Scientist Stories: Tullia Bruno, Tertiary Lymphoid Structures and Immunotherapy

Hello everyone, and welcome to today's webinar, Tertiary Lymphoid Structures to the Forefront of Immunotherapy. What are they good for? I'm Christina Jewell of LabRoots, and I will be your moderator for today's event. Today's educational web seminar is presented by LabRoots and brought to you by Akoya Biosciences. To learn more about our sponsor, please visit akoyabio.com. Now, before we begin, I would like to remind everyone that today's event is interactive. And we encourage you to participate today by submitting any questions you may have during the presentation. To do so, simply type them into the Ask a Question box and click Submit. We will answer as many of your questions as we have time for at the end of the presentation. You may also submit any technical issues there as well if you have trouble seeing or hearing the presentation. I would now like to welcome our speaker, Dr. Tullia Bruno, Assistant Professor, University of Pittsburgh Hillman Cancer Center. For a complete biography on our speaker, please visit the presenter tab at the top left of your screen. Welcome, Dr. Bruno. You may now begin your presentation.

Thank you so much, Christy, and thank you to Akoya and LabRoots for putting this webinar on today. It's really a pleasure to be here presenting to you live about tertiary lymphoid structures and how they are changing the landscape of immunotherapy and what we know about them currently. As mentioned, my name is Talia Bruno. I'm an assistant professor of immunology at the Hillman Cancer Center in the University of Pittsburgh, and my lab has been focused in this space for the last 5 years, really trying to think about not only tertiary lymphoid structures, but B cells in and outside of these structures and how they function within the tumor microenvironment. I'd like to start out by drawing your attention to the picture that is on my title slide, because I think that really starts out by setting the framework of what we think about in the field currently. There's different types and tiers of structures within the tumor microenvironment, but within the lab, we use multispectral imaging to tier the structures into 2 buckets at first, at first pass and tumor types. And so immature TLS and mature TLS. And so, The reality is, is that these mature TLS that you see on the right-hand side, they are the rarity within the tumor microenvironment. Actually, mostly what you see in the majority of structures are usually in these more immature state. And what that is defining that is the absence of a germinal center. So, this is going to be a germinal center, and we'll talk more about that later today. The reality is, and what we'll talk about, is that there's actually a very large spectrum of these structures, and it's not just going to be immature, mature, and it's also not going to just be the spectrum of formation and composition, but also the activity of these structures, which we'll start to get into a little bit today. But I think the field is actively trying to answer that question. I always like to start out with my acknowledgments. I'd like to thank my lab, both present and past. So those in italics were previously in my lab. I'll highlight some of, um, all of their work today because they do work as a team, but In particular, I'll be talking about a postdoc project in the lab, um, captained by Ian McFawn. I'd like to thank all of our collaborators because we're pretty agnostic as far as tumor types that we think about for tertiary lymphoid structures. It's actually a strength, I think, to think about these different structures in different tissues, and we'll talk more about that today. And then, because this is really focused on imaging and also some spatial transcriptomics that we're doing in the lab, I want to thank first the University of Colorado Imaging Core that is really— it's really the immune monitoring core there that does a lot of spatial imaging. And they helped us get started on our platform. And now we work with our TPIL core locally at the University of Pittsburgh in the Hillman Cancer Center. And that is captained by Mary Ann Joyce. So, a really big shout out to her. And then also, I'll show some bioinformatics spatially today. So, I want to acknowledge our bioinformatics team and the core that I run and the staff in there that does all of that work. And then finally, the funding to do all of this work. So I wanted to start out by taking a step back and thinking about secondary lymphoid organs, such as a tonsil, a lymph node, a spleen, because this schema really shows what happens within those lymph nodes and within those tonsils. And what it allows for is really optimal immunity. You have this crosstalk and education of B cells and T cells. You have, uh, clonal selection of these B cells to become antibody producers and really high affinity antibody producers. And in the case of cancer, this would allow for targeting the tumor if it worked properly. The T cell also gets, um, educated and relies upon this really sophisticated sort of interaction. And the reason I bring this up is that in these, um, secondary lymphoid organs, you're really seeing a very, very nice orchestration of these structures. You can see it here with the CODEX system that we have in the lab, and we've utilized to do some multispectral imaging with more markers. And, and basically, you can see these really nice germinal centers here that then are proliferating. They have all the key subsets that we think about, and they're really, really nicely formed. This is not what happens in most tumors. In fact, This is, again, as I mentioned, the rarity, but I just wanna put this up here because this is what's gonna generate memory. And I think that the question in the field is, do you always need a mature TLS, or are those more immature TLS and just having a structure important? And so those are some of the questions that we've been thinking about a lot in the lab. So, just defining a tertiary lymphoid structure, I think that there's, 2 ways to define it, and they are quite complex. And so I would say that the left-hand definition is really a more pathological, histological definition that the field uses to really get a quick look at how many structures might be in a tissue before doing multispectral imaging. And in fact, in the lab, we use this as a kind of screening method for tumor types because we wanna know before we run multispectral imaging on everyone if there's actually structures present. And that's important for tumor types where there's not a lot of patients that have TLS. And so, we're gonna talk mostly about ovarian cancer today. And so, that's a good example of a solid tumor type where only about 10 to 15% of the patients even form structures. So, this, um, histological or single-plex definition of a TLS is helpful to get a read of how many structures are in these patients. So, we do this by CD20. Uh, it's pretty accepted in the field that CD20 is a good single-plex marker, which is gonna mark B cells spatially for these structures because those are are the cells that are gonna really gravitate to these structures very, very readily. And they're a major component of them within the human microenvironment. And so what you can see here then is that you can annotate intratumoral, peritumoral, or non-tumor regions. And in the case of head and neck cancer, which this is the snapshot from, that's important because these tissues are gonna have a lot of lymphoid tissues that might be present at baseline because a lot of times in the oropharyngeal space, these tumors will grow in the tonsil or the base of tongue. And so you can then mark with singleplex and then a pathologist can go through and say, okay, this is a germinal center. This is more of an immature TLS. These are sitting on the peritumoral border, which is where most structures form, or they're actually getting into the tumor. Uh, a lot of people also use H&E and have trained H&E to do this. I would say that while you can do that, there's some papers actually by Catherine Slutz-Friedman that recently came out that shows that really having a singleplex at least of CD20 is pretty important, uh, to really understand how many structures you have, and H&E is quite incomplete. So I would say the field is moving away from just that. Where the field is also going is multispectral imaging. So, uh, being that, you know, a lot of what we're gonna talk today about is looking spatially and, and thinking about these structures, you have to use multiple cell types to really identify these structures. They're very complex and there's not just B cells in there. And so this is a snapshot from a paper that was published in 2021 showing multispectral imaging of several different cell types. So CD4 and CD8 T cells, which, um, are also around or in these structures. I would see CD4 or more in, and CD8s are often around. And then CD21 and CD23, which mark follicular dendritic cells, which really govern interactions between B cells and CD4 T cells and are the hallmark of a germinal center forming essentially in these structures. One thing that's not pictured here are high endothelial venules. These high endothelial venules are also quite important. I would say that They're the number one accepted hallmark of these structures, even without having a germinal center. So if you have a high endothelial venule, it's pretty well accepted in the field that you have a structure. So those are the complexities of this different cell types. And so in the lab, we've actually combined several of the hallmarks into multispectral imaging to really try to identify not only the key cell types that I just mentioned, but also to look at activity of those structures. And so Ian, when he came into the lab, he helped captain this panel which was actually a really large group effort. And we put together this panel to look at TLS formation and to look at TLS maturation actually in the tumor microenvironment. And so, you can see here some of the markers that I've already mentioned. So, CD20 and CD4, which are the majority of the cell types. And you can see here pretty clear zoning of those cells. And then what we also have in there is CD21. That's gonna allow to mark for formation of a germinal center., and then high endothelial venules by PNAD. And so we've also layered on within this image AID and Ki-67. So this is where really where the activity comes in. Ki-67 and AID, which I'll show you later, really pair nicely with CD21 formation. So AID's for class switching of B cells and Ki-67 is just for proliferation of B cells and T cells. But we're also finding in the lab that the Ki-67 also just allows for you to look at B and T cell activity within these structures, even if there's not a germinal center. And so, we're seeing that in some of our tumor types as well. So, I'll talk more about this staining panel a little bit later when I talk more about Ian's project. And so, what are some of the other biological— biological, I guess, evidence or data that would say, okay, we know what TLS are and what