Read the Full Video Transcript

Andrea Miyahira: Hi, everybody. I'm Andrea Miyahira at the Prostate Cancer Foundation. Here with me is Dr. Steven Kregel from Loyola University Chicago. He will discuss his paper, "Clinically Relevant Humanized Mouse Models of Metastatic Prostate Cancer Facilitate Therapeutic Evaluation." This was recently published in Molecular Cancer Research. Dr. Kregel, thank you so much for joining us today.

Steven Kregel: Thank you. This is a pleasure. This is work that we've been working on for quite a while now, that recently got published by my group. I'm a new assistant professor at Loyola University Chicago. I'm just going to walk you through what we've been doing, the problems with modeling prostate cancer, humanized mouse models for metastatic prostate cancer, responses to AR-targeted therapy and immune checkpoint inhibition in humanized mice, and then I'll finish up with conclusions, challenges, and future directions.

So I kind of start most of my talks with this slide. The problem with modeling prostate cancer in vivo is that humans and mice are just fundamentally really different. You can look at human prostate here on the left compared to the mouse prostate on the right, and you can see that just grossly, there's big differences. The human prostate is kind of this one homogenous organ, and the mouse prostate has multiple lobes, and these lobes have different gene expression and different responses to androgen inhibition. Whereas in the human prostate, it's mostly, again, more homogenous. But we have different zones within the prostate.

And during carcinogenesis, humans get prostate cancer all the time, and mice rarely develop prostate cancer. And you have to give a lot of genetic alterations to mice in order to drive prostate cancer development. And ultimately, when they do develop prostate cancer, it's very rarely metastatic. And human prostate cancer almost always goes to the bone. And mouse prostate cancer, it often loses AR expression as the mice progress, and it doesn't respond to AR-targeted therapies. So what we do is we usually rely on a lot of xenograft models where we take an immunocompromised mouse and we inject human cancer cells into the mouse to allow it to grow. And those models are nice because they metastasize and they'll respond to AR-targeted therapy, but you're really missing the really important components of having an intact human immune system. So that's kind of what brought us to what we're doing now with these humanized mice.

So we got mice from Taconic, and they have a platform where they can take human hematopoietic stem cells and engraft them into severely immunocompromised mice. And those mice actually over time develop most lineages of the human immune system. And this is just the baseline model where you can actually see a lot of human CD45 cells that are maintained. And this is about 12 to 16 weeks after the initial injection of the hematopoietic stem cells, and actually they have different backgrounds. And that's what we used for our paper, where we have just the standard human, the standard immunocompromised mouse with an engraftment of hematopoietic stem cells. And then another background that they have that actually has human IL-3 and human GM-CSF knocked into the mouse background. So then that allows for a lot of the myeloid support that you see.

So the mouse myeloid cells kind of dominate in the standard model. And then in this other model called the huNOG-EXL, there's a lot more mature myeloid and granulocyte cells. And actually, we see that these models ultimately respond differently to things like AR-targeted therapy. When we give them human cancer xenografts, in the standard model, T-cell infiltrate tumors, and it looks like a hot model where enzalutamide activates these T-cells, they're responsive to immune checkpoint. And in the huNOG-EXL model, the myeloid cells really kind of create an immunosuppressive environment, and you see a lot of them in these tumors and there are fewer T-cells overall. And the T-cells that you do find are more T-reg or naive stem cell-like.

So let's get into some of the data that we have. This was in our best bet model where we have just the T-cell skewed model where we don't have a myeloid support. We actually can follow up metastasis by luciferase tagging in these cells and grafting them subcutaneously in the backs of these mice, and then allowing for spontaneous colonization of multiple organs. And we used in this experiment 22Rv1 cells, which are just a typical prostate cancer cell line, and it's of human origin and it's very bone metastatic. So you can see over here on the right, these are the femurs that we've taken out ex vivo and imaged them, and we can actually see where the cancer cells have gone to. And we can quantify this with the luciferase.

