Read the Full Video Transcript

Michael Morris: What a great whirlwind tour. Thanks, Matt. All right. So, I'm going to be talking about VISION and TheraP, and obviously, everybody here has been really involved in these two trials and couldn't have been done without everyone's hard work. I only have seven minutes per trial, so don't feel offended anybody if we cruise right through here. All right. First of all, VISION. Just to recapitulate a little bit of what Matt was just talking about, this was the registration trial for lutetium PSMA-617 patient population where those patients who had both progressed through an ARPI at least once and had progressed through at least one or two taxane regimens. These patients were all PSMA positive on a gallium 68, PSMA 11 PET CT. We're all familiar with the criteria that were used. There were relatively lax criteria, so most patients did not screen fail. They went right in. 13% of the patients did screen fail. There was no dual scanning for this patient population.

And I would say that in terms of the design of the trial, I know that there's been a lot of discussion about the design in terms of the control arm, but this is the way that it worked. You were registered to the protocol, your doctor then assigned you to a standard of care that the physician felt was appropriate. That standard of care could have been anything except for chemotherapy, immunotherapy, and radium. Because this was a combinatorial study. There was minimal amount of safety data on those combinations. So, it was a protocol-defined, but loosely applied standard of care with only exclusion criteria in terms of what the patients couldn't receive. But what they could receive was anything from another ARPI to steroids, to palliative medications, to whatever else the physician wanted. The stratification factors were ECOG, LDH, liver METS, and whether they had received an ARPI as a standard of care. There were two primary endpoints. One was rPFS per Prostate Cancer Working Group 3, and the other was overall survival.

And either could have been positive and the trial would have been declared a success. So as Matt just summarized, both of the primary endpoints were ultimately positive for survival of a 38% improvement and a difference of a median of four months. In terms of the survival, a 60% improvement in the rPFS with a hazard ratio of 0.4. A 50% improvement in time to SSE. And then in terms of the secondary more intermediate endpoints, there was about a 50% CR and PR rate for those patients who had radiographically evident disease and a maintenance that favored lutetium in terms of maintenance of quality of life and maintenance and non-deterioration of pain, as assessed by the BPI-SF. So, if you think of the cornerstone of FDA approval as feel, function, and survive, the trial met the hat trick of endpoints, feel, function, and survive all improved in favor of lutetium with the standard of care. And then PSA, probably the least relevant of the endpoints, was superior in favor of lutetium as well.

So what is the cost to the patient in terms of side effects? Well, nothing that's particularly surprising, both with this patient population and with the mechanism of action of the drug. 40% xerostomia, that was all grade. There was zero high-grade xerostomia, 40% of nausea and vomiting, primarily infusion-related with very few of those patients actually having high-grade GI side effects. And then for thrombocytopenia, 8% grade three or four. And for anemia, 13% grade three or four. All in all, for a patient population that is sort of at the end of the line, this would certainly be what we would consider to be an acceptable trade-off for feeling better, functioning better, and surviving longer in this patient population. A little bit about what predicts for a response, and this relates to what we were talking about yesterday. So, we have two nomograms. These have been submitted for publication and these were presented by Ken at ASCO last year.

In terms of nomograms, I think this is particularly important because we don't have otherwise nomograms that are based on survival-based prospective randomized studies, but again, you can see that the SUV mean is heavily determinative of the patient's outcome. But I think that this warrants some examination because I hear a lot in the clinic about, "Well, are the SUVs good or are the SUVs not good?" And how we should use that in order to determine whether the patient should go onto Pluvicto or not. So, first of all, on the top left here, this was from Andy Armstrong's presentation at ASCO two years ago in which rPFS looked like it was related on a quartile by quartile basis to PSMA SUV mean. So, the poorest outcome was associated with the lowest quartile. But with PSMA PET and SUV mean and its relationship to survival, this is the SUV mean of 9.9 or higher.

