Read the Full Video Transcript

Zach Klaassen: Hi, my name is Zach Klaassen. I'm a urological oncologist at the Georgia Cancer Center in Augusta, Georgia. I'm delighted to be joined by Dr. Michael Hofman, a nuclear medicine physician at Peter MacCallum Cancer Center in Melbourne, Australia. Michael, thanks so much for joining us again today.

Michael Hofman: Pleasure to be with you again, Zach.

Zach Klaassen: So, we are going to be discussing some exciting data you presented at ESMO, looking at the prostate cancer working group four and the PSMA PET-CT, sort of serial response evaluation. So, maybe just walk us through some of the background of how this project and this study came to be.

Michael Hofman: Yeah, sure. So, this is PCWG, which stands for Prostate Cancer Clinical Trials Working Group, which many in the clinical trials area will be familiar with. But outside of that, people actually aren't familiar with this. It's a set of consensus guidelines that were developed around imaging endpoints, particularly CT and bone scan. They're used in just about every clinical trial that has led to an FDA-approved product in the last few years. Currently, this uses CT RECIST, typically every 12 weeks or so, combined with a bone scan. And what PCWG two and three did was establish a bone scan as a reproducible biomarker to define radiographic progression-free survival. This is important because prostate cancer often spreads to bone, and you really can't evaluate the bone on the CT scan. However, there's been a shift towards PSMA PET-CT, particularly in places like Australia, where I'm from. But this has now really spread around the world.

In the other half of my job, I report a lot of bone scans, as we do as nuclear medicine doctors, and we know the limitations of bone scans. Firstly, they're not great for patients. It's a four to five-hour procedure. The patient comes, has an injection of the bone scan agent, waits four hours, and then the actual scan for a whole-body bone scan can take half an hour. It's quite a long scan. And then they typically come back on a different day for a CT scan. So, I don't think we necessarily realize how much of an impact this is on patient time. Compare that with a PSMA PET-CT, which incorporates the CT as part of the PET-CT. That can be given with contrast, and this can all be completed really in under 20 minutes, quite readily. Even under 10 minutes on a next-generation PET-CT device.

So, this is a much better workflow for our patients. They come, have a single scan. But even more important than potentially just that patient workflow is the reliability of the scan. And we all struggle with bone scans. It's a little bit nonspecific. You're looking at the bone's reaction to the cancer, rather than the cancer itself. Whereas PSMA images the tumor itself. However, to date, we have no standardized criteria on what defines progression on a PSMA PET versus response. So, the aim of this research is really to design that standardized criteria, so that every person reporting a scan in any center can come up with the same answer.

Zach Klaassen: No, it's a great introduction. I think it's super timely, right? I mean, we've been using PSMA PET scans forever in Australia and certainly in the last decade in the United States and across the rest of the globe. So, tell us about the study design for the data you presented at ESMO, particularly looking at the PRINCE database for this initial evaluation.

Michael Hofman: Yeah, so this was really to start with preliminary work. We've had a lot of meetings over the last two years led by Mike Morris, Howard Scher, and the rest of the Prostate Cancer Working Group committee, to come up with criteria. And that's involved meetings with the FDA as well. So, we came up with a criteria and then we needed a data set. One data set that incorporates a regular PSMA PET-CT was our PRINCE trial. This was the trial of Lutetium PSMA-617, in Australia, combined with Pembrolizumab.

But the actual trial doesn't matter. What matters is that we had serial CT bone scans, PCWG3, and then we added in a PSMA PET-CT. However, this PET scan was just collected, it didn't change patient management. So, it's ideal to compare a new criteria and see how it compares with the CT bone scan.

Just some principles about how we interpreted it. We wanted something simple. This is a visual criteria. For PCWG3, we use something really simple. We say, if there are two new lesions on a bone scan, that may be progression. And if that's seen again on the next scan, then that's confirmed progression. Anyone can do this criteria. For PSMA PET-CT, we have more nuance because, as you know, on PET scan, we've got quantitative biomarkers, SUVs, we can quantify all the tumors. But we thought about this long and hard, and we didn't think these quantitative biomarkers were ready for prime time yet. And we wanted something that a community radiologist, imaging specialist, or even a medical oncologist could look at the scan and say, is this progression or not progression? So, we went with a similar philosophy of PCWG3. Which was that, you should have new lesions or lesions should disappear if you're having a response.

So, we've opened a new category because, with bone scanning, you've just got two categories: progression and no progression. We don't actually define response on a bone scan. However, with PSMA PET-CT, since we're imaging the tumor, we've said, yes, you can define a response. So, if everything disappears, then we can define a complete response. So, we now open a response category for the bone's element component, as well as progression. So, you need to have new lesions or disappearance of lesions. We don't consider changes of intensity in any form, whether it's SUV or visual. That will not define a progression or response.

However, we know that quantitative PET biomarkers will be very valuable. So, we say, if you're doing a clinical trial, you should try and centrally collect all the imaging data so that we can do the quantitative biomarkers, try to come up with some validated methods. But that's really going to be work for the future. And some deep learning algorithms may come around that help us there as well. So, in this trial, the first thing we wanted to do is define reporter agreement. With the new criteria, new lesions or lesions disappearing, can different experts come up with the same answer?

