Hi, I'm Brian McKelligon, CEO of Akoya Biosciences, and on behalf of all of my colleagues and fellow Akoyans, welcome to our second annual Spatial Day. You know, before I get started, some necessary disclaimers and cautionary notes. I'd invite you to go to our website to read this in more detail. Maybe just quickly getting to the agenda. We've got a full agenda today, basically in four distinct parts. I'm gonna give a little bit of an introduction on the company, on the market, on our products, and some of our priorities going forward. Then my colleague in the next section, Dr. Niro Ramachandran, our Chief Business Officer, is gonna dig in a little bit more and talk about where we're taking our platforms. Then our colleague, Dr. Oliver Braubach is gonna dig in and give you some examples of how the technology is being used. In each section we're gonna have some guest speakers, and you can see as listed, Dr. Elizabeth Neut man and Dr. Arutha Kulasinghe from UC Davis and Queensland respectively, are gonna give some examples on how they're using our technology. We're gonna pivot into a separate section, really focusing on the larger and longer term opportunity in the clinical market. That session is gonna be driven by our colleague, Dr. Gavin Gordon, where he's gonna spend some time talking about our clinical vision. We're really honored to have three really powerful guests with us today and users of our technology, Dr. Laura Esserman from UCSF, Dr. Scott Rodig from Dana-Farber/Brigham and Women's and Harvard, and then also Manuel Salto-Tellez from Queen's University Belfast. Like the prior section, we're gonna have a round table at the conclusion of that. Then we'll follow everything with a longer-term Q&A. That's the agenda today. Let me just kick off with kind of a high-level summary of spatial, of Akoya. First, spatial biology, as exemplified by the image on the left, really is transforming not just discovery research, but downstream translational and clinical research. Akoya's approach here is really to build best-in-class platforms that meet the, not just the requirements, but the distinct requirements of these markets. Distinct requirements in the discovery market versus those distinct requirements in the translation and clinical market. We have unique solutions and platforms to serve those market segments. In each case, they're complete end-to-end solutions with instruments, software, and reagents. Probably the best metric or metrics for us as a measurement of our success is the fact that we have nearly 900 instruments at the end of Q3 installed worldwide. As a by-product of that or as a driver of that, we have almost 700 publications as well. Those are some of the key metrics that we follow. Let me spend a minute just to kinda back up a little bit and pull out the aperture, pun intended, to talk a little bit more about what spatial biology is. The concept is fundamentally simple. The goal of spatial biology is really to drive a deeper understanding of biology. Doing so by advancing and delivering next generation tissue analysis. You know, today we can sequence the genome, we can sequence the transcriptome faster, cheaper, with higher volume, really following Moore's law. There's current and next generation that proteomics technologies that are giving us ability to understand the proteome. What matters most, what matters equally is not just understanding what's present, but where it's present. The goal of spatial biology really is to simply map this tissue architecture to give you a snapshot of the biology, again, to understand not just what is present and how much, but where. It's understanding this tissue architecture. It's understanding this disease biology, taking that snapshot that helps us understand how these cells and proteins are interacting, how they're driving response to therapy, how they're driving disease progression. Really, that's the goal of spatial biology, understanding this in an unbiased whole tissue manner. To look at it through another lens, again, the goal of Spatial is to put these biomarkers in their spatial context. If you look at some of the profound technologies that have impacted life sciences, that have impacted patient care, they include proteomics and next-gen sequencing, single cell analysis, flow cytometry, even single marker classic immunohistochemistry have really transformed our understanding of biology. What Spatial is doing is giving us the ability to keep that tissue on the slide, not peel it off and disassociate it so you can run these methodologies, but keep it on that slide to do this high plex biology. That's sort of the how and why, these are some of the customers that are migrating their tissue analysis methods from some of these very classic powerful methods to using spatial phenotyping on platforms like Akoya. That's really what Spatial Biology is. To drill down another click, effectively what we're doing is we're mapping the whole tissue cell by cell to really unlock this understanding of disease. What we do is we stain the tissue tens, hundreds of markers or more, analyze that tissue at single and subcellular resolution. Not just creating an image, but creating a computable map. We can informatically understand causation and correlation between these spatial measurements. Again, what's driving the disease and what's driving response to therapy. Akoya as a company, we are focused on balancing the needs of delivering really high plex and really high throughput. I'll get into this in more detail, but as that trade-off, that's the difference between what you need for a discovery market and what you need for a downstream translation and clinical market. To date, our focus has been on protein. We will continue to perfect, improve, and advance our solutions to do protein spatial phenotyping, while at the same time investing in and giving that large install base the ability to do more, to look at RNA, and eventually, to look at RNA and protein simultaneously and not make arbitrary decisions between protein or RNA, use the right analyte to ask the right question. That's our vision kinda overall of where we're gonna take our platform. In the next section, Niro's gonna talk a little bit more about that. Let me pause and just give you kind of the landscape for a refresher on kinda what we have today in terms of our solutions that enable the discovery of these spatial biomarkers, the validation of them, and then the hopeful clinical use. At the far left is the PhenoCycler platform, now paired with our newest instrument launched this year with the Fusion. That is really the platform that's designed for discovery research, and it's built to maximize plex with the necessary throughput for a discovery project, 10, 20, 30, 40 samples, to get through those samples in a week or two. That's the goal of the PhenoCycler-Fusion, to really serve that discovery market. What's unique about that design of that platform is we intentionally designed it modular, so it can really be a two-for-one. You do your discovery in the PhenoCycler with the Fusion, and because that Fusion has the underlying powerful optics that exist in the HT system, you can then use that Fusion as a standalone to focus your question and do a much higher volume, higher throughput validation study. You can see the metrics at the bottom of the slide. As you advance that discovery farther and farther downstream to translate those discoveries to the clinic, and that's what translation is. It's translating to the clinic to impact patient care. That's where the HT system is so powerful, built under design control, deployed in a CLIA setting, foundational to our current companion diagnostic partnership with Acrivon. That HT system has the throughput and validation necessary to serve that clinical market. That's our goal. We sit here today with a suite of solutions that's solid, that's fixed, and our goal now moving forward is to continue to build on that foundation to make them more powerful, to make them faster, to have more applications, to expand from protein to RNA. There's a lot we're gonna do to expand on this instrument and to expand on this install base. A little bit more details about the install base. You can see that based on the numbers, we continue to accelerate going from 550 to about 700 over the last year. This year, at the end of the 3rd quarter, we had about 860, 870 or so products on the market installed. The breakdown there is on the right. We've got 229 of the PhenoCyclers. Again, that Fusion was launched this year. Many of those are paired with the PhenoCycler, as I noted in the prior slide. We've got about 550 of our PhenoImagers on the market. About 2/3 of those are that newer HT system. There's some historical instruments that are still out there being used in the Mantra and Vectra. That's kind of the landscape of the products and how many we have across that continuum from discovery to translational to clinical. Our publications follow a similar trajectory. In fact, those publications are distributed across that whole install base, across all of the market segments. We're hitting nearly 700 publications year to date, up from about 490 last year. Again, continuing to accelerate. Those publications span areas like oncology, immuno-oncology, transplant medicine, neurobiology, infectious disease, but also they span across multiple market segments. There's publications that talk about novel ways to identify spatial relationships using this concept of cellular neighborhoods. There's publications that talk about biomarkers that are predicting response to therapy in the translational setting. There's publications that support our clinical initiatives and our clinical aims, where they're validating multi-site use of our HT system. They're validating that system against current best practices in standard immunohistochemistry, showing that correlation. Publications mirroring that install base and also mirroring the TAM. As you look at this meaningful TAM, you can see about a $14 billion total, and it descends as you go from discovery to translational to clinical. Part of the reason why, as you go from discovery to translational and clinical, the number of samples certainly grows, but equally important, the value of those samples is certainly true. A clinical sample impacting patient care, impacting treatment is incredibly valuable. There's another thing that we've learned across that install base as we serve these various market segments. There's a series of sort of compounding requirements that are increasingly embodied in our products. Meaning that in the discovery market, what's really necessary is an unbiased question, is an open-ended question, is exploring that tissue to the maximal extent possible across the entire slide. As you move to the translational market, what becomes really important is to do that same thing with speed. Again, those are related. The faster you can do that in the discovery market, the higher probability you have something that can be translational. As you move those translational discoveries to the clinic, it needs to be on a platform that's got the clinical robustness and the viability to be a clinical instrument. What we aim to have is, again, as I noted, a suite of platforms that has a continuity of methodologies, a continuity of reagents, so you can own that biomarker journey from discovery to translational and clinical. That, that's kind of the makeup of the TAM and how we see our portfolio fitting into that. Just this last month, actually last few weeks, a really interesting paper by the team out of Deciphex where they looked at how much of this TAM is actually gonna be penetrated and what are some of the main drivers over the next five years. A couple of take-homes, and I invite you the link's there on the bottom, and I invite you to take a look at that and download it. They estimate about 10% of that TAM is gonna be realized by 2020. If you dig in a little bit deeper, 30% CAGR year-over-year. If you look at the numbers, if you look at the dark blue, and if you look at the light blue on the top, the translational clinical markets ultimately are gonna make up the largest segment of the spatial market. If you look at what's growing the fastest, as projected, it's the clinical, then the translational, and then discovery. That sort of makes common sense. As you understand some of the biomarkers that are relevant in discovery, it's gonna accelerate the value and participation in the downstream translational and clinical markets. Another interesting takeaway is that the drivers of the faster-growing clinical and translational TAMs is multiplex immunofluorescence, meaning protein. RNA is absolutely valuable as a discovery engine, but in terms of an analyte that's gonna translate to a clinical use, protein seems to be the takeaway from this, from this, report. In summary, Akoya's objective over the coming year plus and our strategic priorities is to continue to invest in our workflow. Invest in the workflow so it drives applications in the discovery market, invest in the workflow so it drives utilization in the translational market, and ultimately proves successful in the clinical market. Double-clicking on that, the workflow improvements are gonna happen across the entire portfolio, improving the speed and simplicity of that workflow, expanding the menu of applications in terms of protein content and ready-to-make panels in RNA, and driving improvements in the throughput and the speed across the entire suite of products, but also continuing to streamline the real-time data analysis and downstream tertiary analysis to accelerate the time to answer. Byproduct of that, particularly for the HT system, it's gonna support our desire to accelerate our clinical journey. A key part of that is our CLIA lab, where we're partnering with our key biopharma partners on clinical trials. That really is sort of the apex of our clinical strategy with the HT system. In parallel, achieving our goal clinically begins now, and that begins with delivering to our partners at Acrivon on their companion diagnostic. That is an important milestone for the company. Success breeds success. Success with our team at Acrivon, our partners at Acrivon, is gonna enable us to deliver and announce more high-value partnerships in the clinical setting. That's a summary of where we are at Akoya. Next, I'm gonna hand it to my colleague, Niro, to dig in a little bit more about where we're taking the platforms. Thank you. Thank you, Brian. My name is Niro Ramachandran. I'm the Chief Business Officer at Akoya, today it's my pleasure to share our product roadmap for next year. Before I do that, the first thing I wanna do is share with you where we really come from so you can understand where we're going next. We became the spatial biology company by launching two seminal products. The first, the PhenoImager HT, the fastest imaging technology in the market today, was launched in 2018. In 2019, we launched the first cycling platform, the PhenoCycler system. Combined, what these systems are designed to do is address the number 1 application of spatial biology, which is spatial phenotyping. Requires that you look at every single cell across the entire tissue section, so you can understand the location and what that location means. In order to do that, what we realized was using conventional imaging technologies would take way too long to image millions and millions of cells. Last year at this time, we announced the launch of the PhenoCycler-Fusion. The Fusion was a standalone microscope that has all the capabilities of the HT on a smaller footprint, as Brian McKelligon said. It also connects with the PhenoCycler, making it the fastest spatial biology platform in the market. The ability to image about 1 million cells in just about 10 minutes. The question is, where are we going next? Next, we're gonna focus on building the simplest spatial biology workflow. We wanna make sure any customer, any researcher in any lab can easily perform a spatial biology experiment and with speed and at scale. Let me take a moment to tell you how we plan to do that. There are three things we're gonna think about as we think about the next generation of products we build and launch. First, we're gonna focus on panels and content. We're gonna build an expansive menu of content that supports a wide array of applications. Not just that, they're gonna be optimized content, so they work right out of the box. Second, we're gonna streamline the workflow, make it faster and faster so that it matches the scale of experiments that you may wanna do. Lastly, we wanna think about data analysis. It's a major challenge in spatial biology, and we wanna meet where our customers are. We wanna be able to provide flexible solutions that meets their needs, and today I will talk about all three of those. A great example of where these three things come together is really in what I call the PhenoCycler-Fusion 2.0. When we launched PhenoCycler-Fusion just about a year ago, as I said, it was the fastest spatial biology system. All these three things are now gonna be packed in the rollout of the PhenoCycler-Fusion 2.0. Let me give you some details about what's in the new rollout for the PhenoCycler-Fusion 2.0. Before, we had a large menu of à la carte content where customers can pick and choose the targets they want, and then be able to build the assay for the application. We'll still continue to do that, but now what we're going to do is take all that content, package it in very application-specific themes that are ready to use out of the box with very little development, very little optimization. We're also gonna expand that capability not just for protein, but also RNA as well. The next thing we're gonna do that's different from PhenoCycler Fusion 2.0 versus 1.0 is that we're increasing the speed. We're going from the ability to process about 10 samples a week to now processing 20 samples a week. We already had sophisticated proprietary compression algorithms to make sure the data was manageable in the 1.0 world. In the 2.0, we're gonna have an ecosystem of partners who can use and provide data analysis solutions for our customers. Let me start to going through each one of these stages one by one. The first is an update on where we are on the RNA side of things. You may recall we announced a partnership with ACD at the beginning of the year. ACD has the number 1 chemistry for spatial RNA analysis, with nearly 6,000 publications supported. Akoya is the number 1 company for protein spatial phenotyping, and this partnership allows us to bring the best of the chemistries and the workflows together. ACD's capability historically has been built with looking at 1 to 4 RNA targets in a particular tissue using conventional microscopy technology, including compatible with our PhenoImager HT as well as our PhenoImager Fusion. This partnership is not about that. What this partnership is really about is expanding that capability, not just 4 targets, but 12 targets. Creating a workflow for 12 RNAscope assays that can be fully automated, walk away on the PhenoCycler-Fusion, enabling entirely new spectrums of applications that can be done with the RNAscope technology. Let me show you some examples of the RNAscope chemistry, the 12-plex assay run on the PhenoCycler-Fusion. What you're looking at here are four cycles that are run on the PhenoCycler-Fusion. Each cycle reveals three unique RNAscope targets, measuring targets. What you'll start to see is you can see low expressers, high expressers, medium expressers. The RNAscope chemistry works beautifully on the PhenoCycler-Fusion at high resolution. These do not show regions of interest on the tissue. We're actually enabling you to look at the entire tissue and look at RNA expression across the entire tissue section in every cell. The reason we can do that is because we have the fastest imaging technology that allows you to do not just protein in every cell across an entire tissue section, but also allows you to look at RNA across the entire tissue section at high resolution. Now what's different about the RNAscope technology is as you zoom in, you will start to see gene expression levels within a cell, punctate expression spots that are different from what you'd typically see with protein expression, where you'll see entire cells coded. What this allows us to do is now we've combined the best of capabilities. We've got the fast imaging technology that's been historically proven for protein assays. With our partnership with ACD, we allow completely flexible RNA solutions. More color on this. What the ACD partnership allows us to do is enables customers to pick any RNA target up to 12, design the assay for targeted validation or screening experiments, and be able to run that on the PhenoCycler-Fusion in a fully automated manner. This capability will roll out in the first half of next year. At the same time, we know there are customers who wanted to do discovery experiments with RNA. They may wanna look at 50, 70, or 100 RNA targets. On this front, we are working on a proprietary chemistry that takes us to that dimension, and we'll start to make more announcements on the availability of that chemistry. We expect that to roll out in the second half of next year. Now let's talk about the chemistries. There's chemistry for RNA, there's chemistry for protein, there's assays for discovery capabilities, there's assays for signature development. The chemistry that ties them all together, we used to call this the universal chemistry, because it works across our platform suite. Today, I'm really excited to announce the brand of this universal chemistry, PhenoCode. What PhenoCode represents is a premium brand for spatial phenotyping panels and assays. The promise of PhenoCode is that these assays are optimized and ready to go out of the box. They don't require any development and very little optimization to run your samples. This is how we're gonna make our assays simpler for our customers. This is important because when we think about all the applications that Brian talked about in the discovery market, the translational market, clinical research market, our goal is to make sure the PhenoCode chemistry spans the continuum. The hardware is a continuum, same imaging technology that we use in discovery versus validation, versus clinical, same chemistry flows through as well. It's really important to have that continuum so you can drive those discoveries and translate them into the clinical setting as quickly as possible. Let me give you an example of what these things look like. There are gonna be discovery panels, which are high plex panels that allow people to discover novel information. There are also gonna be signature panels. They're small, targeted panels that allow customers to develop signatures for clinical, downstream clinical use. Early in the next year, I'll tell you more about the rollout of the discovery panels. Today, I wanna focus on the signature panels and what they're about. What really is a spatial signature? Using the PhenoImager HT or the Fusion, you can look up to 6 markers across an entire tissue