Read the Full Video Transcript

Matthew Cooperberg: Hello. I'm Matt Cooperberg from the Department of Urology at the University of California in San Francisco. It is a pleasure to welcome everyone to our third Prostate Cancer Journal Club for Patients. And for those that have not joined us for the first two, a journal club is a format that we have had in medicine for many, many years whereby we present and discuss game-changing or groundbreaking publications, papers that have come out that we think are going to have a great impact on either the care that we provide or new directions in research.

So since last year, we have been periodically highlighting game-changing papers and inviting the authors of those papers to lead a discussion about the findings targeted for patients rather than for peers and other researchers. So for our third, we are highlighting a paper from Jesse McKenney and Jane Nguyen at Cleveland Clinic, which I think, and many others think, is really big news.

This is the paper here, "Proposal for an Optimized Definition of Adverse Pathology, Unfavorable Histology, That Predicts Metastatic Risk in Prostatic Adenocarcinoma Independent of Grade Group and Pathologic Stage." And this is, I think, the culmination of many years of work that Jesse and his group have led to really rethink how we consider grade.

What pathologists see under the microscope when they look at prostate cancer, I think is really a big deal in trying to figure out which prostate cancers do and do not need treatment. So it is a real pleasure to welcome both Dr. Nguyen and Dr. McKenney to discuss the findings of the paper. We're then going to have some comments from our discussant, Dr. Cornelia Ding, with our group at UCSF.

And then we will take Q&A led by our patient advocates at UCSF, who are going to introduce themselves momentarily. And then we will have open Q&A from our attendees as well. So if you have questions as we go, please use the Q&A function in Zoom. And we will field the questions after we've heard the presentations. So for starters, why don't we just do quick introductions from our advocates, if you guys want to just say hello?

Bruce Zweig: Yeah. Hi. I'm Bruce Zweig. Leszek and I share the chairmanship of the committee that we're on.

Leszek Izdebski: I am Leszek Izdebski, the one that Bruce just mentioned.

Dan Rosenfeld: Dan Rosenfeld, Advocate Support Group leader.

Nathan Roundy: Sorry. Hi. I'm Nathan Roundy, a 19-year prostate cancer survivor.

David Schwartz: David Schwartz, a 27-year prostate cancer survivor and advocate. Not cocky, but hopeful.

Matthew Cooperberg: Wonderful. All right. So with that, it is a pleasure, again, to welcome Jane and Jesse. Please, guys, take it away.

Jesse McKenney: OK, well, let me share my screen. OK, perfect. Well, thank you very much for giving me the opportunity to talk to you. It's a real pleasure. And it's actually rare for something published in a pathology journal to get a lot of clinical attention. So we're really excited about this. And I just wanted everybody to know that I'm not only a genitourinary pathologist, but I'm also a testicular cancer survivor. So I've been on all sides of this for many years.

What I'm going to do is—the presentation of our data, I could do in 90 seconds. But for you to really understand this, I really have to go through the background year by year to show you how we got to this point. And if I do that, then I think you'll really understand why this is important and why it's different. And then I'll talk a little bit at the end about future directions.

Basically, I think there comes a time when you have to think about things differently, or you get stuck in a rut. A big part of this is, what if we scrap the current system, the way we're thinking about prostate cancer, and we thought about it a different way? As many of you know, many prostate cancers will remain clinically occult for the entire life of a patient.

However, some patients will develop recurrent disease and metastasis. And that's why there's still controversy about screening, need for treatment, and optimal methods for treatment. This is complicated. So that brings up the big question of how can we predict which patients actually need treatment? And you can look at all sorts of variables. I'm a pathologist, and I've always been focused on microscopic patterns of prostate cancer.

So when I look under the microscope, what am I seeing in terms of what the tumor cells look like? And you can think of this as modified Gleason grading. And I've become convinced over the years that the histologic patterns of cancer in prostate are much more important than other types of cancer. It's just exceedingly more predictive of outcome than, say, for example, pancreatic cancer.

