Genomic Risk Classification in Localized Prostate Cancer - Matthew Cooperberg
June 20, 2022
Matthew Cooperberg, MD, MPH, FACS, Professor of Urology; Epidemiology & Biostatistics, Helen Diller Family Chair in Urology, The University of California, San Francisco, UCSF
Alicia Morgans, MD, MPH, Genitourinary Medical Oncologist, Medical Director of Survivorship Program at Dana-Farber Cancer Institute, Boston, Massachusetts
Alicia Morgans: Hi, I'm so excited to be here at AUA 2022, where I have the opportunity to speak with Dr. Matt Cooperberg about genomic risk classification, particularly in localized disease. So thank you so much for being here.
Matt Cooperberg: It's a pleasure. Nice to see you.
Alicia Morgans: Nice to see you too, and in person, which is always a good thing.
Matt Cooperberg: In person. It's wonderful.
Alicia Morgans: So, just, would love to hear your thoughts. Starting out very basic, what are some of the opportunities that we have to risk stratify using molecular characterization of our cancers, and where might these be most valuable?
Matt Cooperberg: Sure. So, I mean, there's two basic buckets, or categories, of these markers in the pre diagnostic and post diagnostic space. In the prediagnostic, we have a range of liquid tests. So blood and urine tests, some of which are based on genes, some of which are based on proteins and isoforms of PSA, all of which help us make better decisions about whether to proceed with a biopsy for men with elevated PSA. And multiparametric MR very much sits in that space as well. After diagnosis, we have a range of tests, which are mostly tissue based, looking at RNA expression to try to make better decisions about who needs treatment and how intensive that treatment needs to be. There are liquid tests being developed in that space, but they're not quite prime time yet. And again, imaging also sits in the same space. And we think about imaging as biomarkers, right alongside the blood, and urine, and tissue ones, because a lot of the same rules and considerations apply.
So when we talk about genomics, most of the focus is on the post diagnostic space, where there are three tests which have been available for coming in on a decade now. Not quite there, but next year it'll be the 10th anniversary of the first one, I think. There's Prolaris, Oncotype, and Decipher, all of which look at RNA expression in the cancer tissue itself. Prolaris and Oncotype are based on RTPCR, on fixed gene sets. Prolaris focused on cell cycle progression genes. Oncotype based on four different cancer related pathways. And then Decipher is 22 genes that were selected by more of a computational process really intended to predict early metastasis. All three of them have been pretty well validated, at this point, and are available for clinical use. Where a lot of the questions remain, is how exactly to optimize their use. Which decisions, specifically, will they help with? Which patients are the best to order them? And how, exactly, do we react to the results of these tests?
Alicia Morgans: And so where do you think about these being most useful? Now, as I think about very low-risk disease, this is not necessarily where I think about using some of these tests. Maybe it's a little bit higher-risk than that. Where do you think about it?
Matt Cooperberg: Well, first of all, very low-risk doesn't exist anymore, right? According to the AUA, at least, which is terrific news. And it's an important point, honestly. Before we really even get into any kind of biomarker discussion, it's really important that we maximize the information that we have from the clinical chart for free. And that means doing more than just using a risk group.
So we have lots of tools, nomograms, we've got the CAPRA score, now the STAR-CAP, which look at the standard clinical information and are quite accurate, actually, in terms of predicting long term outcomes. We can do even more with, it's not even fair to call them advanced anymore, with more, with other tools like PSA density, like looking at the subtypes of Gleason pattern four, or the amount of pattern four. We can actually do a pretty good job with the clinical information. So before we go to the trouble and expense of ordering any additional tests, we need to make sure that we're doing the most we can with the clinical information at hand. And we don't really do that with the risk groups. So it's not a question of saying, okay, all low-risk, or all intermediate-risk, or favorable intermediate-risk. It really needs to get a little bit more personalized than that.
So which patients are going to be the most beneficial? It tends to be cases where there is some sort of gray zone or uncertainty about the right way to go. And I say that in very general terms, because it's going to depend on the individual patient and the individual provider's sort of thermostat for surveillance versus treatment. In some cases where either the patient or the doc, or both, are not all that comfortable with active surveillance, getting a biomarker may just give confidence and, sort of reinforce a decision for active surveillance, which would have already been very appropriate.
In a practice like ours at UCSF, where we've been doing surveillance very comfortably, and very routinely for decades, at this point, I rarely get them for low volume Gleason Grade Group 1 tumors. I find them helpful for borderline cases. So high volume Grade Group 1, low volume Grade Group 2, or cases like moderate volume tumors in really young men, or folks with strong family history, sometimes African American men, other situations, germline, BRCA mutations, situations where the standard surveillance path might be a little bit different. We might think about the patient a little bit differently. So they're good as tiebreakers in those cases. But there's not much call, at least at this point, to order them routinely on all men in any given risk bucket. At least, according to the risk groups.
Alicia Morgans: And that makes a lot of sense actually. That's where I use them too. When I'm seeing that type of a patient, especially like you said, sort of this low volume Grade Group 2, anyone who's really kind of on the cusp of which way do we go. So I would love to hear you sort of just walk through a case. What if a patient has a BRCA2? It's known in their family, they're tested, they have this, but they have a low volume cancer, or low-risk cancer. What do you say in terms of molecular testing?