section at single cell resolution. It will tell you what types of cells are there and where they're at. A spatial signature is more than that. It's looking at relationships about which cells are sitting next to which other cells, and what can we infer about the function. On the equation below, I show you an example of a spatial score. It looks at the relationship or the distances and how they matter. If there's a tumor cell and a T-cell sitting next to that tumor cell, then the patient may have a better response to immunotherapy. If that T-cell is sitting next closer to a regulatory cell, then it's gonna have a suppression function, and the patient may not respond as well. These types of algorithms will allow us to develop new class of diagnostic assays going forward. Now let me talk a little bit about what are these PhenoCode panels that we're launching. We've intelligently designed five panels that are really designed to look at the tumor microenvironment and interrogate the tumor microenvironment. Some of these panels look at the immune profile within the tumor microenvironment. They look at the TIL status. They look at to see if there's immune exhaustion. The idea is that it captures very specific biology that allows our customers to develop signatures for downstream clinical use. These panels have five unique targets and one spot where a customer can add their unique target of interest. This creates flexibility for the customer, as well as these optimized panels are ready to use right out of the box. They take away all the development and optimization efforts customers have to go through to build their own assay. On top of that, we're doing more to increase the speed of our systems. Specifically here, when I talked about the PhenoCycler-Fusion 2.0 rollout, how are we getting to a faster system? There's a couple of things we're doing. First, we've optimized the microfluidics flow through that system to make sure things are going faster through the system, exchange rates are faster. On top of that, what we've done is created a multi-slide carrier, so you can do more than one sample at the same time. The ultimate product is the entire system is now 2x faster. Twice the amount of targets, twice the number of samples that you can look at in the same unit of time. This upgrade also allows the PhenoCycler-Fusion to be able to run the RNAscope technology as well. I said before, the Fusion 2.0 upgrade will roll out to our customers in the first half of next year. The question is: why are we doing all of this? Why are we developing content that's ready to use out of the box? Why are we making the system go faster? The reason is because we want our customers to explore the entirety of the applications in the world of spatial biology. Whether it's looking at phenotyping, rare cells, functional state, looking at signatures or spatial neighborhoods. We want all that to be possible. Having the content in a fast, seamless workflow allows our customers to do that. We're not done yet. There's still one more thing we have to do, and that is the data analysis problem. There are two challenges in data analysis in spatial biology today. First is that different users have different needs. If you're interested in doing discovery research, you may want the most flexibility in your data analysis tool because you wanna explore the space, the tissue, look at correlations and interactions. If your interest is more in the translational clinical research, you may want something a little more streamlined, robust, scalable. Those are fundamentally two separate types of needs that we have to serve with data analysis solutions, and how might we do that? That's problem number one. What's the second challenge in spatial biology? The second challenge is big data. Of any data sizes in life sciences, spatial biology generates the most data per sample in the order of terabytes of data per sample. When you have terabytes of data being generated per sample, it serves as an anchor in your research, really slows down how fast you can move. Data that size is too big to move, too big to store, too big to analyze. What we've done through our workflows is created a proprietary algorithm that compresses that file about 34 down to gigabytes. When you go from terabytes to gigabytes, the entire world of data analysis solutions opens up. How might we use that? First, when we think about our customers, we have customers who like desktop solutions. There are customers who may prefer cloud capabilities, or they may prefer a surface provider. Some customers prefer open source systems, some want bespoke services and capabilities. You can think about differences for an academic researcher versus a biopharma or someone who's doing a clinical trial. Widely ranging needs of data analysis solutions. How do we make sure we provide the right solutions to the right customers? Fortunately for us, there are world-leading companies that have solved these problems. Companies like OracleBio, Enable, QuPath, PathAI, Visiopharm, and Indica Labs. All of them provide best-in-class solutions to serve very specific needs in these very specific markets. What we're able to do is because of this proprietary compression that generates these QPTIFF that are only gigabytes, we can access these and even more data analysis solutions. We can actually meet our customers where their needs are with providing data analysis solutions for every one of these capabilities. If a customer is interested in high plex data analysis and they wanna do that in the cloud, we have a partnership with Enable Medicine that will do that. If the customer is interested in doing a flexible open source, we have QuPath that addresses their needs. If they're interested more in machine learning and AI capabilities, then we have partnerships, PathAI, Visiopharm, and Indica Labs that support our customers through that process. If customers are interested in more of a bespoke service, we have a partnership with OracleBio that does that. The way we solve these problems for data analysis is making sure we partner with the right technology provider and provide the simplest, easiest solution that fits the customer's needs. Now, I'm gonna show you a video that's gonna explain exactly how our data solutions work and how they meet the customers where they are. Data is beautiful. Its life begins raw as a random collection of information waiting to be refined. The stories data can tell are limitless. These stories drive innovation and action. Managing large swaths of data requires software solutions that are accurate and fast. Akoya Biosciences innovates powerful software solutions for comprehensive workflows spanning across spatial phenotyping, rare cell discovery, and cellular neighborhood analysis. Traditional spatial imaging technologies can generate terabytes of data and take days to transfer. Akoya's proprietary algorithms utilize a compressed QPTIF file, providing analysis-ready data in a matter of hours. This significantly smaller image file may also be quickly uploaded to one of our many software solutions. Once uploaded, every cell within the tissue must be accurately segmented. An essential step for phenotyping requires accuracy and precision and directly impacts the quality of phenotyping. To avoid the challenges of cell segmentation, Akoya offers AI-based solutions that map out each cellular boundary across an entire tissue section. High-quality images are the ultimate tool for the assessment of phenotypes, making Akoya's PhenoCycler-Fusion platform the ideal choice for examining all relevant cell types and rare cells for truly unbiased phenotyping. Machine learning can be used to teach the computer how to recognize cell types of interest. This process is capable of whole slide analysis and is user-friendly. Morphology and biomarker expression data can be cumbersome. However, cell classification and clustering algorithms can be leveraged to identify cell populations of interest, expediting hypothesis-driven experiments. As the story of cellular interactions take shape, we can explore the cellular neighborhood. Cells that organize into distinct spatial patterns that may influence disease progression. Some common methods of cellular neighborhood analysis include raster scanning, the cell-based method or pairwise data. To better understand how cellular neighborhoods interact and communicate, several techniques are available. One being UMAPs, which is a dimensionality reduction technique that shows the co-localization of cells and indicates the similarities between cellular neighborhoods. Another is heat maps, which show the composition of cellular neighborhoods. The most exciting part of the story our data tells us is the statistically significant relationships that ultimately become actions. These powerful software solutions are already in use today, helping to build cell atlases, discovering novel cell types, and developing a spatial phenotypic signature. The beauty of data makes it so we don't have to ask what if, but what else. Learn more at akoyabio.com/software. All right, with that video, let me wrap up all the product launch announcements I shared with you today. Our goal is to go from having the fastest spatial biology system to now having the fastest and the simplest workflow. The announcements I shared with you today were around the PhenoCycler-Fusion 2.0, the multi-slide carrier that increases the throughput by 2x. We talked about the launch of pre-designed panels that are optimized and will require very little development work. These are the PhenoCode Signature Panels coming now and discovery panels coming soon after. We talked about some really exciting partnerships that are enabling an entire world of application tools and capabilities. The partnership with ACD for the first time, a fully automated 12-plex chemistry that's ultra-flexible. On the back of that, we'll talk about our proprietary chemistry that enables you to go to the 50 to a 100-plex RNA. We talked about how we solve the data analysis problem by partnering, enabling a unique proprietary compression algorithm that makes the data manageable to work with across a wide variety of partnership solutions. With that, I wanna thank you for your attention. I'm gonna pass it on to my colleague, Dr. Oliver Braubach, who's the Director of Applications, who's gonna talk about some of the discoveries that are being made off our solutions. Particularly, he's gonna focus on the PhenoCode Signature Panels. Hello, everyone. Thank you for lending us your time today. My name is Oliver Braubach. I'm the Director of Applications Development at Akoya Biosciences. I'm going to speak to you today about our PhenoCode Signature Panels. Before I do so, I would like to dive into the idea of spatial phenotyping and spatial signatures a little bit more deeply than Niro had in the previous presentation. Spatial phenotyping, spatial signatures, it's a new way of looking at tissue biology. It's interrogating the tissue not only for the presence of a biomarker, but trying to understand exactly in which cell that biomarker is expressed, and by knowledge of that biomarker being expressed in thaT-cell, inferring the functionality of thaT-cell, and then not only in thaT-cell, but all of the other cells around it. The function on the tissue is the crux of cancer biology. If we can understand this by means of spatial phenotyping, we will get a big step closer to truly understanding what is happening during this disease, and hence, also coming up with potential treatment strategies. This slide here is 1 example of how this works. On the left-hand side, you can see a comic, a representation of two types of immune cells, a green T-cell and a blue T-cell. Then there's a brown cell, which is a cancer cell. The blue cell's job is to attack the brown cancer cell. However, it can only do so if it's exerting an effector function. That is, it's not being suppressed or it's not being held back by the green cell. You can look at biomarkers, you can see that these cells are there, and you can call it a day. You can actually measure the interactions between these cells, and this is what you do with spatial phenotyping. You know precisely which cell is there, what biomarker is expressing, and what the associations between these cells are. We can now say that the blue cell, by virtue of being closer to the brown cancer cell, is able to attack this cell and hence kill the tumor. If the blue cell is, on the other hand, closer to the green cell, it is being immunosuppressed. If you take this data and you calculate out, you see what happens here, you come up with a spatial signature or a spatial score. That spatial score shown on the left will show you that a low score in a responding patient is significantly lower than in a non-responder patient, which has a higher score. We now have a metric, a biomarker that we can apply to responders or non-responder to stratify them, to see whether or not they will respond to immunotherapy. This is absolutely. It's groundbreaking. It is very, very important to also understand that in this experiment in which they are done, which they studied cutaneous T-cell lymphomas, other more traditional biomarkers such as a PD-L1 score, tumor mutational burden, interferon gamma screen, they all failed to stratify patients. Spatial phenotyping, however, by means of this spatial score, was able to detect a significant difference between patient cohorts to say whether or not a patient will respond or not to immunotherapy. That is amazing. That's groundbreaking. This is a whole new potential world of new biomarkers or spatial biomarkers that we at Akoya are digging into. Others have taken notice of this, there's a lot of research going on right now in trying to characterize tissues. This slide shows you an approach in which you're using an unbiased graph-based means to break down a tissue. Every single cell is cataloged. Connections between every single cell are measured. If you do so, you can take a tissue. Moving from the left to the right on this slide. Left-hand side, you can see in bright red the expression of an epithelial marker that may be tumor. We know we can measure biomarkers. We've been doing this for many years, but that knowledge alone is not sufficient to know what's going on in that tumor. You can break down every single cell relationship to every other cell, and you can come up with a concept that's like, has a spatial phenotype or a cellular neighborhood. This image on the left-hand side, the red markers, has turned into green and yellow and blue. This is showing you individual aggregates and cells of neighborhoods that you can now quantify. You know that these spatial phenotypes are present, how many there are present. If you then go back and you look at your patient populations, the left-hand side showing you the mere biomarker expression in a survival curve. This curve shows you, as it goes down, the patients aren't faring so well. There's no difference between the three curves. This curve shows you that there's no difference in knowing the biomarker expression for these different types of cells. The mere presence of a cell type is not enough to stratify your patient populations. Moving on to the right-hand side, though, we're now looking at these spatial phenotypes. Unbiased methods, finding out how the tissue is composed and how it perhaps functions. You can see clearly that the yellow line, that patient population has done differently. This is the same dataset analyzed in different ways. Biomarker or cell type expression on the left versus spatial phenotyping on the right. Spatial phenotyping shows you something that conventional means have not done. This is revolutionary and really groundbreaking way now to look at cancer samples. From a pathologist's view or a scientist's view, really try to understand the cancer biology and advance immunotherapy. To that end, really excited today to talk about our PhenoCode Signature Panels. These panels are designed with everything I've just told you in mind, be able to spatial phenotype, to do so in millions of cells in a sample. To do so in an environment where the biomarkers are constantly changing. Every day, we learn about new cell types or checkpoints. Every day, we have the urge, or we have the need to incorporate new markers into our panels. That is what these panels are designed to do. The PhenoCode Signature Panels, they're relevant. We designed these panels after mining hundreds and hundreds of publications for content. We know what the discovery in the translational community are using when they're doing these types of experiments. We've included these markers in our panels. This makes them highly relevant. They're fast. Adoption of these panels should not take you months. It should take you merely one or two weeks. This is much faster than you have been able to do this in the past or what you can do with competitor panels. These panels are flexible. We know that every scientist, every lab, has a marker they're interested in that may or may not be something that we have in-house. These panels will accommodate these markers very easily with off-the-shelf a la carte options. These panels are scalable. Scalable meaning that they've been designed around our Opal or our PhenoImager HT technology. This will allow you to not only study 10, hundred, thousands of samples and make clinically or translationally significant observations about potential spatial signatures. This is what these PhenoCode Signature Panels look like. I'm showing you here five different ones. Each one has a purpose that it is designed to do. On the top left, you can see an immune profiling panel. On the bottom, you can see a panel that will allow you to look at macrophage polarization. Each panel consists of five basic markers. These are the core of the panel. A la carte options are then available to plug and play into your panel. Let's see how this works based on this immune contexture panel. Here you have an immune contexture panel that allows you to look at some very important, very basic immune cell lineages: CD8 T-cells, macrophages, but also has the expression of PD-L1, FoxP3, and of course, PanCK to mark epithelial or tumor. What you can do now is you can swap in each individual marker to generate a completely separate panel that will allow you to answer a different question, starting with where are the helper T-cells? Are the T-cells exhausted? Is there any proliferation going on, and so forth and so on. There's about 24 a la carte markers that will be ready, that are now ready at launch. We know that this is not enough. Therefore, as you can see in the red, the red question mark on the bottom, there will always be the option to incorporate your own marker by conjugating a barcode to this marker and using it. Whatever marker you have in the lab that you are interested in that we don't have yet, you can incorporate these and plug them into these panels. The amount of biology that you can now study is really it's quite amazing. There's a lot you can do with this design. This is one example here. Once again, we're starting with the immune contexture panel on the top left. If you add CD20 to this panel, you're not just profiling the contexture of the immune panel. You're not just looking at T-cells. If you add CD20, it allows you to look at B-cells. You can see on the top right-hand side, bright hot orange color, tertiary lymphoid structures. You have now detected these just by adding one antibody. Tertiary lymphoid structures we are now starting to understand are incredibly important in mediating the response to certain immunotherapy agents. Knowledge of them being present or not could really give a lot more information about the outcome of a treatment. On the flip side, if you wanna know whether or not the T-cells are exhausted or not, you add PD-1 to the same basic panel. You're now completely changing the theme of your research. Easy to do. Shouldn't take more than a swap in of an antibody and a serial section, and you're ready to go. It's incredible, you can probably just move on from there and swap in any other marker you want. There's a lot of questions that can be answered, a lot of potential spatial signatures that can be discovered. We've vetted these panels very closely. We want to make sure that their performance is stable across the board. This is what you see here. Here you have two designs of the same panel, same immuno contexture panel, you're adding in PD-1 on the top and CD20 on the bottom. The idea here is that you should collect the same information about the differenT-cell types, no matter what the swap-in is. Adding an antibody in should not change the behavior of your core panel. It does not, as you can see in this histogram. You're finding the same number of CD8 T-cells, the same number of macrophages, the same number of FoxP3 cells. No matter what you're swapping around, your core is always the same. The same is true also for the swap-in. If you swap in a CD8 marker onto one of our core panels, always find those CD8 T-cells no matter which core panel you work with. Moving on. We've also extensively vetted these panels against traditional immunohistochemistry. You can see here two rows of staining, two panels, immune contexture and the immune profiling panel. On the top end, you can see the immunofluorescence images that were acquired with our system. On the bottom, you can see the representative DAB IHC. The staining patterns are the same. They have been vetted by a pathologist and have been deemed to be representative and the same. We're very confident that our immunofluorescence technique is capturing the same as conventional DAB IHC. On the right-hand side, you can see a distribution of intensity values. Essentially, what this shows you is a uniform staining, a uniform distribution of this signal across tissues. It's not high in some regions and low in other regions. You get a signal, a readout across a tissue that is stable. That, of course, is extremely important to make accurate observations and essentially statements about a tissue. This is what we can demonstrate here. Last but not least, I would just like to say that the workflow here, this is a quick overview. There are a couple of key steps that make this workflow different from conventional Opal workflow that some of you may be familiar with. If you walk through the steps from 1 all the way until 8, the key differences have been highlighted in purple. Step number 4, rather than applying an antibody, imaging it and stripping it again, that is the primary antibody, we cocktail all of our antibodies now. All of them have been barcoded, all of them can be applied onto the tissue in 1 step. This both saves time and preserves your tissue. It's a big advancement. The other 1 is we don't use secondary antibodies, as you can see in step number 6. Our antibodies are in an oligo tab. HRP molecule, which is used to pull down our Opal dyes, has a complementary oligo and can be applied this way. This is both more gentle and much faster to achieve than primary, secondary antibody labeling. Last step, number 8, is the hybridization and