PSMA-Targeted PET Imaging Metastatic Disease, Response Assessment, Future Directions, and Read Paradigms - Steven Rowe

November 5, 2021

In this Society of Nuclear Medicine and Molecular Imaging (SNMMI) Satellite Symposium entitled PSMA-Targeted PET Imaging and Interpretation: What Urologists Need to Know Steven Rowe, MD presents on PSMA-Targeted PET Imaging: Metastatic Disease, Response Assessment, Future Directions, and Read Paradigms. In his talk, he focuses on PSMA PET imaging in the context of metastatic disease, both small volume oligometastatic disease, as well as more widespread metastatic disease, response assessment, and future directions. 

Biographies:

Steven Patrick Rowe, MD, PhD, Associate Professor of Radiology and Radiological Science, Johns Hopkins Medicine


Read the Full Video Transcript

Steven Rowe: A few other things, that may be salient to an audience, primarily composed of urologists. I'm going to focus a little bit more on metastatic disease, both small volume oligometastatic disease, as well as more widespread metastatic disease, and the role of PSMA PET in those contexts. I'm going to talk a little bit about response assessment, go over a few future directions that are starting to evolve out of the current milieu of PSMA PET data, and then, I'm going to wrap up with some read paradigms that have arisen, to try to address some of the pitfalls and aspects of PSMA PET imaging that can be a little bit challenging. And then hopefully, come together, with what looks to be an evolving sort of definitive final read paradigm, that we will hopefully, have in the near future.

I do have a couple of disclosures, as well. The most relevant of which, is the first one, that I have received research funding and served as a consultant to Progenics Pharmaceuticals, which is now a wholly-owned subsidiary of Lantheus, the marketer of DCFPyL.

So this isn't anything that you haven't already heard from Dr. Gorin.  Prostate-specific membrane antigen, again, it's a transmembrane carboxypeptidase, so it's highly expressed on prostate cancer cells. Approximately 95% of prostate cancer tumors, at least in the initial staging setting, will have PSMA expression. The expression on a cell to cell level is a bit more heterogeneous than that. So you will sometimes hear numbers that are quite a bit lower than that, but at the tumor level, it's about 95%, and there does seem to be some correlation between an expression of PSMA and tumor aggressiveness. Although, that's almost more of a histologic finding, and I don't know that we often see that at the level of the scan.

PSMA does have a large extracellular domain, which has been the primary focus for the development of small molecules in more modern antibodies. As Dr. Gorin mentioned, the smaller intracellular domain was the target of ProstaScint, which has fallen out of favor in recent years.

And then, to perhaps, recapitalize some of the things that Dr. Gorin mentioned, and what will be my jumping-off point for the rest of my talk. What do we really know about PSMA PET so far? We know from multiple, both single-center and multi-center trials, that there is a moderate sensitivity, but exceptionally high specificity, for preoperative nodal staging in men with unfavorable, intermediate, and high-risk diseases.

There's a high detection efficiency for sites of punitive disease and biochemical recurrence.

And then, there are some data to suggest that this is an effective means for guiding focal therapy for oligometastatic patients, and also, for selecting patients for endoradiotherapy.

So, let me show you a little bit of data, in terms of where the idea that PSMA imaging is highly sensitive for metastatic disease, and can be effectively leveraged for both oligometastatic and widely metastatic patients.

So in the first study that was ever done with DCFPyL, the first-in-human study that was ever done, with DCFPyL, nine patients with rising PSA, and radiographic evidence on conventional imaging of progressive disease. These were all known metastatic patients, a mix of hormone naive and castrate-resistant disease.

And a wide range of PSA's. These men underwent imaging, and they had contemporaneous conventional imaging to compare these, the findings on the PyL scans. And there are a couple of different types of lesions that really stood out on the DCFPyL scans that were occult on conventional imaging. And across the top row here, are retroperitoneal lymph nodes, very small retroperitoneal lymph nodes, on the order of one to two millimeters. These are obviously, not [inaudible 00:03:50] on CT. They are visible on CT, but no one would think they were pathologic.

And the other type of lesions that the PSMA targeted agent really helped with, were lesions that were infiltrative, or marrow-based, within the bowel. So these are lesions that are not causing a sclerotic reaction. They don't take up MVP, they don't take up sodium fluoride, and they are not visible on CT. They can be visible on MRI, but whole-body MRI is not necessarily used commonly in this context. So here is a way to do whole-body imaging to pick up these very subtle bone lesions.

