PSMA Imaging - Is Conventional Imaging Obsolete - Steven Rowe

December 26, 2021

In this Society of Nuclear Medicine and Molecular Imaging- Special Session during the 2021 LUGPA CME presentations, Dr Steven Rowe discusses a number of different scenarios that we might encounter clinically where conventional imaging is clearly being replaced by PSMA based imaging. In the last part of his talk, Dr. Rowe focuses on some future directions where PSMA PET may have some interesting things still to teach us that we haven't quite uncovered yet.


Steven Patrick Rowe, MD, Ph.D., Associate Professor of Radiology and Radiological Science, Johns Hopkins Medicine

Read the Full Video Transcript

Steven Rowe: Thanks very much for the introduction. I look forward to this discussion. I wish I could be physically present with everyone, but still happy to be joining you from Baltimore.

My topic today is PSMA Imaging: Is Conventional Imaging Obsolete? Probably. All right. That's really all I have to say, thanks for your attention and happy to take any questions. Of course, I'm kidding.

I'll be happy to discuss a number of different scenarios that we might encounter clinically where conventional imaging is clearly being supplanted by PSMA based imaging. And then in the last third of the talk or so I'll focus on some future directions where PSMA PET may have some interesting things still to teach us that we haven't quite uncovered yet.

I do have a few relevant disclosures. The most relevant of which is that I receive research funding and have a consulting agreement with Progenics Pharmaceuticals, a wholly-owned subsidiary of Lantheus. And Lantheus markets PYLARIFY, which is the nationwide approved and available PSMA agent. Much of the data I will present to you was obtained with PYLARIFY, but before it was called that, when it was still more generally known as DCFPyL or PyL, and I may call it that at times, just out of habit.

Starting first with Prostate-Specific Membrane Antigen, and why it's a good target for molecular imaging and prostate cancer. It's a transmembrane carboxypeptidase, and it is highly expressed in prostate cancer cells.  At the tumoral level, expression is probably present in more than 95% of prostate cancer tumors. Although there's more heterogeneity than that at the cellular level.

And at the histopathologic level, there is a direct correlation between expression levels and tumor aggressiveness. Although I don't know that we necessarily see that manifest on scans because there are so many other factors that go into the amount of radiotracer that would localize to a given tumor.

Being a transmembrane protein that has an extracellular domain, and then the extracellular domain can be leveraged for imaging and therapy with a variety of small molecules and antibody-based radiotracers and therapeutic agents.

What we know about PSMA PET so far is that it has a moderate sensitivity, but very high specificity for pre-operative nodal staging. It has a high detection efficiency for sites of biochemical recurrence and is effective for guiding focal therapy for oligometastases and also for selecting patients for PSMA targeted endoradiotherapy. I won't really talk about that last topic, but I will go over some of the data behind these other indications for PSMA PET and how PSMA PET compares to conventional imaging in these contexts.

If we first look at the preoperative staging of patients with high-risk disease, there were a number of single-center studies that were done with Gallium PSMA, an agent that is commonly now known as Gallium PSMA- 11, and was variably known as the Heidelberg compound and other things over the years. And at the meta-analytic level, these single-center studies found a sensitivity that was pretty consistently about 60% of specificity that approached a hundred percent.

Let me show you some of our data with PYLARIFY again and before it was known as PYLARIFY when it was at DCFPyL. This was a 25 patient prospective study, and all patients underwent PET scanning and then had a radical prostatectomy with an extended pelvic lymph node dissection.

Most patients looked like the patient here on the left, where there is a high uptake within the prostate itself, but no evidence of disease outside of the prostate. Occasionally, patients would look like the patient on the right, where they would have unsuspected systemic disease, and in this case, tiny lymph nodes including M1A lymph nodes as high as a left supraclavicular fossa. But these were all so small that they weren't callable on conventional imaging. And then there are patients that were in between that had pelvic lymph node involvement but did not have systemic disease.

