Novel Immune Checkpoint Axis To Understand BCG Resistance and Improve Treatment in Non-muscle Invasive Bladder Cancer - Amir Horowitz & John Sfakianos
May 10, 2021
Amir Horowitz, Ph.D., is an Assistant Professor of Oncological Sciences and a member of the Precision Immunology Institute and the Tisch Cancer Institute at the Icahn School of Medicine at Mount Sinai, New York, NY
John Sfakianos, MD, Assistant Professor, Urology and Urologic Oncology, Icahn School of Medicine, Mount Sinai, New York, NY
Ashish Kamat, MD, MBBS, Professor, Department of Urology, Division of Surgery, University of Texas MD Anderson Cancer Center, President, International Bladder Cancer Group (IBCG), Houston, Texas
Ashish Kamat: Hello, and welcome to Uro Today's is Bladder Cancer Center of Excellence. I'm Ashish Kamat, Professor of Urologic Oncology and Cancer Research at MD Anderson Cancer Center in Houston, Texas, and it's my pleasure today to invite Dr. Amir Horowitz, who is a part of the Department of Oncological Sciences, and Dr. John Sfakianos, who is a part of Department of Urology at Mount Sinai. They're both joining us today to talk about novel immune checkpoint axes in non-muscle invasive bladder cancer and enlighten you, the audience, on what's new and upcoming in the field. With that, guys, the stage is yours.
John Sfakianos: Thank you. Thank you very much, Ashish, and to Uro Today for giving us the opportunity to discuss a little bit of non-muscle invasive bladder cancer today. We'll quickly try to go through a story that we've been developing over the last few years about a novel immune checkpoint access, specifically the HLA-E and NKG2A axis in non-muscle invasive bladder cancer, and as it relates to BCG and BCG resistance.
I'll start off by discussing a little bit about the actual idea that we had with Amir about how we can answer some of these questions, and really, this was one of our first diagrams that we came up with and we've used now, subsequently, in a lot of study designs. The idea was, how can we maximize our patients and our sample collection in a way that it can give us the most information from a variety of different novel immunological experimental techniques that we have available to us?
You can see here, we really took the idea of six induction BCG treatments, and then follow them with maintenance, as we do give to the majority of our patients here at Mount Sinai, and really tried to collect tumor blood and urine from these various time points and use them to investigate BCG therapy, but also non-muscle invasive bladder cancer in detail so we can understand it in a way that we can maybe target it.
In some of these experimental techniques, as you can see here, includes spatial sequencing, mass cytometry type of imaging, CyTOF, whole exome sequencing, something called Olink, which is a proteomics approach to serum and urine. And we've now been able to do this and have accumulated enough samples in patients, where we're actually getting some good data. I'd like to then pass it over to Amir, who can then go over the highlights of some of the results from our experiments.
Amir Horowitz: Sure. Thanks, John. And thanks very much, Ashish, and to Urology Today for the very kind introduction. I'll only present some vignettes from within the story to keep to time, but essentially what we did, my background is as a classical immunologist where I study innate immune system, specifically natural killer cells, and the ways that regulate natural killer cells, it forms this immunological conundrum when you consider what drives a CD8 T cell to respond. The HLA class 1 system presents peptide antigens to CD8 T cells. Many of these same alleles encode proteins that contain conformational epitopes that engage NK cell receptors, many of which are inhibitory in function.
And so what we did here was we compared a cohort of non-muscle invasive bladder cancer patients before and after BCG therapy. We further compared these cohorts to muscle-invasive disease as well. I'm showing you here one snippet of data. It's representative from a single patient that was BCG-resistant, so from the time of cystectomy to specimen. We use markers like EpCAM and pan-cytokeratin to identify the tumor, the tumor micro environment, and what you can see right off the bat is that in non-muscle invasive disease, we found two predominant subtypes, which based on these markers, we're calling luminal and transitional.
One of the first things that stood out to us, very clearly, was that HLA-E, which is a non-classical HLA molecule, it was found at just normal levels on the luminal tumors, there was nothing special about it. However, on these transitional tumors, it was acutely upregulated. We did not see this at all in muscle-invasive disease, we did not see this at all in the pre-BCG time points.
We used fibronectin as a marker for fibroblasts, but also, they're known to lack expression of HLA-E, so it offers some contour. What was most striking to us in this setting, and the caveat here is that where these are small cohorts, so these are exciting preliminary signatures, but we are finding almost 100% concordance in these cohorts where the HLA-E bright tumors are showing increased infiltration of CD8 T cells and NK cells. Shown here on the bottom right plot, you can very clearly see strong infiltration by CD8 T cells. Yet, if you were to zoom in on these tumors, they're not killed. And so, something's getting these T cells and NK cells in, but they're not doing the job when they get there.