Altogether, the DCFPyL PET scan at a lesion detection level wiped the floor with both bone scan, diagnostic CT, and the combination of the two. The readers that reviewed these scans, also were much more confident in the PSMA PET findings, than they were in the bone scan, or diagnostic CT findings, where oftentimes, particularly, on bone scan, it was a question of, is this a real lesion or is it degenerative disease?

And even doing some statistical tricks to take into account intra-patient correlations, so sort of clustering effects in patients with a large number of lesions, undoubtedly, the PSMA PET was still demonstrably superior to conventional imaging.

This was also something that Dr. Gorin showed you, but in comparison to a bone-seeking agent, we've been able to identify more sites of disease with a PSMA PET scan. This particular patient was a young patient with a very aggressive, bad phenotype of the disease, he was rapidly progressing at the time that we imaged him. And what that allowed us to do was to find these very, very tagging sites of putative disease in the marrow. So this is again, a lesion that is not going to be causing a sclerotic reaction, or otherwise, going to be apparent on conventional imaging. But we were able to pick it up with the PSMA PET scan.

In going back and looking at [inaudible 00:06:01] has been the uptake of this lesion, and considering the number of PSMA molecules that are expressed on cells in prostate cancer.  We've been able to derive an estimate, that this probably represents about 10 million prostate cancer cells. That may seem like a lot, but with the current state liquid biopsy techniques, such as ctDNA, oftentimes, there need to be 50 million cells or more for the fertile lesion to be sort of producing those liquid biopsy findings. So this really is a very, very small site of disease, when we look at it on the spectrum of cancer lesions.

Now, what do we do with this high sensitivity? Are there implications for patients with metastatic disease? And as we've touched on, there certainly are. So we may be able to take at least, a subset of patients with limited metastatic disease, true oligometastatic patients, identify all their sites of disease, and maybe, even in a small percentage of patients, cure them with focal therapies to those sites of disease.

And then, for patients with more widespread disease, being able to identify the extent of their disease, and that their disease is PSMA expressing, potentially, makes them eligible for treatment with PSMA targeted endoradiotherapy. And we will touch on that a few slides, as well.

Let's first talk about oligometastatic patients, and these examples come from a couple of colleagues in radiation oncology, Dr. Tran and Dr. Radwan. This is a patient that was status post prostatectomy, had a PSA of 3.9 at the time of imaging, and had only one visible site of disease, this presacral or perirectal lymph node, that you can see here, on the axial images. No other sites of disease are present. This patient was treated with SBRT to that solitary site of the disease, with no systemic therapy. His PSA became undetectable, and a couple of years later it was still undetectable.

I don't think anyone has any idea yet, if this patient is a potentially, curative disease, but might be, in the context of having a pelvic limited recurrence, and again, a long time undetectable PSA. Even if he's not cured, this is still a lot of time that this patient has been away from systemic therapy and the toxicities associated with systemic therapy.

Here's a patient at the other end of the spectrum. This patient underwent a bone scan and CT. And there was one lesion, in a rib on bone scan, that was thought to be suspicious. Other findings on bone scans were thought to be traumatic or degenerative. And the CT scan in the bones was actually normal.

That may be hard to believe when I show you this, which is the PSMA PET corresponding image from this patient. So here you can see, there are actually hundreds of lesions. These were all in bone and they were all of these marrow-based infiltrative lesions, so no sclerosis was associated with them. So this patient obviously, not appropriate for treatment on an oligometastatic paradigm, definitively needs systemic therapy, but would have potentially undergone a futile attempt at controlling what was thought to be a limited metastatic disease. Admittedly, this patient also did have a PSA that was over 100. So the PSMA scan is much more in keeping with what we know about this patient biochemically than the bone scan has been.

Here's a patient that is sort of in the middle, so PSA at 10.9, pelvic and retroperitoneal lymph node disease. This patient also underwent treatment with SBRT and had no systemic therapy. He was imaged again, a couple of months later. His PSA had continued to rise, is now up to 19.8, and the patient had disseminated metastases in the bone. These are relatively subtle on the scan. You can see, they are not super high uptake, but they are definitive still, because of the little background with this radiotracer. There does seem to be a pattern of these patients sometimes, failing in bone, when they receive therapy to just lymph node disease, that's visible. And so whether that is something that we can intervene on, or whether we can start to select patients appropriately, with either imaging biomarkers, or other biomarkers, to figure out who is going to have this pattern of failure, remains to be seen, but it is something that people are obviously interested in.

