Roche Holding AG (ROG) Earnings Call Transcript & Summary

Ladies and gentlemen, welcome to Roche's virtual event on key oncology data presented at 2021 ASCO annual meeting. My name is Henrik and I am the technical operator for today's call. Kindly note that the webinar is being recorded. [Operator Instructions] At this time, it's my pleasure to introduce you to Karl Mahler, Head of Investor Relations and Group Planning. Karl, the stage is yours.

Yes. Thanks a lot, Henrik. Welcome to our fourth science call this year. We have reserved 1.5 hours. So half of the time will be for Q&A and half of the time for presentation. So we all are some experts in the meantime, as I know. [Operator Instructions] So very much looking forward to an interactive event. For this year's ASCO, we have chosen a bit different format, I have to say. So first, we will focus on exciting data presented at this year's ASCO for obvious reasons like the IMpower010 or any kind of updates on the bispecifics. Second is, we also wanted to give you an update on our exciting programs in oncology for the early stage. So I think this is sometimes really overlooked. We have pRED and gRED representatives with us today. And last but not least, we'll also give you an update on the exciting program in late-stage Phase II/III, the pivotal development. So basically, all which is really very close to the clinical regulatory approvals. This is the program for today. We have William Pao, the Head of Roche Pharma Research and Early Development with us. He will focus on the pRED Assets. Ira Mellman, Vice President, Cancer Immunology, Genentech Research & Early Development. He will represent the exciting stuff from gRED. Then we have Levi with us, the Chief Medical Officer and Head of Global Product Development. And last but not least, I have to say, Heather Wakelee to represent the IMpower010 primary results. She is a Professor of Medicine at the Stanford University, and she's also Deputy Director in the Stanford Cancer Institute. So next slide, please. Yes. So reflecting on the quality of research development at Roche, you can see that we have 37 breakthrough therapy designations overall since 2013. So this is the largest number of breakthrough therapy designations, 4 of them so far were even done under the so-called, Real-Time Oncology Review, RTOR program, a program which was introduced by the FDA in 2018. And I'm happy to announce today that we have one more, which we will use here for RTOR for this very special way of getting a fast track, which is for the IMpower010. But we not only have the largest number of BTDs -- next one, please. But we also have the largest number of technology platforms in oncology. And you can see that we have basically a differentiated therapeutic platforms for small molecules. We'll hear more about today about the SERD today, about the KRAS; the bispecifics; monoclonal antibodies; mosunetuzumab; glofitamab; and a little bit overlooked to have to say cevostamab fusion proteins; monoclonal antibodies, the TIGIT; antibody drug conjugates, Polivy; and personalized neovaccines maybe -- or maybe not all people know that we have first tied up with Biogen -- with BioNTech to develop this approach years ago. I mean everybody is talking about the mRNAs now, but we were basically the first one to really collaborate here and we also touch a bit on the personalized T-cells, the programmed T-cells for patients neoantigens for antitumor immune response. And this one, with this platform introduction, I wanted to open the presentation for William. William, over to you.

