Matthew Cooperberg: So now, we're going to dive into our questions. It is my pleasure to once again introduce the UCSF patient advocates, who have really been the drivers of this journal club effort. So we're joined by Leszek Izdebski, Stan Rosenfeld, Nathan Roundy, Bruce Zweig, and David Schwartz. And we will now dive into questions.

Nathan, I think you have the first. But you're muted.

Leszek Izdebski: You're on mute, Nathan.

Nathan Roundy: Got it. So thank you to the doctors for this great presentation. It's very, very interesting. So patients should remember that this test is only approved for radiotherapy patients trying to decide if they should get hormone therapy or not. But can we extend the use off-label as a more general risk-stratification tool?

Jonathan Tward: Yeah. So it's interesting. I think you've got three physician panelists, Dr. Cooperberg, Dr. Austin and I. I think we're all probably fairly aligned, but probably differ a little bit in our enthusiasm for it.

So from my perspective, the general principle here is any biomarker, any biomarker that is trustworthy, and Dr. Cooperberg mentioned UCSF-CAPRA, for instance, which he developed. If you have a biomarker and it can prognosticate metastasis for you, if that number is really low, my position, my opinion on it, is if that number is 1% or 2% risk, then that metastasis prediction is what you should potentially be using to make the decision, let's say, on surveillance or not.

So even though Dr. Ross is completely correct that the Artera test I showed was only done in radiation patients. And in fact, none of them got surveilled, that is 100% accurate. So we don't actually know if it was safe to use them in the surveilled patients, I think it's reasonable to consider that. Other doctors may not think that it's very reasonable.

All of the commercial biomarkers on the market, that's Artera, Decipher, Prolaris, and GPS, will give a man on a score report their 10-year risk of metastasis. So one thing that maybe you could use to compare all these biomarkers is that one thing because they report that on the same score report.

But interestingly enough, on an active surveillance question, Prolaris, for example, did in fact develop their biomarker in that cohort. And so maybe one might have more assurance on that particular question. But I mentioned earlier the God-like test versus the coin flip.

And Dr. Ross was very correct in that what happens if there's a scratch on the slide or it's faded or whatever? And what I'll say is there's a lot of additional information that needs to be uncovered and a lot of questions and concerns. But when you look at those so-called AUCs, which were those numbers, that bakes those problems into the cake, and these new biomarkers, whether you talk about the molecular ones or Dr. Cooperberg's completely free-to-use CAPRA, they do better than NCCN.

And so they account for the heterogeneity. But none of these biomarkers are integrating MRI information, or PSMA scans, or your germline genetics. And I think Dr. Ross and Dr. Cooperberg would probably all agree with me that these biomarkers are just one tool, but they're not the only tool. We have to integrate the entirety of the individual that's in front of us and not treat people like a population statistic.

Nathan Roundy: Thank you.

Matthew Cooperberg: Yeah, I would add on to that, and I agree completely. And the last comment is one of the most important. These are tools. And one of the things that has irked a lot of us in this space, we have been doing biomarker validation work at UCSF for over 15 years at this point, and we're working a lot with Artera at UCSF, I should say.

And these tests get developed in academic collaborations, like the one Jonathan has described working with institutions like us, like others, with a lot of terrific statisticians and physicians. And we put together the science behind the test. And then the report that you all see generally gets written by the marketing department at the company in question.

And that is honestly not the best case. It's not the best—it's not the best situation. It is how these things go. But the reality is when you look at these reports, Artera and Prolaris do this, too. There's a little box at the bottom that says, this patient should go on active surveillance. The newer version of the report does this.

And that, in my mind, is completely inappropriate. None of these tests should be construed as telling you what to do. They are providing information. In a case like this and based on the evidence in the paper we just saw, they provide additional valuable information above and beyond what we knew before we ordered the test. But this is another data point that you need to think about with your physician team, with your urologist, your radiation oncologist, the medical oncologist, if you're seeing one. It is our job to integrate this information with everything else that we know to try to help you come to the decision that is right for you. None of these things is a pregnancy test. That's what everybody wants. Turns blue, you take out the prostate. It doesn't turn blue, we leave it in. That is just not the way the biology works. We will never have a test that works that consistently. It's just not the way the biology is. All right. Next one, I think, is Bruce.

Bruce Zweig: All right. Yeah, thank you. That was a great presentation, Dr. Tward. You really made this stuff understandable. The question I have is, can Artera identify how much a patient's risk of metastasis will decline if the patient has ADT with his radiation therapy?

And then there's another side to that, which is, does it depend on where you are on the scale? Some people are kind of closer to the cutoff points between low risk and intermediate risk. And just how does that factor in as well?

