Read the Full Video Transcript

Andrea Miyahira: Hi everyone, I'm Andrea Miyahira at Prostate Cancer Foundation. Today I'm talking with Dr. Jessica Hawley, an assistant professor at the Fred Hutchinson Cancer Center and University of Washington, and Dr. Alex Obradovic, an associate research scientist at Columbia University. Their group has just published the paper "Anti-PD-1 Immunotherapy with Androgen Deprivation Therapy Induces Robust Immune Infiltration in Metastatic Castration-Sensitive Prostate Cancer" in Cancer Cell. Dr. Hawley and Dr. Obradovic, thanks so much for joining us today and sharing this work.

Jessica Hawley: Thanks, Andrea, for having us.

Aleksandar Obradovic: Thank you very much.

Jessica Hawley: All right, well thank you for that introduction. We're glad to be here today to share with you some of the results from our paper, which this is very appropriate timing, Andrea, I don't know if you knew this when we scheduled it, but the paper just came out online yesterday actually. So, looking forward to hearing comments from the research community. Just to kind of give a brief overview of what we did in the study, this is the graphical abstract from the paper and it's showing... I'll direct your attention up on the left, which is showing that we had 10 patients who were enrolled on a phase two clinical trial, which I'll talk more about in the next slide. And we asked all patients to undergo baseline and on-treatment biopsies. And then those biopsies were from metastatic sites across the spectrum of where prostate cancer tends to metastasize to.

And using those fresh tissue samples, we analyzed them using single-cell RNA sequencing, which Alex is going to talk more about how he used the VIPER inferred protein activity algorithm and what benefits that has. And then we also did quantitative immunofluorescence on the samples to provide some orthogonal protein validation of our findings. So these data come from the phase two trial called The PRIME-CUT Study, which was an investigator-initiated trial that we began pitching to companies back in 2018, and Regeneron was kind enough to see the possibilities of this work and funded the trial. And so this is the schema of the trial showing key eligibility criteria. Essentially patients with metastatic hormone-sensitive prostate cancer that had still robust testosterone levels. And we treated all of the patients with Degarelix first, followed by phased administration of Cemiplimab, which is Regeneron's anti-PD-1 therapy, followed by standard of care at that time, six cycles of Docetaxel.

And shown here in the blue are the baseline biopsy samples that we asked all patients to undergo. And then we randomized patients to two on-treatment time points either after ADT alone or after the combination of ADT and Cemiplimab. And this table comes from the paper, I'm not going to dive into it in too much detail other than to say that this patient population comes from Washington Heights and is a very diverse patient population. And we also just call out here that we were able to achieve good cellular yield from most of the metastatic tumor biopsy samples. I'll pass it over to Alex now.

Aleksandar Obradovic: And our analysis of the single-cell RNA sequencing of these tissue samples included a pipeline of inference of protein activity. So this is the VIPER algorithm to get a better signal-to-noise, allowing improved sub-clustering of some of these cell populations than you would typically get from just gene expression alone. As well as interrogation of key signaling proteins and regulatory molecules by the effect of their downstream pathway activation. Especially since a lot of these molecules of interest immunologically don't have necessarily very high direct expression levels but have significant downstream effects on the phenotype of a cell.

So this has allowed us to get the clustering that we show in the paper, which moving on, showing just a heat map that we won't dig into because we dig into it more in the manuscript itself, but just showing the strength of the signal-to-noise of some of the top markers that differentiate each of these sub-clusters of both tumor cells and immune cells very distinctly from each other, which we then dive into and sort of analyze by tissue site, as well as by time point and show that first just at the baseline pretreatment time points, there's a significant difference in immune microenvironment composition by tissue site. Especially the lung we found had significantly less immune infiltration than other metastatic tissue sites.

And the composition of those metastatic tissue sites of the immune microenvironment in those tissue sites also differed. Digging into those specific sub-clusters we identified by protein activity, the bone was mostly populated by plasma cells and monocytic cells. Liver had quite a lot of CD8 T cells, although mostly of an exhausted-looking phenotype and lymph node had B cells, Tregs, and macrophage infiltrate as well as other T cells. And then the lung was overall the most immune-depleted at baseline. Then the effect of the treatment is what is one of the major findings of this is that the combination of ADT with anti-PD-1 led to a significant expansion of T cells across the board, although most markedly of CD8 T cells and activated phenotype of CD8 T cells as well as CD4s and Tregs. So sort of dramatically increased T cells with this combination therapy.

Which we then validated orthogonally by immunofluorescence as just mentioned, looking at the fold expansion of T cells with combo therapy by immunofluorescence and by single-cell RNA sequencing, we get a two to threefold expansion of T cells induced by this treatment and a four to fivefold expansion of CD8 T-cells in particular. And then finally, the protein activity approach also allowed us to sort of dig further into the sub-phenotyping of the tumor cells themselves, which we found to be fairly heterogeneous, but with phenotypes that were represented across patients commonly and had certain shared characteristics. So these further dissected in the paper, but for example, have different levels of androgen receptor activity but in different tumor cell subpopulations. And the subpopulations with less inferred androgen receptor activity actually tended to survive the combination therapy at a higher rate. So this is some additionally delved into in the manuscript.

Jessica Hawley: And then I'll take it back from here. So this spider plot shows how we broke the patients into three treatment response groups, and we classified them as early responders, stable disease, and late progressors. And this was based on the PSA fold change from baseline. So those who had a PSA decline to below one percent of their pre-treatment levels, we called those early responders. And the late progressors were those who had initial treatment response, but then very gradual progression around week 28 as you can see here. We excluded the patients with stable disease for purposes of the analysis, but in some of the reviewer's comments, we did dive into those as well. The reason for doing that treatment group categorization was we could then go back to the baseline sub-clusters and determine if any baseline features were associated with clinical response. And so on the top we're looking at the immune sub-clusters and the bottom are the tumor sub-clusters.