And so we are essentially, as of now, the only lab that is identified a reliable antibody clone to profile the receptor NKG2A, which is the cognate receptor for HLA-E. What this is showing is a consecutive section, essentially just 5 microns away from the immunohistochemistry, where if you see on the left side around nine o'clock, you have that black absence of void, this is essentially showing where the architecture on the IHC staining is. All of these white arrows are pointing to NKG2A-positive cells. What this showed us was that, while something about HLA-E is actually attractive to CD8 T cells and NK cells, there are actually many of the cells that are infiltrating are expressing this inhibitory NKG2A receptor, suggesting that would not be a good outcome in the context of combating these tumors. This is just a representatives section from a muscle-invasive tumor, just to show it's very similar levels to what we find on the luminal tumors in the non-muscle invasive disease.
The other things that we had access to was the ability of doing what we call in situ hybridization, so working with a company called HTG Molecular. They offer these predefined panels. In this one, it was custom built for us based on approximately 2,900 genes. And so, it allows us to do just bulk RNA sequencing in situ on the actual tissue. Here, again, it's a relatively small cohort, but we compared matched samples taken from BCG-resistant patients, both before therapy, as well as at the time of cystectomy after tumor recurrence, and then we also compare them to a separate cohort of BCG non-recurrent patients, where the median time of recurrence in this cohort was about 3.9 months following the start of therapy.
And very counter-intuitive to a traditional immunologist, is that where you see the gene pathways that are maximally enriched and upregulated in the post-BCG setting, these are prototypical anti-tumor responses. This really didn't make any sense to us at the time, and so the most up-regulated pathway in these samples was interferon gamma signatures. You can see here signatures for cytotoxic CD8 T cells, tumor inflammation, checkpoint modulation.
What we believe this is suggesting at this level of the analysis, is a picture of adaptive resistance by the tumors. And so, this is this long-standing, not really a debate, but discrepancy in our understanding that tumors may be recurring for reasons completely unrelated to a lack of response to BCG immunotherapy. And in fact, that's what we see. We are seeing a very strong local immune response followed by systemic immune activation in all patients, yet there seems to be this discriminating signature that, due to so many repeated doses of intravesical therapy, you're inducing chronic activation, which leads to immune dysregulation and eventually leads to loss of function or exhaustion.
What we had from the same exact patients was access to their serum and their urine, and so, as John mentioned in the introduction, we performed this analysis called Olink proteomics. It's essentially fancy qPCR that lets you multiplex 92 analytes. What we show here is a collection of 8 non-recurring patients where you can clearly see type 1 cytokines as defined by Flt3 ligand, IL-12, CCL4, CCL3, IL-8, et cetera, and they are in front of interferon gamma. So this is what we believe is a good, a healthy, immune response to BCG, where interferon gamma comes up at some point on the back of a healthy primate. And the patients, we don't have time to dive into all the data, but the patients that are recurring, they seem to have interferon gamma production much earlier, and we think that that's a very important signature for people to start following.
We took from a similar cohort of non-muscle invasive bladder cancer patients that were put on BCG treatment with John, and we took the tumor tissue and we isolated the tumor infiltrating lymphocytes. I have, growing in my lab, Mycobacterium bovis BCG. It's not the same strain as what's used in the vaccine, but biologically, it offers that organism to study, how does this stimulus shape these tumor-infiltrating lymphocytes here? What we found was really quite striking.
We ran analysis by CyTOF, where we had 44 antibodies. Shown here is a collection of these antibodies targeting different markers of CD8 T cell differentiation, T cell activation, as well as some of these NK cell-like receptors. And then where you see functional markers, we define the clustering based, essentially, on these four functional markers. XCL1 is a very important chemokine that activates and mobilizes antigen cross-presenting dendritic cells. We know these are critical cells for a priming CD8 T cells and presenting neoantigens to them. Granzyme B is a marker of cytolytic killing, and then TNF and interferon gamma are these two hallmark antitumor cytokines.
What we found was five very clearly specialized clusters or subsets of the CD8 T cells that seem to be defined on different patterns of their functional attributes. When we took a careful look at these with respect to NKG2A expression, shown to the left of the heat map as those percentages, in each one of these clusters, NKG2A Is expressed. So when we looked at each cluster based on the functional response, for example, clusters 2 and 3, which are defined exclusively by XL1 expression, we then asked, is the signal coming from the NKG2A-negative subset of CD8 T cells, as the blue dots, or rather from the NKG2A-positive?
