Read the Full Video Transcript

Zach Klaassen: Hi, my name is Zach Klaassen. I'm a urologic oncologist at the Georgia Cancer Center in Augusta, Georgia. We are live at the AUA 2024 in San Antonio, Texas. And I'm delighted to be joined by Dr. Todd Morgan from the University of Michigan, Chief of Urologic Oncology. Thanks so much for joining us, Todd.

Todd Morgan: Hey, Zach. Thanks a ton for having me.

Zach Klaassen: Always good to have you. So we're going to talk today about new data you presented at this meeting looking at the prostate AI test and looking specifically at new data post-prostatectomy biochemical recurrence. So maybe just highlight for us why there's a need for prognostic biomarkers in this disease space.

Todd Morgan: Sure. So let's talk about the disease space. As you know, patients recur after prostatectomy at a pretty substantial rate, honestly, especially with the shift towards operating on patients with high-risk disease. It's somewhere around 20 to 40% of patients, again, depending on disease risk, who ultimately have a biochemical recurrence. Put in a little plug for our new AUA/ASTRO/SUO salvage therapy guidelines. So we have a new guideline in this space. It's a really important space. We see a lot of patients, and there's a need to better understand who has more aggressive disease, who has less aggressive disease, and how should we treat those patients.

Zach Klaassen: And so we'll talk about the data, but I think as you mentioned, and it's in the new guideline as well, we're going to be integrating biomarkers, really good biomarkers with PSMA PET imaging. This whole disease space is really changing, isn't it?

Todd Morgan: It's changing like crazy. When we think about, there was a guideline in this space 10 years ago that is just dramatically different than the guideline that we put out and talked about yesterday at this meeting.

Zach Klaassen: Absolutely. So tell us about the model development validation and maybe walk us through some of the key results.

Todd Morgan: This is the first post-prostatectomy MMAI, multimodal artificial intelligence, model. Many people may know about the ArteraAI model for patients who've undergone biopsy. And so it's something that we can use for prognosis, say, in a newly diagnosed patient. But for this model, the development relied on two trials, RTOG 9601 and then the SPORT trial, both of which enrolled patients with biochemical recurrence who were undergoing salvage radiation therapy. And then they were randomized to undergo hormone treatment or not, undergo hormone treatment, more or less. And so we use those datasets, and what we use is both the clinical data and the digital pathology. So that means that the H&E slides were scanned in, interpreted by the MMAI model, and then an entire model was put together that took all this data and is prognostic. And so the cohort was developed, was separated into both a development cohort and then a validation cohort. And ultimately, it validated really well with really amazing performance characteristics.

Zach Klaassen: That's great. I think it just goes to show that we're taking biopsies... These are old trials. We're taking old biopsy tissue and making phenomenal artificial intelligence biomarkers out of it. It's kind of wild, isn't it?

Todd Morgan: It is. And we're a little bit used to this now, I think, because we have done this with genomics. There are genomic models with newly diagnosed patients. There are genomic models with patients who have undergone prostatectomy. And really, this is the first foray into this post-prostatectomy space with this digital pathology type platform.

Zach Klaassen: Absolutely.

Todd Morgan: And it works.

Zach Klaassen: Absolutely. And I think I want to delve into probably arguably the most important result that came out of that study, which was risk stratifying people for hormonal therapy at the time of salvage therapy. Maybe you can walk us through that and discuss some of the implications for how we manage our patients.

Todd Morgan: So we talk a lot about prognostic biomarkers. Prognostic biomarkers are ones that tell us whether somebody has higher risk or lower risk disease. Are they more likely to develop metastasis or less likely to develop metastasis? Of course, what we really want, no matter what biomarker setting we're talking about, are predictive markers, ones that actually tell us the given patient in front of us is going to benefit from a given treatment and then another patient is going to benefit from a different treatment. And so that was the goal as part of this study. Because we had randomized trial data, patients who underwent hormone treatment and patients who did not undergo hormone treatment, we were able to ask the question, can we develop a model that allows us to see which patients actually benefit from hormone therapy and which don't?

And what we saw is there's a real spread of basically response to hormone therapy or benefit from hormone therapy based on the ArteraAI model. So patients with low-risk AI disease did not seem to benefit from hormone therapy, and those with high-risk AI disease did seem to benefit. Now, I'll tell you that one of the statistical nuances as we look at an interaction term, and the interaction term was not statistically significant. But when we see these curves, on the one hand, the high-risk patients have a real significant difference in likelihood of progressing with hormone therapy or no hormone therapy. And for the lower-risk patients, these two curves are right on top of each other.

Zach Klaassen: It's very impressive.

Todd Morgan: So we see some compelling data. Is this something that's ready to be brought into the clinic? No, but this is ultimately the goal, and I think this type of platform can do that.

Zach Klaassen: Absolutely. I think, let's take a step back and just note, it's been a big couple of years for ArteraAI prostate tests. The NCCN guidelines have their own table just specifically looking at it. Maybe just walk our listeners through briefly some of the other spaces where we've seen this test be beneficial.

Todd Morgan: So it's really in that newly-diagnosed disease setting, and the NCCN guidelines do a great job. Congrats to all the folks, everybody who's on that guideline. It's a great guideline to be able to participate in that, in developing a table. And actually, the newest guidelines go into this space in real detail in terms of which biomarkers, which, for example, genomic tests. And it does include, at least as a row, germline testing, and then ArteraAI, spelling out what's the disease setting and what evidence is there to support one of these tests. And so the ArteraAI test for newly-diagnosed patients is in the NCCN guidelines for help with risk stratification. And it's a largely prognostic marker, but also there is, again, some really tantalizing compelling data, Dan Spratt presented this, suggesting that we can maybe use it to help guide the use of hormone therapy for patients undergoing radiation as their primary treatment for prostate cancer.

Zach Klaassen: And the guidelines even have examples of where you could use it. It's really the change from 2023 to 2024 was really quite vast in terms of how they're breaking it down. It's quite helpful.

Todd Morgan: Big shift, including the Simon criteria, which not everybody's familiar with, but is a way of trying to categorize the evidence level for a given biomarker. And the Simon criteria level was high for this Artera test.

Zach Klaassen: That's great. Last question, just looking to the future a little bit. Next 12 to 24 months, where do we see the ArteraAI prostate test going, whether in the biochemical recurrence setting or just in general?

Todd Morgan: My sense of the lay of the land is that in the newly-diagnosed patient, this platform is going to start having a real role. And it's been commercialized. It's available. We can easily imagine those patients that may benefit from additional risk stratification information. We're already using genomic tests in that setting. Of course, I'll plug a randomized control trial we're performing in that setting called G Major where patients are randomized to undergo genomic testing or not genomic testing in favorable risk, newly diagnosed disease. And we can imagine ArteraAI going right into that space and us really understanding over time how these types of tests move management. And so again, that test is now available. In the biochemical recurrence setting, this test that we're reporting on here at the AUA, it's not yet commercially available, but I suspect it will be commercialized at some point here.

Zach Klaassen: That's great. Great discussion, as always. Any take-home messages for our listeners today?

Todd Morgan: It's really amazing to be able to be involved in this type of work and to collaborate with a company like Artera. The ability to use digital pathology to understand disease risk is really incredible. And it opens up doors not just to prognosis, but also prediction and maybe even understanding other molecular aspects that go beyond what we can see with the actual eye. And so I am really excited to be able to work in this space.

Zach Klaassen: That's great. Thanks so much for your time, as always, Todd, and for your expertise.

Todd Morgan: Thanks, Zach.