And so those patients are clearly doing better, but this is any other quartile. The survival looked pretty much the same for any other quartile. So, you might think, "Well, I'll only put patients with an SUV mean on Pluvicto." But that would be a mistake because if you take this same graph and you put it over here and match it head-to-head by quartile with how the patients did on the control arm, everybody did better with Pluvicto than not. So, if you see that the PSMA PET SUV mean... And I recognize this isn't a standardly reported feature of the PSMA PET. But it doesn't mean that these patients shouldn't get Pluvicto. It just means that they may do less well than these patients, but they will do better than patients who don't get Pluvicto and get some other standard. I think that's a very important subtle point, but you might, when you look at this, say, "Maybe there is some alternative that these patients could get in which they would be highly favored to do well."

And that brings us to TheraP because this is the only study that we have that compares head-to-head, although not as its primary endpoint, Pluvicto with cabazitaxel. I think this is a very, very important data set for us to think about because this is the natural alternative for patients who have not received two taxane-based chemotherapies, and so many of us are left with making this decision in the clinic chemo or lutetium for those patients who have not yet received two prior chemotherapies. So, this was a randomized phase two study. These patients did get dual imaging and as a result, you would think that the patient population would really be enhanced for being responders to the lutetium. And indeed, the screen failure rate was higher than with VISION because these patients were more rigorously screened for PSMA avid disease. These patients were randomized to either lutetium at 8.5 gigabecquerels every six weeks, which diminished with each cycle up to six cycles, or cabazitaxel at what is a standard dose of 20 milligrams per meter squared.

The primary endpoint here was PSA-based, which is important but appropriate for a randomized phase two study. So, what we talk about in terms of either rPFS or OS, those were secondary endpoints. The trial wasn't designed to really show those differences with statistical significance. But nonetheless, it's the only data that we have in terms of this comparison. So, it's important to analyze this carefully. But the primary endpoint was to assess with 80% power a 20% absolute difference in PSA 50 decline for each arm. And it was clearly a positive trial in that respect, greater than a 20% absolute difference in the PSA 50 rate of the two arms. And if you look at rPFS beyond six months, it looked like the curves were beginning to separate quite nicely. So, this was all looking quite good for looking at an ultimate OS benefit. And indeed, when you looked at PSA SUV mean, it was looking pretty predictive with everything except for the lowest quartile of SUV mean, which were those who had an SUV mean of less than 6.9.

But pretty much the higher the SUVs, the more likely you were to respond. Less so, but still trending that way for rPFS and for PSA progression-free survival. But when you ended up looking at overall survival at the Kaplan-Meier curves, or whether you looked at this as a function of the restricted mean survival time, you can see that survival is pretty much the same in both arms. And again, with the caveats not a primary endpoint, not a phase three, etc. It's still an important thing to remember that you have two standards of care in this space. And that even with selecting rigorously for success with Pluvicto, that in many ways cabazitaxel is a player here. What's interesting as well, going into the subtleties of this, is that if you look at the PSMA and the FDG assessment of tumor volume, you can see that this really is prognostic for OS.

Regardless of what arm you're on, the patients who have a higher SUV mean are doing better than the patients who have a lower SUV mean, and that's regardless of treatment and as well for the groups combined. And for the FDG, the more tumor volume that you have, the worse off you are regardless of the treatment that you receive as well. In other words, there's a lot of subtleties between VISION and the use of the PSMA scan and the SUV mean and TheraP, and we have a lot of work to do before we really understand how to utilize the predictive models to predict how patients will do with these treatments. Safety, I think, is a really important outline of the differences between chemotherapy and lutetium. I personally, if I had to live through each of these, would prefer the side effects of Pluvicto over chemotherapy, but there's certainly more neuropathy, GI side effects, hematuria with chemotherapy relative to lutetium.

So briefly, just to compare these and wrap it up, these were both for their primary endpoints, positive trials. There are a lot of subtleties here that would suggest that we are not really ready for prime time in terms of understanding what's predictive and what's prognostic for lutetium. Remember that PSMA PET is not a biomarker. I've heard this a couple of times yesterday and today. PSMA PET is a technology. The biomarker is the quantitative readout that you associate with survival, and I don't think that we know exactly what that is or how to use it and how to model that with the other predictors of survival quite yet. We need more data. But fortunately, PSMAfore, PSMAddition, and other registration trials should furnish that information. Thanks very much.