And we find that it's universally, for XCL1 in clusters 2 and 3, as well as for TNF alpha and interferon gamma, coming from the NKG2A-positive subsets. And so, we believe that these cells are proven to be the most activated, and, in the process, what we're chasing down now, is are they officially becoming immune dysregulated in the face of tumors recurring, such that the tumors can be responding to the interferon gamma signaling and increasing expression of their own checkpoints as an evasion strategy?
Just in conclusion, essentially what we have here is developing a model of tumor resistance to BCG therapy that is largely mediated by interferon gamma production. If we just follow the simple schematic based on the red numbers, number one, TURBT of non-muscle invasive cancer, as most of the audience here knows, these are the tumor types that patients are eligible for. Once the tumor is resected, John will administer intravesical instillation of BCG. It's important to note that the majority of the surface area of the bladder epithelium is not tumor, and we know that BCG readily infects bladder epithelial cells.
So what we're finding is a very strong local immune activation in the bladder, that is largely mediated by dendritic cells, macrophages, CD4 T cells in the face of type 1 cytokines, and this is leading to systemic activation of NK cells and T cells, and causing them to be recruited to the bladder tumor microenvironment. They have been chronically activated over so many doses of repeated treatment, they are increasing their own checkpoints. And in our analyses, we're finding predominantly NKG2A and PD-1 are going up, and this is correlating with robust production of interferon gamma as they're coming into the tumors. So, where tumors are recurring for whatever reason that they're recurring, they're responding to the interferon gamma signaling. And we have data that we just didn't have time to share, in vitro data treating primary tumors with both BCG, as well as recombinant interferon gamma, showing these exact consequences occurring.
In light of all of this, what we're finding is that NKG2A-positive NK and CD8 T cells are enriched in bladder tumors of patients treated with BCG. So the question is, is there actually a strategy here that we can exploit these cells, these pathways to better realize their antitumor functions? And so, where we sort of find ourselves now is trying to consider repositioning combination checkpoint blockade in settings of BCG therapy, where, at the beginning of 2020, for the first time, in more than 40 years, the FDA approved the use of pembrolizumab in settings of BCG-resistance for carcinoma in situ. The question is, perhaps by adding in NKG2A blockade, this may prove to be a very effective strategy in BCG resistance settings.
John Sfakianos: Yeah. I think, from the clinical perspective, the most striking part of all this is really the idea that the chronic inflammation may actually be doing harm to patients, instead of necessarily benefiting patients. We've seen this in the past with a variety of different models, especially with Matt Albert's work using a mouse model and priming the immune system with BCG versus not priming and the timing of the priming and the timing of the BCG instillation.
So I think we've used six doses historically, and Ashish probably comment on this more than I can anecdotally, because from what I understand, six vials came in a box, but the idea is that some patients it may work with, some patients it not. And so I think that there's this idea that we need to better understand how much inflammation is actually in the patient's bladder prior to treating them, and this is the concept where maybe some patients will do better with chemo, some will do better with BCG, and those who have a lot of information and don't do well with BCG, there are these novel checkpoint axes that we can target and it's going to take more than one target, so combinations of two or three, I think, to really knock down the inflammation and the immune response to be appropriate for fighting these cancers.
I think this is really exciting because we're starting to look at it and think about it, maybe not only has the mind of a urologist, but the mind as an immunologist combined. And when we go back and forth, there's a lot there that we don't think about as urologist, and immunologists don't think about, so this combination has really been fruitful.
Ashish Kamat: Now, thank you guys. That was a very succinct presentation of a lot of data and I want to thank you for keeping the time. It's interesting as we progress. BCG has been used forever and we've always lamented to the fact that it's poorly understood, except for the fact that we actually understand it really well, we just don't have the tools interrogate the axes. And now that we have the appropriate tools, we're finding things such as the NKG2A and some folks are looking at NKG2D and the monoclonal antibodies and what have you. So it's really exciting to see this kind of work being brought to the forefront and the collaboration between the two of you is really, really great.
A couple of questions, in interest of time I won't dwell too much on it, but a couple of quick questions for our audience, just so we can understand. Are you proposing that the NKG2A expression and the inhibitory receptor interactions is specific to antitumor activity or BCG? Or would you postulate that that would be something that would regulate the efficacy of BCG across all its different immunological functions?
