Oliver Sartor: Hi. I'm Dr. Oliver Sartor. And I'm here with UroToday. And we have a very wonderful guest, one that I always enjoy speaking with. And that's Alex Wyatt. Alex is an associate professor at the Vancouver Prostate Cancer Centre. But oh, so much more. I think he's taught me more about circulating tumor DNA than anybody on the planet. So welcome, Alex.

Alexander Wyatt: Thank you, Oliver. And thank you, UroToday, for having me. A pleasure to be here.

Oliver Sartor: You recently published at the end of May a manuscript that I thought had great consequence. And I'd love to talk it over with you, because this is the TheraP trial. And we'll, of course, explain what that is. And the circulating tumor DNA biomarkers and being able to compare the two randomized arms, which are the cabazitaxel and the lutetium-177 PSMA-617. So I'll tell you what. I'll set the stage very quickly for the protocol and then we'll go forward.

So this protocol came out of Australia. It was a very important phase two randomized trial, large phase two that included a variety of endpoints with PSA being a primary endpoint, but also some PFS and OS endpoints. And comparing heavily pre-treated patients who have had an ARPI and docetaxel, comparing cabazitaxel to lutetium-177 PSMA-617. And so that sets the framework. These heavily pre-treated patients. So Alex, tell us a little bit about what you found with the ctDNA. And I'll interrupt you now and then just to understand or clarify.

Alexander Wyatt: Yeah. Absolutely. So I think an extra piece of context around the TheraP trial is the imaging selection, which I think is quite important, and especially when we think about the generalizability of our findings as well. So as you know, if you had discordant imaging results for FDG and PSMA PET, namely that if you had FDG-positive lesions that weren't PSMA positive, you actually weren't eligible to enroll. There are some patients that essentially we're not capturing, even though this is third-line aggressive disease. There's probably some aggressive disease that actually isn't in our cohort. And that's kind of an important clarification there.

So I mean, there are multiple aspects of the results here. I think one of the headlines was that if you had low levels of ctDNA in your blood, so the proportion of your total cell-free DNA that's tumor derived, if you had low amounts, you seem to benefit considerably more from the lutetium PSMA arm than you did on the cabazitaxel chemotherapy arm. And that was true for PSA50. So biochemical response, which was the primary endpoint of the trial. But it was also true for progression-free survival, but not for overall survival. So for overall survival, we saw the typical prognostic relationship that we normally see with ctDNA fraction.

And actually this relationship was true in multivariable analyses. So even after accounting for PSMA SUV mean, which is a kind of prognostic predictive factor in this setting. And also it seemed when we added it together with PSMA SUV mean, you got even better stratification. So I think there's potentially very high value for ctDNA fraction.

Oliver Sartor: Yeah. So let's talk a little bit about this FDG and PSMA, because the TheraP trial incorporated both imaging modalities. And as you pointed out, in order to be eligible, you could not have discordant lesions because that led to an exclusion when you had discordance. So we have two sets of PET scans, which is atypical from the American perspective. But you found a really interesting relationship between FDG and ctDNA and PSMA PET and the ctDNA. And I wonder if you might talk about these two different PET scans and how they relate to the circulating tumor DNA in distinct ways.

Alexander Wyatt: Yeah. I also found this fascinating. We've never had the chance to look at ctDNA fraction in the context of imaging like this before. And we've always thought, well, OK, you have high ctDNA fraction. Your tumor must be really proliferative and aggressive. But we've never been able to have this synchronous measurement at the same time. And so I think what you're referring to is partly that if you had high ctDNA fraction, we saw high FDG metabolic tumor volume, which we know is a prognostic factor in and of its own right. So those two things correlated together.

But fascinatingly, then when we looked at PSMA PET, actually, when we looked at the SUV mean of PSMA PET, we saw the inverse relationship with those patients with high ctDNA fraction who had lower PSMA SUV mean. And those patients with low ctDNA fraction had a higher PSMA SUV mean. So I think it speaks to some biological characteristics associated with ctDNA fraction.

Oliver Sartor: When I saw that data, I said, wow. The intuitive nature of the FDG going up when the ctDNA was going up, that was my expectation. But the PSMA SUV mean with the circulating tumor DNA in essence going down when the SUV mean for PSMA went up, that was like, whoa, that's really intriguing. So thank you for covering that. And thank you for-- nobody's ever made that relationship before. I thought that was a really interesting finding from the trial.

Now, you mentioned a little bit about the discordance between the PSA, the PFS, and the OS. And of course, let's talk a little more in depth about the OS because it's really, I think, where the rubber meets the road. We love PSA. We love PFS, but we love OS even more. So let's talk about the differences in the signals that you picked up between the first two and the OS, the PSA and the PFS versus the OS. So let's talk about that in a little more detail.

Alexander Wyatt: So I mean I think there is-- obviously, the independent signal starts to be diluted when you look at OS. You don't see quite the same strength of relationship that we do when we look at the progression endpoints. But I think there are still some trends there actually. And I wonder about whether a larger trial might uncover something.

I think the other thing we have to think about, especially when we're looking in a third-line setting, and perhaps we'll come to this if we can talk about some of the genomic alterations, is how important immortal time bias and context dependency is here. That when we look really early in disease, some of the most powerfully prognostic factors may not actually be as relevant when you look a little bit later. And that's partly because unfortunately, those patients with the very worst disease don't live to benefit from treatments in the third-line setting.

So that's a factor I'm always thinking of here, is who is not represented in our population? But I think ultimately what the response data point to is that in settings where you have multiple choices, multiple life-prolonging choices available, then even if there isn't an OS signal, then I think this is helpful information to pick the ideal treatment for each patient. But maybe you can tell me, what do you think about the OS data?