removal of the complementary oligos. Once again, this is a little more gentle. There's no stripping involved, and it's much faster to achieve than what you would have done before. Overall, these few key modifications that we've made have both aligned our antibody portfolio across the board, all the way from our PhenoCycler-Fusion systems to our PhenoImager HT and Opal technology, and also made it significantly faster, easier, and dynamically possible, frankly, to incorporate new targets into your panels as you like, without incurring a huge amount of cost and time. Well, with that, I would like to thank you for your attention. I would like to now show you a really nice video that we've made that introduces this exciting new product. Thank you. It's my pleasure to welcome our next speaker, Elizabeth, Dr. Elizabeth Newman. She's an Assistant Professor at the Department of Chemistry of the University of California, Davis. Her research focus on understanding the molecular and cellular architecture behind neurological diseases. This highly interdisciplinary research involves developing analytical tools and multimodal imaging methods for understanding complex biological phenomena. Before joining UC Davis, she was a National Institutes of Health Postdoctoral Fellow at Vanderbilt University in Nashville, Tennessee, where she was developing an open global atlas of the human body at the cellular level. Dr. Newman received her doctorate in analytical chemistry at the University of Illinois Urbana-Champaign. Welcome, Elizabeth Newman to our Spatial Day. Thank you for the lovely introduction, Hiro. Again, I'm Elizabeth Neumann, and I'm coming from the chemistry department at UC Davis. What my lab does is develop multimodal imaging approaches consisting of things like highly multiplexed immunofluorescence and MALDI mass spectrometry imaging. To begin, part of my postdoc was in HuBMAP or the Human BioMolecular Atlas program. This is an NIH program or funded program that focuses on building a cellular molecular atlas of different organ systems. That's a mouthful and doesn't mean much. I actually get a question of what even is that? If we think about a world atlas or a map and say we are interested in Norway, a good world atlas gives you multiple views of where Norway is at. It's not just, okay, Norway is here. It tells you where Norway is at in conjunction of the entire globe. It also, though, gives you a higher spatial resolution view with these roads, rivers, and maps. There's even some topography. You know where Norway is at in conjunction of other places, so on and so forth. We're kind of doing a similar thing with the kidney, where we can have a kidney. Kidney is broadly divided into the cortex, medulla, and or outer medulla and medulla. What we're trying to do is say, okay, given a normal healthy adult human kidney, what does it look like? What kind of cells and molecular features does it have, and how does this vary as a function of space resolution? For those who aren't renal pathologists in the room, which is me too, on a bad day. The workhorse of the kidney is this nephron unit, where it has a glomerulus on the, on the cap, and then different tubular segments that are involved in reabsorption of different nutrients that are filtered from your blood. Again, we're trying to build an atlas that shows us the cellular and molecular construct spatially within the human kidney, but also as a function of different spatial resolutions. To give you an idea of what kind of data set we're looking at, and I'll be walking you through, we have patients, I think it's about 25, I lose track every time I go through this slide. But about 25 patients that range from age from 20 to 77. A mixture of men and women as well as mostly white patients. In future experiments, I think it's more important to include multiple races and diversify our patient population, although we're running into roadblocks there. It's definitely an area that the NIH and a lot of other labs are trying to supplement. Lastly, we have their height and weight, so we can calculate their BMI. We have a mixture of BMIs that are considered healthy BMIs or some that are considered unhealthy. We then again have handed our tissues off to the Fogo lab where they go and very critically annotate different areas of disease or disorder. Even if the kidney is functioning well, it might still have pockets where there's glomerular or glomeruli that are sclerosed or other types of lesions, especially as individuals increase in age. It's important to supplement our cellular and molecular profiles with this kind of spatially resolved pathological information. Here is a Conics multiplex immunofluorescence image of a human kidney. Again, while I'm showing just 1 single patient, the data towards the end will be a composite of all 25 of our human patients together. To orient you, here's the division between the medulla and the cortex, where the medulla is on this side and the cortex is on this side. If you remember nothing else of our nephron unit, this is the glomeruli or the beginning of the nephron, where the blood is basically filtered within your kidney. These are different blood vessels. There's 7 different markers displayed here. We had a total of 25 antibody panel. What I want you to appreciate is that every discrete color is a differenT-cell type. Even looking at only seven markers, the diversity of colors you see in the zoom-ins is quite stunning. In fact, we get that spatial resolution across our entire image. Again, we're spanning very single cell-oriented resolutions to whole tissue segments. We can then use this information to semi-quantitate cellular neighborhoods. I've shown heat maps here. The numbers are Spearman correlation coefficients. What I want you to really take away from that is that the higher or the closer to one a number is, the more correlated those cell types are. These are often correlated with one another because I've narrowed our scope to things that should be correlated or glomerular markers. Basically we have a cell, and we ask the computer how often another cell is present within a 20-micron radius. Like a 2-layer cellular neighborhood. We use CytoMap or a slightly hacked rendition of CytoMap. How you read this is a cell expressing nestin has a 12% chance of having a cell expressing vimentin within its neighborhood. Likewise, a nestin cell has 60% or 64% of nestin-expressing cells also have a PARP-1-expressing cell in its neighborhood. This is really neat. The power of this data is we're not just looking at a handful of cells from one patient. We're looking at demographical information across 25 patients for millions of cells. This is a really large data set. We can subdivide our demographics based on sex or on BMI and actually look to see cellular neighborhoods getting reordered as a function of one of these two variables. Yes, we've included some of the other comorbidities, so we can say, "Oh, okay, well, it looks that men have a differential expression of KDR than women. Is it actually something else that we're not controlling for? This is an experiment." We have all of those comorbidities online, so you can look at their individual heat maps. Then while really exciting and powerful, we like mixing things together in a multimodal lab. This technology is mass spectrometry imaging. What this allows us to do is look at the spatial orientation or localization of discrete lipids within a spatial context. What I want you to appreciate is pixels with red have a high abundance of this PA, green, this PE, and blue, this PI. In this first image, you see almost no red, green, or blue pixels. That's because it's a dynamic abundance of these three lipids that create this rich architecture. You'd actually miss this without having the spatial context. You're just doing bulk analyses. Again, we're looking at hundreds to thousands of different features. Here's about a fourth of the dataset that I'm able to actually put a discrete name to. This is after removing isomeric species or isotope distributions, where each thumbnail is a slightly different view distribution of this m/z value, which correlates to a lipid. While really exciting, it's actually hard to make sense of this really rich chemical data. That's where combining the two datasets makes sense. In this case, we have our nephron, where again, we have a glomeruli in different tubule segments. Here on the right is some violin plots with the lipids detected by MALDI. By using multiplex immunofluorescence, we're able to segment out these different functional units and then use machine learning to determine the top features associated with, say, glomeruli compared to every other pixel. There's a high abundance of this sphingomyelin within glomerular pixels as opposed to all other pixels. While I've shown glomeruli here, we can do this for anything we've stained, so the proximal tubule down here. Can also segment based on sex or in this case, BMI, and use this to quickly screen different features associated with these different descriptions or characteristics. What's neat about this data is you can have very tight distributions associated with individuals of high BMI and much narrower distributions with those of healthy BMI. Largely the lipidome is unexplored. This is one way where we can start narrowing our data into testable hypotheses where we can say, "Oh, okay, these are the top 4 discriminators of sex or of BMI. What are these perhaps doing?" For instance, it turns out this oxidized phosphatidylcholine actually is high abundance in kidneys undergoing renal failure. We can then backtrack and look at these pixels associated with a high abundance of this oxidized phosphatidylcholine and see if there's any comorbidities present there. Finally, we're not again limited to looking at whole kidneys. We can separate based on functional unit. We have like a glomeruli, or what a distal tubule of an obese individual looks like and how their neighborhoods change as a function of functional tissue unit. With that, I hope that's shown you that you can use CODEX and mass spectrometry imaging to look at very rich chemical and cellular information within human tissues. With that, again, I'll highlight the Vanderbilt team here. I've since moved on. I know Jeff and Richard are both still working on the HuBMAP project. Everyone who's highlighted in italics here, were wonderful research colleagues and did a lot of the work that I'm presenting here. With that, if you're inspired by research, we're always looking for collaborators and students. Here's my email right here, and I look forward to answering some questions and maybe establishing some collaborators. Thank you, Elizabeth, for that great presentation. Our next speaker is Dr. Arutha Kulasinghe. Dr. Arutha Kulasinghe is a Peter Doherty NHMRC Research Fellow and leads the Clinical Omics Lab at the University of Queensland. Dr. Kulasinghe has pioneered spatial transcriptomic approaches in the Asia-Pacific region, contributing to world's first studies for lung cancer, head and neck cancer, and COVID-19. His research aims to understand the underlying pathobiology by using an integrated multi-omic approach. Dr. Kulasinghe has published his research in over 70 manuscripts and is supported by the National Health and Medical Research Council, Australian Academy of Science, Cancer Australia, Cure Cancer, US DoD, and numerous philanthropic and hospital foundations. It's our pleasure to welcome Dr. Arutha Kulasinghe to our Spatial Day presentation. Thank you, Neeraj, for the introduction. Today I'll be talking to you about ultra-highplex spatial phenotyping, a new lens of discovery for IO response. My name is Arutha Kulasinghe. I'm an NHMRC research fellow from the University of Queensland, and we'll be talking about some of the really exciting work with the Akoya Biosciences. Spatial biology gives us new insights into disease biology. We've written a couple of pieces in GenEng Biotech on this topic, how spatial biology is reorientating the biopharma development, top 10 spatial biology companies and a few other articles here. We've done a real deep dive on the applications of this, both in the IO space but also in the COVID-19 space, looking at multiple organ studies. The immunotherapy landscape, we know that IO therapy is highly effective in a number of tumor types, specifically led in the field by the work done by Jim Allison in melanoma and then a few other tumor types. Where we know a subset of these tumors appear to be highly responsive to immune checkpoint inhibitor therapy. Non-small cell lung cancer, head and neck, bladder are other tumor types where an immune responsive microenvironment is present, but really being able to identify these patient responders from non-responders is a current unmet clinical need. This paper came out in Nature Communications last year, which showed that tissues and not blood are where immune cells function. It was a really nice piece which showcased both needing to look at the proteomic profile and the genomic profile of these tissues, but also being able to simultaneously gain the imaging data from them. We do a lot of studies in this space. This is using a targeted panel of markers looking at PanCK, PD-1, PD-L1, FoxP3, CD8 and CD68, where we can look at the composition of tissues, whole slide and tissue microarrays. We can look at the interactions and the distance mapping, which you can get. This is a non-small cell lung cancer study where we looked at 90 lung cancer patients across this cohort in tissue microarray form, a single biopsy per patient sample. We could look at the expression of these markers both within the tumor but also within the microenvironment, and then start to look at the co-expression of these markers as we moved into the tumor and as we moved away into the microenvironment. Features of co-expression within these. The largest study here is now documented. 45 were treated with immune checkpoint inhibitor therapy in the second-line setting. 45 were treated with standard of care chemotherapy. We looked at the cellular neighborhoods defined by this 6-plex panel in these cohorts, and we could identify 9 differenT-cellular neighborhoods. A neighborhood is saying that my cytokeratin cell might be close to a CD68 cell, might be close to a CD8 cell, and how many of those we have per core or per patient sample, and then extrapolate that into the whole cohort. What that provided us was we could ask questions from the data based on the neighborhood analysis. We could ask the question, what was the enrichment of these nine neighborhoods across the entire 90-patient cohort? We could see the CD68, PD-1, FoxP3 cell interaction that was highly enriched for in the non-response group. It shows you using an unbiased total slide, total TMA profiling of 90 cores, we can start to identify, even using a targeted panel, really interesting cellular phenotypes in the data that are predictive of response and resistance to IO therapy. This work is ongoing currently in the lab. I'll quickly go on to some really exciting work in skin cancer, where we're profiling almost at 1.4 million cells, right? That's a massive number of cells at 35 plus. We're getting expression profiles for 35 different markers for each one of those 1.4 million cells. We can then start to cluster the data to really identify the differenT-cell types within them and the features associated with the tumor microenvironment. What we can see in this tumor in the purple is the pancytokeratin CD68 and the macrophages in the blue, CD20 are the lymph nodes. This is skin cancer that's metastasized to the node from an immune responsive patient that even at 2 years out had a complete response to therapy. This is what an immune responsive tumor microenvironment looks like. What's really powerful now with spatial is we don't need to build t-SNE plots. We don't need to build UMAPs. We can see this directly on the tissue. This is now mapping those cell types back to the tissue in situ, which is really powerful, right? Previously, a couple of years ago, we'd be looking at the t-SNE plot, which is on the bottom right, now we can see those cell phenotypes on the tissue directly, right? There's no need to infer them. There's no need to, you know, imagine where the, where the signal is coming from. We can visually see that on the tissue in situ. That's really powerful for the field. When we saw this data initially, why this was really powerful for us was now we can go channel by channel, whole slide, 1.4 million cells, and look at the individual markers across this tumor microenvironment in an immune responsive tumor microenvironment for skin cancer. We can see pancytokeratin in the bottom right, Ki-67 in the bottom left, you can start to see the other sort of co-expression of these markers. The red being high, blue being low, CD44, HLA, CD4. What's really neat is you can go through these single channel by channel. You can imagine how powerful this would be for a pathology tool to really be able to start to see the expression of 35 markers, but you can also plex this higher to really define these tumor microenvironments. Where we think this will head to is we can now look at, you know, immune responsive microenvironments and resistant microenvironments at high plex. What we want to be able to do is when we look at these extremes of IO therapy, is we'll be able to distill down protein signatures that are associated with response and resistance. If we can define those, you know, 6-plex panels, now we have 6-plex panels that define a CR, that define a progressive phenotype. I think that's what we want to be able to do with these high plex studies. I'll go into a really exciting head and neck study we've collaborated with Akoya Biosciences on. This is really a real showcase of what's possible using these tissues. This is an H&E section from an HPV positive head and neck cancer oropharyngeal, this tumor was from a patient that was given immune checkpoint inhibitor therapy and had a partial response to IO therapy and died after 14 months. The question that arose in the lab was: Can we look at the tumor microenvironment and see features of resistance, right? Is there something in the tissue that defines this partial response? Are there areas in there that, you know, that we can see that might give us clues as to why this patient didn't have a complete response and had, you know, some response but not a CR. This is now looking at that 103 plus, and you can instantly see the architecture of the tumor microenvironment, right? You can see the multiple tumor regions. So this is just looking at four colors here. So the red defining the CDA, PD-1 in the purple, and so on. So the architecture is really powerful, right? If you look at an HME, you really need you can't really see the architecture from this tissue, and you really need to be, you know, well-trained to be able to see that. What we can see now as we plex, we use high-plex panels, is really being able to see thaT-cellular architecture. We're able to generate not just these pretty pictures of expression, we're able to then infer the cell types based on the clustering of the data. We're able to go from these multiplex immunofluorescent images on the left to these cell phenotypes or cell atlases that are defining these tumors. What really stands out is these red and yellow bits. These red and yellow areas are germinal centers in the middle, so that's normal tonsil in the middle. At the top right, you can start to see this red and yellow structure infiltrating that tumor on the right. That's a tertiary lymphoid structure that's infiltrating the bit of tumor on the top right exclusively. It's absent in the other areas of the tumor. That's. That was the first, you know, thing that came to our mind. You know, why aren't these TLSs throughout the tumor? Why are they only in one bit of the tumor? Based on Ki-67 co-expression with a few other markers, we can now break this tumor down into four different tumors. We can start to see these tertiary lymphoid structures are exclusively present in tumor region 4 only. They're absent in tumor region 1, 2, and 3. We know that these tertiary lymphoid structures sustain an immune responsive microenvironment, and there's a lot of literature now in melanoma, head and neck, and lung cancer which show that these TLSs sustain an immune responsive microenvironment. What's really key with the tertiary lymphoid structures is they need to be structured TLSs. They can't be naive or immature TLSs to sustain an immune responsive microenvironment. When we look at this tumor, we can break that tumor down into about 900,000 single cells in situ, and we can look at the subset of this tumor. The bit in the orange, green, blue, and purple. We can look at the cell phenotypes within these areas. When we look at the M-one macrophages, which are immune responsive, all of the M-ones are in this yellow bit, the same area where the tertiary lymphoid structures are. The M-two macrophages are in the green, blue, and purple region. The M-two macrophages are immune suppressive, right? On 2 dimensions, we can start to understand this partial response to immunotherapy. The lack of TLSs throughout the tumor and it only being exclusively in part of the tumor. The patient had inherent resistance, right? You can see that. Also based on the disparity between the localization of the M-one and the M-two macrophages gives you that additional dimension of macrophage biology, which is now starting to define this immune suppressive microenvironment. If we looked at CD68 in the green, it's highly infiltrated, right? It's throughout the tumor. If we looked at CD8, which is typically what a pathologist would look at, it's highly infiltrating the tumor. This patient's tumor would be called highly infiltrated every day of the week. You can see it right in the whites in this tissue. Where it got really, really exciting for us was when we started to look at the metabolic panels or the metabolic assays for this tumor. It's on a backbone of about 20 different proteins, we can start to look at the metabolic profiles in situ for the first time. What's fascinating in this data is that you can see that glucose-6-phosphate dehydrogenase, which is a marker of metabolic activity, is highly expressed in tumor region 3, the area which lacked TLSs and had the M-2 macrophages. Bax, which is a pro-apoptotic marker, is highly expressed in tumor region 4, 2, and 1. Bax is showing areas where there's necrosis, it's all dying or dead. You know, you can start to see that in situ metabolic data, which is potentially really transformative in that what you're seeing at a metabolic level, we can see paralleled at an immune infiltration level. It's the first time we've been able to do this, and it's a fascinating demonstration of really multi-omic assessment for IO therapy and how we can now, for the first time, see inherent resistance in situ. We've been able to develop now a spatial metabolic map of this of this tumor, and you can see the pentose phosphate pathway highly upregulated in tumor region 3. It's bordered by DNA damage and fatty acid metabolism, all of tumor