AstraZeneca PLC (AZN) Earnings Call Transcript & Summary

Welcome, ladies and gentlemen, to AstraZeneca's deep dive on lung cancer. Before I hand over to AstraZeneca's management team, I'd like to read the safe harbor statement. The company intends to utilize the safe harbor provisions of the United States Private Securities Litigation Reform Act of 1995. Participants on this call may make forward-looking statements with respect to the operation and financial performance of AstraZeneca. Although we believe our expectations are based on reasonable assumptions, by their very nature forward-looking statements involve risks and uncertainties and may be influenced by factors that could cause actual results to differ materially from those expressed or implied by these forward-looking statements. Any forward-looking statements made on this call reflect the knowledge and information available at the time of this call. The company undertakes no obligation to update forward-looking statements. Please also carefully review the forward-looking statements disclaimer in the slide deck that accompanies this presentation and webinar. [Operator Instructions] With that, I will now hand the call over to AstraZeneca's Executive Vice President, Oncology Business Unit, Dave Fredrickson.

Thank you, operator, and thank you to everybody who's had an opportunity to join us today for this investor deep dive into lung cancer and QCS following on sets of presentations and discussions that have been taking place at the WCLC. We've got about 30 minutes of prepared remarks that we'll go through today. The agenda, you can see, is as follows. We're going to spend time talking about our ambition in lung cancer, then really double-click into the work that we're doing to lead in ADC development and really moving antibody-drug conjugates into lung cancer to displace classic chemotherapy with more precise approaches. We'll talk about a second major trend that we are looking to lead on, which is in bringing next-generation IO to build and improve upon what we've seen historically with IO, and then we'll move to a Q&A session that we expect to be about 30 minutes in length. If you can move to the next slide. You can see that Susan and I will handle the prepared remarks, and then we will be joined by Sunil, Matt and Leora Horn in the Q&A. If we move to the next slide, please. And the next slide as well, please. So you'll all be familiar with this slide that Pascal had the opportunity to present at our Investor Day, laying out the ambition that we have as an organization for achieving $80 billion in total revenue by 2030 and really setting ourselves up with platforms that are differentiated and that allow for growth beyond 2030 that is sustainable and meaningful. We start, obviously, at $45 billion-plus in sales in 2023. I think it's important, of course, to note that there's some but manageable and more limited IRA impact and loss of exclusivity exposure that we know that we need to work through. But as we move from there, we have an existing portfolio with significant life cycle management opportunities that we see having the opportunity to unfold between now and 2030. I'd also like to highlight that LCM has often been underappreciated by, I think, investors and the investor community. Examples of this are TOPAZ and HIMALAYA, where I think that the opportunity that we've had to really maximize the number of patients and create shareholder value from those studies, I think, has been very, very important and something that has really allowed for growth on important brands. We have recent also examples like ADRIATIC on small cell, limited stage for Imfinzi amplified with Calquence. And many of our portfolio medicines have opportunities to continue to drive meaningful sales through LCM. On top of that, we have 15-plus significant new molecular entities that have the opportunity to launch and meaningfully contribute to our $80 billion ambition, which is a risk-adjusted ambition that we have by 2030. I Say risk-adjusted because, obviously, not all of the NMEs that we have on this list will necessarily launch. But we think that we've got a very ambitious and sensible opportunity to drive towards the $80 billion and, as we do that, really lead in some of the critical trends that we believe are going to change the treatment of cancer: ADCs and radio conjugates, cell therapies, T-cell engagers, advancing forward within gene therapy, which obviously is highly relevant within rare diseases, next-generation IO bispecifics and then importantly, weight management and risk factors. If you move to the next slide, please. Specifically within lung cancer, you've heard us articulate an ambition that one in two patients being treated with a medicine for lung cancer are being treated by 2030 with an AstraZeneca medicine. And I think that there are 4 established brands that have launched, with Tagrisso, Imfinzi, Imjudo and Enhertu, where we've already begun to establish and show both clinical as well as commercial leadership in advancing this ambition across each of these medicines. To this, we also know that we've got additional ambitions to be able to advance forward further in antibody drug conjugates with Dato-DXd. We'll go into a deeper dive on the studies and the data that were presented at WCLC for Dato. Of course, as well, our two IO bispecifics that are presented here at this World Congress in lung cancer, volrustomig, our PD-1 CTLA-4, rilvegostomig, our PD-1 TIGIT, and really the opportunity to think about these for a deepening response relative to what traditional checkpoint inhibitors have been able to do. And then lastly, while not something that we're going to spend as much time on today, we also have within our pipeline and portfolio our ATR inhibitor, ceralasertib, which has a study importantly, LATIFY, underway to address IO resistance. Next slide, please. We've had strong Phase III pipeline momentum already in 2024. I talked about the life cycle management of all of our existing medicines and how that has continued to develop a steady drumbeat of important readouts that have allowed for us to grow our brands that -- many of which are already blockbusters or multi-blockbusters even further and to use those as the primary workhorse for getting to the work that we need to do to achieve our $80 billion ambition. But you can see this year in lung cancer alone, some really, really important readouts with LAURA, the FLAURA2 OS data, ADRIATIC, and the [ TL0-10S ] data, each of these on their own being very important in terms of being able to help us drive towards that one in two ambition that we have in lung cancer. Additionally, below the line we see outside of lung cancer progress in GI with EMERALD-1, HIMALAYA, progress in bladder cancer with Niagara. And I think Niagara, importantly, when we look at that in the context of also POTOMAC and VOLGA and NILE, represents a blockbuster-plus opportunity. I mentioned previously AMPLIFY, which together with ECHO positions us well to continue to establish Calquence as the leading third-generation TKI. And all of this has just been happening this year in 2024. So if you go to the next slide, with that as the backdrop, I'll turn it over to Susan, who will talk us through a double-click on the work that we saw here at WCLC on advancing our ADC leadership.

