Elena Castro: Our next speaker is Joaquin Mateo, who is the head of the Prostate Cancer Translational Research Group at the Byron Institute of Oncology. He will discuss genomic and clinical markers in non-metastatic CRPC.

Joaquin Mateo: Thank you, Elena. Good morning, everyone. Good morning also to those of you who are connected online, following us remotely. This is my first time here at this meeting. I'm really enjoying it, so I really wanted to thank you for inviting me and congratulate you for creating such an engaging space for discussion, particularly this year, where we were able to meet again.

So, I'm going to be discussing the possibility of using molecular markers to improve the management of non-metastatic prostate cancer. If someone is in a rush and has to leave, I'm going to give you a summary, okay? There is no data suggesting that nowadays we can use molecular profiling to improve the outcome of patients or improve the management of non-metastatic CRPC patients. But if you can stay, we are going to go together to identify areas where we can probably improve this.

So, these are my disclosures. In this session, we have been discussing non-metastatic CRPC, but this is actually a snapshot of the disease taken at one moment of the continuum. So, I would argue that if we want to understand how to guide decisions there, based on biomarkers, we need to understand the evolution of these biomarkers throughout the course of the disease. Clearly, patients with prostate cancers, these tumors can take different routes. Prognostic biomarkers are those that inform us of which route one individual tumor is going to take and how fast it is going to advance through that pathway. Predictive biomarkers are those that can help us make decisions about treatments that can change the course that the disease is taking.

There are some junctions in these courses of the disease where we classify the disease based on different bins, M0, M1, [inaudible], CRPC; we were discussing it yesterday. Some of these junctions, like the M0 versus M1 CRPC, are a reflection of different biologies. As Dr. Morris was saying, also the interaction between the biology and the impact of the treatment, but in fairness, are not necessarily neat bins, but are a range of possibilities where we classify the patient into one or another bin based on the sensitivity of a given assay, as we have been discussing yesterday and today with regards to imaging.

So, this is a simplified schema of how we manage non-metastatic CRPC today. We basically base our decisions, beyond of course the patient's overall status, on imaging biomarkers and PSA kinetics. Imaging helps us put the patient into one of the bins, M0 versus M1. And we have been listening to how this can change depending on the imaging technique that we use. PSA kinetics help us select those patients that may merit intensification of treatment with an AR inhibitor, and then both PSA and imaging are used to monitor that patient along the treatment and decide when it's the optimal time to stop and change therapies.

So, how can biomarkers help us at these different time points of the decision-making? And of course, there is lots of work going on in this side of the slide, and I'm not going to be covering this today, but basically, when it comes to classifying patients in M0 versus M1, we can use biomarkers to more precisely classify patients within that range of opportunities that are now simplified into two bins.

When it comes to deciding who needs treatment, we can use biomarkers to improve the performance of PSA kinetics, to select who needs intensified treatment, who can maybe spare treatment and toxicities, but also maybe to identify some patients that may merit a completely different approach to therapy at this stage. And obviously, here is a space where we can also do better, and we can use biomarkers to improve how we monitor these patients and make the decision to switch therapies at a given time. So, running studies in non-metastatic CRPC is very hard. As [inaudible] said, there is a gap of knowledge in the molecular encyclopedia of prostate cancer. And this is in part because it's very hard to run studies when there is no contemporaneous tissue available for the disease state that you want to study. However, the three pivotal phase three trials represent a very good platform to run correlative studies that can help us guide these decisions better.

And here, Dr. Fang and Dr. Esmol and colleagues did a very interesting effort to pursue RNA profiling based on a microarray assay to subclassify patients receiving Apalutamide or placebo in the SPARTAN trial. Using the Decipher assay, they demonstrated that the genomic classifier was prognostic. They also showed prognostic value for the PAM50-based basal versus luminal classification, but even if the magnitude of benefit may be slightly different between the subgroups, at the end of the day, they all benefited from adding Apalutamide. Hence, I would argue that it would not help us to decide who to treat; it has prognostic value, but not predictive value. And something to highlight in this study, not in this study but in all of the studies in non-metastatic CRPC, is that because of this lack of contemporaneous tissue availability, the samples were taken around seven to eight years before the patient was treated.

