Hello everyone, welcome to today's Meet GSK Management event, focused on the next wave of our pipeline. This is an interactive event with a presentation sent to our distribution list by email, and you can also find it on gsk.com. Next slide, please. This is our usual safe harbor statement. Next slide, please. Our Chief Scientific Officer, Tony Wood, will take you through the presentation before Hesham Abdullah, Kaivan Khavandi, and Luke Miels join for Q&A. With that, I'll hand over to Tony. Thanks everyone, and welcome. Today I want to talk to you about our next wave of R&D innovation at GSK. At the pace we're now working, some of these medicines will contribute meaningfully to our growth this decade, while others will reach patients in the 2030s. At the core of our approach is deepening our expertise in science related to the immune system, taking it beyond our current understanding of its role in autoimmune disease and infection to a deeper understanding of fibrosis and autoinflammation for precision treatment, and ultimately to target interventions that support healthy immune system aging. We're doing this through investment in scientific partnerships and advanced platform and data technologies to substantially increase our use of genetics, functional genomics, biomarkers, and immunophenotyping, all of this to identify the right target, the right drug, and the right patient. Our aim is to develop significant competitive advantage for GSK with deep expertise inside the company and full advantage of a bespoke world-class network of partnerships. Data collaborations with experts like Oxford, Cambridge, and Boston universities, Ochre, and FinnGen, together with platform tech partnerships like Flagship, Wave, and our recent acquisition of Elsie, are enabling us to understand underlying disease processes, reach previously inaccessible targets, and better identify patients for treatment. With the progress we've now made, I believe we have a clear path to develop a differentiated pipeline of first and/or best-in-class medicines and vaccines across a range of significant diseases that will deliver benefit at scale to patients and ultimately sustainable value to shareholders. Next slide, please. Today I'll be focusing on our approach in oncology and respiratory immunology. These are two areas where we have significant opportunities for growth in 2025 with prospective launches of Blenrep and depemokimab, and where we're increasing and prioritizing capital investment to accelerate key pipeline assets. In oncology, our plan is to rapidly expand beyond our current focus in hematological and gynecological cancers to develop antibody-drug conjugates, or ADCs, for the treatment of solid tumors. And in respiratory and immunology, we're building on decades of knowledge in inflammatory mechanisms to lead in COPD and to target fibrotic lung, liver, and kidney disease. We plan to provide a similar update on HIV in the second half of 2025 with infectious diseases following in due course. But for today, let's start with oncology. Next slide, please. Just last week at ASH, we presented statistically significant and clinically meaningful overall survival results for a Blenrep combination with a 42% reduction in risk of death over current standard of care, which may translate to giving patients up to a median additional three years of life based on projections. With an acceptable safety profile, including on eye-related side effects, we believe Blenrep has the potential to transform treatment for people with multiple myeloma. We've already completed seven regulatory filings, including in the U.S., where we expect a decision by July of 2025. Next slide, please. We've already seen promising results in newly diagnosed patients as well. For example, from the BelaRd trial, which showed a 100% response rate. So we're increasingly confident in Blenrep's first-line potential and plan to start the phase III DREAMM-10 study in newly diagnosed, transplant-ineligible, multiple myeloma patients this month. The trial will compare a Blenrep combination with standard of care, looking at Minimum Residual Disease, or MRD, negativity and progression-free survival as primary endpoints. We expect initial results by the end of 2027. Blenrep reinforces the significant potential of antibody-drug conjugates as targeted cancer therapies. Next slide, please. In addition to Blenrep, we have two exciting opportunities with our B7-H3 and B7-H4 assets, as well as the potential for further BD in this space, as demonstrated by our exclusive option agreement with Duality Biologics recently announced for DB1324 in GI tumors. First, let's look at our GSK227 ADC, which targets B7-H3. The B7-H3 antigen is overexpressed in a wide range of solid tumors, including lung and colorectal tumors. These tumors are sensitive to topoisomerase inhibitors, which means we can deliver potent medicine targeted to cancer cells, sparing healthy tissue, unlike traditional chemotherapy. The topoisomerase payload in 227 is clinically validated, and preliminary data suggest potential anti-tumor activity across a range of indications. Next slide, please. These early data from the ARTEMIS-001 phase I trial in advanced solid tumors were presented by our partner, Hansoh, at ASCO in 2023. They show promising initial clinical activity in small cell lung cancer, non-small cell lung cancer, and sarcoma, with multiple confirmed responses and a manageable safety profile. Next slide, please. This slide shows further results from ARTEMIS-001 presented at the World Conference on Lung Cancer in September this year in patients with extensive stage small cell lung cancer, with disease progression on or after platinum-based chemotherapy. The trial showed a 61% and 50% overall response rate for two different doses of 227. For patients at this extensive stage of disease, standard of care is typically single-agent chemotherapy with an expected response rate of under 20%, illustrating the impressive nature of these results. B7-H3's transformative potential has also been recognized with breakthrough designation from the FDA. phase I and 2 trials will start in coming months with a comprehensive development program, including in lung, GI, GU, and beyond. Next slide, please. Like B7-H3, B7-H4 is expressed across a number of solid tumors, particularly in endometrial, breast, and ovarian tumor types. Our B7-H4 ADC, GSK584, has best-in-class potential in ovarian and endometrial cancers, with additional opportunities in other solid tumors. We're exploring a range of potential biomarkers in our development program to determine the patients most likely to respond to treatment and using AI/ML to understand the mode of action of ADCs in different tumors to guide and accelerate combination development, including with Jemperli. Next slide, please. phase I data presented at ESMO in 2023 showed promising proof of concept for GSK584 in patients with advanced solid tumors refractory to standard therapy. This trial primarily enrolled breast cancer patients with a focus on triple-negative breast cancer, but also included small numbers of patients with ovarian and endometrial cancers. Across the study, subjects were heavily pretreated with a mean of five prior lines of therapy. As you can see here, GSK584 showed broad anti-tumor activity in advanced solid tumors, with encouraging clinical activity in triple-negative breast cancer. We believe these data are representative of the type of activity expected in gynecological tumors, an area of strength for GSK. We're moving quickly to generate further evidence, including in combination with Jemperli. Next slide, please. Given their potential, we are prioritizing and increasing investment in 227 and 584 in 2025 and 2026, and accelerating their development with a plan to deliver the first phase III results in 2027. For B7-H3, we're developing this asset in lung, colorectal, head and neck, and prostate cancers and evaluating other solid tumors. Together with Hansoh, we expect to share updated small cell lung cancer and osteosarcoma data at ASCO and dose escalation at ESMO next year, with pivotal studies planned for Q4 2025. For B7-H4, we're developing this asset as monotherapy and in combination across multiple indications in ovarian and endometrial cancers. We expect to share early ovarian and endometrial data at ASCO and ESMO in 2025. Those expansion data are also anticipated during the year, with a targeted pivotal study start in 2026. For both programs, we'll apply extensive biomarker work, supported again by AI/ML, to determine the patients who will benefit most from treatment. Next slide, please. As you can see, the ADCs complement and strengthen our oncology pipeline, expanding our presence across multiple tumor types with significant opportunity. ADCs could disrupt conventional treatment regimens alone and in combination with PD-1s, complementing our Jemperli programs and replacing chemotherapy in later lines. Next slide, please. Our clear focus is to prioritize and accelerate development of assets with the greatest potential, with effective stage gating and checkpoints in place to drive timely investment decisions. Gynecological cancers remain a cornerstone of our portfolio with Jemperli and now our B7-H4 ADC. We're moving into additional solid tumor types, including lung, colorectal, and head and neck, with both our B7-H3 ADC and Jemperli. As a reminder, we expect initial results from the AZUR-1 and AZUR-2 trials exploring Jemperli in rectal and colon cancer in 2026 and 2027, respectively, with the phase III JADE study in locally advanced head and neck cancer expected to read out in 2028. We're confident that B7-H3 will serve as a significant cornerstone in many of these cancers. Beyond these, we continue to evaluate our TIGIT monoclonal antibody, belrestotug, mindful of data we've seen elsewhere, and our development strategy will be informed by additional data generated