region 4, 2, and 1 is high for a number of pro-apoptotic pathways. It's really defining a new biology that we're starting to see with high-plex data, which is really exciting for the field. We think that being able to develop this further is really gonna give us new insights into immune responsive microenvironments and immune suppressive microenvironments. With that, I'd like to thank all the collaborators, the patients who've been on these studies, and the funding agencies. Thank you. All right. Well, let's bring back our speakers, Elizabeth and Arutha, to talk a little bit more about spatial biology. Thank you both. Great presentations. What I want to start maybe with, you know, think a little bit about standardization, 'cause as you both know, in a variety of research fields, when new methods come in, once as a community, we come together and start to standardize. It makes it easier for other researchers to adopt a technology because they can then figure out which tools are right for their research. I think some of the movements we're seeing, particularly, Elizabeth, things you talked about, HuBMAP and others, where you're trying to coordinate large datasets, not just from one lab, but different labs, but also different platforms. That seems like the kind of initiative we wanna see more of to drive standardization and use of these new techniques. I wanted to maybe learn from you a little bit about in the spatial biology world, where do you think we can use more standardization that will really help the community in trying to figure out where to use spatial tools going forward? Absolutely. These are all great points, because for every scientist you have in a room, you have about 100 opinions. When you start to standardize things, it can get wild. If you think about you're trying to standardize antibody names, and people can have 4 or 5 different names for the same antibody, it's wild. I'm a mass spectrometrist by training, so, like, going back and trying to even learn from what other people have done, and they call the same thing different names, it's a, it's crazy. What I think is helpful is, 1, if the metadata is actually aligned. Mm-hmm. That, like, if you think about it from a data analysis perspective, someone just has to build 1 script, and they can then extract what they need, and it's aligned from different platforms. The other things that are helpful is, I don't actually have a good answer for this, is aligning and providing, like, raw data or semi-processed data so that it can be reprocessed at a later time. This gets problematic when you have giant data, which I think we'll talk about later in the roundtable. Also, like, calling things the same name. If you get a mass spectrometrist like myself, we're gonna call stitching and tiling, hopefully, stitching and tiling, but that might mean something else to the spatial transcriptomics people and even something else to the highly multiplexed antibody-based community. Thinking about how to actually reduce the number of acronyms and names, I think is also helpful. No, thank you. Well, Arutha, you know, Elizabeth brought something there in terms of, especially, I know you've used a variety of different spatial platforms as well. In terms of thinking about standardization, so you can actually compare data across different platforms on the data side of it, how do you think about that today, and what do you think needs to happen to get better standardization happening across multiple sources of spatial data, if you will? Yeah. It's a great question, Niro. I think to Elizabeth's point, I mean, we very much feel like cowboys, right? When we generate these datasets, and we put it out there, and then it's an afterthought. I think the challenge for the field is really being able to standardize every, you know, from your, you know, pre-analytics all the way through. You really need to define, you know, your assay, what are you measuring? If we just name an instrument or a tool, we assume the workflows to be the same, right? The ways you've captured your whatever, your antibody profiles or whatever it is, but every experiment is different. One of the challenges we face daily is standardizing both the pre-analytics, the running of the samples, the, you know, the data QC, the normalization, and having those pipelines, even the clinical metadata, right? If you've got clinicians curating, you know, clinical databases in 20 different ways, you need to standardize that for your bioinformatics team, right? A zero or a one might be very different for a live date, for example, right? I think just being able to standardize every bit of that workflow is critical. We see that... I think we'll get there in time. I think the challenge at the moment is we're running a lot of siloed experiments, and these efforts need to be collated. Efforts like the HuBMAP and sort of those initiatives will put this at the fore, I think. Really putting the data out there, which again, will become a challenge as the imaging data becomes quite large. Hosting these data sets is gonna become increasingly more and more difficult. Yeah, there's no easy way, Niro. I think we're all acutely aware of it, we're finding ways of standardizing them. You know, when my lab collaborates with other labs, we ask for a standard data format, standard, you know, what have you done? What was the QC? What was the normalization method applied? We have a set of parameters that we define, and then we work through that. There's always a lot of back and forth in really figuring out what was done in the experiment. Yeah. If I, if I step back, you know, several years ago, for folks who are thinking about doing spatial biology research and getting into thinking about how they might think about doing spatial biology, there was always this challenge between, you know, do I wanna look at small regions of interest at ultra-high plex or, you know, not at single-cell resolution? Or should I be looking at things at single-cell whole slide? More and more, it sounds like the appetite to do single-cell whole slide is becoming sort of the dominant format of doing spatial analysis. The, the next foray is thinking about, well, what's the analyte I should be starting with? You know, how important is protein to my research versus RNA? Liz, from your mass spec world, and you look at different types of analytes, how do you think about protein RNA? I know you're looking at lipids as well. I'm curious to hear how you look at these different analytes and how do you think about adopting which analyte to which question you ask, right? Absolutely. This goes, and I actually cut this part of my presentation 'cause I didn't wanna be long-winded. I usually start with a couple slides that are like, what even is a biological system? You can get all cute and look at the central dogma of biology, and it's DNA goes to RNA, goes to proteins. That like, we know that that undersells what's going on, and it's way more complicated than that. Like, that neglects the metabolome, which everyone's kinda just ignoring, partly 'cause it's too complicated. I show this big slide that has about 200 different technologies you can use to assay DNA or RNA or proteins or metabolites. Really, I think where a lot of people struggle is they don't actually understand what one class gives you and why it's beneficial to probe that class. It even happened in the meeting I had this morning where they're like, "Everyone's focused on transcriptomics." It's like, "Well, yeah, 'cause you can amplify that." They're like, "Well, why does that matter?" It's like, well, you bypass the dynamic range issue of most instrumentation because you can look at high abundance things, separate it out, and then look at low abundance things 'cause you can amplify the bejesus out of it. You can't do that with proteins. There is a discrepancy between gene expression and what's actually happening in the cell, especially when you look downstream to metabolism. That's why I really push for multimodal approaches because you can look at then multiple snapshots of the tissue at once. You can say, "Oh, okay, here's gene expression," but also what metabolites are actually there. Does this make sense? Does it not? Why does it... Usually, when it doesn't make sense, it's 'cause another cell's providing it. I think only through, like, a really open lens are you gonna really be able to understand complex biological systems or phenomena like learning and memory. Got it. What about you, Arutha, as you think about... I know you've done, you know, all kinds of spatial in terms of looking at whole transcriptome all the way down to 6 proteins. When you think about the dynamic range of analytes and plex, how do you think about the analyte requirements that you have for your projects and the plex level relevancy questions you ask? Yeah. No, it's a great question. To Elizabeth's point, I mean, we initially when we want to learn new biology, we go with the whole transcriptome approach, right? COVID was a great example, you can distill that down into, you know, your top hits, we typically validate using proteomics, spatial proteomics. What's happened for us That's been our traditional approach, right? What's happened to us in the last year and a half is that as the plex level on protein increased, now protein's becoming discovery too. Protein's really exciting for us because it plays in discovery. When you can get, you know, 50, 100 markers on protein and potentially capture metabolic data- Yeah. That becomes interesting because now that's a lens of discovery, but also translational, right? If you can, you know, do a screen using a 50 to 100-plex protein assay, and you can distill down what a complete response looks like, what a progressive phenotype looks like for IL, for example, we can then screen using a 6-plex or a 5-plex or even a lower plex than that, right? I think for us it's this transition from traditionally running to whole transcriptome, but now it's like, do we actually need that? Can we, you know, pivot and look at proteins, which is probably closer to the clinic and more translational and closer to the pathology community, which really excites us because that is likely gonna be the adoption of these technologies. It's not gonna be the RNA assays, unfortunately. We know that. I think being able to... Yeah, the high plex protein really excites us and some of the metabolic stuff is just fascinating. You know, it's the most exciting biology we've seen in the last year and a half. Being able to develop that and being able to screen on that is really powerful. There's this transition we're seeing, and we're actively now, looking at this, protein high plex. To Liz's point as well, that multi-omic assessment, I think, is going to be, you know, the next bit that we need to integrate. How do you think about the right blend of the RNA and the protein integrated multi-omic assay that gives you, like, the sweet spot of where you wanna start your work? That's a great question. I think to a large extent, it's what's around you, right? Like, I have a mass spectrometry system a floor down, so that's a really low bar for me. Whereas, if you don't have a mass spectrometry system, it's $1.5 million, and that's an incredibly large barrier to overcome. I think part of it's actually just practicality, and I think as long as you know the limitations of your approach, you can actually start in any one of those different fields and then start building a hypothesis, right? 'Cause you can get a list of proteins that are important, but if you don't have their PTMs, you're missing part of the story. But it's impractical to get PTMs for 1 billion different proteins that have wildly different concentrations in the cell. I think starting anywhere is a good place. I think starting in the most convenient spot for an individual's lab is great. slowly actually building out and reaching out to community members in, like, orthogonal communities about how you actually get integrated and do these experiments correctly, I think is the next major step. that's why we publish, like, community standards in a couple Nature-related journals. never ask someone, like, "Can you just do the mass spectrometry or can you do the multiplex immunofluorescence?" Like, that's the worst way to start. I get that a lot. Like, no, these are all highly complicated things. I think it takes an amount of collaboration and really thinking about, this is the data I have, how do I then take it and supplement it? starting at a point is also fine. Whatever is most convenient. Yeah. Arutha? Yeah, Niro, great question. For us, I mean, we look at RNA and then protein on serial sections, right? We try to collapse that and integrate those data sets. If you could get both those assessments on the same tissue, that's very powerful. That multi-omic single slide, you know, we haven't seen data coming out in that space, so that's very exciting for us. I think if that's possible at any plex would be really, really powerful, because you know that you're not moving in space, you're looking at the same piece of tissue. Whether that's 50 plex, whether that's 100 plex, I think that information is going to be really, really informative, because we've, you know, the field's traditionally done it on single slides, and then we're inferring down. Being, you know, the sweet spot again, it's sort of that golden question, right? Everyone wants more, I think there will be a trade-off of cost and, you know, the availability of the material to really do that. I think if you're in the 50-100 plex RNA protein range, that's gonna be really informative. It's hard to comment at this stage because we just haven't seen the data sets evolve, being able to get all that information on a single slide is potentially transformative. Yeah. Awesome. Well, let me ask this last question 'cause I know there's a lot of emerging discussions around artificial intelligence and deep learning and, you know, a lot of technology and companies starting and publications. I'm trying to think about where could that knowledge and that technology be best used to move spatial biology forward. In your opinion, if you were to pick sort of one area where this new analytical capability could be applied to help our learning get sooner, faster, where do you think that would be? I'll go back to you, Liz. I mean, my heart is in neuroscience. I'm an analytical chemist, but I'm a junkie for the brain. I think it's because if you think about some of the most complex things, the brain is one of them. Especially when you start throwing in, like, God, PTSD, like you give someone, at this point, it's trial and error. You give someone a drug, and you're like, "Do you feel better?" They're like, "No, it's actually much worse." You're like, "Well, that's cool." If you think about neuroscience and then, like, neurological diseases, I really think you need some of these high complexity data sets to really get a more complete picture of what's going on because these single plex data sets are not doing it. I mean, that's why we don't have, like, therapeutics for things like really effective for PTSD, especially non-combat PTSD. We don't even know anything about that. Even depression, which is common. Mm-hmm. We have nothing that's like, we actually just, you give drugs and then see what happens and then try again. I think that's really where the future is can you use these high complex data to actually look at some of the most complex systems that are out there. Great. Arutha, for you, where do you think the biggest applications are? Yeah, Niro, I think I'll give you an application of what's been the most exciting for us using these, you know, technologies and applying it has been in, you know, when you're able to host these data on some sort of, you know, cloud computing network or something like that, and you're able to sit around on a Zoom call just like this, right, with a pathologist, a virologist, an immunologist, you know, basic translational scientists and clinicians, and see that disease in high definition and then walk through that and incorporate everyone's opinions into the, you know, understanding what that looks like. That's been the most exciting thing for me, right? And then, and to Liz's point, you know, the neuro field is certainly, you know, Alzheimer's and all of those diseases now we can profile using these high dimensional modalities. It's been fascinating being able to see that, right? We're not sending, you know, figures to our collaborators. We're seeing these data sets at the same time, and we're understanding it with the lens of our whatever training background is, right? It's just fascinating to get that insight on those calls with them and understand that, you know, we've done it for COVID, we're doing it in the IO space, but it's just that, you know, being able to see that data, and it's powerful for every one of those individuals to see that data at the same time. I think being able to integrate, you know, your machine learning AI into that is gonna be key and sort of open source platforms. That's been the most exciting bit for us. Excellent. Well, I think, you know, every year, spatial biology seems to get more and more exciting. I wanna thank both of you 'cause for your presentations really reveal that we're just in the early stages of discovering information with the tools that we have. I'm really excited to see over the next three to five years what new capabilities come out and even the capabilities that we have to do biology at the scales that you're both imagining of. Thank you both for your time, and thanks for making the presentation today. Thank you, Niro. Nice to see everyone. Hi, my name is Gavin Gordon. I'm the vice president of Clinical Market Development here at Akoya, and I'd like to spend the next 10 or 15 minutes telling you about our clinical aspirations. First, our clinical vision, which is no less than establishing the standard for clinical protein multiplexing applications. What this means is the use of Akoya's platforms and technologies in clinical laboratories being used by clinicians to make routine decisions to decide patient care, diagnosing disease, determining which cancer subtype you have, assigning therapy. In order to realize our clinical aspirations, Akoya will need to evolve and is evolving as a company, moving from not just a life sciences tools company, which is the leader in the spatial biology market, but to a medical company capable of entering regulated markets around the world. Not just from an imaging platform on the product side, providing translational and clinical researchers with a means to discover the next generation of diagnostics, but actually implementing those diagnostics on a platform that meets requirements of FDA and other regulatory agencies around the world. Finally, and perhaps most importantly, providing the value to the ultimate customers that we serve in the clinical market, which are clinicians themselves. They make decisions to inform patient care. There's multiple places along the patient journey where laboratory testing on tissues can be utilized, and we believe here at Akoya that the most value that we will provide to the market is in diagnosis and predictive testing. The way this is done today in diagnostic pathology is the pathologist will measure the expression of a given protein on a single tissue section one at a time. If they need to look at 6 different proteins to make a cancer diagnosis, that means 6 different markers, 6 different tissue sections. There's several pain points here and inefficiencies that Akoya's platforms are capable of solving. The first is that even though the amount of tissue available for laboratory testing has remained relatively constant over the years, the number of markers that needs to be utilized in order to direct patient care has escalated dramatically, especially if you consider protein, DNA, and RNA. Number two, the readout from these traditional single-marker IHC tests are largely qualitative. They're not quantitative, unlike what multiplexed immunofluorescence can provide. As therapies are developed that require diagnostics that be more quantitative in terms of assessing the amount of protein present, not just whether it's present or absent, routine IHC falls a little bit short. Finally, in order to realize the clinical vision and the ultimate aspiration of protein spatial phenotyping, you need to have all of the markers measured in the same cell on the same tissue section at the same time. Spatial biology is really not possible or even practical using one marker at a time, single marker IHC. You can imagine a future where whether it's chromogenic colorimetric-based detection or multiplexed immunofluorescence, protein spatial phenotyping makes a huge impact on diagnostic pathology. Now, the case for predictive testing is a little bit more straightforward and is the subject of a large body of evidence in the scientific literature. In fact, today, protein spatial phenotyping using multiplexed immunofluorescence is already done in a number of laboratories. The reason is that it's easier in the case of immuno-oncology, for example, to better assess whether PD-L1, common IO marker, is present in the macrophages or the tumors using immunofluorescence. It's a much more challenging discernment to make using single-marker IHC. Today the report still contains single marker at a time in the readout, so you're assessing and evaluating a single marker at a time. You can imagine a future where all of the markers together with the spatial data, come together and provide binary output, either yes or no, in terms of predicting response to therapy. This feature is not too far away. Papers such as the Johns Hopkins study in Science and others around the world have shown that the promise for protein spatial phenotyping is closer than you think in terms of realizing the ability to utilize this technique to both develop and clinically validate algorithms that take the spatial component into context as well as the expression of multiple protein markers. In immuno-oncology specifically, the clinical opportunity is really rich, not just in terms of the medical scientific need, but also in the business opportunity. Here at Akoya, we've estimated the clinical total addressable market at about $7 billion, which is equal to the discovery and the translational markets combined. The overwhelming majority of the contribution to this clinical TAM is provided by oncology, and not just oncology, but immuno-oncology. What's going on in biomarkers and laboratory testing in IO today? As you can see on the right-hand side of the screen, two of the FDA-approved biomarkers for IO, PD-L1 IHC, as well as tumor mutational burden, are insufficient. They're associated with about a 40% sensitivity. That is their ability to identify true positive patients, and they have an unacceptably high false negative rate. You can see the ideal test at the top, which really identifies about 80% of the patients who are true positives and has a false negative rate that's ideally below 10%. We're not there yet. There's room for improvement. That's the