Now we take the term grading for granted. So these are like Webster's dictionary definitions. But grading basically means that it's the identification and grouping of microscopic features. And what that means is, things that we see under the microscope that we can record. We identify those patterns, we group them together, and we create a scale that conveys a range of how aggressive a tumor is.

And the other way to think about it is it's actually an outcome prediction model based on these features that we see under the microscope. So at the bottom, I've just put three images on the left. That's nice, well-formed glands. That's a low-grade prostate cancer. In the middle, you start to see cells fill these white spaces. And on the right, the tumor just grows in these sheets.

So prostate cancer is very variable from cancer to cancer and from area to area. And we can record and study all these variations in the histologic patterns. Now this goes back to the 1960s, when the Veterans Administration Cooperative Group wanted to study prostate cancer. And they wanted to study it across institutions.

But at the time, every center was using their own grading system. So they asked the Chief of Pathology at the Minnesota VA to come up with a grading system. And he was really the first person to come up with a method to categorize these histologic patterns. This was an original drawing that he made. And he basically recorded the patterns, sent them off to a statistician who had the outcome. And they grouped them into five different groups.

And that became Gleason pattern 1, 2, 3, 4, and 5. So you'll see that when I show you our methodology, we didn't create anything novel. We just went back to the basics that Gleason did originally. Now, over time, from the '60s to 2016, the grading system has evolved. So there's been gradual changes over the course of time.

So the way he graded in 1966 is not necessarily exactly the way we grade in 2016 or today. And the main reason is because we entered the PSA screening era. So we went to Gleason's time, when really almost all of the cancers were pretty aggressive because the men presented with symptomatology, to patients who were discovered at screening. So you bring in a lot more cancers that are indolent.

So you went from an era where it was basically good news, bad news to we're going to try to use this grading to decide who to treat or who to have surveillance. And that really changes the game. And it really puts the grading to a test that it hadn't gone up against yet. And if we go back—this is kind of philosophical—but if we go back to the way we're taught science in school, everyone tells you, well, what we're teaching you is the current model or the current hypothesis.

And unfortunately, in medicine, because we're dealing with patients and the realities of the difficult things they're facing, we get away from that and think of it more as an objective reality when we really are talking about the current model or current hypothesis. So if you have any major change in how any disease is screened or diagnosed or treated, then the way that the pathologist classifies the tumor may be no longer optimized for the clinical setting. And that's where the problem we got into with prostate cancer.

Now this is just a collage. I'm not doing a review of the literature. But the pathology literature is overwhelming that Gleason pattern 4 prostate cancer is not equal, some are very indolent, and some are very aggressive. And that's really the basis of what we're looking at. If you look at the big study of grade group, you get this nice split.

So this is biochemical recurrence-free survival. So grade group 5 had lower survival compared to grade group 1. And these are all statistically significant. Now this is based on 6,000 patients. If you were to substratify these by the type of 4, I can tell you because we've done this, that in grade group 2, you're going to have some of those patients that are almost identical to grade group 1, and some that are almost identical to grade group 3.

And the more you subtype the pattern 4, the more these lines converge into two groups. OK. So what I would say is the grade groups work for a large population, but they're not optimized for the individual patient. So you're always going to have a measurable misclassification rate that's most easily recognizable within grade group 2.

So our hypothesis was based on some incorrect assumptions. So all Gleason pattern 4 is aggressive, we think is not true. And AJCC stage pT3, so that's extraprostatic extension, is a marker of aggressive cancer. And the overwhelming data shows that neither one of these have specificity. And so the question is if these don't have specificity, is there anything that does? And we think the answer is yes.

So the picture on the left shows what looks like round circles. And that's a well-differentiated or low-grade prostate cancer. The one in the middle, the round circles start to become filled by cells, so you're gradually becoming more complex. And the one on the right is completely filled with all these white spaces that kind of looks like Swiss cheese. And that one on the right is what people refer to as cribriform gland morphology.