So, in the PRINCE trial, small data set, only 36 patients, but with serial paired PSMA PET-CT bone scans, that's actually 142 paired scans, and then include CT measurements as well for RECIST, and we looked at reporter agreement. And this was substantial. If we considered each category like complete response, partial response, stable disease, progression, with a kappa of 0.7. And if we considered just progression or no progression, more binary, then there was almost perfect agreement between five reviewers, with a kappa of 0.9.

So, this was encouraging. It tells us that PSMA PET-CTs can be reproducible, that we can all follow a new criteria. I hesitate to say, as someone who reports bone scanning as well, when this hopefully takes off in the future, it's going to be more reproducible. We sometimes really struggle with a bone scan. Is this a new lesion? Is it a fracture? It can really be very tricky.

The next thing we looked at was whether the level of response correlated with prostate cancer working group three. And what we found is that, when you throw in the PSMA PET-CT, you generally define progression a little bit earlier. This is not unexpected. And differently from PCWG3, we still required a confirmation at that second time point. So, if your first scan showed, let's say, two new lesions, that did not define progression, we wanted that repeated. But if down the track, you were on your fourth PSMA PET-CT and then you had new lesions, we did not require progression. This is different from PCWG3, which kind of always requires a confirmatory scan.

So, we defined progression earlier and then we correlated it with overall survival, because we've got the OS data after this trial. And both PCWG3, and when we threw in a PSMA PET without the bone scan, also correlated with overall survival. Actually, the correlation was a little bit tighter for the PSMA PET-CT, which was encouraging. But this was not statistically significant, this is a small data set, so this is really preliminary work. But it looks like it was performing at least as good, probably better in my eyes, but we can't say that yet.

Zach Klaassen: Excellent. It's a great breakdown of the data. Thank you for that. So, it sounds like we've got excellent inter-reader variability between the five readers for progression and some interesting data with progression and overall survival, and certainly the response will continue to be developed. So, how do you see each of these, sort of buckets, going forward, in terms of additional work, additional external validation? What are the next steps?

Michael Hofman: Yeah, look, it's worth saying that there are some concerns about putting PSMA PET-CT as a response criteria. And this relates to its increased sensitivity. There is concern that maybe it's too sensitive. Yes, it might give us the correct answer, but if we take patients off... Particularly this is for clinical trials. If we take a patient off the treatment early, just because there's a new lesion on the PET scan, maybe that's too sensitive. So, we are cognizant of that and the committee has continued to meet several times after this work, and we're continuing to refine the criteria. So, Mike Morris is leading that effort and we hope to have a final criteria with a manuscript out soon. This will define a new landscape. But then, we do acknowledge that there is a lot of work needed to validate this. It's one thing to have criteria to show that, yes, five readers can agree with each other, but it's another thing to show that it's robust and that it correlates with overall survival and that we can believe this data to change patient management.

We're also mindful that there aren't too many clinical trial data sets with PSMA PET-CTs done every eight to 12 weeks in a similar fashion. So, what we're planning to do, and in conjunction really with FDA advice, is take data from many trials. It may be a proportion of trials, where a proportion of patients have had regular PSMA PET-CT. If we combine the patient data from five or six clinical trials, so that we have several hundred patients with serial PSMA PET-CTs, then we think we'll be able to validate this.

And this is really repeating work that Mike Morris and others have done with bone scanning. This was done, it was a labor-intensive exercise to validate that bone scan two plus two. We want to do the same with PSMA PET-CT. So, when this criteria comes out, we don't want people acting on it straight away. But we do want to encourage industry and academics, cooperative group trials, to incorporate PSMA PET-CT into the trial so that we can collect the data, perform the validation. And then hopefully, in the not too distant future, we can get rid of the bone scan and have a one-stop shop, which will be a PSMA PET-CT with contrast. And I also want to stress the "with contrast" is needed because we all know that there are PSMA-negative lesions that could be a liver metastasis that's growing on the CT, that's not PSMA-positive. That is clearly progression. So, this criteria, even in this work, we took the PSMA PET findings but instead of the bone scan, we added it to the RECIST findings from the contrast-enhanced CT.

Zach Klaassen: Excellent. Well, you certainly have a global group of experts putting this together. Congratulations on this early work. Maybe just a couple of take-home points for our listeners today.

Michael Hofman: Yeah, I think the take-home points are that agreement for PSMA PET-CT in this preliminary work is very good. It's almost perfect if we consider progression from non-progression. This is a marker that the images have good tumor-to-background contrast. The new criteria detected progression earlier than prostate cancer working group three, but this still correlated well with overall survival. This is an ongoing project. The final criteria should be out shortly. And then, the real work begins when we get all these data sets together from around the world and try to validate this criteria.

Zach Klaassen: Wonderful. Thanks so much, as always, for your time and expertise, Michael. And thank you again for sharing this data with our listeners.

Michael Hofman: My pleasure. Thanks for covering it on UroToday.