And in between that time, the patient became resistant to ADT. So, we are studying the outcome of patients with castration-resistant prostate cancer based on samples taken years before. And that has some limitations. So, if we want to ask, could genomic profiling better stratify patients with non-metastatic CRPC? It's very difficult to answer because we don't know what is the genomic landscape of non-metastatic CRPC. We know, or we know a little about the genomic landscape of these patients when they were hormone-sensitive back in the day.

And that's relevant because castration resistance is the reflection of a heavy selective pressure on the tumor. Prostate cancer changes dramatically from hormone-sensitive to castration-resistant at a molecular level. There are many studies showing this; this is just one that we presented at GU ASCO, where we analyzed the genomic landscape of over a thousand samples of men with metastatic prostate cancer, by means of whether the biopsy was taken before or after different treatments. And what we saw, and many others have shown this, is that patients that develop castration resistance, those biopsies are enriched for TP53 mutations, RB1 loss, PTEN loss, MYC amplifications. All those markers have prognostic value, both in hormone-sensitive prostate cancer and in CRPC. So, if we are studying the potential value of these biomarkers in non-metastatic CRPC, we cannot be using information taken before these biomarkers may have changed over time.

So, to answer this question, we need to understand better what is the genomic makeup of non-metastatic CRPC, both at that time but also importantly, we need to understand what molecular subgroups of prostate cancer will evolve towards the path of becoming non-metastatic CRPC and how they are different from those patients that are going to develop bulky metastatic CRPC. And identifying them upfront could be a fair step towards better managing these patients.

So, okay, we don't have access to contemporaneous tissue; could liquid biopsy help in improving the management of these patients? So, theoretically, liquid biopsy could help in refining the classification within this range and could complement imaging. Imaging is not the only way to measure things. Liquid biopsy clearly could help in or complement PSA kinetics in selecting who needs treatment and who deserves more intense treatment. And definitely, liquid biopsy has shown a role to monitor patients along the treatment course; CTCs and cell-free DNA have shown to be prognostic and response-resistant biomarkers in the castration-resistant setting.

The problem is that the yield of CTCs and cell-free DNA normally correlates with the amount of disease, right? So, for example, running liquid biopsy studies in localized prostate cancer is very hard, at least with the sensitivity of the assays that we are using nowadays in clinical practice; probably will get better. And it may be the same case for non-metastatic CRPC. So, Gert Attard and Matthew Smith and others did this interesting exercise that was presented at ASCO a couple of years ago, of looking how often patients with non-metastatic CRPC have sufficient ctDNA in the blood to run studies. And two interesting points, and I'm not going to enter into what is detectable, what is the yield threshold; that's not the topic for today. They compared it with the TITAN trial, also using Apalutamide in the metastatic hormone-sensitive setting, and they used the same methodology; that's what we need to understand today.

So, first message: only 7.5% of the patients starting treatment for non-metastatic CRPC had detectable ctDNA. So, it's very unlikely that we can use qualitative information on these samples if only 7% of patients would have material. But this 7% of patients did much worse, regardless of whether they were on Apalutamide or placebo. So, it has prognostic value; maybe we can use it to understand who needs treatment. Of course, because these trials only treated patients with fast PSA doubling time, we don't know how this applies to the other segment of the population and whether we can rescue some patients for treatment, despite the PSA doubling time being over 10 months. Second observation that I find very interesting: at progression, so when becoming metastatic CRPC, 27% of patients had detectable ctDNA, compared to 63% of patients that had it in the TITAN trial, also becoming metastatic CRPC. Again, showing that there are different biologies here.

Second conclusion: therapeutic indications for AR inhibitors in non-metastatic CRPC are nowadays based on prognostic markers, and there is no data on predictive biomarkers for tumors in this space. We can use functional imaging to redefine the disease state. Right now, we don't have sufficient information to use genomic biomarkers, and we need to better understand the biology differences between these tumors. Liquid biopsy could provide additional opportunities, but we need to study it better. And at the risk of getting the other bell, just one last message: yesterday, today, and at all the conferences, we keep hearing about the performance of single biomarkers. Thank you. But we need to focus on the disease and integrating the different biomarkers. Thank you very much for your attention.