Credentialing Imaging as an Outcome Measure in Prostate Cancer - Lawrence H. Schwartz and Michael J. Morris
August 15, 2019
Michael J. Morris, MD., is the Clinical Director, Genitourinary Medical Oncology Service & Prostate Cancer Section Head, Division of Solid Tumor Oncology, Memorial Sloan Kettering Cancer Center, New York, NY. Dr. Morris is one of the leaders of the Prostate Cancer Clinical Trials Consortium (PCCTC). This initiative is designed to increase patient access to clinical trials across the country. He is also the chair of the Genitourinary Committee of the National Cancer Institute’s cooperative group Alliance for Clinical Trials in Oncology, which facilitates testing new drugs nationwide.
Lawrence H. Schwartz, MD, is the James Picker Professor and Chairman for the Department of Radiology at Columbia University Medical Center. He also is the Chief of the Radiology Service and Attending Physician at New York-Presbyterian Hospital.
Charles J. Ryan, MD is the B.J. Kennedy Chair in Clinical Medical Oncology at the University of Minnesota and Director of the Division of Hematology, Oncology, and Transplantation.
Charles Ryan: Hello from ASCO 2019. I'm delighted to be joined today by Mike Morris, who's a member at the Genitourinary Oncology Service and head of the prostate cancer program at Memorial Sloan Kettering Cancer Center, and Larry Schwartz, who's a professor and chairman of the Department of Radiology at Columbia.
You two have really done a lot of work around imaging, the issue of credentialing imaging as an outcome measure in prostate cancer. It's really one of the most important issues we face when we do clinical trials, and honestly, I think it's one of the issues people understand the least.
How do we credential images? How do we perform them? How do we interpret them?
So, Mike, you just presented a Phase 3 trial, a very important study done in the cooperative group system. In addition to the clinical results of the trial, there was a very heavy component of determining PFS from imaging. Let's start there and talk about your results and how PFS and overall survival relate from that study.
Mike Morris: Sure. Well, as you've been... A significant component of [inaudible 00:01:04] and participant, there's a big effort in prostate cancer to make imaging a key component of new drug approval in the United States. And, really in close collaboration with the FDA, we have really made an attempt to clinically qualify radiographic progression-free survival, which is defined by Prostate Cancer Working Group 2 and 3, sufficient, as a clinical trials endpoint, to justify the FDA approval of a new drug.
So, we've now conducted three prospective, preplanned studies to look at the relationship between radiographic progression-free survival and overall survival, and that has now involved over 4,000 patients. My presentation yesterday was the third of those studies, which reaffirmed that RPFS for the patient population getting AR-directed therapy as first-line metastatic CRPC therapy, a really tight relationship between radiographic progression-free survival and overall survival.
And what that, we hope, will do is allow the agency ultimately, then, to approve new drugs in this space of that mechanism, on the basis of radiographic progression-free survival, and not wait for overall survival.
Charles Ryan: You pointed out in your discussion yesterday that RPFS is an individual outcome measure for an individual patient-
Mike Morris: Yes.
Charles Ryan: ... and its correlation with that particular patient's overall survival.
Mike Morris: Right.
Charles Ryan: And I think that's an important point to bring up, because that sort of covers both the regulatory issue as well as the practice of medicine issue, out in the community.
Mike Morris: Yeah. I think that that is an important point, in a sense that these associations are tight enough that we can say, with a fair amount of confidence, that a patient who's radiographically progressing at a faster rate likely will die at a faster rate, and conversely, somebody who has a long response to therapy will probably die at a later time.
That seems like it's intuitive, but on the other hand, in our world, what one believes or intuits is different from what one proves, and so now we have really good prospective evidence to prove that point.
Charles Ryan: So, Larry, it's 2019, and I say that because I'm going to come to the bone scan issue. And we've done a lot of work with bone scans as sort of the measure of outcome in prostate cancer. How should they be used in 2019, both in the clinical trial setting and from the doctor in the community treating the patient?