And the literature is one of the most overwhelming clinical outcome sets of data I've ever seen. The cribriform is associated with high risk for biochemical recurrence and metastasis. And in fact, it's difficult to show that it has any difference from Gleason pattern 5. So let's look at what we did. We call this the "Canary" methodology because this came out of our research that was supported by the Canary Foundation in California.

We decided that we were going to go back to Gleason and break down everything into its individual parts and try to rebuild it. So over the course starting around the mid-2000s to the late 2000s, I started to record all the types of patterns that I thought I could reproducibly identify. And this was my final hand-drawn schematic diagram.

We converted this into something more official. And we created these arbitrary designators of letters. So you could refer to each pattern by a letter. When we say Cx here in our group, we all know exactly what pattern that is, because we wanted to take each individual pattern or groups of patterns and see how they associate with clinical outcome. And we published our first study in 2016 as a proof of principle.

So after classifying radical prostatectomies with these patterns, we could have an algorithm to make groups based on which patterns were present. And if you looked at clinical outcome, you could see that the different groups could stratify. So again, this is biochemical recurrence. So the black line is the lowest biochemical recurrence, the blue line is the best. And you got distinct groups by looking at the pattern. So again, we just wanted to prove that this would work.

Now the big problem is there's all this data on cribriform. But people talk about what's large cribriform and what's small cribriform and what's the difference in the definition and the clinical outcome. And the definitions range all over the map. And we didn't have a good definition. There were two consensus groups that defined cribriform. But unfortunately, they didn't define large. So this was still the biggest hurdle to creating a new system.

So Dr. Chan, Emily Chan, who's now at Stanford, she was at UCSF at the time, took some of our original data from the Canary Group. And she went back and measured the dimension of every cribriform. And we did statistical modeling to see what was the optimal cutoff if you looked at biochemical recurrence. And so you can see, A, here, the red line here is cribriform absent, cribriform present.

So cribriform is associated with worsened biochemical recurrence, which has been shown by almost every study. And as you look at different iterations of the sizes, you get down to the two most optimized groups. And that's a size cutoff of greater than 0.25 millimeters. And I was very excited when she sent that data to me, because one of the best post-radical prostatectomy cohorts that I've ever seen showed that the best predictor of outcome was the size of perineural invasion.

And I can tell you, by looking at tens of thousands of prostatectomies in my life at this point, the only way to get large perineural invasion is if it's cribriform. So basically, they were measuring cribriform. And they got the exact same number we did, which was very exciting to me. We put this up in a validation cohort and showed that it was very good at predicting metastasis-free survival.

So the patients who had no large cribriform and no Gleason pattern 5 had the best. There were only six patients with metastasis. Large cribriform with no pattern 5 and the presence of any pattern 5 both had much worse in biochemical recurrence but were very close.

Matthew Cooperberg: Jesse, can you explain these plots? Explain the Kaplan-Meier plot for 30 seconds here.

Jesse McKenney: Sure. So basically, distant metastasis-free survival—so these patients, when you get to 20 to 30 years, less than 5% had biochemical recurrence, compared to patients with large cribriform or any pattern 5, where you're getting, after 20 years, less than 50% of the patients were metastasis-free. Is that sufficient, Matt?

Matthew Cooperberg: Yeah. And just for those who are not familiar with looking at data in this way, these are called Kaplan-Meier plots. We use these a lot in clinical research. You want to stay at the top of one of these plots here. OK. The percentages that we see here, this is the percentage of patients who have not had the outcome. We call it survival.