over the next 12 to 18 months. With that complete, now let's turn to our respiratory and immunology portfolio. Next slide, please. Our unique understanding of the role that inflammation plays in airway disease is based on decades of research in chronic respiratory conditions like asthma and COPD. In respiratory, our focus on the underlying biology and heterogeneity of inflammation, notably in COPD, is leading to a differentiated pipeline of long-acting options with multiple mechanisms of action. In immunology, we're researching how inflammation contributes to the development of fibrosis in the lung, liver, and kidneys. Fibrotic diseases are thought to account for up to 45% of all deaths worldwide, so this is a major area of need. As I'll share through examples of our work in COPD and liver disease, we're transforming our approach with the application of advanced technologies. Next slide, please. As with our work in cancer, our approach starts with understanding human biology, in this case for COPD. This disease affects more than 300 million people globally and is the third- leading cause of death worldwide, excluding COVID. It's a complex and heterogeneous condition driven by multiple inflammatory pathways. We were pioneers in establishing the role of IL-5 through our work with NUCALA, and we're delighted with the positive headline results of MATINEE, our phase III trial in patients with COPD with high eosinophil levels. We're applying the unique insights we've collected in the NUCALA phase III COPD studies to progress our wider COPD pipeline, including for depemokimab, where we aim to start phase III development in COPD next year. Next slide, please. Both IL-33 and TSLP also have strong genetic evidence supporting their potential as promising targets for intervening in COPD. People with naturally occurring lower expression of IL-33 and TSLP proteins have a reduced risk of COPD, and lower activity of both TSLP and IL-33 seems to decrease susceptibility to COPD in an additive manner, providing strong rationale for a combination therapy that targets both of these pathways. Next slide, please. By integrating a range of diverse data from disease phenotyping, human genetics and genomics, cell biology, and insights from our own scaled clinical studies, we've built a sophisticated and unique understanding of different COPD populations. Together with insights from partners like Cambridge University and Boston University, we've amassed a vast data set to help answer key questions on COPD progression, predictive outcomes, and biomarker measures for disease stability. Factors including T2 inflammation, eosinophil levels, and a host of others influence both disease progression and response to treatment, and we're now using modeling to map prospective treatments to specific patient types. This work strongly supports the need for multiple mechanisms in addition to IL-5 to help as many patients as possible. Next slide, please. This slide shows the range of mechanisms of action in our pipeline, which target different biological pathways to reach the broadest range of COPD patients. The chart is split by eosinophil levels and disease severity, but you can also map against factors like smoking status and exacerbation history. It's important to note that while we're already well positioned to compete in COPD with our IL-5 assets, these will only be suitable for about 40% of patients whose disease is driven by eosinophils. Next slide, please. There remains significant need for improved biologic options in COPD to stop patients exacerbating and improve persistence with therapy. We see ultra-long-acting medicines as a critical differentiator, providing sustained suppression of inflammation to impact disease progression for new efficacy benchmarks such as exacerbation and hospitalization reduction, as well as adherence and convenience benefits. We have deep understanding of patient payer and HCP preferences in this space. As the only company with this range of mechanisms in ultra-long-acting format, we believe we're very competitively placed to extend our leadership in this disease. Next slide, please. In the next two to three years, we anticipate a series of pivotal trial starts in COPD, which could transform the treatment landscape. Next year, we're targeting approval for COPD through NUCALA and expect to begin a phase III trial for depemokimab with a potential filing by 2031. Our IL-33 program is currently in phase I, and we're targeting a potential phase III start in 2027. For TSLP, currently in phase II, we expect to begin phase III around the same timeframe, whilst also evaluating the combination potential of these two molecules. This is an ambitious development plan, conducting multiple large-scale pivotal programs in parallel. But given our significant experience and strong heritage of working with respiratory investigators, we believe we're well placed to deliver. Next slide, please. As I laid out earlier, our work in human genetics and phenotyping is also generating insights that are informing moves into other areas, including liver disease. One example I'd like to share is our progress on steatotic liver disease, or SLD, a chronic progressive disease that affects around 5% of the global population and has significant health burden. Steatotic liver disease includes several conditions associated with the accumulation of fat in the liver, including metabolic dysfunction associated steatohepatitis, or MASH, also known as NASH, and alcoholic liver disease. Over 12 million patients globally have MASH, and a further 26 million have alcoholic liver disease. This accounts for half of liver-related deaths in developed countries. There are currently no pharmacological treatments available for ALD, which is associated with significant emotional and psychological issues. Many patients also struggle with alcohol addiction, making lifestyle changes difficult. These conditions have typically been seen as difficult to treat. But new modalities, including oligonucleotides, have shown real promise in targeting liver diseases due to their ability to specifically modulate gene expression in the liver. Next slide, please. HSD17B13 is a gene involved in lipid metabolism and is primarily expressed in the liver. Human genetic studies show that naturally occurring variants in this gene are protective against both alcohol-related and non-alcohol-related liver disease, providing a 30% to 50% risk reduction in carriers compared to non-carriers. This is one of the strongest levels of genetic validation we've seen for any target. And reassuringly, this protective effect is maintained even in the case of continued alcohol consumption. Next slide, please. GSK'990 is our second oligonucleotide in the clinic, following Bepirovirsen for chronic hepatitis B. 990 is an siRNA therapeutic designed to selectively target HSD17B13, reducing its expression and slowing or halting disease progression in ALD and MASH. In phase I, it has demonstrated robust target knockdown and encouraging reduction in markers of liver injury. Next slide, please. We've now started the HORIZON phase II study in MASH patients with advanced fibrosis, and we're initiating a phase II trial in patients with ALD called STARLIGHT. GSK'990 is another fantastic example of our tech strategy in action, building out our oligonucleotide platform. The program's also been strongly informed by genetics, and we've used single-cell imaging to confirm the expression of the target in liver cells, an approach which roughly triples the chance of a target reaching phase III. Our clinical development program for 990 is being informed by biomarkers, so we're able to predict the patients most likely to benefit from treatment. We've identified several well-established and non-invasive tests that have excellent predictive value for clinical outcomes. Next slide, please. Looking beyond 990, we see further opportunities in liver disease enabled by our growing expertise in this space and enhanced by our collaboration with Wave Life Sciences. We're evaluating two additional targets which complement GSK'990. These both have strong genetic association with a range of steatotic liver diseases, and oligonucleotide tools have shown promising early activity in validation studies. While this works at an early stage, we see future potential for an emerging portfolio of complementary mechanisms of action informed by genetics to reach a broader group of patients and address the significant burden across steatotic liver disease. Next slide, please. So we expect a growing pipeline in respiratory focused on COPD and an exciting evolution of innovation in new areas, just a few of which we've covered today. These exemplify our approach to science and technology. We anticipate a flow of data readouts and potential approvals for the remainder of the decade. In the near term, this encompasses approvals in a range of indications for depemokimab, the first long-acting biologic to reduce severe asthma attacks leading to hospitalizations by over 70%, and NUCALA in COPD. New assets in COPD and liver disease offer significant growth opportunity in the longer term, among other early-stage programs focused on fibrosis. Next slide, please. So to finish, we're making strong progress in R&D with 67 pipeline assets, 18 of which are in late stage. Our deepening expertise in immune science, the use of advanced technologies, and world-class strategic partnerships is leading to a differentiated and exciting early-stage pipeline that will deliver growth to and beyond 2031. With strong momentum in oncology, a clear path to extend our leadership in respiratory, and exciting new prospects in immunology, our pipeline offers significant growth opportunities and high-potential medicines. As a result, we're increasing and prioritizing investment in both oncology and respiratory and immunology, while continuing to pursue major pipeline opportunities