problem. Well, what's the potential solution? Well, a group of authors published a study recently in JAMA Oncology Journal looking at this exact issue. This was a meta-analysis looking at different biomarker modalities, intensive studies across thousands of patients to see how each performed relative to the other. The way this is done is using receiver operating characteristic curves, and you see one on the left there. Where on the y-axis, you plot the true positive rate or sensitivity of the assay, and on the x-axis, you plot 1 minus the specificity, which is a measure of the false positive. If you were simply to flip a coin and assign patients to a responder or non-responder group, you would get this dashed line, no better than chance alone. The ideal curve is in the upper left, where you identify all of the true positives and have a zero false positive rate. Of course, we're not there yet. But when you look at assessing each of these, one way to do it is by measuring the area under the curve. The area under the curve, higher is better. The area under the curve for PD-L1 IHC is about 0.65. Protein spatial phenotyping using multiplexed immunofluorescence is associated with a dramatically increased area under the curve, about 0.85. And other modalities such as NGS or RNA expression fall in between. This does a couple of things. Number 1, it shows that the promise for protein spatial phenotyping in the clinic is there. This was a very good study, an incomplete study, but it shows the potential. Number 2, the number of markers in all of these studies, for protein spatial phenotyping were typically 2 to 3 at a time. Furthermore, it wasn't the same 2 to 3 markers. This provides broad support, not just for the approach, but for adding relatively small numbers of markers. You don't need tens of markers. Just 2 or 3 can dramatically increase your diagnostic accuracy, for example. That's what's happening on the biomarker space. What's happening in the drug development space? Well, pharma companies, biotech, biopharma, are investing $ billions annually into developing better therapies that improve on what you currently get with monotherapy checkpoint inhibitors targeting PD-1, PD-L1 axis. As you can see in the graph, over the last several years, the number of combination trials has increased at least 3 to 4x, right? Now we're including multiple drugs into the mix and evaluating them in clinical trials, but our success rate is still very low, typically about 1% to 2%. That's a problem, and it's a problem that a lot of pharma companies believe can be solved by better biomarker strategies. Not only are they shifting their approach to combination therapies, they're now acknowledging a significant unmet need in the biomarker space to better stratify patients. Now, Akoya's platforms and technology are already used in most of the major pharma companies today to drive translational research, and we'll continue to do that. To support our clinical market development strategy, we also have a very unique attribute, an asset that we can provide to support pharma partnerships. In fact, pharma partnerships are fundamental to our success in the clinical market. That asset is our clinical lab services group called the Advanced Biopharma Solutions team. This is a CLIA-compliant wet lab that does more than just facilitate laboratory testing to support clinical trials for pharma all the way across the spectrum from exploratory research use only, all the way through companion diagnostics and FDA submissions. It's not just a wet lab. It's a full-spectrum support model composed of scientists, pathologists, bioinformaticists, experts in various regulatory matters and jurisdictions, program managers, operations, and entire infrastructure to support our pharma partnerships. We've used this laboratory to support partnerships over the past 3 to 5 years, primarily in exploratory and feasibility studies to discover and validate new biomarker signatures. More recently, as you'll hear, to begin to develop companion diagnostics. Being CLIA certified allows us, as well as our also CLIA certified CRO partners, to provide clinical trial testing and support drug therapy trials to meet the highest level of stringency required to support FDA submissions. One of our clearest, most significant proof points in our pharma partnership strategy was announced this past June in a press release describing a companion diagnostic co-development agreement that we signed with Acrivon. This press release does 4 things. Number 1, it shows that the future is now for protein spatial phenotyping in the clinic. It's not 5 years from now, it's now. Number two, it shows that our ABS lab can actually meet the requirements to support a companion diagnostic partnership with pharma. In fact, prior to announcing this agreement, we worked with Acrivon on feasibility and pilot studies to bring their assay into ABS and use it to develop a clinical trial assay, which we analytically validated and provided with the IND application of Acrivon to the FDA, which was considered when they cleared the IND, and that was press released at about the same time. We've shown that we can set up a laboratory to support pharma across the full spectrum, and now we have some tangible proof points that we're doing exactly that. Number three, it shows that Akoya is the clear market leader. Pending FDA approval, this will be a first-in-class diagnostic. There are no other FDA-approved multi-marker spatial phenotyping assays out there. What does this assay actually do? It's an assay that's based on the PhenoImager HT workflow, that incorporates the Acrivon OncoSignature test into a companion diagnostic that, pending FDA approval, will be used to assign patients' eligibility to the Acrivon drug. The Acrivon drug, interestingly enough, is a targeted therapy. It's not even an immuno-oncology agent. This just shows that the opportunity for protein spatial phenotyping is even broader than in immuno-oncology. In fact, we think it's actually beyond, broader, has broader relevance beyond cancer as well. Importantly, from a commercial standpoint, Akoya will be the sole provider of this companion diagnostic, pending FDA approval. We partner with other pharma companies as well, and one that we've also recently announced is with AstraZeneca. This is actually in immuno-oncology, and it's focused on building the foundation for a biomarker discovery and validation platform that can be deployed to support AstraZeneca's clinical programs, but more broadly, immunotherapy developers more generally. We've done a lot of work developing this platform and conducting biomarker discovery studies with AstraZeneca. With our CLIA certification, last year, we're now in a position to provide support for clinical trial assays and enrollment assays, not unlike what we're doing with Acrivon in targeted therapies. Finally, in conclusion, bringing this back to the big picture, we believe that Akoya is uniquely situated to take advantage of the clinical protein spatial phenotyping market that's nascent at present, but has some very clear proof points that we've begun to establish progress against. The PhenoImager HT meets requirements for the clinical market, not just in respect to throughput, but also with robustness, repeatability, reproducibility has been shown in multiple multi-site studies as well as individual studies as exemplified by the MITRE study publication referred to here. Akoya, the company, has taken the necessary steps to become and will continue to take additional steps to evolve into a medical company. This is largely around our quality management systems, including ISO certification as well as CLIA, so that PhenoImager HT and associated reagents can meet requirements to be an FDA-cleared medical device. Finally, we've talked a little bit about the foundational effect that pharma has on our clinical strategy. Although we're focusing on pharma, it's not to the exclusion of other ecosystem companies such as PathAI, which you've heard about, and our partnerships with academic medical institutions, which you will hear about from our next series of speakers. Okay, moving into the clinical section of Spatial Day, our first speaker will be Dr. Laura Esserman. Dr. Esserman is a professor of surgery and radiology at University of California, San Francisco, and the director of the UCSF Breast Care Center. Her work in breast cancer spans the spectrum from basic science to public policy issues. Dr. Esserman is a recognized thought leader in cancer screening and overdiagnosis, as well as innovative clinical trial design. She led the creation of the University of California-wide Athena Breast Health Network, which is a learning system designed to integrate clinical care and research as it follows over 150,000 women from screening through treatment and ultimately clinical outcomes. Finally, in some ways, most importantly, she is the leader of the innovative I-SPY trial model, which is designed to accelerate the identification and approval of effective new agents for women with high-risk breast cancers. Dr. Esserman. Thank you. It's a pleasure to be here today. I'm gonna talk about how we've tried to unlock the science of response prediction through some of the spatial immunodiagnostics and some of the lessons from the I-SPY trial. Really, to give you a sense of that, I thought it'd be important for you to understand, you know, a little bit more about this platform and how these important platform trials can really help advance science, advance diagnostics, and bring precision therapy to patients. My disclosures, I sit on the Blue Cross Blue Shield Medical Advisory Panel, and I'm a founder of the board of, and a board member of Quantum Leap Healthcare Collaborative that is the sponsor. It's a not-for-profit group that sponsors the I-SPY trials, and I have a DCIS vaccine that's funded by Mark. One of the things that really drove my interest in trying to really tailor therapy was the use of imaging. Back in the 1990s, when I first joined UCSF, I was approached by Nola Hylton, who was an MRI physicist, writing the sequences for a new study that would then become breast MRI. She came to me and asked me, you know, how could I imagine using such a tool? The first thing that I thought about was, well, the best thing to do was to look in these clinically advanced patients and see whether their tumors all look the same and whether the response to treatment was the same. We found this really interesting background, and we actually still have not sorted out what is driving all this change. One of the things that was really clear is that every patient didn't respond the same, and all these new sequences was bringing incredible precision to or the possibility of precision, saying, "Wow, people were having different therapeutic responses." At that point, we had to say, "Well, there's no way we should be treating every patient as if they're the same, even if they're molecularly high risk or not," that we had to be thinking about doing something different. That really is what drove us towards building this platform, which was saying, "Look, we need to standardize the way we do imaging, pathology, biomarkers, and data collection, and democratize the way we use the information and get everyone to start." We initially started with anthracycline, started by paclitaxel, but when we wanted to start adding drugs, we switched it because paclitaxel was easier to put drugs together with. One of the things that we noticed was that the absence of tumor after neoadjuvant therapy, that complete resolution of disease or PCR, complete pathologic response, was an optimal early endpoint, but really only for molecularly high-risk disease. It worked better for by subtype. That the high-risk disease is different from the low-risk disease, and that's true for the immune phenotypes as well, that they are not prevalent in the molecularly low-risk disease and probably not an important target for those patients. I bring this up since it is, of course, World Cup time. It really was back when the American women first won their first cup, so this was back in the 1990s, but this really set the stage for collaboration and what has become this longest-running platform trial, I think, in medicine. The whole idea of this was to have a more efficient platform for understanding biology and being able to figure out how to tailor treatment. Our goal was to improve the way we evaluated new treatments. Again, this has been replicated and applied to different diseases. There's nothing specific about breast cancer except that it was a great place to start. We were able to accelerate knowledge turns and drive urgency and innovation and have trials that incorporated disease heterogeneity from the beginning, taking the understanding, especially in the setting of immunotherapy, that the metastatic or end-stage disease may not be the ideal place for drug development, much better in the early-stage setting than in the metastatic disease. The idea is to try and move drug development into the earlier stage setting when a complete response would lead to cure. Our whole goal was to identify these early endpoints that could be captured in the course of care. So reverse the order of therapy. I'm a surgeon, so I could say, "Don't operate on people. That's not the best thing to do. Let's get the systemic therapy right, then the surgery can be, you know, less intrusive and a good way of capturing that early endpoint, and whether or not we've had a good response. Looking for big signals, and it was a place where there were a lot of advocates involved in wanting to be part of trying this new approach, and that made breast cancer a great place to start. You know, our purpose was to identify agents that would improve that chance of PCR in combination with standard chemo in electrically high-risk stage 2, 3 breast cancer. We again showed that PCR was a very good surrogate for 3- and 5-year EFS and DRFS, disease-free survival with a hazard rate of 0.18. Again, this sets up the stage for having companion diagnostics that help us figure out how to tailor therapy. Prediction is improved when looking by subtype. Additional information is gained when looking at residual cancer burden. The strategy of using biomarkers for adaptation and graduation and testing qualifying biomarkers has led to the identification of new ways to characterize tumors, and that actually is leading to the chance that we can improve the chance that each patient can get to a PCR. And we've used functional tumor volume, which has been turned into a biomarker from MR, as a good predictor of response and has led to graduation decisions that have stood the test of time. We put 24 combinations through the I-SPY trial, with the graduation of pembrolizumab in 2016 to 2017, it really led to many more IO combinations that we've had the opportunity to learn quite a bit from. One of the things that we understand is that here's an example of a complete response that you that we see, now the tumor is here, and now the tumor is gone. This is at 3 months. That this is part of the pembrolizumab. This is triple negative patient that had all her tumor go away in the first 3 months of therapy. The reason why this matters is that you can look here. Here's the response. You can see that in the HER2 negative patients, in both hormone positive and hormone negative. Here's the triple negatives, and here's the hormone positives. You can see that there's a benefit, but that the large amount of benefit is here in the triple negatives, but there is some population here. I'll show you a little bit more about this. Turns out to be that there's about a third of patients in the hormone positive patients that turn out to be basal or have an immune phenotype. That actually turns out to be super important. Why? This is the average PCR rates. Why that matters is because, this is from Doug Gey's paper, that if you have a complete response, PCR, your outcome is much better where you stabilize here. This has stood the test of time, whereas you continue to do very poorly if you do not get there. It turns out that the degree of residual cancer burden makes a difference where this is a continuous outcome. It's not just getting to PCR, which is this blue line, that's 0, 1, 2, and 3. The degree of residual tumor does make a difference. This becomes another important way to sort of drive things forward. This trial is the biomarkers of response and non-response will drive progress. We have over 90,000 specimens annotated with short and long-term outcomes. That is really what the power of a platform trial really can do, and it informs our approach and the way we want to prevent metastatic disease. We started off with these standard assays, and we actually, you know, started out by looking at MammaPrint, which was we used the FDA-cleared 70 gene assay when we started. That was one of the requirements from the FDA. We actually had an IDE on the 44K array, so we had that back where we could investigate different ways of using and qualifying these signatures. We had DNA repair deficiency, AKT, HER2 immune signatures, and we had the immune, you know, the Vectra multiplex staining environment as one of our platforms for trying to understand how best to categorize the biology and the response. Over time, circulating DNA has now moved to qualifying, and the immune signatures have now moved to standard, which I'll show you in a minute. One of the most important thing that really came out of this trial is looking the first 990 patients. This is a paper that came out in Cancer Cell about five months ago, really highlighting the fact that we could see that the way we categorize tumors now using the receptor subtypes, ER, PR, HER2, is not sufficient. This is the group, this is Laura van 't Veer's group that led it, Denise Wolf and Christina Yau. Again, one of the most important things here is that, you know, people say, "Oh, the immune response or the immuno-oncology agents work best in the triple-negative subtypes." In fact, they don't work in everyone. They work only in people who have an immune signature. We can actually go more into that in a minute, and I'll show you that in using the spatial profiling. What's important is that really about 60% or 70% of the triple-negative patients turn out to have an immune phenotype, whereas about a third of the HER2 negative hormone positive patients turn out to have an immune type. It doesn't matter, these are more basal-like. These patients here, you can see these are all the first 9 drugs that we looked at. This is pembrolizumab with about a 75%-80% response to therapy. These are turning out to be super important for us to understand, okay, how can we identify who's gonna respond? This is actually important because these immune drugs come with toxicity. They come with real problems like adrenal insufficiency, 7%-10%. If you try and combine immune agents in the early stage setting where people's immune systems are intact, you can double that chance of getting an immune response. I mean, an immune related serious event. You don't wanna be giving these drugs to the people that don't need them, or they're not gonna respond to them. You definitely wanna be able to be giving them to the people who do. Then you could probably spare them other therapy. That's just one example of how this works. It's also allowed us to sort of go from where we started, which was a 19% overall response rate, to where we believe that we should be today in about a 60% of all patients being able to achieve PCR. That second part of the trial was that we were able to show that PCR predicted distant recurrence-free survival regardless of subtype or therapy, that the degree of response is important, and that there are many agents that can improve subtype-specific PCR, and it works better when you use these molecular markers for treatment assignment. That has given us the impetus to be able to say, "Well, if we really understand the biology, let's start using some of these targeted therapies, including immunotherapy, in the patients that we know are gonna be most likely to benefit, and start with them first. Then if they get to a complete response, they can go to the OR, and if not, we give them the rescue with the best of I-SPY 2 for their rescue. This gives us a chance to really test these special targeted agents, and then only if they don't get a complete response to the second treatment would they go on to AC. One of the things that we've been working on is Mike Campbell and our team, as well as Sandy Borowsky, have been working on really trying to compare the gene expression profiles of reverse phase arrays and multiplex immunofluorescence. This paper is in publication. Anyway, that's our whole opportunity here in thinking about it. Within the HER2 types, there's a group that's gonna respond well, and then the HER2 negative and the immune positive. In this group here, you know, it's possible that immune expression profiling may be complemented or replaced by a streamlined multiplex spatial profiling. You know, the simpler a tool we have to get to that prediction, the better. We are embarked on this exciting experiment. We have 28 main sites, so we have 12 more sites opening the next 4 months. We opened up at the, at the beginning of July, middle of July, we opened up I-SPY 2.2. We have already 100 patients enrolled on this study, and we're super excited about it. We continue our platform to make the process more efficient and effective, bringing all our stakeholders together, solving problems in real-time, and making sure that we have source data captured to reduce the friction from running trials so that we can make new and better and more personalized treatments available faster at a time when patients need the most. By better, we mean higher distant disease-free survival and less toxic, personalized by matching patients' biology, faster using early endpoints and continuous learning. We hope that what we learn here will be very easily applicable to many other cancers and disease types. Just to say that the whole I-SPY program is a huge effort, hundreds of people involved across the country from working groups, chairs and working groups and site principal investigators, program oversight, and the whole Quantum Leap Healthcare Collaborative site collaborative staff. To our many participating organizations and funders that have been part of this. Thanks so much for your attention. Thank you, Laura, for that presentation. Our next speaker is Dr. Scott Rodig. Dr. Rodig is an anatomic and molecular pathologist at the Brigham and Women's Hospital in Boston, Massachusetts, and director of the Tissue Biomarker Laboratory of the Center for Immuno-oncology at the Dana-Farber Cancer Institute. He's also an associate professor of pathology at Harvard Medical School. Dr. Rodig's laboratory focuses on the development and application of novel tissue-based biomarker assays to improve the diagnosis of cancer and to guide therapy selection. His lab utilizes a variety of techniques and assays, including multiplex immunofluorescence and protein spatial phenotyping, coupled with digital