Larry Schwartz: Sure. So, I think that the bone scan today is still a very useful test. Certainly, we have new molecular imaging agents, many of which Mike is pioneering, in the clinical trial space, but right now, the bone scan is what is ubiquitously available, really globally. It is a very important test, and we've certainly been using it for decades.
The challenge has been, especially as RPFS becomes an important endpoint, that the way that that was interpreted was very qualitative. You're used to seeing radiology reports very, well, let's say touchy-feely, sometimes. "A little bit worse, a little bit better." "Can't really tell." You know, the hedge, maybe.
All of that is accurate, but certainly, in a multicenter clinical trial where results have to be compared and results have to be compared historically, we need a better fundamental way of assessing that.
Now, certainly, the field of radiology is becoming more quantitative. Computers will help us. But we also think that the interpretation should be more standardized and quantitative. And what the Prostate Cancer Working Group 2 and 3 really put together was an attempt, and it was a successful... no longer attempt... but was a successful means of doing that. Successful from the fact that it's simplified, but more importantly, successful from the fact that it's both a reproducible measurement and it's a measurement that actually correlates with outcomes.
And the statement that Mike just made, that correlates with outcomes, means that both in the clinical trial setting as well as in the clinical setting, that type of tool can be used.
Charles Ryan: Mm-hmm (affirmative). So, should radiologists and should oncologists in the community be using bone scan worksheets, and should we be looking at lesions, and should we be counting them and guiding our therapy based on the number of lesions and where they are? Is that realistic?
Larry Schwartz: So, I don't know that that's entirely realistic, but I think some of those principles can be applied, and hopefully one day, we will evolve to a more standardized approach such as that. But certainly, the attempts, the successful implementation of a more structured report, the successful implementation of more quantifying results, is really important, even today.
Charles Ryan: One lesson that I think people continue to learn is, as I say, sometimes bone scans can only tell us that what we're doing isn't working. It can't really be used as a measurement of response, and sometimes you talk to people and they're still looking for a response on bone scan and things like that, and so there's a lot of education, even around a very old instrument in the technician bone scan, that still needs to be done.
Great. I'm going to toggle back to Mike because I want to focus on the question of the clinical decision that comes from what Larry just said.
So, we have criteria on progression. They've been published. People can read those papers. Should we be using those criteria for progression that we apply to drug development and clinical trials to guide when we change therapy with enzalutamide or abiraterone?
Mike Morris: I think that's a really great question. So, Working Group 2 and Working Group 3 were designed as clinical trials tools, and the application of those tools, in terms of routine practice... There, you, of course, need to bring in all of the tools that you have available to assess how a patient is doing. How is the patient feeling? Is the patient clinically progressing? What are other biomarkers that we might use in terms of informing how the patient is doing and whether he's responding or not?
The key thing is that what Working Group 2 and 3 taught us is don't rush for therapies that are effective and have a relatively good toxicity profile. Don't rush to pull that patient off of treatment. That still applies, but on an individual basis, you're obviously going to have to take all the characteristics of the patient into consideration.
What I think that is an important point is that what we did do is at least highlight the minimum criteria for what an important change is on a bone scan, and, as we move forward... because I think we all recognize, as Larry pointed out, that bone scans are probably yesterday's technology... is that we need to, for new tracers and new imaging modalities, be equally rigorous in terms of both analytically validating the performance characteristics of those tracers, so that what we understand is noise, that can be separated from signal to inform a patient, and then standardize the inquiry into what's a meaningful change on, let's say, a PSMA scan or a fluciclovine scan, so we understand what we're talking about when we tell a patient this is a meaningful change, and this is an actionable change. We're not there-
Charles Ryan: So, tell us about the work that's going on now, in trying to credential these scans, the PSMA PET scans, et cetera. We hear a lot about them. Everybody wants them, but I think more of us want them than really know what we'll do with them what we get them.
Mike Morris: Yes. I think that's a really, really great point. So, first of all, let's deal with the first question first, and then what do we do with them.