Jonathan Tward: Yeah, that's a— that's such a good question because it gets to the difference between prognostic and predictive tests. So in the question of ADT use, the two tests on the market that directly speak to that with some kind of good evidence are the Artera test and Prolaris.

And full disclosure, I also developed the risk thresholds and the ADT work for Prolaris. I've worked with Decipher. I work with everybody. But on the Artera test, we'll start there. Regardless of the risk of metastasis, if you're an intermediate-risk patient specifically, whether that's favorable or unfavorable, the Artera test had that so-called predictive signature.

And it can simply tell you whether the drug will work or not, regardless of the risk of mets. It's a futility test. That's a predictive test. So it won't do what you asked, which is the difference in metastasis using one or the other, whereas the Prolaris test does it differently, which is they looked at patients who did or did not receive hormone therapy independently, and then they also used a lot of really interesting modeling based on NRG trials to give you that specific answer.

So it gets a little muddied here. Because Artera is only validated for, let's call it, futility of ADT in the intermediate-risk space. It won't help you in the high-risk patient. It won't help you in the low-risk patient. Whereas Prolaris's claim, and again, you know I'm fond of it because it's actually my work, is across the risk spectrum, it should give you a really good idea of that exact difference.

It might go from 12% to 8%. But I think what a lot of men don't understand, and this is a very critical point about hormone, is that as a general principle, what we've historically thought of as hormone therapy is no matter what risk group you're in, if you get radiation therapy and you use hormone therapy, it'll reduce the risk of metastasis by about 40%.

And so it doesn't matter if you start with a 3% risk, it'll be, let's call it, close to 1.2%. If you start with a 30% risk, it'll go to close to 20%. It is the same effect, effectively. But you got to understand that initial risk.

And that's why that discreet precision on all these tests or nomograms is helpful. It's nice to know if it's only 2%. Because then you'll want to know if it's worth decreasing that by 40% or not. Most men, most men, when polled—not just men, but experts agree that the majority of men will accept all the side effects and risks of hormone therapy when you drive the absolute risk down by 5%.

That's where most men are. But most men aren't you, an individual. Men deserve to have the information in front of them so they can make the individual decision. And that's why I like nomograms, and biomarkers, and tests that can give men an idea of in you, sir, it's exactly this. That makes for a much more informed decision regardless of what tool you're using.

Matthew Cooperberg: Let's take— I want to take a couple out of the Q&A because there's a theme here asking about Decipher versus Artera. Both. Is this going to replace Decipher? And what should a man do who has had both an Artera and a Decipher and they are discordant?

And I'm seeing these two. High Decipher, lower Artera. I think more common than the other way around in my experience so far, but thoughts from both you guys.

Jonathan Tward: Yeah, I think— Yeah, Dr. Ross, please. You first.

Ashley Ross: Yeah, a couple of things. One, and I know it's hard for the patients, and it's confusing. I won't go back to the previous question, but there's a lot of nuance for the different biomarkers. I actually don't agree that Prolaris was developed in the active surveillance cohort. It was a watchful waiting cohort.

It was a registry, very different than active surveillance. I don't think the level of data is high for Prolaris for using ADT decisions. That was also done with a lot of modeling and retrospective analysis and not linked to randomized trials. I don't think the Prolaris signature looks at ADT signatures.

It's a cell cycle progression signature and score. I do think the strongest data for ArteraAI is in that ADT decision-making group, but in a narrow window of intermediate-risk patients and needs more validation. So the problem is there's a lot of controversy, even among prostate cancer experts, in how we understand and interpret the data and interpret for you, the patients. And so that becomes a lot of issues.

But I'll tell you this for the Decipher, Artera questions. As I mentioned before, when I look at the data I have and maybe extrapolate it a little bit dangerously, what I would tell you for Artera is if it's high, it's bad. If it's low, I don't know.

The low guys, like Dr. Tward said, these are all guys that got treatment. We don't know if we're thinking about a surveillance decision. You got a lower Artera. I don't know if you're safe.

Because the data I'm looking at is people who got treatment, even in my prostatectomy paper from the PLCO with them, those guys got prostatectomies. These guys all got radiation. The Decipher, if it's high, it's bad in most cases. In the Artera, if it's high, it's bad.

If I had a low Decipher and a high Artera, my antenna would be way up. And so in the absence of data, if I'm getting a signal that something's going on, so if the smoke alarm goes off in my house or if some alarm goes off, I don't see smoke, I don't smell smoke, my anxiety is way up, and I'm maybe going to get out of the house, even if I can't find something until I look for it to burn down.

But if everything's good, and I have a detector that's not a great detector, and it's telling me everything's fine, but I'm smelling smoke, I'm seeing smoke, I get pretty worried. So I think when there's discordance, play towards the higher-risk thing to keep you out of trouble. But that's, again, one provider's opinion.