in infectious diseases and HIV. We're very confident in achieving our prospects for growth and global impact on health. Today, I shared just a few examples and look forward to sharing more in due course. Looking forward, we have multiple data and regulatory catalysts to come, including five regulatory approvals next year and an increasing flow of innovation throughout the remainder of the decade. With that, we'll now open for questions. Thank you. Thanks, Tony. And we'll now begin the Q&A portion of the call. If you'd like to ask a question, please indicate you would like to raise your hand using the Zoom toolbar, and I will unmute your line. Our first question goes to James Gordon. Hello, James Gordon from J.P. Morgan. Thanks for taking the questions. Two questions, please. One was AI machine learning. So I think, if I understand it right, that you're going to use that to sort of work out cutoffs for your ADCs. So can you help us out in terms of how big would the eligible treatment populations be for these two ADCs, so 227 and 584, based on the biomarkers you envisage using? And what are the actual biomarkers that you'll be looking at? Is this a bit like IHC that we've seen for data in terms of how you assess what's being expressed? So if you could elaborate on that, please, how it works and what the eligible populations might be. And then the other one was just a clarification on the biologics for COPD. So is the idea that the IL-5, TSLP, and IL-33, they're all going to be six monthly, or do you need to see on that? Thank you. Hi, James. And let me start just with the ADC point. I'll make a broad point, and then Hesham can perhaps hand over to you in terms of how we're thinking about bringing on some additional precision to the ADC landscape. And then Kaivan, over to you for pharmacokinetics for the COPD assets. James, the first thing I'll stress is that clearly one of the attractive features of both the B7-H3 and B7-H4 ADCs is that they are broadly expressed across various treatment populations. And so what we're interested in, and we're not going to give away a great deal at this point, as I'm sure you will understand, are other features associated with the cancer that might indicate a greater or lesser response. And Hesham, I might just hand over to you to add a little bit to that. Yeah, no, absolutely. I mean, James, I think one of the things that we've learned certainly with antibody-drug conjugates and the targets that they actually selectively go after is the target expression levels are not the only thing that really matters. So I think the field has moved beyond that. So certainly looking at not only target expression, its natural distribution, its density, and how the ADC is actually being processed within the cell are really critical factors. And so as part of our translational strategy, these are elements and variables that we will prosecute and look at. I'd like to maybe just call your attention to the fact that, at least based on the data that's currently available to us from the phase I study that's being conducted in China, both for B7-H3 and B7-H4, we're seeing data that says that there's clinical activity across a broad array of tumors, irrespective of biomarker or at least target expression at this point in time. So we're going to learn more. And we certainly have already initiated our own development program. And we have a translational strategy set in place that helps us even go after bigger treatment effects potentially that could be defined through this biomarker strategy. But as of now, the data tells us that the drugs are active across both high and low expression levels. Thank you, Hesham. Moving over then into COPD and PK, perhaps you might just give James and everyone listening a sense of where we are on the qualification of the pharmacokinetics for the three assets. Yeah, thank you. So for depemokimab, of course, the protocol has been established in the SWIFT studies as a six-monthly administration. For the anti-TSLP, the data that we had at deal signing from exposure in China and Australian participants demonstrated through modeling confidence that that will also be amenable to a twice-yearly dosing regimen. We're taking that into a phase II study in asthma, which we've just had IND confirmation that we're safe to proceed in Q1 next year for dose selection. Then for the anti-IL-33 program that's currently in a phase Iab, we're moving into a patient cohort next year where we're going to do repeat dosing. The first dose will be after 12 weeks. And then we're going to follow up after the second dose for over six months to confirm whether it's going to be a three or six-monthly dosing regimen, both of which would meaningfully differentiate versus other anti-IL-33 molecules. Thank you, Kaivan. Clear. I think we've moved on to the next one. Thank you. Next question is from Simon Baker. Simon, your line should be unmuted. Thank you, Mick. Thank you, everyone, for the presentation. Three quick questions, if I may. Just starting off with Blenrep. At ASH, a week or two back, it was shown that the dose reductions related to oculotox led to a fairly rapid resolution and did not impact efficacy. So in light of that, I just wondered if you could give us the feedback you've had from physicians on their enthusiasm for using Blenrep and their comfort with the side effect protocol. I kind of get the sense that physicians are probably a bit more relaxed than we are. So any color on that would be handy. And then on 227 and B7-H3, there was a poster at San Antonio last week showing B7-H3 being targeted through CAR-T. You didn't mention CAR-T today, so I just wondered if you could give us your thoughts on that as a modality for GSK. And then finally, moving on to the long-acting IL-33, there was work published earlier in the year showing potential utility in cystic fibrosis. I wonder how interesting that is to you, given it's an obvious adjacency to your respiratory work and an adjacency which has orphan drug potential. Thanks so much. Yeah, thanks, Simon. Let me see if I can get through all three. Please remind me if I miss any component of it. Why don't we start, Hesham, with you just quickly on the physician feedback with regards to Blenrep and the emerging profile over efficacy versus safety effects? I might then come back and just make a quick comment on CAR-T before we hand over to Kaivan. So over to you, Hesham. Yeah, thank you, Tony. I mean, certainly the feedback that we've been getting from physicians that have had experience with Blenrep is it's relatively easy to administer, especially given the context around, of course, the fact that we have 70% of patients, multiple myeloma patients, that are sitting in the community. So the ability to be able to administer this on an outpatient basis, the ability to not require hospitalization, is really important and critical for them. And the fact that it can be given immediately. So there's no need to wait for certainly patients going through apheresis, lymphoid ablation, or, of course, for them to really have to experience some of the side effects that are associated with other therapeutic modalities like CAR-Ts or T-cell engagers. Now, again, it's a familiarity that the physicians have to acquire, of course, just as is the case, of course, with any new class of drugs that emerges in oncology. I think you think back to, let's say, for example, checkpoint inhibitors, right? And this was a class of agents that initially, when it came onto the scene in 2014, was associated with these immune-mediated adverse events. Not everyone was very familiar with them. But once everyone realized that they were inflammatory in nature, can be managed by steroids, and that they can certainly interrupt the dosing until they resolve and then resume it afterwards, you look at it now. At least, no one's really talking about them at this point in time when they're thinking about checkpoints. The same holds true here in terms of starting out with a 2.5 milligram dose, induce the depth of response. And we've seen that, whether it be the MRD negativity, whether it be the duration of response. And then basically, over time, increase the dose and extend the schedule. You saw the data at ASH when the dosing intervals were extended to every eight or 12 weeks. The response was sustained and maintained. And you saw that the ocular toxicity in terms of its incidence and discontinuations resulting from it decreased over time. So I think that's the message and the feedback that we're getting from the physicians as well, is their ability to effectively use the dose reductions, the dose interruptions, help them to manage the ocular side effects quite well. And look, Luke, I'll give you a chance to speak to this feature of developing an understanding of managing the treatment population as well. But before I go there, Simon, just quickly on the CAR-T question. Obviously, whether it's ADCs, CAR-T, or TCs, what we're doing is exploiting the same antigen proposition, but bringing a different feature associated with the cytotoxin component. One of the things that we find particularly exciting about the B7-H3 and B7-H4 ADCs is that we're able to deploy them to cancers for which we already know topoisomerase inhibitors are effective. So we feel that given that background, the ADC approach is one that is likely to bring the most effective therapeutic index. And indeed, as Hesham stressed earlier in the comparison, the more straightforward process for treatment. With that said, Luke, I don't know if there's anything you want to add just on the question over. Yeah, please. Yeah, I think Simon completely agree with the direction he's taking. And practicing hematologists, oncologists, I mean, they are empirical by nature. And I think, I mean, firstly, the overall survival, the profile that we've got in DREAMM-7 and DREAMM-8, I think essentially resets the discussion in the minds of