And we found that at the patient level there was about a 70% sensitivity and a 90% specificity for pelvic lymph node involvement. So the sensitivity here may not look that great. It's not a magic bullet. Certainly, the staging for patients with high-risk disease is still predominantly a surgical and pathological gold standard and not an imaging gold standard. But in order to qualify for this study, these patients all had to have an abdomen and pelvic CT or a pelvic MRI that showed no evidence of nodal involvement. So in this particular study, the sensitivity of conventional imaging was 0%. Relative to 0%, 70% is pretty good. And I think clinically relevant in this particular situation.

Now, if we take one more step sort of down the journey that many men face as they go through prostate cancer, and we look at patients with biochemical recurrence, again, there are a lot of early single-center studies in the literature done with Gallium PSMA-11. And I think many of these studies overstated the detection efficiency or sensitivity that we could expect with PSMA PET. And that was because many of these studies weren't necessarily limited to just patients with biochemical recurrence. Many of them included patients that were overtly metastatic. So they not only had rising PSAs or detectable PSAs after surgery, but they also had visible findings on conventional imaging.  So it was possible to have inflated sensitivities by including a lot of patients that had overt metastatic disease. I don't think that was done intentionally, but it did tend to overstate the sensitivity that could be achieved with these PSMA based PET agents.

We did a prospective single-center study with PYLARIFY. It was a 31 patient study. All of these patients had to have either a pelvic MRI or an abdomen and pelvic CT and a bone scan. And all of those conventional imaging modalities had to be negative. So we were going into this study with a 0% sensitivity for sites of disease based on conventional imaging. As you might suspect, the median PSA in this patient population was low, it was 0.4, and it ranged from 0.2 to a little over 20.

We found that in about two-thirds of patients, there was an identifiable site of uptake that was consistent with prostate cancer. As you go to higher PSA levels, the detection efficiency of the scan improves, and the number of patients with more than one site of punitive disease also increased. So none of those are probably surprising findings, but again, two thirds, so perhaps not the 85 or 90% sensitivities that were bandied about in the early days of PSMA imaging, but nonetheless, relative to a 0% sensitivity of conventional imaging, certainly demonstrably superior to what we have been using in the past.

And here's an example of the kind of biochemical recurrence that one can detect with PSMA based PET imaging. This is a small local recurrence. You can see the relatively subtle uptake behind the excreted radiotracer activity in the bladder. And so, it is important for the interpreting radiologists to window appropriately, and to really hunt out what can be relatively subtle local recurrences. The most common finding in this patient population is going to be one or perhaps two lymph nodes in the pelvis, but local recurrences are also very common.

Now, if we move again, sort of another step down the journey into those patients with metastatic disease, a single-center study, again, prospective, nine patients. This was actually the first-in-human study with PYLARIFY. These patients all had rising PSAs and they had to have findings on conventional imaging of progressive disease. So we are not looking for biochemical recurrence here. These are all patients with positive CTS and bone scans. And it was a mix of hormone naive and castration-resistant disease.

As you might expect, the PSAs here are a bit higher. It's quite a wide range, but some of them had very high PSAs. And here's a few examples of the types of lesions that one can detect above and beyond what you might see with conventional imaging. So small lymph nodes down to one to two millimeters and infiltrated bone lesions that are centered in the marrow, and they haven't caused any sort of sclerosis yet. So they do not show up on the bone scans and CT as negative for these lesions.

And not only does PYLARIFY or another PSMA based imaging agent, not only will it detect a lot more lesions in conventional imaging, but it will also detect them more definitively. So while conventional imaging is in this particular patient population, a third of lesions detected on conventional imaging were equivocal. It's relatively uncommon to have equivocal lesions with PSMA PET, more common perhaps than in this patient population with advanced disease. But nonetheless, relatively uncommon, and most lesions are going to be definitive with PSMA imaging. And this restates that same case only with some advanced statistical trickery.

All right. So how about relative to sodium fluoride PET? So sodium fluoride PET, we tend to think of as an incredibly sensitive imaging modality, and it is, but particularly in men with relatively rapidly progressive, aggressive disease. Sometimes they are lesions that wind up being these small marrow-based lesions. They simply do not show up even with the sensitivity of sodium fluoride PET.