Survival is a statistical term here. Metastasis-free survival means you are alive and well without the metastasis. And you want to stay at the top, this means nearly 100% of the patients are alive without metastasis at a certain time point, at 30 years here, right? Whereas here, at 10 years, for any 5 curve, for example, half of all men have—we use the term failed, meaning they now have a metastasis at 10 years. Just in terms of how to interpret these graphs—

Jesse McKenney: Thank you. So the thing that was interesting to us is that patients with no large cribriform and no pattern 5, six had metastasis. So is there some kind of signal there that we could have recognized to better classify those patients that has never been seen before? So after about 15 years of studying this, with all this data, we thought that it suggested that you could have a dichotomous classification for predicting metastatic potential.

Cribriform seemed to be driving at least 90% of the bad outcomes. The problem is, in the pathology literature, people have tried to combine cribriform with grade, and you get these extraordinarily complicated systems with way too many categories. And I think that's the main reason this has never been adopted. It's buried in the pathology literature. And it's too complicated. So we ask ourselves what nomenclature has worked well.

Dr. Bruce Beckwith was one of the world's experts in pediatric renal tumors. And he came up with a system for pediatric tumors that started with Wilms tumor, which is a malignant kidney tumor of children. And he came up with dichotomous—favorable histology and unfavorable histology. And this is just to show you the way it plays out. So neuroblastoma is a malignant tumor of children.

And I just want to prove the point that—look how complicated the classification is. It's within the boxes. And basically, they've defined it into favorable histology and unfavorable histology so that it's very easy to communicate which patients need more aggressive therapy and which ones don't. And we thought that this way of approaching it was probably going to be the easiest for terms of communication.

So now, this leads us to our publication that came out this year. Based on all of our data, we decided that these patterns within the red boxes would be our unfavorable histology. So Dr. Nguyen actually entered this equation in about 2016. She came to the clinic and started learning the Canary system as she was really learning sub-specialization in pathology.

So she had the unique experience of learning the Canary methodology at the same time she was becoming a sub-specialist in GU pathology, which is a unique way to learn prostate cancer. And so she's been an enormous help to me in getting these projects pushed through and being able to talk to somebody about these Canary patterns. So she scored whether these patterns were present or absent in 419 patients who had undergone radical prostatectomy.

And this is just showing some of the histologic patterns. I won't go into great detail. But all of these are unfavorable histology. So let's look at these Kaplan-Meier curves again. The top one is overall biochemical recurrence. So the blue line is our patients with favorable histology, the red is unfavorable histology. So you can see that a lower line means worse biochemical recurrence. So more of these patients had biochemical recurrence.

When we look at distant metastasis, you have this same nice split between favorable and unfavorable. But the interesting thing is you actually have 0 metastases in the favorable group. Now, what if we break that up into—we look at it in subgroups? So let's talk about stage, because a lot of publications use stage as their marker of what adverse pathology is.

So if we look at tumors that are limited to the prostate, unfavorable histology has significant metastatic rate, even without extraprostatic extension. And if we flip that to the bottom one, patients who have favorable histology and have extraprostatic extension still lack metastatic potential. So this is showing that this classification, histologically, is outperforming the way we stage patients.

When we break it up by grade group, you see the same thing. In grade group 2, the favorable histology had no mets. Unfavorable histology had metastatic rate. Now, grade group 3, it's very rare to have grade group 3 favorable histology in a radical prostatectomy. But even these few cases that had favorable histology, none of those metastasize.

Now, if you remember, I alluded to there was a small group of patients who had mets in the original study who didn't have large cribriform. And we think that we've identified what those patterns are, and it's something that we've always called—this is just an arbitrary designator. But this is Canary pattern Cz. So all of these patterns would not qualify for large cribriform based on standard definitions. But this is high-risk prostate cancer.

And can you improve specificity with an unfavorable? So this is going to be the next big step, is once you know who's unfavorable, you can risk-stratify within that group. So the more unfavorable histology you have, the higher the metastatic rate. And the other thing we want to add in that we think is going to be important is, the higher dimension of cribriform you get beyond the 0.25 millimeters, at least our early data is showing, the higher your metastatic rate. So when you get up to 0.5 millimeter or 1 centimeter, your metastatic rate is going to be much higher.