Thank you, Dave. Can we go to the next slide? So our strategy for Dato-DXd in lung cancer has 4 points to it. And TROPION-Lung01 is the first step to show efficacy and improvement over well-established docetaxel chemotherapy in the second, third line setting on small cell lung cancer. The next very important step though is to advance novel combinations with our IO agents into both the first line and the early-stage setting. And I think the data that we presented that I'll go into more detail on NeoCOAST-2 is very encouraging in that regard and supplements the data we previously presented from TROPION-Lung02 and 04 to support our ongoing Phase III trials in the first line, which you see listed here. In addition, the combinability profile that we have with Dato-DXd, which is differentiated from other TROP2 ADCs, is enabling us also to build on the strong evidence we already have in the EGFR mutant space with the FLAURA2 data with Tagrisso combination and again was supported by the previously presented TROPION-Lung05 data in patients with actionable genomic indications together with good combinability data from our platform study ORCHARD, and that's supporting the TROPION-Lung14 and 15 trials which are ongoing. The final part of the strategy is really to optimize the patient selection using our quantitative continuous scoring technology, or QCS, and I'll go into the details of the data that show that this is predictive for the progression-free survival in the TROPION-Lung01 and also helps explain the difference that we've observed with Dato-DXd across histologies in non-small cell lung cancer. Lets go to the next slide. I'm going to start with the TROPION-Lung01 overall survival data that Dr. Jacob Sands presented earlier today. These are the data from the overall ITT population with a numerical hazard ratio of 0.94, which did not reach statistical significance in the ITT population. If you go to the next slide. What he did show is that building on the data that we've seen with the co-primary endpoint of progression-free survival and the enhanced response rate that we've seen in the non-squamous population, we've also got a 2.3 absolute increment in the overall survival in the non-squamous population for Dato-DXd with 14.6 months, has a ratio of 0.84, confidence interval 0.68 to 1.05. And what you also see from these data is that we've got no new adjudicated drug-related ILD events or deaths since the PFS database lock. And as we've noted here that the benefit risk in the patient population that is nonsquamous is favorable, with a 1.7% rate of the ILD-related deaths compared to the higher rate that we've seen in the squamous population. Going to the next slide. So now I want to move on to how we optimize the patient selection and -- please go to the Slide 14. So first question is, why isn't standard IHC for TROP2 expression sufficient to predict Dato-DXd benefit? And I think it's very important to understand the protein that we're talking about here, the TROP2. This is a protein that's highly expressed across segments of non-small cell lung cancer. And you see on this plot on the left that this is equivalent to IHC 2+ 3+ levels of expression. And this expression level is both high but also overlapping across all of the response data. And the reason for this is that TROP2 is a highly post-translationally modified protein that actually undergoes cleavage. Not everything that you see and that you can stain on the outside of the cell is relevant to the mechanism of action of an ADC that depends on internalization. So because some of it is cleaved, what you need to look at is the fraction of the overall surface expression that ends up in the cytoplasm. And this also points to the differentiated mechanism of action of Dato-DXd versus other ADCs. What's different about Dato-DXd is it's got a highly plasma-stable linker. This is supported by its half-life of 5 days compared to other TROP2 ADCs that have less than 1 day half-life, and so you're just getting exposure to free payload. Dato-DXd, differently, binds to the membrane TROP2 and then needs to be internalized for its cytotoxic payload to be released. So because not all of the TROP2 protein on the cell surface ends up in the cytoplasm, you need this assessment on both the cell membrane and the cytoplasm to better predict the response to Dato-DXd. So that's what we've attempted to do with the quantitative continuous scoring AI-based computational pathology solution. So let me break this down into pieces. The first thing is that we have an IHC assay for TROP2 developed in collaboration with our partners at Roche Tissue Diagnostics. The second thing is that you then undergo whole slide imaging with a digital scanner. And then there is this fully supervised artificial intelligence algorithm, which does a number of things. First, it's been trained by pathologists to identify the cancer cells on the slide. And you can see that those slides on the -- cells on the left-hand panel at the bottom part of this slide that are green, those are the cancer cells, and this is the algorithm-identified cancer cell versus normal tissue. Secondly, for each of those cancer cells, you can then measure the staining intensity, which is reflected as the optical density, OD, for each dimer cell, and you can do that for each pixel in the membrane and each pixel in the cytoplasm. What the algorithm has then done is calculate the ratio of the membrane to the membrane plus cytoplasm optical density. And because this is a ratio, what we're looking at is the highest cytoplasmic expression gives you a lower normalized membrane ratio. And so the biomarker cut point has then been determined from the TROPION-Lung01 data set. There's more than 75% of the tumor cells that have this normalized member ratio, NMR, of less than or equal to 0.56. So if you go to the next slide, this normalized membrane ratio is just illustrated again on the left-hand side. So for the higher cytoplasmic staining in the context of non-small cell lung cancer where all of the cells have high membrane expression, the lower ratio is what is predicting the higher response to Dato-DXd. This is because Dato-DXd must be internalized to release its cytotoxic payload, which I showed you on the previous slide. So on the right-hand side, you show the data from the ITT population and how we got to the biomarker-evaluable population. The biomarker-evaluable patient population is all of the patients with evaluable TROP2 QCS-NMR results. And in TROPION-Lung01, we didn't require tissue to be analyzed for this at the beginning of the study. The optimization of the biomarker was then optimized based on the non-squamous non-AGA patient population, which is the largest part of this subgroup of the ITT population. We go to the next slide. Just looking at the prevalence of this biomarker of the non-squamous patients, you can see from this slide that 66% of the non-squamous patients were biomarker-positive, 62% in the non-squamous non-AGA, 76% in the AGA subgroup and a smaller number, 44%, of the squamous population. Going to the next slide. This is the -- now the Kaplan-Meier curve for TROPION-Lung01 in the biomarker-evaluable patient population in the non-squamous non-AGA subgroup. Let me draw your attention to a few points. First of all, we've got a response rate of 36.8% in this biomarker positive group versus 15.3% with docetaxel and higher than in the biomarker-negative group. Secondly, you've got a median PFS of 7.2 months, which compares to 4.1 in the biomarker positive for docetaxel and compares to 4 months for Dato-DXd in the biomarker negative. The PFS hazard ratio in the biomarker positive is 0.52. We go on to the next slide. This is now in the overall biomarker-evaluable population, which includes the squamous patients as well as the AGA patients that are biomarker evaluable and very similar data with a 32.7% response rate in the biomarker positive, median PFS of 6.9 compared to 4.1 months for docetaxel and a PFS hazard ratio of 0.57. So now I'm going to move on to the neoCOAST-2 data. So this is a neoadjuvant platform study which has enabled us to look at a number of different next-generation IO combinations. Combination with oleclumab plus Imfinzi plus platinum-doublet chemotherapy in Arm 1, monalizumab plus Imfinzi plus platinum-doublet chemotherapy in Arm 2, and in Arm 4, Dato-DXd plus Imfinzi. The volrustomig plus chemotherapy arm is not fully mature yet and will be presented at an upcoming congress when those data are mature. Primary endpoint of NeoCOAST-2 is pCR rate and safety and tolerability with secondary endpoints of major pathologic response rate and event-free survival. Obviously, the numbers in these cohorts are small for event-free survival, but also importantly, feasibility of getting to surgery and the completeness of surgery were important secondary endpoints. Go to the next slide. What we've seen in the Arm 4, which is what I'm going to concentrate on for today, is that we get a very encouraging pCR rate of 34.1% in the Dato-DXd plus Imfinzi plus platinum chemotherapy. And also very encouraging, major pathologic response rate of 65.9%. Cross [indiscernible] comparisons are always difficult, but just for your reference, you've got the AEGEAN data set for Imfinzi plus doublet chemotherapy there. What we also observed in this study is higher pCR rates with increasing PD-L1 expression, as you can see on the right-hand side. And as the presenter Dr. [ Gasconi ] reminded, these are relatively small numbers in these subgroups. So