Exploring PSMA Expression Variations in Metastatic Prostate Cancer: The Role of Reversible Epigenetic Changes - Michael Haffner

May 9, 2023

Michael Haffner and Andrea Miyahira discuss a JCI Insight publication, "Reversible Epigenetic Alterations Mediate PSMA Expression Heterogeneity in Advanced Metastatic Prostate Cancer". The study investigated PSMA expression heterogeneity in advanced metastatic prostate cancer, finding that the expression pattern of PSMA varies significantly between different metastatic sites and patients. They also discovered that the gene encoding PSMA is regulated by both DNA methylation and acetylation of the histone H3K27. The team found that histone deacetylase (HDAC) inhibitors can reverse the epigenetic repression of PSMA expression, leading them to develop a clinical trial that aims to increase PSMA expression in patients with low levels of PSMA. The trial will involve epigenetic priming with an HDAC inhibitor before administering lutetium PSMA.


Michael Haffner, MD, PhD, Assistant Professor, Human Biology and Clinical Research Divisions, Fred Hutchinson Cancer Research Center, Assistant Professor of Pathology, University of Washington, Seattle, CA

Andrea K Miyahira, PhD, Director of Global Research & Scientific Communications, The Prostate Cancer Foundation

Read the Full Video Transcript

Andrea Miyahira: Welcome, everyone. I'm Andrea and I'm the Senior Director of Global Research and Scientific Communications at the Prostate Cancer Foundation. Today I'm joined by Dr. Michael Haffner, an assistant professor at Fred Hutchinson Cancer Center. Dr. Haffner and team recently published the paper, "Reversible Epigenetic Alterations Mediate PSMA Expression Heterogeneity in Advanced Metastatic Prostate Cancer and JCI Insight." Dr. Haffner, thank you for joining us to discuss this paper today.

Michael Haffner: Andrea, thank you very much for having me. So it's really a great pleasure to share with you some of our recent results regarding the epigenetic regulation and heterogeneity of PSMA expression. And maybe what I want to start out with is really think a little bit about PSMA as a targeting molecule. We know that over the past couple of decades there's been a major interest in developing new therapies that specifically target cell surface proteins. And PSMA for prostate cancer has really been on the forefront of this development. And this really culminated in the recent FDA approval of a radioligand drug that specifically targets PSMA. And here I'm showing you outcome data from the Pivotal Vision Trial, which expressly demonstrates that this radioligand lutetium PSMA shows a increased progression-free survival. So overall, there's really great enthusiasm for this new type of targeting approach.

However, we also have to consider that even in this positive trial progression-free survival was still relatively short with less than nine months on average. The overall survival is still poorer in this advanced patient population. And [inaudible 00:01:51] responses were relatively rare with about 50% of cases.

So one of the things that we really want to think about when we're talking about targeting cell surface proteins in oncology is that expression of the cell surface target really matters. And this has been demonstrated for PSMA in multiple different ways. Here, I'm showing you data for a PSMA ADC-based targeting strategy where it is absolutely dependent to have high levels of PSMA expression to get efficient tumor targeting.

So what we thought we will see is that there's a major need and in terms of two separate aspects. So one is, we really want to assess the expression pattern of PSMA in patients with advanced metastatic prostate cancer. And then we want to determine, what are the mechanisms that underlie the repression of PSMA expression in tumors where, in this case by IHC, we don't see any PSMA labeling?

In order to do that, we leveraged a very unique cohort of samples that were available at the University of Washington. And these come from a rapid autopsy cohort where patients that died of advanced metastatic disease underwent the rapid autopsy to procure all the different metastatic tissue sites. And in this particular cohort, we included 52 man. We sampled in these 52 man up to 20 different metastatic sites per patient. And this totaled over 830 tumor samples, which allowed us now to really study in very, very great detail, the heterogeneity and the expression pattern of PSMA and advanced disease.

And what we started to appreciate is that there was a very high level of intra patient expression heterogeneity. So what does this mean? In these micrographs here, I'm showing you PSMA IHC micrographs from a single patient from a liver lesion, a bone lesion, and a lymph node lesion. And it's very easy to anticipate just looking at the brown staining, that there's a dramatic difference between the different metastatic sites in terms of PSMA expression.

So to summarize this cohort, here's a graph that shows you all 52 patients. And each metastatic site is shown here in a dot. And all of the dots are color coded based on the molecular phenotype, which we determine on IHC for androgen receptor and also neuroendocrine marker expression. And what you start to appreciate is that there's a relatively large group of patients, about 25% that have absolutely no PSMA expression in any of the metastases. And this group is strongly enriched for tumors that show neuroendocrine small cell differentiation.

