Deciding Against Active Surveillance: How Genomic Testing Impacted Treatment Decision-Making Among a Predominantly African American Population - Peter Gann and Adam Murphy

August 3, 2022

Peter Gann and Adam Murphy join Ashley Ross to discuss how the use of the Genomic Prostate Score (GPS) may intersect with patient health literacy to affect decision-making regarding prostate cancer treatment. The discussion begins with an overview of genomics in prostate cancer, particularly the lack of validation of the Oncotype DX assay on African American men. From there, the conversation then moves to how the recognition of this gap ultimately resulted in the study led by Gann and Murphy, which focused on a predominantly African American population at minority-serving institutions, or “safety-net” hospitals. The conversation explores key questions about factors in decision-making regarding active surveillance versus treatment, including population, the importance of continuity of care, patient trust, and how physician communication impacts patient decision-making.

Biographies:

Peter H. Gann, MD, ScD, Professor of Pathology in Urology and Co-Leader of the Prostate Cancer Working Group at UIC. His primary appointment is Professor in the Department of Pathology. He serves as Program Leader for the Carcinogenesis and Chemoprevention Program of the University of Illinois Cancer Center, and as Academic Director for Tissue Imaging in the Research Histology and Tissue Imaging Core (RHTIC) facility at UIC.

Adam B Murphy, MD, MBA, MSCI, Assistant Professor of Urology and Preventive Medicine (Cancer Epidemiology and Prevention)

Ashley Ross, MD, Ph.D., Associate Professor, Department of Urology, Northwestern University, Feinberg School of Medicine, Chicago, Illinois



Read the Full Video Transcript

Ashley Ross: Hi, this is Ashley Ross for UroToday. I'm a urologist at Northwestern Feinberg School of Medicine, and I'm really delighted tonight to be joined by two experts in the field of prostate cancer and in research that has to do with health disparities as well. This is Dr. Adam Murphy. He's an assistant professor of urology and preventative medicine at Northwestern Feinberg School of Medicine. One of my friends and colleagues there. Welcome, Dr. Murphy. And also Dr. Peter Gann, he's a Professor of Pathology, Department of Urology, but also co-leader of the Prostate Cancer Working Group at the University of Illinois, College of Medicine. Welcome, Dr. Gann.

Peter Gann: Thank you.

Ashley Ross: They recently collaborated to publish a really important paper in the Journal of Clinical Oncology and summarized data that they did around their ENACT trial, which was a trial where they're engaging newly diagnosed men for clinical decision-making in prostate cancer. And the population was predominantly a minority population, African American. It talked about genomics in that study. Dr. Murphy, maybe talk to us a little bit about what are genomics in prostate cancer and how does it apply to specifically Caucasian men, African American men? What do we know about its utility there?

Adam Murphy: So it's interesting. Let me just caveat this with saying that I was fortunate enough to work with Dr. Gann on this. It was his brainchild, and we came together because we had just complementary skillsets. I was recruiting patients in other studies working at the VA and at Cook County Health and Dr. Gann was the biomarker expert and we actually found that we were working on studies that were conflicting. So we decided to play well together in the sandbox and he came up with this alternative idea, which was really kind of fun. And he helped me do a validation of the Architect DX Assay, which is a genetic assay in African American men. So I will give him a lot of credit for that and I will defer some of my discussion of the background on the study to him.

Adam Murphy: But what I'll say about genomics, in prostate cancer was Oncotype DX with its relative gene expression assay that measured 12 cancer genes and then five housekeeping genes, and it essentially was across four different cancer relevant pathways. They put all these different pathways into basically a linear regression formula with higher numbers, meaning that the cancer was acting more aggressively from a genetic perspective. It was a fairly well-validated assay, but there were very few African Americans in the initial studies. And so Dr. Gann helped me figure out how to validate it in an African American population.

