Integrating Genomic Testing with MRI Enhances Stratification of Surveillance Patients - Eric Kim & Mark Sultan
December 6, 2024
Mark Sultan and Eric Kim discuss research on predicting prostate cancer progression during active surveillance using Decipher® genomic classifier scores and MRI characteristics. Their study reveals that while both high Decipher® scores and high-risk MRI findings independently predict progression, these markers are weakly correlated, suggesting they capture distinct risk information. The conversation emphasizes how this data can help personalize surveillance protocols, potentially reducing unnecessary MRI frequency in low-risk patients while identifying those who need closer monitoring. The researchers stress the importance of integrating multiple data points, including genomic testing, imaging, and clinical patterns, to make informed decisions about active surveillance, while maintaining focus on individual patient preferences and circumstances in treatment planning.
Biographies:
Eric Kim, MD, Chief of Urology, University of Nevada, Renown Health, Reno, NV
Mark Sultan, MD, Research Fellow, University of California Irvine, Irvine, CA
Zachary Klaassen, MD, MSc, Urologic Oncologist, Assistant Professor Surgery/Urology at the Medical College of Georgia at Augusta University, Wellstar MCG, Georgia Cancer Center, Augusta, GA
Biographies:
Eric Kim, MD, Chief of Urology, University of Nevada, Renown Health, Reno, NV
Mark Sultan, MD, Research Fellow, University of California Irvine, Irvine, CA
Zachary Klaassen, MD, MSc, Urologic Oncologist, Assistant Professor Surgery/Urology at the Medical College of Georgia at Augusta University, Wellstar MCG, Georgia Cancer Center, Augusta, GA
Related Content:
SUO 2024: Clinical Progression of Prostate Cancer from Active Surveillance is Predicted by Decipher® Genomic Classifier Score on Index Biopsy Independent from Risk Assessment by MRI Characteristics
Investigating the Relationship Between MRI Findings, Genomic Markers, and Gleason Score - Eric Kim
AUA 2024: Decipher® Predicts Clinically Significant Upgrading on Final Radical Prostatectomy Pathology
SUO 2024: Clinical Progression of Prostate Cancer from Active Surveillance is Predicted by Decipher® Genomic Classifier Score on Index Biopsy Independent from Risk Assessment by MRI Characteristics
Investigating the Relationship Between MRI Findings, Genomic Markers, and Gleason Score - Eric Kim
AUA 2024: Decipher® Predicts Clinically Significant Upgrading on Final Radical Prostatectomy Pathology
Read the Full Video Transcript
Zachary Klaassen: Hi, my name is Zach Klaassen. I'm a urologic oncologist at the Georgia Cancer Center in Augusta, Georgia. I'm delighted to be joined for a UroToday discussion looking at data presented at the recent SUO 2024. I'm joined by Dr. Mark Sultan and Dr. Eric Kim. They're going to be discussing clinical progression of prostate cancer from active surveillance predicted by Decipher genomic classifier score on index biopsy, independent from risk assessment by MRI characteristics. Mark, Eric, thanks so much for joining us today on UroToday.
Mark Sultan: Thank you so much. We're happy to be here, and we're grateful to share our research. As you mentioned, the goal of our research was to understand how now, as active surveillance is being used for more aggressive disease than what was traditionally recommended with either very low risk or low risk, we now have some intermediate favorable risk disease that is also being managed with active surveillance. And we wanted to try to understand what clinical variations we can pick up at the start of their active surveillance journey that can help predict how we should continue our surveillance regimen for them, for a more personalized protocol.
So the goal of our project was to understand how, now that we have access to a genomic form of risk stratification for which patients are more likely to progress during their active surveillance management timeline, and how we can integrate that with imaging, which—prostate MRI has become so ubiquitous that it is easily accessible for all of our patients. And how can we easily combine that information to try to understand which patients are going to require a more aggressive surveillance regimen, versus which patients it may be more safe to allow a more personalized approach where we can extend the timeline between repeat MRI.
