Disadvantages of PSMA PET/CT in Advanced Prostate Cancer Imaging Presentation - Ian Davis

October 1, 2019

As a follow-up to Dr. Stefano Fanti’s presentation on the advantages of PSMA PET-CT imaging, Dr. Ian Davis from Melbourne discussed some of the pitfalls and disadvantages of PSMA PET-CT in advanced prostate cancer imaging. 


Professor Ian Davis, MB, BS, Ph.D., FRAC,P FAChPM, is a medical oncologist, cancer immunologist, clinician-scientist, and NHMRC Practitioner Fellow. He is Professor of Medicine, Monash University and Eastern Health, and Head of the Eastern Health Clinical School (EHCS), where he is responsible for all clinical research within Eastern Health through the Monash Eastern Clinical Research Unit (ECRU).


Read the Full Video Transcript

Ian Davis: Thank you very much. Thank you for the opportunity again, and for the invitation. Hard acts to follow this morning (Pritchard, De Bono, Eeles, Fanti), and I'm terrified of those bells, so I'll try not to get there. So, Stephano's given you a great introduction to PSMA PET imaging, and I'm not going to do anything to downplay that.

I will just put up some of my conflicts of interest here. I'm the chair of ANZUP Cancer Trials Group. We're doing a number of imaging and therapeutic studies, and also I want to acknowledge Louise Emmett and Michael Hofman, who've provided some imaging for me to use. I'm a medical oncologist, not an imager, so thank you for that.

So, agree with everything that you said, Stephano. Ga-PSMA PET/CT is fantastic, but I'm here to pour a bit of cold water on it, and put a bit of balance on it. And GA-PSMA PET is not the Messiah. Sometimes it can be just a very naughty boy. I'm going to show you where some of the pitfalls lie in terms of false positives and false negatives. Sometimes we do see true positives but who cares? Or how sensitive a technique do we really need to have to find very, very early-stage disease? Particularly, if it does not alter management, or, even worse, if it leads to inappropriate changes in management. And I'll close with other pitfalls.

And, because I couldn't think of anywhere else to put it, I've just put this other point here, that the CT component of PET/CT is also important, and who's reporting that. Particularly if we're going to move away from what we've been calling conventional imaging, we still need to have some imaging of the soft tissue components aside from those that we can pick up on the PET component. So, the quality of the CT and who's reporting it remains important.

So what about false positives? Well, the issue, as you all know, is that prostate-specific membrane antigen is not specific to the prostate. And we see an uptake in a whole range of non-prostatic tissues. Depending on the clinical scenario, this may or may not be a problem, but if you're working on a patient with disseminated malignancy who happens to have a bit of a high PSA, it can be misleading. Particularly if there's asymmetry. We also see the situation where we can see upregulation of PSMA expression with AR inhibitions. I'll show you some pictures here. This is taken from published papers. So, the issue here when you're seeing uptake in different parts of the body, is that you could misdiagnose these false positive images as metastatic prostate cancer. It doesn't happen often, but it can happen. And it means that you may be missing another critical diagnosis, and that you may be blinkered into treating the prostate cancer, when in fact there's another condition that you should be looking at.

So, at the top right there, we're seeing uptake in salivary glands. Just below that there's uptake in an accessory right parotid. And the bottom left, not highest UV, as you can see, but there is uptake in a left celiac ganglion, and an inexperienced radiologist or nuclear physician might report that as significant. And in the bottom right there you can see uptake of the right pulmonary hilum. What does that mean in the setting of prostate cancer?

Some other images here. At the top left, I've taken a costochondral junction. I'm not going to try and point to the screen, because it might not show up. Recurrent lung cancer, and that would be a clinical scenario, where it might be a little bit less confusing. But bottom left there, rectal adenoma-carcinoma. The larger panel there, top right, is uptake in hepatocellular carcinoma, very high uptake. Next to that marrow uptake in polycythemia rubra vera. And the bottom right is the bone.

Any of us who've worked in multidisciplinary meetings, where PSMA is frequently used, we are seeing uptake in ribs. So, just here, adjacent to the liver we will find this imaging finding very commonly. And Stephano actually showed an image like that as well. Almost always this turns out to be a benign refracture. It's easy if you've got a line of several ribs involved, but very frequently it's one of these. There might not be a history of trauma, and we've biopsied more of these, than I care to admit. And this is a real nightmare for us in our clinic.

Here's another point just to be aware of, and that's the imaging findings might change depending on the context of treatment. So here's a patient, thanks to Louise Emmett for providing these slides, where you can see uptake in the spine there. Probably the central image at the top left is clearest, and the coronal image there where this patient was placed on androgen deprivation therapy. PSA fell, but the SUVmax went up. You can see at the top right, there that there's increasing intensity, and there are more lesions lighting up here. Now, this is obviously disease that was present before and was not being picked up despite the sensitivity of the method. But does not represent progressive disease in this short time, where this patient is also having a PSA response. So, this is in relation to inhibition of AI signaling. You can sometimes see upregulation of PSMA expression, which can be detected on the scan.