Adam Murphy: We did a comparative effectiveness study. So that was one angle of our work together. And Dr. Gann's idea was, well validation really shouldn't just end as showing that it has a high area under the curve for something. The real validation is how it is clinically utilized and how it affects in this case treatment decisions. And one of his ideas was that it was a probability-based test, which is unusual for tests that patients typically see. They get a pregnancy test, you're pregnant, you're not pregnant. Your PSA is high, it's not high. So I think that was one of the unique angles for why this study was a unique trial.

Ashley Ross: Maybe I'll ask Dr. Gann, the original Architect DX, which is run on the [inaudible 00:03:56] fixed, paraffin-embedded tissue and like Dr. Murphy said is a gene-based analysis. It was originally developed to predict adverse pathologic features. And maybe just for our audience, tell us about adverse pathologic features after prostatectomy and why that might be important in a population thinking about surveillance versus treatment.

Peter Gann: Right. Well, when we speak about adverse pathology features, we're really talking about the findings after removal of the prostate at a radical prostatectomy, and the key things there, of course, are the Gleason Group Grade, as well as the extent of growth of the tumor, whether it's extra prosthetic or not. So the assay was really validated to make a prediction about what would happen if the particular patient goes to surgery and then has the resulting pathology. And exactly as Adam said, I mean, in our current environment, our regulatory environment, the FDA does not even consider whether or not a laboratory-developed test such as this. So this is a test that's going to be done in California at the laboratory of the vendor of the assay. It's not going to be licensed for use in hospitals and clinics all over the world.

Peter Gann: In that circumstance, your audience really is third-party insurers. And so we never ever had a trial, which said, wait a minute, the real clinical utility, if this is a decision support tool, which is what it is, is how well does it improve the decisions that patients make and what are the factors that determine how it operates within that context? So I have to say, we checked the literature on this and couldn't find a single randomized trial of that nature. So apart from the prostate cancer context of this, it's a very unusual trial in general, because it's very simple. We randomly assign men to either receive a GPS test plus standard NCCN risk-based counseling, or just the standard NCCN counseling regarding risk alone. So that's the information that both the urologist and the patients were given and our primary endpoint was really to see, well, how does it actually affect the decisions that are made? There were a lot of other things that went into the analysis that some of which actually surprised us, but we can go into that later.

Ashley Ross: To go into the study. So, it was men with, we'll call it favorable risk disease that might be candidates for surveillance. And in the past, what we had seen with the Architect DX prostate and some of the other genomic tests for prostate cancer, primarily in Caucasian populations, was that if they had those tests and it showed that the risk of the disease molecularly was maybe a little bit lower, that active surveillance was often chosen as a preferred option and they would go on to surveillance.

Ashley Ross: Now there's a lot of differences in your study. It's a randomized study where they're getting the genomics, or they're just getting counseling and the population is also different. So Dr. Murphy, tell us a little bit about the population for your trial, and then also maybe highlight some of the differences from the patient population. And also the, I think there's also differences in the provider population compared to the rest of the real-world data that led to respective assays.

Adam Murphy: Well, that's an interesting point too. So one of the things that I thought was really unique about our study is that we had three really high minority-serving institutions involved. So there was Cook County Health, the County Hospital, Jesse Brown VA, which is the VA system is all a safety net system. And then the University of Illinois at Chicago, which is, of course, an academic medical center but it also, depending on the definition, fits the definition of a safety net hospital or, and it's definitely a minority serving institution. So that's one thing that I think was really great about our population because you almost could think that this assay could help people choose actual surveillance. Because in the initial time when we were writing this grant, the idea was that urologists weren't very comfortable with who should get active surveillance, and using these tests could help encourage both patients and urologists to go that direction.

Adam Murphy: This group is also a more high-risk group where 55% of patients are going to be diagnosed with cancer on the biopsy and of those over half are not necessarily eligible for active surveillance. So it's a higher risk group doubly. So what genomic tests would do to this group was unclear at the time, too.