image analysis. Importantly, his group has discovered and characterized the disease mechanisms underlying Hodgkin lymphoma, diffuse large B-cell lymphoma, specialized forms of lung cancer, as well as melanoma that have resulted in new diagnostic tests to direct clinical trial enrollment and lead to personalized therapy. Dr. Rodig Thank you for the kind introduction and for the opportunity to present some of the work that we've been doing at my institution. What I'm gonna speak about today perhaps is a little bit different than what some of the other investigators have spoken to you about. I'm gonna talk about an effort that we have run over the past three years to bring multiplex immunofluorescence and quantitative imaging into a clinical assay, something that we call ImmunoProfile. This is a joint venture between the Dana-Farber Cancer Institute and Brigham and Women's Hospital's Department of Pathology. What are we doing? We are looking to quantify immune cell infiltrates into the tumor compartment using tumor biopsy samples that are generally taken for diagnostic purposes. We now know that host immunity regulates cancer initiation, progression, and survival, and especially response to different immune checkpoint blockades and other types of immunotherapies. The biopsy samples that we get in pathology departments do have a clinically relevant snapshot of an ongoing patient's antitumor immune response or lack of an immune response, and this can be exploited for therapeutic purposes. For those of you who are not familiar with what I'll be showing a number of different images. These are pathology images. Just to quickly explain, in this particular type of image, these actually little blue dots are the nuclei of cancer cells. The little brown dots are actually a special stain that we do for a particular type of immune cell known as a CD8 cytotoxic T- cell. On the left cancer over here on the left panel, we have what we call a cold tumor, meaning there's lots of tumor cells but no immune cells. On the far right, we have what we call a hot tumor, which is cancer cells infiltrated by immune cells that are, in part, supposed to recognize and eradicate the tumor but by various mechanisms are shut down and unable to do so. In the middle is what we call a warm tumor that has a modest amount of immune cell infiltrates. Now, that's one set of markers. There are other markers that are clinically relevant that are expressed by the tumor cells themselves. Here in brown is perhaps the best-known tumor marker. It's PD-L1, which is the ligand for PD-1, which is the target of immune checkpoint blockade, primarily, you know, KEYTRUDA and others. Tumors that express high levels of the ligand for PD-1 or PD-L1 are associated with better response to PD-1 blockade than those tumors that don't express the PD-L1 ligand. We now know through a lot of different clinical trials that just the immune cells themselves that are infiltrating tumors are predictive of immunotherapy, and that there are biomarkers on the tumor itself that is predictive of response to immunotherapy. Based on these data, there's been a real interest in how can we quantify these data and present it to patients. What we do know is that for some tumors, the response to immunotherapy is highly correlated with select biomarkers expressed within the tissue biopsy sample. What we don't know is which tumor types is this microenvironmental signature meaningful. We do know it for some cancer types, but certainly not all. We don't know whether different cancer types have different types of immune microenvironments and that whether or not there are additional immunological proteins still to be discovered that are relevant or targetable in that immune microenvironment. Importantly, how can we integrate these observations into a clinical score that we deliver to our clinicians and to the patients? Also, how do we integrate these cellular findings into the genetic findings that we also now take more routinely in characterizing patients' cancers? The biggest problem for pathologists is that although we know through a number of clinical trials that tissue biomarkers are predictive of clinical response, these are typically not quantified in any type of reliable or reproducible manner. Current biomarker testing is a qualitative test interpreted by individual pathologists and also generally a single marker at a time rather done as part of a multiplex package. One of the solutions to this conundrum, and to bring quantitative assessment to the immune microenvironment in tumors, is to use multiplex immunostaining with digital image capture and AI-assisted image analysis guided by pathologists, so we can get, quantitative and reproducible data back for each individual tumor to our clinicians and to the patients. For this assay, we brought on and validated the Akoya Phenoptics or Polaris platform. Our overall goal of this clinical enterprise, clinical research experiment was to provide a clinically validated quantitative assessment of the tumor immune microenvironment using tissue biopsy samples taken for diagnostic purposes and do it in a CLIA-certified laboratory. These are laboratories that are allowed to report out clinically actionable laboratory results. Ideally, we wish to do this for all patients that came to the Dana-Farber or Brigham who underwent OncoPanel or genomic testing. For a variety of operational reasons, in our cohort of 2,200 patients, only about half had genetic testing to go along with it. Our specific purposes was to do a cohort study to determine the immune landscape of cancer broadly, not just the cancer types that we know respond to immunotherapy. To develop and validate a companion test to genomic testing, one that captures an immune score. To establish a framework for clinical trials in immuno-oncology that's clinically actionable because it's done in a clinical laboratory. Provide the first quantitative assay for PD-L1 assessment and hopefully start moving away from the qualitative assays that are currently being used in the laboratory. These are very important criteria that distinguish what we're doing from the basic science discovery laboratories that use multiplex immunostaining and image analysis, or even translational projects, many of which are published in the New England Journal of Medicine or other places, where the assays are actually done in research laboratories without a clinical certification and are not clinically actionable going forward. These are important distinctions. We, this is a joint venture between the Dana-Farber, which funded quite a bit of the assay development, and the Brigham Pathology Department, which provided space, infrastructure, and pathologists. What are we talking about? When I first showed you sort of those blue and brown pictures, those are qualitative assays looking at a single marker, whether it's CD8 T-cells or PD-L1 on the tumor cells. Here, what we did is went in to develop a pretty simple but robust multiplex immunopanel where we can look at multiple biomarkers simultaneously in a tissue biopsy sample. For this first iteration of this assay, this clinical assay, we had a tumor marker, which I won't go through the details, but it can be specific to individual tumor types. Markers of specific T-cell subsets, CD8 T-cells, which are the activated T-cells that kill tumor cells, FoxP3 positive T-cells, which are suppressor immune cells, they help suppress an immune response. PD-1, which is the target of KEYTRUDA and other PD-1 inhibitors, PD-L1, the ligand that's expressed by tumor cells, and we know from qualitative testing is associated with clinical response to PD-1 blockade. I just show you a couple of examples here, not showing all the markers, but just some select markers. Cytokeratin to highlight our cancer here in pink, PD-L1, the ligand expressed by tumor and immune cells, and CD8, our T-cells that perform the execution function when they recognize and kill tumor cells. Here is the qualitative picture that we see when we do the multiplex assay. With AI-guided image analysis, we can now start using pathologists in a more useful way to quantify thousands of differenT-cells within the tumor microenvironment in a quantifiable and reproducible manner. This is for individual T-cells and for the ligands expressed on tumor and immune cells, in this case, in a case of bladder cancer. Because we did this in a CLIA-certified laboratory, we went through a very long and involved effort to validate this assay against clinically approved biomarkers. On the left are a whole series of clinical biomarkers, single markers in brown applied to tumors from our clinical, from our clinical laboratory, and on the right is the multiplex immunofluorescence platform with digital image analysis that we validated against the clinical assay. We had to do this. This is just some examples to show that the immunofluorescence-based assay is equivalent to what we get by the standard assay. Of course, with the standard assay, each marker is done individually and interpreted by a pathologist qualitatively. Whereas on the right, all the markers are done together, and with pathologist assistance, are quantified using computers. We went through a very involved process. We ended up with a 60-page, excuse me, validation document, indicating full CLIA and clinical validation of this assay. We had to install this assay into a workflow within our clinical pathology department. If there's perhaps one thing to take away from my talk today, it's that when you bring on a assay such as digital pathology, you're not just bringing on a single technology, you're bringing on a full workflow. This requires involvement from many stakeholders within the hospital. In our case, we wanted consent, a research consent from every patient that received this test, and we wanted the clinicians to request the test within the electronic medical record. This means we needed to actually have software that we could then ensure consent for every patient that we ran the test had been done and ensure that there was a requisition by the clinicians in the electronic medical record. Cut them in our histology lab, move them to the ImmunoProfile laboratory, where we actually had QC done to confirm tumor content, and that was adequate for testing. We tested the actual sample, then it went to a pathologist who reviewed all of the assays and the scoring. Went back to senior scientists and technicians to generate the quantitative results, and then the data had to be exported into an accessible data file that the clinicians and researchers could access. In our case, it's called Matchmaker. What I'll say is that this assay, the automated staining and then the automated staining and image capture, essentially never fails. We did approximately 2,200 cases over the course of three years. I think the failure rate is less than 5%. It essentially always works. This is part of a larger research enterprise. It's part of our effort that really bridges between sort of advanced diagnostics, which traditionally in our department and the Dana-Farber has been around genetics, called Profile, and pure research, which we do in non-clinical laboratories and just basic research laboratories, in our case, the Center for Immuno-Oncology. We border between both in the hopes of transitioning from the research lab into the clinical lab. Finally, I'd just like to say this is a team effort. Really enormous number of people have worked on this project, and we're still analyzing quite a bit of data. Perhaps for me, one of the biggest and most unexpected aspects to this project was the significant amount of computer software support we have required in order to integrate all aspects of this project into a clinical laboratory. Thank you, and I'm happy to answer questions. Thank you, Scott, for that presentation. Our next speaker is Dr. Manuel Salto-Tellez. Dr. Salto-Tellez is Chair of Molecular Pathology at Queen's University, Clinical Consultant Pathologist at the Belfast Health and Social Care Trust, and Deputy Director of the Centre for Cancer Research and Cell Biology in Belfast, Ireland. Professor Salto-Tellez is a recognized international leader in molecular and digital pathology, with a research focus on the relationship of phenotype to genotype, with a specific interest in the pathway of cancer biomarker development from discovery all the way through to clinical application and adoption. His advocacy and work on digital pathology and morphologic molecular diagnostics across the UK, Europe, and the US has helped shaped the landscape of modern pathology today. Professor Salto-Tellez was instrumental in establishing PMC at Queen's University Belfast, which is a new clinical laboratory bringing together high-throughput genomics, digital pathology, and big data analytics in a fully integrated and automated fashion. He aims to accelerate the translation of potentially relevant diagnostic, prognostic, and therapeutic findings into clinically actionable information by applying state-of-the-art technology in a clinical laboratory environment. Dr. Salto-Tellez. Thank you so much for the kind introduction. My task today is to talk to you about MIF and a potential roadmap to clinical implementation. The way I'm going to do this is essentially trying to ask a very fundamental question. How suitable is multiplex immunofluorescence for diagnostic purposes today? I would like to answer that question by explaining some of our experience in using MIF in colorectal cancer and using MIF in lung cancer, and then ending with some comparative analysis of how the experience that we have accumulated in bringing genomics into the clinical practice can help in the MIF paradigm as well. Let me start with colorectal cancer and let me start with the reason we started really embarking on MIF. It was by seeing results like the one I'm showing you here from our group generated by Stephanie Craig and Mark Humphries. This was an analysis of more than 18 individual proteins in a chromogenic way, analyzing, as you can see, more than 1,500 colorectal cancers from 3 very important studies, study cohorts within the U.K. What we saw is that from those almost 20 biomarkers, there were 3 that had a very strong clinical significance, CD3, CD4, and CD8, as you can see, even more than the traditional CD3, CD8 approach. What we saw as well is that the application of this prospectively could have a significant therapeutic advantage for our patients. That tool that was generated with traditional digital pathology was then revisited with an artificial intelligence tool as the work here again in our laboratory by Jasmine Makalou, where as you can see, there is a significant improvement in bringing artificial intelligence tools to the direct application of digital pathology and immuno-oncology. So much so that in our setting now we have something called TruT, which is a collection of all the information that we have in colorectal cancer, on hundreds of cases over the years. Whereas you can see by analyzing and by including 5 different variables, including the TruT status as indicated by CD3, CD4, CD8, we can immediately calculate what would be the potential prognosis of a patient in our own paradigm with our own tools. Obviously, if you believe the promise that multiplex immunofluorescence is going to bring a significant degree of complexity to the way we practice pathology, then in our minds, the best way to deliver a tool like TruT is to make it mIF TruT. The reason of this is what really made us think that we really had to develop tools to validate multiplex immunofluorescence with the standards of diagnostics that are relevant in our routine practice. To be able to bring all the annotations, all the gold standards that we have generated with chromogenic information, and bringing them into multiplex immunofluorescence to make our analysis more meaningful and substantially richer. To do that, we had to set up the tools to make that happen. This is what this study by Anali Pitharam and Stephanie Craig in our lab showed. As you can see, we started the pathway of how we should start validating multiplex immunofluorescence with the level of rigor, with the level of reproducibility, with the level of accuracy that is important to be delivered for patient care. What we end up was with a very specific roadmap on how to do this so that we would get the most of the quantitative information that this process will generate in this particular case in colorectal cancer. We saw the importance of establishing chromogenic singleplexes for establish our own gold standards to identify what would be the appropriate analytical thresholds. We understood the importance of performing singleplexes to determine the perfect Opal antibody pairings and obviously all the other considerations, analytical and pre-analytical, that make a tissue analysis meaningful and robust. Needless to say, at the same time, we understood some of the challenges of bringing this technology into routine. One of them, as many of you know, is the potential batch variation that we identify in some of these analysis, which may be Opal-specific or maybe biomarker-specific, and many other things that we know could go wrong, like in any other tissue hybridization test, by the way, some of them related to the quality of the tissue, some with the process, the wet lab process of a multiplex immunofluorescence, and sometimes, as you can see, with the treatment of the digital images, in this respect. If I were to try to answer the question today of, is MIF a diagnostic tool today, I would say yes, it's ready, but perhaps trying to bring the following improvements. A higher degree of automation in the pre-quantitative step, perhaps automation in the detection of those artifacts, which is something that is going on in my laboratories, and also automation in batch variation co-correction. You can see very close to having a technology that could be rolled into diagnostics, in a, in a meaningful manner. We learned a lot from the lung cancer paradigm as well. It was very interesting from the point of view of pathology to note that genomic-targeted therapy was definitely buying our patients longer lives and better lives, but for the vast majority, these patients will come back with disease that was more difficult to treat. While immune checkpoint therapy, for some cancers and for some patients within those cancers, was able to deliver long-term sustainable survivals like we've never seen before in advanced cancer. It was very interesting for the pathology community to realize that the leading biomarker that started this venture was not a complex genomic testing in any way. It was something as simple as PD-L1 in an analysis in lung adenocarcinoma. We know, though, that PD-L1 scoring is intrinsically difficult. We know that there is a logic to it because we know by the work of David Rimm and others that MIF could potentially be the best predictor of therapeutic response to checkpoint blockade. We know from our own work as well that PD-L1 MIF has extremely good correlations with the chromogenical standards. In fact, when we start putting together chromogenic and immunofluorescence analysis, we end up with pathways, diagnostic pathways that are able to analyze by digital means, the vast majority of the patients that we get in our clinics. This is essentially because of the extreme accuracy of this technology and the capacity of looking at co-localizations, co-analysis, and diversify the expression in a way that, as you can see on the right-hand side, chromogenic immunohistochemistry has difficulties with sometimes. We also know from the lung cancer paradigm that it is good to consider lung cancer from a broader point of view, and this from a very practical point of view, has to do sometimes, not infrequently, with the availability of the sample that, as you know, sometimes is minuscule. Being very small, the questions, the interrogations that we take that sample through are absolutely phenomenal, not only for therapeutic intervention, but even for routine diagnostics to establish a diagnosis. Obviously, the promise of multiplex immunofluorescence running multiple biomarkers in one section is extremely relevant in a scenario like this. However, if we take that into account, the question that we need to ask ourselves is really the way of building MIF clinical panels. Should we build one for colorectal, one for lung, one for melanoma, one from ovary, one for prostate, one for endocrines? Is that the way? What the genomics experience is telling us is something slightly different. Remember, genomics tests vary broadly. Even if we know that a small percentage of the genomic information is going to be applied to a small percentage of the patients in a way that is going to be clinically and therapeutically meaningful. We are used to apply large panels, and those large panels are applicable to many of the different cancer types. It's very interesting that while genomics tests broadly, we are still thinking of small panels that are very much cancer-specific. Indeed, this may be the first way of applying multiplex immunofluorescence routinely, but the question comes as to whether we could find a match for the genomic broad testing in the tissue hybridization testing that we're beginning to apply. As you know, there are technologies already that are giving us the promise of being able to analyze 50, 60, perhaps 100 biomarkers in a way that parallels in some way the genomic information that we are generating in some patients. This is important, and I would like to finish with this reflection. We've been saying for a couple of years that after the first revolution in pathology, immunohistochemistry, and the second revolution in pathology, molecular testing, we are experiencing a third revolution in pathology, the one that comes with digital pathology and artificial intelligence. The question that we are asked, not infrequently, is what comes next? I'm sure after what I have told you may agree with me that perhaps the next revolution in pathology is going to be dictated on how we integrate complexity in our analysis. Only then we are going to be able to bring a totally new generation of biomarkers for a totally new generation of drugs that will be extremely beneficial for our patients. With that idea, with the idea that perhaps there is a fourth revolution coming very soon, I would like to finish this talk. Thank you all very much for your attention. Okay, wonderful. Thank you, Dr. Salto-Tellez, for that presentation. I'd like to now invite back all of our speakers for a round table discussion. Okay, my first question is to Dr. Rodig. Dr. Rodig, the degree and pace of adoption of multi-marker protein spatial phenotyping by the medical community is going to depend on the diverse views of a large set of stakeholders. You've developed a very broad awareness of the value of this approach in the pathologist community for sure. Can you tell us what other types of medical or even non-medical professionals must be similarly influenced to realize this vision? Maybe just a word or two on what's important to each of them. Sure. You