Matthew Cooperberg: 

With that, we're going to open up for questions. Please, again, use the Q&A in Zoom if you have questions you want to pose. We're going to open it up for our advocates. I have one question I want to lead with, which is that in parallel with this paper, we are rapidly entering the era of AI-assisted pathology. And there are more papers and more tests than I think we can even keep track of, proposing AI-assisted pathology solutions in prostate cancer and elsewhere.And I'm just curious for all three of our pathologists, how you think this new classification is going to interface with Artera, with IRA Prostate, with [? Happy ?] AI, with all the other systems that are rapidly hitting our clinical armamentarium? Is this something that the computer will learn to do? Is this something that will be a new--will set a new benchmark for us to improve on with AI? How do you see this evolving?

Jesse McKenney: 

I'm happy to start with that. So AI certainly has a lot of potential. So far, the AI tests seem to be very limited in scope. So they're looking at one small thing. I think if we really want to know, then any AI test needs to be put up against unfavorable and see if it beats it. It basically needs to be put head to head.So it's hard to get this fast enough through, but I think any new test should be shown to improve on unfavorable histology before it's adopted. So I'm happy--if anybody wants me to review a cohort and go head to head against any test, I'll do it. Dr. Nguyen?

Jane Nguyen: 

I mean, adding to what Jesse just said, so it's similar within the genomic testing as well. Our clinicians currently use a lot of Decipher classification on biopsy specimens, where we may report unfavorable or favorable, and then they'll order a molecular classifier on it to see.I would say, majority of the time, there is concordance amongst what we see between molecular classifiers and what we see in the pathology. What the big question is, when there's a discordance, what does that mean? I mean, how do the urologists act upon that? And I think that's the question of the correlation with the final RP and outcomes.And I think the same thing we see is happening with Artera. We have a lot of cases that do get sent for Artera testing, and there are discordant results. So what does that mean? And I think the data is still being collected. I think there's a lot to be done still. And we'll see how it all shakes out.We did have a question in the Q&A. I'm sorry. Yeah, go ahead. Go ahead.

Jesse McKenney: 

Along those lines, exactly to what Dr. Nguyen is saying, what I think is happening is so unique to prostate cancer is when prostate cancer acquires metastatic potential, there's an end histologic type. And there's no way around that. So no matter what molecular pathway you go through to get to metastatic potential or aggressive prostate cancer, it all has to go through that histologic type.And so anything that you test likely is going to be a surrogate for cribriform, large cribriform histology or unfavorable histology. The data is so overwhelming. It's very unlikely that anything's going to have prognostic value that's not going to be associated with those findings.

Matthew Cooperberg: 

We did have a question in the Q&A about Decipher specifically. Do Decipher scores reflect histology? For anyone that's not familiar, Decipher is one of the molecular classifiers we've been using for the past 10-12 years now, which looks at 22 different genes, actually looking at how the genes are expressed in the cancer tissue itself. And Decipher has had, by far, the most research associated with it among the tests that are on the market. It's a great test. It definitely helps us separate out patients into higher and lower risk buckets.Does the Decipher reflect the histology? On average, higher grade-looking cancers have higher Decipher scores, and that's been shown in many, many papers. But there is a wide range. There are some cancers that actually don't look so bad under the microscope, but have pretty aggressive-looking genomics. There's other cancers that look concerning under the microscope, but have relatively favorable genomics.

So there is a range.Generally speaking, we get additional information by looking at the genes and the pathology together. In the unusual cases, where there's a real discordance, the genes do seem to have important information. So the cases that look reasonably OK under the microscope and have a bad Decipher, we think that this is the crystal ball telling us the future. Sooner or later, that patient is going to show up with something that looks worse under the microscope. Not universally, but usually.And on the other hand, patients that have a bad-looking tumor but a favorable genomic score--this is not that common. But when we see it, at least at prostatectomy, we see those patients may recur. They may have a rising PSA after surgery, but they very rarely show up with metastatic disease. So there's definitely independent information by thinking about the cancer from these different vantage points. Bruce, do you want to lead our--you have a set of lead questions concatenated from the advocates. Do you want to take us through the next few questions here? I think you may be muted.

