Integrating Genomics into Localized Prostate Cancer - Todd Morgan and Dan Spratt
Todd Matthew Morgan MD Associate Professor Specialties: Urology Area of Practice: Prostate Cancer, including robotic prostatectomy and active surveillance; Bladder Cancer, including bladder preservation approaches as well as open and robotic radical cystectomy; Kidney Cancer, including open and robotic approaches, and nephron-sparing (i.e. partial nephrectomy) when appropriate; and Penile Cancer, including robotic inguinal lymphadenectomy. Michigan Medicine Urology Oncology Clinic | Rogel Cancer Center
Daniel Spratt, MD Associate Professor Laurie Snow Research Professor of Radiation Oncology Associate Chair, Clinical Research Chair, Genitourinary Clinical Research, Rogel Cancer Center Director, Spine Oncology Program, Rogel Cancer Center
Charles J. Ryan, MD, B.J. Kennedy Chair in Clinical Medical Oncology at the University of Minnesota and Director of the Division of Hematology, Oncology and Transplantation. He previously held the position of Professor of Clinical Medicine and Urology and the Clinical Program Leader for Genitourinary Medical Oncology at the UCSF Helen Diller Family Comprehensive Cancer Center
Charles Ryan: Hello from the PCF 25th Anniversary Retreat in Carlsbad, California. I'm really delighted today to be joined by two colleagues from the University of Michigan, Rogel Cancer Center, Dan Spratt, a radiation oncologist, Todd Morgan, urologist. You guys are doing a lot of really exciting work at the University of Michigan. I've been following it for a long time, and it's really great to have an opportunity to talk to you about it.
Let's start with your approach to localize prostate cancer. You two are really helping to lead the way for decision making and the integration of genomics into decision making. So, Todd, tell us about your approach as you sit with a patient and you're trying to decide which direction you should go, and how are you using genomics today?
Todd Morgan: My approach to a localized prostate cancer and it will be a great Dan's as well. The challenging decisions often are around patients with kind of borderline high volume, low-risk disease and lower or favorable, intermediate risk disease. And we know that for very low-risk patients, those patients should be on surveillance. And we know that really the vast majority of low-risk patients should be on surveillance. But there probably are some low-risk patients who should undergo surgery or radiation. And there are definitely some patients with favorable intermediate risk, meaning Gleason 3 + 4 disease so it can safely be managed on surveillance. The question is, how do we do that appropriately? And there's a lot of use of MRI. MRI does a good job of finding, say the three plus four lesions or the four plus three lesions that you might miss.
But there's no question that genomics can be helpful too. And the question is, who, when, how often. For me and my practice, I really use genomics for patients who are in that favorable intermediate-risk side of things. And so those patients where they're interested in surveillance, they've undergone a biopsy. Maybe they've already undergone a repeat biopsy, it shows low volume 3 + 4 disease, especially low percent pattern 4 disease. Then I find genomics to be really helpful and so, there are issues around cutoffs and thresholds and maybe we'll talk about that some more and how that really influences decisions. And we have some data of showing how that's been done in Michigan that we can definitely touch on.
Charles Ryan: So the clearly active surveillance patients, the Gleason 3 + 3s or with the low PSA is you don't need any genomic test to put them in surveillance, the clearly high-risk patients, you don't need a test to know that they're high-risk based on Gleason and PSA and others. It's those patients who are in the middle. What's your sense as to how much these tests are being utilized out in the world?
Todd Morgan: Yeah. So we've looked at that in the state of Michigan and we have the incredible luxury of having the Michigan Urological Surgery Improvement Collaborative or MUSIC and through MUSIC, we collect data on basically 90 plus percent of the urology practices in the state, share the data with each other. It really is a transparent collaborative. Everybody has access to the data. And for the last couple of years, we'd been collecting use of Prolaris, Decipher, and Oncotype. So what we've seen is that the majority of patients are managed without undergoing these tests. But there are some practices that use these tests all the time. There's some practices that are ordering some type of genomic tests on over 50% of their newly diagnosed prostate cancer patients.
Now, there are a lot of practices that don't use these tests at all. And there were a bunch that are in the middle there that seem to be using it for in specific instances. But at this point, there's just huge variation in how these tests are being used. There's also huge variation in the use of MRI. And so we're trying to learn what's the most informative, what's the best in terms of the predictor of upgrading or adverse pathology, at surgery? Safety of surveillance, and just what's driving decisions more than the other tests?
Charles Ryan: So you're a urologist, you're a radiation oncologist. Is there anybody saying that the genomics tests can say we should take this patient from the urologist and send them to the radiation oncologist, that he should have radiation or he shouldn't have radiation? How does the genomic world, you can answer this question, Dan, help answer this age-old question of surgery versus radiation?