whilst the trend is of interest, the absolute numbers are to be taken with caution. If you go to the next slide, what you can now see is the waterfall plot looking at the change in tumor area with viable tumor cells on the Y-axis across here, and they're color-coded. Those patients with pathologic complete response were in light blue, with major pathologic response in dark blue and with non-pCR/non-mPR in the maroon color. And you can see you've also got underneath indication for PD-L1 stage and histology that are color coded. What was also encouraging in this data set is that we had a low rate of grade 3, or more treatment-emergent, adverse events of 24.1%, treatment-related adverse events of 18.5% in the neoadjuvant segment and a low rate of adverse events leading to discontinuation. Any SAE with outcome of death was 0 in the neoadjuvant phase. And for the 1 patient that [ saw this ] this post surgery, this was a patient who had preexisting colmonofibrosis, which was in fact an exclusion criteria. So this was a protocol violation for inclusion of that patient on this study. These are very encouraging in relation to the other cohorts in this study and historical IO plus doublet chemotherapy trials. If you go on to the next slide, the NeoCOAST data for efficacy and safety go together with the TROPION-Lung02 and TROPION-Lung04 data sets that we've had from the first-line setting of metastatic non-small cell lung cancer that you can see summarized on the left-hand side of this slide. And I think, together, this adverse event profile is different for Dato-DXd compared with other TROP2 ADCs. Because it's got a plasma-stable linker, we have lower bone marrow toxicity, which is something that correlates with exposure to free payload. So this lower bone marrow toxicity enables us to combine with platinum chemotherapy and lead to these encouraging pathologic complete response and overall response rates. It's a convenient dosing, which aligns with chemotherapy, one IV infusion per cycle, and we have now several first-line trials ongoing with ongoing data monitoring committees, which have not seen any significant safety concerns with ILD. So in terms of the summary of the overall learnings, we feel that TROPION-Lung01 final OS data are consistent with the previously reported response rates and progression-free survival with an absolute clinically meaningful improvement of 2.3 months versus a well-established chemotherapy of docetaxel in previously treated patients. The QCS-NMR as a biomarker supports the potential of TROP2s measured by QCS as a predictive marker for Dato-DXd and helps to build confidence in the AVANZAR and TROPION-Lung10 first-line patient populations. We also have a broader ambition to employ QCS across our ADC portfolio with the goal of developing these predictive markers to enhance patient selection across our portfolio. And finally, I think the NeoCOAST-2 data are very exciting and reinforce the confidence in the first-line non-small cell lung cancer alongside the data that we previously presented from the first-line Phase II studies. And we think, together, these demonstrate the potential for the combination of Dato-DXd plus IO and chemotherapy in patients with early-stage non-small cell lung cancer. So with that, I'm now going to move on to our next-wave IO bispecifics. And just as a reminder, we have 2 bispecifics where we're presenting some data at this congress, and those presentations are happening just after this call. So I'm going to stick to the data that has been revealed in the abstracts predominantly. Cartoons on the left show the design of both of these molecules. And I think what we're starting to see across bispecific next-wave IO molecules is the opportunity to have this combined cooperative binding of 2 different targets that can lead to different biology. Both of these have been designed with the relative affinities of the PD-1 or the TIGIT or the CTLA-4 so that we get this coordinated binding. Rilvegostomig, in addition, is Fc-attenuated to avoid unselective depletion by Fc-mediated antibody drug ADCC killing within the tumor microenvironment. And again, volrustomig is designed to fully inhibit PD-1 while preferentially inhibiting CTLA-4 only on the activated T cells in the periphery. We have with both of these bispecifics a robust Phase III program with ongoing Phase III trials that you can see illustrated on the right-hand side of this slide. Next slide, please. So the data that's going to be presented is from the ARTEMIDE-01, which investigates rilvegostomig in the first-line and second-line setting of non-small cell lung cancer. And these were data that we presented at ESMO 2023 as a reminder, showing the spider plot on the left-hand side, and again, low rate of treatment-related discontinuations and very well tolerated, which I think is important for this TIGIT molecule and potentially differentiating from some of the co-formulations that you've seen. Go to the next slide. ARTEMIDE-01 demonstrates that rilvegostomig was well tolerated in a checkpoint inhibitor-naive lung cancer population. So of this patient population, we had 16.7% had prior chemotherapy for metastatic disease, 13.5% had liver metastases and nearly 22% had brain metastases. Again, a low rate of grade 3 or more treatment-related adverse events and a low rate of treatment-related discontinuations and no difference in terms of the safety profile between the 750 and the 1,500 milligram safety profile. In terms of efficacy, we're encouraged by the response rates in the checkpoint inhibitor-naive setting. So you've got here separated out on the left-hand side the 1% to 49% PD-L1 expressers, with a 30% to 35% rate of response in this setting, which is very encouraging, and even stronger response rates, of course, in the PD-L1 high expressing of more than 50% that you see on the right-hand side, 62% to 68% based on local or central assessment. Importantly, these responses are highly durable, and at the data cutoff, nearly 54% of patients have remained on treatment with a response ongoing in 80% of the confirmed responders, which is encouraging. Next slide. Now going to move to the volrustomig data, again starting with a reminder of the data we presented at ESMO 2022, where we showed strong [ efficacy ] in combination with chemotherapy, particularly in the PD-L1-negative non-small cell lung cancer, and you see the waterfall plot here and the response rates of 44% in that PD-L1 less than 1% population. Overall, we saw that the safety of volrustomig at 750 milligrams plus chemotherapy was improved versus the 1,500 milligrams plus chemotherapy, and that's the dose level where we continue to evaluate volrustomig. Next slide. So what we're now showing is that a large proportion of patients that are enrolled have been in this less than 1%, which reflects again that unmet need in the patient population. So again, in this patient population that we're presenting, 74% male, 68% ECOG performance status 1, 15.7% with liver metastases and 15% with brain metastases. In terms of all-grade treatment-related adverse events, there's been a learning that we have done through this trial in terms of the inclusion/exclusion criteria, the grade 3/4 treatment-related adverse events is higher than we see with rilvegostomig at 76%. And in the initial cohort, Cohort 1A, there were 7 treatment-related deaths, 2 of which were attributed to volrustomig. But based on the learnings that we've had, we've improved both the treatment management guidelines and the inclusion/exclusion criteria and encouragingly in Cohort 1b, 54 patients, we've had no of those treatment-related deaths and 1 in the Cohort 2, which is [ out ] of 20 patients. Go to the efficacy now. What we've seen is promising clinical activity, particularly again in that patient population with high unmet need of PD-L1 less than 1%, a response rate of 44% and the disease control rate of 85% in the non-squamous population on the left, with a good sample size of nearly 120 patients. And on the right, in the squamous population, smaller numbers, 65% response rate and 95% disease control rate. And then if you take that number and look at just the patients with the less than 1% PD-L1 on the right-hand side, again, those numbers are 42% and 50% in the non-squamous and the squamous population. What we've also seen in terms of the translational medicine is greater T cell proliferation and memory T cell activation with this when we compare with an anti-PD-1 plus chemotherapy regimen. So on the next slide, just to summarize the totality of the data, I think the bispecific design allows for this coordinated synchronized PD-1 and TIGIT blockade for rilvegostomig with the Fc-attenuated reducing the potential for some of the toxicity and depletion of T cells within the tumor microenvironment. And volrustomig allows for the optimized safety profile compared with co-formulation of CTLA-4 with a PD-1, so that you're only getting CTLA-4 blockade in the presence of PD-1. What we've shown in this data is encouraging response rates and progression-free survival for both rilvegostomig and for volrustomig. And we have learned on the safety profile of volrustomig about the manageability of this. We have multiple ongoing Phase III trials in both settings and, again, close data monitoring committee involvement with that, which is going well. So I now want to move to the final slide and hand back to Dave Fredrickson, who is going to summarize our overall presence here at World Congress on Lung Cancer. Thank you, Dave.