do they look like within the tumor microenvironment? So, one thing is, is that TLS composition varies. And so, this is a snapshot of the different types of TLS that we see in lung cancer. And so, what I just wanna point out is that I talked about those two buckets that we put structures in, immature and mature. But the reality is, is that there's a spectrum and that not all patients have TLS, even if it is a tumor type like lung cancer where a good portion do. And so, here's a good example of no TLS. And then as the TLS go up, you can see that the cell types here, from the cellular outputs from this multispectral imaging with the Vectra platform actually allows for you to see the different cell types that are in there. And so you can see B cells obviously flux in as those TLS form. The other thing that I think has been really helpful in the field is that TLS often correlate with increased CD8 T cell influx into the tumor and into the stroma. So this is just one example of that, um, pathologically scoring CD8 T cells, but there's been papers that have actually pointed out that it also increases CD8 T cell function. And so some of the papers that came out in Nature in 2020, which we'll talk about here in a minute, they actually showed that when you have more structures or when you look at T cells that are associated with those structures, that T cell activity is better as well. So it's not just the amount, but the functionality of those cells seem to follow suit when a TLS is present. And then finally, some work in our lab from 2021 have shown that the type of structure is quite important. And so Again, we come back to this mature structure here that has a germinal center versus one that is more immature and doesn't have a germinal center. And really what these heat maps are showing, they're kind of small, but the point is, is that these structures actually can govern not only CD8 T cell influx into the tumor, but CD8 Treg interactions. So those tend to go down when you have more sophisticated structures present and within the tumor microenvironment. The other point I would make here is that these more mature structures correlate with better survival. In most solid tumors, and we can talk more about that in the Q&A if we'd like to. So, one thing I would just have you thinking about as we're going through this is that we use TLS as an example, right, of what we want to try to understand as a cellular neighborhood. And that's probably one of several cellular neighborhoods that are present. And so, one thing that I would challenge you with is that, is TLS always the cellular neighborhood that you need? And again, back to the question, is a mature TLS always the cellular neighborhood that you need? And I think those are open questions in the field, but something to continue to think about. Okay, so if the biological data isn't enough, there's a lot of data related to prognosis of patients, as well as response to checkpoint blockade. So this review is quite nice. It has, um, a prognostic benefit for B cells in most solid tumors, and also TLS or CXCL13, which is often a surrogate marker for TLS in a lot of tissue types. And you can see here that the green means that it's favorable, and the level of green means more favorable versus mostly favorable. And then, um, also I think what really put these on the map is that these cells were indicated not only in, in disease like melanoma, but in kidney cell and soft tissue and sarcoma, as well as now bladder cancer and lung cancer, that these cells and that TLS help with response to immunotherapy. And help is kind of a broad term. I really think what these studies have shown is that they correlate with better response to immunotherapy. And I think where the field is going is thinking about how these therapies also regulate these structures, which I think is an important question. And so when these papers came out in 2020, I was invited to write the News and Views. And so that was really fun because the field was really just getting started, I think, in thinking about this, this, uh, kind of question in a very, very large scale. I would say that there's been several people that have been working in this field for much longer than me. And so I think I think this was also quite validating that these structures came to the forefront and they really helped to correlate with response to immunotherapy. This is where we first defined that immature and mature axis, and that if you have more mature or sophisticated structures, you would indicate that there's better T-cell activity as well as better B-cell activity. But I think we're still trying to understand that in, in patients. And now we've got a lot of mouse models, which we won't get to talk about today, that can help us understand this. So, I think the question in the field is, how do we maximize TLS function if they're present? Or how do we basically make TLS within patients? And so, it starts out by understanding TLS within the tumor microenvironment. So, now I want to show a more complete spectrum of what TLS look like in the tumor microenvironment, or how we would hypothesize they look like. And there's the cellular outputs that come from this. And so, this is from a book chapter that was put together by my previous graduate student, Ayana Ruffin. And what we can see here is that there is a spectrum of these TLS. And again, we talked about these very mature TLS, but really, um, this middle kind of group are more of these immature TLS. And that's what we see at the majority in, in the tumor microenvironment. And we also have more rudimentary structures that we've started to coin lymphoid aggregates, which I think a lot of people in the field use this. Uh, and so now we've got 3 kind of buckets. Aggregates that we'll come back to today. So lymphoid aggregates, early TLS and primary follicle-like TLS are gonna be those more immature that are then going to this mature, which are those secondary follicle-like. But one thing that we do know, and this is from some work in my postdoc time, is that when you put B cells and T cells together, and this was in lung cancer patients, and you ask the B cells to educate the T cells with antigen through an antigen presentation assay, that you do get that antigen presentation and you do get that T-cell response. But when we go back and correlate with TLS, we see that the, the patients that had the best activity actually had more structures regardless of the type of map, type of structure they had. So it's pretty clear that these can also help predict function within the tumor microenvironment. And I think more data is, will be coming out on that. And so some take-home messages from today and some things that I've already hit is that the tissue matters. And we'll talk about this a lot today with with the, the data from our ovarian cancer cohort. Cellular neighborhoods are quite important, and I've already kind of emphasized that B cells and T cells aren't going to be created equal, and a lot of this is going to be dictated by where they sit in the tumor microenvironment. I won't get a chance to talk to you about this today, but this is a large effort in the lab, thinking about the types of B cells that are in the tumor microenvironment and where they're located in relation to TLS. And obviously, the tumor microenvironment is never canonical. So, while we can take cues from secondary lymphoid organs or other diseases where these ectopic tissues form, that's not going to be the end-all, be-all. We have to also consider that the tumor is there. We need to think about what might be different about that. Okay. So, some additional information for the field, because I really wanted to make sure this was comprehensive and not just about my data, and we will get to the imaging here in a minute, is that a lot of the data in the field from, from not cancer has shown that that certain cues like lymphotoxin are quite important to make TLS. And so, what I'm trying to get back to now and, and leading into is that it's not just the cellular composition of the TLS, it's also other cues that are important to generate a TLS and maintain a TLS. And I think that's what makes these so complex and difficult to completely understand in patient specimens. And so, while we're doing the best we can here, I think that a lot of these studies in mice that are not in the cancer space have given us a lot of cues of what could be important. So, lymphotoxin is one of them, and it has been for a very long time. But the reality is, is that based off of secondary lymphoid tissue, as well as based off of what we know in other ectopic sites where these structures form, it's not just the cells. As I mentioned, there's a lot of initiating and maintaining factors. So, I've talked about lymphotoxin beta. CXCL13 is right up there on the list. And then, there's a lot of studies showing modulation of IL-21 and how they might regulate these structures. But there's a whole host of these. These are not the only 3. And then antigen is quite important. So I'll show a little bit of data from my lab and others that virus and the microbiota can dictate how these structures form. But in general, any sort of immunogenic cell death or any sort of very stimulating antigen will help to maintain these structures. Not only— it's not just the formation, but keeping them around. Antigen's important, and the stroma is very important. So, we'll talk about this at the very end today, what we're doing in ovarian cancer. But that stroma is quite important, and it's complex, and it can be very heterogeneous from patient to patient. And so, trying to understand this and how it regulates TLS is quite important. And so, when we think about what's different about the TME— oops, let me just go back one click too quickly— is that we start with patient tissue. And I think that a lot of the work has been done in patient tissue, and that's important because that's where we've seen structures the most, and it's been it's taken us time to get the right mouse models around, but I think it's happening. And so, I think that having really great cohorts is important. And so, a lot of what I'll talk about today is more in, in primary setting, and then thinking about some more metastatic sites, but really understanding metastasis is going to be important. I think having the right models, and we won't get to talk about this today, but happy to chat offline as well if we run out of time. But there's mouse models that are being thought about to really study this within, with a more controlled setting. And then I think a lot of the state-of-the-art techniques that have come online have really helped us to understand these structures in a more complex way. So multispectral imaging is part of what we'll talk about today. But also we roll that together with spatial transcriptomics. And I think that's quite important. But I also want to say that we can't forget about the power of techniques such as single-cell and flow cytometry, which allow us to pair some of our spatial techniques with cellular immunology. And I do think that that's also quite important. Important. Okay, so