Ashley Ross: And so, a couple things struck me about this group. One, like you said, it was primarily minority. I think it was 70% African American.

Ashley Ross: The second thing about this particular group was that there, and let me know how you measured it, but their health literacy was lower than what would be presumed from some of the other studies that had been published respectively. Again, they didn't measure it the way you guys did. Maybe either you or Dr. Gann, tell me how you measured health literacy and what the percentage of men, where their health literacy lay on the spectrum.

Peter Gann: The primary instrument that we used for health literacy was a brief health literacy questionnaire, just because for practical reasons. I mean, the long-form health literacy which we used in a subset of the patients, it takes a long time. It requires the patient actually read text and then their comprehension's tested. And it simply wasn't practical to keep people in the clinic that long. But what we did do was we did our own internal validation and we found that the brief health literacy instrument correlated very well. And as it did turn out, health literacy was a very important factor in modifying the effect of the GPS on treatment choice. Because in fact, men among a low literacy, men below the median, were seven times less likely to choose active surveillance compared to the control group, if they were in the low literacy group. Whereas in the higher, above the median group, there was no difference at all between GPS versus control.

Adam Murphy: And the median health literacy level was already on that scale at what would be considered marginal literacy. So there's insufficient, marginal, and then sufficient literacy. The median was marginal. So compared to other trials where they're in active surveillance cohorts and they're talking about treatment decisions, our folks to baseline were less literate than many trials would show.

Ashley Ross: Yeah. So I thought a couple unique features in the study, one, the health literacy of the patients was lower than average. Two, the providers were much more apt because you selected providers that would be willing to place such patients on surveillance. And it's turned out that your surveillance rates were exemplary. Even with people choosing the treatment, it was about 80% on surveillance, even when you're considering NCCN low risk and then this category you guys melded together, which we'll call low intermediate risk. They were still 80% surveillance rates, which was great, which really moved the bar from 40%, what we were used to and saying, well, this test encourages surveillance to 80% on surveillance. And then as we see some people come off surveillance, we obviously won't know from the study if that was the right thing for them or not because, but it's an interesting juxtaposition.

Ashley Ross: And then the third thing I want to talk about before we go more into the detail of the study is that because it was a lot of African American men, maybe you can speak towards, either you or Dr. Gann, can speak towards the idea that, do African American men who have low-risk prostate cancer, do they tend to have higher genomic scores than Caucasian men, or at least in your population did it tend to be higher? I'll just preface this by saying there's other genomic assays out there like Decipher and there's a study called the Van Dam Study that was presented at ASCO, but it has yet to be published, that's Kashi Imoda's work. And that already shows some hints at ASCO. Your previous work actually, Dr. Murphy, had shown that the high-risk scores in this low-risk population, genomically, I think they're a little bit right-shifted compared to the general Caucasian population, but maybe you can speak to, one of the two of you can speak towards that a little bit too.

Adam Murphy: Well. So this is one of the other projects that Dr. Gann and I worked on. So we really believed that the score should be higher because the data that had come out preliminarily from the company, which was then called Genomic Health, was that the distributions between Blacks and whites were the same. And I didn't believe that. And I don't think Dr. Gann believed it either because we knew that rates of adverse pathology seemed to be higher stage for stage per NCC and risk group for Blacks. Especially in the lower end of the spectrum, where you find more aggressive tumors, when you think that they're going be localized or low risk. So we expected it to be different, but actually in our own analysis, which we published in Neurology two years ago, the distributions were in fact the same. If you looked at NCCN low, very low and low combined, and then the intermediate risk, the distribution between Blacks and whites were the same.