The ORIOLE clinical trial, which was carried out primarily at Johns Hopkins, looked to try to get a handle on how useful treatment of oligometastases was in prostate cancer patients. Now, this was a prospective clinical trial, but the data I'm showing you here, was actually, a post-hoc evaluation of that prospective trial.

So patients underwent treatment to their oligometastatic sites. They were selected, and the treatment was planned, based on conventional imaging, so not based on a PSMA scan. Most of the patients in the study also underwent PSMA scans. And so that was what the post-hoc analysis was. Did the disease that was found on conventional imaging entirely match the disease as it was seen with the PSMA scan? Or whether there are additional sites of disease that were visible with the PSMA scan, and what was found was, that if all the sites of disease that are visible on the PSMA scan were treated, then those patients had better progression-free survival, and better additional distant metastasis-free survival, than patients that had untreated lesions. So this certainly makes an argument, a little bit of a roundabout argument, but an argument nonetheless, that we should be selecting patients with our highest sensitivity agent, which at this current time, is going to be PSMA-based PET imaging.

Let's talk, for just a moment or two, about PSMA-based therapies, because I think, these are things that, even if not administered by urologists, you are likely to have patients that may need access to this treatment. And it is going to be, I think, an important part of the array of systemic therapies that are available for men with prostate cancer, going forward.

PSMA-based therapies have mostly been labeled with a radionuclide, known as, lutetium-177. This is a beta-emitting radionuclide, and you can think of it as analogous to say iodine-131, which is given to thyroid cancer patients. A nice thing about lutetium-177 is, it's less energetic than iodine-131, and so patients require fewer special precautions. They are distinctly radioactive after getting this therapy, and will be for up to probably, a week after getting this therapy, in terms of any meaningful amount of radioactivity. And so there are some precautions that they do need to undertake. But, they do not need to be hospitalized, and it does not need to be done as an inpatient procedure. They are just things that the patients need to be aware of once they are discharged from an outpatient infusion.

How effective is this therapy? Well, if we take a look at a meta-analysis of primarily retrospective studies that were done, mostly in Europe, but also some in Australia, patients that were metastatic castration-resistant at the time of being treated with lutetium-177 PSMA, about 40% of those patients had an objective PSA response, with a drop of at least 50% of their PSA. That is pretty good for a heavily pretreated population. If you look at any PSA drop, it goes up to about 70% of patients having some biochemical benefit from this therapy.

And it comes with the cost of relatively manageable toxicity. So there is high uptake of these agents in the kidneys, and patients at the rate of about 12% to 15%, will have generally transient nephropathy. So they may have some bump in their creatinine or a drop in their GFR.

And then, xerostomia also occurs maybe, in about 20% of patients. This again tends to be grade one or two, self-limited and high-grade xerostomia is highly unusual with this agent. This is in contrast to the actinium labeled PSMA agents, which do have.  They are more effective, actinium is an alpha emitter, and it is as hopped up as you can get with endoradiotherapy, but it does have correspondingly higher toxicity. And in many patients that are treated with actinium, can have a permanent, and sort of life-altering xerostomia. So that may need a little bit more study before it's quite ready for prime time. There are other alpha emitters that are being investigated, and it may be likely that one of those is going to take hold of that aspect of the endoradiotherapy market.

But what do we have, in terms of prospective clinical trials? Well, we have a couple now. The one that was carried out partly in the US, and partly at European sites, was the VISION trial. This was with an agent called, Lutetium-177-PSMA-617. And this was recently published in the New England Journal of Medicine.

Patients had better progression-free survival, better overall survival, and better survival to a first symptomatic skeletal event, when treated with PSMA-617, than when treated with what would otherwise be, standard of care.

There is also a nice prospective trial, known as the TheraP trial, that was carried out in Australia, that compared the same agent, Lutetium-PSMA-617, to cabazitaxel, and found that the PSMA-based therapy was both, more effective, and also better tolerated than cabazitaxel. So exactly where this is going to wind up, in terms of the spectrum of patients that are being treated with systemic therapy for prostate cancer, I think, is anyone's guess. It's mostly been in the metastatic castration-resistant setting, but is it eventually going to find its way into castration-sensitive therapies? Will it be a part of other therapies, such as a combination of PSMA-617 with immune checkpoint inhibitor therapy, along those lines? We do not have any idea yet, but those studies will certainly, be carried out.

