Bringing Pathology into the Digital Age – An interview with Michael Bonham, MD, PhD, Chief Medical Officer of Proscia – Part 1 of 2
Pathologists have used the same technology to make diagnoses for the past one hundred and fifty years - tissue sections mounted on glass slides, examined under the light microscope. Will pathology be able to go fully digital and enter the twenty-first century? We’re joined by Michael Bonham MD, PhD, Chief Medical Officer of Proscia, a company looking to perfect cancer diagnosis through intelligent software.
Joseph Anderson, MD: Mike what's going on? Why haven't we been able to go fully digital in pathology? The technology's been with us for nearly twenty years. Since the early 2000’s, we've been able to scan slides.
In terms of speed, efficiency and accuracy, many pathologists would tell us that they're doing just fine without it, which may be surprising to many folks, who would naturally assume that technology can generally improve performance - but not necessarily so in this case. The secondary benefit of being able to collaborate and share cases with colleagues across the world has been nice, but certainly still somewhat inconvenient. Image analysis, which many thought would move us forward, hasn't really added much. So where do things stand now?
Michael Bonham, MD, PhD: I think that's a good starting point. For everyone to be on the same page, let’s just walk through what pathology is and how it's currently practiced. Pathology is part of seventy percent of all medical decisions. Much of what happens to a patient is really dependent on the pathologist’s diagnosis. Most people would be surprised to know that the practice of the pathology is still remarkably outdated.
Just to walk through what happens, let’s say a patient goes to a dermatologist and they have a biopsy taken. The tissue from the biopsy is packaged up in a jar, which is then sent away either to the basement of the hospital or perhaps even to a lab that's hundreds of miles away. When the tissue arrives at the lab, it gets cut into thin pieces - very thin – and mounted onto glass slides. The slides are then delivered to the pathologist – either across the hall or across the country. The slides then sit on the pathologist's desk. You could imagine that there are actually piles of the slides - hundreds of glass slides - sitting there. Then the pathologist sits down to make a diagnosis. They sit at their desk and place the glass slide under the microscope. They have to look at very small and subtle features. It can somewhat like looking for a needle in a haystack in order to find one or two cells that could signify whether that patient has cancer.
The microscope technology we use is not that much different than what you may have used in eighth grade biology class. We’ve been using these microscopes for over a hundred and fifty years.
JA: In all fairness, it has been a tremendously effective and tremendously accurate. But there's got to be more. What challenges do we face now and how are we going to get to that next level?
MB: Pathology has been very effective in terms of what it's been asked to do. The question is, “What does the future hold?” In terms of challenges, it is very clear that our current technology, workforce, and technology infrastructure will not be able to support the global demand for access to pathologist expertise throughout the world. Labs don't have enough staff to meet the demand that will be required. That's a very dramatic problem to consider.
Beyond that, there are core challenges to the practice that limit what we can do. I’ll sum them up as: 1) pathology is a manual process, 2) it is a subjective process, and 3) it is largely an inefficient process, especially in light of the vast array of incredible technologies that we have available to us in other parts of our lives.
JA: People would be extremely surprised to see how the diagnoses of their biopsies are made. It's essentially a subjective opinion - hopefully a well-informed opinion - but ultimately it is just the opinion, in some sense, of the pathologist looking through the microscope.
MB: Exactly. People want to think that when they send a sample back to the lab, that it's run through a computer and you get a yes or no answer. Unfortunately, that's not quite the case. At the end of the day, it is still dependent on a human being looking through a microscope trying to recognize patterns. These patterns can be difficult to interpret. For this reason, the process is inherently subjective. It's sort of analogous to two people looking at a piece of art. Quite often, they may have a different perspective. Sometimes there may not even be a right answer as to what something represents.
Let’s use prostate as an example. Two pathologists are looking at the same slide. One might say, “Oh, I think there's a low-grade cancer here.” The urologist treating the patient would then say, “I’m going to leave this cancer alone and not operate now.” Another pathologist, looking at the same slide, might say, “I think there's a high-grade cancer here.” The urologist then says, “We need to do surgery; we need to remove this prostate immediately.” There have been numerous studies that have looked at this. We know that the disagreement rate can be well over 30%. It's something that is a recognized problem. Clearly, it can be solved through getting second opinions and access to expert pathologists, but that's not always easy or is it available to many patients.
JA: “Pathology has performed well in terms of what pathologists have been asked to do.” I think that’s, perhaps, a good point of departure. When we are able to integrate predictive and prognostic features into our assessment, such as either Gleason grading in prostate cancer or Bloom Richardson grading in breast cancer, that’s where we can begin to add value. That's where I think there's going to be so much potential to go above and beyond what we've done before. Even pathologists may probably be surprised to realize that grading is currently not even included anywhere in clinical practice guidelines for breast cancer. There's so much more opportunity for pathologists to add value.
MB: Definitely agree. Tasks like grading tumors is a continuum. When you ask a pathologist to say, “Is this a “1,” “2,” or “3?” You give them only three choices. Often times, that's not how biology works. You might have a “1.5.” You might have a “2.5.” How do you classify that? That's where we run into difficulty. The human eye is only able to classify cases on very broad levels and not able to quantitate the vast amount of information that is present on a slide under the microscope. That leads us to a large number of cases, where the diagnosis is open to interpretation. We're being forced to place patients into buckets. Part of the reason why, in my opinion, grading is not that successful in predicting outcome, is because of this subjectivity on what is the ground truth. This is part of the promise of digital pathology, as I'm sure it will get into.
JA: Before we go too far down that road, you also mentioned challenges with staffing in the work force. We certainly hear that there are challenges regarding a shortage of pathologists. I'm not sure what to believe in the United States; it seems to be a very contentious issue. But is this true, as far as you can tell, in the United States? And worldwide is it a bigger problem?
MB: It is definitely a bigger problem worldwide. I can only speak to what was recently published. A study found that the number of pathologists in the work force decreased eighteen percent over the past ten years. Meanwhile, the average pathologist’s workload - as defined by number of new cancer cases they have to look at - went up forty two percent. To me that doesn't seem like a sustainable trend. Worldwide, there are dramatic differences. The United State has probably the most pathologists per capita - although Canada may be next. Certainly, in parts of Europe, there are dramatic shortages in pathologists. You only have to read what is happening with NHS. They are making gigantic investments in technology as the only way to address the shortage of pathologists. I talked to a pathologist in one country in Europe. He said, “Don't tell anyone this, but I have a three-week pile of cases that I have gotten to.” Those patients are waiting three weeks for results. That’s just unbelievable. Would we tolerate that in the United States?
JA: That's astounding. We certainly wouldn't want to hear about that with regards to a family member or loved-one waiting three weeks or more for a diagnosis.
MB: No, you wouldn’t. Obviously, you’d want to know right away, because everything, from that point forward, that you will do is based on that pathology result. And, furthermore, when you look at countries beyond Europe - when you look at Asia, for example - the number of pathologists is orders of magnitude lower. These are also the countries that are growing the fastest. When you look at China and India, these countries are not only growing in population, they’re growing dramatically economically. When you look at the next billion people on this planet to move into the middle class, I think seventy to eighty percent of them will be from China and India. They're going to want good health care. Who are the pathologists that are going to provide that is really a question. I think a lot about how we are going to solve this problem.
Further, there are countries in this world where there are practically no pathologists. I was recently talking to someone from a country in Southeast Asia, who said they had only four pathologists in their entire country. I said, “Four? I’ve got four pathologists just down the hall from me.” You really have to put it into perspective, just how large this global problem is.
Continued in Part 2…