Read the Full Video Transcript

Louise Emmett: We will get right in. I want to say thank you so much for inviting me to talk today. It's such an honor to be here. I'm not going to talk about something quite so clinically relevant. Perhaps, it will become so in the future. I really want to talk about improving prostate cancer diagnosis with PSMA and using the PRIMARY trials. Here we go. I'm going to talk about the series of PRIMARY trials that we've been doing. These are my disclosures.

In prostate cancer diagnosis, at the moment what we really have is evidence for MRI and MRI guided treatment with these two great trials, PRECISION and PROMISE. MRI guided treatment both improved diagnostic accuracy and reduced the number of required biopsies for prostate cancer diagnosis compared to just biopsy. But there are questions that remain about the safety of not biopsying clinically high risk PRIMARY2's. We know in the PROMISE trial, about 25% of patients who were PI-RADs 2 had clinically significant malignancy on biopsy. And with the PRECISION trial, many of the biopsies undertaken are still negative even though they're not biopsying PI-RADs 2.

So there's a question, is there a way that we can actually improve or safely reduce biopsy? So that in the end for prostate cancer diagnosis we're only biopsying those patients who have clinically significant malignancy, essentially for staging purposes. Certainly, looks like PSMA can be used for that purpose. And in fact, it was Matthias Eiber in about 2015, published a really nice PET MRI paper showing that it could differentiate between pattern three and pattern four disease. And that really started the whole process of thought for me with the PRIMARY trial. So the PRIMARY trial is a prospective phase two trial that we undertook in Australia. Multi-site trial, 296 men who were having a multi-parametric MRI PI-RADs two to five, and who were also going for a trans perineal template biopsy. Trial was really simple. We really just slotted in a pelvic only PSMA PET in between these two procedures.

The question from the PRIMARY trial was can PSMA PET combined with MRI better diagnose clinically significant malignancy? The PRIMARY endpoint for the trial was negative predictive value. And what the study showed was PSMA improved, or the addition of PSMA to MRI improved negative predictive value from 72% with MRI alone, up to 91% with MRI and PSMA. Similarly, sensitivity improved as well from 83% up to 97%, if you use the two modalities together. The other thing that was really interesting was intensity is predicted of gray group. So the more intense the primary lesion on the PSMA pit, the higher the gray group that you saw on histopathology. So one of the questions that the PRIMARY trials, can we actually safely further reduce biopsy over an MRI guided paradigm? And in PRIMARY we had a particularly high risk group of PI-RADs, 2 and 3 because they were going for transperineal template biopsy anyway.

So 28% of patients with PI-RADs 2 in the PRIMARY trial had clinically significant malignancy. And 47% of patients with PI-RADs 3. Of those 90% were detected by PSMA PET. So if you look at it, if you take patients who are PSMA negative with PI-RADs 2, 3 in the PRIMARY trial, about 38% of those patients probably didn't need to be biopsied, or 19% of the total population. And then can we actually de-intensify biopsy in those patients who we know definitely have clinically significant malignancy? Can we be more certain about whether a patient has cancer on imaging alone? Once again, I think the answer is probably yes. If you had an SUVmax of greater than nine on the PSMA PET in the PRIMARY trial with PIREDs 4 or 5, all of those patients had clinically significant malignancy. And if you had an SUVmax greater than 12 independent of what was found on the MRI, all of those patients had clinically significant malignancy.

But one of the problems at the PRIMARY trial is that we used based on an interim analysis and an ROC curve analysis, an SUVmax of four to define clinically, or positive, or a negative PSMA PET with a clinical read. But as we've spoken about this morning, we use multiple different tracer agents. We have multiple different cameras that we image on. And it's really not feasible to use an SUVmax of four. So one of the questions fairly early on in the PRIMARY trial, is there a better way to report PSMA PET? And we know from PIREDs and also from histopathology studies, that in fact about 76% of all malignancies arise within the peripheral zone, yet most benign changes happen within the transition zone. So the question was, could we use that information within the reporting of PSMA PET to improve what we are doing?

So we looked at a number of things. We looked at diffuse versus focal pattern. We looked at anatomical site, transition versus peripheral. And we looked at intensity. And then we modeled that within the PRIMARY trial mathematically to come up with what we call the PRIMARY score. And Courtney I know has gone through this a little bit before, but basically the PRIMARY score is a five-point scoring system. And we used different patterns, anatomic site, and intensity to define whether we thought it was benign or malignant. So PRIMARY score one is no significant pattern within the prostate. PRIMARY score two, what we did was differentiate transition and peripheral zone. And the way we do that is by overlaying. So we always look at the fused image. And if you look at the fused image, you can see that the transition zone and the peripheral zone, you can see a halo around this activity.