So we performed a single-center retrospective review of cases that received Decipher genomic classifier testing on biopsy specimens and MRIs existing between 2016 and 2023. Our cohort included 338 patients with a mean PSA of 7.4. We had about one-third of our cohort progress to curative intent treatment over a median follow-up of 26 months. About 50% of our patients did have a PI-RADS 4 or 5 MRI. And 28% of our patients had a genomic classifier score that would be considered intermediate or high risk. Our cohort was an even distribution of Gleason grade group 1 and Gleason grade group 2.
However, there was also, interestingly, about over 10% of patients that did have greater than Gleason grade group 2 disease. What our data showed that was most interesting to me was the fact that we had both a Decipher GC greater than 0.45, as well as a PI-RADS 4 or 5 MRI, were both associated with progression. However, though this was expected, the correlation between the two results was weak, which potentially means that both imaging and genomic information have different forms or different ways of assessing risk that allows us a more comprehensive picture of the patient's clinical pattern.
And so, given that the correlation between high-risk genomic and high-risk imaging was weak, we wanted to understand how these tests performed in the context of one another, meaning how well does a Decipher GC predict the potential for future disease in the situation where you already have an existing high-risk lesion on MRI? What we found was that, in patients who already had a high-risk lesion on MRI, a Decipher GC greater than 0.45 did still add adequate value to our patient's clinical picture in the sense that it was associated with progression.
However, interestingly, when we looked at the data of how well MRI performs in the context of Decipher GC scores, we found that, unlike a Decipher GC, a high-risk lesion on MRI, meaning PI-RADS 4 or 5, was not associated with progression in the setting of an already existing Decipher GC of intermediate risk or greater. Really, to tie things together, what we found was that both a Decipher GC of greater than 0.45, meaning intermediate risk, or an MRI of PI-RADS 4 or 5 category, were both associated with treatment and active surveillance.
Though the correlation between both were weak, given that they were both associated with progression, we can conclude or believe that they capture distinct forms of information that should be used in combination, given the fact that MRI has a very strong negative predictive value but a very poor positive predictive value, based on what the literature shows. And that would allow us to better understand how well patients who already have a high-risk lesion on MRI may do in a surveillance protocol.
So, typically, these results would validate a Decipher GC greater than 0.45 as a predictor of progression, regardless of PI-RADS classification. However, an MRI was not. And so, really, we believe that active surveillance is going to be augmented by including genomic testing to help stratify between patients who are more likely to progress at a sooner interval. That's what our data has shown. Thank you all for your time.
Zachary Klaassen: So, Eric, as we know, there are many versions of active surveillance algorithms. And I know, from the 2018 EAU active surveillance statement, there's almost a recommendation of MRIs up to every two years for people that are on surveillance for many years. So in the context of your data, how may this maybe adjust or decrease the number of MRIs that we're getting in our patients that are on surveillance for a number of years?
Eric Kim: Yeah, for sure, Zach. I think originally, if we look at MRI AS in Europe and probably what—not to shout them out—but what Dr. Klotz is doing in Toronto, the first question, first bridge that I think we've probably already crossed—I don't want to speak for everybody—but a bridge I think many of us have crossed is when and how often do you have to do confirmatory biopsies in these men? But I think you're getting at the next question.
MRIs are obviously expensive in America. They're not going to get cheaper. How do you really rationalize use of that resource that's somewhat limited? Certain systems only have so many 3 T MRIs. Scanner time isn't abundant. How do you choose which patients really need that MRI at a year, versus two years, versus maybe not really even in that time frame?
So, no, I think for someone who's low-volume disease, whether that's grade group 1 or low-volume grade group 2, and if the MRI is negative to start and their PSA is remaining the same and their Decipher low, suddenly, you got a bunch of checkboxes in the low-risk column. Maybe they don't need another MRI unless the PSA changes or something clinically changes that alarms you. And maybe that's the way we should go.
Zachary Klaassen: It makes a lot of sense. I mean, you think about our patients' active surveillance. There's not many times where that MRI changes in two or three years, I mean, if we're looking anecdotally at our data. And I'm sure there's stuff published on it. But I think your point is good. I mean, if you had a negative MRI, the likelihood of that changing, unless there's a huge change in PSA, is pretty low.