In this study that we published, we also had a separate cohort of 250 patients post-radical prostatectomy, all with pathologically proven metastatic disease. And now, this is a biased study because we knew they had mets. But we couldn't find a single patient who had 100% pure favorable histology. And in our anecdotal experience, our group has never seen a metastatic prostate cancer with only favorable histology.

And probably my personal cohort of what I've seen is probably over 20,000 RPs at this point. So I think that data is pretty strong. So our conclusions are that unfavorable histology at radical prostatectomy is associated with potential for metastatic disease. The majority of this is driven by large cribriform gland histology.

It has a sensitivity of about 100%, but the specificity is only about 51%. So that's what we want to work on next. It has better predictive value than grade group and stage. Any amount of unfavorable denotes some risk. And so where we see this—the big problem is the Gleason summation method. So particularly for Gleason 3 plus 4 equals 7 or grade group 2, there's a lot of misclassification that takes place.

So the next things we're doing—we have a validation cohort of about 450 patients. That should be done by the end of the year. The big study preoperatively, what's the best predictor of a patient having unsampled unfavorable histology? And so that's our next big study. We have a cohort of 700 RPs, about 550 with in-house paired biopsies. We're probably seven to eight months away from having that done.

And I think the big questions are, should unfavorable histology define progression in active surveillance? Should unfavorable histology define adverse pathology in all studies? And can we develop a clinical nomogram that's really based around unfavorable histology? And can we convince the NCCN to make some improvements based on all of this? So I really appreciate your interest in this and your attention. I hope that by giving you all that background information, it makes it more clear what we did and how we got here.

Matthew Cooperberg: Wonderful. Thank you very much. All right. Cornelia, do you want to spend a couple of minutes just giving another perspective on these findings?

Cornelia Ding: Yes. Thank you for this opportunity. This is definitely one of the most exciting papers, I think, in this year in the prostate world. A little bit about myself—I joined UCSF in 2023. And so I just started. But I'm training. So I actually spent two weeks with Dr. McKenney and Dr. Nguyen at Cleveland Clinic last year to learn this pattern.

So a little perspective is that for most pathologists, it's difficult to adapt to a new system. However, just within that two weeks, I feel comfortable to apply that in the daily cases I encountered at UCSF. And actually, Dr. McKenney made a very nice training set. And so when I came back to UCSF, some of my colleagues, who never heard this pattern before or have a little knowledge because some of Dr. McKenney's collaborators at UCSF, actually could spend two to four weeks to become familiar and comfortable reporting these patterns.

So from a pathologist's perspective, it's actually not that hard to adapt, although definitely some difficulties at the beginning, worrying about if this is not easy. However, I think this is really important for patients because basically, we are summarizing these complicated histology patterns into something easier to communicate. So instead of five grade groups, we are actually talking about three or two groups to facilitate clinical management. So this is great.

I think one comment I will have is that this paper is focusing on radical prostatectomy. But we know a significant portion of prostate cancer patients may opt for different treatments, like chemotherapy or radiation. So they might not have radical prostatectomy. So how to apply this system into patients who only have biopsy for histology evaluation will be a really interesting question, which I know Cleveland Clinic is working on. And we are also trying to validate in our patient cohort too.

And I guess one final comment is that no system is perfect. We are basically trying to put different patients into two buckets, unfavorable versus favorable. And we all know that every patient is different, has different biology and different life priorities. So this is not perfect. However, I hope this is a newer system that could make communication between different sub-specialties and with patients easier and standardize our reporting. OK. Thank you.

Matthew Cooperberg: Yeah. Wonderful. Thank you.

Jesse McKenney: And I will say, it is true that no system is perfect. But I've been in this game a long time. I've been doing work with prostate cancer biomarkers and risk stratification for a long time. I'm not sure I've ever seen any test that splits the population so cleanly between a group that is and is not at risk of prostate cancer metastasis as this classification, which is why, like I said at the beginning, this is a really big deal.