And we see in this model, the immunocompromised mice, which are these NOG control mice, actually get more metastases as you increase the level of AR inhibition through castration in the red here and then finally in castration plus enzalutamide treatment. And in the presence of an immune system, which is from two different humanized donors, you actually can block the colonization. And we did this in a different approach with a different cell line. These are VCaP tumors. And we treated these tumors with both enzalutamide and pembrolizumab after we castrated them. And pembrolizumab is the immune checkpoint inhibitor PD-1. And in our NOG controls, which are our immunocompromised mice, we don't see any difference in any of the growth in any of these conditions. But in our huNOG mice, where we have the immune system, we actually block growth with enzalutamide and pembro alone. And the combination actually completely clears the tumors from these mice. So we can see that our treatments can actually impact the immune system of these mice, and it mirrors what we see in patients a little bit better since we have actual effects with enzalutamide.

So in summary, we have a human cancer model and it's capable of responding to standard-of-care AR-targeted therapies. Because it is a human model, it metastasizes in mice to clinically relevant locations such as the bone. And it has an intact human immune system, and you can kind of adapt it or modify it depending on which backgrounds you want to use. And it can model many distinct human prostate cancer models. So we used two different prostate cancer xenografts in our paper. And it models both the hot setting, which is kind of the rarer setting in prostate cancer, as well as the cold setting, which I think is more common, meaning that there's less response to immune checkpoint inhibitor and it's more myeloid cell dominant. And I think it really has the potential to facilitate a lot of these drug development questions that standard immunocompromised mice don't seem to be able to address, as well as how GEMs, which are often not as different in terms of their tumor alterations and how many neoantigens, for example, they seem to accumulate over time—so genetically engineered mouse models.

So the big take-home message is that these data suggest that humanized mice better recapitulate the patient castration and enzalutamide responses. And we see decreased metastasis with enzalutamide treatment, particularly in this huNOG background, which is more consistent with what we see in patients with non-metastatic CRPC. That's kind of, I think, one of the best examples of the utilization of these mice.

So the challenges with using these: they're very expensive. That's one of the things. And it's a lot of moving parts in the system and has a more competent immune system. But it's not completely, completely intact. It doesn't develop with a human thymus or liver or all these other organ sites that are necessary in a normal functional organism. And we could HLA match, meaning that we can adapt the MHC class I molecules in particular to the donor. However, in these cell lines, there's very little expression of the class I, meaning that there's probably very little presentation. There's probably other things that are happening, and we're trying to investigate those now.

And there are sometimes graft issues with these hematopoietic stem cells where you do get graft-versus-host disease that happens in the mice. So these grafts, Taconic has a pretty good hold in this where they have models that can maintain their grafts for a really long time. And you have to irradiate these mice to get the graft to take. So ultimately, you do get a reduced lifespan, but with these metastatic models, it doesn't matter because the prostate cancer usually progresses pretty rapidly.

So with that, I'd like to thank everyone. I especially would like to thank the Prostate Cancer Foundation, which has really championed this work and has funded me as a young investigator. And I'm sure they're probably sick of me talking about this all the time, but I really appreciate their support and all of my colleagues and collaborators. And if you have any questions, you can contact me.

Andrea Miyahira: Well, that was terrific. Thank you so much for sharing that with us, Steve. So can these models be used to evaluate differences between efficacy of various non-immune cancer therapies when you want to compare your immunocompromised mice with your mice with an intact humanized immune system to find out if there's any immune contributions to efficacy, say for radiotherapy or radioligand therapy, etc.?

Steven Kregel: Yes, that is the goal. I think we're really not appreciating how much the immune system is playing a role in the efficacy of these therapies. I will say that the background, the NOG background, that is a common gamma chain knockout, is really sensitive to radiation. And when they do irradiate, they're very careful with how much they dose, and they don't need a lot to clear the mouse bone marrow in order to allow for the engraftment to occur. So when you do use radiation in these models, you have to dial down the dose because the background itself is pretty sensitive. So that's the one thing that I think is a little bit lacking in this model in particular is that the radiation and the radioligand responses are probably going to be a little bit skewed because you're going to get some toxicity. But in terms of other therapeutics such as PARP inhibitors or other things that we're trying to investigate, I think they'd be really useful for that.