Amir Horowitz: Yeah, so if you take these actual T cells or NK cells, throw them in a dish, and just give them cytokines for 6 days, you see the exact same pattern emerge. It's in response to the tumor, it's in response to BCG. Clinically, it's a very hard situation to tease out precisely what's driving what. But what you can say conclusively is that in this setting, we're seeing both local and systemic immune activation that's driving increased expression of NKG2A, and it's standing in the way of efficacy of BCG therapy as tumors recur.
I'll just add onto this briefly, when you're taking a cystoscopy every 12 weeks, for all we know the BCG non-recurrent patients really had a recurring tumor and we just missed it. The immune system cleared it. I'm not necessarily proposing that's happening, but what is clear is that everybody is responding to BCG and that tumors are recurring in some and not in others. And when they are recurring, they're definitely responding to the activation, which is leading to increased checkpoint expression.
John Sfakianos: What I'll add to that, Ashish, especially for the urologist, what this got me doing is I pay attention more to what's actually happening in the bladder at the time of your initial TURBT. You go into these bladder sometimes and you just see all this bullous, edematous tissue, and even non-tumor portions of the bladder are all inflamed. And then you'd go into these other bladders where it's the papillary tumor or the cecal tumor, and just that area is what you see is inflamed and the rest of the bladder looks actually pretty good. To Amir's point, I think that there is a combination of tumor inflammation or changes in the water because of the tumor, as well as because of our treatment. It's a combination of both, where I feel like the already inflamed tumors might not do well with BCG because they've already been exhausted and we're just making it worst.
Ashish Kamat: What you're saying is actually an absolutely correct. One of the mistakes that essentially the urologic community has made for many years is, for example, you give 6 weeks of BCG and then you look in the bladder 6 weeks later and the patient had persistent cancer, that was called a BCG failure. But if you actually look at the studies, you just wait another 9 to 12 weeks and one-third of those tumors just disappear. You don't do anything. You just wait. You let the immune axes correct itself, the Th1, Th2 respond. So what you're proposing actually makes perfect sense.
I could chat with you guys for a long, long time, and maybe we'll do that off the record, but the interest of time, let me give you guys both the closing stage. So quickly, from your point, John, and then Amir, I'll let you finish up. What are some of the closing thoughts you want our audience to take away from your findings?
John Sfakianos: I think the idea that I'd like for everybody to come across is that we need to start changing our thinking. I think, like to your point, Ashish, were all taught, you do 6 weeks of BCG and then you get a cystoscopy 6 weeks later and then you do maintenance and then you just move forward. Well, I think, and this concept has been brought up before by many urologists and many scientists, but we need to start thinking about, is 6 weeks of BCG, really what we need to give to patients.?Is there a way that we can think about either combining with chemotherapy or some patients benefiting from chemotherapy versus others for BCG, just from a clinical standpoint? Is there inflammation? Is there not inflammation? I think we need to really start pushing the field forward a little bit more by really questioning the historical concepts and data and moving forward as a group.
The biggest thing I think that we need to do, and I know that many people are trying in the back, is really as urologists, come back together and build these banks of blood tissue and serum, because not one place has enough patients where you can actually move the field forward with quick timing. It really needs everybody come together and really think about this differently and move the field forward.
Amir Horowitz: Yes. I couldn't agree more with what John said. As a non-clinician and a scientist, I'm very passionate about the work, we've got a really exciting signature, and of course I want to get access to more material. What I would say that strikes me in closing that I'd like to impart is that T cells and NK cells serve highly redundant functions, but they operate at very different windows of immune activity. And so, one of the hallmark differences these cells is that T cells require prior sensitization and NK cells do not. And so, what we're seeing here is almost this paradigm shift that T cells are starting to appropriate NK biology. And this is a very novel axis that the world has been positioning today, exclusively around a double whammy inhibitory signal. Yet, it positions ourselves that when we understand how HLA class 1 is modulated, which is a very frequent event on tumors, and we can profile these innate NK cells alongside these adaptive T cells, we can start understanding how they work in concerted efforts and really start breaking this down.
I would love to see a study starting to systematically address the dosing criteria for BCG and allowing us to, in a stepwise manner, piece together what's going on before BCG therapy, while on BCG therapy. Yeah.
Ashish Kamat: Great. Once again, thank you guys both for taking time out from your busy schedules and spending it with us talking about this for our audience. I really hope that we can get to see each other in person sometime soon, but until then, guys, stay safe and stay well.
Amir Horowitz: Thanks so much.
John Sfakianos: Thank you.