Read the Full Video Transcript

Tanya Dorff: I'm really excited to share with you some of the work that we've been doing at City of Hope. But obviously, also some work that's being done elsewhere on PSMA-targeted immunotherapy. And I'm going to go a little bit rogue because this is PSMA and beyond. So I will touch briefly on some other targets.

So this is just a table showing some of the CAR T that have been developed targeting PSMA. And you can see there's even a CAR NK. So obviously, a good target. The CD3-targeted bispecific is targeting PSMA. And there's also at least one that's CD28.

So these are really complex platforms that you can do a lot with. You can make all kinds of different modifications that can impact efficacy and toxicity. So hopefully, you'll see that as we go through just a couple.

So the Poseida P-PSMA-101 CAR T was one that we worked with. And this has a nonviral transfection to change the T-cell receptor. And it tends to produce a higher rate of stem or naive T-cells, which are really, really vigorous. So think about a teenager compared to you and what you can do in an afternoon, right?

So these T-cells are really, really vigorous. And they also have a kill switch, which became very important because of just how vigorous the proliferation was with this agent. And so we are trying to get it published now. But these were mostly taxane pre-treated mCRPC patients.

A couple had had the PSMA bispecific T-cell engager before entering onto the trial as well. But none had had lutetium PSMA because it just wasn't available when we were running this trial. So we started with dose escalation, but there was an unfortunate grade 5 toxicity, so then we had to actually de-escalate the dose.

CRS was only about 57%. But some of it was grade 3 and higher. And then we also saw a little bit of ICANS with this, which we don't always see with some of the other targets. So I wonder if that's a PSMA-specific issue.

And in fact, not only did we have to use the rimiducid kill switch. In one patient, we had to use it twice because those cells just kept coming back and proliferating. So you can see on the swimmer's plot, what I want to call your attention to is the yellow and green dots. Those are PSA responses.

And you can say that they're often followed pretty closely by a black dot, which is progression. So unfortunately, the big issue in this trial was the limited duration of response. The waterfall plot shows that there is definitely some PSMA as well as some PSMA-- sorry, PSA and PSMA PET responses. And we showed one of the exceptional responders with the PET scan changes there on the right.

That patient did stay in remission for close to a year. And when we did biopsies, we did see presence of CAR Ts infiltrating into the tumor. So proof of principle that the CAR T-cells were there. The question is then why does the efficacy wane? And I think that's unanswered. It doesn't necessarily look like it's due to antigen downregulation.

The UPenn group had a PSMA CAR T that has a TGF beta receptor dominant negative construct. Again, very, very vigorous. These cells are designed to survive and thrive in this hostile microenvironment. And they saw some good PSA responses shown on the waterfall plot. They show you some images. But they also had excessive proliferation that led to a grade 5 toxicity.

So the issue is really threading the needle. What we don't see with PSMA-targeted CAR T is dry mouth. And we do see a bit more liver and neurotoxicity, which I do think is antigen-related. So we'll have to look at other targets for CAR T, I think, and also some adjunctive strategies and different dosing strategies to get there.

But on the other hand, you get these exceptional responders, but very few of them with CAR T. With the bispecifics, it seems like almost everyone responds. It looks a lot more like a radionuclide therapy kind of a response curve there.

However, you can see that cytokine release was about 100%. I mean, this AMG 160 was very active, but also quite toxic. And at that 0.9 dose, which was as high as we got, we did see hearing and vision toxicity.

So again, the CNS signal that we don't see with other targets, and I don't think can be explained simply based on immunotherapy. But a very different side effect profile than what you get with your radioactive particles.

J&J also had a CD3 bispecific. And you can see really in the IV cohorts, not responding. And that was due to antidrug antibodies. They changed to subQ and saw some responses. So this was also an issue with AMG 160. We saw a lot of antidrug antibodies. And I wonder if that's part of what's going on with the CAR Ts and the lack of durability of response.
There's just a little more rejection when we go after this target. So high rates of response, not durable. And mostly, I think, we're going to end up looking at other targets. I think PSMA may not be as good for immunotherapy as it is for radioligand.

So there are some other bispecifics. And I just wanted to highlight AMG 509. Because it's really exciting. This is now in phase III. So the phase I, we showed a really high response rate. This targets STEAP1 and it has a 2 to 1 construct. And whether that's the key to having a lower rate of cytokine release or whether it has to do with STEAP1 versus PSMA—unanswered question.

But the cytokine release happens mostly in cycle 1 and it wanes by cycle 2 and 3. But you do get these musculoskeletal toxicities, which were not predicted based on expression of STEAP1. So we learn a lot about our targets as we target them in different ways.

But the phase III is open. We've put a bunch of patients on already. 2 to 1 randomization to xaluritamig versus physician choice which does include cabazitaxel. So this will be a registrational and really impactful trial.

If you have patients in the LA area who are motivated, please send them our way because I think this will cross the finish line, which is really exciting. I've seen it help a lot of patients. I also did with the AMG 160. I had a patient really who was at death's door. And it stopped the disease in its track. And then he went on to get the CAR T and responded. So I think sequencing remains an open question.

I just wanted to highlight Rahul Aggarwal's presentation on the DLL3-targeted T-cell engager. Because not every prostate cancer has PSMA or STEAP1. They get dedifferentiated. And I think what we learned from this trial is you have to target patients who have the antigen you're trying to target, right? The response rates don't look great. But the patients who did respond were strongly DLL3 expressing.

So in our CAR T trial with PSCA, we do stain tissue. Whether that's the best way, whether CTCs, whether imaging, what's the best way to show that the patient has the antigen you're trying to target is an unanswered question. And it may be different when we look at different modalities as well.

There was also a second DLL3-targeted T-cell engager. Neuroendocrine prostate cancer is an unmet need because they just have different antigens. And PSMA is not going to work for them. I did want to highlight this provocative presentation about resensitization to pembrolizumab after a PSMA bispecific.

But if you note, up at the very top, they did give radiation. And I think that could actually be maybe what resensitized. And that's something we've been interested in at City of Hope is looking at the role of radiation in various forms and how it could prime the tumor microenvironment by getting rid of some T-regs and increasing antigen presentation so that our immunotherapies can work better.

There's some concern about all the dexamethasone that we use in these trials. In preclinical models, it looks like it abrogates the effect compared to toci. But AMG 509, we're giving a ton of dexamethasone and we're still seeing these really gratifying responses. So I think it remains an unanswered question.

So PSMA-targeted CAR T and bispecifics, I'm not sure are going to be the way we get there with immunotherapy for prostate cancer. Unless we really modify these constructs. Maybe it's the 2 to 1. Maybe in the CAR T, we just have to balance proliferation and energy with those toxicities and find some adjunctive strategies. But there are also these other targets, which I think are fairly promising as well.