer in because then you can serve patients by getting more effective treatments because you have better diagnoses and you can monitor treatments. The hope is to eventually use it to predict who will respond and what their prognosis is, and even more to identify who might be at risk and how you then can prevent disease or manage the patient.

I believe in AI. It will be beneficial to people, to patients, and that’s why we’re doing it.