What about false negatives? I won't talk about this very much. I think we've heard a fair bit about the sensitivity. This is really not much of an issue in the city of overt metastatic disease. If you already know this patient's got metastatic disease, PSMA PET really offers very little. Stephano did allude to some of the sensitivity. We know, again, from work from several groups, that PSMA PET has low sensitivity when the degree of nodal involvement is below four millimeters. And we know that most node metastasis, particularly in the setting of clinical, the clinically localized disease, are less than five millimeters in size. PSMA PET because of the spatial resolution cannot detect below two millimeters. And between two and five millimeters has about 60% sensitivity. So, if you're looking at this to try to influence your treatment decisions in a setting of localized disease, you need to be aware of the limitations of the technology.

Also up to 10% of prostate cancer do not express PSMA at all, and particularly in the advanced setting. Michael Hofman and others have shown very, very clearly, be very wary of the PSMA negative, FDG positive prostate cancer. These ones do very poorly. So the pitfalls here is that, if you see a false negative, you may actually radically treat patients who actually do have incurable patients. This is the flip side of the advantage, as Stephano had pointed out before. And we may end up giving unnecessary multimodality treatment for prostate cancer that we thought was, in fact, localized, when it is actually metastatic.

Again, I was seeing some of this data in the previous talk. I won't belabor it, this information from three studies here from California, Brisbane and Sydney. Looking at the sensitivity. And sensitivity is very good. And positive predictive value of these series is also very good. What's happening, though, is we're seeing clinical decisions being made on this, and so in this Californian study at the top right there, they proudly proclaimed that if you treat those lesions, you can see a substantial PSA drop in more than 80%. Which is terrific, but we already know that this may be the tip of the iceberg and are we actually making a difference then? Again, from the Brisbane and Sydney series, we're seeing very high specificity. Sensitivity really depends on the size of the lesion, and the negative predictive value is 80%. So, not as high as we would like it to be.

A strategy before that PSMA expression can go up with therapy, can also go down, of course, as you're wiping out have prostate cancer you're going to have less cells there, and say, here's a patient showing a reduction in PSMA uptakes. So, if you're doing the imaging after commencement of treatment, you may be underestimating the degree of cancer that is present.

So, not all prostate cancer is PSMA-avid, thank you to Michael for providing this slide. Here we've got a patient, whose primary is showing no uptake of PSMA on the left there, but has a pyrites five lesion on MRI. And, when the whole man's specimen was taken and looked at in pathology, you can see, shaded in black, are the areas of malignancy. And that area there that was not showing up on scan. We've validated the assay now, as we heard earlier this morning, here's the immunohistochemistry load standing. So, not all prostate cancers are PSMA-avid. What's going to happen to this patient when and if this cancer metastasizes? This was Gleason five plus five prostate cancer. So obviously nasty and aggressive cancer may behave differently.

What about the situation, true positives, but we don't really care. So I've already alluded to the point of having known extensive metastasis. Is there really any additional value? I don't think so, in this situation. What if we know the patient has known likely metastasis, but we're planning local therapy anyway. So, high-risk cancer, high PSA, almost certainly this patient has got metastatic disease, but we haven't yet found it. Does it really matter if we're going to be planning to treat that patient with ADT and radiation anyway? The converse, though, may be more important. And this is often something that comes up in our meetings as well. When you're looking for a reason not to give radical therapy, noncurative therapy therapy. In that situation finding a PSMA disease outside the pelvis may be of value.

Sometimes it can lead to inappropriate changes in management. So I have seen, on several occasions, the patient with PSMA detected metastasis in the setting of high-risk disease with this patient has been treated with ADT alone with no therapy to the primary. We know, from previous data, that that is suboptimal treatment. We're also now seeing the situation where we're extrapolating findings and clinical decision making, based on high or low-risk definitions to PSMA PET findings. So, if patient has low volume or low-risk metastatic disease by CT and bone scan, but apparently high volume disease on PSMA PET, what does that mean? Is that patient going to respond to docetaxel, or not? It might be less important for abiraterone or the AR inhibitors.

We're seeing unnecessary additional investigations. As I said, it influences decisions on trial participation. We discussed this yesterday. So in patients who were, eligibility is related to conventional imaging, but has more extensive disease on PSMA PET. What does that mean for trial participation? I think we should be including those patients but analyzing them separately. And just to flag a little bit of controversy. What about treatment off-study of PSMA-detected... Dammit. Sychnronous oligometastases. I am winding up. This was a slightly scary study. I was an Australian study showing overall there was a 51% change in treatment plan based on PSMA PET. But look at it more deeply here. The biggest change here was in diagnosis of oligometastatic disease, went from 10% to 38%, and the conclusion was these can be treated with SBRT. They can, is that the right thing to do? We don't know the answer, but many of those patients did have that. And here's a patient with disease was found in the nodes, on that left-handed scan. Had surgery, and then, four months later we're seeing more distant disease. So this patient had unnecessary extensive nodal dissection.

So some of the other pitfalls. There are situations that exist with PSMA pet is clearly of value, but sometimes that adds no value to the management plan, particularly if you've already decided what you're going to do. Sometimes the information comes, and it's incomplete. So we're making decisions without the context, without knowledge of the histology, or genomic data, without a parallel FDG PET. In the advanced disease setting that's dangerous. And, sometimes, if the CT component is not of diagnostic quality. And increasingly in Australia, and elsewhere in the world, we're seeing this issue of a new PSMA neurosis taking over from PSA neurosis. So thank you for your attention.