High Risk Prostate Cancer and Genomics - Paul Nguyen & Alan Pollack

November 18, 2021

Paul Nguyen and Alan Pollack join Alicia Morgans in a discussion on risk prognostication in high-risk prostate cancer.   Dr. Nguyen discusses the role of genomics to risk-stratify high-risk prostate cancer patients and the role of Decipher at the time of biopsy.  He reviews data from an oral presentation he delivered at the 2021 American Society for Radiation Oncology (ASTRO) annual meeting titled: Validation of a 22-Gene Genomic Classifier in the NRG Oncology/RTOG 9202, 9413, and 9902 Phase III Randomized trials: A Biopsy-Based Meta-Analysis in High-Risk Prostate Cancer".   The purpose of the study was to validate the performance of the Decipher 22-gene genomic classifier in pre-treatment biopsy samples collected in three randomized phase 3 high-risk definitive radiotherapy trials: NRG/RTOG 9202, 9413, and 9902. This is the first validation of any gene expression biomarker on the pretreatment biopsy samples from prospective randomized trials of high-risk prostate cancer.  This trial demonstrated an independent association of the Genomic Classifier score, which was a Decipher score and was prognostic for distant metastasis, prostate cancer mortality, and overall survival on univariate analysis. In looking at any of the trials, it was highly prognostic. Each of the endpoints was significant. High-risk prostate cancer is a heterogeneous disease to personalize shared decision-making for treatments.  They also discussed a currently enrolling clinical trial, NRG-GU009/PREDICT-RT (NCT 04513717) will further determine the optimal therapy based on the Genomic Classifier score.  


Paul Nguyen, MD, Radiation Oncologist, Brigham and Women's Hospital, Dana-Farber Cancer Institute, Boston, Massachusetts

Alan Pollack, MD, PhD, Radiation Oncologist, University of Miami, Miami, Florida

Alicia Morgans, MD, MPH, Genitourinary Medical Oncologist, Medical Director of Survivorship Program at Dana-Farber Cancer Institute, Boston, Massachusetts

Read the Full Video Transcript

Alicia Morgans: Hi, my name is Alicia Morgans and I'm a GU Medical Oncologist at Dana-Farber Cancer Institute in Boston. I am so excited to be here today to talk about risk prognostication in high-risk prostate cancer. Joining me today I have two colleagues. Paul, why don't you introduce yourself?

Paul Nguyen: Great. Thanks, Alicia. I'm really excited to be here. I'm Paul Nguyen, Professor of Radiation Oncology at Dana-Farber, Brigham Women's Cancer Center at Harvard medical school.

Alicia Morgans: Thank you so much, and Alan.

Alan Pollack: Thank you. I'm a Professor and Chair of Radiation Oncology at the University of Miami.

Alicia Morgans: Wonderful. Well, thank you both. Paul, why don't you start your presentation, please?

Paul Nguyen: I'm going to get started talking about high-risk prostate cancer and genomics, which is the focus for today. NCCN high-risk disease as everyone knows includes Gleason 8-10 or PSA greater than 20 or clinical T3 or higher. Even though it represents only 15% of diagnoses, it turns out to be 67% of the deaths from localized prostate cancer, which is why we want to focus on this. The NCCN guidelines tell us that we should be adding long course ADT to external radiation. The options for high-risk prostate cancer include radical prostatectomy or external beam radiation plus 1.5 to three years of ADT. I will notice, they do mention you can use potentially up to as little as one year of ADT if you give a brachytherapy boost, but that duration has actually never been tested in a randomized trial against other durations. So I do not personally advocate for the one-year duration with brachytherapy boost. So in general, 1.5 to three years with the external beam.

We know from various studies, multiple randomized trials, that long-course ADT is better than short-course ADT for high-risk prostate cancer. This includes the RTOG 92-02 and the DART 01-05 studies, which showed us that 28 months of ADT improves prostate cancer mortality, or overall survival compared to just four months. The Bolla study published in New England journal from the EORTC showed that 36 months of ADT improves overall survival compared to six months. To make sense of this, most of the national and international trials have adopted 24 months of ADT as the standard duration for high-risk prostate cancer. So if you are enrolling on a trial nationally or internationally, typically now it's 24 months.

If external beam and 24 months of hormones are the standards for all high-risk patients, what about the fact that high-risk disease is actually quite heterogeneous? This was a study done by Dr. Vinayak Muralidhar, who was a resident in our program and is now in practice in the Portland area. He looked at high-risk prostate cancer and broke it down and found that if you have patients with just a Gleason 8, and a low PSA and T1c disease, these patients, these screen-detected patients actually have a much lower risk of prostate cancer mortality than the other high-risk patients. So these are the patients in red here, the favorable high-risk patients, their risk of death is comparable to those with unfavorably intermediate-risk disease and much less than other patients with high-risk disease. So the question is, should we really be giving one size fits all therapy for high-risk patients?