Bruce: 

Thank you. Yeah, sure thing. So the first question I had was, all the patients in your study, they all had standard of care at the Cleveland Clinic. So they were followed after their prostatectomies. If they had a BCR or biochemical recurrence, they were treated. Just assuming that that's the case.And so that means that anybody who has a favorable histology after their prostatectomy, they still got to go get their PSA tested. And if their PSA rises, they're going to have to get treated. They don't get a free pass just because they have favorable histology. And I'm just asking if you could confirm that.

Jane Nguyen: 

I can answer that, Bruce. So, yes, all the 419 patients that were reviewed for the unfavorable histology and favorable histology all had clinical follow-up at the Cleveland Clinic. Yes, they all had follow-up PSA and follow-up visits with their urologists to follow if they had biochemical recurrence, or metastasis, or so on and so forth.Now, I think what you're hinting at is if in the future, in terms of treatment and follow-up, should the patient have favorable pathology on their final radical prostatectomy? Should the follow-up procedure change? Do we need to subject our patients to multiple office visits, follow-up PSA? That's a great question.Given that our study showed that there was no evidence of any recurrence, that could be something that may come down the line. I think more studies need to be done to warrant changes like that within the guidelines. But that's a great question posed from that management standpoint.

Bruce: 

OK, thanks. And I just had one other, which is that the studies are about patients who had radical prostatectomy. But even if you didn't--if you have unfavorable histology on a biopsy, that's still unfavorable. I mean, it's just the same as if you had had a radical prostatectomy as far as your probability of metastasis goes.

Jane Nguyen: 

Absolutely. 100%. And our clinicians, when we do report unfavorable on the needle core biopsy specimens, they are aware of that aggressive potential and would counsel their patients accordingly.

Bruce: 

But if it's so-called favorable on a needle biopsy, that doesn't mean so much, or at least as far as your research has gone right now.

Jane Nguyen: 

Right. Because on the needle core biopsy due to sampling bias--and we don't know what's not sampled, so there is that complexity that adds to, let's say, you only see favorable. Now, you could argue that would still--the patient would be amenable to active surveillance and a repeat biopsy. And I think with the advent of MR target, and we talked about Decipher, and some of the other tests available, the clinicians can make a decision as to is there any progression or not or potential for treatment?

Jesse McKenney: 

I would just say, we had to start with radical prostatectomy studies because we had to be able to evaluate the entire prostate to know what patterns were present, which ones weren't. So that was the only way we could put all the patterns against each other. And now, we're trying to go back and see how this can be used in biopsy, and what could predict unsampled unfavorable. Now, I have my hypothesis about what's going to win that game, but we'll hopefully know that in about a year.

Bruce: 

I'll just take one more real fast, if I could. So in your sample, 48% of the patients who had unfavorable histology ended up metastasizing. But that doesn't really apply to the population at large. It's probably a different number. And it depends on the size of the cribriform. And so there's some large range, I guess, in that probability. Just double-checking that.

Jane Nguyen: 

Absolutely. So, I mean, what we did in the study is we actually stratified for percentage of unfavorable. And we also looked at the size, and so on and so forth. So the different types of unfavorable as well. So I think there is some variability as to what may tease out. Like you said, about just under 50% of those did end up having adverse events.However, I think we're currently doing a follow-up study, looking at--trying to tease out some of those elements a little bit further to see if there's something better in terms of predicting for adverse events.

Jesse McKenney: 

I think in our validation study, my hypothesis is if you have someone who has less than 10% unfavorable histology and the cribriform is barely over the 0.25 limit, that's going to be much lower risk than 50% cribriform gland morphology that's a centimeter in diameter. So I think there's going to be a lot of possibility to sub-stratify the unfavorable group, as Dr. Nguyen said.

Bruce: 

Thank you. That was a great study, by the way. Thank you.

Cornelia Ding: 

Yes. I can address some questions in the Q&A. I think many patients, I think probably mostly from UCSF, have questions about their own situation. I guess I could say if you are a UCSF patient and you want your specimen to be re-reviewed for CANARY pattern, talk to your doctor, like Dr. Cooperberg, Dr. Carroll, and they can trigger that action.But if you are not a UCSF patient, you need a second opinion, or you are at UCSF and you want a second opinion, you could send them to Dr. Nguyen and Dr. McKenney's way. They do take consultation. I hope that answered. I think there's five questions about their own situation. So hopefully, that covers most of them.

