Read the Full Video Transcript

Andrea Miyahira: Hello, everyone. I'm Andrea Miyahira at the Prostate Cancer Foundation. Today, I'm joined by Dr. Alex Wyatt, an associate professor at the University of British Columbia. Dr. Wyatt's team recently published the paper, "Multiregion sampling of de novo metastatic prostate cancer reveals complex polyclonality and augments clinical genotyping," in Nature Cancer. Dr. Wyatt, thank you so much, and I look forward to your presentation and our discussion.

Alexander Wyatt: Thank you, Dr. Miyahira, and thank you to UroToday for inviting me to speak today. Yes, thank you again, UroToday, for hosting and allowing me to explain some of these recent findings to you. The paper falls under the theme of how we genotype de novo metastatic prostate cancer. As Dr. Miyahira mentioned, this was really based around a report that was published in Nature Cancer in early January 2024, a study that we collaborated with Piet Ost in Belgium on. There are three co-first authors, Kim Van der Eecken, who's a pathologist and clinician in training with Piet at the time, and two bioinformatics PhD students in my group, Evan and Andy.

And of course, like many studies from my group, this has been generously supported by the Prostate Cancer Foundation. We've been fortunate to receive over the years some challenge awards and also several authors on this paper, including Dr. Edmond Kwan and Dr. Nicolette Fonseca, have been supported through Young Investigator Awards over the years, so thank you very much to the PCF.

This manuscript refocused on what we call de novo metastatic prostate cancer, which is treatment-naive at the time it first presents. This is really a disease that, when it is very first diagnosed, it's already spread beyond the prostate. These individuals essentially walk into the clinic with metastatic disease. They don't go through that typical paradigm of presenting with localized disease and maybe recurring later on.

I think that these patients that present with this de novo metastatic disease comprise perhaps only five to ten percent of all diagnoses, but they have been chronically understudied throughout the years, especially when we compare to localized disease where we have often the entire prostate resected to be able to study, and castration-resistant disease where PCF and other organizations have helped support large metastatic biopsy programs. In both of those settings, we have a really good understanding of the molecular features of the disease, and any associations between DNA and RNA, and protein changes, and disease aggression and even treatment outcomes.

But I would say for these patients that present with metastatic disease upfront, we really don't understand the complexity of their disease or the way that the disease is spread through the body, the difference between metastasis and the primary site, and that's mostly because these patients don't provide prostatectomy specimens. It's not standard of care in this population. And typically, they only will have a prostate biopsy, maybe a few cores taken from their prostate at first diagnosis.

I think it's important to recognize that we now have clinical guidelines around the world that say we should be genotyping these individuals, we should be looking at the somatic mutations that comprise their cancer because they're potentially important for the way we treat and manage these patients. So we ask the question, "Well, how do you genotype a patient with treatment-naive metastatic disease when all you may have is a sample from the prostate?"

We teamed up with Piet Ost in Belgium to study a really unique cohort of 43 patients who presented with distant metastatic disease, so hormone-sensitive untreated disease, but still underwent prostatectomy in a couple of clinical trials. And so, we had the whole prostate and multiple lymph nodes from these patients. What we did is a comprehensive pathological review of all of these regions, and we ended up sampling about 600 different tissue regions from across these 43 patients. We subjected that to a combination of targeted and whole exome sequencing to understand mutations, copy number changes, structural rearrangements, and generally those genomic features across each region.

As you may appreciate, this is actually a very aggressive disease from a genomic perspective. When you can aggregate together 20 or 30 samples per patient, you really start to get a picture of that. We saw that tumor suppressor loss was really common. DNA repair defects are common. And actually, this is a disease that sort of resembles castration-resistant prostate cancer, so what we would see in the really late stage, except that the androgen receptor is not yet altered because we haven't yet begun to target the androgen receptor.

Even in this small cohort, some of these molecular features actually associate with clinical outcomes. I'm highlighting TP53 here, which is consistent with work from larger studies of individual prostate biopsy specimens, which do suggest that genomic features of this disease may be able to help clarify clinical outcomes in this population. I really think there's potential here for the future.