Dan Spratt: It's an excellent question. I think right now there's definitely a lot of people in our group and others have tried to find potentially genomic biomarkers that may potentially portend increased radiation sensitivity. There's some been created such as PORTOS, Dr. Felix Feng from UCSF created that. There's a signature called RSI, created from Moffitt Cancer Center. Unfortunately, I think these are very early on and haven't really been prospectively validated. And so I think at University of Michigan we really use the genomics to help inform risk stratification. Do you need treatment even? If you need treatment, then we have kind of an equipoise type discussion of them and use other factors. Age, comorbidity, urinary symptoms, the patient's preference to help decide. So I think right now, I don't think genomics are really helping guiding that discussion between surgery and radiation, but I think they're definitely helping to help guide potentially additional treatment that we can talk about, especially patients who do undergo surgery and potentially this discussion, do they warrant to have radiation afterwards?
Charles Ryan: Okay. So let's talk about Prolaris first. Tell us a little bit about when Prolaris is used and what scores are significant and how would you recommend that the treating physician watching this would use Prolaris in his or her practice?
Dan Spratt: So it's a great question. So the three tests Prolaris, Decipher an Oncotype, Dr. Morgan and I have shown that essentially they each have different cutoffs, but all of them kind of define their own low-risk group. And so you kind of need to use their risk groupings of the low genomic score.
Charles Ryan: The report tells you this is actually low risk, you don't have to interpret the number necessarily. It'll basically deliver you the interpretation. Right?
Dan Spratt: A little of both depending on the test. Usually, they'll give you an absolute risk of some type of endpoint and each test is different. It may tell you the risk of adverse pathology, risk meaning sort of upgrading if they were to undergo surgery, risk of metastasis, risk of dying from cancer. And often they give you the absolute percentage points. Sometimes it's a little challenging I think for us in practice and probably for other practitioners given that the way we treat patients is based on their NCCN maybe CAPRA risk grouping, which isn't really tied in very finely to the absolute risk of recurrence. Sometimes it creates this discordance or unclear decision making of how we act upon these tests. But I think to answer your question, I think in general if you have a question, like Todd said, where it's this sort of favorable intermediate risk and they have a reassuring and the way sort of we've defined it is sort of by the company's specified reassuring value, that should make you think about more conservative management such as active surveillance.
Charles Ryan: Right.
Todd Morgan: The issue of cut point is really fascinating and we've run into trouble with that. In the past we think about using a PSA cut point of four, PSA of 3.9, basically it's the same risk of PSA four, and so drawing these lines in the sand are difficult. That's it. That's how we practice clinical medicine. The decisions are often binary. Yes, do something. No, don't do something. And so the three tests do each have their own cut points. Again, they give you the continuum of risk, but they also have a proposed cut point.
The problem we've found when we looked in MUSIC data was if you looked across all Gleason 6 patients, for the three different tests, about 15% of patients with Gleason 6 disease who underwent Prolaris testing had a test above the threshold, but much higher percentage is 40 and 50%, if they underwent Oncotype or Decipher testing. Intuitively, it shouldn't matter which test you get. The percentage of patients who are have their management changed if, assuming that you think that all three tests do relatively well and the three tests do have really strong validation data. And so that's a big disparity there that we're trying to hammer down, and Dan has done a lot of work around this clinical genomic risk grouping and to try to translate genomics into a more NCCN style classification. I think that's where we need to go with all of this.
Charles Ryan: I asked you about Prolaris, I was asking you about Decipher, is there a case to be made that we should be getting one or the other or should we ever get more than one on an individual?
Todd Morgan: I don't think so. I don't think anybody could tell you right now that one test is better than the other. There's no comparative data-
Charles Ryan: But they don't validate each other is what you're saying also. You could get a Gleason 3 + 4 that says on Prolaris this is low risk. You could use Decipher, they might not say it's low risk.
Todd Morgan: We have some preliminary data that says that they do tend to be aligned, they correlate again along that continuum of risk, they correlate with each other. Now if you use a given threshold, you can be totally disparate, but in general, there's definitely differences. But in general, for the patient that's higher risk, Prolaris are going to be higher along that continuum for Decipher. We don't know which one's better than the other. They all again have pretty strong clinical validation data.
Charles Ryan: What's the utility of Oncotype?
Todd Morgan: It's in the mix.
Charles Ryan: It's just another test?
Todd Morgan: Yeah. Again, here's some who would say, "Well Oncotype it's validated for an adverse pathology endpoint. That's what I want to drive my decisions with for the newly diagnosed patient." Some would say, "Well, but Prolaris gives a metastasis and death endpoint. That's what's really important in prostate cancer and any cancer when we talk about metastasis and death." You can argue both sides of that. And realistically what you need a tool for risk stratification, all three of them give you a tool for risk stratification.
Charles Ryan: Right. And what's also interesting about these tests is that these are all panels of different genes and the genes themselves are rarely discussed. I think a lot of us out there don't even know what's on the panel of genes. If I asked you to list all the genes, you might be able to do it, but most people wouldn't. And, are these genes that are biologically telling us a story about the propensity for metastasis, about proliferation, whatever. I'm unclear on that question myself.