Thanks, Susan. Appreciate you going through all those data. If we could go to the next slide. We've spent quite a bit of time with all of you over the last many months articulating our approach to attacking cancer in general, but lung cancers more specifically, from multiple angles and through novel combinations, which really leverage the pioneering work that we are doing to build winning and differentiated platforms that we think will meaningfully transform the way that we can treat this disease. At this year's WCLC, Susan has just gone through and nicely summarized progress with our bispecifics, progress with our antibody drug conjugates, progress in moving into earlier lines through combinations and how we're advancing all of our efforts that we are making meaningfully within our efforts to scientifically and clinically differentiate our portfolio. I would also say that I think that it's worth pointing out that with Tagrisso, we also -- while we weren't presenting any specific Tagrisso data, we did get an opportunity to understand the competitive landscape. And I think within that context, we continue to believe that Tagrisso will maintain its position as the clear backbone standard of care for patients with EGFR mutated disease across all stages, but including and specifically within frontline advanced disease. I think that we heard very compellingly from the podium the importance of continuing to have monotherapy as an important option for many, many of the patients, but also the different toxicity profiles will be important as clinicians consider how they want to think about treating those patients who are most relevant for combinations. We also hear from the podium a real assertion that in some of those patients, like patients with brain mets, the importance of the Tagrisso data that is emerging and is very strong with FLAURA2. Finally, on the commercial side we've seen, that with great data that we are presenting and having an opportunity to share, that we're able to convert that into patients in the real world that are being able to benefit from those treatments. And that's through the efforts that we've got to deliver successfully companion diagnostics in the real world and to be able to really deliver on the promise of precision medicines, along with efforts that we make to really ensure that we're working with multidisciplinary teams and transforming the way in which care teams are considering our data sets and able to drive towards best possible outcomes. Next slide. As we think about the paths into 2025, we've got a strong cadence of oncology readouts that are anticipated through '25. We've got 4 pivotal trials in lung cancer with AVANZAR, SAFFRON, which is an opportunity to look at combination with Tagrisso and savolitinib in the second-line context, continued work with Enhertu, and I had alluded to before the LATIFY study that's looking at a ceralasertib opportunity in the post-IO non-small cell lung cancer. Of course, there's many, many other things happening beyond lung cancer. You can see those here. Again, I spoke about the efforts being made in bladder cancer, POTOMAC and VOLGA, collective efforts being made across multiple studies to move Enhertu into earlier HER2-positive breast cancer settings, 3 really important studies with 09, 11 and 05 reading out, Imfinzi into GI with MATTERHORN, and then importantly, an opportunity with camizestrant to hopefully be able with SERENA-6 to have a chance to get our first positive readout, which would be, of course, then leading to the opportunity for the first approval for that medicine if positive and we're able to have the right conversations with health authorities. Lastly, before I turn it over to Q&A, if you can go to the next slide, please. One quick reminder that in Barcelona for ESMO, we will have an event to meet the AstraZeneca management. It will be on Monday, September 16, and that will be at 20:00 hours Central European time. So with that, I'm going to turn it over to our Q&A.