Matt Cooperberg: It's a perfect example. This is coming up more and more because we're doing so much more genetic testing. And I have many more men in my clinic, just in the last couple years, who are in exactly this situation. They've got a mutation. Maybe they haven't even been diagnosed yet, or maybe they have low-grade cancer at a pretty young age. And there's this controversy about, do we do active surveillance in men with BRCA mutations, basically driven by the one paper from Hopkins that showed an earlier, or greater likelihood of progression within the first six years for men with the mutations. And there's an argument, based on this, that we should be treating all these men immediately, which I personally do not hold with. First of all, it was about a 50% reclassification rate by six years. Well, that's 50% who didn't reclassify by six years. And reclassification just means, okay, well now we treat, and we treat with every expectation of cure.
And it's really important to remember this with active surveillance that a delay in treatment, as long as the treatment is still curative, is absolutely appropriate. But it's important to set that backdrop because if the mindset is, well, we're just going to treat all the BRCA mutations, there's no point in getting a marker. But that is a good situation for markers, in my view, because these cancers may be more aggressive. They may not in the setting of a germline mutation. We tend to use Decipher, for the most part, at UCSF not because it's necessarily a better test, but we get more information from it because they're running a full Affymetrix array on every clinical specimen, rather than just doing RTPCR on the fixed genes. We get a lot more data in the back end, which as a research institution, is fantastic for us. We're basically getting genomics as a side benefit of getting this clinical test. And that's a huge value-add for us. But it could be Oncotype, it could be Prolaris sort of similar clinical validation, and validity, in this setting.
So any of those tests is totally reasonable in this sort of scenario. And then you look at the result. None of the test reports is perfect. And it's an interesting phenomenon that these tests get developed typically in collaboration with academic urologists, and geneticists, and genomics experts. And it's a pretty good scientific collaboration. And then the Marketing Team takes this and develops a report. And there is a science to decision making. And there is a science to the way quantitative information should be presented to patients with very variable numeracy, and literacy, and that rarely comes into play.
So the reports have evolved over time, and they've generally gotten better, but they can still be pretty misleading. Especially, they've finally fixed this, but for years, the archetype report, for example, showed the patient and the doctor this sort of risk box reclassification from low to unfavorable, intermediate unfavorable, intermediate to, I mean, it was often very misleading. And it was actually a paper that Adam Murphy led, which was in JCO a year or two ago, showing that in low literacy patients getting an archetype actually increased the treatment rate for low-risk disease. And I'm sure that this sort of box changing on the report was part of the explanation for that.
So to answer the question, we get the test, but then the way we interpret it with the patient actually gets a little bit more subtle. And it's easy when the score is low. It's easy when the score is high. We've got our answer. It's very often in the gray zone. And then we say, well, you where you were before, which is, we don't know exactly what to do, but we can probably follow this for the time being with the surveillance.
Alicia Morgans: Well, that's really good to hear. And it's good to hear too, that sometimes these results even leave you still in the gray zone, which, of course, is not what we want in medicine, but tends to happen sometimes. And even in a center like UCSF, this can be something you think about. But that's why it's so important, I think, to have some practice around using these, and counseling patients, and knowing that you may get an answer that's low or high. But you may still be in the gray zone. So you want to know, before you go into that conversation, what you're going to do with that information.
Matt Cooperberg: Absolutely. That's the reality of the biology, right? I mean, when I give talks in this, I have the slide. Everybody wants a pregnancy test. If it turns blue, you take out the prostate. But the test that comes back maybe pregnant is a little bit less helpful. But that's often what we're talking about here, because the biology is a continuous spectrum. It is not the case that there are low-risk and high-risk prostate cancers. So nothing in between. The quote, unquote, "intermediate-risk category" by NCCN or UA standards is incredibly broad in terms of its biologic and clinical capacity to metastasize and to cause clinical problems. And we don't begin to capture that diversity with the unfavorable versus favorable dichotomy. So within that group we know there are the folks, the 5% fused gland subtypes that genetically and genomically are often indistinguishable from the Gleason Grade Group 1s. And then we've got the high volume four threes, which are quite aggressive. And then in between, we've got a huge range, and that's reflected in the genomics.
And the other thing we have definitely noted over the years is that it's uncommon to have a major discrepancy between the clinical pathologic information and the genomic information. So for the most part, as you would expect, higher grade higher volume cancers tend to have higher scores. But the interesting thing is when they're discrepant, the genomics tend to be on the correct side of things. Meaning in the uncommon cases where you've got a favorable looking tumor with a bad genomic score, those tend to progress over time. And conversely, when you have a bad looking cancer under the microscope with a favorable genomic test, they tend to have a little bit more of an indolent clinical course. So it's not that common. We've always estimated somewhere between 5% and 10% of the cases, that's a guess, but it's probably about right, where we really do get real information that we did not have before from the genomics. But it's hard to predict in advance which those patients will be.
Alicia Morgans: I can imagine and certainly see that in my own clinical practice. So thank you so much for walking us through all of that, and really kind of giving us a sense of these tests in a real clinical practice, and with the mindset of someone who is incredibly facile with them. So thank you for your time and for your expertise.
Matt Cooperberg: Always a pleasure.