many physicians. The other thing is, and Hesham, correct me if I'm wrong, if you look at the percentage time that a patient actually experiences blurred vision or ocular events, it's a single-digit percentage of the total time they're on the drug. So it's not a sort of ongoing durable disturbance, blurred vision, etc. It's a relatively confined point of time. And as Hesham has said, it's reversible. I think the other thing is if you just print out the PI, the safety profile of CAR-T, where you've got all sorts of novel things, parkinsonism, excess early mortality, unusual and partly irreversible AEs with CAR-T. So you've got excellent efficacy, but they are sophisticated and difficult to use. And as Hesham said, in the community, it's just practically not accessible. And if you look at the bispecifics, I mean, we're really getting a sense of the profile there. I mean, the MajesTEC-2, okay, there was a difference in terms of the rates of infusion, but that had 22% infusion deaths, right? I mean, other data earlier, I think neutropenia was around a third of patients got grade three. So I just look at this and I try and be as objective as we can in talking to experts using the drug. In that context, Blenrep is looking increasingly like an accessible product. And as Tony said, we're going to do a lot of work to manage that interface between the optometrist and the ophthalmologist. We need to see what the REMS that we get from the FDA is. But the benefit of the product is the benefit risk of the product has dramatically changed now that we've got overall survival against the CD38 regimen. Thanks, Luke. Simon, we'll move over now and deal with IL-33 in cystic fibrosis. I mean, I guess, Kaivan, we might talk more broadly about obstructive lung disease and what we see in the opportunities from the phenotypes there. Yeah, absolutely, so I think in eosinophilic, you often see TSLP and IL-33 occupying the same space together as an upstream alarmin, but actually, Tony presented some of the multi-omic approaches that we're able to pursue, and what we see is that whilst TSLP, IL-13 and IL-5 map together, you get a clear bifurcation where IL-33 is clearly moving towards more TH1, TH17 pathways and therefore maps more closely to this neutrophilic COPD obstructive lung phenotype. We're actively exploring the possibility of evaluating anti-IL-33 in non-CF bronchiectasis, and we're fully enabled to start that study next year if we wish. Of course, consideration can be made whether we include CF patients within that same trial design. Thanks, Kaivan. Thanks, Simon. Our next question will come from Seamus. Seamus, I'm sorry, Seamus Fernandez. Seamus, if you just unmute your phone. Hi, this is Akdan. I'm on for Seamus Fernandez. Thank you for the questions. So just to start off with, it looks like you're not pursuing combination strategies with IL-5. Can you explain the rationale behind this? And then shifting over to MASH afterwards, do you believe that your pivotal trials will require a biopsy? And more broadly speaking, what do you see for the opportunity in MASH as this market evolves? And then third and last question, just curious, Tony, what do you think is the most underappreciated asset that you presented on today? Thank you. Great. Thanks for the questions. Kaivan, I'll hand straight over to you on the IL-5 and MASH questions. Obviously, if you can include the context of the sort of temporal nature of the evaluation of all of that, and then please, in the context of MASH, address the fibroproliferative components and post-GLP-1, and I will, in the meantime, take a look among all of these exciting assets and figure out which one I like the most. Yeah, so Tony presented earlier a genetic risk score that we've created for IL-33 and TSLP, and you saw there that the combination of activity results in additivity for those pathways. I just described some convergence of TSLP, IL-13 and IL-5. Therefore, it's unlikely that we pursue a TSLP anti-IL-5 combination. However, there is a rationale for an anti-IL-33, anti-IL-5 combination strategy, and that's something that we would consider. Moving on to MASH, we're conscious that there's a number of mechanisms that are exploring anti-steatosis mechanisms in MASH. The unique differentiated value of HSD17B13 is that we've got really quite exquisite human causal data from genetics that we presented showing that this pathway is most relevant in advanced MASH and fibrosis and cirrhotic segments of disease. And those are the segments of the natural history that are going to be underserved by emerging standard of care. And what we see is that we're able to pair that genetic confidence with a cell-level understanding of what the pathway can do to ballooning of hepatocytes, inflammation, and fibrosis. And so the population we're studying in the phase II-B Horizon trial is MASH patients with F3 fibrosis, F3 stage disease. And we also include an F4 exploratory population in that study. We will be biopsying, of course, because we need to demonstrate either MASH resolution or improvements in fibrosis, and given that therapeutic hypothesis, that's particularly relevant to alcohol-related liver disease, where you see a very fast progression through to fibrotic phenotypes and where, of course, there's currently no pharmacological treatments for what is the number one cause of liver transplantation in the U.S. Yeah, in answering your question with regards to which ones I'm most excited about, I think the first point I would make is the one I did right at the beginning of the presentation, that these are all areas in which we are investing in because we see the potential for significant growth driven by a combination of the medical need and indeed the underpinning science. So if you don't mind, I'm going to answer it with two examples because I think they illustrate nicely the sort of features that science and programs which are now emerging from our research activities into clinical from our BD focus are going to carry for the future. Obviously, B7-H3 represents a very exciting opportunity. Hesham laid out the broader context of that. But what we have there is, through phase I studies, the opportunity to look for early efficacy, the confidence that the topo inhibitor brings, and then a range of additional either genotyping or phenotyping approaches that will, in time, begin to disclose, which will give us a sense of who's going to be the most likely patients to respond, even within the breadth of the antigen expression. So very excited about that one, given both the opportunity and the link back as well to the translational science that we're building within Hesham's group under Tony Ying, for example. With regards to Kaivan's world, for me, HSD17B13 is a really great example of something that carries tremendously strong genetic credentials. You can think about this one as being PCSK9-like. But what we have on top of it, and we didn't have time to go into detail, is single-cell work through spatial transcriptomics that then links the transcripts that are expressed as a consequence of that genotype against the cell types, ballooning hepatocytes, and indeed looks at it relative to alcohol as, for example, a given insult. So we're reaching a degree of precision in understanding target modulation and the patients who are most likely to benefit from these agents in both oncology and in MASH, which is going to characterize the approach that we take in the future. And for those two programs, they for me represent probably the pinnacle of what we've presented today. But it's important to stress that this is not a portfolio against which I'm making relative choices. Thank you. Our next question comes from Emmanuel Papadakis. Emmanuel, you should be free to speak. Yeah, thank you for taking the question, sir. Hopefully, you can hear me okay. So a few, if I may, please. Firstly, on oncology, the pipeline. So if I recall correctly, Blenrep was discovered internally, but you've now largely had to resort to in-licensing the next wave of assets from Chinese companies, both ADC. I mean, even beyond B7 molecules we talked about today, cobolimab, dostarlimab, or belrestotug came from outside the company. So what's been the holdup on building upon the early success of Blenrep, either within ADC or beyond in oncology in the last few years? That's question number one. Question number two, 990 in MASH. Thank you for the evidence on genetic validation. What about translational rationales? So what does the B13 protein actually do? If I missed it, you didn't tell us. What would suggest it should deliver a functional benefit when modulated and do any better than previously promising genetic targets you've outlined? I'm thinking of things like CCL17. And then a very quick one on IL-33. Is that actually a totally new molecule, or is it the same backbone as the molecule you discontinued several years ago for aspirin? Thank you. Okay, man, thank you. I will, again, brief answers and then perhaps some in fact, actually, in this case, I'll hand over to you, Hesham, to talk about the internal build in oncology discovery. I would say, first of all, this is not a reflection of a lack of internal effort, but rather a reflection of a deliberate strategy in which we recognize that by focusing on the sorts of technologies that we've just been describing that enable us to make better choices of targets and patients and combinations, then our efforts are more effectively deployed there. It gives us an insight into what, as you, I'm sure, are fully aware, is a very active potential BD landscape. Now, sitting behind that, of course, we are continuing to develop internal programs and follow-ons, for example, in the Pol-theta arena that Hesham, you might just briefly mention. And then I'll take up the genetics and translational rationale in a moment. But why don't you just mention a little bit about the internal earlier portfolio, Hesham? Yeah, no, I'm