Matthew Cooperberg: 

And David, you had some similar questions--some logistic questions along the same lines, right?

David: 

Yeah, I did. But I think for the patient--it sounds like for the patients--that a UCSF patient, as you say, contact Dr. Carroll or the others that you mentioned. What about somebody that isn't a patient from UCSF? How can they find out if they've got an unfavorable histology? And is it common for pathologists to detect and note this? In other words, at this point, how pervasive is the awareness, and how does it impact people who are not at UCSF as patients?

Jesse McKenney: 

This is very early, and it's not standard of care. So at present, this is only research. So we report it on every case at the Cleveland Clinic because our urology group has agreed that we would do this. And we've had a lot of conversations about what it means and how they're going to use it. But it's most--almost no one else is doing this.

David: 

What would you guess--may guess estimate in terms of when it might become standard of care?

Jesse McKenney: 

Unfortunately, trying to change the system that's been used for 50 years is an incredibly high hurdle. So after the validation--so when we do the validation, we're going to try to go to as high of a clinical journal as we can to try to get it in the clinical realm. But then we just have to sit back and see how other people adopt it. It's hard to drive this through in a forceful way, so we just have to see how it evolves. But I suspect it's not going to go fast.

David: 

Does that mean 5--10 years?

Jesse McKenney: 

I don't know. I don't know. No one has tried to change the Gleason grading system this radically since 1966. So I can just tell you it's not going to happen fast. There'll be a lot of resistance.

Matthew Cooperberg: 

Having said that, what is your guys' sense of--so most academic or GU subspecialized pathologists now that get referral cases will certainly read out larger, expensive cribriform. They will read out intraductal. And it seems like those two explain the majority of the unfavorables, as opposed to the complex tremorgenic and some of these other, the Cz's or the CY's. How much of the battle do we win just by getting a clear designation, whether there is large cribriform, whether there is intraductal, which most of your colleagues around the country seem to be doing?

Jesse McKenney: 

It just makes it easier for communication. The other thing is, I think--I'm trying to think how to say this best. I think that in general, there is a tendency to be too aggressive in designating something as Gleason pattern 5 in a biopsy. So that's where you're going to get the most discrepancies. So people are going to be calling things unfavorable based only on a distinct type of Gleason pattern 5, and I think that the outcome data suggests that that's too aggressive to be doing that on biopsy.

Matthew Cooperberg:

It is a good take-home, though, that I hope everyone carries from this, is you do want your pathology read by a GU expert. Whether--ideally and obviously, we're hosting this because we at UCSF are very enthusiastic about this new system. But if you have access to a referral center that is not necessarily using it, you at least want the subspecialist view on your slides. We can see some significant changes from the initial community-based read to a subspecialist read, which can have real implications for treatment.

Terrific. I think you're next.

Leszek Izdebski: 

So my question is related to something you already started alluding to, which is negative biopsy. So wondering, at this point, are there any possible changes or recommendations related to either protocol based on imaging, for example, studies, or any other predictors that in case of negative on unfavorable histology, meaning not finding unfavorable histology, you can still say, we feel that there might be relatively high probability, if it's histological finding, or better directing of the needle biopsy based on imaging correlation with potentially some of the cribriform or other negative or unfavorable histology?

Jane Nguyen:

That's a great question. And I think probably Dr. Cooperberg, because he's a practicing urologist, knows a little bit more of the algorithms. But I can tell you how here at the clinic, we approach situations like that. So the standard here is if a patient comes to clinic with an elevated PSA and there's a concern for rule out prostate cancer, the patient will undergo, perhaps, a standard 12-pack needle core biopsy just to see if there's anything atypical or cancer in the initial examination.

