Read the Full Video Transcript

Alicia Morgans: Hi. I'm so excited to be here today with Professor Matthew Smith, who is joining me from Massachusetts General Hospital in Boston, as well as joining me recently in Lugano, Switzerland for the APCCC 2024. Thank you so much for being here with me today, Matt.

Matthew Smith: Happy to be here.

Alicia Morgans: Wonderful. So you gave an overview, an introductory lecture in Lugano about our considerations of prognostic and predictive factors in choosing treatments for patients with metastatic hormone-sensitive disease. I wonder if you could share a little bit of that with us now and then we'll have a conversation when you're through.

Matthew Smith: Happy to do so. My topic at APCCC was to introduce metastatic hormone-sensitive prostate cancer, also known as castration-sensitive prostate cancer, focusing on relevant prognostic and predictive factors. And I think it is really an important issue as we think about optimal patient selection for intensification, and then also as we begin to consider choice of available therapies in this setting. So my aims were to describe the clinical imaging and molecular factors associated with survival in mCSPC and to summarize factors that have been associated with the treatment effect in mCSPC, recognizing that to date there are no validated predictive factors in this disease state. A couple of important caveats: I focused on mCSPC and overall survival, not other outcomes. And this presentation was intended as an introduction, not a comprehensive review. So don't be offended if I don't cover your preferred gene or profile or pathway.

So there are a number of clinical factors that predict outcome prognosis in patients with mCSPC. This is really nice data from the STAMPEDE control arm of patients with metastatic prostate cancer treated with the ADT alone. As we all know, these patients have a relatively poor outcome with a median overall survival of less than four years, and their survival is related to anatomic sites of disease. So patients with bone and soft tissue metastasis do worse than patients with soft tissue disease alone. So if you think back on this, you could say that really bone scans are an important prognostic factor in mCSPC.

We all know as well that overall survival is linked to disease volume and manner of clinical presentation. This is nice work from Chris Sweeney and colleagues showing outcomes by volume of disease at diagnosis with metastasis, and whether patients had de novo or recurrent disease outcomes vary widely from a median survival of only 43 months for patients with de novo high-volume disease going all the way up to a median of 92.4 months for patients with recurrent low-volume disease. And you can see the intermediate groups in the middle.

Genomic alterations have also been implicated in prognosis in mCSPC. This is data looking at a composite biomarker of P53, PTEN, and RB. Patients with a genomic alteration in one or more of those genes had worse survival than patients with no alterations, and their survival worsened if they had more than one alteration in this sort of three-gene panel. Really, I'd say, state-of-the-art work has now been reported by Gerhardt Attard and colleagues looking at this comprehensive profiling of patients who were enrolled in the STAMPEDE-Abiraterone trials.

Really, this figure is representative, an extraordinary amount of data looking at 57 different gene signatures. They sort of locked in the gene signature and then did the analyses looking at survival for patients with metastatic disease on the left and localized high-risk disease on the right. Decipher scores rose to the top as being most predictive of overall survival for patients with both localized disease and metastatic disease. Decipher also performed very well in patients with localized high-risk disease. This is a real tour de force. So look at this reference in the full public. This is sort of a pre-publication and the full manuscript will be coming out soon, but really extraordinary. And you see a number of other gene signatures that performed very well as prognostic markers in mCSPC, again including PTEN and P53.

We have a different story in thinking about predictive biomarkers to date. There are no validated predictive biomarkers in mCSPC, but there are a number of factors at least associated with differential treatment effects. This is a very nice meta-analysis, reported by Riaz in JAMA Oncology, looking at forest plots for ARPIs on the top and docetaxel on the bottom according to volume of disease at presentation. What you see is that there's a very consistent effect for ARPIs, whether the patients have high-volume or low-volume with hazard ratios of 0.66 versus 0.58 respectively.

What a different story with docetaxel, where most of the benefit appears to be in patients with high-volume disease with a hazard ratio of 0.73, dropping off to 0.91 in low-volume disease. And this sort of supports the original CHARTED observation that most or all of the benefit resides in patients with high-volume disease. So while this doesn't rise to the level of being a statistically significant validated predictive biomarker, it does support the idea that patients with high-volume disease benefit most from docetaxel, whereas the benefits are more consistent for high and low-volume disease for ARPIs.

And in that same study, a similar story sort of emerged by manner of presentation with de novo versus recurrent disease. A number of genetic markers have been evaluated, genetic profiles have been evaluated as potential predictive biomarkers. This study by Hamid AA et al. shows a separation according to PAM50 profiles, with patients with luminal B genotype having a better treatment effect with the addition of docetaxel to ADT compared to PAM50 basal where there's really, you see almost complete overlap of the docetaxel and ADT alone curves. Decipher scores have also been evaluated in this context to evaluate the benefit of docetaxel in mCSPC. And you can see that more of the benefit with docetaxel, a bigger treatment effect, is seen in patients with higher Decipher scores.

Data seen that's more compelling to date is data from Abiraterone and STAMPEDE. This is the same study by Gerhardt Attard and colleagues looking at differences in overall survival by addition of abiraterone to ADT by Decipher scores and graphed over time. Really seeing that much of the benefit is in patients with higher Decipher scores shown in red, with much less benefit in patients with Decipher scores below the median. So in summary, there are a number of important prognostic factors in mCSPC including disease presentation, anatomic pattern of metastasis, and disease volume. Certain single gene alterations, including PTEN and P53, are also associated with shorter survival, as are higher Decipher scores. To date, while there's some early interesting work, there are no validated statistically significant predictive biomarkers in this disease state. So that remains an area of important unmet need. Thank you.