happy to. And Emmanuel, thank you very much for the question. I'll start off first by saying even I'd probably say industry leaders in oncology have relied on, certainly in licensing, to build parts and complement parts of their portfolio, including certainly other organizations that have, for example, a leading PD-1 inhibitor as well. Right? So I think we just have to keep that in mind. So the ability to have the right mixture and balance between business development, but also in-house research and development activities is really important and critical. I think with that in mind, I think, Manuel, you highlighted, of course, Blenrep was developed internally. We've complemented that now with B7-H3, B7-H4. And then, of course, a recent deal that we're doing, an option deal that we're doing with Duality on another ADC with another validated linker payload technology for a target that's directed towards gastrointestinal tumors. With that in mind, in parallel, we've been developing an internal linker payload platform as well. And you'll be hearing more about this over the next 12 to 18 months as you see the first asset from that platform come into the clinic. So that's one. Two, we do have T-cell engagers. We actually have gotten access to them and are in the process of developing both bispecific and trispecific T-cell engagers that will be applied across prioritized tumor types and areas, whether it be in solid tumors or hematologic malignancies for that matter. Then three, what I would highlight, of course, to Tony's point, we're continuing to look at small molecules as a means of synergy and as combination partners for some of our assets, including antibody-drug conjugates. I think one area to highlight, of course, is DNA damage response and the potential for that, of course, to synergize with the topoisomerase warheads in our ADCs. The Pol-theta program is one example of that. It's currently in the clinic and being explored in combination with Niraparib across gynecologic malignancies and breast cancer as well. But you could imagine that it could be certainly a key combination partner for antibody-drug conjugates moving forward, along, of course, with the dostarlimab, no doubt. So we found really a good balance between giving time for us to be focused and selecting the right technology platforms within our research portfolio, but then complementing that with clinical stage assets that we can certainly get access to through business development. Yeah, and just to reiterate then, of course, not unusual for late-stage clinical pipelines to come in at least 50% from BD-related activities and a deliberate choice on our behalf to focus our early-stage work on technologies, which will give us a better opportunity to make those selections based on an understanding of the patient landscape and potential combinations, coupled with very focused internal work in areas where we already have a clinical lead that informs discovery programs. As far as moving on to the other two questions are concerned, let's do the oligo program first. I mean, Kaivan, you can add a little bit more to this, but one of the big advantages of oligonucleotides is that we don't need to understand fully the biology associated with the target, and I'm sure you're probably aware that for HSD17B13, actually, that still remains to be fully illuminated, but the key advantage we get from oligonucleotides, once we understand the direction of the factors, the effect is one can recapitulate the phenotype simply by addressing the target at a message level. As far as IL-33 is concerned, it is the same molecule that we had earlier. It carries half-life extension technology, and there was no basis, given the profile that we'd accrued of the molecule, to want to search for an alternative. But Kaivan, anything else you'd like to add on HSD17? Yeah, so I think whilst the mechanism not fully identified, what we do understand is that it's able to target the lipid particles that sit on the surface of hepatocytes. And I briefly described some spatial transcriptomics work that we've done in primary human hepatocytes that shows that targeting that pathway reduces ballooning, which is the most powerful prognostic cell phenotype that is linked to outcomes. And so that really represents quite a distinct pathway from any other mechanism to target inflammation and fibrosis in those advanced stages of disease. All right. Thank you. Thank you very much. Excellent. Our next question will come from Peter Wellford. Peter, you should be free to talk. Hi, thanks for taking my questions. Apologies if some of these have been asked, but I joined a bit late. Firstly, just on Jemperli, I think you mentioned for AZUR-1 and AZUR-2, but I may have missed this. Was it 2026 and 2027, you said, for the readouts for that? Or was it 2027 and 2028? I guess, curious, given your recent breakthrough therapy designation you've got for rectal, is this possible to file based on just one of those readouts, or is it even possible based on an interim read before then, do you think, to potentially consider discussions with the FDA? Just moving on then to the respiratory portfolio. For depemokimab, presumably, we should be assuming for COPD that there are two phase IIIs planned, and you can't possibly piggyback on any of the NUCALA data for depemokimab at all. But just wanted to confirm that. Then just for IL-33, I guess two things here. One is when you talk about the combination, I'm right to understand that's two separate drugs. You're not considering a bispecific or any single modality that hit both of those targets. And if you looked at all of the biologic rationale, when you think about the phase III for IL-33, are you targeting the ligands? And if so, have you potentially looked at whether or not you should be targeting a prior smoker's population? Thank you. All right. Thank you, Peter. Let me start, Hesham, with just timelines for the Jemperli program and the clinical strategy for rectal. I'll hand straight over to you on that one. Yeah. No, thank you, Tony. And I think let me start off by just reiterating the data, of course, which is extremely promising, great news for patients. That was presented at ASCO 2024, and specifically 42 patients with locally advanced dMMR/MSI-H rectal cancer all had a clinical complete response, 100% clinical complete response, which is quite, of course, unheard of, but just shows how exquisite the science is here in terms of the sensitivity of these patients to checkpoint inhibition. With that in mind, of course, we designed the AZUR-1 and AZUR-2 studies. The AZUR-1 study, of course, is a phase II study. It's a single-arm trial in locally advanced dMMR/MSI-H rectal patients. And then AZUR-2, of course, is a phase III study in neoadjuvant locally advanced colorectal cancer patients. So the readout for AZUR-1 anticipated in 2027. And then the same holds true for AZUR-2. So both of them, actually, Peter, are planned for 2027. Bear in mind, of course, that if AZUR-1 data is available early, we will plan on, of course, being able to file off of that. So both studies are not necessarily required to make a filing, only one of them. If you may recall, of course, we did have a discussion at an ADCOM, an Oncology Drug Advisory Committee meeting in 2023, whereby we got agreement on the single-arm design of AZUR-1 as well. Thank you, Hesham. Just moving over then to the questions on COPD, where we might start, actually. Obviously, the standard will be two phase III clinical studies, but it might be worthwhile commenting on what we've learned or how we will be designing that relative to lessons from NUCALA, Kaivan, at a high level, bearing in mind that we're yet to disclose those data. That will come in the first half of next year in May. And then as far as IL-33 is concerned, perhaps we might just talk a little bit about what we see with regards to the behavior in COPD for the non-smoker and ex-smoker populations and our view of whether or not they are distinguished phenotypes for the future. I might, Peter, just say one thing with regards to targeting the ligand versus the receptor. Obviously, you'll appreciate from our experience with IL-5 that one can make arguments on both sides of that equation. Typically, though, when you're targeting the receptor, it's often the case that you see higher turnover of those molecules. And given that we are focused on a long-acting portfolio, that was a factor in our decision. Although the total pharmacology associated with each side of that axis obviously remains to be determined. We will continue to examine the phenotypes associated with the diseases in question as the development programs proceed. Kaivan, over to you just in terms of high-level lessons, thinking about the depemokimab-COPD program. Yeah. Just to reemphasize, given that it's definitely come out, of course, a different molecule, the assumption would be that we would have two replicate pivotal studies. What we're exploring is whether we could use one of those studies to evaluate populations that could provide a differentiated label. One area that we're exploring is earlier disease and predicting those at risk who have either had no exacerbations or one exacerbation in the year prior to enrollment. Obviously, with respect to the variables that we identified in the MATINEE study with NUCALA, what I'll say is that if you have a like-for-like comparison with T2R mechanisms and COPD, we're very competitive. There are variables that we'll take through as learnings for the design of the phase III for depemokimab. As regards anti-IL-33, I think what you will have seen is varied data across, again, peer companies that may be targeting receptor or ligand, and then reaction, I'd say, to relatively small empirical data sets for subgroups in phase II. What you saw in Tony's presentation is that we're taking a much more data-driven approach and using these proteomic