And our first question that we have is from Steve Scala at TD Securities. Steve?

Two questions. Curious on TL01's lack of nominal significance on OS and what does that imply for its approval. And secondly, apologies if I missed it, but what does TL01's OS look like in the QCS biomarker-positive subgroup?

Susan, can I suggest that those are two good questions for you to tackle?

Yes. So thanks for that. Again, what we've shown is that the -- with the OS data that we've got a clinically meaningful trend to improvement with 2.3 months median improvement, and that's just very consistent with what we have in the primary end point. Because this is not alpha-controlled -- alpha analysis, the data is going to be considered in the context of the overall benefit risk. Discussions with the FDA are ongoing. We don't have an update to share with you other than what we've presented in terms of the OS on that front. In terms of the QCS OS, obviously we just presented the final analysis of OS. We have previously looked at OS on the interim analysis, similar trends that we've seen that we presented with the PFS, but we'll obviously also then present the OS data at an upcoming congress. And we'll share with that with you as soon as we have that available.

Next question is from James Gordon at JPMorgan.

James Gordon, JPMorgan. Two questions and a clarification, please. The first question was on the TROP2 biomarker testing. Just how practical this would be in the real world, like if you could contrast for a physician or a pathologist how much more onerous would this be than, for instance, doing a HER2 test? I know in the presentation yesterday, there was a comment about maybe like differences versus if this was done centrally versus individual labs. So what is actually done centrally versus done at local labs and how much more onerous would this be? Second question was just a -- or clarification. In the presentation yesterday, there seemed to be a focus in the non-AGA population. And that's where you seem to have optimized the TROP2 biomarker. But can you confirm the [ 201 ] filing is still looking for approval in AGA and non-AGA patients? And do you still see a role for Dato in patients with AGA? So like I know you're doing TL14, which is, I think, just in the patients with EGFR mutations. So is that still something where you're confident? And then maybe third and finally, so you now have got the final TL01 OS, and given what we know about the upper confidence interval and the magnitude of the benefit, does this change your confidence in approval versus the data we had at ESMO last year on OS? Does it make you any more or less confident?

Okay. So thank you, James, for those questions. And why don't I again hand over to Susan. I would say that on the first question, we could also, Susan, invite Sunil to talk about the work that we're already doing to understand kind of work that we would need to do to implement a QCS biomarker in the real world and then also the two other questions that James...

Yes. So thanks for asking the question. So first of all, we announced in the press release that we have a development and commercialization agreement with Roche Tissue Diagnostics for this test to be used on the navify digital platform that they have. So the vision is that we would be enabled to use this as a centralized test where simply 3 slides can be shipped and get a biomarker positive or negative read back using that system. Obviously, the plan will then be to think about how we then get to decentralized testing. And I'll ask Sunil to answer that and then I'll come back and answer the other two questions.

Thank you, Susan. I think, James, as you can imagine, the lung cancer clinical community is really excited about this data, as was noted by the discussant and as well as the Q&A. They have been pioneers in personalized medicine, and they see the value of this in bringing personalized medicine to the field of ADCs. So as Susan mentioned, the initial aim is to build these centers of excellence to have this available with [ stained ] -- this would be IHC stained locally and then read centrally. And I think that's very feasible, that's very doable. And then, of course, then our aim is to make it more scalable to have local testing and local measurements done. And I think that pathway is what we are building and the road map is where we're building. But our goal is by day 1 to have the central labs ready to go.