Thanks a lot, Karl, and hello, everybody. So in the next few minutes, I will give an overview of what's going on pRED oncology. Next slide. So I know the topic is oncology, but just -- this is just to highlight the contribution of the molecules from pRED to launches for Roche over the past decade. And in addition, the molecules that are in late stage under full development. Just wanted to highlight, in particular, the approval of Evrysdi, a small molecule that came out of the Basel group here. And then we look forward to the filing of faricimab in ophthalmology this year. And then we're also particularly proud of the scientific contributions of pRED to the Regeneron CoV-2 antibody as well as the AT-527 small molecule for COVID-19. Next slide. So today, we're here to talk about oncology. Within pRED, we have a focus on cancer immunotherapy and molecular targeted therapy. Today, I'll predominantly talk about cancer immunotherapy, large molecules, but we're also working on small molecules and molecular targeted therapy and then we have investments in game-changing innovation. Next slide. Over the past few years, we've signed many collaborations with external partners, so harnessing external innovation from wherever it comes from. On the left, you can see the in-licensing projects. And then on the right, multiple partnerships with academics. I won't go into detail about these, but we'll talk about the Tusk acquisition, the CD25 antibody, and then multiple other programs are in discovery and development. Next slide. So this is an overview of the pRED oncology development pipeline. You can see the vast majority right now is large molecules within the cancer immunotherapy space, multiple molecules in Phase I and II. And then within the cancer immunotherapy space, we -- previous slide, we separate into immune cell generators, T-cell engagers and immune cell modulators. Next slide. So building on the success of our 2:1 format T-cell bispecific, which has 2 binders to the target antigen and then 1 binder to CD3, we are expanding our portfolio here and building on the molecule Cibisatamab, which targets carcinoma embryonic antigen for colorectal cancer and then glofitamab, CD20 in non-Hodgkin's lymphoma. We have multiple other molecules in early clinical development as well as -- these are standard format T-cell bispecifics and then we also have 2 TCR-like or T-cell receptor-like T-cell bispecifics, and I'll talk about one of them. Next slide. So at this year's ASCO, we presented updated data on glofitamab, so I won't go into too much detail about that. Levi will mention that. In the latest data, in the step-up dosing, we're seeing a 71% complete response rate in relapsed refractory aggressive non-Hodgkin's lymphoma. These data are getting very competitive with CAR-T cells. And I just wanted to highlight here some of the additional activities that are going on with glofitamab. One is to remind the audience that because of the 2:1 format, we actually can combine directly with concurrent rituximab and obinutuzumab. This is because we can compete for binding with an antibody that has 2 Fab fragments, and you only need about 2% of the target antigens occupied to get T cell-mediated killing from the T-cell bispecific. As proof of concept then, we have done studies preclinically as well as in the clinic to show that with continuous obinutuzumab or Gazyva, you can actually get an enhanced effect. And here, we're taking into account multiple mechanisms. Gazyva has direct killing of CD20 positive cells as well as ADCC-enhanced cytotoxicity. And then we also have the T cell-mediated killing. So we're trying to take advantage of multiple modes of action against the target cells. In the preclinical data, we show the combination of continuous Gazyva with CD20-TCB leads to superior activity compared to either agent alone. And then in early clinical data, with continuous obinutuzumab given with glofitamab, we see antitumor activity actually almost in all evaluable patients. This is early data with 10 milligrams and 16 milligrams, and we're now examining the combination with the new step-up dosing regimen. Next slide. In addition, we're trying to learn more about the mechanism of action of glofitamab and, particularly, in T-cell bispecifics, what happens exactly with the T-cells in various locations. And for that, we've initiated a collaboration with ImaginAb to look at noninvasive whole-body monitoring using a CD8 positive nanobody to monitor CD8 T-cell activity. Shown here are preclinical studies with the CEA T-cell bispecific as proof-of-concept. On the top, you can see that with vehicle and a molecule CEA-4-1BB ligand, which provides co-stimulation to the T-cells. You see minimal activity then when you add the T-cell bispecific, you see some activity and then the combination leads to enhanced activity. On the bottom are shown 2 representative PET images from animals, taken 40 hours after administration of the nanobody. And the bottom line is with the combination of 4-1BB ligand plus the T-cell bispecific, you see the most CD8 T-cell infiltration into the tumors. The white areas denotes the