Dan Spratt: So for each of these tests, each of these companies, they use a different algorithm, how they selected the genes. And so for Oncotype and Decipher, primarily what they've done is using a training cohort, testing probably thousands of genes, they then, basically shortened it down to a small gene list, typically under 50 genes and then validated it for the endpoint of interest and sort of optimize their then cut point and thresholds of weighting each gene. They start looking at proliferation, invasion, et cetera, but that's not necessarily a priori. They said we want to just look at proliferation, invasion, genes.
Prolaris, is a little bit different because it's been sort of developed in concert sort of not necessarily just in prostate cancer because it's looking at cell cycle progression. So it's sort of looking at like a Ki-67, some people say on steroids, it's looking at genes really related to proliferation. It wasn't necessarily created and then optimized in prostate cancer for the endpoints and reports, but because it's more tied to potentially biology, it validates as well.
So I think similar to Oncotype for breast cancer, although they kind of bend them into these biologic pathways, ultimately it's really just a gene expression pattern that seems to correlate with risk.
Charles Ryan: I think the next step, I mean it's great that these tests are out there, you're using them clinically, they're available on a day-to-day basis. It's great that they're being disseminated through the state of Michigan and presumably other states. What I hope also is happening is we're integrating these into our clinical trials, that we're doing in localized disease and in adjuvant settings and things like that, so that we at the end of the day can evaluate the efficacy of new therapies with regards to their genomic profiling. And I think both of you are involved in some studies that are looking at that and just your thoughts in terms of the future, as you think about new therapies in prostate cancer and how we can integrate genomics.
Todd Morgan: I agree. I mean this retrospective observational data is really interesting and important, but there's so many biases and we need to know what the right thing to do is. And that's what we all want, that's what the patients our patients want, to know what is the right thing to do, what is the most useful way to approach these problems? And so, we feel very strongly that prospective trials need to be performed in this space. And so one of the trials we have ongoing that just completed enrollment is called the G-MINOR trial. Genomics-
Charles Ryan: Who's the musician at Michigan who keeps coming up with these acronyms? It's great.
Dan Spratt: Dr. Morgan.
Todd Morgan: Stick with the theme. It took me a little time. But yeah, so genomics in Michigan Impacting Observation or Radiation, not even that much of a stretch to hit that acronym. It's a prospective RCT for men who've undergone prostatectomy and are at high risk of local recurrence. So pT3 disease or positive surgical margins. These are the patients where we're asking the question, do these patients need adjuvant radiation or not? And there's, of course, a lot of data is saying that the vast, vast majority of these patients do not get adjuvant radiation, despite lots of randomized controlled trial data saying that we should strongly consider it.
And so, we randomized, we just completed enrollment this past summer. We randomized 350 patients across 13 different MUSIC practices. And so academic and community practices, randomized to getting the Decipher test or not getting the Decipher test. And the kind of the other hook is that we use the CAPRA-S score in both the control and the intervention group have a free nomogram available too, because the question isn't just do biomarkers help us, but really do biomarkers help us above what's free and readily available to us in terms of the clinical tools? And so we use CAPRA-S as the clinical predictor, so all patients have access to CAPRA-S, half the patients have access to Decipher. The primary endpoint are the decisions around the use of adjuvant radiation. But all of these patients will be followed long term to see if there's any impact on metastasis or even survival down the road.
Charles Ryan: Great. Great. Excellent work. So, any other comments on the prospective use of genomic testing and in prostate cancer studies underway? Integration of new therapies?
Dan Spratt: Absolutely. So we have, myself and Dr. Felix Feng or the Co-PIs of an NRG trial. So a cooperative trial in North America assessing again in patients who have actually undergone surgery already who've experienced a recurrence. And all these patients, they get the Decipher test done, but that we're not actually using those 22 genes in the Decipher test, we're using PAM50 signature, which we have some retrospective data saying it may help us guide who benefits intrinsically or a predictive biomarker of the benefit of androgen deprivation therapy. And so this trial is one of the first trials in prostate cancer actually to stratify the trial by their PAM50 if they're luminal or basel type to see if not only do we see if next generation hormone therapy benefits these patients with radiation, but can we finally actually identify predictive biomarker?
And so I think, the ongoing trials are trying to better understand if these prospectively validate and are strong prognostic markers to help. Should we intensify treatment, but can we actually get predictive biomarkers to use these to personalize treatment?
Charles Ryan: Right. Of course, this is already happening in breast cancer, right? You use genomic profiling to say you don't need chemotherapy, you do need hormone therapy or vice versa. And we talk about intensifying therapy and that's wonderful and we need to do that for people who are destined to progress. But we also need to be mindful of the fact that we're able to de-intensify therapy for women with breast cancer by using genomic testing. And that's a great relief because let's face it, a lot of guys are getting radiation they don't need, and a lot of people aren't getting radiation that they should get. So that's what we need to do, and I'm very grateful to both of you for working in this field, and I look forward to hearing the wonderful music of your data emerge and hopefully it won't be a B flat.
Todd Morgan: Well played.
Charles Ryan: Terrible joke.