know, bringing on new testing modalities within pathology, of course, requires adoption by the pathology community, first and foremost. As you've alluded to, there are other stakeholders in this and for widespread adoption and then routine use, it requires really an alignment of benefits for all of them. I guess I would list a number of people. First and foremost, our oncology colleagues and oncologists. For them, you know, the meaning of new testing is really to allow for some type of clinical actionability. If a test comes on that's new, that's necessary and required for some type of change in clinical trial, I mean, in clinical practice, usually arising out of a clinical trial scenario, that of course is an impetus for a pathology department to work out the added costs of bringing on new testing. Yeah, I would say, you know, one of the things that matters, you know, you have to have real value. I think increasingly, it's going to be important to understand how to target immune directed therapy. The cell, you know, the immune checkpoint inhibitors are great, but the immune checkpoint inhibitors come with their side effects. They come with a considerable burden of adrenal insufficiency, thyroiditis. I mean, these, especially as they move to the earlier stage setting, you're gonna find, you know, that, you know, 7%-9%, especially the combinations of these, are gonna really increase the burden of permanent side effects of adrenal insufficiency and thyroiditis, you know, rates 10%-20%. These can be treated, if you're not going to respond, why you still get the side effects. You don't want the side effects without the benefits. Increasingly, as we can start to target and we can say, "Oh, well, you know, the spatial proximity of, you know, PD-1 T-cells to any, you know, or to any, you know, PD-1 positive cell is important." If this really turns out to be an important predictive biomarker, that is going to make a big impact. While there are gonna be genomic predictors, there's lots of predictors, I think it depends on the ease of use, the ability for places to do it. We need to simplify it down to something that can work. If it's really very predictive, I think that's gotta be key and that's gonna make a big difference. Great point. I'll just have to want to use it too. I mean, it's the same way we've learned to use other genomic predictors. Seems to be lots of lessons from the NGS journey and any other diagnostic or platform modality. Definitely. Well, thank you. Dr. Esserman, Just to follow up on that too, I mean, I completely agree. I would just argue that it has to be very clear, you know, where it's going to change therapy. If it's a predictive. Yes. -marker, and the clinician has an option of perhaps not using immunotherapy because it's very likely to cause severe toxicity and not be effective, and they have an alternative, there, you know, it makes sense. There's certain, you know, situations where maybe there are no alternatives. Right. It won't matter. Mm-hmm. It won't matter, they won't order it, and it won't get reimbursed. It will very much depend on the specifics of what the test is used for in terms of decision trees. Wonderful. Thank you. I'd like to shift gears a little bit, continuing with the theme of clinical adoption of protein spatial phenotyping using multiplex immunofluorescence and turn to Dr. Salto-Tellez. We've had a very IO-rich discussion. I'd like to invite Dr. Salto-Tellez in his answer, if you'd like to comment on other types of applications clinically for this approach or even diagnostic pathology. Dr. Salto-Tellez, with the growth of AI and digital data in pathology, adoption and acceptance of electronic review and sign-out of cases is essential if pathologists are to evolve from traditional glass slide review of cases with a microscope, right? There's absolutely a digital pathology element baked into this approach. My question for you and the other panelists as well is what lessons can we learn from the digital pathology journey that would encourage more broad adoption? Also, how do you see spatial biology in everyday diagnosis as a driver of this adoption, if at all? Yeah. Very important questions, Gavin. I would say, I'm answering your first question, that the very clear message to the pathology community has to be that those pathology departments that have adopted digital pathology do not look back. What we are seeing in pathology departments that have adopted digital pathology as their modus operandi for diagnostics is that 90% of the pathologists, now certain faculty adopt digital reporting for the majority of their cases, somewhere in the order of 90%-95% of those. We know that the turnaround time is similar for biopsies and is significantly better for resection specimens. We know that the added value that it brings to the, to the routine pathology in terms of double reporting, ease of consultation, preparation of MDTs, training of residents is phenomenal. Although the adoption of digital pathology has probably been slower than expected in the pathology community, it's very clear that those departments that have followed this path do not look back. You know, it's so interesting. I don't know if you agree with this, but I think COVID has been a huge boon for digital pathology. I mean, even at UCSF, we had minimal adoption, and we're like now, like 90%, 100% is all on digital slides. It's fantastic. In tumor boards, you know, we have the pathologist there. You know, before you had to prepare, and it had to be this big rigmarole if you're gonna review things. Now you can review things in a minute. If I wanna see something and I'm looking for a margin, I can call a pathologist, and I can Zoom and see it in 2 minutes. It facilitates discussion. It makes more people look at slides, look at what the pathology is. All of a sudden, you've got this incredibly rich ability to ask lots of different questions. You know, in I-SPY. We have everything digitized now and, you know, everything is uploaded. Our ability to do that changed dramatically with COVID because when people couldn't come in, it just. I mean, everyone's workflow changed, and I thought it's just remarkable. I think that will. I mean, this brings a whole revolution forward, right? Whenever you digitize, right? The interesting question from Gavin, obviously, and I fully agree with that, Laura, was how the, you know, improved by the general adoption of digital pathology. What I would say, I'm probably paraphrasing what Scott and Laura mentioned earlier on, is that there are probably two main determinants of that. One is how applicable is the technology in diagnostics? Sure. How we make it as seamless as possible so that we can operate according to CAP standards, ISO standards, et cetera. Mm-hmm. The second thing is what is the clinical applicability of the biomarkers that we have decided to deliver in a multiplex fashion? Until that is not clear, and until there is a clarity that those tools are delivering something that pathologies are not delivering by more conventional methods, I think that the whole adoption is going to be, it's going to be tricky. I completely agree. I mean, I think from my perspective, I mean, I think we're gonna have genomics as well as, you know, image analysis for every patient. I think, you know, they're gonna provide complementary information. One set of information about the tumor itself, one set of information about its microenvironment, and of course, one can influence the other. I think they provide complementary information, so those will eventually, I think, work into routine practice. Yeah. I think that there. I think even if you can get the analysis down to two or three you know, multiplex, you know, summary slide, which I think is fairly straightforward to do, you have the analytic tools. The throughput and the analytic tools have got to be much faster and, you know, much, which they will. I mean, the technology will evolve, but that has to be a pretty easy analysis. I think you have to sort of look at what am I gonna gain from a genomic analysis and what am I gonna get from an immunofluorescence? Do I need both of them or are they teaching me different things? Sometimes they will, and sometimes they won't. I think the other real opportunity also is to think about, oh, well, if it works. I mean, I think that there are immune phenotypes that are common to tumors, and they're independent of organ of origin. I think that's gonna help us run more organ-agnostic trials and continue to refine how we target these immune environments. Mm-hmm. To get benefit, which I think is very exciting. I think, you know, for sure these tools are gonna have a role as a catalyst in making change. The question is, at the end of the day, is there enough value or is there enough, that's added as a routine or does it replace something else that's also expensive? Instead of a genomic test, we could say, "Well, let me just run the immune test. If that works and that's dominant, then that's what we would do." I can see that that could make a big difference, in choice of therapies. There's always that, you know, tension between, you know, I know that the, you know, the pharma companies also, you know, want their drugs given to more people and for longer. It's not clear to me that that's what's needed. What patients want is effective things that can be done less. You know? Yeah. Great point. We want precision. We want less toxicity and, you know, this ability, you know, people are putting up with the toxicity now because they're so grateful to have solutions. There's gonna be, I mean, there's all kinds of cool things coming down the pike, you know, where, you know, you know, the anti-CD47 drugs, the, you know, antibody drug conjugates, all of these things that we are gonna start figuring out how to use them. I think immune multiplex have played an important role in figuring this out. The question is, can we come up with something simple that, as a, you know, a simple enough construct to sort of make that be, as you said, Scott, the, you know, one of the key things that we do on every patient to, you know, as we're making choices. Great points, Dr. Esserman. Brought us back full circle, in a sense, just acknowledging through Dr. Rodig's points as well, the importance of an ecosystem of things. That patients and pharma companies and pathologists and even hospital administrators, you know, all have different things that they're interested in. The pathologist needs to be empowered, and the patient needs to be provided the best possible care, for example. So, you know, a tipping point, pathologist empowerment, lots of things, clinical utility, lots of things very important in driving forward a future scenario where like genomics or NGS even, this approach, even though it's fundamentally familiar, to clinicians, is used in a multiplex digital pathology fashion to drive patient care forward. I'd like to end the roundtable exactly there with a bit of a look towards the future. We'll just go maybe in order, starting with Dr. Rodig. I'd like to ask all of the participants to look in their crystal balls, if you will, and predict in whichever way you'd like, what the future may look like for clinical spatial phenotyping, maybe over the next two, three, four, five years, something, the near-term future. What indications will benefit the most? Is there anything beyond IO, beyond cancer? What might lead to a tipping point to broader adoption? Whatever you think is important, I'll just leave it open to the panelists. Maybe we'll do Dr. Rodig, Dr. Esserman, and then we'll end with Dr. Salto-Tellez. Yeah, actually, I mean, I'm pretty optimistic. I mean, you know, our institute is now undergoing digitization of pathology and incredible interest in AI and machine learning approaches just to H&Es. It's of course gonna get expanded into multiplex analysis as well. It is reaching a tipping point, I do believe, and it has been slow in coming, but digitization is very actively occurring in almost all major academic, you know, pathology departments in the United States anyway. With that, I think, with the advent of AI, machine learning, and these multiplex methods, I really do believe in the next three to five years, we will get a pretty comprehensive look at not only the immune microenvironment, but any other type of, you know, biomarker that we want to interrogate, you know, with spatial resolution in tissue. There's many even beyond the immune system that we already look at by crude immunohistochemical methods that will get transferred into these new technologies. Yeah, I fully agree with you, Dr. Rodig. I think that in the field of cancer, I think it's very clear that, you know, to us, that, you know, it didn't work out that the immune or even though people still treat people and say, "Oh, if you have triple-negative breast cancer, you should get an immune therapy." It turns out that 60-70% of these high-risk tumors have an immune phenotype, and the others don't. The, you know, these are expensive and come with toxicities. Don't give drugs that don't work. In the same way, the hormone positive, about a third of them are actually more basal phenotype and have immune infiltrates that you can manage. I think as we go forward, it's gonna be clear, and we're gonna start to find that same pattern across tumors regardless of organ of origin. I think that's very exciting because it's gonna really open up our understanding of how to appropriately target these tumors and give people, you know, much better options and with less toxicity. I'm really looking forward to it. I'm really looking forward to even figuring out how to target the few people we earlier in the course of their care. Maybe even stage 1, you can figure out who needs what. It will also direct the ability to do intratumoral injections and change that microenvironment. I think all of that is super exciting. I think I wouldn't be surprised if you could see it in other diseases where that are using. There's a lot now, you know, in rheumatoid conditions and, you know, inflammatory bowel disease, et cetera. In places where you're actually getting tissue for diagnosis, you're understanding it. In the inflammatory bowel diseases, I wouldn't be surprised if you start to see these same kinds of assays and tools to help you better understand, you know, which drugs should be used and in what sequence. That's really, I think, the key. Having more platform trials, more ability to test things in sequence or different orders, trying to understand it, trying to sort out all that complexity, that's where I see these tools really helping us really, you know, pave a better path for patients. Thank you. Dr. Salto-Tellez, what do you think? Thank you, Gavin. I've tried to predict the future in the past, and my degree of success has been variable, let's put it that way. If I had to, you know, to take the challenge, I would say that in the near future, we will need to embrace tissue hybridization complexity like we have embraced genomic complexity in diagnostics. Yes. It is something that will need to happen because as Laura has been reminding us, there is increasing evidence that that is going to be clinically very, very relevant. Also taking the lesson of genomics, I think that we are going to have to test very broadly, like genomic is testing as well. Remember, in genomics, we are applying single applications with 3, 400 gene targets to identify 2 or 3 that may be applicable to a fraction of the patients that are having genomic analysis today. That is not wasted. That is the level of testing that we need to put in place to deliver personalized medicine. I think we need to get used to the fact that with tissue hybridization complexity, we are going to have to do the same, that our panel will be broader, our digital pathology analysis will be more and more accurate, and at the end of the day, it will be a fraction of those large panels that will be applicable to our patients today and tomorrow. Embracing- Wonderful. Yeah. Complexity embracing broad testing, I think, is going to transform very much the way we practice. Wonderful. Thank you. With that, I would like to thank all of the panelists for their time, for their presentations, and for their incredible, thoughtful discussion here in the roundtable, sharing your views on the current status and, the future potential of clinical protein spatial phenotyping. Thank you. Thank you. Thank you. Thank you so much. Okay, that's the end of the first part of the program. I want to first thank all of our guest speakers for the presentations for the roundtable. I want to thank all of you for your time. I want to thank my colleagues at Akoya for the hard work in putting this together. Now we're going to transition to the Q&A. I invite you to stick around. You can submit your questions online, and we'll do our best to answer those. With that, we're going to transition right to the Q&A. Thank you. All right. We'll now start taking questions on the phone. Our first question from Kyle from Canaccord. Kyle, are you online? Yeah. Hey, guys. Can you hear me? Yeah. Hey, Kyle. Perfect. Hey, congrats on the day. That was really good color, good overview of all this stuff. I guess I just want to start on the software side. That was some great charts there, talking about the different markets and how you were using these different vendors and all that. Just sounds like a lot of moving pieces, though. Any thoughts on, like, the interest or kind of benefit in kind of offering, like, one proprietary software platform that kind of offers the best of all these worlds? Like, kind of jumping on that, I'm just wondering, you know, when you think about it, like, who kinda like pays for all these different software vendors, whether it's QuPath or Enable Medicine, OracleBio? Like, would it help to have just one kinda platform too? I'll let Niro. I apologize. I got a little frog in throat. I'll let Niro speak to this specific question, Kyle. I just think thematically, maybe just to take some liberties with your question. I think what we're able to do now thematically, Kyle, given that spatial as a market is really beginning to explode, equally, if not more important, because we have such a large install base, there's really a great opportunity, and I mean this in a positive way, to leverage partnerships as catalysts, because we're sort of ready-made to do that. Bio-Techne is an example. Our instruments are ready-made to partner. Expanding our panels and having such a large install base provides incentive for antibody partnerships. Our software, because we come out with a standardized format that many of these large vendors like Indica and Visiopharm have been using for years, it's really easy then for to extend that into the HyFlex space. I think as an organization now, and hopefully the theme came through, we're really now at a point where our ability to leverage our scale as a growth driver is a real differentiator and something that is a focus of our time in 2023. On the software question specifically, I'll hand it to Niro. Yeah. Kyle, you know, when we think about creating a single sort of software solution, then in that design, we have to figure out which customers we serve and which ones we won't, right? Because of the inherent choices we have to make. A great example would be if we decided to build a single solution, and then we have to decide if it's a desktop or a cloud. If we went with cloud, there are a number of customers from their institution, they can't have cloud access. They have to have on-prem. That would eliminate those customers. There are some customers who are price sensitive who rather use a free solution as opposed to a service, if you will. I think the challenge is you could create one, but it may not serve all your customers. Our customers are pretty diverse from academic to biopharma to doing discovery to clinical research. For that, it's better to offer them the best kind of tool that serves their specific needs, if you will. Kyle, this is not a unique phenomenon to spatial. If you look back historically, when NGS really started to explode, the same dynamic occurred where you had this ecosystem of freeware and commercial providers that were building purpose-built software solutions to serve these different market segments, both discovery and translational, ultimately clinical, and some free and some fee. It's sort of a well-worn path of life sciences tools and DX software development going back to NGS, microarrays and other areas like flow. Yeah, that's a good point. There might be, like, a misconception that something like BaseSpace is like what I'm getting at, but that's really more of a collaboration platform. This is, like, much more intricate, sort of, and it does matter by market, like Niro kind of was mentioning. Yeah. When, yeah, just another question here. You know, interesting slide where you're moving to like a diagnostic or like a medical company that's pretty cool, I guess. Life science tools, though, is pretty relatively safe. It's a simple business model. Diagnostics, much more complicated. How do you think about reimbursement, FDA approvals, marketing, publications, integration with EHR eventually, things like that when it comes to diagnostics? What can you do in the next year or near term, I guess, to kinda get ready for that? Yeah. Getting ready for it, really, you know, the tip of that spear is really what Gavin spoke about with our Advanced Biopharma Solutions groups. The services that they're providing are not just about frankly driving business and partnerships and broader clinical trial participation from some of the panelists you heard from, but it also is really the front on which we're tackling many of these things to become a medical company. The partnership with Acrivon is really the mechanism by which we're tackling all of these things, like doing FDA submissions, understanding the regulatory environment, preparing for the commercial side of that launch, and forming these companion diagnostic partnerships where you're exerting those expenses as part of a partnership that's helping fund that, but also it's helping de-risk the actual clinical adoption. What you didn't hear us talk about is doing what you've seen, again, in some of these larger NGS companies, which is self-funding large clinical studies for diagnostic applications. The tip of our spear, really, and the mechanism by which we'll become a clinical company will be through partnerships with biopharma, forging companion diagnostic agreements more and more over time, building out that menu. That's how we're trying to both de-risk it and maximize the probability of success and focus on one relevant clinical assay at a time that's tied to a specific therapeutic. Okay. That was helpful. If I could just squeeze one more in on the, on the products. I guess first, when is the, like, simultaneous protein and RNA imaging gonna take place? Can you, like, roll that out, you know, midyear, maybe something like that, in 2023? Secondly, could you just confirm if there's no RNA markers for the PhenoCode Signature Panels? It looked like it was just protein. Yeah. Kyle, on