Ashley Ross: And so then this brings us to it. Now, a couple things about the study that I want to wrap my head around and see what we learned. The first thing that I think was a strong point that was overarching was that I was very impressed. And maybe it's just a credit to you, Dr. Gann and Dr. Murphy, that you were able to conduct this study in this sort of underserved population, that the surveillance rates were really high in this underserved population. You mentioned in the text that this shows that it's feasible and actually that doing these types of studies that are pragmatic studies are not only going to inform us, the community, but also is going to allow probably, potentially, best care or better care for some of our patients. I think that's a very important point.

Ashley Ross: But now to get into the study, what you'd want to see is for the people who got the genomic testing, if they were higher risk, they go maybe towards treatment, lower risk, maybe they stay on surveillance. I know that you said the major driver that, and it really stuck out, was health literacy. If health literacy was low, they were going towards treatment, but how was that influenced? How about if health literacy was low and the genomic score was lower, low on the low end, were they going towards treatment still? Or were they going towards surveillance?

Adam Murphy: Dr. Gann, listed about 20 plots of this, trying to figure out this issue. I'm sure you'll have something cool to say, I'll let you go first.

Peter Gann: Yeah, no, I mean that sevenfold higher choice of treatment versus AS was the result of a statistical model, in which everything, the risk level was part of the model, it was adjusted. So we're talking about an effect that is independent of NCCN. And I think we really need to talk about the role of the urologist in this because one thing we know for sure is the urologist has an enormous influence on what the patient decides to do. One thing we don't know is where is this health literacy effect actually happening? Is it happening at the patient level, or is it happening at the urologist level? It's entirely possible that a urologist who did not see the health literacy score—because they weren't shown them—perceives that a given patient is going to have difficulty understanding the GPS score.

Peter Gann: They're interpreting the GPS score and thinking that this is a patient who may not do well on active surveillance because the demands on the patient on AS are much greater than simply showing up for your surgery and the follow-up after that. So that's future research that I think really needs to be done. We have some data on the doctor/patient interaction, but unfortunately, I don't think we really capture enough of it to really address that question.

Adam Murphy: I have another view of that same issue, which is—which I agree with Dr. Gann, that's definitely a part of it. The other thing is that William Dale, from the University of Chicago, when he was there, showed that people coming for their prostate biopsy thought that if they were African American, they thought that they had a very low risk of having prostate cancer and had a very low risk of having aggressive prostate cancer when they showed up for the biopsy. So my sense of it is that some folks with low health literacy actually just believe that they have low risk of this cancer being significant, especially when they say, you have a low risk tumor or a favorable risk prostate cancer. And then you have this test that show that you have a 35% risk of adverse pathology on average, that changes their perspective just by showing them probability data. And I also think that can affect people. So I'm not sure which version of it is true or if it's both, if it's a mix, but I'm worried that both of them are at play.

Ashley Ross: Two questions around this, one is—and I don't know all the nuances—but you also looked at decisional regret among the participants. And there's a lot of factors as you've highlighted. Decisional regret, what did it look like? Can you expand on that a little bit? How did people feel that when they did go to treatment or they did go to surveillance? Can we glean anything from that?

Peter Gann: Okay. Yeah. I mean, so decisional regret was measured, I believe it was on visit four. So the patient had had their treatment at that point, and they were able to use a standardized instrument, which again, had been validated in previous research to comment on their level of regret about their choice and essentially in the overall analysis, which is what we published, we really didn't see any effect at all. There appeared to be no effect on regret. Now, the real story is a little more complicated than that. We're still working on it because there are some very interesting things with regard to who expresses regret, who doesn't, but in terms of the randomized intervention, we didn't really see any impact.

Ashley Ross: Yeah, I think a couple other takes, so there's a couple of next steps/take homes that I at least got from this, or at least, or I would think provocative things that I thought about. And for Dr. Murphy as a urologist, I'd like to get maybe some of your comments here.