some evidence that the tumor microenvironment can lead to TLS formation and maturation. This is again some imaging with the Vectra platform that shows differences in head and neck cancer. This is a published paper, so I'm not gonna show a lot of data from this, but just to point out that these structures exist in both microenvironments, but are more pronounced in the HPV-positive microenvironment when we match the tissue site. And we also showed, as I mentioned, that this correlates with better survival, both in red and purple for TLS that are mature that come HPV-positive sites. So patients do better in general in HPV-positive, but even better when you bifurcate by more mature structures versus less. But this is also a benefit in HPV-negative tumors. So having these mature structures are helpful from the survival standpoint, but I think having a structure in general is important versus not having one. And then finally, we have used some markers like CXCR5 as a singleplex to really understand, um, if there's certain cell types that are more pronounced, or could be more pronounced within the tumor microenvironment. So CXCR5 are gonna mark things like T follicular helper cells. They're gonna mark germinal center B cells, which memory B cells, which all gravitate to these structures. And so, this HPV-positive cohort, we can see that CXCR5 is much more pronounced. So, that's another good marker to really look at cells that are inside and outside TLS. And then some work that we did collaborating with collaboratively with Tim Hand's lab, and actually the first author of this paper was Abby Overaker-Delgoff, who now has her own research program, actually showed that even one microbiota can influence TLS. And this is in a mouse model of colon cancer, and giving HF, you can see that the TLS go up as a result of having that in this mirroring model. And the other thing that was shown in this paper is that these specific bacteria-specific CD4 T follicular helper cells reside within those structures. And so, I think this really speaks to the importance of having the right cells there, having some sort of immunogenic antigen to have these structures form. And I think having these either tumor-specific or antigen-specific cells are not only important for the response or for the formation, but we also think about what they can do for maintenance and for helping TLS formation. And so Tom Schumacher's group wrote a really great paper showing that these antigen-specific T cells, they have a lot of hallmarks, but one of the hallmarks that really piques our interest is that they produce CXCL13. So, Unlike in normal lymphoid tissues where some of these different factors that help to induce TLS come from, from non-immune cells, quite frankly, that's what's being shown here. In the tumor microenvironment, we are seeing that immune cells such as TFH, even Tregs, or tumor-specific CD8 T cells can produce factors like CXCL13 and help with this formation. And we're looking at this with spatial transcriptomics and trying to understand if we look at a lot of different factors, and this is a curated gene list with spatial transcriptomics. Transcriptomics, and we take regions that are TLS or not TLS and a very simple question. And we say, what is the expression of some of these factors? You can see here, it's difficult with the heat map because there's a lot going on. But this box here, or this like block here that you can see, most of these are TLS or purple regions. And you can see that some of those signals that are highlighted here are some of the signals I've been talking about, in particular CXCL13, lymphotoxin beta, they are present in those TLS regions, but there's a lot of TLS that have incomplete expression of these factors. And in fact, they are not always expressed within those factors. And so that starts to get us thinking that in the tumor microenvironment, there might be a paucity of these factors being expressed, and that might be also dictating the type of structures you get and, and how long they're around. And then finally, I just wanna highlight the complexity of the stroma. I know I've already said it and we'll come back to this, but there's a lot of questions in the tumor microenvironment experiment about how some of these pre— these cells that are really important for TLS, such as FDCs, where do they come from in the tumor microenvironment? And are they properly differentiating so that we can have this stroma maturation and have really nice organization of these structures? Okay. So now into our data, because I wanted to give a very thorough overview. So we have been interested in several tumor types, and I'm gonna talk about ovarian cancer today. And I, I think what really really struck us about ovarian cancer, and, and Ian in particular, is that when we compare the number of structures in high-grade serous ovarian cancer to something like head and neck, which I've already shown you, they have a lot of structures, they have a lot of mature structures. You can see that HPV-positive, HPV-negative TLS are here, or TLS from those patients rather. But in high-grade serous, you can say, you can see that there's actually not as many structures. And that's despite the fact that there are B cells that would come from these structures, particularly germinal center B cells and antibody-secreting cells. They're present within the tumor microenvironment. Now, granted, they're present in a higher amount in the ovarian tumor microenvironment by flow cytometry compared to healthy donor blood. But when we compare this to normal lymphoid tissues like tonsil or lymph node, it's obviously less. So it's important. It's, it's very, very important to think, to say that they're somewhere in between. But those, those cells are there. They're just not forming structures. And further, when Ian went to the TCGA and he looked for B-cell signatures, he saw that patients did better when they had high B-cell infiltration. And then he used the 12CK score, which is a chemokine score that's been pretty well accepted in the field to predict response and also be a good surrogate for TLF. Less. And when you look at both of those patients, ovarian cancer patients do better, but it's really when you combine the 2 of them that they do really, really well. And so we started to think about, okay, it's not just about, again, the cell types, but it's about the other pieces that are important for these structures to form. Okay, so before I get into some of his data and talk about what we did, I want to introduce how we're doing this spatial— these spatial methods at Pitt. And so I have to emphasize that we did not do this alone. We, as I mentioned, we use the University of Colorado Immune Monitoring Core, which, which is led by Kim Jordan. Really, really a great facility that helped us get this going. And now we have our own kind of facility at the University of Pittsburgh and Hillman Cancer Center. And Mary Ann Joy has been instrumental in helping us do this. And so I would really encourage you to pair with the right people, and I can help you put, put you in touch with that. And when we initially started, we started with the older platform, the Vector platform. And that Vector platform is great, but you, as you know, you can do 6 to colors, but you have to unmix the images. And so there's an extra step with scanning, and then you have to pick, you have to pick ROIs, and then you have to rescan and unmix. Now we've moved completely to the Motif platform, which allows us to get an image, and when we get the image, we don't have to unmix. We can scan once, and then we can go straight into data analysis. The other thing that I would mention is that you can do manual staining, or you can do auto staining. The manual staining, if you have like a cohort of of, you know, 20 to 30 patients can take you 3 to 4 weeks to do it by hand if you're doing one at a time. Or if you're doing, um, if you're doing one person staining that cohort, it's probably because you can comfortably do 7 to 8 slides at once without really messing up. Or you can get onto an auto stainer, which is the Leica Bond, and we have one of those in our facility, and you can run 30 slides overnight. And so it really just depends on where you want to put your effort and how many slides you're running. And so this is just one example of a, of a new motif panel that we put together here, which allows us to look at some of the key subsets within the tumor microenvironment. But we also put PNAD into this so we could start to mark TLS. So it's just one example of a panel we have. We have several now that we utilize. So I'm going to go through some of, some of these steps today. And so the spatial biology method, if you, if you don't know it, I think what we really like about the motif and the, than the, or the previous Vector platform is that it amplifies signal very, very nicely. It's probably one of the best that I've seen. You can only do 6 colors, but it really is a workhorse. It's tried and true. It gives you really beautiful images and you can really get a really, really great read on the cell types that are in there and their interactions. And so this is the amplification that occurs with that platform if you're not familiar. And this is one example of our staining template in the lab, how we know what dilutions to use, how to stain day 1, on day 1 versus day 2. And then if we do this on the auto stainer, obviously this is different. So I would say that everyone in my lab learns manual staining so they understand the process and they understand how to troubleshoot. Then once they basically understand that staining process, then they can move to the auto stainer and focus on image analysis. And image analysis starts in PhenoChart. So this is the software that comes with, it's part of the Inform package. You can actually put this without having Inform on your computer so you can see images. So if people send me images, I can open them in PhenoChart. And then, so you can see it, the example staining from this panel here, and you can layer on different filters. You can look at the different channels individually. You can look at the tissue at large. You can zoom in to the tissue. So this is a tonsil, which we always have this positive control when we're staining, particularly for TLS. It tells you if your staining worked, it tells you if it didn't work, it tells you if there's something that went wrong. And then you can toggle on and off, as I mentioned, the different channels. So you can see different cell types. So this is gonna be FOXP3. This is gonna be your CD4. You can see then your PanCK, which is gonna be not tumor in this case because it's, it's a, it's a tonsil, but it allows you to see signal. You can then start to see your PNAD structures, your B cells layered on, et cetera. So it's a really nice platform to be able to see. And then there's kind of two points where you can go. You can do a lot in Inform and, and this This is nitty-gritty, but I won't talk about all of this today, but you can, you have to do tissue, you can do tissue segmentation. I'll show