Next, we saw a larger group of about 44% of patients that show high levels of heterogeneity in PSMA expression. So what would we mean by that? So in these patients, we see several metastases showing no or very low level of PSMA expression while other metastases show high level of expression. And then another group was identified that showed relatively uniform high PSMA level expression. But overall, the picture that you get is that there's a large fraction of tumors that show highly heterogeneous expression of PSMA on the protein level.

And to dig a little bit deeper into an individual lesion, here I'm showing you micrographs from one large liver metastases. We started to appreciate that even within one metastatic deposit, you can see a very high diversity of expression of PSMA with some tumors cell clusters expressing high levels, others show complete absence of PSMA expression.

So this really prompted us to try and understand a little bit more, what are the underlying mechanisms that regulate PSMA expression? And prior work has suggested that the endogen receptor plays a key role in potentially suppressing PSMA expression. And we found some evidence for that as well. But this alone could not really explain the striking differences of PSMA expression that we found in different patient-derived xenografts, but also in our patient cohorts.

So, we started to perform epigenetic analysis and search of potential epigenome pattern that would inform us about PSMA regulation. And what we started to see is that in cases with very high PSMA expression, in the case of this PDX line LuCaP 77 for instance, we saw a strong enrichment for the activating histone mark H3K27. And a large stretch of DNA high poll methylation right around the transcriptional starts at bleeding into the first couple of entrances of FOLH1, which is the gene encoding for PSMA.

However, in cases with no PSMA expression, we saw that this region uniformly was densely methylated and the enrichment for H3K27 was completely gone. So this really suggested to us that there's potentially an epigenetic mechanism that centers around both DNA methylation as well as acetylation of the histone H3K27 that controls PSMA expression. And we were able then to corroborate these findings, in particular on the level of DNA methylation, also in other clinical cohorts, again, showing the same picture that methylation of first introns of the FOLH1 gene is associated with repression of PSMA.

So, we know that a lot of these epigenetic marks often act in concert in repressing and regulating gene expression. And over the past decades, there's been a major effort in developing specific inhibitors to either readers or writers of these epigenetic marks. So we reasoned that maybe it would be a potential approach to reactivate PSMA expression by using some of these already available molecules that target key components of the epigenetic machinery. So we performed a series of in vitro experiments in cell lines with different PSMA expression at baseline. And we treated them with different inhibitors for methyltransferases, histone deacetylases (HDACs), and other epigenetically active compounds.

And our search really led us to the observation that HDAC inhibition results in different cell line models in re-expression of PSMA on the protein level. And again, this is independent of the baseline expression of PSMA in these different models. And you can see here in these flow plots, a variety of different pan-HDAC inhibitors, including panobinostat and vorinostat, are very effective in increasing PSMA expression. And this is also recapitulated in these fluorescent images to the right that show you a strong re-expression of PSMA after HDAC inhibitor treatment.

And we were also able to demonstrate the re-expression of PSMA in vivo experiments. And in this case, we used patient-derived xenograft and tumors that express low to very little PSMA at baseline. And after a four-week treatment period with the Pan-HDAC inhibitor CUDC-907, we observed robust re-expression of PSMA in all treated animals.

So this really led us to conceptualize a proof of concept clinical trial that is led by my colleagues Mike Schweizer. Even YU, And Delphine Chen, where patients with low level PSMA expression based on PSMA-PET would be enrolled and given a 28 days HDAC inhibitor, epigenetic priming. And then PSMA expression would be read out again by PSMA-PET to look for potential conversion of patients with low levels at baseline, but then increased expression after this epigenetic priming therapy.

If patient would convert to high PSMA expression, they would then be able to go on to receive lutetium PSMA. And of course embedded in this clinical trial, are several correlative studies including multiple tissue biopsies to correlate PSMA-PET findings with tissue staining as well as several blood-based biomarkers.

And the study would've not been possible without the great support of the members of my laboratory. As well as a wonderful list of colleagues from both the Fred Hutchinson Cancer Center, the University of Washington, Weill Cornell, and UCSF. And I thank you very much for your attention.