Alicia Morgans: Matthew, thank you so much. That was a wonderful tour de force and really just a great way for all of us to be updated on some of the newest work as we try to understand prognostic biomarkers at least, and prognostic factors as we treat these patients with metastatic hormone-sensitive disease. I wonder, when you go into clinic, what are the things, if you had to name the top few, that you're thinking about right away when you're making decisions in clinic with patients when they come in with mHSPC, biomarkers aside, fancy stuff aside?

Matthew Smith: Yeah, absolutely. It's a great question and your question's spot on because it highlights the nuances of clinical practice versus just guideline-based medicine. What I try to think about for the patient in front of me is what is this man's cancer-specific survival? And to consider that, we have to think about the disease characteristics that I reviewed some of here, and also their age and comorbidities. So that's so important. In presenting these prognostic markers, it highlights the fact that there are a variety of ways we can think about that: presentation, disease volume, all that, but it's not that simple. And then we also have to consider the age and health of the patient and his chance of dying of prostate cancer versus something else. And that's going to inform the decision about whether or not to intensify. The default should be to intensify, and you're going to find some exceptions where that may not be necessary or appropriate. And then the secondary decision is how best to intensify.

Alicia Morgans: Yeah, I think that's exactly how I think about it. Really, there are all of these factors that come into play and the disease characteristics are really just one of them. And sometimes the ones that drive us most are those associated with the patient. But getting back to your talk, and really it was such a nice review of some of the data that is up and coming and more and more is evolving and coming out each day. Where are you most excited to see blood-based or tissue-based biomarker development go? What do you think is going to be our next potential predictive biomarker? And obviously this is a guess and there's no money on the table, you're not actually placing a bet, but I would love to hear your thoughts on what might fly.

Matthew Smith: I was really impressed with the work that's been done by Gerhardt Attard in his group, it's really extraordinary because they took this comprehensive and agnostic approach, they said, "Here's the data we have, here's what we can do. Let's do the sequencing, lock in the profiles, and then let the statistics sort out how these various biomarkers perform." And that really should be applauded because the more common approach is to say, "Let's pick my favorite biomarker profile and then see how it performs." And that has many limitations.

So this really was extremely helpful. And while Decipher scores didn't quite meet the statistical requirements to be called a predictive biomarker, they performed very well. And I think there's considerable potential as larger data sets are looked at and more informative events are available, that we may have that kind of information. And particularly around the edges of that patient who's maybe a borderline candidate for intensification, it could be very powerful in making decisions about whether or not that patient should have treatment intensification. So I'm very excited about that. And then using the same platform, I think they do have the capacity to do biomarker discovery, which of course is looking for the unknowns that may be even better at predicting outcomes.

Alicia Morgans: Yeah, I agree with you. I love the work that comes out of that group and really sincerely look forward to seeing where things go. One of the things that's so important about metastatic hormone-sensitive disease too, though, is that we continue to have an evolution of treatment approaches. And I wonder what your thoughts are as we consider biomarkers. How specific do you think those biomarkers might be to the treatments that are being applied and where do we go in terms of future research in a thoughtful way as we're continuing to evolve and strategize with new therapies?

Matthew Smith: Yeah, it's a great question because if you think about it, as nice as that data is, and then we got the suggestion from other work that high Decipher scores were at least associated with a differential treatment effect with docetaxel, we know that high Decipher scores are also prognostic, so these are patients who had worse clinical outcomes benefiting from intensification. So it's not a pathway-specific observation, it just may be a more precise way of identifying patients with poor outcomes with ADT alone. Nonetheless, it could be very meaningful, particularly for a test like that, that could be relatively straightforward in its application.

But to your point, we're going to always be a little bit behind. So now if you think about the next generation of studies looking at treatment intensification, for example, with PSMA-RLTs, we're going to have that clinical data before we're going to have the potential predictive biomarkers. And we already know from the experience in mCRPC that there are vast differences in outcomes of patients treated with not only PSMA-RLTs, but certainly specifically PSMA-RLTs. And so we don't have a handle on that yet at all. And now those therapies could be moved much earlier, but we'd really like to know, of course, which patients benefit most and patients who may not benefit at all, so we could be much more precise in our treatment recommendations.

Alicia Morgans: Absolutely. So lots of work to go, but thank you so much. This was really a wonderful update on where we stand today. If you had to sum it all up, what would your message be to listeners in terms of prognostic and predictive biomarkers in the metastatic hormone-sensitive setting today?

Matthew Smith: Yeah, I would really say the default, your starting point should be evidence-based medicine. NCCN guidelines, for example, are an outstanding source, but practical decisions in clinic need to be a bit more nuanced. And there are going to be patients where a good understanding of prognostic markers is going to allow you, I think, to make better decisions about which patients to intensify therapy, and then even beyond that, how best to intensify systemic treatment.

Alicia Morgans: Wonderful. Well, it is always a pleasure to talk to you, Matthew. Thank you so much for your time.

Matthew Smith: Likewise. Thank you.