Read the Full Video Transcript

Phillip H. Kuo: All right. I'd like to thank the organizers for inviting me to come give this talk on how to crush a PSMA PET report—guidelines and scoring systems. Actually, I think I'll be able to catch us up on time, because a lot of various topics here have already been touched upon. Just in way of disclosures, I do consult or speak for a variety of radiopharmaceutical companies.

So this is a very nice proposed standardized template. Many of the authors of this paper are in the audience. And I'd like to start with it, because I want to emphasize that there's a lot of crossover between PSMA PET reporting for clinical trials versus standard clinical care, but there are some differences.

So let's start with this very nice proposed standardized template. As you'll see at the top, it mirrors what you would do in a more classical nuclear medicine or radiology report, where you start with the indication and some history and some technique. And then you get into the findings. And the findings start, like many reports do, with the primary prostate. The prostate bed then goes to lymph nodes, to distant metastatic disease, and other items. But also, as you can see here in this proposed template for clinical reporting, you'll see things like sightedness, SUVs, and even measurements on CT, particularly if you're a site that does diagnostic CTs or MRIs with your PET/CT or PET/MR.

So let's compare that framework for one that was proposed for clinical trials, which has already been discussed here. So I'll go quickly. And this is the PROMISE version 2. And again, you start with the goal with the imaging, which is to get a molecular imaging TNM framework. All right. And also we have these intensity scores, which, as we talked about, can do visually or quantitatively.

Again, starting off with what we did here, similar to the PRIMARY score—initially, of course, for initial staging or for biopsy purposes, you really want to localize the regions for the prostate gland. Not so important when you're talking about metastatic castration-resistant prostate cancer. And again similarly, we're going to talk about nodes and distant metastases. And I want to point out here that when you do the PSMA expression score, you have 0s and 1s, which are benign, and 2s and 3s are thought more eligible. But also for the purposes of assessing for candidacy for radioligand therapy, people will have adopted generally what was used in the VISION eligibility PSMA PET criteria, which was greater than liver by purely visual assessment.

OK, this is the template from the PROMISE version 2. As you can see, it's very similar highlights to what was seen in the proposed clinical standard templates, where you'll, of course, deal with the lesions in the prostate, then lymph nodes, then visceral metastases, broken down by bone and visceral metastatic categories.

All right. So next step, PSMA PET use and reporting—context matters. It's very, very hard, and we should probably not try to do something that's one-size-fits-all—it can get you into trouble. And what this slide is supposed to show you from this article is, focusing on the left-hand side, all the different various indications, from diagnosis, staging, biochemical recurrence, metastatic hormone sensitive, metastatic castration-resistant prostate cancer, where you're selecting patients for different types of therapy, specifically PSMA-targeted therapy. You could actually tailor your report to each one of these different uses to really get at answering the question, which comes from the context.

So we've talked about the big picture, the different templates, and how they hit the different categories. And let's not forget, when you're looking at an individual lesion, it comes down to your confidence in the individual lesion. Sometimes that matters, sometimes that doesn't. Particularly if you're doing targeted radiation therapy, your confidence in a specific lesion can matter. And this has already been brought up. It started with BI-RADS, breast imaging and mammography, and it's now gone into TI-RADS for the thyroid, and now it's PSMA-RADS, where you have increasing scores with your level of suspicion. So you'll start with BI-RADS 1 and 2, which is thought to be benign, 3 is unclear, then 4 or 5 are high confidence of metastatic disease.

All right. Let's go through an actual case. This is a patient. First thing you want to know is the context. This patient was being evaluated for lutetium PSMA radiopharmaceutical therapy—radioligand therapy. And this is the maximum intensity projection images. On your left-hand side is the anterior projection. On the right-hand side is your lateral view.

All right. So if you're doing this for lutetium PSMA RPT, first thing you want to do, of course, is assess if you have lesions that are PSMA-positive, visually greater than liver—no need to measure any SUVs. As you can see, you have a multitude of lesions that are greater than liver. You also, of course, need to exclude that there is any significant PSMA-negative lesion. And let's just take for granted that we did that. And let's look at some slices.

All right. So there's very intense uptake here in the primary prostate. There's seminal vesicle extension—not so important now that it's mCRPC but would have been more relevant early in initial staging. You also see something very important here, which is spread to the anterior rectum, which could be important for radiation therapy planning. But also you notice that the disease is quite PSMA-positive, so if the patient were symptomatic from that local disease, they may benefit from lutetium RLT.

There is a positive right external iliac node here. And I make a relevant point here that this left focus is actually the ureter—it's not a lymph node. This is the lymph node on this side. I don't know if people can actually see the arrows. Sorry. And then numerous PSMA-positive bone metastases—one axial slice here, but you can see on the MIP there's many.

All right. So you can imagine filling out the table or going through your standard dictation methodology and delineating all that disease. But I'm sure there's more than a few imagers here that picked up on something, because a lot of us will train our trainees to look at the kidneys—whether it's a bone scan, if you know that we do them anymore, and the PSMA PET scan. And look at the kidneys. There, you actually see you can get blooming artifact, but it looks like you have hydronephrosis on the left and hydroureter. And on the right side, it’s even more alarming that you don't see any radioactive urine

And so this was a prior PET for comparison. But as you can see, this patient has new hydronephrosis and hydroureter, and it's actually particularly scary on the right side, which—that is not a lymph node, that is the ureter. So massively dilated with no radioactive urine, which means a high-grade obstruction.

So barriers—time, time, time. I got to go back over a PET list that's over probably 200 long by the time I get back. Whether you're radiology or nuclear medicine, time is an incredibly valuable thing to you. I think in the future, we got to incorporate these into standard dictation systems to make this easy. Artificial intelligence really probably can help with this identification, disease segmentation, quantitative analysis, but also hopefully auto-filling these CRFs or these standardized templates—generate findings, generate impression points. And just—I’m a scientist, so a picture is worth a thousand words.

So what you're trying to do is essentially get all that information from that PET/CT or PET/MR and put it into all the data points in a form. And nuclear medicine folks and radiologists, we like to use the term innumerable or too numerous to count. When I was an internist, we had a running joke: how high can a radiologist count to when they say “too numerous to count”? But now I'm on the dark side.

So a picture is worth a thousand words. I started counting up all the number of boxes and fields here, and I stopped counting at over 200. So if you have a complex patient, you'll spend most of your morning just doing one of these. And that's just not something we can do yet. But hopefully we can overcome some of these barriers and really allow for a standardized template which, when adopted, will make life easier for everyone.

So in quick conclusion, there's overlap and differences between trials and standard care for reporting. Context matters for both. Standardized templates are important tools, but should not supplant thinking. And don't lose the forest through the trees, because if you start going—first thing you do is start to fill up one of those 200 boxes—you can miss some important non-oncologic findings that are actually critical to patient care. Thank you very much.