some of that today. You can do cell phenotyping, but then, and then that pairs with this Phenopter reports that you can get graphs and analysis and even do nearest neighbor. But we've also started to use QPath for some of our analysis because you can do whole slide analysis. And we've just found that for some analyses, it's just easier to use the QPath function. So that's also a free software and, and you can download it. So, to show you some of the data and coming back to the TLS part, this was done in QPath. And so, so really what we started to think about is a lot of what has been done in ovarian cancer and the data that I showed you previously to going through some of the technical pieces, given the nature of this webinar, I, you know, we started to think about, is it just the ovary? Right? And a lot of what we do is in just the ovary site, essentially. So we actually got sections from fallopian tube as well as omental tumors. So we looked in all 3 sites. And so what Ian did is he created an algorithm where he can train the program to identify lymphoid aggregates within the tissue using this, using the QPath software. And so you can see here, everything that's green is a structure. So we vet this with the pathologist, right? To say, okay, are we capturing the right things? But now we can do automated counting with this platform for TLA. Less, which is quite powerful for us. So we don't have to sit there and count everything. And so what you can see just by eye, and I'll show you the numbers here in a minute, is that the fallopian tube tumors have a lot more structures. You can see those structures are more sophisticated. They're larger. In the omental tissue, you have probably the middle point of it, and you can see the structure composition here on the inset. But in ovary, you can see that some struc— some patients just have one structure, one green dot. And so we have a lot of cases in the ovary. Tissue compartment where you have patients that have no structures and then patients have one or more with there not being that many within this site. And so we really got thinking about the tissue, right? And what's happening within these tissues that might be different. And so it's not just, um, oh, sorry. I wanted to also show this. Excuse me. So this is the imaging of the different structures. And then this is the quantification. So here we can see fallopian tube and omental tumors. You have a lot of structures compared to ovary, but we also We looked in benign fallopian tube, we looked in benign ovary, we looked in STIC samples, which are a precursor to fallopian tube disease. It's gonna be serous tubal in situ carcinoma. So basically this is gonna be the precursor and you can see some, but not as many as when it's a full-blown tumor. And then I already mentioned the size, but this is the quantification of size that he did with QPath. So he is able to measure the average size of these structures. And you can see the fallopian tube structures are the, are the largest compared to the omental tissue. And then the ovary, And actually, size has been indicated as a pretty good surrogate for TLS sophistication. And we're starting to see in this dataset that that's true because we paired a lot of this with spatial transcriptomics, which I won't get into today. But we, we have been able to see that. And so, what he was really interested in was not just the numbers, right? So, he did this beautiful algorithm to look at that. But what about the composition, the cellular composition within those structures? Are they the same or are they different? And so he used Inform to really do this unmixing segmentation and then phenotyping to understand this question. And so— May I have your attention? May I have your attention? So, so— This is the hotel fire safety director. Oh, sorry. We are conducting a test of our fire safety system. Please disregard all sirens and alarms. You will be notified if this was a real emergency. Thank you for your cooperation. Sorry, we had a fire drill in the building. I just wanted to make sure I didn't have to leave. Um, so this is the composite image here that you can see, and then what he's able to do is he's able to train tissue segmentation. So what's nice about this is he can train stroma, he can train tumor, anything that's just glass, and then he can, he can train the TLS. And so what now happens is you get this trainable phenotyping where you can look at the cell outputs based off of these different masks, which is quite quite, quite great. And so this is showing you some of this. So basically what we found, the bottom line is, is that the TLS composition is quite varied as well, depending on the site of ovarian cancer. So here we're comparing fallopian tube and ovary. We're still putting the omental tissue, um, analysis together. But basically what you see is that if you compare to STIC, which is gonna be that precursor to fallopian tube versus fallopian tube and ovary, which we already saw overt differences with number of TLS, you can see here that the cell types are different as well. And in fact, um, ovary have more Treg cells, they have more CD8 cells, they have less B cells. And so this started to get us thinking, well, why is it that you're seeing these differences here? And could it be related to high endothelial venules as well? And so, um, that's something that Ian did. He used this panel to look at high endothelial venule quantification, and he showed that when a high endothelial venule is present with nearest neighbors, it likes to interact with B cells the most. That's the other than itself. It likes to be interacting with those B cells. And then when we actually quantify those high endothelial venules in these different sites, you can see that the ovary has, um, fewer high endothelial venules compared to the other two sites. So it's possible that the B cells aren't getting in because those high endothelial venules are, are less frequent. We see more CD8 T cells. So we're actually currently testing the hypothesis that maybe those CD8 T cells are expressing CXCL13 and trying to signal for the B cells to come in, and it's just not working. There's a lot of possibilities here, but all of this spatial work has allowed us to uncover some of these differences. And so, I just want to come back to this kind of, kind of solidifies what Ian saw in the TCGA, this slide that I showed you earlier, which is that it's not just about having the right cues for TLS to form. And in this case, it's the 12 chemokine signature. We've tried other signatures. The same thing actually comes through for TLS, but the 12 chemokine is not enough. You have to have B cells present. Present too. And so that's quite important. Okay, so it's not just about TLS frequency and composition, it's also about the TLS maturation. So I spent a lot of time talking about TLS maturation early on and what that means. And I told you that we now bucket the cells into 3 kind of groups in the lab. We do lymphoid aggregates, we do TLS, which will be TLS without a germinal center. So those would be more immature, that's tier 2. And then we did TLS with germinal center, so that's tier 3. And so another feature of this maturity panel that I showed you earlier on at the very beginning allows us to train for a germinal center as well. So this has also been really great for the lab is that we not only can train for TLS, but we can train for GC, and we use CD21 to do this. And so you can see here very clearly those 3 tiers, and then this is the criteria that we use for this at present. It might change as we learn more. And when we quantify, we can see that the ovary has the most mature— has the excuse me, mature TLS, they actually have a lot more lymphoid aggregates than mature, and only about 10% of the structures are mature, whereas in the fallopian tube and the lumen, there is 40 to 50% that are, are quite mature. So the site matters, and it's not just composition, but it's maturation. And as I mentioned to you earlier, we can look at how Ki-67 and AID follow suit when there is a germinal center present. And you can see here that Ki-67 and AID go up when there's a germinal center compared to not. However, there is a spectrum and we're trying to understand the spectrum because it's not always like they're, they're gonna have a great activity with the germinal center being there. And this is in relation to the B cells. Okay. So that's how we've used a lot of multispectral imaging and different platforms and different, I guess not platforms, but different panels to really look at this and understand it. And, and I think that it's really powerful to understand interaction, cellular neighborhoods, et cetera. But we're now pairing that with other spatial platforms so that we can understand a little bit more about the TLS. And that's gonna be the spatial transcriptomics. And so what we like about Vectra or Motif now and DSP is that you can basically mimic your image and then you can pick these regions of interest, which are gonna be your TLS, and then you can even segment for those different regions and look at different cell types via different outputs. And so for NanoString, they have multiple outputs here. They have protein code sets, they have attenuated gene sets, but they also have expanded sequencing platforms, which is really what we're using. And I'm gonna show you 18,000 gene readout today. And so basically what's happening here in this image is that we are doing a morphological staining for TLL. TLS, and we can use 3 markers, and then you can segment out for different cell types. And so what happens here is that you can collect CD45 cells that are, that are not B cells into one well, B cells into another, tumor cells into another, and then by negative selection, you can collect stroma. And I mentioned to you that we were gonna come back to the stroma, and you'll see why in a minute. And so we were able to, for each region of interest that was a TLS, we're able to collect those 4, um, regions and look at the sequencing. And then we can also collect regions that are next to a TLS to really understand what is happening adjacent to a TLS. And so the initial comparison of just us looking at TLS versus no TLS actually allowed us to generate a prognostic signature that came directly from the TLS that's predictive of survival in these patients. And I'm not showing it 'cause we're still running some further tests, but really it shows better survival in these high-grade serous patients, and they're under 55 actually, which is also quite interesting. And it out— and it outperforms that 12 chemokine signature when we use it in the TCGA. And then some other nice things about looking at the CD20 segment from this, from this platform is that you can look at the top genes that are different in different types of TLS. So if you use your staining that I just showed you to tier your structures based off of maturation, you can say, okay, B cells that are in more mature TLS, what do they look like compared to those that are or not. And you can see here several genes that pop up that are probably not surprising, BCL-6 in