Andrea Miyahira: Are these models restricted to human prostate cancers with certain HLA types?

Steven Kregel: So you can match them. A lot of these companies—I worked with Taconic, but there are other groups out there that also run similar models—you can match them if they have the donors. And they usually have a pretty extensive bank of donors. So it's as simple as giving them a genotype and saying, "Hey, do you have this HLA?" And they'll come back and say, "Yeah" or "No" or whatever. But for me, I don't know how much that response is necessary because especially in prostate cancer, 80 to 90% of prostate tumors basically lack HLA expression in the metastatic setting. So maybe we'll see a primary response. But in patients, we see a massive downregulation.

But I will say that there are other groups, including our group that's kind of looking at the androgen control—I know Felix Feng's group is looking at how HLA is regulated by the androgen receptor because the androgen receptor does everything in prostate cancer. And actually, I think if you do inhibit really strongly, you do get increased HLA expression. But I'm sure the cancer has other ways of figuring out how to evade the immune system. So that's, I think, one of the kind of exciting and tricky things in terms of treating patients. But one of the exciting things from a research perspective is to try and figure how this is occurring and how the immune system is being evaded or even co-opted by the cancer.

Andrea Miyahira: Okay, thanks. You talked a little bit about how tough these models are, but for those who are interested, do you have any other tips?

Steven Kregel: Yeah, I would say try... I mean, right now we are in the process of trying to develop a matched setting. But I will say that they are pretty user-friendly. The one thing is that you have to be quick with them. When we do model xenografts and use immunocompromised mice, we usually use six-week-old mice. And that's kind of a really artificial thing because we're modeling basically a pubescent mouse with late-stage prostate cancer, which isn't the case in people. But you can grow slow-growing xenografts or PDX for a really long time in a lot of these immunocompromised mice and they'll live, especially if they're not metastatic, they'll live for up to a year.

You don't have that long when you use these mice because usually they're about three months old by the time that you get them because you have to wait for the graft to take, and the company won't send you the humanized mice unless they know that the graft is good and they see a lot of human immune cells. So you do only get... From that point on, you can get nine to 12 months. And I think after that, things start kind of going off the rails and you do develop some issues ultimately. But for most therapeutic response questions or most prostate tumor models, especially the cell line models which are really aggressive, that's plenty of time for most of the things that you have to ask, most of the questions you'd ask.

Andrea Miyahira: Okay, thanks. And what are the next steps in your studies?

Steven Kregel: Well, we're going to expand which tumors we model. We're going to be modeling PDX; we'd like to develop patient organoids and have matched HLAs and see if we could mirror what we see in the clinic in terms of their responses. And also just expand the therapeutics that we can use to see ultimately how the immune system is affected by these drugs. We know that the immune system seems to be sensitive to androgens and AR antagonists, at least in sort of anecdotal systems, and some work out of Amy Moran's group at OHSU. They've really shown that there are these responses, and I think that we don't really appreciate them as much. And especially in a patient that comes in that's been castrate for so long, these changes have adapted or whatever has occurred, and the cancer's got around it.

But if we hit that window maybe of a combined inhibition of the androgen receptor plus a checkpoint, maybe in a pre-castration or even a non-metastatic setting, maybe we can induce an anti-tumor response that would be durable that we see in other tumors where they're actually responsive to things like checkpoint inhibitors, where prostate is notoriously bad and cold and does not respond very well, at least for the majority of tumors.

Andrea Miyahira: All right, well, looking forward to all the next studies that will come out. Thanks so much for coming on today.

Steven Kregel: Thank you so much. This was great. It's a pleasure as always.