I do want to talk a little bit about response assessment with PSMA PET imaging. I will say, that for things like the endoradiotherapies, that we just talked about, or other distinctly cytotoxic agents, response assessment with PSMA PET imaging is probably not too difficult. The tumors tend to go away, there is no more abnormal PSMA expression, or at least, the PSMA expression goes down, and the tumors become smaller. And there is evidence of objective response.

However, if we are treating patients with androgen signaling axis-targeted agents, the response assessment becomes much more complicated. Inhibition of the androgen signaling axis leads to increased PSMA expression. So while you may be actually positively impacting the tumor cells, and they may be dying off, they have a transient increase in PSMA expression. So in the short term, response assessment in men that have been started on androgen axis-targeted agents, it is quite challenging with the PSMA PET imaging.

However, within these challenges, there may be some real chance for the development of prognostic imaging biomarkers, based on changes that can be seen in PSMA uptake, when men start these agents. And let me show you a couple of examples, sort of the first-in-human experience, and then a couple of more recent trials, that have been trying to get at some of these ideas.

So the group from UCSF were the first folks to observe in a human being, that if they image the patient, prior to beginning androgen deprivation therapy, and then a couple of weeks after androgen deprivation therapy, the lesions that were visible pretreatment, became hotter, more apparent, and more apparent on the follow-up scan. And there were also a number of additional lesions that were now visualized.

Unfortunately, subsequent patients, it turned out, it wasn't as clean as this, and some lesions would get bigger, some would get smaller, some would get more intense, some would get less intense. So it became a more complicated picture, that was immediately apparent, despite these promising earlier results.

In a two-center, single-arm study of men starting abiraterone or enzalutamide, a total of 18 men underwent this study, patients were imaged with DCFPyL immediately prior to starting Abi or Enza, and then two to four months later. And then they were followed for about four years to look for evidence of progression on conventional imaging, time to therapy change, and overall survival.

As you can see here, the overall changes in PSMA uptake on the PET scans were all over the map. So some lesions did get hotter, some lesions became distinctly more apparent, and even new lesions would appear on the follow-up scan. But other lesions would also show lower uptake, they'd be distinctly smaller, less visually apparent.  And men sort of fell into a few different patterns here. There were men that had a purely mixed response in their lesions, there were some that had a mixed response, but mostly, the lesions were more apparent or a mixed response. And mostly, the lesions were less apparent, and every other imaginable sort of combination in between.

So we developed a couple of metrics, and these are relatively arbitrary metrics, that let us look at whole-body changes in uptake, and to see if those metrics provided us any insight into associations that may be important for figuring out if the patient was responding to therapy, or not.

On a waterfall plot, the one thing that maybe stands out is, that men that had a mixed, but predominantly increased uptake on their PET scan, generally, had an immediately increased PSA, and so they were failing therapy. Whereas, men that had the other patterns of disease, generally, had at least, some biochemical response. Many of them had an objective biochemical response.

If we take a look at these whole-body tumor metrics and changes in tumor uptake, between the baseline scan and the follow-up scan, there was an association with time to therapy change, with generally, men having more uptake on their second scan, having a shorter time to therapy change, than men that had lower uptake on their followup scan.

The same trend held up for overall survival, where there was an association again with men who had larger increases in their uptake on their second PET scan, having shorter overall survival.

Baseline PET scan metrics didn't necessarily have those associations. So, in looking at overall tumor burden, there was no change in time to therapy change, and although, a trend in a similar direction with overall survival, to what we saw, in terms of the changes in whole-body tumor metrics, this did not meet a statistical significance threshold.

A little bit cleaner study, and maybe one that is a little easier to understand, but it still shows the complexities of trying to follow a response assessment with PSMA in men undergoing androgen targeted therapies. This was a study of six men who were starting bipolar androgen therapy. They were imaged with DCFPyl again, right before going on BAT, and then two months after starting therapy.

And these six men, again, small numbers, broken down into two distinct groups. Their existing lesions tended to go all over the place, much like we've seen in other examples. But, there were three men that had distinctly new lesions appear on the second scan. And there were three men that did not have new lesions appear on the second scan. And when we followed those men out with conventional imaging, the three men that had new lesions on their short-term follow-up PET CT, had early objective progression on conventional imaging at six months after the initiation of therapy. Whereas, those men that did not have new lesions on their PSMA PET scan, actually went until nine or 12 months before they had objective progression on conventional imaging. So it does appear, again, with the caveat of small numbers, that PSMA PET does have some prognostic value in this context, and that the appearance of new lesions is an imaging biomarker for a poor response to therapy.