Can't we do any better and try to sub-stratify these patients more? In our interest, what we are talking about today is genomics. Can we use genomics to better risk-stratify high-risk prostate cancer, and try to personalize therapy? Now we're going to talk about the Decipher test, which was developed as a post-prostatectomy biomarker. Initially, you would have your biopsy, you would have your risk assessment, and if you chose radical prostatectomy, you could then have a Decipher genomic test to help you figure out if you should be getting adjuvant radiation or wait until the salvage therapy. That was the initial formation of the test. But in recent years, Decipher has moved to the biopsy setting where you can actually test it upfront on biopsy samples before you've decided on your treatment. This has really magnified its utility, as it helps us think about which treatment should we be using.

That leads to the presentation that I gave at ASTRO this year. I really appreciate the opportunity to represent it here for the audience. This was presented at ASTRO as an oral, and it is the Validation of a 22-Gene Genomic Classifier in some very important NRG, randomized trials, NRG Oncology RTOG 92-02, 94-13, and 99-02. These are all phase three randomized trials, and this is a biopsy-based individual patient meta-analysis. These are my disclosures here. The purpose of this study is to validate the performance of that Decipher 22-Gene Genomic Classifier, which we will sometimes refer to as GC in pre-treatment biopsy samples collected in three randomized phase three high-risk trials, definitive radiotherapy. These samples were collected, I should note up to 29 years ago. These are some pretty old samples, and I'm quite amazed that we were able to get some results out of this.

Here were the methods. These were all archive tissues, and after central review, the highest-grade tumors were profiled on clinical-grade whole-transcriptome arrays, and these GC scores were obtained. Then we use Cox multivariable analyses to validate the independent prognostic ability of the GC to prognosticate distant metastasis, prostate cancer mortality, and overall survival. These are the baseline characteristics and really just what I want you to see here, this is a pretty high-risk cohort. These patients had a median PSA of 25.8 and 44.9% had a Gleason 8-10, 38.9% had Gleason 7. Here are the bottom-line results. The GC score, which was a Decipher score was prognostic for both distant metastasis, prostate cancer mortality, and overall survival on univariate analysis. If you look at any of the trials, it was highly prognostic. Each of the endpoints was significant. And you can see this across each individual trial and across the summation of the trials.

Now, what about multi-variable analysis? Well, here's the thing on multi-variable analysis, the genomic classifier was still prognostic for distant metastasis, prostate cancer mortality, and overall survival, even after adjusting for all of the clinical factors. You can see that the GC score started to dominate. For example, looking at prostate cancer-specific mortality, the GC score was the only significant prognosticator of prostate cancer mortality, even trumping Gleason score, which was no longer significant in the model. This just illustrates the cumulative incidence of outcomes split up by risk groups that were devised for the purpose of this study. These aren't the standard commercial cut points, but this is just to illustrate the differences that you can get an outcome by the Decipher score. Really what this leads to for us, I think, is, it really opens the door to the idea that "Hey, you can take patients with high-risk prostate cancer, and you can use genomic testing to help identify which patients are likely to do well and which patients are likely to do less well."

That means that it's prognostic, but I think to really get to where we need to go, we need to figure out whether it's predictive, whether we can use that biomarker to actually change what we do and personalize therapy. And that's what the NRG-GU009/PREDICT-RT randomized trial is supposed to do. This is led by myself and Dr. Oliver Sartor. This is looking at high-risk prostate cancer patients. The trial is open now and... every one with high-risk prostate cancer gets a Decipher score. If you are in the bottom two-thirds of the Decipher score, you get randomized to the de-intensification trial, for these patients, we want to lessen the burden of treatment. They get randomized between the standard of radiation in 24 months of hormones, versus radiation and just 12 months of hormones seeing if we can reduce the treatment burden for these patients.

Whereas patients in the upper one-third of the Decipher score or node-positive patients, we are going to intensify their therapy. So they are being randomized to the standard of radiation plus 24 months of hormones with, or without apalutamide.  For these patients, we want to do more whereas, for the genomic low-risk patients, we want to try to do less. I think the idea of cutting down therapy to a shorter duration to me makes a lot of sense for these patients with lower genomic risk because remember those trials that I showed you, the benefit of longer duration ADT on survival has actually been quite modest. Even in genomically unselected high-risk patients, four versus 28, bought you a 5% prostate cancer mortality, six versus 36 bought you a 3.8% overall survival benefit, and 18 versus 36, you couldn't even see a difference.