particular, CD79B signaling, et cetera, that would be present when the structure is more sophisticated. So this is really a nice thing to pair with what we're doing with imaging because we can take our imaging and then we can understand genes that are governing these different states. And then what other— the other thing that we did and what's been really exciting is we were able to create a differentiation trajectory for that immature to mature state. And actually what popped out of that is this intermediate state here in green that you can grab these more immature structures that go to mature structures, but then there's this intermediate state that's popped out from the spatial transcriptomics that we probably wouldn't have captured with the imaging. And so now we can utilize these gene sets and then put it back onto the imaging platform to look at it and at a protein level. And what was really interesting here from a biological perspective is that those that were more immature or intermediate were from the ovarian tissues, which those that were more mature came from the omental. So it kind validated what we saw with our imaging is that ovary has the least number of mature TLS. And so we started to think about the different sites and, and what that means for the spatial transcriptomics. And we started to compare to lymph nodes, normal lymph nodes, because that's where you have a lot of secondary lymphoid organs. And so we started to look at gene sets that were different in the omental and in the ovary compared to the lymph node. It's not probably surprising that there's a lot more that are different in the ovary than in the omental. And that's because these are gonna be less mature than these. So that was one thing that we saw with the spatial work. And what was interesting though, is that the top genes that are here, that are present, those genes are going to be the same. So it's as if when they get to that more mature state, essentially, the genes are similar. So those are gonna be genes associated with metabolism and actin mobilization. So the number of structures aren't the same, but the genes seem to be similar. And overlapping. And so we really wanted then to look at these signatures of immature versus mature. And so you can see here that when we look at our vector imaging and we grade these accordingly, we can then take the mature signature and it pops out pretty nicely. And so all these genes in red really govern that mature signature. So we're using our imaging to then look at this in a spatial transcriptomic way and we get genes that are correlated. Here. And the genes that pop out, there's a lot of them, but some of them that are quite impressive are those associated with focal adhesion and motility, actin cytoskeletal interaction, cell matrix adhesion, all things that we would think about in the stroma. And so, this takes us back to the stromal compartment. And it's probably not the only thing here that's dictating what's going on. I mentioned some other things that could be helping or hindering here. But I think that we come back to this idea of what are the stromal precursors in the tumor microenvironment, and can they dictate what's happening with these support cells that are necessary for TLS formation, but really for that maturation, essentially. And so we paired with a really talented physician scientist. This has been one of my favorite collaborations at Pitt. Um, Dr. Long Kaufman, she's fantastic. She focuses on the stromal compartment of ovarian cancer. And so we started to look at the different types of these mesenchymal stem cells, which are precursors to lymphoid tissues like follicular dendritic cells, follicular reticular cells. And so she's really done a lot of work to define cancer-educated mesenchymal stem cells, which really are defined by WT1 compared to the other markers that are utilized to mark these mesenchymal stem cells in general. And these more suppressive cells, they've actually promote tumor initiation. They make a more immunosuppressive microenvironment. Patients do worse as a result of having them. And we can see that these are not good to have around. Whereas if you have more normal MSCs, or even bone marrow-derived MSCs, these are going to be that soil that lays the foundation for the, the stromal compartments that are gonna be important for TLS formation. And so, we went back to our spatial transcriptomics, and we started to look at, unbiased way, at TLS that had stroma proximal or, or TLS distal stroma. And we compared these two regions with that spatial transcriptomics. And so, what we saw was that in ovaries, the top stromal signature is not a mesenchymal stem cell. Actually, it's in the omental tissue that it is. So this says that MSCs in general are more abundant in omental tissue. We don't know what type of MSC here. And what we started to do then was we use spatial imaging. So pairing the Vectra platform with immunofluorescence, and now we're rolling these together in one motif panel, is we started to look at where these MSCs are. And so these cancer-educated MSCs are quite absent when there's a more mature TLS, but the bone marrow-derived ones are present, which you can see here adjacent. And what we did with the spatial transcriptomics, that is, we said we looked at that hallmark WT1 for these cancer-associated MSCs, and we saw that they were distally more concentrated than adjacent to the TLS. And even gene sets associated with WT1 were more distally associated. And so, we can start to look at those differences at a gene level.. And we used Lon's signature for CAMSCs, and we showed that as you move away from the TLS, you have less— you have, excuse me, as you move away from the TLS, you have more of these cancer-educated mesenchymal stem cells. So more of these suppressor ones, whereas if you're closer, you have less of them. However, there is a spectrum. And so we've started to pair this with cellular immunology. So this is my plug for doing a lot of spatial, but then going back in the people and then also looking at what's happening functionally. And so Lon's able to isolate these, um, cells from the patients and we can look at cancer-educated mesenchymal stem cells versus normal mesenchymal stem cells. And what we can see here is that the cancer-educated mesenchymal stem cells don't, um, allow for B-cell adhesion as nicely as the normal ones from patients. Migration tends to be somewhat similar, but you can see here it's slightly increased for the, um, normal MSCs. And now we've created an FDC differentiation assay in the lab where we can look at if if these, um, different mesenchymal stem cells differentiate to follicular dendritic cells in particular by using CD21 and CD23. And then this is just one graph from that output of that in vitro assay showing that at day 6, basically these normal mesenchymal stem cells, they have more CD21 expression compared to the cancer-educated ones. However, they seem to normalize and level out. So we're trying to now understand these kinetics both in vitro as well as using, as I mentioned, a multispectral imaging platform that rolls a lot of these markers together from our spatial transcriptomics. But also we have a mouse model that we're testing this in as well that I don't have time to talk about today. So with that, I'm going to wrap up. I'm going to put some take-home messages here. I talked about a lot of this along the way. But really, I think that what we, we know is that the multispectral imaging is quite key, and there's a lot of ways you can use the imaging to really look at these structural in a sophisticated way. I think the spatial transcriptomics is necessary to pair with this because there's a lot we don't know. And I think that's what's being revealed here. And then we can go back and make protein panels to look at this in a more granular way. And, and really try to understand what's going on in patients while we further develop mouse models. And so what I would leave you with is, is that if these more sophisticated structures are our endpoint, let's just say that that's what we're saying at this present time. There's a lot of things to consider and, and you could get more CD8 T cells and then have better response response to PD-1 blockade. And this would be particularly helpful for ovarian cancer patients because they've historically not responded to PD-1. And they would also benefit from the feedback loop of having the CD8 T cells there because they make a lot of CXCL13, as I showed you, and you could kind of continue to have these structures present with the right CD8 T cells that could then be used in a combinatorial therapy. So, our lab, as I mentioned, we really think about the complete microenvironment and how they govern these structures, the tumor, the the immune cells and the microbiota. I talked about some of this today, but there's a lot of projects underway in multiple solid tumor types in the lab, as well as metastases that we're really trying to understand this question. And I think in general, despite all the data that I've shown you today from others and myself, there's still a lot of unanswered questions in the field. And these are some of them, some of what I hit on today, but some that I think are in progress. So, understanding antibodies, understanding how, um, immunotherapies modulate, how other therapies modulate. And this is really the focus of our Mark Foundation grant that we have, which is a team grant. And it just shows that you can't do this on your own. So, we do a lot of collaborative work with a lot of individuals to try to understand this in humans and mice. And this is us at our first retreat. We just kicked off the grant a year ago. So, it's been really fun trying to dig into this in more detail. So, with that, I'll leave you with this image. This is an image from the same book chapter that Ayana wrote. It really laid out nicely different ways in which might be able to think about targeting TLS. And I think this is now probably incomplete, given all the literature that's come out in the last couple of years and a lot of the work that people are doing. But it just kind of gets you thinking about these structures, thinking about these cellular neighborhoods and what might be important and necessary. With that, I'll thank everyone again. In particular, I want to thank the patients and their families. All of this work has been done from patient material, and we can't do it without them. So, a big thanks to them. In addition to all the others I thanked earlier. And then finally, if you like what you heard today and you're interested in this topic, there's 2 ways that you can join us. We have an online consortium that meets monthly to talk about B cells and TLS in cancer, but also in other tissue sites and normal tissues, other indices. But we also have a Keystone meeting that's going to be happening in October. It's October 1st through the 4th. So if you're interested, you should go submit an abstract and get signed up because, um, the abstract just opened and registration just opened. With that, I will stop and I will start to take questions.