So, the first question is, do we know what is signal and what's noise, for PSMA imaging? And even though there's quite a bit of data, now, in terms of specific contexts of use for PSMA imaging, like pre-op imaging or rising PSA context imaging, we really don't have great prospective validation studies looking at very... precisely comparing images with images, or images to tissue on a large-scale registration level data. But that is actually changing now, so for example, for the F-18 PyL PSMA tracer, we're going to be presenting later this afternoon one of the cohorts of the prospective registration trial that was termed OSPREY. That was a study that looked for analytic validation using tissue as a comparator, microscopy, for the pre-op population, looking at nodal detection. So, those patients got a PSMA image, then they went on to a radical prostatectomy and lymph node dissection, and the tissue was compared.
What we're going to be presenting today is the metastatic cohort, taking metastatic patients and then imaging them and then biopsying them for their bone and their soft tissue metastases and correlating those findings so that we really have the confidence to say that a positive is likely a true positive at a specific rate.
So, there's another study that's being done, as well, specifically for the rising PSA population, called the CONDOR trial. That is accruing well, and so we are analytically validating the performance characteristics of PSMA imaging for the purposes of FDA approval so that we can say, "Look. We understand what is signal and what is noise when we advise a patient."
And most importantly is that if we see a positive on a scan, which is what actually changes the needle on clinical decision-making... because if you have a negative, you're no worse off than if you got a standard bone scan or CT... but it's that positive that makes you perhaps change surgery or change radiation or how you deliver it or if you deliver it, or change what is metastatic progression, or change what therapy you have a patient who has metastatic disease, understanding the real rate of positivity for that.
Charles Ryan: But Larry, that positive that Mike alludes to is a snapshot in time. It's, the patient has a lymph node here that we should target with radiation or surgery or something like that. What do we know and what will we know, or can expect to know, about the change in the PSMA PET or the novel imaging... shouldn't just limit it to PSMA PET... the change in the novel imaging analyses over time, with a given therapy. Will we have to go through that whole credentialing process where we do what we've done over the last 20 years with bone scans, or is it possible that we can sort of redefine what RECIST means based on PSMA PET, something like that?
Larry Schwartz: So, I think we can. I think the advantage is that we have a construct now for which to do that, so it won't take as long. We've learned a lot. But certainly, we will have to go through a certain level of validation to understand what a change is, is it real, is it real from the perspective of the acquisition modality, the acquisition modality's sensitivity and specificity, and then is it real in terms of correlating within outcomes?
So, the good news is that we have the construct, the mechanism, to do that, but we will have to do some of those studies again with these-
Mike Morris: If I could ask Larry, just as an expert radiologist, do you feel confident that you know how to read a PSMA scan? What is a positive? What should the reference organ be? What is an SUV that constitutes positivity? Do you feel like you have the tools available to do that now?
Larry Schwartz: Not entirely, and I think that that is derived from time and that's derived from additional information from some of these studies that are currently being done. I think that we'll be able to understand that much better. If we go back in time, we didn't know that either even for simple FDG. We weren't aware of that.
When you do a change analysis, that's certainly important, but it has to be done in the context that a change that you see is a true clinical change, or correlates with something. So, one doesn't come before the other, but it's really a combination of putting together the clinical abilities of the test and then backing in or moving forward to understanding what's the minimum change that's going to be real.
Charles Ryan: I'm glad you mentioned FDG because there are some advocates for FDG-PET saying we missed a window of opportunity. We should actually be using this in standard practice. What do you think? And it actually is done in many community practices, where people are using it because they don't have PSMA PETs, and maybe they're used to using FDG in other malignancies.
Larry Schwartz: Mm-hmm (affirmative). Well, I think that's just the point, that they're using FDG in other malignancies and one malignancy isn't exactly the same as another, as you know. I think, in this setting, I think FDG actually has some serious limitations. Some of them are technical, related to signal to noise, related to basic tumor metabolism in this setting. You can certainly get good information from it, but we, as you know, we haven't studied it really rigorously in the proper setting to do that.