Let me talk about a couple of future directions that will dovetail with some of the things that we've already discussed. So one thing is artificial intelligence. So this is sort of the boogeyman for all of the imaging right now, in terms of, what is AI going to do? Is it going to put radiologists out of business? It's not quite at that point yet, but there are some short-term things we think it can provide; lesion classification, whole-body tumor burden assessments, and prognostication are likely all things that we will be able to leverage AI for in the very near future, and even already can, to some extent.

And then, I will talk a little bit about non-prostate malignancies, as Dr. Gorin alluded to earlier, particularly clear cell renal cell carcinoma.

Right now, where we're at with artificial intelligence, in this case, machine learning, as a sort of subset of that, is that machine learning is very good at picking up hotspots on the scan. It's moderately good at classifying those hotspots. But, one of the problems that we run into is that things that aren't prostate cancer, so say, we are picking up an incidental second malignancy on a scan. That is a relatively rare event, and it's hard to have enough data for a machine learning algorithm necessarily, to really bite onto that, and be able to provide a definitive answer when that rare occurrence does happen.

Now, if we compare this to self-driving cars, self-driving cars that are run with artificial intelligence and machine learning algorithms, they have to be prepared for very rare events. So even though a child chasing a ball onto the street would be a rare event, a self-driving vehicle still has to be able to identify that and deal appropriately with it. So we are going to have to get to that point for this to be a viable way forward for automated scan interpretation, and things like that. And I think we will get there, and there are methods of data augmentation that will probably help that out.

In our group, our goal, with the work we do in AI, is to try to automate lesion classification, and try to get the AI, not only just to find hot spots, and maybe, spin out a putative staging, based on an examination. We really do want the rare time that a patient shows up with lung cancer that is PSMA avid, we want the AI algorithm to figure that out, and not just think that the patient has lung metastatic prostate cancer. We have a long ways to go for that, that this degree of nuances is not particularly easy. And we will probably take the pooling of thousands and thousands of scans across numerous centers to really drive this forward as a viable approach.

Okay. Let me switch gears a little bit, and talk about PSMA expression in non-prostate cancers, with a focus on conventional or clear cell renal cell carcinoma, which, in the metastatic setting, has essentially, universal neovascular expression of PSMA.

This was a small study of a few patients that had metastatic clear cell renal cell carcinoma. And you can see that the PSMA-based radiotracer works like gangbusters in this context. So in particular, on that bottom row of panels, a very small subcutaneous metastasis was visible on the CT, it was even avidly enhancing on the CT, but still very visible, it was not visible on the PET, nice light bulb, bright lesion. And even though PSMA-based radiotracers do not generally cross the blood-brain barrier, once there has been a blood-brain barrier breakdown, the identification of brain metastases is definitely possible.

And relative to FDG, which has sort of a bad reputation in the context of renal cell carcinoma, but is actually, not bad. Nonetheless, PSMA finds more lesions, and the lesions that are found by both PET radiotracers, are definitively hotter with PSMA.

And in case there is any question about the specificity of these findings, this was a manuscript that was written by Dr. Gorin, our first speaker today, and it was quite a Herculean task. So this was a gentleman who was, unfortunately, dying of widespread metastatic clear cell renal cell carcinoma. He has enrolled in hospice actually, at the time he agreed to undergo a PSMA PET scan, and a rapid autopsy. And so, he passed away not long after the PSMA PET scan, and the number of sites of disease that would have otherwise been inaccessible to biopsy, or to resection, were able to be investigated, and essentially, all of them were true positive for sites of disease. So, this one patient has contributed as much as almost anybody, to the field of nonprostate cancer imaging with PSMA.

One thing that is still out there is kind of a tantalizing clue, but we haven't quite honed in on yet, is response assessment, in these neovascularized tumors. Response assessment can be hard, particularly, with all the different agents that are out there now, immunotherapy often being a first-line agent. But many patients are still going to see a neovascular targeted agent, such as tyrosine kinase inhibitors, at some point in their course of therapy.

If PSMA uptake is a readout for neovascular density, then it ought to be predictive of those patients that are going to respond to neovascular targeted therapies, and we should be able to see the very early response. This has been postulated in the literature. There are some clues that this may be true, but not yet, a good study that necessarily, definitively shows us, but hopefully, we will have at some point.