tumors and the yellow areas denote areas of uptake elsewhere in the body. So we got first patient in, in a study with glofitamab and this CD8 tracer, and we look forward to seeing how this teaches us about the mechanism of action as well as if we can use this as a measure of early response to therapy. Next slide. We also have a vision to basically make treatment of non-Hodgkin's lymphoma chemo-free. And with that, we've started a combination of the glofitamab, molecule with a CD19-4-1BB ligand molecule, next-generation or CAR-T-cells, version 2.0, take advantage of 4-1BB signaling in addition to the T-cell mediated killing. And here, we provide Signal 2 with a separate molecule antibody, a CD19-4-1BB ligand. And you can see in the middle here in preclinical models that the combination in a very refractory tumor cell line leads to enhanced activity compared to either agent alone. And then on the far right, you can see from immunohistochemistry studies that the combination leads to massive T-cell infiltration compared to either agent alone. So this combination has also started in the clinic, and we look forward to seeing the results in the future. Next slide. So this is an example of the T-cell receptor-like T-cell bispecific that I was talking about. This is a molecule that recognizes an intracellular oncoprotein Wilms' tumor 1 that is presented in the context of MHC Class I. So it's HLA-A2 patient restricted. This potentially opens up a whole new space for us to target tumor cells because we're here, we're looking at intracellular proteins that are presented in the context of MHCs. Notably, there's only about 2,000 of these molecules on the surface of the tumor cells. So this is also an engineering feat where we at least preclinically see very minimal, if little, toxicity and lots of cytotoxic T-cell-mediated killing. On the far right, you can see in tumor models that we see dose-dependent killing of tumors that express the WT1 in the context of HLA-A2. This molecule is currently in Phase I as well. Next slide. Now I'll pivot to talk about developing next-generation or PD-1 molecules. We have 2 bispecifics that target PD-1 and LAG-3, which are both co-stimulatory molecules and then PD-1 and TIM-3. We believe that the bispecific format has advantages compared to monotherapy agents alone. And in preclinical models, we can show superiority compared to the monotherapy combinations as well. Both of these are in Phase I dose escalation, and we anticipate starting Phase II studies any day now, and we will compare against a PD-1 as the control arm. Next slide. We also have a PD1-IL-2 variant molecule, which I will spend a few minutes talking about. This was presented at the AACR earlier this year with an oral presentation. Next slide. So some of you might be familiar with our IL-2 variant molecule. This is a molecule that's engineered to eliminate binding to CD25 IL-2 receptor alpha, which is also called CD25. This is because T regulatory cells have high expression of CD25 and also CD25 is expressed on endothelial cells. And although IL-2 is approved, the wild-type IL-2 is approved in renal cell carcinoma and melanoma. The molecule does have side effects, which we think are due in part to activation of endothelial cells as well as an influence of T regulatory cells. So here, we've engineered the IL-2 variant to eliminate binding to CD25 alpha and, therefore, provide agonism of only beta and gamma. Now the twist on this molecule is that we've also engineered PD-1 -- a bispecific against PD-1 or a regular antibody against PD-1. And this fusion protein then provides cis-signaling of the IL-2 variant on PD-1 positive cells. So you can see in the middle, basically compared to a nontargeted or a non-PD-1 targeted IL-2 variant, we have 40-fold increase of stat signaling, which is a measure of activation of the tumor-infiltrating lymphocytes. On the far right, this depiction summarizes a lot of in vitro and in vivo data, but we believe that this molecule is superior compared to PD-1 or PD-L1 alone in activating and expanding the resource T-cells, which recently have been shown to play a large part in mediating the antitumor activity against cancers. Next slide. In preclinical models, we've seen very encouraging activity. On the left is an orthotopic pancreatic cancer cell model, where the Kaplan-Meier curves show basically dose-dependent survival of mice treated with the highest dose of PD-1 IL-2v. In fact, 707 mice survived and all of the other arms were inferior. And so this was compared to PD-1 alone as well as a non-targeted IL-2 variant. And then in the far right, in collaboration with the Doug Hanahan Lab, in a separate mouse model, we've also shown the molecule to be active. But interestingly, in this model, when you treat with PD-1 IL-2v, you see upregulation of PD-L1. And so the combination of PD1-IL-2v plus PD-L1 actually leads to superior antitumor activity. And in the far right are ultrasound images of one of the animals that had a complete response. Notably, these are usually very refractory to cancer immunotherapy. Next slide. Another molecule I wanted to highlight is anti-CD25 