I think that GU-009 is a major opportunity to use genomics, to see if we can safely cut the duration of ADT in half for these men in the lower two-thirds of genomic risk. I think these patients are very unlikely to be in that three to 5% who benefit from long-term ADT. I think this is going to be safe for these patients. I do want to just make a shout-out to everyone. This trial is open around the country, but massive enrollment is needed. We are going to need 38 patients per month to enroll in this trial. For those of you who are NRG affiliates, we do ask you to please open the trial and enroll patients early and often.

In conclusion, this study is the first validation of any gene expression biomarker on pretreatment biopsy samples from prospective randomized trials of high-risk prostate cancer and demonstrates an independent association of this genomic classifier with distant metastasis, prostate cancer mortality, and overall survival.  We know that high-risk prostate cancer is heterogeneous. And now we know that the genomic classifier can improve risk stratification and help personalize shared decision making. I implore everyone that the GU-009/PREDICT-RT trial, I believe will further determine the optimal therapy based on GC score. I think that is the dream for the future, is that you can use this score and decide what treatment you are going to give patients, but we need that randomized trial to make these conclusions. I want to acknowledge the co-authors and the funders, the institutions, and of course, all of the patients who enrolled in these studies. I'll turn it back to you, Alicia,

Alicia Morgans: Thank you so much for that presentation, Paul, I think it's really, really exciting to consider that we might be able to prognosticate for these patients to understand which patients are going to be at the highest risk. I wonder, Alan if you can really put this into context for us, how does this come into play now in your patient conversations? Hopefully, you're thinking about enrolling in this trial, but what else are you thinking about?  How else are you talking to patients about the kinds of abilities to prognosticate?

Alan Pollack: Well, I just first wanted to thank Paul for a great presentation. Also, I think the behind... I know that the behind-the-scenes time that is required to accomplish such a large trial.  And to see it come to fruition and inform us and also set the stage for his clinical trial GU009. What does it mean in the meantime, before we get the results of GU009? I think as Paul nicely outlined, we do know that certain patients are going to do worse based on their Decipher.  In addition to their NCCN classifications and... some of his prior work has shown that you can stratify patients into unfavorable, but less high-risk, high-risk, and very high-risk. And then when you consider those classifications and then further stratify them by Decipher you still have the questions.

A lot of this depends on patient feedback too. There are some patients that come in, even with high risk, and say, "There is no way I'm going to take any androgen deprivation therapy". So you use the Decipher as a way of saying, "Hey this cancer is very aggressive. You really need to consider this. The standard of care is one and a half to two years for what you have". It is used in that sense. Also, if you have, say a situation where it's a low volume, Gleason 8, still high-risk, but you're not really sure if the patient needs long-term hormone therapy, the Decipher score can then be used to help us to decide, well, maybe say it's a mixture of four-plus four, and four-plus three, and the Decipher comes back as being the more intermediate-risk.  That helps you to say to the patient, well, maybe you can do six months, a year instead of one and a half to two years of androgen deprivation therapy.

These are the kind of things that we go through with patients using Decipher, and the study GU009 is going to help us... to do this in a much more informed way, but it is going to be a while before we get to that point.

Alicia Morgans: I wish we would have the information immediately, of course. But as you said, it will take a while. What would you say to sites if they are trying to decide, that there are so many patients they need to enroll each month, is it worth it to open this trial and put patients on?

Alan Pollack: Well, absolutely. Because we don't really... right now, it's a little bit of a guessing game. We know that patients have a worse prognosis, but we do not know exactly how to treat them. So we need these kinds of trials to define how to apply the information to risk stratify. I think de-intensifying treatment is something we all want to do, but we want to do it in the right patient. Then there's the patient who is... very high-risk, and a high Decipher or high-risk and a high Decipher who we know is going to do poorly with standard treatment and that we need to add a second line androgen deprivation therapy agent. This study will also help us with that because the other randomization on the intensification side with apalutamide is going to be very informative. We predict it's going to be a very positive study on both ends, but you do not know until you do this study.

Alicia Morgans: Paul, one of the things that I think is so interesting, and this actually builds on some of the work you've done over the last few years, actually, maybe a decade or so of work that you've done. Thinking about competing risks for men who are receiving ADT.  From my perspective, having a better understanding for an individual man about what his prostate cancer-specific mortality maybe, I think is really, really helpful as we are trying to have these conversations about intensifying with ADT. What are your thoughts on that as we do face other competing risks like cardiovascular disease?