So a peripheral halo with no activity in it is the peripheral zone for us. And PRIMARY score two is diffuse activity within the transition zone. PRIMARY score two is also this activity that you see in the central zones around the ejaculatory ducts. Not involving the posterior margin of the prostate, so not involving the peripheral zone. About 15% of malignancies actually arise within the transition zone. And so focal activity well above background counts in the transition zone counts as PRIMARY score three and is considered a malignant pattern. PRIMARY score four is any activity that is focal within the peripheral zone. The peripheral zone including all of the apex. And PRIMARY score five is in SUVmax greater than 12, independent of where it is in the prostate. And as you can see here, there's a nice gradation of the incidence of clinically significant malignancy as you go from score one to five.
And this is what it looks like in terms of score one to five and grade group. And one of the nice things about the score is about 69% of the patients who have clinically significant malignancy within the score five are actually grade group three or higher. So there's a nice gradation not only of whether or not you have clinically significant malignancy, but the grade group of that malignancy goes up as well. So you can't just implement into clinical practice as Tom was saying, a clinical scoring system or an analysis on imaging without having validated it in a separate imaging procedure, or data set. So this is a retrospective study that's just been published of 227 men who had MPMRI and PSMA PET prior to transperineal template biopsy. Gallium PSMA PET was undertaken in all cases and each scan was double read with a third reader for discordance.

So we had six readers for the PSMA PET. And six readers who were expert in multi-parametric prostate MRI. The PSMA PET readers each had a single one-hour education session explaining what the PRIMARY score was with a few cases. So when we look at the Kappa score between MRI and the PRIMARY score, the PIRADs had a Kappa score of 0.55. And PRIMARY score had a Kappa score of 0.7. So substantial agreement between the six readers. Accuracy was very similar between PSMA and MRI in this patient cohort. And looking at sensitivity between and specificity between the PRIMARY score, PIRAD score, and the combination 86% and 76% sensitivity and specificity for PRIMARY. 89%, 74% for PIRADs. And once again, just like with the first PRIMARY study, we found an increase in sensitivity with the combination of up to 84% with a specificity of 68%. And an increase again, in negative predictive value if you use the two together.

We did have one case that was negative by both essential readers, which is this case here. And you can see if you were going on PRIMARY score criteria, this is focal activity, it's confined to the apex, technically it's a PRIMARY score four. But these readers hadn't really used the PRIMARY score much before. It was a one-hour reading session. So all readers were fairly new to it and it definitely takes some education. So the question is, when we're reading intra-prostatically, how should we read? Should we read the PSMA PET and the MRI together? Should we put PIRADs and PRIMARY together as a score? Is that going to be better? We've definitely shown the accuracy is higher. Should we read PRIMARY score alone? We always have an MRI available and how should we be reporting? So we have also modeled within the PRIMARY study a composite score of using PIRADs and PRIMARY together and then validated that within the other score system.

So if you do develop a score that uses both PIRADs and PRIMARY, even if they're read separately, you get really nice stratification in definitely have malignancy and definitely don't have malignancy. So looking at this composite score with PIRADs and PRIMARY together, if you had a PRIMARY score five on the composite score, 93% of those patients had ISUP three or higher. And all patients had clinically significant malignancy. If you have a score of four 97.5% of those patients have clinically significant malignancy. And if you have one or two, almost all of those patients do not have clinically significant malignancy. There was one miss, one patient who did have clinically significant malignancy at a score one or two, and that was the patient that I show you previously with the PRIMARY score four that was missed by the two readers. When we look at the area under the curve for using a composite of PRIMARY and PIRADs, we get very high AUC of 0.928 compared to 0.836 with the PRIMARY score alone. And clinically significantly better or statistically compared to PIRADs as well.

Even more interesting is when you look at the sensitivity and specificity for ISUP three malignancy, which is the clinically significant malignancy considered in most MRI studies, sensitivity was 99% with a negative predictive value of 98%. And it was just that one patient that was missed in this study. So what's the cost-effectiveness of using PSMA PET intra-prostatically prior to prostate biopsy? I know it's certainly not done here and it's not done a lot in Australia yet. So the PRIMARY2 study is a randomized phase three study that we're undertaking across Australia looking at the additive diagnostic value of PSMA PET in men with negative or equivocal MRI.

So in 660 men, patients with negative or equivocal MRI who have clinical risk factors such that they're going for transperineal template biopsy, get randomized to either standard of care, which in Australia is a transperineal template biopsy, or they get randomized to have a pelvic-only PSMA PET. If the PSMA PET is positive, they get a targeted MRI PSMA biopsy only. So no template biopsy. And if it's negative, the patient does not get biopsied. We just hit 50% target for randomization on the study yesterday. So we are 330 patients in, hopefully we will be finished in the next 12 months. It's a co-primary endpoint of the study, which is looking at detection of clinically significant malignancy. And also, the percentage of patients who avoid biopsy because of the addition of the PSMA PET. Also, really focusing heavily on health economics, cost-effectiveness, and patient-reported outcomes in this study. So PSMA PET in prostate cancer diagnosis, I believe combination PSMA MRI is more accurate than either modality and isolation. Particularly, in high-risk populations who have clinical risk factors for a malignancy.

We need to prove the addition of PSMA to MRI's cost-effectiveness and impactful to patient outcomes and the PRIMARY2 study should hopefully achieve that. I do think there's a potential role in biopsy avoidance or minimization, or perhaps changing the way we biopsy in men who definitely have malignancy based on imaging. Do we need to do 30 cores? Can we just do a single core? Do we need to do any cores at all? And that needs further work. And I think this will have a lot of value in identifying patients for focal therapy and also for active surveillance in the future. Thank you.