Read the Full Video Transcript

Andrea Miyahira: Hi, everyone. I'm Andrea Miyahira, the Senior Director of Global Research and Scientific Communications at PCF. Joining me today is Dr. Himisha Beltran, an associate professor at Dana-Farber Cancer Institute. Dr. Beltran and team recently published the paper Landscape of Prostate-Specific Membrane Antigen Heterogeneity and Regulation in AR-Positive and AR-Negative Metastatic Prostate Cancer in Nature Cancer. Dr. Beltran, thank you for joining us to discuss this paper today.

Himisha Beltran: Thank you so much for having me. It's good to see you Andrea. We are super excited to share our new paper Nature Cancer, as mentioned. This paper, the study was led by Martin Bakht, the post-doc fellow in my lab, and it's been a huge team effort with multiple collaborators at Dana-Farber, Memorial Sloan Kettering, and Weill Cornell.

The rationale for this study was that we had observed, and others as well, that when you get to late stages of prostate cancer, there can be quite a bit of heterogeneity at the phenotypic level. Some cancers can develop neuroendocrine features, an area that we've been particularly interested in for many years, and some tumors lose androgen receptor, as well as PSMA, expression. And so, not much had really been known about why and how PSMA expression is lost in these cases. We know, of course, an exciting advance in the clinic is that we now have Pluvicto, the PSMA targeted radionuclide that's approved for patients with advanced prostate cancer with confirmed PSMA-positive disease on PET.

But those clinical trials also revealed that a significant proportion of patients do have PSMA-negative or PSMA-low disease on PET scan and are excluded from receiving Pluvicto. Even within PSMA-positive lesions, work by Johann de Bono, Michael Hefner, has shown that there can be tumor heterogeneity, and the degree of heterogeneity on PET scans SUV means it correlates with response to Pluvicto. So these are all setting the clinical stage of why we thought it was important to try to understand PSMA regulation. There had been a whole body of work looking at how androgen receptor can regulate PSMA or the full H1 gene by binding the enhancer and that cell lines that are AR-positive or PSMA-positive, and cell lines that are AR-negative tend to be PSMA-negative, so it's always been thought that AR is required for PSMA expression.

The questions that we sought out to ask, again, led by Martin, is, how does this loss of PSMA expression occur? Is it due to lineage plasticity and transformation to neuroendocrine prostate cancer? Is it really true that all neuroendocrine prostate cancers are PSMA-negative, which has some clinical implications there? And then, are there better ways to image PSMA-low prostate cancer as thinking of thinking about how we may be able to use molecular imaging to develop more refined biomarkers of biology.

When he looked at androgen receptor-positive metastatic prostate cancer, what we found is that there was varied levels of PSMA expression across metastasis in individuals, and particularly lower PSMA expression and liver metastasis compared to other sites of disease, suggesting that maybe there are features of the microenvironment that are contributing to lower PSMA expression or maybe it's that PSMA-low clones tend to preferentially metastasize to the liver.

So we developed a couple orthotopic models, and I'll tell you about one of them today. And of course there's more on this in the paper if you're interested. But one of the models that we developed is a 22Rv1 PSMA-positive, androgen receptor-positive prostate cancer orthotopic model, where we took the cell line and put it in the prostate of a mouse. And you can see this develops an AR-positive, PSMA-positive prostate cancer, which subsequently metastasizes, including, interestingly, PSMA-high, PSMA-low, and PSMA-negative liver metastasis. And these were all androgen receptor-positive, so suppression of PSMA was not due to loss of androgen receptor.

We also confirmed that this PSMA-low tumor was stably PSMA-low when imaged by gallium PSMA PET imaging compared to the control 22Rv1 tumors. Perhaps this PSMA suppression or heterogeneity liver METs contributes to some of the inferior outcomes that we see in patients treated with Pluvicto that have liver METs as something that I think could be explored further.

Interestingly, when we took this PSMA-low liver met and put this back into the prostate of a mouse, multiple metastasis developed, again, that were all PSMA-low, despite still retaining androgen receptor, suggesting also that maybe PSMA-low lesions can give rise to other PSMA-low lesions and that, perhaps, the microenvironment is important, but at some point the seed may be more important than the soil in driving this clonal dominance. In these cases that are AR-positive but PSMA-low, what we found is H3K27 trimethylation and likely epigenetic suppression of PSMA in this context.

We were also interested in this neuroendocrine prostate cancer, AR-negative situation where we often saw loss of PSMA. And, indeed, when we looked a little bit more broadly at clinical datasets and our patient cohorts, we found that most neuroendocrine prostate cancers are PSMA-negative, but a small fraction were actually PSMA-positive. Here are five neuroendocrine prostate cancer xenografts, for instance, and two of these five were PSMA-positive, yet all being androgen receptor-negative. And here you can see confirmed strong PSMA uptake in this neuroendocrine model that was AR-negative. So this was really surprising to us.
First of all, this, I think, implies that maybe we shouldn't be excluding our neuroendocrine prostate cancer patients from getting PSMA imaging and being considered for Pluvicto, but also suggests that there's this disconnect between androgen receptor and PSMA and that maybe there are alternative regulators of PSMA. And so through a series of cistrome analysis and functional experiments, we honed in on HOXB13, which we know to be a androgen receptor co-regulation, but it identified a novel enhancer where it can regulate PSMA. We found this to be relevant in our neuroendocrine models that were AR-negative, as well as in our AR-positive models. And so the model here is that, while most cases when you lose androgen receptor, you lose PSMA expression, in some cases, particularly these neuroendocrine tumors that have no androgen receptor but high PSMA that HOXB13 can still bind to the cancer to regulate PSMA.