Paul Nguyen: I think it's such an important question, Alicia. I think we know that even patients with high-risk prostate cancer still face probably a higher risk of dying of cardiac disease in many scenarios. Trying to figure out that balance between prostate cancer mortality risk, and cardiac risk and how much we can lower prostate cancer death risk with ADT versus how much that might worsen our cardiac factors is always an important thing to consider. I would say at this moment, I would say that we... when we face a patient with high-risk prostate cancer in general, I think we still need to be giving hormone therapy. I think we know for sure, based on randomized trials that hormone therapy cuts the risk of overall death in half. We do not yet have data from any randomized trial that tells us that ADT increases the risk of cardiac death.

It certainly increases the risk of cardiac events, but not death. I think that at the moment, when I'm trying to compare those two things, I would still never withhold ADT from a patient with high-risk disease, knowing that it's going to cut the risk of overall death without, increasing the risk of cardiac death right now. I think it will be interesting as we move into the future with drugs like relugolix, which do not cause as many cardiac events as standard GnRH agonists do, that we may be able to even reduce that cardiac harm even more. But at this moment, I'm still fully in favor of ADT for high-risk patients.

Alicia Morgans: I agree. I think it is important though, just to understand what the risk for an individual is in terms of his prostate cancer-specific mortality. Because as Alan said, sometimes patients say "Doc, I just, I look at these risks and I think about cardiac disease and it's just something I'm not sure that I want to take on this additional treatment." But you're going to be able to tell them now with this information, at least, you have quite a high risk of prostate cancer mortality, and so we do need to intensify. Alan, when you think about this in your real-world patients, are you using things like Decipher to help you with those clinical conversations as you talked about earlier?

Alan Pollack: At our practice at the University of Miami, we use it routinely across all of our specialists, and it is a key factor that's considered in our tumor boards because it is such a powerful and independent predictor of outcome in addition to Gleason score or grade group. Yes, it's routinely used and it is the center of discussion about how best to manage a particular patient, considering all the factors, including comorbidity, and age. When people think about prostate cancer, they're thinking, "Oh, it's not a very difficult cancer to manage", but there are a lot of complexities involved in terms of how best to manage a patient considering other factors like age and comorbidity. There is also the influence of concerns about sexual function, the changes that occur in body habits that are very... I think, overriding factors to many patients. And so we need the added information to better present to them more definitively what their outcome might be.

Alicia Morgans: I think that's a great point. Paul, I will ask you the final question before we do a wrap-up at the end. I think it's really interesting that Decipher has been shown in your work to be helpful to prognosticate these outcomes for men with prostate cancer. It's also been useful in some other analyses in helping to understand which patients might ultimately benefit from hormonal therapy intensification, which I think is really, really important as we ultimately try to match the right patient with the right treatment. Do you have any thoughts on where we might go in the future with these kinds of assessments, all being gathered through the Decipher platform?

Paul Nguyen: I think that it has really opened up a whole new world for us in radiation oncology, to be able to have the genomic analyses from the Decipher seeing, as you were alluding to, for example, in the postop setting, the study that our colleague Dr. Felix Feng did on the postop RTOG 96-01 samples that suggested that patients with a lower Decipher score had less benefit from the addition of ADT in the salvage setting than patients with a higher Decipher score. These kinds of studies are fantastic, I think to really help us personalize the use of ADT. I think we have some pretty compelling data from... reanalysis from that randomized trial in the postop setting. What we hope to be able to do is develop this prospective data in the intact setting with the PREDICT-RT trial. I think this is the direction that we are headed in and we're excited to go there.

Alicia Morgans: Great. Well, Alan, I'll let you have... some final remarks. What would your message be to listeners?

Alan Pollack: We definitely need more information on the genomics of cancer to make better decisions. I would like to add as well, why this is so important, not just for overall survival, but when patients recur after primary treatment with prostate cancer, if it's not at the right intensity, then it usually means lifelong treatment. This carries with it a tremendous detriment in quality of life. The importance of better risk stratifying patients, so that we reduce morbidity upfront, but also intensifying so that we reduce morbidity down the road is extremely important. Paul's study really illuminates how important this could be for men with high-risk disease.

Alicia Morgans: Well, thank you. I could not agree more. Paul, I'll give you the final comments. What is your message?

Paul Nguyen: Thanks again. I would just say that I think that radiation oncology is really moving into the future with personalization. I think that genomics is going to help us better stratify patients and figure out which patients we need to de-intensify therapy and which patients need to intensify therapy moving on. I think that is what patients want and I'm excited that our field is going there.

Alicia Morgans: Wonderful. Well, thank you both. Congratulations again, Paul, on this fantastic work and I wish the two of you luck as you continue to enroll in the PREDICT-RT trial. It's really, really exciting when we finally get this information, which hopefully will happen sooner than we think because people will enroll and enroll and enroll. I think that this will absolutely inform our practices and help men with high-risk disease. Thank you for your efforts and for your time today.

Paul Nguyen: Thanks so much.

Alan Pollack: Thank you, Alicia.