Okay. So let me address your second question -- thanks, Sunil -- on the non-AG versus AGA. Obviously, in the actionable genomic altered group, we've got 2 data sets now. We've got the TROPION-Lung05 Phase II data set and then the subset from the TROPION-Lung01 data set, where again, the hazard ratio is very encouraging for both PFS and OS. And the response rate is actually similar to in the QCS positive patient population with a sort of high 30% response rate in that group. So I do think that that is a segment that is of continued interest. And as I say, that supports our TROPION-Lung14 and 15 Phase III trials that we've got in combination with the fact that we've got the safety profile that is highly combinable with Tagrisso that we have from the ORCHARD data set. So the reason though that we optimized the QCS on the non-AGA, subgroup of the non-squamous is because that's the biggest patient population. And again, in terms of the biomarker-evaluable numbers, we've optimized it on that basis. There clearly is already a biomarker for the AGAs, it's the genomic alterations. So I think it's very practical to think about them as separate populations in that way, but it doesn't reflect a lack of interest in the AGA subgroup. And then in terms of the third question about the approvability, again, I'll go back to what I've said, overall we feel that in the non-squamous population, which this difference between histologies now has the difference in the biomarker prevalence to help explain. And then we've seen that not just in the TROPION-Lung01 patient population, we also saw it in the TROPION-PanTumor01 data set. It was also seen in terms of a difference between nonsquamous and squamous benefit in the ICARUS study, which was presented at ASCO. This is a single-arm externally sponsored research study run by the Institute of Gustave Roussy in 100 patients. They also saw that. And we also saw it in the China data set which was presented at ASCO. So 4 different data sets that have seen that difference in activity in the -- with better activity in the non-squamous versus squamous. And also, this difference in prevalence that we have observed that we reported in TROPION-Lung01, we've also seen in other internal lung cancer data sets such as the POSEIDON and the NEPTUNE data set that we have available. So I don't think this is a random chance event within the TROPION-Lung01. I think it's a consistent biological event. And it's, again, different for Dato-DXd versus other TROP2 ADCs because of the difference in the highly plasma-stable linker that we have with the Dato-DXd design. So I think we have to put all of these together to really understand that difference. And within the nonsquamous patient population that we presented, we've got consistency in terms of response rate, progression-free survival, and now the overall survival trend that we have seen, and that was the basis of the filing with the FDA.

Thank you, Susan. And I think that James, while maybe not exactly your question, I think it's an important follow-on, which is, as a result of the data sets that we have been able to see here at WCLC, we have building evidence base for confidence for data in the frontline setting. And I think this is a really important piece. We're understanding more about histology, as Susan laid out, with consistency around PFS response rate, overall survival within the non-squamous. We add to that the opportunity for an additional shot on goal to build into the AVANZAR study with the biomarker QCS population, which gives us, in addition to the other 4 opportunities that we have with frontline studies in 07, 08, AVANZAR now, we've really got a very nice opportunity to be able to incorporate the biomarker into that study. And then, of course, we also know that the combinability that we see out of NeoCOAST-2, together with the TL02 and TL04 data that we've presented in the past, we can also see that other ADCs don't appear to have the same combinability with them. All of this together creates opportunity for building confidence in the program more broadly in the front line, which is where really an enormous amount of the opportunity to displace chemotherapy lies and the value lies as well. Sachin Jain, Bank of America. Over to you, please.

Sachin Jain, Bank of America. A few questions, please. First just to follow up on the last one. I just wanted to clarify, is the TROP2 biomarker part of the FDA debate for TL01? And my question is given the lack of ARS benefit in the non-AGA population, where the hazard ratio was just 0.89. And then two questions just to follow on from your commentary on front line, if I may, Dave, and this may be for Susan. So firstly, on AVANZAR, QCS was being developed while AVANZAR being recruited. You sort of referenced there's an opportunity to build into that study. So I just wanted to check what -- where you were in the process of having TROP2 as a primary endpoint? Is that biomarker agreed with the FDA? And will you have data on QCS for the majority of patients [ within ] AVANZAR? And then the second question is a question I've had a lot from investors today, which is how do we think about the correct population for [ data ] in frontline, i.e. how broad within non-small cell lung. The questions incoming are off the tilt of we should be thinking about non-squamous non-AGA TROP2-positive, which is really sort of 40% of the total population. So I wonder if you could just comment on that.

Sachin, could I just ask you for -- maybe Susan caught this, but for my benefit, could you reask your first question? The other two are clear to me, but if you wouldn't mind on question one.

Yes, apologies. Is the TROP2 biomarker data you presented for the non-squamous non-AGA part of the FDA debate given the lack of OS benefit in the total non-AGA population, I think you referenced a hazard ratio of 0.89, with a 1.3-month delta in OS.

Great. Thanks. So Susan, I do think those are 3 questions for you, and perhaps Leora can also talk about some elements as well within the third question that Sachin just asked around the frontline and connecting it there.

Okay. So the intent with the development of the QCS biomarker is that this will be help us to optimize the patient selection within the first-line study and [ requires ] prospective validation in the first-line study. I think it does help to explain the observation of the differences by histology that we've seen in TROPION-Lung01. And in terms of your second question about AVANZAR, we have made sure that we've access to biomarker-evaluable tissue from all of the patients that are in AVANZAR, so we would expect that we have a high proportion of biomarker-evaluable for QCS on that point. And we are intending to look in both the ITT and the biomarker-positive group within AVANZAR. And in terms of the third question, I'll start, and maybe, Leora, you can give some context on that, about how do we think about the overall patient population? So again, I think in that first-line setting, there are -- we have taken the learnings from TROPION-Lung01 into AVANZAR and enriched for the non-squamous population within that and [ capped ] the number of patients that are squamous patients within the AVANZAR design. So I think we've taken some of those learnings and done that. But I do think there are -- there's a biomarker-evaluable subgroup within the squamous population that will be -- that we're also, over time, have the opportunity to look at. Leora, do you want to give some context on the overall what you think about the percentage of the first-line patient population?

Yes. So I don't think that we believe that TROP2 expression levels change with line of therapy. And so the data that we're seeing from TL01 in the non-AGA population, approximately 2/3 of patients are TROP2 biomarker positive. And so as Susan mentioned, while TL01 was not designed to have the TROP2 QCS biomarker evaluated, we will have all the patients in AVANZAR who are evaluable for this biomarker. And then also thinking about the frontline population, we also have a first-line study TROPION-Lung14 with Dato-DXd and Tagrisso in the EGFR mutation population. And this data from TL01 also shows us at about 75% of the AGA population are biomarker positive. So it's actually a significant number of the nonsquamous non-small cell lung cancer patient population that will be evaluated, and then we'll get more information about the squamous patient population as we get a larger sample set of the subgroup.