Ashley Ross: I think as a field, we still struggle to understand what's a win on surveillance. So if you have a patient with say, low risk disease and a high genomics score, who, as Dr. Gann mentioned before, it might be a hardship for them to make it to multiple appointments, have multiple serial biopsies and say the man is younger, say 55. And the urologist tells him, "You're going to have multiple [inaudible 00:20:53] material biopsies, but your chance of progression in 10 years to treatment to say 437 or something like this and then we have to do treatment is high, 80%.”

Ashley Ross: It could be a tough choice for any individual and certainly for one of more limited means to stay on surveillance, knowing that that might be an inevitability. And I don't think Dr. Gann, that we've come to grips with what we consider a win. For me, I usually tell my patients that if I think that there's a 50% chance you'll still be on surveillance in five years, it's worth doing because you don't have any side effects of treatment. But if I think that you're going to have worse disease in one or two years, certainly I want to take you off of surveillance. But as a community, and we've had debates and discussions even on UroToday and others, I don't think that we've settled on what an act of surveillance win is. So that's one thing I want you to talk towards.

Ashley Ross: And the second is, how can I better assess and speak to my patients who may have lower health literacy? Because I take it for granted, and quite frankly, I sometimes can be blindsided by the fact that I've been talking over the head of the patient and they're lacking understanding, and I try to make up for it by doing something that you did in your study, which is having multiple touch points with the patients. You had multiple visits and you assess at different times, but let's tackle those two questions. How should we think about a win on surveillance? You could do this from your perspective. And then the second is, how do we talk to patients who might have lower health literacy and help them understand data?

Peter Gann: Yeah, well, you referenced indirectly the long-term follow-up studies on AS which do show that over a long period of time, the probability of either progression or just simply patient choice going to treatment is very high. But it also shows that there's a lot of benefit in deferring that treatment if it's long enough.

Peter Gann: Personally, I don't believe that there is ever going to be any single threshold or decision about what a win is. Like a lot of things in medicine, I believe this is going to be an individualized situation involving each patient. With regard to how you arrive at the best communication and the best, hopefully, the best decision for the patient, continuity of care is critically important. I think building trust with a patient and being able to see them over time, it's not important only for this, but virtually anything really important that I can think of in medicine. I personally started out as a primary care physician and I can attest to the value that I experienced in being able to continually see the same patients over time.

Peter Gann: I think that deferral, it's a quality of life issue like anything else. If somebody can defer surgery for, let's say five years or six years, but they're anxious and pretty miserable the whole time, it's really hard to consider that a win. So I think we've got to look beyond the numbers and that's where it gets very individualized, very tricky. You really got to look at what's the quality of life of that patient while they're on AS.

Ashley Ross: And Dr. Murphy, I think two follow-up questions. One is, what did you learn about, and how should we maybe think about approaching the patient with low health literacy? I like what Dr. Gann said a lot, which is continuity of care, establishment of relationship that has a lot of trust and openness and has many touchpoints is critical.

Ashley Ross: And then the second question, after reading your study, should we be doing genomics in these men? Does it help? What are your thoughts on that?

Adam Murphy: I guess what I learned from this study was, it was very important to have continuity of care. When you have low health literacy populations, I really believe that the first parts of it are all based off of the fact that they trust what the urologist is saying, they're relying on a relationship. And they're using the tone of voice, they're using this probability data, they're using, maybe what someone in their family or friends have said to make a decision. And so actual surveillance can be dangerous in a low health literacy population if you don't have the support to keep these people in care. And so one thing that, when Dr. Gann and I finished this study, we were really wanting to, obviously, finish the papers, but also do a part two to the study because we thought that the research coordinators acting as patient navigators was critical to us having the loss of follow up, the low dropout rate. And I think that's what really keeps people safe in the situation because I don't think that they understand.

Adam Murphy: Most people do not understand what it means to have a biopsy every few years when they're making this decision. They just know they don't have to have surgery or radiation right now. And so I think that their health literacy is building over the serial visits, their comfort, or their anxiety around this builds over the time afterwards, too. So to me, in a safety net hospital, having research infrastructure allows for basically you to have a registry to keep up with these folks.