I think, with some of the newer molecularly targeted agents, it's probably not necessary to do that right now, and I think that we should really just move ahead.
Charles Ryan: So the FDG ship has sailed, so to speak.
Larry Schwartz: I believe it has. That's not a universally held belief, and, again, there's a lot of factors that you have to weigh, including availability, including global availability. So, I don't want to sell it entirely short, because, for some patients, that might just be the option, and it might be better, as a modality, than what we've used.
Charles Ryan: So, how can those of us who treat prostate cancer work better with our radiology colleagues?
Larry Schwartz: So, call Mike.
Charles Ryan: We'll get his number on the screen.
Larry Schwartz: Yes, yes, exactly. No. No, no. I think we could work together by kind of complementary engagement, and the reason I like this space and the clinical trial space so much is this is where you can really learn so much. This was a truly missed opportunity for decades, where we were using standard tools, where we weren't using, necessarily, the right tools, where we weren't testing new approaches at the same time.
Charles Ryan: We were also trying to credential PSA as a surrogate for survival for a long time, and I think that took our eye off the radiology ball, so to speak.
Larry Schwartz: Right.
Mike Morris: I think that we are used to, for example, talking to pathologists about difficult diagnosis, but the reason that it's so important now to have engagement with the radiology community for mutually educating each other about these tools and about prostate cancer, and it's because scans have assumed primacy in clinical decision-making, because we understand that PSA is really not such a great readout for many issues, in terms of whether a patient's responding or benefiting or regressing, but because imaging is becoming such a key component of understanding whether a patient's treatment should be switched, there's no excuse for you and your radiologist to not have an ongoing dialogue about nearly every patient that's getting a scan. To really understand each other's priorities and how you think and what the images actually look like, as opposed to just reading the impression on the report. It's really important to get, if you're going to understand how your patient's reacting to the therapy that you're giving, you need to understand the data behind it.
Larry Schwartz: Right. And once people understand that critical nature of the images, not just as a kind of ancillary and not just a companion piece of the equation, then you back into all the other questions related to imaging, about the technique, about the interpretation forms, about whether you're having a machine do it or a radiologist do it, about the reproducibility of the technique. So, all those add or become even further critical to the interpretation.
Mike Morris: What's also an interesting phenomenon is now, when I pick up the phone with a radiologist to talk about a scan, I am as likely to change my mind, in terms of what my impression is of the image, as I am to say, "Okay, now I understand." Because these discussions do lead to a different sense than when you walked into that discussion of what really is going on.
Charles Ryan: Yeah. A very important lesson.
Larry Schwartz: And then for the radiology community, what's in it? Basically, as these techniques are developed, as they're tested in clinical trials, to one of your first questions, they obviously become, then, part of clinical routine. Now, they may be used slightly differently. We have to work through many of those issues, but that really creates a critical mass for the imaging community.
Mike Morris: Unfortunately, though, what I think that is happening is while the radiology community is working with medical oncology in terms of a more rigorous and prospective data-based assessment of how to read scans and how to understand their clinical import, now the medical oncologists are making decisions without that rigor. So, for example, see a lesion, irradiate the lesion, or treat the lesion, see that it goes away, and decide that nothing further needs to be done.
We haven't necessarily worked together with our radiation community and our surgical community in being equally rigorous in terms of what we do, knowing that we're building a database that actually supports that decision in prospective trials. We're now overreacting to imaging data in the hopes that we're doing good, but not actually working together to say, "Hey, let's make a plan to actually study this and make sure that those treatment approaches based on molecular imaging are actually rational and data-based."
Charles Ryan: Right. Well, thank you. This is a great discussion. Really important points. I want to congratulate you both and thank you for your leadership on this area, in defining the space for us, defining how to think through this, and there's obviously a lot of questions that we'll continue to answer. Maybe next year we'll get together and talk about how you answered them. So, thank you for joining us.
Mike Morris: Thank you for having us.
Charles Ryan: Thank you. Okay.
Larry Schwartz: Thanks a lot.