Eric Kim: Extremely low. And, obviously, there's those one-offs, where you have a negative MR and aggressive disease. But, again, we shouldn't treat—we should stop treating the zebras. That's what we should get out of our minds.
Zachary Klaassen: So what do you guys think in terms of—we're discussing active surveillance with a patient in the clinic. And we sit there, we have our biopsy. They're low, very low-risk disease. They have a Decipher score greater than 0.45 and less than 0.45. Based on your data, how do those discussions change at this point?
Eric Kim: Well, at least for me, Zach, every patient is—every patient is an individual. Mark made a great point, showing that in our data, a relatively large series of patients that had both tests, Decipher and MRI, are relatively independent. So more data is more data.
At the same time, it's kind of up to the patient. This data to me would say, if your GC is high, Decipher GC is high, and you want to consider a short period of active surveillance, you're actually doing deferred management, which is OK. Maybe you just got remarried. Maybe a year to delay your treatment is actually a big win in the context of your life.
Zachary Klaassen: Yeah, that's true. I mean, and you have a 55-year-old. The GC is 0.6. The conversation may be, we're going to do treatment at some point. We're going to delay it for now, but we're going to keep a close eye on it. Versus the guy that's got a GC of 0.2, 68 years old. You're hoping never to do anything to that patient.
Eric Kim: For sure, yeah, for sure. I think that's absolutely right.
Mark Sultan: I think it also allows providers a bit of confidence in knowing that though you have a high-risk lesion on your MRI, I can safely watch you for a period of time. And the intention to treat doesn't have to be so imminent.
Zachary Klaassen: Yeah, absolutely. No, that's a great point. This is really, really impactful data. I think almost we can take this to the clinic tomorrow and use it. Maybe a couple take-home messages from each of you guys. I'll start with Mark.
Mark Sultan: I think the biggest take home for me, personally, was the fact that both a Decipher GC and an MRI that is high-risk were weakly correlated. That means the same patients who had high-risk genomic testing did not have high-risk imaging. And so, by not performing one of those assessments, you are limiting yourself with the amount of information with which you can truly gauge the risk.
And I think the future of health care is going to be to learn how to integrate such a wide spectrum of information, which includes genomic testing, imaging, clinical pattern, et cetera, in addition to what the patient's preferences are, of course, but to be able to comprehensively get a true, better picture that we are more confident about in terms of how their disease will progress.
Zachary Klaassen: Absolutely. Eric, you have the final word.
Eric Kim: Oh, geez. No, Mark, I think that's very well-stated. I think, again, like we said earlier, more data is more data. And at the end of the day, active surveillance really does come down to a lot of coaching. Zach, we see this all the time. There are patients that you can, I don't want to say steer, but you present the data in a way that really is objective and rational, and they will see the logic.
And so, a young patient who's hell-bent on active surveillance when they have a somewhat concerning MRI and/or low-volume, favorable intermediate-risk disease, Decipher really can be that litmus test telling you, hey, is that a safe, reasonable decision? Is it not a great decision, but we're going to safely do so for the interests of the patient? Or, potentially, should we steer them away from that as a decision?
Zachary Klaassen: Yeah, great points. I think you guys have made several points, including more variables, more data points to discuss. Shared decision-making, always at the forefront of active surveillance, and this certainly helps us. I want to thank you both for your time on UroToday. And congratulations again on this excellent work.
Mark Sultan: Thanks, Zach. Thanks for having us.
Zachary Klaassen: Hi, my name is Zach Klaassen. I'm a urologic oncologist at the Georgia Cancer Center in Augusta, Georgia. I'm delighted to be joined for a UroToday discussion looking at data presented at the recent SUO 2024. I'm joined by Dr. Mark Sultan and Dr. Eric Kim. They're going to be discussing clinical progression of prostate cancer from active surveillance predicted by Decipher genomic classifier score on index biopsy, independent from risk assessment by MRI characteristics. Mark, Eric, thanks so much for joining us today on UroToday.
Mark Sultan: Thank you so much. We're happy to be here, and we're grateful to share our research. As you mentioned, the goal of our research was to understand how now, as active surveillance is being used for more aggressive disease than what was traditionally recommended with either very low risk or low risk, we now have some intermediate favorable risk disease that is also being managed with active surveillance. And we wanted to try to understand what clinical variations we can pick up at the start of their active surveillance journey that can help predict how we should continue our surveillance regimen for them, for a more personalized protocol.