And as shown here in green are those that were associated with early response. So the CD8 T cells that had high activity levels of LAG-3, which Alex alluded to, were also increased with the combination therapy. Similarly, the Treg population that had high protein activity levels of GITR was associated with early response. Whereas a CD4 T cell population that had high protein activities of TNF-alpha was associated with late progression. Shown below are some of the tumor sub-cluster findings as well. So the REF-EPI 1 sub-cluster was a sub-cluster that had high levels of androgen receptor activity. So, not surprisingly, that was associated with early response.

And so we'd just like to thank all of the patients and families for contributing not only to this study but to all studies and trials, especially those that are using translational science, calling out some of the members from Columbia and Fred Hutch who contributed to this work. Obviously, all of the co-authors and our funders, without whom this work would not be possible. Regeneron for sponsoring the trial, Conquer Cancer, Prostate Cancer Foundation for support of the young investigators featured today. And the National Institute of Health for providing a lot of support to our cancer centers and core facilities, and Cell Press for taking an interest in our work and sharing our findings with the larger community.

Andrea Miyahira: Thanks, Jess and Alex, for sharing this work. So, some questions. You saw different tumor microenvironment changes in different metastatic niches. What do you think drives those differences, and how should we develop biomarkers to investigate immune changes or response to immunotherapies as we go forward?

Jessica Hawley: Yeah, that's a great question, Andrea. Thanks. We think a lot of the changes that we saw across the different metastatic niches were a function of the tissue context in which those metastases were taken from. And so I think our paper is really helpful in sort of shining a light on beginning to understand how these different tumor microenvironments are different across the different metastatic niches. And the changes that we see with treatment are different across those different metastatic niches. So, might we start to begin to think about offering clinical trials and different treatment therapies and precision oncology based on where a patient's metastases are? So maybe not a one-size-fits-all, but different types of therapies based on where the metastases may be.

Andrea Miyahira: Thank you for that. What does it mean that tumor cells increased in certain sites like the bone and lymph node after combination therapy while decreasing in other sites, soft tissues such as liver and lung?

Aleksandar Obradovic: Thank you for that question as well. I think, calling back to what Jess just said, this relates to two separate issues about the differences between these different metastatic sites. One is that the baseline microenvironment characteristics of these metastatic sites are different from site to site. And so the response to combination therapy with respect to how well that therapy is engaging the immune system against those tumors differs by the baseline characteristics of those sites to begin with. And then the tumor cells that are surviving are also phenotypically variable and may also likely have different tropism to those different sites. So part of this is which tumor cells seed those sites, and the other part of this is what the microenvironment of those sites looks like. So when you're giving a combination immuno and hormonal therapy, both of those factors come into play, and it's impossible to truly deconvolute the effects of both in this study. But it's something that's definitely worth digging into further.

Andrea Miyahira: Really interesting. And did you examine the impact of ADT plus or minus anti-PD-1 on PD-L1 expression in the tumor microenvironment, or did you look at other targetable checkpoints, for instance, B7-H3?


Aleksandar Obradovic: Yeah, so one of the nice things about this dataset is that it serves as a resource for other investigators to also interrogate checkpoints of their own interest, checkpoints that have been previously described, checkpoints that are newly being described, and probe those further. We did look specifically at PD-L1 and we looked at B7-H3. The PD-L1 expression that we observed here in the tumor cells themselves was relatively flat. There wasn't a component, a sub-cluster of tumor cells that was defined as a PD-L1 high cluster. There was PD-L1 expression in the myeloid cells to varying degrees as well that can be dug into further. But there wasn't a specifically up or down regulation of PD-L1 in the tumor cells induced by this treatment.

However, the B7-H3, there were changes induced in B7-H3 expression, which also isn't only present on the tumor cells. And so part of the value of this dataset is going to be digging into some of those further because these checkpoints we find are not exclusive to the tumor cells or are not exclusive to the T cells. And the cleanness or specificity of those checkpoints is variable. B7-H3 in these data in particular, to the extent that we've examined it, we find mostly on the endothelial cells.

Andrea Miyahira: Thank you. And what next steps do you think are most promising toward the development of ADT plus immunotherapy for advanced prostate cancer?

Jessica Hawley: I can take that, Andrea. I think what our study really highlights here and what's exciting to us is that we saw a lot of activity and a lot of changes in the tumor microenvironment in the hormone-sensitive setting, which I think a lot of work has been done in the castration-resistant setting, at least in terms of clinical trials. And as you know, we haven't appreciated a huge impact for patients yet, but we're hopeful that this dataset will provide a strong foundation for further clinical trial investigation in the castration-sensitive period, peri-ADT, peri-castration, to sort of leverage the changes that we are seeing in the tumor microenvironment.

So how you give all of that therapy to a patient safely, I think, is another important area of investigation. But there's so much work happening with bispecific antibodies and patient safety in the late-stage setting, and there are earlier single-agent studies happening in the castration-sensitive setting. So, I envision a future where a lot of these different novel agents are being moved earlier in the disease setting, and we may indeed get to a cure.

Andrea Miyahira: Okay. Well, thank you so much, Dr. Hawley and Dr. Obradovic. I think this is a really rich dataset, and it's nice to see explorations of the tumor microenvironment on this level. So, I look forward to what we're going to learn next.

Jessica Hawley: Thanks so much, Andrea.

Aleksandar Obradovic: Thank you very much.