Thanks, Leora. I think, Sachin, just again coming back to this and maybe to summarize, I think that we see we know non-squam is 70% of the population. We know that what we've seen, at least with the QCS data, is that a biomarker is the best approach to be able to take to identify those patients if you've got 2/3 of those. We also, though, know that in the AGA population, and we saw this with the data that was presented from the overall survival in 01, that there's a very good response in the AGA population as well just within AGA and in an un-biomarker, if you will, selected way. So I think that we still have a considerable number of patients that we see as being suitable for Dato, especially in the frontline setting, and we've got multiple studies that are underway to be able to really understand that, and we've incorporated the lessons that we've learned on histology, we've incorporated the things that we're seeing around combinability, and I believe that the more that we're able to identify the best patients who can benefit, the things that we'll get out of it are faster uptake, longer durations of therapy, across the globe almost certainly better prices as we can identify those benefits. And so I think that we've got collectively here quite a few data points that lend us to the belief that there's a large number of patients that remain suitable for Dato. With that, Richard Parkes, please. Please go ahead with your question.

So yes, two questions. First one, I just want to explore the likely eligible patient population in first line in a slightly different way. So when we look at the squamous patients, I think there were still 44% of those that are biomarker positive. Do you think that accounts -- fully accounts for the lower efficacy seen in those patients? Or is there something else associated with squamous biology that makes Dato-DXd less effective? Again, it just ties into thinking about whether Dato is a drug for biomarker positive patients overall or biomarker-positive non-squamous patients? So that's the first question. Then secondly, on the survival trend in TL01, obviously, you saw more than 35% improvement in the non-squamous population, but a more modest benefit on survival. That doesn't seem to be accounted for by confounding by post-progression treatments, given there are limited options for those patients. So I just wonder why you think the improvement in PFS only led to a more modest overall survival benefit in that population?

Thanks, Richard. Susan, I think those are two questions for you, please.

Okay. So in terms of the potential for activity within a biomarker-positive subgroup of squamous, I think there is potential for that. It's possible that the optimized cut point across histologies might be different. And I think we need a larger patient population to further optimize that, because that was a relatively small proportion of the overall TROPION-Lung01 patient population. And when you look at the biomarker-evaluable numbers, I don't think it was possible to optimize that. But I do think that there is a potential for this. I do think that if you look at the overall benefit risk in the squamous, it's not fully explained by the difference in prevalence of the biomarker. As I've indicated on the slide, there was a higher rate of ILD in those patients, including deaths from ILD. And obviously, that's something that is important. Now I would say a couple of things. Just as we've done within Enhertu, as we've gone from the initial trials we have learned about the treatment management guidelines and were able to educate physicians and patients better. That's point number one. As you go into the earlier lines, you also see a lower rate of some of the comorbidities that predict for ILD. And again, I would point out that the TROPION-Lung02, TROPION-Lung04 and NeoCOAST-2 data sets together are very encouraging for the overall safety profile of data in those earlier lines of treatment and we now have hundreds of patients that have been treated in the Phase III trials with ongoing data monitoring from that perspective. So I do think there's more than one factor that affects the overall benefit risk profile in the squamous population, but I do think it's going to be possible to identify a subgroup that can potentially benefit with the technology, but we probably need a little more time to fully develop that. In terms of the OS question about the post-progression, you're right, that I don't think there was a lot of confounding difference in terms of the post-progression. But remember, this is a relatively heavily pretreated patient population, And I think, again, as you go into the earlier lines, the potential for overall survival benefit in combination is greater.

Okay. Now I'm going to turn over to Peter Verdult at Citi. Peter, please go ahead with your question.

Peter Verdult, Citi. Just two questions. Just some more clarifications on AVANZAR. Susan, when you say you've enriched the population, should we assume, given the natural etiology, that [ 80% to 90% ] of the patients involved are going to be non-squames. And how [ fully ] are you actually going to be testing [ NMR ] TROP2 QCS -- [ NMR ] status during the trial? Is it a selection criteria or something you will set at trial end to interrogate the primary end point? And then secondly, just more big picture on the competitive landscape in non-small cell lung cancer. Just your thoughts on the [indiscernible] data, PD-1 that measured the -- sorry, VEGF bispecific data versus KEYTRUDA first-line lung cancer. Impressive, or would you caveat that, that data was only generated in Chinese patients?

Great. Peter, thanks so much. So question on AVANZAR are and a question on landscape as it relates to the VEGF bispecific. Susan, take both of those, please.

Yes. So given the rapid accrual for AVANZAR and the data that we've now got from TROPION-Lung01, we are, as I've said, using an ITT and a biomarker positive design for AVANZAR. So this is something that we can look at retrospectively. And as I've said as well, we've taken the learnings from TROPION-Lung01 and enriched for the nonsquamous patient population and kept the squamous numbers. So I can't reveal the total percentage today, but we have enriched based on the learnings of TROPION-Lung01. But I would say that we have multiple data sets also available to optimize the QCS biomarker, including the TROPION-PanTumor01, the TROPION-Lung01 and other data sets as well, that help build our confidence in the ability to select the patients and improve the overall benefit/risk profile in the first-line trials. So in terms of the ivo data that was presented, again, I think, first of all, I think the data look encouraging. As the discussant mentioned, obviously the standard of care in the 1% to 49% PD-L1 positive in the -- is something that would require potentially a different control arm in that group of patients. And obviously, I think it's also true to say that regulatory authorities may want to look at data sets that are not just accrued in China, but are more broadly more representative of the patient population for which the indication will happen. So I think both of those things need to be taken into account. But I do think that this idea of coordinated biology of bispecific molecules is something that we are also seeing with our bispecifics. And I do think that's an interesting observation, and I think the data sets are encouraging. I would note that we are combining rilvegostomig with VEGF bevacizumab in our ongoing GEMINI Phase II platform studies in hepatocellular cancer and in gastric cancer. So we've got good combinability and we have the opportunity then to optimize for both PD-1 TIGIT and the VEGF biology within that context.