So the goal of our project was to understand how, now that we have access to a genomic form of risk stratification for which patients are more likely to progress during their active surveillance management timeline, and how we can integrate that with imaging, which—prostate MRI has become so ubiquitous that it is easily accessible for all of our patients. And how can we easily combine that information to try to understand which patients are going to require a more aggressive surveillance regimen, versus which patients it may be more safe to allow a more personalized approach where we can extend the timeline between repeat MRI.
So we performed a single-center retrospective review of cases that received Decipher genomic classifier testing on biopsy specimens and MRIs existing between 2016 and 2023. Our cohort included 338 patients with a mean PSA of 7.4. We had about one-third of our cohort progress to curative intent treatment over a median follow-up of 26 months. About 50% of our patients did have a PI-RADS 4 or 5 MRI. And 28% of our patients had a genomic classifier score that would be considered intermediate or high risk. Our cohort was an even distribution of Gleason grade group 1 and Gleason grade group 2.
However, there was also, interestingly, about over 10% of patients that did have greater than Gleason grade group 2 disease. What our data showed that was most interesting to me was the fact that we had both a Decipher GC greater than 0.45, as well as a PI-RADS 4 or 5 MRI, were both associated with progression. However, though this was expected, the correlation between the two results was weak, which potentially means that both imaging and genomic information have different forms or different ways of assessing risk that allows us a more comprehensive picture of the patient's clinical pattern.
And so, given that the correlation between high-risk genomic and high-risk imaging was weak, we wanted to understand how these tests performed in the context of one another, meaning how well does a Decipher GC predict the potential for future disease in the situation where you already have an existing high-risk lesion on MRI? What we found was that, in patients who already had a high-risk lesion on MRI, a Decipher GC greater than 0.45 did still add adequate value to our patient's clinical picture in the sense that it was associated with progression.
However, interestingly, when we looked at the data of how well MRI performs in the context of Decipher GC scores, we found that, unlike a Decipher GC, a high-risk lesion on MRI, meaning PI-RADS 4 or 5, was not associated with progression in the setting of an already existing Decipher GC of intermediate risk or greater. Really, to tie things together, what we found was that both a Decipher GC of greater than 0.45, meaning intermediate risk, or an MRI of PI-RADS 4 or 5 category, were both associated with treatment and active surveillance.
Though the correlation between both were weak, given that they were both associated with progression, we can conclude or believe that they capture distinct forms of information that should be used in combination, given the fact that MRI has a very strong negative predictive value but a very poor positive predictive value, based on what the literature shows. And that would allow us to better understand how well patients who already have a high-risk lesion on MRI may do in a surveillance protocol.
So, typically, these results would validate a Decipher GC greater than 0.45 as a predictor of progression, regardless of PI-RADS classification. However, an MRI was not. And so, really, we believe that active surveillance is going to be augmented by including genomic testing to help stratify between patients who are more likely to progress at a sooner interval. That's what our data has shown. Thank you all for your time.
Zachary Klaassen: So, Eric, as we know, there are many versions of active surveillance algorithms. And I know, from the 2018 EAU active surveillance statement, there's almost a recommendation of MRIs up to every two years for people that are on surveillance for many years. So in the context of your data, how may this maybe adjust or decrease the number of MRIs that we're getting in our patients that are on surveillance for a number of years?
Eric Kim: Yeah, for sure, Zach. I think originally, if we look at MRI AS in Europe and probably what—not to shout them out—but what Dr. Klotz is doing in Toronto, the first question, first bridge that I think we've probably already crossed—I don't want to speak for everybody—but a bridge I think many of us have crossed is when and how often do you have to do confirmatory biopsies in these men? But I think you're getting at the next question.
MRIs are obviously expensive in America. They're not going to get cheaper. How do you really rationalize use of that resource that's somewhat limited? Certain systems only have so many 3 T MRIs. Scanner time isn't abundant. How do you choose which patients really need that MRI at a year, versus two years, versus maybe not really even in that time frame?