These findings are also led us to try to figure out, well, what are some other clinical implications and can we better image these tumors? We looked both at neuroendocrine prostate cancer and our PSMA-low liver MET, and we found that one of the most upregulated pathways in PSMA-suppressed tumors where amino acid transporter activity, and we know fluciclovine is taken up by amino acid transporters and fluciclovine PET imaging is an approved imaging modality that was being used in recurrent prostate cancer and now largely replaced by PSMA. But what we found is that in our PSMA-suppressed models, that we see strong uptake of fluciclovine PET. This is our liver met model of our 22Rv1 that was PSMA-low, as well as our neuroendocrine prostate cancer models that upregulate L-leucine, the amino acid. And we see high SUV means and maxes in our neuroendocrine models for fluciclovine PET, as opposed to the more variable PSMA expression, as mentioned.

Based on this, we are initiating a pilot clinical study at Dana-Farber in collaboration with Heather Jacene from Nuclear Medicine, where we'll be doing fluciclovine PET imaging in patients that have PSMA-low disease on PET imaging or have neuroendocrine prostate cancer. We'll be obtaining blood samples and biopsies, and we hope to learn a lot about how this may complement PSMA and what are the features that characterize fluciclovine-avid prostate cancer, as we start thinking about phenotypic subtyping in prostate cancer, where I envision a future where we'll probably see more in the lines of molecular imaging, liquid biopsies, and other non-invasive approaches.

I just want to acknowledge Martin, who led this work with lots of members of our lab who contributed. Collaborators, again, at Dana-Farber, Memorial, Weill Cornell, and Blue Earth who provided the tracer. I want to give a special shout out to Tony, QD, Jason, and Kishore, our imaging collaborators. I've really learned such a tremendous amount by working with them, so I think this is really truly a multidisciplinary study. We're really excited, and so please read the paper. Thank you PCF, Andrea, for all your support.

Andrea Miyahira: That was awesome. Thank you for sharing that excellent presentation Dr. Beltran. Just a few questions. I'm really intrigued by the lower expression of PSMA in liver. Do you see that in patients and do you think there's a functional reason to have lower PSMA in liver?

Himisha Beltran: Yeah, this is a really good question. We did look in patients and we do see more heterogeneity in liver, and I think when you look at tumors, we definitely see lower overall expression in liver. This is predominantly, I think, in the context of AR-positive disease, due to epigenetic suppression. The question is really, like you said, why are we suppressing PSMA? Why do we see downregulation of folate and upregulation of these amino acid transporters? This is kind of taking us in a new direction, which we're building upon this to look more at metabolomics of these PSMA-low lesions and liver METs and try to understand how these features are driving progression and how we may be able to modulate this. I know Michael Hefner has been interested in looking at, can we de-repress PSMA using epigenetic strategies? I think that's also quite a fascinating avenue of, we can understand what's causing it to be suppressed. Maybe we can modulate that.

I do think, though, that in the setting of AR-negative disease, that or PSMA-negative and HOXB13-negative, that it will be very quite unlikely to re-express PSMA when you don't have androgen receptor present at all, because that's just been our experience in the context of EZH2 inhibitors and DNA methyltransferase inhibitors in the setting of neuroendocrine prostate cancer, while we try really hard to reverse this plasticity.

Andrea Miyahira: Okay, that's really interesting and I'm looking forward to seeing the results from those next studies. What next steps do you envision for optimally choosing patients for Pluvicto?

Himisha Beltran: Yeah, I think one thing I'm really happy about is that we have PSMA scans that are being used in conjunction with Pluvicto because many investigators, PCF-funded investigators in Australia, Europe, and US have been really leveraging those scans to optimize what we can learn from them, as far as quantifying uptake and disease burden using AI approaches and integrating that with clinical features. I think that's going to be really important in developing clinical predictive biomarkers to help us optimize our patient selection and understand resistance. And I think the work that we're doing looking at biology and how we regulate PSMA can compliment that and hopefully will provide strategies for combination approaches, complimentary biomarkers. With the approval, it's been around for a while and lots of investigators have been setting this, but I think we have a major opportunity moving forward to leverage what we're learning in the clinic to guide those studies and also understand resistance.

Andrea Miyahira: Yeah, I was going to ask, do you think any of the mechanisms you've identified here are contributing to Pluvicto resistance? And if you have any thoughts on how we could counteract those?

Himisha Beltran: Yeah, I think that PSMA heterogeneity and loss of PSMA, I think, does contribute to resistance and we're not necessarily excluding all of those patients, because PSMA-low patients are still eligible for Pluvicto. Not that I'm saying that they shouldn't be, but I think we can learn more about how we may be able to maybe combine with something else and understand that.

I think that it also will help us understand acquired resistance and whether loss of PSMA is going to be a mechanism in some patients. I don't think it's going to be the only mechanism. Similar to what we've seen with AR-targeted therapies where we can see lineage plasticity, are we going to see similar lineage plasticity when you have an effective PSMA-targeted drug? And what is that going to look like? And then, of course, there's mechanisms of radioresistance and other features that may also drive resistance and subsequent therapies when we're thinking about actinium or other PSMA-targeted approaches. So, I think that we still have a lot to learn, but I think that I'm hopeful that I would encourage people to do progression PET scans and biopsies in the context of research studies as much as possible, because I think that's how we'll really learn.

Andrea Miyahira: All right. Well, thank you so much Dr. Beltran. This was a really informative study and I encourage anybody who's interested in more details to read the paper. Thank you for joining me today.

Himisha Beltran: Thanks so much for having me. Thanks.