Now, as you get to more advanced, heavily pretreated disease or more indolent metastatic disease, sodium fluoride PET and PSMA PET probably become about equivalent for bone lesion detection. But of course, with sodium fluoride PET, you would miss any soft tissue lesions that are otherwise apparent on CT.

And so if we take that to the logical conclusion, it seems to reason that we might be able to sensitively detect metastatic disease that patients can be offered focal therapy, such as SBRT to their sites of metastases and at least achieve the disease-free interval, if not a potential cure in a subset of patients.

Here was a patient with a PSA of 3.9 after prostatectomy. He had a single site of disease visible on CT and also on PSMA PET. He underwent SBRT to that site, no systemic therapy, and as his PSA became undetectable, his scan normalized, and it has remained that way for a couple of years.

At the other end of the spectrum. Here is a patient with what appears to be a single suspicious site on a bone scan. The CT was negative in this patient, but when he underwent PSMA PET, he actually had hundreds of bone lesions that were hidden in the marrow. This is uncommon. We've only seen this a couple of times, but obviously, this patient is not appropriate for focal therapy, to what had been thought to be a relatively limited volume of metastatic disease, and does need systemic therapy.

And then here's a patient that's in the middle of those two patients where there is pelvic and retroperitoneal adenopathy. His PSA of 10.9 at the time he underwent SBRT, but unfortunately, his PSA continued to rise, and on his follow-up scan, he had disseminated relatively subtle bone metastases.  So there's still some work to be done to identify exactly who will benefit from this oligometastatic treatment paradigm. But clearly, there are some patients that if carefully selected can do so.

This was some post-hoc data from the Phuoc Tran's group at our institution. They selected patients who were selected as having oligometastatic disease on the basis of conventional imaging and then underwent treatment with SBRT. And in this post-hoc analysis, it was determined whether the patients had a concordance of all of their lesions on conventional imaging with PSMA PET, or if there were additional lesions visible on PSMA PET that hadn't undergone treatment with SBRT. And those patients who had all of their lesions treated had improved progression-free survival and improved distant metastasis-free survival. So this certainly argues that relative to conventional imaging, we should be using a more sensitive modality like PSMA PET in order to select patients for treatment for oligometastatic prostate cancer.

All right. So let me switch gears a little bit and talk about those things that we do not know about PSMA PET imaging, but things that we might learn that we certainly can't learn from conventional imaging in the same way. So there is something to be said for how prognostic findings are on these scans. Sensitivity and specificity are great, but how are the patients going to do? There's a complex interplay between androgen signaling and PSA expression, and maybe we can leverage that for some prognostication.  And there might be patterns on that that are important for determining how patients are going to do, and what therapies might be appropriate. Artificial intelligence is going to play a huge role in all radiology going forward, and we will briefly mention this in the context of PSMA PET. And then as these scans become more and more common and more and more people are reading them, is it useful to have a structured reporting framework where we can have a common lingo between ourselves and our referring clinicians?  And I will be talking about the same things in the same way.

There is some really nice retrospective data with Gallium PSMA-11 from Australia, where men with intermediate- and high-risk prostate cancer underwent prostatectomy and pelvic lymph node dissection. And those men that were true positive on the scan and then histologically proven, did worse than those men that had false-negative scans that were subsequently found on histology to have nodal disease involvement.

So this suggests that there is value even in a negative scan, even in a falsely negative scan. And I think it tells us that with PSMA PET, those men who have a normal scan or at least no evidence of disease outside of the prostate, are great candidates for surgery.

This is some work from Tom Hope's Group out at UCSF. So they imaged a patient with Gallium PSMA-11, then gave him a couple of weeks of androgen deprivation, and then imaged him again. And all of his lesions were hotter, and even new lesions that appeared on the follow-up scan. This tells us that with an interruption and androgen signaling, we actually get an increase in PSMA expression and early time point imaging might show us more disease following ADT than before. Now it winds up being much more complicated than this. And in subsequent patients, the same group apparently found that there was a mix of some lesions getting hotter, some lesions getting colder, and everything in between.