So, no, I think for someone who's low-volume disease, whether that's grade group 1 or low-volume grade group 2, and if the MRI is negative to start and their PSA is remaining the same and their Decipher low, suddenly, you got a bunch of checkboxes in the low-risk column. Maybe they don't need another MRI unless the PSA changes or something clinically changes that alarms you. And maybe that's the way we should go.
Zachary Klaassen: It makes a lot of sense. I mean, you think about our patients' active surveillance. There's not many times where that MRI changes in two or three years, I mean, if we're looking anecdotally at our data. And I'm sure there's stuff published on it. But I think your point is good. I mean, if you had a negative MRI, the likelihood of that changing, unless there's a huge change in PSA, is pretty low.
Eric Kim: Extremely low. And, obviously, there's those one-offs, where you have a negative MR and aggressive disease. But, again, we shouldn't treat—we should stop treating the zebras. That's what we should get out of our minds.
Zachary Klaassen: So what do you guys think in terms of—we're discussing active surveillance with a patient in the clinic. And we sit there, we have our biopsy. They're low, very low-risk disease. They have a Decipher score greater than 0.45 and less than 0.45. Based on your data, how do those discussions change at this point?
Eric Kim: Well, at least for me, Zach, every patient is—every patient is an individual. Mark made a great point, showing that in our data, a relatively large series of patients that had both tests, Decipher and MRI, are relatively independent. So more data is more data.
At the same time, it's kind of up to the patient. This data to me would say, if your GC is high, Decipher GC is high, and you want to consider a short period of active surveillance, you're actually doing deferred management, which is OK. Maybe you just got remarried. Maybe a year to delay your treatment is actually a big win in the context of your life.
Zachary Klaassen: Yeah, that's true. I mean, and you have a 55-year-old. The GC is 0.6. The conversation may be, we're going to do treatment at some point. We're going to delay it for now, but we're going to keep a close eye on it. Versus the guy that's got a GC of 0.2, 68 years old. You're hoping never to do anything to that patient.
Eric Kim: For sure, yeah, for sure. I think that's absolutely right.
Mark Sultan: I think it also allows providers a bit of confidence in knowing that though you have a high-risk lesion on your MRI, I can safely watch you for a period of time. And the intention to treat doesn't have to be so imminent.
Zachary Klaassen: Yeah, absolutely. No, that's a great point. This is really, really impactful data. I think almost we can take this to the clinic tomorrow and use it. Maybe a couple take-home messages from each of you guys. I'll start with Mark.
Mark Sultan: I think the biggest take home for me, personally, was the fact that both a Decipher GC and an MRI that is high-risk were weakly correlated. That means the same patients who had high-risk genomic testing did not have high-risk imaging. And so, by not performing one of those assessments, you are limiting yourself with the amount of information with which you can truly gauge the risk.
And I think the future of health care is going to be to learn how to integrate such a wide spectrum of information, which includes genomic testing, imaging, clinical pattern, et cetera, in addition to what the patient's preferences are, of course, but to be able to comprehensively get a true, better picture that we are more confident about in terms of how their disease will progress.
Zachary Klaassen: Absolutely. Eric, you have the final word.
Eric Kim: Oh, geez. No, Mark, I think that's very well-stated. I think, again, like we said earlier, more data is more data. And at the end of the day, active surveillance really does come down to a lot of coaching. Zach, we see this all the time. There are patients that you can, I don't want to say steer, but you present the data in a way that really is objective and rational, and they will see the logic.
And so, a young patient who's hell-bent on active surveillance when they have a somewhat concerning MRI and/or low-volume, favorable intermediate-risk disease, Decipher really can be that litmus test telling you, hey, is that a safe, reasonable decision? Is it not a great decision, but we're going to safely do so for the interests of the patient? Or, potentially, should we steer them away from that as a decision?
Zachary Klaassen: Yeah, great points. I think you guys have made several points, including more variables, more data points to discuss. Shared decision-making, always at the forefront of active surveillance, and this certainly helps us. I want to thank you both for your time on UroToday. And congratulations again on this excellent work.
Mark Sultan: Thanks, Zach. Thanks for having us.