Oliver Sartor: Well, first of all, I think it's prognostic in an important way. And part of the problem is-- and I might have worded it slightly differently, but I'm not sure if I'm using different concepts. Maybe just different words --is when you reach these very high volume patients, the probability of response goes way down. And cabazitaxel actually over-performs in this setting. And the Pluvicto under-performs in this setting. And it was fascinating to me that the cabazitaxel seemed to perform almost equally despite the distinction in the baseline ctDNA. But that was not true for the Pluvicto. And this interaction between the biases move the curves together in a way that doesn't discriminate. So I don't know if I'm using different words or different concepts, but I think it all points in the same direction.

Alexander Wyatt: Yeah. You're absolutely right. I agree with that assessment. I think initially, we were actually surprised not to see more stratification in the chemotherapy arm. But I think we've seen other data, certainly, internally that you don't get as big of a split for response in terms of chemotherapy. And I think you're absolutely right that those very aggressive, the worst disease is not getting a long benefit from lutetium.

Oliver Sartor: Yes. Unfortunately, it's something that we're learning. And by the way, Louise Emmett has presented some data that would corroborate your findings. You were first, but she's finding additional information that would in my mind tie together to make part of a larger story.

Alexander Wyatt: Yeah. And actually, Louise's study is going to be fascinating because she did also do dual PETs. But she didn't then select on that basis. So we will actually have those patients included in that population. So that's going to be really interesting.

Oliver Sartor: Yeah. And then she also has a combination therapy. When you add enzalutamide in addition to the lutetium, it changes matters in a very provocative way. She presented at ASCO GU. I don't know if you saw that presentation or not. Yeah. It was a really fascinating relationship with the SUV mean and the interaction with the monotherapy and the dual therapy.

And enzalutamide, as it turns out, patients respond better with a low SUV mean as compared to the high SUV mean with PSMA lutetium. So it's a fascinating interplay between the hormones and the lutetium. Anyway, cool stuff. Let's get down to something you alluded to earlier about the individual genes. We don't have time enough to do a giant drill down. But which ones do you think are most important? And maybe I'll get you just to talk about PTEN in particular.

Alexander Wyatt: Yeah. So I mean, I think I'll preface it by saying that over the years, we've recognized that the best way to hone in on ctDNA and individual genes is to do it by first controlling for how much ctDNA you have and then looking at exactly how many alleles are disrupted, how many are remaining. And I think this is really important in prostate cancer, particularly for genes like PTEN. Maybe it's relevant if you still have one intact copy of PTEN or you have both of your parental alleles disrupted.

So in this paper, we tried to use a framework that allowed us to both adjust for ctDNA fraction and count the remaining functional alleles of a gene. I think PTEN was what really stood out to us for two reasons. Firstly, because there was actually a fascinating relationship with imaging for PTEN. It was modest. But we saw that increased FDG metabolic tumor volume if you had PTEN loss, which I think fits with the idea that PI3K pathway upregulation leads to increased glycolysis. An interesting finding there.

But what we saw is that actually-- so if you have PTEN loss, and it didn't matter how you cut it, what type of alteration, you did worse on cabazitaxel. And it was mostly because of a relationship with poor outcomes in cabazitaxel rather than better outcomes in lutetium PSMA. And I think this was the case for the progression endpoints, but also for overall survival. So it stood out even for the OS endpoints, which was particularly interesting.

And then on the flip side, we saw that there were some patients with ATM mutations that did pretty well on lutetium PSMA. And there has been some precedent for that in breast cancer, where ATM mutations are thought to sensitize to radiation. And then I think the last thing to mention is that it was actually interesting to see absence of relationship for things like TP53, AR, BRCA2. Some of these alterations earlier in disease, we would expect to see prognostic effects, at the very least.

But no evidence of any interaction between the arms there. Although, I think in OS, TP53 remained prognostic. But they didn't stratify differently. And so that would be totally different if you looked in, say, first-line ARPI context. You'd expect to see huge splits in the context of those genes. So it really points to how we've talked about how important context is by line of treatment. But I think genomic alterations, we can't take a finding from one disease setting and extrapolate it to everything. It's so context-dependent.

Oliver Sartor: Yeah. Alex, we're going to need to finish up here in just a moment. Any final comments you'd like to make before we finish up?

Alexander Wyatt: I think just to-- we're really excited about these findings, but I think they have several limitations. It's a phase two trial, image-selected. We need to look in other populations. We need to look in other lines of treatment. And PSMA radioligand therapy is here to stay. So I think we're going to have lots of opportunities over the next few years. So that includes multiple trials that you're involved with that are integrating ctDNA biomarkers, but also some phase IIs around the world, both in Australia and here in Canada.

And the other layer of information that we're going to be able to add is the epigenomic aspects. So we don't see huge remodeling of the genome in the context of acquired resistance to lutetium PSMA. So that points to more epigenomic factors being relevant, how are those cells adapting to that radiation. Are they changing neuroendocrine expression? Are they upregulating features associated with radioresistance? I think that's going to be the next frontier of biomarker analysis.

Oliver Sartor: Well, if you crack the radioresistance code, please call me first. I'd love to know. Alex, thank you so much for being here today. I personally always enjoy speaking with you, and I thank you for teaching us, the field, so much about circulating tumor DNA, the interactions with the therapies, the interaction with the patients, interactions with the scans. You're a master of what you do. Thank you very much.

Alexander Wyatt: Thank you so much, Oliver.