Given that we have this comprehensive map of the prostate, and the lymph nodes, and we have the circulating tumor DNA at the same time, we asked the simple question, "How would our impression of each person's disease change if we used just one prostate biopsy core?" Which is what you typically do in clinical practice.

Very interestingly, what we saw is that if you just rely on a single biopsy core, you may really underestimate the genomic alterations that are present in that person's metastatic disease. That's highlighted on the right-hand side, that for something like PTEN loss, if you took a single biopsy core, you may find PTEN loss in about 20% of patients. But if you have multiple biopsy cores and you're able to aggregate that information, you can actually find it in upwards of almost 40% of individuals. That's telling us that there's potentially heterogeneity in your ability to detect somatic alterations across the prostate and lymph nodes.

This was driven by what we call intra-patient genomic heterogeneity. It's a combination of biological and technical differences between the different regions of these people's prostates and lymph nodes. At the mutational level, each different region, while they typically share mutations, they also have differences. There were, in fact, seven patients that had independent cancers within their prostate, so multiple unrelated tumors, only one of which would make it out into the metastatic environment.

This heterogeneity was mirrored at the copy number level and the whole genome duplication status as well. It was particularly noticeable for tumor suppressor genes, PTEN, RB1, in particular. That heterogeneity was mostly isolated to the prostate, although that was not always the case, and the metastatic lymph nodes were more homogeneous than the prostate. It tells us, I think, that the prostate is potentially a pool of diverse clones.

And that's indeed what we saw when we started to create family trees for each person's prostate, trying to map how the prostate cancer initially developed. The branches represent each population coming off the main trunk. What we typically saw was those branches were very dense within the prostate, and maybe only one branch would reach out into the metastatic niche and colonize the lymph nodes and the distant regions. So, that tells us that maybe the prostate is actually, as I mentioned before, a pool of diverse clones, not all of which are present in the metastatic niche.

But interestingly, in about one-fifth of patients, multiple primary site populations had independently colonized the metastatic niche. So, that tells us that the primary tumor actually probably continues to evolve and colonize the local metastatic niche at least, even after that initial seeding event. There may be waves of metastasis that can happen over time. This is particularly interesting in light of data that we've seen from large phase three trials, where ablating the primary site through radiation has a benefit, an overall survival benefit, in individuals with low volume metastatic disease. This is speculation, but potentially that effect might be partly due to the fact that you are removing a pool of diverse clones that hasn't yet spread to the metastatic site. So, you're hamstringing the cancer by getting rid of that pool of diversity.

Finally, we asked the question, "Well, how could this primary site heterogeneity impair precision oncology? How can it change or negatively influence the way that we genotype and treat patients on the basis of those results?" This slide here illustrates that if you have very homogeneous disease, it doesn't really matter where you put your needle for biopsy; you're going to get the same result. But what we actually saw is very typically the image on the right-hand side, which is that there are multiple populations within the prostate. Just biopsying in a single region, you might not actually get what's representative of the metastatic disease.

We saw this was actually the case in our patients. If you just studied a single prostate biopsy core, you're very likely to miss events in, for example, tumor suppressive genes, TP53, PTEN, and RB1. But recognizing that obviously in a clinical workflow, we cannot expect everybody to be independently sequencing all these different prostate biopsy regions. We did actually test a strategy where you could combine together multiple biopsies from a patient prior to sequencing. So it comes out about the same cost as sequencing a single biopsy region, but you get a much better aggregate of the patient's tumor.

There are some negatives to this, of course, because you lose spatial information, and the average is probably slightly worse than the best sample from that person's prostate, but you don't know upfront which is the best or most representative prostate. We think this is a strategy moving forward for how we might be able to incorporate the findings that we report into clinical practice.

With that, I'll just thank the funding bodies, including, of course, the PCF and Piet Ost, who's been my partner in all this work over the past few years. Thank you very much.

Andrea Miyahira: Well, thank you for that wonderful presentation, Dr. Wyatt. A few questions. First, did you evaluate whether germline alterations contributed to tumor evolution?