We looked at a similar situation imaging men with PYLARIFY, giving them two to four months of abiraterone or enzalutamide, and then imaging them again and found that similarly, some lesions got hotter, some lesions got colder, new lesions appeared, some lesions went away. It was really a mixed batch. But if we took a sort of a holistic whole-body tumor level approach, and we developed some metrics that were based on changes and uptake between the baseline and the follow-up scan, that we could start to bend patients into various categories of how their disease changed on the follow-up scan, correlate that to PSA and start to see that there were probably some changes that portended a poor prognosis for the patient.

And we could take those whole-body metrics and associate them with time to therapy change, and even with overall survival in a way that just a baseline total tumor burden didn't have a statistically meaningful association with those two oncologic outcome measures. So there's something about the actual change in uptake on a whole-body level that is telling us about how the patient is going to do with a second-generation antiandrogen agent. We obviously need much more data than this to really make use of it, but it is tantalizing to think that we may have true imaging biomarkers that are available to us with PSMA PET.

Real briefly, we can do something similar with men that are going on bipolar androgen therapy. So if we image them before and after initiation of that, we find we can again, bend them into categories. So their individual lesions may be all over the place, some get hotter, some get colder, but in those men who had definitively new sites of disease appear on the follow-up scan, they subsequently failed on conventional imaging in those same sites. And they progressed in those same sites on conventional imaging. It's just that the PSMA PET detected it earlier than conventional imaging was able to.

Just real briefly on artificial intelligence, everyone in radiology worries about the fact that we're going to be out of jobs once computers get good enough. That won't immediately happen.  But in the near term, we can expect AI to provide us with some really important things. So lesion classification, we can more easily do these whole-body tumor burden assessments, and then that leads to prognostication and decision making. Again, in a way that's hard to imagine, we could do with conventional imaging alone.

Now we can't incorporate conventional imaging into the PSMA PET. And if we do that, algorithms are already at the point where they are probably about as good as radiologists are at just detecting hotspots on the scan. Now, not everything that's hot is going to be cancer. So the exact utility of this may not be immediately apparent. But, we're already again at the point where we can pretty quickly arrive at where the abnormal sites of uptake are on a scan.

A perhaps more nuanced way to think about that is to try to take hotspots on the scan that are detected by a machine learning algorithm, and then classify them according to how likely they are to be prostate cancer or not.  For instance, here's a patient with a bland compression deformity. This wasn't a pathologic compression fracture, but nonetheless, it has uptake. Perhaps a famous example from the literature is various ganglia, dorsal root ganglia, or celiac ganglia. They can look like lymph nodes or bone metastases but are just a type of physiological distribution of a radiotracer.

There are a couple of different read paradigms that have been proposed at this point. PSMA-RADS, I'll touch on briefly, which was our proposed system. There's also the PROMISE system, our molecular imaging TNM staging system. There are advantages and disadvantages to both of these systems. Although I think they're going to converge to some kind of consensus-type statement in hopefully the near future.

PSMA-RADS is a lot like BI-RADS or PI-RADS. It's a five-point Likert scale of the likelihood of cancer in an individual lesion. It has high inner reader reliability and lesions that are equivocal on PSMA PET and categorized as equivocal on PSMA-RADS, which turns out truly are equivocal. Although equivocal lymph nodes will manifest as true positive for cancer about 75% of the time on follow-up imaging, whereas bone lesions, it's only about 20% of the time that those turn out to be true positive for the disease. So still a long way to go with this. But I think if we can all get on the same page, the overall satisfaction with PSMA-based PET can improve because everyone will be speaking the same language and categorizing things in the same way, and providing the same output to their referring providers.

So in conclusion, there are already multiple indications where PSMA-based imaging is, I think definitively superior to conventional imaging and those include primary staging and biochemical recurrence in men at risk of metastatic disease. And we really need larger studies that define what the imaging biomarkers on PSMA PET are and how we can prognosticate for our patients based on PSMA PET. I'd certainly like to thank a large number of people I work with. Here's the subset of them, as well as funding. And hopefully, I've wrapped up in time to answer a few questions. And I certainly thank you for your attention and thank you for the opportunity to speak today.