and multiomic and genetic approaches to be able to evaluate explainability of reacting to small data sets. In our phase one study of anti-IL-33, we're stratifying by smoking, and we're doing a number of mechanistic assessments within that, including our in-house clinical unit in Addenbrooke's in Cambridge. So I think that we're going to be making data-driven decisions that might be based on more mechanistic rationale than just reacting to small subgroups in clinical studies. Thanks, Kaivan. I wonder, Luke, if it's worthwhile giving you an opportunity just to talk about how we see the long-acting portfolio in lung disease at this stage. Thanks, Tony. I mean, I think there's a degree of natural synergy, of course, in having others validate using short-acting programs to target, as we've done ourselves, obviously, with Nefaludomab, NUCALA, with depemokimab, which attempted to do that with TSLP. And if we could replicate that with the IL-33 program, yeah. I mean, there is also the synergy, of course, in terms of EOS and T2 inflammation positioning the IL-5s and 33s in, obviously, in people who are former smokers and then stratifying them based on their eosinophil level. So yeah, I mean, I think the market research that we have is that there's a potential to disrupt the relatively low penetration of biologics in respiratory disease. I mean, it's still below 30%, which I think is somewhat disappointing considering the profile of these assets. And so shifting into products which can be administered in the U.S. context by physicians in office obviously has a lot of benefits in terms of just barriers to injection, compliance, and other parameters which are favorable to that method of adoption for physicians and also the patients. So the market research is very encouraging. We now need to get out there. And I think the profile of depemokimab is very, very competitive of what we're seeing so far. Thanks, Luke. Excellent. Our next question comes from Graham Parry. Graham, you should be free to speak. Great. Thanks for your question. Full and respiratory, actually. Depemokimab, the phase III data, I think you showed 74% reduction in exacerbations leading to hospitalization, which looks like the best comparator to the existing data. When you go out with that one, this is a question for Luke. Do you see that as a best-in-class, or is it just a convenience argument you'll be looking to commercialize that on? And then are you looking to sort of match existing COPD data that we've seen in the COPD study, or do you think, again, that could be a best-in-class asset there? And are you looking at a broad population in COPD, so across all biomarker subtypes? And then on the IL-33, just wondering if you have any data to say whether it works on both oxidized and reduced forms of IL-33. Does it prevent oxidation of IL-33? So any impact on RAGE EGFR pathway, which is one of the differentiating points that one of your peers points out for their product. And then on your combo strategy on IL-33 TSLP, is it co-formulation, or is it actually just the safety efficacy data in combination that's the rate-limiting step in taking that any further forward? Thank you. Okay. Thanks, Graham. Let's make a start with the Depi data in phase III. And I presume then, Graham, this is a bridge from asthma to COPD, and it might be worthwhile just emphasizing, Kaivan, before I give you a chance to speak to that, that as I've said on a number of occasions, what we see there is more the advantage associated with the breadth of coverage and across different phenotypes. When one looks at comparable populations, the efficacy is probably slightly lower than for Dupi, for example, in the high eosinophilic bronchitic population. But you'll see when we disclose the data next year that we have advantage with breadth of coverage. I'll come back to Luke at the end to talk about the advantage for best-in-class for the longer-acting agent mix. So please remind me to go there. I want to, before we go into Kaivan, just make a point on the co-formulations versus bispecifics because I forgot to answer Peter's question in that context, and at the moment, we very much see this as a matter of, first of all, establishing a clear understanding and proposition for the contribution of components, both in terms of appropriate doses and the most responsive patient populations. Taking that then into co-formulation will obviously depend on understanding just exactly what the doses are. One really can't make a start on that until you have a solid grasp of that. In terms of bispecifics, look, we've evaluated these as potential areas. They always bring the challenge of how one gets the dose proposition accurate on both of the arms. So we prefer, at the moment, to take a more optimized approach, particularly in an area where it may be that different patient subgroups are more responsive to one pharmacology versus another. So Kaivan, you might want to add a little bit more just in terms of COPD and a comment on the oxidized IL-33 pathway. Yeah, sure. So I think you hit the main point, Tony, that the MATINEE study included patients with chronic bronchitis and those with emphysema. If you look at the group with chronic bronchitis, again, you'll see this data in detail next year. It's competitive and comparable with the best data that's available for other mechanisms, but really, the differentiator here is that we recruited these patients with severe distal airway disease and emphysema, and we're seeing efficacy in that population, and those are patients who, of course, have the worst outcomes, and importantly, of note, in the recent GOLD guidelines, there was a stipulation for IL-13 being recommended for those with chronic bronchitis. Of course, the data set that was generated for NUCALA wouldn't have that same restriction in terms of societal guidelines. For IL-33, again, I think I touched on it earlier, that some of the perhaps overly simplified arguments around smoking, oxidation, and what we're seeing clinically might require a more nuanced description. Our molecule binds to IL-33 in its reduced form, but it also, of course, and therefore blocks the ST2 signaling pathway and the direct pro-inflammatory effects of the cytokine. We're seeing the same data preclinically that we do have the ability to modulate oxidized forms and therefore EGFR and RAGE. We say even with our own data, there's caveats around those translational models and whether you're having the correct expression of EGFR and RAGE in the relevant airway and tissue. So in summary, I think that we've got the correct molecule with respect to not being limited to a single receptor binding the ligand. And also, of course, most importantly, in a population that's got multi-morbidities, the dosing administration, which is, I think, very clinically important in COPD. Yeah. Of course, Graham, you'll recall that, for example, RAGE is something that has been actively targeted in the past without a great deal of success as well, so very much part of the emerging picture we have and understanding the precision of the disease phenotypes. Why don't I hand over to Luke any comment you'd like to add in terms of how we see the depemokimab positioning? Yeah. I mean, I think, Graham, I mean, if you look at the pooled data of 52 weeks, it's about 54% from memory, which is in the range of NUCALA. It's very consistent, which is no surprise considering the heritage. If you then look at that with Fasenra, it's very much in the range at their 41-week data in the 300 EOS population. I think Dupixent has been a little bit more creative in terms of how they cut the data in terms of eosinophil level and 24 versus 52 week, but our feedback through market research is it's seen to be comparable. And I mean, the fact is, it's exactly as you say, the main driver is the frequency of administration, but also, what we've learned from NUCALA is once you take these drugs into a real-world setting, we know there's a huge drop-off of biologics once patients get after about six months of treatment. It's a big drop-off, and there's a lot more control, of course, in a medical benefit physician-administered context where you essentially lock in six months of compliance for a single shot. And you look at two shots a year versus I mean, Dupixent at 26 shots. So yeah, I think, firstly, short-term, it's comparable, and it will be shorter burden. Secondly, over time, as we do real-world evidence and see patients, how they operate actually in the wild, I think that we'll see the frequency of injection and the injection burden translate to efficacy. But we need to do those studies. And we have a big program that's starting out to do that. Thanks, Luke. Thanks. Our next question comes from Jo Walton. Jo, you should be free to speak. Can you hear me? Yes. Excellent. I have three quick questions, and they all are from us. So some going to James and Tony's question. Can you talk a little bit more about biomarkers? Because you mentioned them as being important, but then you go on to say, well, other things. Just from a simple mind first, how are we going to decide what the appropriate populations are? Are there going to be biomarkers, or does everyone have to have some genetic test? And then you say, usually, this is the drug that you should try first. Secondly, to go back to Simon Baker's question about other modalities, just to spitball that you have to be interested in radioligands or radiotherapy because that seems to be the flavor of the day. And then finally, in your presentation about having knowledge that tripled your chances of getting a product to phase III. I wonder if you could tell us a little bit more about that and how you've measured that. If you've tripled your chances in getting things to phase, you must have done quite a lot of products to get those sorts of statistics out. So just tell us a little bit more about why what this is and whether the unit is