Andrea Miyahira: Thank you Michael, for sharing that with us. I was wondering, have you compared your PSMA IHCH scores with PSMA-PET SUVs? Or also do you know if PSMA low H scores would correlate with FDG PET SUVs? As we know that Michael Hoffman and others have shown that when we're trying to select patients for Pluvicto, if we do a double PET then we're able to see PSMA high versus PSMA low lesions and better select patients.

Michael Haffner: Andrea, this is really an excellent question and something that we're very, very interested in investigating. Unfortunately right now, there's very limited data on the correlation between PSMA-PET findings and tissue based IHC studies on matched lesions. And I think this is also the type of study that's extremely difficult to execute in the right way.

So one of the things that we are planning on doing right now, the study is already ongoing, is to try and use our rapid autopsy cohort to really perform rapid autopsies on patients that had very, very previously undergone PSMA and FDG PET. And this would really allow us to procure tissues, not just from one lesion but from multiple different lesions, and study the correlation between PET findings and tissue-based findings at a much, much greater level. And again, this would also allow us to dig much deeper, not just look at IHC findings, but also really understand, what are the underlying molecular alterations that we find in tumors within a given patient that show different PET avidity pattern?

The other thing I think that's also important here is, as I've shown in my slides, there's a lot of heterogeneity even within a given tumor deposit. So it really makes me a little bit worried that a single biopsy from a larger lesion might not really reflect the heterogeneity of PSMA expression. And therefore I think these autopsy based studies that address this very, very crucial question for PSMA biology are really needed to understand much better the correlation between PET findings and tissue-based findings.

Andrea Miyahira: Yeah, thank you. I think that would be very enlightening. So you show the epigenetic alterations in PSMA seem to suppress PSMA expression. Do you know if epigenetic alterations may mediate primary resistance and, or required resistance to PSMA targeted therapy?

Michael Haffner: So again, excellent question. I think the primary resistance question, we would speculate yes, because in a lot of, at least the PDX models that we know do not express PSMA and are therefore also resistant to PSMA targeting agents, the PSMA locus shows the same epigenetic alteration that we have described in this study.

Now, the treatment emergent resistance is a very good question, and right now we don't have data, but again, we're collecting actively, patient samples that have failed lutetium... from patients that have failed lutetium PSMA therapy. So hopefully in the next month, we'll be able to analyze those and report in potential epigenetic changes in these tissues as well. My gut feeling would be that knowing that there is this mechanism to reduce or shut off PSMA expression already operative in prostate cancer cells at baseline, I would not be surprised if we see that same mechanism also playing a role in treatment induced resistance.

Andrea Miyahira: Okay, thank you. So HDAC inhibitors, they were found to increase PSMA expression in your model, and I think that's really interesting idea, your trial. Do you know yet if HDAC inhibition might increase PSMA and non-prostate cancer tissues and contribute to toxicities?

Michael Haffner: So I think this an excellent question. And again, one of the key objectives of the clinical trial is really to evaluate the expression pattern of PSMA in patients that have received HDAC inhibitor priming to see whether there are other tissues that start to be PSMA-PET avid. However, based on our cell line and in vivo experiments in mice, we have not seen PSMA re-expression in non-prostate lineage-derived models. So it still, I think, remains to be critically tested in patients. But from our preclinical models, this does not seem to be a major problem.

Andrea Miyahira: Okay. When do you think your trial might open?

Michael Haffner: That's a wonderful question. We hope to start in the next couple of months.

Andrea Miyahira: Okay, cool. And do you have any other major takeaways that we can learn from your study as far as using PSMA as a theranostic targeting for imaging or for treatment?

Michael Haffner: Yeah, I mean I think the bigger picture is that target expression heterogeneity is really a major challenge, I think, for all of these cell surface marker-based approaches that we'll see hopefully coming down the pipeline in the next couple of years. But I think once we start to understand the regulation of these proteins, we can intervene as we have proposed here with potential other co-targeting strategies that enhance, in this case, PSMA expression.

I think the other thing that's really intriguing as well, and this was part of our study that I didn't present here today, is that in PSMA negative tumors, we see a whole host of other potentially targetable cell surface markers being expressed at higher levels. And what stood out from our data was that MUC1, for instance, a cell surface protein that has been explored for therapeutic targeting and many other solid tumors, was specifically upregulated in PSMA low tumors. So again, suggesting that maybe a co-targeting of not just one, two or three different cell surface molecules might really be the way to go to increase efficacy of these therapeutic approaches.

Andrea Miyahira: Okay. Well, thank you so much. And thank you for coming on today to share this paper with us.

Michael Haffner: Thanks. This was great. Really appreciate it.