Following that, if they do find something, they will undergo a prostate MRI, looking to see if there's a PI-RADS visible lesion on the imaging. And then they may undergo a targeted needle core biopsy at that point to see what the pathology is within the PI-RADS lesion in and of itself.Now, let's say they undergo all of that, and there is no unfavorable. We only see favorable histology. At that point, the patient will be counseled based off of other metrics, such as PSA, number of cores positive, the volume. Oftentimes, the urologist here may order a Decipher test, like some genomic testing, to get ancillary data. We also offer IsoPSA. That's another initial serum test that they do here. And they use all these testing elements to make a well-informed decision in terms of counseling the patient as to how to proceed.If it is all favorable, oftentimes, low volume and the PSA isn't too high, patients will oftentimes remain on active surveillance, but given the option for other treatment options as well. But I think it varies by the patient to patient, to be honest. And maybe Dr. Cooperberg can chime in as well, given that he is a practicing urologist.

Matthew Cooperberg: 

Look, I mean, there is no test that exists entirely in a vacuum. So any biomarker that we do before or after diagnosis, we are always looking at that result in association with everything else that we know. And actually, it's an interesting point that in the paper--in Jane and Jesse's paper, Gleason grade still actually matters and Gleason score still matters in the statistical analysis. And PSA still matters--patient age, other health conditions, life expectancy. We consider all these things in trying to figure out who needs to go forward with an evaluation of an elevated PSA, for example. After diagnosis, we look at these factors, again, to try to figure out who needs treatment and what that treatment is going to entail.

Do they need a whole gland treatment, like surgery or radiation? Can we think about focal ablation? Can we just watch it? And none of these results is going to be a black and white, yes/no pregnancy test, take out the prostate or don't take out the prostate answer. Even unfavorable or favorable to all the extent that it dichotomizes the population very cleanly, there will still be scenarios where we will watch an unfavorable. If it's a single-core small biopsy in a man in his late 70s, early 80s who already has a bunch of other health concerns--because we have to remember, even the aggressive prostate cancers move very, very slowly. We're talking about over years and decades. So every one of these decisions, at the end of the day, is very much personalized and individualized, as it should be. And this applies for the biomarkers no less than for histology.

And there were some other questions about biomarkers in the Q&A. Decipher versus Oncotype. We've talked quite a bit about Decipher, mostly because of the three tests. There's Decipher, Oncotype, and Polaris. All of them look at gene expression in the cancer tissue. And Decipher, over the years, has done by far the most academic studies. They've done the most research. And they've been gradually increasing in market share because they give us more information than the others. They all work. I mean, they all do successfully sub-stratify risk. And which one gets used in a given urologist's practice is more about preference than anything else. I think the academic centers have worked heavily with Decipher in particular over the years, and Decipher has been gradually gaining market share. But Oncotype and Polaris certainly are still very much interpretable and are still in the mix, too. And Nathan, do you have any--

Nathan: 

Yeah, I do.

Matthew Cooperberg: 

--you had questions about biopsy.

Nathan: 

Well, so I do want to point out that, for example, I'm a member of five support groups, and there are hundreds of support groups around the USA. I think that this is going to become a pretty interesting topic in the support groups. We're going to get a lot of men who are going to be requesting these further testing to see if they have this characteristic or not.In terms of biopsy, a man who is diagnosed with cribriform is going to want to know if that diagnosis is--if he should change his treatment protocol. And there's an interesting paper by--that fellow down in the Netherlands, where they looked at cribriform and determined that if they were in the favorable characteristics, that they would be a candidate for being in active surveillance, even if they were Gleason 3 plus 4. Do the presenters here agree with that analysis?

Jane Nguyen: 

So if I understand the question correctly--so you're asking if in a grade group 2, Gleason 3+4=7, if there is cribriform present, are the patients still eligible for active surveillance?

Nathan: 

Yes.

Jane Nguyen: 

So currently, here--and again, this is how--this is very much like a urology practice type of question, because I think every urologist practices differently. But I can tell you, based off of how the urologists practice here at the Cleveland Clinic, if we report out a 3+4=7, grade group 2 without cribriform, they are definitely still eligible for active surveillance. And the percentage of pattern comes into play as to how the patients are counseled.Now, if we do report out any elements of unfavorable or cribriform, like large cribriform morphology, on the prostate biopsies, our patients here are technically not eligible for active surveillance here at the clinic. They are counseled that given that the pathology shows unfavorable or large cribriform morphology, they would be recommended for definitive treatment at that point.