uni-everybody is doing it now or will be doing it in the next few years. Okay. Thanks, Jo. Look, we couldn't hear you very well, so I'm going to make my best attempt at directing the question. I think the first one was about biomarkers and how one really navigates the journey of, first of all, identifying and then making them a practical proposition when you reach late-stage studies and patient selection. And I might actually just begin, Hesham. You could describe the journey we've been on in the context of dMMR-positive cancers for Jemperli in both, particularly, of course, in the endometrial setting. Because I think, Jo, that sort of illustrates the general path here that one starts in early translational studies with a range of hypotheses that are tested and then validated in later-stage studies with different instruments, which are more amenable ultimately to the market setting. Hesham, you can make a quick comment on radiotherapy and our focus. I mean, Jo, essentially, we have an exciting portfolio in front of us with regards to the areas that I described and Hesham's answered questions on. And we don't see radiotherapy as an area of interest for us at the moment. And then let me just quickly deal with the comment on the improved survival in early phase one studies from the single-cell data. This is work coming from Sarah Teichmann's lab, and it's a broad assessment of the performance of industry portfolios with and without an ability to localize targets into expression in individually very carefully characterized cell types. Nick, we can send Jo the reference, but that data very clearly identifies an improvement in survival in phase one associated with targets that carry those characteristics. But let's go back. First of all, Hesham, perhaps you make a comment on biomarkers in the context of our oncology portfolio. And Kaivan, you might just describe the journey for IL-5 in eosinophilia as an example of how that might develop elsewhere as well. So over to you, Hesham. Yeah. Thank you, Tony. And I think probably a really good example of that is, of course, the RUBY study and specifically looking at dostarlimab when it was combined with chemotherapy in first-line endometrial cancer. I think everyone's well aware, of course, that early on, we were really exploring two different hypotheses, right? The first is how the combined assessment of dostarlimab plus a platinum-based doublet chemotherapy regimen could have a treatment effect relative to chemotherapy in the dMMR/MSI-H segment. And we saw tremendous treatment effect there, right, in terms of the interaction between the biomarker and the combination effect. And then, of course, the outcome in patients, whether it be on progression-free survival or overall survival, quite dramatic treatment effects. We didn't stop there as well, too. So we also looked at the combination in the intent to treat patient population, which basically included the remainder of the broader patient segment, so about 75% of patients, which are the pMMR/MSS patient segment as well, too. And we also saw a treatment effect for PFS and OS. Now, while not necessarily as dramatic as that seen in dMMR/MSI, it was clinically meaningful. So I think, Jo, what we really have to do is really look at each tumor type and the patient segments that exist, whether it be based on certainly molecular characterization of those patient segments, whether it be based on proteomic characterization of those patient segments, and really ensure that we have different treatment options available and to better characterize the treatment effect of different types of therapeutic modalities. I believe you were referring to antibody-drug conjugates in terms of how we would think about biomarker selection for them, and I think it wouldn't be necessarily any different. There may be certain patient segments that require combinatorial-based approaches based on their biomarker status. There may be patient segments that require monotherapy-based approaches. There may be patient segments that require combinations with current or existing standard of care agents. And that is the way that we're thinking about and choosing to evaluate and assess biomarkers in terms of best tailoring treatment options for patients. And at the end of the day, we're in the era of precision medicine, and certainly oncology is leading the way there. And hence, how we're trying to make sure that we provide options to patients. Now, just one point around the radioligands to Tony's point as well, too. We don't necessarily have any interest there. I think the bottom line is for us as an organization, we're really focused on making sure that we impact 2.5 billion patients by the end of this decade, of course. And if you bear that in mind, I think what you're really thinking about is treatment options that are scalable and accessible. That is very important. And so when we think about the therapeutic modalities that we're introducing across our oncology portfolio, it is a priority for us that treatments are scalable and accessible. Some of the challenges that exist with radioligands, whether it be manufacturing of the isotopes, whether it be the supply chain, whether it be the facilities that are required to potentially manage these isotopes and the radiation, it's a difficult space to navigate. But it's important, of course, to continue to look at different treatment options that utilize different technologies for patients. Yeah. And look, the only other thing I would add to that, Hesham, of course, is we have an advantage with the collaboration that we have with Tempus, which helps us understand across a broad range of tumors and molecular characteristics that feature in the early identification of potential biomarkers. Let's just finish with a quick comment on eosinophilia and the T2 and adjacent populations, Kaivan, just to help Jo see, again, an example of how this has worked out in practice. I'll describe the NUCALA example, and if I may, a forward-looking view on how we're approaching this in the pipeline. With NUCALA, of course, using eosinophils as a surrogate for T2 high inflammation allowed for the correct patient selection and, of course, represented the first precision medicine approach in respiratory medicine and even actually in COPD metrics, and it really did pave the way across the industry for that approach within COPD. Coming back to that broader point around genetics reducing attrition, I think the key development at GSK is that we're now able to match genetic confidence with what Tony described from Sarah Teichmann's work, which is the cell-specific confidence. That's really important because it allows you to prosecute those questions in clinical developments. And if you think about the HSD17B13 program, we had genetic confidence, but then we're able to match that with a cell phenotype and then an organ phenotype and a clinical phenotype. And so when we look at the alcohol-related liver disease study, we're looking at digital pathology evaluated using machine learning, but then able to pair that with spatial transcriptomics to look in the clinic at what the treatment effect is to the cell, and then looking at other clinical measures such as FibroScan, which is able to predict outcomes in compensated alcohol-related liver disease and MR elastography. So you've got that continuum from genetics to cell to patients. And Jo, just to finish off, of course, across that continuum as well, and increasing, hopefully, simplification over time as we recognize the reality of treatment decisions that will be made in a more distributed sense in the future. Okay. I think hopefully that answers your questions, Jo. Thank you. Our next question comes from Rajan Sharma. Rajan, your line should be coming live. Thanks. I've got two questions on the oncology ADCs, actually, and Tony, and I guess Hesham as well. Begin to get your take on the mechanistic rationale of PARP and topoisomerase inhibitor combinations and whether that's actually feasible from a safety perspective. And I guess ultimately the question is whether there's combination potential with 584 and ZEJULA. And then the second one was just on B7-H3 and how you expect 227 to compare with the other assets that are in development. I guess kind of the most advanced looks like it's Merck and Daiichi's assets. So do you think there's an efficacy or a tolerability advantage there? Thank you. Yeah. Great. And look, let me just quickly deal with the PARP combo comments. Hesham, you can add any more detail, and then I'll leave you for the competitive positioning relative to Merck-Daiichi, which is sort of simply put in the context of combination opportunities and the focus within areas of our oncology portfolio where we're building a presence. Look, on the PARP combo combination, there is clearly pathway overlap when you look at the mechanism of PARP inhibitors and topoisomerase inhibitors. One might envisage potential synergies, but what we've seen from the PARP class in general is that that type of effect is occurring right at the top of the dose response and therefore introduces questions with regards to therapeutic index. It is an active area for us, actually, in terms of our detailed translational research going on in Hesham and Tony Ying's teams, but not something that we're necessarily actively pursuing as a foundational thesis in clinical evaluation at the moment. But Hesham, over to you. Yeah. Thank you, Tony. And just to close the loop on that, I mean, we certainly look at, for example, what could be a second generation of DNA damage response, especially with our POLQ or Pol- theta inhibitor as well, too. Better therapeutic index, more well tolerated, and potentially lends itself to being a more combinable agent as well, too, and especially with an antibody-drug conjugate. So something to consider, evaluate, especially as we think about development of these ADCs across certain patient