Thank you. Thank you, Dr. Bruno, for that excellent presentation. We will now start our live Q&A portion of the webinar. And to our audience, if you have any questions you'd like to ask, please do so now. Just click on that Ask a Question box located on the far left of your screen, and we will answer as many as many of your questions as we have time for. We do have an abundance of questions already, so I'd like to jump in and get started. So Dr. Bruno, let's start with this one. Is there one marker to utilize for TLS identification and quantification?

Yeah, I think this is a great question. So I think that I mentioned this in the beginning of the talk, is that you can, you know, start out with the single plex images like CD20 paired with H&E. I think that's the most accepted at current, at present. And you could certainly singleplex for any one of those markers that I mentioned to you to get some sort of a quantification. I think you have to start somewhere, but I would argue that use that as like a triage or as a broad stroking scan of your tissues, right? And then go back and do the multispectral imaging because as I showed you, the multispectral imaging is really going to give you the, granularity that you need for the different types of structures. And then you can fine-tune that if you're doing other studies to really understand how you can curate your TLS definition better.

Thank you, Dr. Bruno. Our next question, and this is two parts: What is the definition of peritumor TLS? Sorry, peritumoral TLS. And if there is a normal tissue between the tumor border and the peritumoral lymphoid aggregates. Should we consider these aggregates as TLS?