Alexander Wyatt: Yeah, it's really interesting because germline, certainly pathogenic mutations in DNA repair genes are fairly prevalent in this population. We did see several individuals with BRCA2, mismatch repair defects, and even an individual with germline CDK12 mutations. And so, they, of course, have very different tumors to those that don't have germline mutations in those genes. Typically, they're more complex, so they have higher heterogeneity, which makes sense because there's greater genome instability. The individual with a germline CDK12 mutation actually had two different CDK12-driven cancers in his prostate, and he was only 44 when he was diagnosed with advanced prostate cancer, so it tells you about the aggression of those cancers when they arise. And certainly, they influence the evolutionary trajectory of the tumor. I think absolutely they need to be taken into account at the same time as the somatic alterations, which is why we recommend both germline and somatic screening of these patients.

Andrea Miyahira: Thank you. You were able to do this study because you had access to a really unique clinical trial. Will you be able to get longitudinal samples from these patients and see how the genomics and the clones change with treatment?

Alexander Wyatt: Yeah, I think that will be a really fascinating thing to look at because what we never had before, was this impression of what the treatment naive complexity was. And we hypothesize that when we use intensive systemic treatment, you're putting a bottleneck on the cancer. And you probably do a better job of suppressing some populations than others, but we don't know which ones they are.

So, you're absolutely right. In already half of the cohort, we've been collecting samples at progression to castration-resistant prostate cancer, and I know that Piet has some nice ideas and plans for even metastatic biopsy or even autopsy in those patients that may consent to it, so that we can really understand how the disease did evolve. We know that it's likely there's going to be some androgen receptor gene alterations, but I think what will really be interesting, is which populations didn't go through to CRPC because that will then tell us about maybe molecular features of good outcomes in the initial diagnosis stage.

Andrea Miyahira: Okay. Well, I look forward to that. You weren't able to get ctDNA from a number of the patients, but in those that you were able to get ctDNA, did it more resemble primary or metastatic clones?

Alexander Wyatt: Yeah, so obviously ctDNA has been really pioneered in the castration-resistant setting, when we know that the levels are typically quite high, although not always. And it remains a question I think, how useful ctDNA will be in these patients that present with de novo metastatic prostate cancer. Partly because in clinical practice, what you do is immediately start hormone therapy. You really want to treat that symptomatic metastasis as quickly as possible, and that likely reduces the amount of ctDNA in patients. There's probably only a short window where you can get ctDNA, but what we did appreciate in those patients where that was possible, was that it is typically more representative of the dominant metastatic genotype, rather than those primary restricted populations.

We certainly believe that in the individuals that have ctDNA present at diagnosis, and obviously the more sensitive the tests become over the years, the more that number's going to grow, that's going to be an alternative for genotyping metastasis. Yeah, watch this space because we have some studies in our lab which are trying to address that question at the moment.

Andrea Miyahira: Thank you. And you did touch on this already in your presentation, but going forward, what is your recommendation for how we should best use somatic genomics as biomarkers in trials or for treatment decisions? And as a follow-up question, do your findings call into question other genomic biomarker studies?

Alexander Wyatt: Yeah, absolutely. The complexity within the primary site really I think does present challenges for relying on one needle biopsy result. We hypothesize that this may explain some of the discrepancies in the previous biomarker-driven trials. I think PTEN is an obvious example to focus on that. I think it would be quite easy to have both false positives and false negatives for PTEN status, in terms of how metastatic disease states. I think that could potentially have influenced prior trials that intervened on the basis of a PTEN deletion or wild-type status.

We think that going forward, of course, this is still a retrospective study that we are presenting, and even though it's of a very rare cohort, I think probably what's best, is that we incorporate these ideas into future trial designs. We need to be able to study multiple biopsy regions from patients that are being treated with biomarker-driven therapy or considered for de-intensification or intensification. I think that ultimately will tell us what to do.

One way we are trying to explore this ourselves, is more comprehensively across hundreds of patients, look at the difference between the result that we'd have from a single biopsy core versus say a pool of multiple. I think that's really going to help clarify. I do believe that some of these genomic features will be relevant, not just for the biomarker-driven therapies, but also helping us understand who is best suited for, let's say, ADT monotherapy or just double it, rather than the triple treatment that is standard of care at the moment.

Andrea Miyahira: Okay. Well, thank you so much for joining us today and sharing this really informative study. I look forward to the next one.

Alexander Wyatt: Thank you. That was my pleasure.