Jonathan Tward: So I have some insight here. And that you're seeing, let's call it, healthy disagreement among experts. I think the data is strong for a lot of these biomarkers. Dr. Ross is skeptical. And that's healthy.

That's what's going to continue to push all these companies to develop better levels of evidence. But what's interesting is the discordant data thing. And this is unpublished, but I've been collecting patients with all three, Decipher, Prolaris, Artera. I'm not going to tell you what I found, but there's a lot of discordance.

And this gets back to the fact that a risk group is, let's call it, 65% accurate at knowing the truth. And any one of these tests is maybe up at around 80% accurate. So one in five times, you're going to get misled. And that's why there's discordance.

And so I don't want to take a position on if I have test A versus test B, who do I trust? I think it goes back to, all right, if you've got discordant information, you need to start looking for other pieces of information. What does the MRI look like? Percent positive cores.

PSA doubling time. PSA density. These are cheap, old-fashioned things. I will tell you that PSA density, I've done a lot of work, is remarkably good at prognosticating outcomes.

So every piece of information we have is discordant in a patient, including Gleason score, which was the most vaunted prognosticator of all time. You give 10 pathologists a slide and have them interpret it, it's going to be discordant. So we deal with discordant data and discrepant data all the time.

Maybe I've drunk the Kool-Aid, but I think that these tools usually add when I'm making treatment decisions. And so I do tend to trust the biomarkers more than a risk group, which is kind of the point. Someone asked if Decipher is going to be killed.

Decipher is a very good prognosticator. And I will tell you, though, that it was developed specifically to determine the risk of death from prostate cancer after radical prostatectomy. So again, and this goes very much to Dr. Ross's point earlier, what were the tests developed for? Who were they tested in?

Can you shoehorn them into every single case? No. So caveat emptor, when you're ordering these tests, what was the population they studied? And is it fair to take their thresholds and move them into every single scenario in oncology? No.

Matthew Cooperberg: I'll make two more comments on this one. Number one is the space is going to converge very quickly. So Decipher, of course, has their own digital imaging effort. Artera—there's talk of co-marketing with Tempus Oncology, which is a genomics company.

What we are seeing is version 1, maybe even version 0.95. And this space is going to move faster than anything we've seen in the biomarker world before, is my prediction. I think many people's predictions here. Decipher will be in better position than many of the others because they have the whole genome.

When you order Decipher, you're not just getting the 22 genes; you get all 30,000 genes. It's just the 22 that go into the score. And Decipher has been looked at, in fairness, for some of the same ADT intensification questions.

Felix Feng published a paper on this from some of these same NRG cohorts a couple of years ago. And it does work. So I think the question of when you use what, where is— we definitely do not know the answer yet. I think this is going to be fertile ground for all of us that research how we take care of prostate cancer patients going forward.

I make one other comment because there's questions here about, what do you do if you're low risk with a bad marker score? You do need to remember that men who have NCCN low-risk disease do not die of prostate cancer, basically full stop, as long as we have accurately staged that cancer.

A first biopsy can absolutely underdiagnose cancers. So we need to make sure that we've done imaging, that we have a good biopsy, that we're fairly confident we have not missed something. But when I see a low-risk cancer, grade group 1 with a high Decipher or a high Artera, this is somebody that I will usually tell you needs treatment at some point. But that point does not need to be today. And we need to be very careful that we are not kicking men out of surveillance and into an overtreatment mindset just based on a biomarker, no matter what that marker is.

Ashley Ross: I totally agree with that. And it should prompt, as Dr. Cooperberg is insinuating to, those bad scores often can prompt an early confirmatory, well-done MRI-guided biopsy. Absolutely agree with that.

Matthew Cooperberg: Leszek, I think you're next.

Leszek Izdebski: Yeah, my question is related to patients who are typically classified within intermediate high risk, intermediate. Very often, and it relates to radical prostatectomy. I'm actually glad Dr. Ross mentioned already looking at some patients with radical prostatectomy in Artera.

Can Artera, for those patients, be used after radical prostatectomy to decide about adjuvant radiation therapy versus waiting until biochemical recurrences at 0.2, and at that point deciding about radiation therapy? And similarly for radical prostatectomy patient adjuvant ADT treatment?

Ashley Ross: Like I said, we looked at—we just wanted to kick the tires a little bit on, is the digital pathology pulling out useful information beyond standard clinical markers? Kind of like what I showed you in a group where we had known outcomes, like prostate cancer-specific mortality.

We looked at the PLCO series. They were nice enough to share some of their digitized slides so we'd have long-term outcomes. And we saw some discrimination. And it just showed a signal. Artera wasn't trained on radical prostatectomy specimens.