Ah, this is a great question. So I think it depends a little bit on your tissue. So I showed the image from head and neck cancer at the very beginning because that's where we see this a lot, where you see clear tumor, then you have a peritumoral border, and we can talk about that in a minute, and then you have the tissue outside. So any of the tissue that's not annotated histology— histologically by a pathologist as tumor, I don't think you can call call that a tumor-associated TLS. You might call it a TLS or a TLO. Pick your, your, the way you want to say it, because it might have all the other hallmarks that I mentioned. So, but it's not going to be one that's associated with the tumor. And I think that's something that we're starting to try to understand with both multispectral imaging, as well as with the spatial transcriptomics, is what's the difference between those two. And so, so that normal space is going to not be associated with the tumor. And then the peritumoral space, I think, is set differently by a lot of different, um, people, but pathologically, usually your pathologist will mark the number of microns away from the tumor. But what we look for is that they're really sitting up against the tumor, essentially, and that's the peritumoral space. And a lot of people still call those tumor TLS, essentially. So I would say that you can work with a pathologist to really understand what they would define as peritumoral. That's what we did. And then we basically, um, looked at the counts both intratumoral and peritumoral. The reality is, is that the intratumoral TLS are, are not that common. They usually, if there's a solid chunk of tumor, they're gonna sit on the peritumoral border. However, you could have some tumors that don't have solid islands of tumor. They're more like dispersed. Then you would see TLS like within those tumor cells and not associated with the stroma. So those are gonna be different. And we just actually talked about this yesterday in the lab. I think that it's a moving target and we're We're going to have to continue to refine it over the next year.