the first one, right now the plan is to have the RNA-only panel launched by the end of the year, and customers will be able to run RNA and protein on 2 sequential sections. We haven't rolled out the true multiomics applications yet, and we'll let you know as the year goes on when the launch for that will be. On the PhenoCode Signature Panels, in the translational space, more in the discovery, the demand has been strong for protein, and so that's what we're largely enabling at this point. It's worth pointing out the RNAscope technology, as it is, their full plex capability is already compatible with our imagers. If people wanted to do RNA, RNAscope might be the preferred way for them to do that. At this point, we're not launching RNA panels yet. Okay. Perfect. Thanks so much, guys. Congrats on the day. Thanks, Kyle. All right. Our next question is from Mason from Stephens. Hey, guys. Thanks for hosting this virtual day. A lot of helpful information in there. First, for your PhenoCode Signature Panels, thinking about the future, how are you thinking about adding additional flexibility to these panels? Could there be an option to scale, maybe down the predetermined markers or add additional unique markers to these panels or maybe the second generation of these panels down the road? Mason, the way the panels are designed right now is that they can use the Fiplex and be able to add one target of choice. If they wanted to design their own panel, we have a labeling kit that allows them to buy antibodies off the shelf and then label them with our unique barcodes and build their assay. We also have our historic Opal chemistry as well that's designed for them to build assays as they see fit. The flexibility is absolutely there for them to build their own panels or use what we do off the shelf. I think the important part about this is that, you know, for our PhenoCycler platform where we have a large menu of barcoded antibodies. The idea is that as we continue to expand our content for our discovery assays, a lot of that content could then be downstreamed used by our customers to create their own signature panels. Got it. That makes sense. Maybe on one more here on the clinical setting. You guys hit on a lot of pain points or hurdles that, you know, have been addressed or still need to be addressed for broader adoption there. One question I had was whether it's clinical research or routine clinical testing down the line, could access to adequate tumor samples be an issue? You know, there's a lot of genetic testing going on for patients, and some companies that are offering that genetic testing often talk about inadequate tissue samples has been a hurdle. Could that be a hurdle for you guys down the line or just multiplex tissue testing in general? It's actually a really good question. I'd actually flip it on its head, that it actually represents an opportunity for us because one of the things that you hear particularly in fields like non-small cell lung, where you're dealing with very small biopsies, and I think as Gavin alluded to it, there's a need for multiple immunohistochemistry slides already. What happens when you embrace multiplexing is you're actually reducing the tissue demand by doing it all on a single slide, removing some of the challenges where you run out of tissue, and you'll hear this from folks, you run out of enough tissue, or to do some of the molecular. Multiplexing actually presents an opportunity to solve for some of the tissue scarcity that you've heard about. Got it. That makes a lot of sense. Thanks, guys. Appreciate it. Thanks, Mason. Our next question is from Cameron from Deciphex. Great. Can you guys hear me okay? Yes. Perfect. Awesome. Thanks so much for putting on this great event. Just have a quick question, in terms of the business model considerations in terms of clinical expansions. You know, we're really excited to see spatial entering, you know, it's a little bit later growth stage. Do you expect, you know, as we see greater adoption of spatial in the companion diagnostics setting, that you'll be able to continue to run tests out of, you know, your central lab, and in collaboration with reference lab partners, or is there any expectation of decentralization down along the line? You know, are we too far out to even have that discussion yet? It's a good question. Maybe. Excuse me. I think what's important, and I actually think you guys nailed this in your report, what you heard from the last series of panelists was a belief, a hope, I'd say, maybe an expectation that this will have clinical relevance. On the way to getting there, what you see is a really expansive and meaningful growth in the translational market as we get all of these learnings. To your specific question, I think ultimately our objective and our aim is for a distributed test. That's why our focus is on the workflow. That's why, you know, I think our aim here is through companion diagnostic partnerships, is ultimately to have a clinical deployed, distributed test in a kit manner. One of the reasons why we continued to make these incremental investments in the PhenoImager HT system is so that that system does have broad deployability, not just for, and as Gavin alluded to, not just for the higher immunofluorescence-based testing, but its capability to also do standard chromogenic does give it a lot of applicability for larger and larger menu creation. Hopefully, that answered your question. Yeah. That's great. Thank you. Thanks, Cameron. Our next question is from Sunil from Enable Medicine. Sunil, can you hear us? Yes. Am I coming through? Perfect. Awesome. Well, first, thank you for a great speaker series. you know, these data sets are fantastic in that they're biologically rich and dense, but they're also large and complex from a data management and analysis perspective. It's great to see the partnerships that Akoya has highlighted, from a software perspective. I'm curious, how does Akoya think about orchestrating this ecosystem that best empowers a scientist to ask and answer questions from their data? How should Akoya customers know how to piece together the solutions from your ecosystem? Yeah. Sunil, I think the first thing I'll acknowledge is I think you've captured the data analysis problem perfectly, right? There are two buckets. One is data is too big, and then if I can get through that, how do I actually analyze the data set? I think the first thing I'm excited by is the fact that we were able to solve the data size problems with our QPTIFF format that opens up sort of this ecosystem of partnerships that we've created. I think the second problem is yet to be solved, right? I think, you know, you guys are doing a great job at Enable and so are the others in terms of figuring out what's the best way to analyze spatial data. I think the ingredient that's required to answer that question is data, right? The more spatial data that you have, that you can learn about how to analyze and train, I think the analysis tools are gonna get easier and faster. I think when I think about Akoya, particularly, the reason this ecosystem is powerful is that right now we have the largest install base that generates hundreds of thousands of samples worth of data every year. What that means is, you know, working with partners like yourself, we can get that data analysis workflow to be better and better and better as the time goes on. I feel like with this partnership and with the partners that we have and the data, the scale of data we generate, it's just a matter of time before the spatial data analysis workflow gets streamlined for our customers. Again, it becomes easier and easier. Great. Thank you. Thanks, Sunil. Our next question is from Julia from JP Morgan. Can you hear me, Julia? Hi. Can you hear me? Yes. Hi, Julia. Okay, great. I have a question on the proprietary RNA menu. We know there's a wide range of offerings on the market in terms of plexity. Where do you see the industry demand fall in terms of plexity? Maybe how does that, you know, look differently for discovery versus translational markets? Can you talk about the long-term roadmap that you're thinking of for the RNA menu? I think right now, Julia, where we look at it is, I think, from the discovery area, there's two places. I think there's places where you can think about, can I do a, you know, a high plex RNA in the 500+ RNA or do I do in the 100-plex RNA? I think when I think about the RNA for discovery, I particularly think about the application area. Currently, our first RNA panel is gonna be in the immuno-oncology cancer area where. There's a lot of great antibodies, and protein is probably a preferred analyte there. For us, in terms of solving the discovery application needs in IO, we're largely gonna focus on RNA and protein at the 100-plex range and integrating that as best as possible so that it works off the same section. I think that's probably gonna be the most powerful discovery platform because in terms of a data, every protein is probably worth 10 RNA, right? I think once you start thinking about going into other areas like infectious disease or something like that, where there aren't great antibodies, then you may have to run larger plex RNA assays. Right now, I think we're really focused on rolling out RNA capabilities in the IO space, where I think 100 plex goes a long way along with our protein capabilities. Got it. That makes a lot of sense. A follow-up on the software, questions. You know, we know research customers have a wide range of needs, you know, standardization is also important for translational clinical applications. How do you balance the need for flexibility and standardization? In light of the partnership model, you know, how do you plan to make sure to stay differentiated in terms of your software capabilities versus competition? Yeah. There's layers to that question. One is, I think the flexibility piece is an easy one, right? There's lots of software tools out there and scripts that are always being innovated on and published that customers can access and use. Once you start getting to the clinic, it depends on, you know, where in the clinical application you are. For example, you know, our partners like Visiopharm and others do have clinical workflows that are much tighter and fit for purpose, and they can be optimized and validated and locked. We can work with partners like that to do customer-specific workflows, if you will, that have the kind of desired constraints one wants, right? In terms of innovation, I think the competitive edge that we have is that because of the for software partners, what they value the most is, do you have a large enough install base for them to be meaningful to work with you and serve your customers? I think because of that install base and the productivity and install base, the competitive edge we have is some of our clients will some of our partners will develop custom workflows that support, you know, Akoya samples and Akoya applications. I think that's where we'll be competitive. I would just triple down, Julian, say that if you just look at the partnerships, the organizations that Niro mentioned, on their own, they're incredibly powerful, inventive groups. The competitive edge we have in software, is the competitive edge that those organizations have on their own right. Indica, Visiopharm, large install bases, professional groups, incredible development staffs. Enable Medicine just storming onto the market with a really powerful solution and deep expertise in PhenoCycler-Fusion. OracleBio being able to take all of these tools and answer a question for you. PathAI really leading the charge in leveraging datasets like ours to extract signatures and meaning. Our greatest competitive edge is that a market is created, and it's being capitalized on by instant standalone institutions that are experts in these relative fields. Great. Excellent. Thanks for hosting a great event. Thanks, Julian. Thanks, Julian. All right, our next question is from Tejas, Morgan Stanley. Hey, guys. Good afternoon. Just a couple of points of clarification here, actually, Brian and Niro. On that 100-plex multiomics, panel rollout, is early access still happening by your end, or is that now sort of at some point next year? Yeah. I think, just to clarify, Tejas, the way we're thinking about the first generation of multiomics, it'll be RNA and protein on sequential sections, but the data can be integrated. We haven't started the early access for the RNA yet. We're hoping to do that sometime in, you know, early part of 2023, but we haven't fully formalized that yet. Got it. Niro, as you think about sort of the mix of, you know, proteins versus, RNA in that 100-plex panel, how flexible is it? Is it initially like, should we think about it more as 5-10 proteins and scaling up over time? Or could sort of users, even in early access, go to as many as, you know, 100 proteins if they wanted to? I mean, the goal is, in terms of the multiomics on two separate sections, we expect people to run somewhere in the order of 50-100 proteins and 50-100 RNA. I think when we start integrating it, we'll probably look at some combined assays, about 100-150 targets when you mix RNA and protein together. We wanna design the chemistry, so it's not lopsided, so it's not like five proteins and 100 RNA. Got it. It's likely to be protein heavy, right, and the RNA in the back, especially in IO. Got it. Just in terms of the modest sort of like slippage in timing here, I mean, is it sort of something to do with technical issues, or is there something else that you're looking to fine-tune before rolling this out, into customer hands? Yeah. I think, right now the priority has really been to roll out our Fusion 2.0 changes, because those are all dependent on getting the RNA up and running because there are software and hardware upgrades that are rolling out. Mm-hmm. The priority is to get that out and as well as with our partners, get the RNAscope chemistry out as well. That will allow us to sort of prime the market, see the early access customers, the RNA capabilities before we come out. The pacing of it is really based on sort of a logical rollout of capabilities that build on one another. Got it. Got it. Okay. Then, you know, I wanna go back to the sort of broader point on market differentiation. It's been pretty interesting how, you know, different players, you know, have positioned their platforms, their in situ profilers sort of differently. I mean, you know, one of them is focusing on plex. The other one is sort of saying we're gonna be best in class throughput. You guys have, of course, sort of committed to, you know, multiomic sort of capabilities. What's the, how should we think about, you know, the market for these, you know, across these buckets, right? You know, going back to what Arusa was saying, and then perhaps even Elizabeth, there does seem to be a little bit of, you know, practicality element here. Just curious if you have any sort of early read on how the market, at least for those early adopters of this stuff, will shake out. I think the first thing, Tejas, I'll probably say is when I, when I think about the competitive landscape, the one you're particularly referring to, you know, is sort of the discovery end of things, right? We also have the translational and the clinical, which have a different competitive mix to that. I think when I look at the discovery market and where things are, the way I would look at it is the discovery market, you could probably break it down to a genomics market and a proteomics market. You know, on the proteomics side, I feel like we're pretty established as a leader, and we're probably the platform of choice. If you think about publications in every metric, I think we dominate in that space. I think the genomics market, there are a number of platforms coming out. I don't think it's settled on which platform is a leading platform on the genomics side. I think there's gonna be this new market where it's sort of the multiomics market, where customers are agnostic on the analyte. I think that part of it is the one that we are very excited about in terms of engaging customers. Right now, I think if customers are value protein, they're probably looking at Akoya as their number one choice. If they have value RNA and they need that today, they may be looking at one of the other platforms. Going forward, I think multiomics might be the major modality for discovery, and that's where we wanna be able to serve our customers the best. Got it. Okay. Just from a science perspective, what needs to happen, you know, on your end to get to sort of, you know, 500-1,000 plex, perhaps, in terms of that multiomic panel. You know, obviously, there's gonna be a degree of platform convergence over time. I mean, you know, your competitors are also starting to talk about co-detection, not with the initial launch, but perhaps with the subsequent iteration of that platform. How are you guys sort of positioning, you know, Akoya to sort of, to compete sort of favorably in that sort of world? I think on the first question, technically, there's no limitation. It's just putting more barcodes out there that gets us to the higher plex. If there's value in doing that, we would. Obviously as you go to higher plex, then you have to trade off throughput, right? Now you're looking at less and less samples per day that you can do. We try to figure out what's the maximum amount of content, the maximum amount of throughput that's a sweet spot for customers, and that's usually what informs the product design. A great example, Tejas, on this one is that we thought there might be an excitement for higher plex RNA, but as we go talk to our customers, we find in IO particularly, the multiomics was far more exciting. That was sort of our biggest lesson coming out of AGBT in terms of the excitement around multiomics, and that's why we're focusing on that as a product concept. Less technology-driven, more market-driven on that perspective. I think when you think about, you know, market convergence, this market, to be fair, this market is pretty diverse in, you know, in multiple sectors, right? You got the proteomics, the genomic side of things. You got the discovery, the translational clinical. When I think about that, there may be some convergence, but I think at the end of the day, you're gonna require multiple companies to serve the broad range of needs in spatial biology. I think there'll be some convergence, but not total consolidation, if you will. I think where we become uniquely differentiated, I think, in this area is that you can see from the portfolio there's two areas. One is, as you try to go to the masses, customers are gonna look for the platform that's the most used, the most accepted, the most published. Currently, you know, we're the largest install base with the largest publications. We expect that to be true going forward. That's always gonna be a competitive differentiation for us. The second one, I think, is as the clinical side of it matures, I think customers are gonna go on and invest in platforms, especially in biopharma settings and top academic centers. They're gonna start to look to invest in platforms that can take them to the clinic the fastest, and I think that's where we're gonna be competitive as well. Brad, do you want to add to that? I would just add maybe from a business commercialization standpoint to expand on what Niro just said, but also to sort of repeat what I mentioned in answering Kyle's first question which is, we have an opportunity/responsibility to look at our current install base and say, "What can we do to maximize utilization of that system? What can we do to drive expanded use, expanded pull-through, higher plex and more applications?" Again, our real focus in terms of our expenditure and how we're gonna spend our R&D and commercialization dollars and product launches is really about not only making the instrument attractive to buy, as Niro mentioned, but making sure that we are kind of doubling down on the applications that our current customers can do. That is really a, kind of a high-level theme and focus of our product development priorities, and hopefully, that's obvious, Tejas. Got it. Thank you, guys. Appreciate it. Thank you. Thank you, Tejas. All right, our next question is from Tim, Capital One. Hey, thanks, Brian, Niro. Really interesting presentation today. You know, I wanted to ask you both, you know, how much focus you guys are likely to put on companion diagnostic partnerships. Obviously, you have one with AcuraBlade. It seems like that there's a lot of focus from the researchers that you spoke today just on the potential of companion diagnostics in the immunotherapy space? Maybe you guys can comment on that. Yeah. Yeah. Maybe just to make sure this is just crystal clear. A companion diagnostic and more of them are really the by-products of a successful near-term strategy, to deliver value to the very speakers you heard in the last roundtable. In their clinical trial and their translational studies and their clinical research efforts is growing our top-line business, growing the install base of the HT, getting broad adoption of PhenoCode as a revenue driver. Those are all near-term revenue activities in the translational market. A by-product of that over the years are additional CDx deals. That is a long-term goal. The near-term revenue driver and the near-term realization of that strategy is revenue in that translational market. It's growing of the ABS service business. That is where we're focusing our effort with, again, a by-product of that being potentially future CDx deals. Does that make sense? Yeah. That helps. I guess, you know, a follow-up to that question then would be, you know, just given the summaries that each of the researchers gave in the latter part of the discussion on the clinical potential for using spatial biology tools, I mean, could you sort of guesstimate, I mean, how many years will it take before we get companion diagnostics in the immunotherapy market that could increase efficacy for immunotherapies like KEYTRUDA? I would defer to those experts who just, I think, answered Gavin's question on their belief that this is years away. It's happening real-time. You heard Dr. Rowley talk about it already. You know, I think it's. I think the DeciBio report is actually also a pretty good window into that timing. You know, I do think it's just in a few years. Again, along the way there, our growth driver is going to be broader and broader utilization as there is more and more recognition of the applicability of multiplexing in many of those IO combinatorial clinical trials. Okay, great. No, that's very helpful. Thanks, Brian. Thanks, Jim. Our next question is from David from Piper Sandler. David, can you hear us? Nope. Okay. Well, that was the last question then, from Jim. We appreciate everybody's time. We appreciate you sitting through our second annual Spatial Day. Feel free to reach out with any other additional questions or comments. Thank you all for your time, and have a wonderful rest of your week and a wonderful holiday season.