Read the Full Video Transcript

Andrea Miyahira: Hi, everyone. I'm Andrea Miyahira at the Prostate Cancer Foundation. With me today is Dr. Ece Eksi of OHSU. She will detail the results of her recent paper, multiplex imaging of localized prostate tumors reveals altered spatial organization of AR-positive cells in the microenvironment. This was published in the open-access Cell Press journal iScience. Dr. Eksi, thanks for joining us.

Ece Eksi: Thank you so much, Andrea. It's a pleasure to be here. So today, I will talk about our recent results on spatial profiling of androgen receptor cells in the prostate tumor microenvironment. And I would like to start by talking about prostate cancer diagnosis and patient stratification. Currently, we're using histopathology to stratify patients into risk groups.

And this is based on an H&E evaluation of a tissue specimen. This could be a tissue biopsy or a radical prostatectomy. And then we assign risk based on some clinical parameters, as well as the Gleason grades and Gleason score. And with the rise of AI-based digital pathology platforms that are out there, such as Artera AI or Paige, we're beginning to capture more textural and architectural components from these H&E images, but we're still relying on an H&E image.

And I will argue that even if we use the most sophisticated pipelines that are out there, this H&E image does not capture the vast levels of molecular and cellular heterogeneity that we observe in these prostate tumors. Prostate cancer is one of the most heterogeneous cancer types that are out there.

And with the developments in single-cell sequencing, as well as imaging, we're beginning to uncover how much variability exists in important disease markers such as ERG fusion or PTEN loss. And it's important to capture these spatial landmarks and single-cell-level spatial associations to better stratify patients and also identify more precise treatment options for patients.

So in this paper, we used 32 protein markers to analyze these treatment-naive localized prostate tumors, and identified 27 cell types in prostate tumors. And we used something called recurrent cell neighborhoods. So these cell neighborhoods are units that show organizational structures between cells.

And the way we define these cellular neighborhoods is we take a single cell and look at the frequency of cell types around that single cell, and record the frequency of these cell types as a vector. And we do this for each and every single cell for more than 700,000 cells in this paper, and do a non-hierarchical clustering of these cell-type frequency vectors, which then gives us these recurrent cellular neighborhoods that exist in localized prostate tumors across patients.

And one of our surprising results is that we observed that these cellular neighborhoods in prostate tumors are driven by androgen receptor expression in the stroma. So we identified several different immune and non-immune stromal cell types, and the organization of these cell types was the key in defining these neighborhoods in prostate tumors.

And these were also the only neighborhoods that showed significant differences based on clinical grade. Then we went beyond these neighborhoods and looked at some of these cell types and their single-cell associations with other immune cells in the microenvironment. Mast cells are one of the most popular immune cell types in prostate cancer patient samples.

And we identified these AR-positive mast cells that show spatial associations with M2 macrophages in the tumor microenvironment. So here, as you can see, with distance, all of these single-cell associations look random or like noise. But at x = 0, we see this close association between AR-positive mast cells and M2 macrophages.

Using another spatial statistics tool called Ripley’s L, we were then able to show that AR-negative mast cells in turn are closely associated with Tregs. So in summary—and this is really a resource paper for the prostate cancer community where we can look at these different epithelial and stromal markers and see how their associations are changing at the single-cell resolution level—but I also want to state that this type of spatial imaging and analysis gives us a chance to then look at the emergent properties that arise from studying the epithelial-stromal crosstalk in these tumors.

And my take-home message is— the first one is we've been testing and validating hundreds of antibodies in the past three to five years. And we published all of the clone information with this paper. It’s in Supplementary Table 2. So please feel free to reach out to us. This is a growing list of antibodies and antibody panels. And antibodies are expensive. There's no need to repeat some of that research.

We’re happy to share all of this information with the community, and I want to emphasize that there are a lot of different types of AR-positive stromal cells in these prostate tumor microenvironments. And the expression profiles for AR change the spatial associations and likely functions of these cells in the microenvironment. And the stromal AR expression is really a key driver for clinical grade-related changes. And thank you so much.

Andrea Miyahira: Thank you so much, Dr. Eksi, for sharing this really interesting study and such a great resource for the community. So can you link any stromal cell organizational patterns in primary prostate cancer that you observed in different patients with tumor genomic alterations?

Ece Eksi: Yeah. So we only had genomic alteration data for 10 patients in this study. So the correlations that we derive are very anecdotal for this specific study. But we are very interested in expanding this work to more patients and larger clinical cohorts. Looking at these 10 patients, we had only one patient that had an RB1 loss and a TP53 loss.

Even though these patients had very similar risk scores when we looked at it’s [INAUDIBLE] and had very similar Gleason grade 4 pattern, it had a drastically different neighborhood structure that was driven by macrophages and an AR expression that was different than the other tumors. So we're very interested in looking further into how RB1 loss in these localized patients before applying any treatments may be showing different neighborhood organizations.

Andrea Miyahira: Thank you. And what more is known about the role of mast cells in prostate cancer? And do we also observe them in tumor metastasis?

Ece Eksi: Yeah. So I've been looking a lot at studies published by Dr. Sfanos' lab in Baltimore, Johns Hopkins. And she has some amazing work that characterized these mast cell populations in different clinical cohorts, including cohorts with African-American men, and showed that there are intratumoral mast cells and extratumoral mast cells. And they may have different functions. And these extra tumoral mast cells are associated with biochemical recurrence and metastasis in the studies that she published.

And she finishes her paper by saying that there are likely heterogeneous expression patterns for these mast cells. But with the spatial imaging tool that we have, we were able to show that these mast cells express different markers, including AR and CD44, which can bind to hyaluronic acids and be involved in extracellular matrix reorganization. They express CD90, which is involved in their maturation. So we definitely see these different populations of mast cells that her previous studies hinted at. And if I could speculate, I would say that these extratumoral mast cells that she was observing in her studies were likely AR-negative.

Andrea Miyahira: OK. And I know you've observed AR-positive stromal cells in your study. So do you also see these cells in healthy prostate tissue? And do you know what role they might be playing in tumor progression?

Ece Eksi: Yeah. This is an excellent question. We definitely see AR expression in healthy prostate. So we had large regions of tumor-adjacent normal tissues in some of these patients. And we see AR expression in mast cells and also in other immune and stromal cell types. What is interesting is that these AR-positive stromal populations are showing different patterns with disease progression.

So I think we really need to delineate what each one is doing. So it’s beyond a collective behavior. It’s possible that Tregs that are positive for AR have different function than mast cells that are positive for AR. And we really need to delineate some of these results.

Andrea Miyahira: OK. Thank you. And what are your next steps?

Ece Eksi: Well, currently we're expanding this spatial imaging study to larger clinical cohorts—specifically, active surveillance patients—where we have longitudinal tissue biopsy samples from the same patients and really observe how the tumor microenvironment is changing with progression versus no progression so that we can identify more precise stratification tools for prostate cancer patients.

Andrea Miyahira: OK. Thank you so much for sharing this study with us today, and I hope people will take the time to check out this paper and all the amazing resources. Thank you.

Ece Eksi: Thank you so much, Andrea.