And then, let me talk about one other cancer that may be of interest to some of the folks in the audience, so transitional cell carcinoma, or urothelial carcinoma. Although the histopathology would suggest that this should work, and there are a couple of case reports out there saying that PSMA uptake is actually pretty robust in these lesions. Our experience was that it was substantively lower than [inaudible 00:29:12] FDG. And that histologically, while there was an expression, the amount of expression, just wasn't to the extent that it was all that useful for in vivo imaging.

And then, of course, there are lesions that are not cancer that also have uptake. Here's a compression fracture in an osteoporotic patient, a nice linear uptake. It doesn't have sort of a focality to it, which would suggest that this is a cancerous lesion, but nonetheless, it does have uptake.

And then, something that has been commonly discussed in the literature, and it was one of the first pitfalls to be recognized with these agents, is that peripheral ganglia will often have uptake. Most notably, the coeliac ganglia, which are high in the retroperitoneum, and have a somewhat lymph node morphology to them. Those can be mistaken for false-positive lymph nodes.

And so all of that, I think, comes together and brings us to the point that, for nonprostate cancers, for pitfalls, for all these things, we really need paradigms that allow us to categorize lesions, and make actionable decisions about lesions. And there is a couple of different read paradigms that have been proposed. Our group proposed PSMA-RADS, which is most useful as an individual lesion categorization method.

And then a group from Germany, and from UCLA and UCSF, proposed the PROMISE criteria, which is sort of a molecular imaging TNM staging. And that is a little bit more agnostic to the nature of individual lesions but does attempt to boil everything down into a staging bottom line answer on the scan. There are advantages and disadvantages to both of these, and I will just show you a couple of slides on each one.

So PSMA-RADS is essentially a five-point Likert scale. There are a couple of subcategories, so there is maybe a little bit of complexity built into the system, but overall, a pretty simple system.  And that's reflected in a high inter-reader reproducibility, where even inexperienced readers, who are aware of the PSMA-RADS system, can generally categorize most lesions correctly.

Indeterminate lesions, which are PSMA-RADS three, have a couple of subcategories, unfortunately, they are actually, indeterminant. So if we look at lymph nodes, about 75% of indeterminate lymph nodes will turn out to be true positive on follow-up imaging, but only about 20% of bone lesions, that are deemed to be indeterminate at a baseline imaging, are found to be true positive on follow-up imaging.  With some of the newer radiotracers that are out there that tend to have higher degrees of false-positive bone uptake, that number is going to be even lower.  And there are a few reconstruction tricks we can do to try to make some of the indeterminate lesions a bit more determinant, and we are working on incorporating those into future versions of this paradigm.

Let's talk about PROMISE a little bit. So PROMISE, the big advantage PROMISE has over PSMA-RADS is that it is incredibly detailed, and the authors were very, very diligent about how they thought about anatomy, and what the implications of anatomy were, for a TNM staging system.

There are cutoffs of a PSMA expression score, which is shown here. These are somewhat arbitrary cutoffs and don't necessarily have anything to do with the biology of the tumor. So this is maybe something that may fall out of future versions of PROMISE, but it remains to be seen.

Now, as I said, PROMISE is very, very granular, in terms of the level of anatomic information that the readers are expected to provide as part of the miTNM process. And I think, this is highly, highly useful for say, big data sets. So AI-driven algorithms, big clinical trials, were being very granular about anatomic information and may be very important to some exploratory endpoint. But the disadvantage is that the system is, as a result, necessarily complex. So this is not an easy system to memorize. It's not necessarily an easy system to apply, clinically. It does, for readers that are experienced, have an eye high inner reader reproducibility. But again, it does come with an upfront complexity that can be maybe, a little bit of activation energy to get over.

The hope is that Dr. Gorin, and myself, and others, all can do a sit down over Zoom. And I think we will hash out a system that has the lesion level of classification, nuances of PSMA-RADS, with the anatomic granular information of PROMISE, where PSMA-RADS is used to categorize lesions, to lead to what is ultimate, a miTNM score. And so, we are just getting started with that, but I do think there is some common ground here, that we will be able to take advantage of.

So just a few conclusions; PSMA PET imaging is useful for men with metastatic disease to guide therapy options, response assessment is maybe, still a little bit nebulous for androgen axis-targeted agents, but will improve with AI, and with larger trials, more questions and answers so far about exactly what the role of AI will be, and what role PSMA PET may have in non-prostate cancers, and then hopefully, we will have a combined adaptable read paradigm on the horizon.

I'd like to thank all of my collaborators, my co-speaker, Dr. Gorin, and again, everyone for organizing this, and thank you for your attention.