Gordon Brown: Dr. Rowe, this is Gordon Brown. I had a quick question, as it relates to, you brought up the concept of using imaging as a biomarker, and I think it's a compelling one. And certainly, PSMA PET seems to be able to exclude patients from certain therapies, which wouldn't benefit them in this context. Do you see this technology downstream potentially being utilized as a marker for response in those patients to either intensify or de-intensify therapy in patients with advanced disease?

Dr. Steven Rowe: Yeah. I think that's a great application for PSMA PET and there are probably a few different places where that can help us out. As you alluded to, there are deal-breaker findings that would exclude patients from what might be futile therapies that the patient who has a rising PSA after radical prostatectomy, if we see something outside of the salvage radiation field, it's unlikely the salvage radiation field is going to be of any benefit to that patient. So we have to either change the salvage radiation field or perhaps pinpoint that lesion and treat it with SBRT.

But I think you're also right that at the systemic therapy level, we can figure out who might be having an early response to therapy or who has an early progression to therapy and find those patients and get them on what might be a more appropriate therapy instead of just flailing away for a few more months while they progress through it.

Ben Lowentritt: Thank you very much, Dr. Rowe. Ben Lowentritt. I think a lot of our members and people in the audience are just getting their initial experience with having available PSMA PET in their markets. And one of the really remarkable things the NCCN came out with is this very strong statement that conventional imaging may not be necessary, or should not be considered a prerequisite for PSMA PET. But they didn't change any of their other recommendations for when to get imaging.

I'm curious if you see this as, is this going to change in the staging, especially? Is this going to change when we consider, how we consider doing things that might be standard for certain groups, like when to do a lymph node dissection, should we always be doing PSMA imaging before we consider doing that, and then maybe not doing the dissections, if the imaging is negative? I mean, what are your thoughts on that? And maybe what is the institutional experience so far or advice you are giving?

Dr. Steven Rowe: Yeah, so I think that as good as PSMA PET is, I don't foresee it ever really replacing a surgical approach to nodal staging in patients that are presenting with newly diagnosed disease. I think there is an argument to be made that any patient would be considered for systemic staging, and I guess that would probably be starting [inaudible 00:22:58] as your favorable, intermediate, and on, that the PSMA PET is a great alternative to that, and probably preferred at this point, just because it does have an additive sensitivity advantage over what we've been using in the past.

I think that we don't necessarily have the definitive data yet, but hopefully, we are already at a point where people are exploring the idea that if you see unilateral nodal involvement, maybe that's a sign that the surgeon could be more aggressive on and perhaps spare some of the morbidity on the other side where we don't see anything. But ultimately, unfortunately, I just do not see it having quite the sensitivity to replace a good node dissection and surgical staging.

One thing that's being explored in Europe, and we are still probably a ways away from having access to in the U.S., would be radio-guided surgery, where we use slightly longer-lived agents that will stick in the lymph nodes, and then provide a means for the surgeon to use a gamma probe during surgery to hunt out what might otherwise be subtle lymph nodes. And there are some promising results that say that is actually a better way than even preoperative PET imaging to identify and stage lymph nodes.

So some exciting things are on the horizon, but maybe we're not quite there yet. But I do think that if a patient does have a positive finding on the PET, there is hopefully value in that to the surgeon in terms of their surgical planning and then prognostically, and a negative scan, again, I think really pretends that patients can do really well with surgery. And even if they do have small volume nodal involvement in pathology, they're still going to do really well and potentially be cured or at least have a prolonged disease-free interval after surgery.

Gordon Brown: I guess one additional question, although the results from the [inaudible 00:24:53] oral trial that you presented would argue against it, I guess one question that does come to mind is, is there a lead-time bias, as it relates to the interpretation of some of these results, from an outcomes perspective? I'd be interested to hear your thoughts and comments on that.

Dr. Steven Rowe: I think that's undoubtedly true and something that will be a challenge [inaudible 00:25:12] in coming years. All of our big clinical trials and everything that dictates how therapy has been done is generally been driven by some combination of CT and bone scan and calling progression on those modalities. And so I think it is going to be a lot of work and we still have a lot of work to do to have PSMA PET intelligently replace conventional imaging for some of the decision-making that we do. But it's again, just looking at the numbers, it's undoubtedly a better imaging modality. And so I think it's incumbent upon us to get it incorporated into clinical trials as fast as we can.