Nathan: 

So today, we haven't talked very much about intraductal. What does that mean, and how is it different from cribriform?

Jane Nguyen: 

Sure. So the intraductal carcinoma is actually--it has different morphologies, but it's cancer that's colonizing the ducts of the prostate gland itself. And for our study, we put that into the element of unfavorable. It's lumped together. And we do report that out as large cribriform as well. So it would be one and the same in terms of if you're considering cribriform morphology. I think Dr. McKenney wanted to jump in and comment.

Jesse McKenney: 

No, I was going to say they're basically --they're equivalent. There's no reason to distinguish intraductal from large cribriform. It just creates--it makes the whole process more difficult for pathologists and everybody involved. So we've just lumped them all together.

Matthew Cooperberg: 

I'd make one question. We'll come to you. I want to make one extra comment about the active surveillance question, which is that we have to remember that active surveillance is not a question of treat versus don't treat. In many cases, we're talking about treating now versus treating later.And like I mentioned before, at UCSF, we don't have a hard line to say if you've got an unfavorable factor, like pathology or high Decipher, you must go immediately to treatment. I think somebody in whom we find these patterns, we will still offer, as judiciously, cautiously, with the understanding that the window of opportunity to cure localized prostate cancer, we typically measure, like I said before, in years and sometimes even in decades. For somebody who's relatively young and has years in front of him, he is going to need treatment.

But the question is, does that treatment have to happen today? And often, that answer is no. So we are still offering surveillance. This is a case where we will do genomics. Somebody who's got a small volume of, say, extensive cribriform in one core, we're considering whether we can safely wait some period of time, one year, three years. This is where I think the genomics can be particularly helpful. Stan, please.

Stan: 

Thank you, Dr. McKenney. So I think you just said that you don't differentiate intraductal from cribriform. But some of my reading seems to say that there is some intraductal that doesn't have cribriform within the duct.

Jesse McKenney: 

So intraductal carcinoma is defined as unfavorable. It's associated with high risk.So we don't worry--if there's an issue, is this large cribriform or intraductal? We already know it's unfavorable, so we don't really spend a lot of time and effort to distinguish those. That's what I'm trying to say. So if you had a case that was intraductal, but wasn't large cribriform, that would still be unfavorable.

Stan: 

But as unfavorable, perhaps?

Jesse McKenney: 

So if it meets the definition of intraductal carcinoma, by our classification, it is unfavorable.

Stan: 

I was trying to differentiate how unfavorable, but thank you.

Jesse McKenney: 

Well, we don't have a real stratification within the unfavorable yet. It's basically--at present, all we know that the larger the percent of unfavorable that you have, the higher your risk.

Stan: 

Got it. Thank you.

David: 

On occasion, where I would get into an elevator with another civilian, a non-medical professional. And the question was, before we get to the 10th floor, what's cribriform?

Jane Nguyen: 

That's a great question. I mean, so from the pathology standpoint, cribriform is a certain architectural pattern, where I think Dr. McKenney showed some pictures of it. It looks like--it's like a circle with a bunch of punched-out lumina, like a Swiss cheese kind of appearance. And it forms this round structure. And it's a distinctive pattern of growth or morphology of the cancer itself. And so that's what we define as cribriform.I gave a talk on what is cribriform? The definition of cribriform is very subjective. And what you want to call cribriform or what is not cribriform, that is heavily debated amongst GU pathologists across the board. So that's a great question, David, but that's my simple way of explaining it to you.

David: 

I think I'll take the stairs.

Jane Nguyen: 

I don't blame you. It would take a while.

Jesse McKenney: 

It's debated because pathologists are making it too hard, not because it's difficult.

Matthew Cooperberg: 

Wonderful. Well, maybe that's a perfect note to sum up on. I'd like to thank you all, again. Thank you very much for our presenters, and our discussants, and all the advocates. This has been a wonderful conversation. And we will look forward to seeing everybody again for our next Journal Club early in the new year. Have a wonderful holiday season, and we'll see you all soon.

---

Biographies:

Matthew R. Cooperberg, MD, MPH, Professor of Urology; Epidemiology & Biostatistics, Helen Diller Family Chair in Urology, UCSF Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, San Francisco, CA