Read the Full Video Transcript

Michael Morris: So the question that I need to answer today is, what are the optimal imaging standards for metastatic hormone-sensitive prostate cancer? And what is recommended in daily practice? These are my disclosures.

This would have been a very easy talk about 10 to 15 years ago because it would be like there is no standard imaging that needs to be done. Just follow the PSA. And when the PSA begins to rise, then you get your first image. And after that, at some point, when the PSA begins to rise as a leading indicator of progression, the lagging indicators are imaging. And then lagging after that are symptoms.

And it's not like this is past history. If you look at the publication date of that figure, it was just actually a couple of years ago. And we see this figure come up over and over and over again. We know what the basis of that figure is. It's that imaging is terrible in this context.

First of all, it's less sensitive to both response and progression relative to PSA. And second, we’re deceived up to 50% of the time by flare phenomenon on bone scintigraphy, which confounds our interpretation, misleads us. And that was really the purpose of prostate cancer working group 2 and 3. It was to shift our attention from response to progression, and progression would be delineated by two or more lesions that were confirmed on a subsequent scan, and we could agree that imaging basically sucked for response assessments.

I hope I said that slowly enough for the non-English speakers. Now let's look at NCCN as it stands today. So a pretty straightforward recommendation of checking PSA every three to six months. And for imaging for symptom onset, that seems pretty obvious. But there's a phrase here that needs to be unpacked a little bit, and that's periodic imaging to monitor for treatment response.

And if you look deep in the footnotes of that phrase, it's basically imaging every 6 to 12 months, something that we should explore. And then at progression, which is biochemical or clinical progression, to get a scan. And NCCN is pretty clear that a PSMA PET could be used in lieu of traditional CT and bone scintigraphy.

So let's look into this a little bit. So we know that upfront now, as we've heard in all of these excellent talks that preceded mine, that we're applying our most potent therapies up front in this castration-sensitive population, our most potent suppressors of AR signaling. And so that should call into question the reliability of PSA as a reporter of progression. And indeed, that's borne out. So let's look at the ARCHES data that Andy Armstrong presented at GU ASCO a year and a half ago.

Remember that this was ENZA and ADT versus placebo and ADT in castration-sensitive disease. And if you look at how patients relapsed, just under 40% of these patients relapsed by radiographic progression only, with no rise in the PSA from the PSA nadir. And just under 70% of patients met a definition of RPFS without the requisite working group 3 definition of PSA progression.

So if you were just following PSA, you would be missing progressive disease in the majority of patients. And the median rise in PSA in that latter group was only 0.77, and that's with Q3 month imaging. Now, we don't have a sensitivity analysis to look at how many patients we would lose if we did imaging at a less frequent rate—6 months, 9 months, 12 months—we don't know what that sweet spot is, but we do know that the implications of missing this are significant.

The patients who progressed by radiographic imaging and not by PSA have an awful prognosis. Half those patients are dead by two years. So we can't really miss this event. And I agree with the investigators from ARCHES that regular imaging needs to be part of assessing response in these patients because so many of these patients are having a clinically relevant event that is not defined by PSA.

So is our savior then PSMA imaging? I'm all for what Oliver said. Let's move ourselves into the future. But it's just not so simple, is it? So let's look at metastatic castration-resistant disease first.

Can PSMA imaging, which should directly image the tumor—and wouldn't it be nice if we could get away from the flare phenomenon and just make it straightforward and simple? And we have a few examples of analyses both for taxanes and for radioligand therapy of the use of PSMA PET. And it does look like if you are a responder, whatever that means on PSMA PET, that your prognosis is probably better.

But we don't have a real identified response parameter that's quantitative, and we don't have an identified clinical outcome measure such as progressive disease or overall survival. And we don't have a clear correlation that we can put down on paper as a quantitative thing and submit it to a regulatory body and say, this is our new response criteria. We do have some proposed very good schema for doing so. We have the PPP, which uses new lesions in order to describe progressions. We have RECIP, which uses the volume of PSMA-avid disease to describe both response and progression.

And these do require further validation. Hopefully, we'll have working group 4 done within the next few months. But it's really in castration-sensitive disease that this is going to be most problematic. And it's hard to say that PSMA is our panacea for response assessments in the future for castration-sensitive disease.

First of all, we have not just flare phenomenon as described by Tom Hope in his publication from about five years ago, but we have PSMA downregulation as described by Louise's paper from 2018. So we have both phenomena happening potentially in the same patient but certainly between patients. And if we look at some quantitative assessments of that flare phenomenon from this publication looking at 25 patients receiving upfront ADT and being followed by PSA, PSMA, and FDG in the first month of their therapy, about 25% of the bone lesions flared by an SUV max of around 50%.

And if you look at the panel on the right, you can see those patients whose PSAs are going down but whose SUV maxes are going up. And if you look at those patients in terms of their correlation with FDG, actually, they're not that FDG-avid. So perhaps these are actually the good-risk patients who are flaring, i.e., who are looking worse. Doesn't that sound like bone scintigraphy to you?

So I don't think that—when you look at the matrix of what response versus progression looks like, this looks like what we were describing before prostate cancer working group 2 and 3. It's really confusing. So we're going to have to understand this, control for it, and accommodate it in whatever schema we ultimately arrive at in terms of response criteria.

Then to make matters more complicated, we have the issue—so that was response assessment; now let's think about progression. Like many solid tumors, including lung cancer, prostate cancer develops lineage plasticity with treatment. And so as patients are progressing from castration-sensitive disease to resistant disease, 15% of those are going to have a lineage switch, which is going to make PSA very unpredictable but also PSMA imaging very unpredictable.

Fortunately, as Mike Hofman pointed out earlier, you have a CT scan to help you out to identify progression. But let's look at molecular imaging with PSMA versus a small cell tracer in prostate cancer. These images were generated by Mark Dunphy at Memorial using a zirconium 89 directed DLL3 targeted tracer. DLL3 is Delta Like Ligand 3, and is expressed in small cell cancers.

Look at that neuroendocrine-treated prostate cancer patient, how effective PSMA is at identifying his extent of disease relative to the DLL3 tracer. It too is pretty bad. So that's a lot of why most consensus criterion organizations are not favoring PSMA at least as a standard right now in terms of response assessments, because we don't really know how to use it on the basis of both response and progression.

So to conclude, first of all, we should get rid of that figure so that we don't keep reusing it. We can't assume that PSA precedes response and progression in castration-sensitive disease. We do need to do serial imaging. And perhaps during one of the coffee breaks, Andy can define for us what the sweet spot is in terms of imaging intervals—3 months versus 6, 9, or 12.

Also, PSMA imaging: We have a lot to work on if we're going to understand using PSMA imaging disease state by disease state and mechanism of action by mechanism of action before it becomes a reliable response or progression indicator. There may be other molecular imaging that's going to be more useful as lineage plasticity emerges. Thank you very much.