Dr. Bruno, going back to your cellular neighborhood slide, what software did you use to determine the cellular composition?

So, we use a mix. So, I kind of buzzed through some of that work because we had a lot to cover, but I would say that we use both. So, you can do the first part, which is, you know, PhenoChart and tissue segmentation and everything like that. You do all of that in Inform first. You have to do that part. Then once you have that, you can do the cellular output based off of Inform. They'll give you those cellular outputs. But we've also found that QPath helps us because you can do the whole slide and you can scan the whole slide a little bit easier. So you can use a combination of the two. But as far as if you want to look at— that just tells you cellular outputs. But if you want to really start looking at cellular neighborhoods, and I showed a little bit of this with the PNAD, you actually have to write scripts in R to do that. And that's something I didn't mention. So I would advise, once you get a good sense of your cellular outputs, your general cellular neighborhoods, just by looking at your staining, I would recommend meeting with a bioinformatician or having someone within your group that has the capability to do this that can then write scripts to look at those cellular interactions. Interactions, because that's the other power of having all of the parameters on one slide, is that you can look at those cellular interactions, which you can't do with singleplex.

Thank you, Dr. Bruno. So the next one is a few parts. How stable are TLS structures? Once formed, do they later resolve, and how quickly? And in terms of inducing new TLS for clinical benefit for oncology, Would there be a risk of inducing a chronic autoimmune disease?

Yeah, so great questions. So I think the first question is really difficult to answer because when in humans, we just get a snapshot, right? So we get a snapshot of TLS and, and it's a piece of tissue and we're not getting it over time. So now we can start to do this in mouse modeling. And so we have a mouse model that is carcinogen-driven that we can look over time at TLS formation. And we kind of stumbled upon this model because it's been been around for a while, but these, these mice, they get it. It's just a very long model. And so that model, actually, what we're seeing, and I didn't show this today because we didn't have time, but basically what we see is that over time, you see TLS formation, and then you do start to see them start to contract a bit. So we probably have to look a little bit longer to understand this a little bit. So I would hypothesize that, yes, they would, because it wouldn't— to your second point, it wouldn't be advantageous to just have a structure around that's active all the time because you don't want to swing in the other direction of autoimmunity. It's the same thing we think about with any immunotherapy. And so that's why I always come back to this question of, do you have to have always have really, really sophisticated mature TLS? And would you always want them all the time? Or would you want to make sure that they were going to contract and go back down to an immature TLS? Or even more so, do these different types of structures have different functional outputs? Which is something that I think the field is thinking about a lot. And so, um, I think that there's always, you always have to consider that. And so, So that's something that we think about a lot by thinking in the mouse models. And now in the human cohorts, I know a lot of the work that I'm doing collaboratively, we're looking at adverse events and we're trying to understand those people that got adverse events, particularly based off of immunotherapy. Do we think some of this might be related to the TLS being around for too long? And so I think it's still an open question, but theoretically, the answer is that you wouldn't want them to necessarily always be around in a really, really sophisticated form because you don't want to swing in the other direction.

Thank you, Dr. Bruno. And I'm going to try to squeeze in 2 more questions, but I do want to let our audience know that those questions we are unable to answer today due to time. They will be answered via email at the email address you used at your time of registration. Okay, let's squeeze in a few more. Question: wondering what's your consideration of tumor heterogeneity when you work with FFPE slides in certain tumor areas?

Yeah. So, this is, this is a great question. So, everything that we did is an FFPE. So, I am sorry if I kind of glazed over that. I think it's something we do all the time, but you use— we use FFPE. And so, the tumor heterogeneity is something that is going to be quite important. And so, we've been trying to understand this more from a pathological standpoint by doing better annotation of the tissue and then seeing where the TLS lie for certain tumor types that we're working on. So, I think the jury's still out. Out on this. I think actually in general, and it's something that we didn't have time to talk about today, we haven't been paying as much attention probably to the tumor as we should with how they dictate what's happening with these structures. And so that question kind of gets at that, like, what are they doing? What's that tumor doing? And when they're different and different heterogeneities, how is it going to affect the formation, the maturation? And I think this is something we can get at in a more controlled way also in mouse modeling.

Thank you, Dr. Bruno. And we'll wrap with this question. Hi, really nice talk. I have a question. The presence of Treg or expression of high IL-10 could affect the maturation of TLS, and if this could be related in cold tumor with low T cells tumor infiltration.

Sorry, Christy, you cut out at the beginning. Can you just repeat it one more time?

I sure can. The presence of Tregs or expression of high IL-10, could it affect the maturation of TLS? And if this could be related in cold tumor with low T cells tumor infiltration?

Yeah, so this is a great question. So the thing about it, the Tregs, and I think we're trying to get a better handle on this in the lab too, is that in general, we see Tregs can be around TLS, but they're not the most abundant cell type. Cell type. But again, in humans, they're obviously— they're not always the most abundant CD4 either. That's very different from mouse modeling. A lot of sub-q models, like 40% of the CD4s are Tregs, and that's not the case in a lot of humans. So I think that, um, they can play a role, but I think what we need to tease out further is there's a subset of T follicular helper cells that are also regulatory, and the sole purpose of those in normal lymphoid organs is to contract and control TLS formation. And so what we're doing now in the lab is we're not just looking at Tregs in general, but we're looking at Tregs that could also be of these helper follicular— helper phenotype that might be helping to control the TLS but not suppressing it. And so, I do think that very much so, the IL-10 and everything could be suppressing for sure in some ways. But I think we just don't quite understand where they play a role quite yet. And I think we've got the phenotyping, but we need to kind of dig into that functional aspect.

Thank you, Dr. Bruno, for your time today and your important research. I would also like to thank LabRoots and our sponsor, Akoya Biosciences, for underwriting today's educational webcast. Dr. Bruno, do you have any final comments for our audience?

No, I just want to say thank you so much for listening, and I'm, I'm looking at all the questions coming in. I'm sorry we didn't get to everything. I should have left maybe another 5 minutes so that we could do a few more, but I'm happy to answer any of them via email, or you can contact me directly at tbruno@pitt.edu, and we can talk further.

Thanks so much. Thank you again, Dr. Bruno. And we did have so many questions that came in and that are continuing to come in. Those that we didn't answer today, and additionally those that do come in during the on-demand period, they will be addressed by Dr. Bruno via the, via the contact information you provided at your time of registration. Now, today's webcast can be viewed on demand. LabRoots will alert you via email when it's available for replay. Now, we encourage you to share that email with your colleagues who may have missed today's live event. Until next time, thank you again for joining us and have a great day.