Ashley Ross: And then a question to you and Dr. Gann is, should we be using these tests in this population, genomic tests in this population? I know how I feel about it after your study, but I'd like to get your guys' view.

Adam Murphy: Well, I think as a follow-up point to what I just said, I think when you use this, I believe that providing both the providers and the patient with this probability data, this risk data, will allow the patient to realize that there is some risk to being followed. And I believe that this will allow for them to be more adherent when they know that there is some likelihood of having a bad pathology in their prostate. That's what I believe to be the case. We have to do that and prove that. But I think that's one of the benefits of either MRI or Decipher or an Archetype DX test is for people with low health literacy to understand that there is a baseline risk to this.

Peter Gann: I think first of all, I'm not a pathologist. I'm actually an epidemiologist by training so I tend to think that way. And one of the things that you learn when you're training in epidemiology is the importance of variation across populations. When dealing with a higher-risk population, it's not surprising that you're going to see some very different things than you're going to see in an average or low-risk population. And we certainly saw that.

Peter Gann: One of the things we showed in the paper is that if you look at the NCCN risk category at baseline prior to the GPS, and then you look at the NCCN risk category that they may have moved into once the GPS was considered, the movement was up. And 69% of the men who received the GPS actually moved from a low intermediate risk to either unfavorable intermediate, or even high risk. It's 69%. So now, do I think that this test should be used or could be used in this kind of ... Absolutely. I think it has a great utility, but its utility is going to be a little bit different because it may be the net benefit is for men to avoid the mistake of going on AS when they really shouldn't. So to me, that's why you can't say there's always going to be one size fits all with regard to how a test is going to perform.

Ashley Ross: Yeah. And I think those are my take homes. One is encouraging. I know that both of you, I think, are doing studies and that's encouraging at diagnosis. We won't even talk about tonight because we don't have time, but MRI prior to biopsy to get, it's techniques to try to get the best biopsy you can and the most fidelity of your diagnosis. And I agree that you want to do these tests to help guide care, and then you want to deliver that message to the patient.

Adam Murphy: Exactly.

Ashley Ross: So maybe just, if I can get some closing comments from first Dr. Murphy, then you Dr. Gann. Dr. Murphy, what's the main take-home if you want to give one message or two messages to our audience from this paper and work?

Adam Murphy: I think one thing that came out of this for me as a urologist that I think I wanted other urologists to take away was that the tests that you use need to be appropriate for the population. You need to consider the impact of the test on the population that you're serving. So if you use a Decipher score, Oncotype DX test, in an average population, you're going to have a different likelihood of having bad disease than if you are in a high-risk group. And I think that being familiar with your baseline probability makes the test have different meaning.

Ashley Ross: And Dr. Gann, any final thoughts?

Peter Gann: Yeah. I would add to what Adam said. Just one thing that I believe you brought up at the very beginning, which is that a GPS test can be helpful, but you don't necessarily need it to encourage use of active surveillance for men who really should be on active surveillance. If, and this is the lesson that I think that is a take-home here—This was a research study. We had very well-trained, very homogeneously coordinated urologists. As Adam said, we had some fantastic clinical coordinators working with these patients. So, what we ought to be thinking about is, how can we provide that kind of support both for urologists and patients in everyday care? And if we can do that, we're going to have better outcomes.

Ashley Ross: Well, I think I couldn't agree more, and I think Dr. Murphy was chiming in that he couldn't agree more. Again, congratulations, it's great work. As you both mentioned, it's just a springing-off point, but it teaches us a lot about some of the tools we use for making decisions in surveillance, but moreover, how we communicate with our patients and how we share decision-making that affects their lives and their health for prostate cancer.

Ashley Ross: So thanks again Dr. Gann and Dr. Murphy.

Peter Gann: Thank you.

Adam Murphy: Thank you Dr. Ross for having us.

email news signup