It wasn't trained for that endpoint that you're talking about. And even though some of these tests, for example, the Decipher, trained originally on metastasis development after radical prostatectomy but then validated for other outcomes—at biopsy, how you do on radiation or in post-op settings—Artera hasn't had that, and we don't know what it's looking at.

It's like it's not—it's not as plug and play, in my opinion, as these other tests because it's not looking at a general characteristic of the cancer, per se. We actually don't know how it's weighting its different pieces. And even there's a different Artera algorithm for different outcomes.

And so it even—that even tells you, because it's trying to optimize a model, which is a good and bad thing. One side of the sword is that they will eventually be able to look at that precisely, and maybe they can get precise optimization on a model just around that that takes the features. But the bad part is it's so individualized that glove may only fit one person.

Jonathan Tward: I'm a radiation oncologist, so I'm always pondering, should I add ADT in a so-called salvage patient? And I think the whole oncologic community has abandoned the idea of what's called adjuvant therapy. We all want to see PSA rising after surgery. But this goes back to, again, what was the tool designed to ask? And so Decipher, being a post-prostatectomy tool at its outset, performs very well—or I should say better than any of these other tests for those particular questions.

And so I do use something like a Decipher test for answering questions like, should I use hormone therapy or not in a man with a PSA that's rising after surgery? As long as that PSA is under 0.6, and they don't have extremely high-risk features like Gleason 9, because they showed very nicely.

And again, some post-hoc analysis of randomized trials that it can help stratify men into people who do and don't benefit. So if there's really a lesson in today's discussion, pick the tool for the job you're asking. There's no one-size-fits-all tool here.

All these tools and tests and nomograms that are available can have a niche. And the commercial interests here would like you to order one thing all the time for every reason. And so you need to have your physician really engaged with you in this discussion.

I saw there was someone in the chat who said they got an Artera, a Decipher, and a Prolaris. And I'm thinking, holy moly. That person has a lot of information, but I think they need a human to help them sit in front of them and interpret all this. Because that sounds like possibly too much information, and you'll just end up with paralysis of indecision.

Matthew Cooperberg: There was a question in the chat. Well, first of all, are we paying—all three of us that are the physicians on this call have worked with pretty much every biomarker company out there to some extent or another.

They have funded research studies at our institutions. In some cases, we've done consulting work for them. I would argue, having known both Dr. Ross and Dr. Tward for a long time, none of us has any inherent bias or financial bias in favor of one company or another. And I hope that's coming clear in these comments.

We all want to see the best test win. And the market has really shifted over the years. Some tests look like they're going to take over and then fall off. Others rise. Things are going to change even faster in the path AI era.

So I think the flip side, the second part of that question, are these expensive tests contributing to health disparities? They may be. One of our former fellows, Mike Liebman, has a great study out there on the unequal uptake of these markers.

There's parts of the country where there have been zero biomarker tests ordered. There's others—not where you would think, like North Dakota, for example—where half of all men with prostate cancer diagnoses are getting biomarkers ordered. That is unnecessary.

So there is a lot of variation out there. There's a lot of disparity out there. And this is why I said before, we have to make sure we're using all the free information that we have first before we spend additional money on additional tests.

And do we really need to order three tests in a given cancer? Usually not. And truthfully, we do not know and will not know anytime soon what to do when we have three tests, and they don't agree with each other. Which is why there is still a lot of art in the medicine of prostate cancer management and will be for, I think, a long time to come. All right. We have time for maybe one or two more questions. I think, Stan, you're next.

Stan Rosenfeld: Really quick. Availability of Artera internationally. Because we do have some patients that join us from Europe, and there are some patients from Australia, et cetera.

Jonathan Tward: I don't really work for Artera, so I don't know the international—I really have no idea, frankly. So I don't know if the other people on the call know about its global availability.

Matthew Cooperberg: I don't know either. I know they are working on it. And I know that, in principle, it will be easier to launch this test internationally than, say, Decipher or Prolaris because you don't need to actually ship tissue specimens across an international border, especially in this day and age.

Who knows what the tariff rates will look like on transmitting a JPEG across international lines? But in principle, I know they are interested, but I don't know the current availability or what that time frame looks like.

All right. We are at the top of the hour. I really want to thank our panelists and all the patient advocates once again for your participation in what I think has been another really fantastic conversation and hopefully another great example of bringing really cutting-edge research right to the patient community and helping all of our patients across the country and around the world make better-informed decisions about what to do with prostate cancer moving forward. Thank you all for joining us.

Jonathan Tward: Thank you.

Ashley Ross: Thanks so much. Thanks