Neal Shore:
So great presentation. We've got several questions from the audience, and I'm just going to combine them. So, on May 26th this year, the PYLARIFY got full FDA approval across the country. I think they are in most markets, and I know CMS is reimbursing and many commercial payers are too. The Gallium PSMA PET got approved at UCLA and UCSF about a year ago now. So there are two questions that have been brought up for you, Dr. Rowe. So compare and contrast the 18F-PyL versus the gallium, and then compare and contrast PSMA PET versus axumin, the fluciclovine scan.

Dr. Steven Rowe: All right. No both great questions. So, my general sense about the two different PSMA agents and I think we're probably going to wind up with half a dozen PSMA PET agents approved at some point, is that they are all great, and they are all better than anything that we've had before. And I think that to power a study that would head to head compare them in a prospective way to some oncologically meaningful outcome is something that's going to take hundreds and hundreds of patients, maybe even more, and be so expensive that it might be prohibitive. So I don't know that we are ever going to have this definitive answer that PYLARIFY is better than Gallium PSMA-11 or vice versa.

But the F-18 as a radionuclide has the certain image quality and practical advantages relative to Gallium. It's got a longer half-life and can be made in generally much larger quantities. So centrally producing it and sending it out to sites is a little more practical, than it is with Gallium PSMA-11. That being said, we do have a distribution model for Gallium 68-DOTATATE for neuroendocrine tumors that have been very successful in the U.S. and Europe, and other places. So I think where there's a will, there's a way, and I definitely expect Gallium PSMA-11 to be available nationwide in the not too distant future.

There are some image quality advantages to F-18. And so looking at the scan side by side, the F-18 is a little cleaner, it's a little less noisy. But the vast majority of findings are going to be visible on both. And again, I don't know that we will have that top-level data that really ever differentiate the two. They are both great radiotracers.

Relative to Axumin, the FDA approval is a little bit broader for PSMA. So Axumin is approved only for recurrent disease, whereas PSMA is also approved for initial staging. So that's an advantage that I guess has been poised sort of upon us by the regulators, but nonetheless might be a consideration for a lot of folks in the audience.

I think in the recurrent population, as you get to relatively high PSAs, two and above, the sensitivity or detection efficiency of all of these radiotracers converges and is better than anything we've had a few years ago. Definitely better than our conventional imaging has been.

I think as you get down into really low PSA levels, the 0.2 to 0.5 range, right when men have biochemically recurred, there seems to be a distinct sensitivity advantage for PSMA. And I think that's maybe a little bit of a clinically relevant window for men that are being considered for a lot of salvage treatments with an intent to cure. So I do think that PSMA has an advantage over Axumin in that context.

Neal Shore: So, one last question that's come up, and I'm going to ask this one to Dr. Lowentritt and Dr. Brown. So it's from one of our colleagues who said, "If I'm right in understanding that [inaudible 00:29:42] technician bone scan is being relegated to an acronym and that the best test with a caveat that cost isn't an issue for getting a PSMA PET for my prostate cancer patients, and I already have a CT scan in place. As many large urology group practices do in the United States." So Doctors Brown and Lowentritt, the question is, should we be looking at getting a PSMA PET for our practice?

Ben Lowentritt: Yeah. I think that's an individual assessment in every group and there are different hurdles in different places. I do think that we all have experience with bringing imaging into our groups. It's certainly something that a lot of us are going to explore if it's feasible in your community from other regulatory standpoints. So I guess it's not like PET is going to go anywhere. It's only going to get better.

I would also say though, that this is one where to get good information, you need good equipment and you don't want to go in on the cheap. You would want to go in with a high-quality tool and a good partner who can read the images for you, because of those two things, and arguably, we might be able to control the readers more in the community, which has been a challenge with our previous experience with new PET agents. I do think there is something really attractive there, but you got to have the right equipment and the right people.

Gordon Brown: Yeah. I would agree with Ben. I think it's a capital expense that would ultimately benefit our patients. However, one, that has to be done appropriately with appropriate support and infrastructure. And to me, it's all about the read and the interpretation.