Next question is from Emily Field at Barclays.

I'm going to ask one annoying one and then one that I hope is far less annoying. Firstly, you previously said that, based on your discussions thus far with the agency on the TL-01 filing, that you had not expected an ODAC meeting. Can you confirm that that is still the case? And then how close to the PDUFA, practically, could they ask for one? And then secondly, on rilvegostomig, we have seen a lot of failures in the TIGIT space, most recently with SKYSCRAPER-06. Some have suggested that also with anti-TIGIT that perhaps looking for a biomarker would help in terms of finding patients that could better respond to these therapies. Is that also work that you guys are engaged on as you're moving that asset forward?

Maybe just on the first question, Emily, what I'd say here is that we -- as we've said repeatedly, we don't have updates to provide on TL01 at this time. What we do know, based on the past and historics, is that FDA can ask for an ODAC pretty close to the end of all PDUFA dates. We've certainly seen that experience with, for example, on PROpel. So I mean, I think within this context, it is one we'll give updates when we have those. But yes, the FDA reserves the right to be able to go into those advisory committees at any point in the process within this. On the second question, Susan, over to you and to Matt.

Yes. And I have Matt to help -- to make a comment. But the whole design of the clinical development plan for rilvegostomig is already based on trying to develop this, particularly in the higher expressing PD-L1, which is the biomarker that is generally available at the moment for the IO checkpoint inhibitors. There are obviously ongoing efforts to further improve on the patient selection criteria. But we think that in that patient population based on the data that we've seen and that we're actually presenting at this meeting, we have very encouraging data in both the 1% to 49% PD-L1 and the greater than 50%. But the PD-L1 less than 1%, I think, is a space that is better reserved for -- this is the patient population where PD-1 and TIGIT on their own don't work as well, even in combination with chemotherapy, and that's where there's the unmet need and the need for something like the PD-1 CTLA-4 to add benefit. But Matt, do you want to add any color?

Anything I would add in addition to what you've highlighted about our focus on the PD-L1 expressing population as well as the efficacy that we find encouraging is that the differentiated molecule may also have an opportunity to differentiate on safety. We're very encouraged by the very clean safety profile, with very few immune-related toxicities or discontinuations, which I think gives us a distinct opportunity for combinability, which we're bringing into our Phase III studies.

Thanks, Matt. Thanks, Susan. Okay. Let's turn to the last question that we'll take for the day from Luisa Hector at Berenberg. Luisa, please.

I wonder if you could just touch upon the potential scenarios for the FDA decision on TL01? And I wonder whether the biomarker data will play a role in the decision. So when we look at the TROP2 negative data, the hazard ratio was [ 1.22 ]. So does that have any bearing, and maybe you could just outline where you see the potential kind of outcomes of that approval? And then on the biomarker. So it's been validated retrospectively in TL01. What are the risks that you see in using that prospectively? Or can we assume that that will just kind of yield similar results as you use it prospectively and the front line is the test still learning and improving. And do you expect the biomarker to be specific for use with Dato-DXd rather than any other TROP2 ADCs?

Thank you, Luisa. Maybe -- Susan, I'll start on the first question and then turn it over to you to add anything you want on the first, and we can move into the second. I think Luisa that we filed with FDA in the non-squamous population. And that's the discussions that we're continuing to have with FDA. We don't have any further updates at this particular time. I think that what Susan did speak to in her remarks that she gave that the biomarker certainly helps to begin to potentially elucidate some of the biology that we are seeing that sits behind histology. And I think that that's an important contribution that it's making to the overall view that we have on Dato. And we'll have to continue to -- just our conversations with the agency, and we'll provide more updates when we have those. Susan, if there's anything more that you want to add on that, but -- please feel free. But otherwise, also please take on the second question.

Yes. So the second question is about the risks of using this prospectively, but that's the intent in terms of validation within the first-line studies is to exactly do that testing. I would say again that we have seen that this particular feature and biomarker is something that is helpful for Dato, not just in the TL01 but in other settings that we're looking at as well. So there is more data beyond the TL01 data set that we will obviously be looking at as well. And then in terms of the specificity for use with Dato-DXd, I've emphasized in the prepared remarks that I made and I'll emphasize again there, I think Dato-DXd is differentiated because of its highly plasma-stable linker, and therefore a higher proportion of its activity is dependent on the internalization and then the subsequent cleavage within the cytoplasm of that linker and release of the payload. That is different from something where you've got a very short half-life, a nonstable linker and just basically a lot of high exposure to free payload. And you see this not just with the half-life, but with the difference in the bone marrow toxicity that you see. Bone marrow toxicity is driven through exposure to free payload typically. So I do think that the biomarker that's related to that internalization fraction is going to be specific for ADCs that have that plasma-stable linker type and the cut point has obviously been optimized for Dato-DXd.

Thank you very much, Susan. So with that, I want to bring this session to a close. For us, this has been a very exciting WCLC, an opportunity to really showcase some of the important data through a pair of presidential symposia that are advancing our ambitions in lung cancer together with the presentations that we're looking forward to at ESMO. Hopefully, we begin -- or we continue to give an opportunity to see the breadth of the pipeline and the steps that are being made and the progress that's being made towards driving our $80 billion ambition that we've set out for 2030 and the opportunity for sustainable growth beyond. We've got exciting data coming up in congresses in the next 12 months, and we invite you to join us, as I said before, at the Meet the Experts Meeting at ESMO. And with that, we'll bring the call to a close. Thank you very, very much.