segments, including those patients that could have homologous or combination deficient profiles as well, too. With that in mind, the question really around the B7-H3 ADC and its competitiveness relative to other assets or compounds that are currently also in the clinic. I would first and foremost, of course, just draw your attention to the fact that there was data that was previously presented from the phase I study conducted in China that showed broad, I would say, clinical activity across a number of different tumor types, and you may have seen, of course, that certain health authorities have recognized that. So we were granted a breakthrough therapy designation by the FDA based on the small cell lung cancer data that was presented at the World Lung Cancer Conference in September of 2024. And then just yesterday, we were also granted a prime designation in Europe, again, for the same patient population as well. So I think the development program is progressing well. I think for us, combinations are really going to play a key part in this. And of course, dostarlimab, which is an asset that's been well benchmarked to the leading PD-1 in the class, gives us confidence in our ability to have combinations with checkpoints across different tumor types, but then also, like I said, with small molecules potentially lending themselves as being key combination partners. We're also exploring potential external clinical collaborations. We know that the current clinical landscape is continuing to evolve, and there may be additional assets that we may want to combine the B7-H3 ADC with. Stay tuned on that front. And then, of course, the tumor types that we're moving into, and you've seen the expression profiles. Some of these tumor types, we have certainly key capabilities and strengths, so they lend themselves quite well in terms of the infrastructure that we've developed there. And then, of course, some of the biomarker work that we currently have ongoing, of course, which we think will be very unique, but also provide, hopefully, over time, a level of differentiation as we think about our development strategy moving forward. Yeah. Look, at the highest level, the competitive environment is something, obviously, that I am very attuned to together with Hesham and Luke and will be a feature of capital allocation to ensure that we're moving forward in a competitive frame. Okay. Super. I think we're on the last question now. That's right. Our last question today comes from Richard Parkes. Richard, you should be free to speak. Thanks very much for taking my questions. Yeah, just three quick ones, hopefully. On B7-H3, and excuse me if I missed this because I dropped the line for a while, but can you just—you're talking about first pivotal study in Q4 next year. So is that going to be small cell lung cancer? And can you give us a view on how that clinical design is likely to look? And will you have biomarkers integrated into that clinical study? So I'm just kind of looking for visibility on what that first phase III program looks like. Then on cobolimab, which I think is down as 2026, has not really been talked about. So should we just see that as a high-risk asset, or are you still very confident about that readout? And then finally, on Blenrep, can you just help me understand how the FDA might deal with the likely dosing recommendations for physicians in the prescribing label? I'm just wondering how that algorithm or how they'll describe how physicians should conduct kind of dosing. If you could help, that would be really helpful. Thank you. Yeah. Super. So Hesham, these are all coming to you, I guess. In terms of small cell lung cancer, obviously, we're not going to disclose the details of our clinical programs, but perhaps, Hesham, you can give some broad line guidance. Similarly for CoSTAR, what I would say for CoSTAR is obviously the second line setting is tough, and we very much view this as an opportunity of looking at dostarlimab in that setting as well. And again, Hesham, you might just describe a little bit about how we're seeing that playing out and indeed the features associated with the study. And then lastly, a brief comment on this prescription proposition. Obviously, we're very early on with regards to label conversation. So broad considerations there only, Hesham. Yeah. No, thank you, Tony. And I'll start off first with B7-H3, and I'll just say we have more than, I'll probably say, one shot on goal to start a pivotal study at the end of 2025. To Tony's point, unfortunately, we can't necessarily disclose which tumor type, but what I can tell you is there's more than one shot on goal. There could be one, possibly two. Stay tuned for that to happen in 2025. The second really around cobolimab, and I believe, probably, Richard, you're referring to the CoSTAR study specifically. This is actually a phase II, three trial that we had initiated a few years ago in second-line non-small cell lung cancer post PD-1, post chemotherapy. With that in mind, what we do know, and this is something that we actually had communicated back in September of 2022, is that the study actually had what was actually gated by certain criteria to move from phase II into phase III. We know that those criteria, when the IDMC, which is an independent body, looked at this data, they basically recommended that both investigational arms in the study actually get expanded and move into phase III. The two investigational arms in the study, of course, are the triplet of cobolimab, dostarlimab, and docetaxel. And then the other, of course, to Tony's point, which is dostarlimab plus docetaxel. And of course, the control arm is the docetaxel. We're just awaiting, of course, now the overall survival data readouts. This is an event-driven trial. So bear in mind that timelines can shift. They can shift upwards. They can shift backwards depending on how fast or how slow events are growing. We have no knowledge, of course, of the data, and we're awaiting the data readout, of course, from that trial. I think what's probably important to highlight in second-line non-small cell lung cancer is that we continue to see certainly a key unmet need emerge in this area. And while other ADCs, and specifically TROP2 targeting ADCs, necessarily haven't demonstrated the success that everyone had hoped for, there continues to be an opportunity for novel agents to be introduced into the space. And then finally, of course, as we think about Blenrep and labeling, of course, we can't necessarily comment on the ongoing regulatory submissions which have been made and accepted in the U.S., Europe, and China and Japan and other key markets and regions. But what we can say is, I think at the end of the day, the dosing and the labeling will be driven by how the drug was administered and, of course, the data that's emerged from both the DREAMM-7 and the DREAMM-8 results. I think probably as you look at other drugs that have been used in oncology effectively, no doubt the dosing decisions being driven by the toxicities that patients experience and how prescribers are able to manage those toxicities on a per-patient, individual patient basis is really important. We've seen that with other labels. I think certainly looking at, for example, the label with palbociclib as an example here, that the ability to be able to, again, titrate the dosing, introduce the dose interruptions, the reductions, and that each patient should be managed individually based on the set of toxicities they experience and how they're responding to treatment as well. And again, still ultimately, of course, a simple framework that is easy to deliver in the clinical setting. And as we've mentioned earlier, that's clearly part of the work that we're doing, but it's a bit premature to get into those conversations. I would say. Okay. I believe we've got one more question left, and since it's Christmas, we should entertain it. We do. Eric. With your final one, Eric. Eric, Look where we go. Your line should be open. Thank you very much for the Christmas gift. One last question then. On the financial aspect, you were talking a lot about prioritizing oncology and respiratory. How should we think about this in the context of the R&D budget going forward? Does that mean that you will prioritize those two over others that will be deprioritized, or should we think about R&D budget going up with those two taking a lion's share? Should we think about any step change in R&D investment going forward to support those two areas? Thank you. Yeah, and look, I'm not going to get drawn on individual capital allocation to specific areas. There are sort of two comments I would make in principle rather, and that is we will continue to fund the portfolio in line with the expected growth of the areas in question, so this is very much a competitive process across the portfolio. As I mentioned in the past, it's done at the highest level at a committee that Luke and I chair together. Long-term, in terms of our R&D budget right now, I'm very well positioned, I would say, in terms of supporting the opportunities that we've described, and of course, we've also mentioned throughout that as the overall company aspirations continue to grow, then there will be an allocation into R&D budget. So for me, the same principles that we've described across the year continue to apply here. And what I would say, though, is that I'm really delighted with the shift in the broad potential value in the later stage R&D portfolio, as I mentioned at the beginning, that we're now starting to see our focus really shift into areas of greater medical need and opportunity. And with that, we can finish the call. It just remains for me to thank you all for joining us on, I know, what is a busy day. I wish you all a restful period over the holidays, and I'll look forward to seeing you in San Francisco if you're there and at subsequent interactions. Thanks, everyone.