molecule. This is a molecule that we in-licensed from Tusk. As I mentioned, T-regulatory cells have high expression of CD25 and we wanted to eliminate T-regulatory cells. But the problem, if you just make an antibody against CD25 is that you may block IL-2 receptor signaling and, therefore, dampen your T-cell response. So Tusk was able to identify an antibody that not -- that binds the CD25, but does not block IL-2 signaling. The data on this were published recently. And we believe that this molecule will really help us answer the question as to whether elimination of T-regulatory cells will have a big effect on -- in cancer immunotherapy. So the idea here is the non IL-2 blocking antibody will bind to cells that have high expression of CD25. Then through ADCC, such cells will be eliminated. And then the remaining IL-2 will be redistributed to effector T-cells. Next slide. We recently presented data on this at AACR, and the data are summarized here. On the far left, we show that essentially this molecule has a preferential binding to induce T-regulatory cells, but not to CD8 T-cells as well as other T-regulatory cells. And then in early data in patients, in dose escalation, we see very nice dose-dependent depletion of T-regulatory cells in the blood at least in patients, and you can see that in the middle graph. On the far right, we just show a representative example of patients treated at the 6-milligram dose where pre and post treatment, we actually see increased CD8 T-cells. As you can see in the upper level is baseline and then the bottom level is on treatment. As you see an increase in Ki67 and CD8 staining, we see a modest increase in FOXP3 and then we also see an upregulation of PD-L1. I say these are early data. Notably, the adverse effects are minimal. So it's very manageable toxicity at this point. And this continues in dose escalation, and we plan to have combinations with atezolizumab in the future. Next slide. Finally, we are working on next-generation T-cell bispecifics. One of the challenges of T-cell bispecifics is that because they harness T-cells, if you have any expression or a little expression on normal tissues, you can get additional toxicity. So we are trying to enhance the therapeutic index of our T-cell bispecifics. And here shown is just one example where we're actually masking the CD3 component of the T-cell bispecific, as you can see on the left, with a cleavable linker. Then we're taking advantage of the fact that in tumors, you have protease specific and elevated protease expressions, such as metalloproteases and other kinds of proteases. And then in the presence of the tumor and the protease, you cleave the cleavable linker and you open up the CD3 [ mildly ] and, therefore, you hopefully will get a tissue-specific activation of a T-cell bispecific. In this particular paper, we use the folate receptor DCB to show proof of concept. The wild-type folate receptor had toxicity in the lung and in other tissues, And then use of this mask actually was able to eliminate such toxicity. And so we think proof-of-concept that this has been a reasonable approach to move in the future. We've also extended this data to a separate T-cell bispecific -- mesothelin T-cell bispecific. We look forward to bringing next-generation T-cell bispecifics into the clinic. And with that, I will conclude, and hand over to Ira.

Thanks, William. Happy to be here. Maybe next year, we'll all be together again. I didn't actually think this would last for 2 years, but here we are. So good morning, and today, I'm going to walk you through some of the activities that we at gRED are engaged in early development and even preclinical development. This is an outline of the various components of the portfolio as it now stands. So on the left, we have a variety of molecules in Phase I. I'm not going to be speaking to all of them, but giving you at least a hint as to what they're doing. You can see here also, I'm going to spend a little bit of time today talking about things that are not directly -- it can be considered immunotherapeutics, but things that we hope to use in combination with them. We have our personalized vaccine, Autogene cevumeran in Phase II and Phase III. You'll hear more from Levi on some of these molecules that are shown here on the right. In addition, of course, we have Tecentriq or atezolizumab, that is serving as the basis for combination for many of the agents that you've heard about thus far and will hear about continuously. And I wanted to mark the fact that it's about 10 years now, although it seems like yesterday since we first introduced atezolizumab into patients. Next slide, please. first, I'd like to talk a little bit about some of the targeted agents that we have been working on. We've long been interested in the RAS/MAP kinase pathway. And just recently, it's not impossible to start to generate inhibitors and drugs against different elements of this pathway, but actually have some promising ability to not only work but also more importantly, to combine with each other. These are very plastic and complex pathways. And I think our firm belief as well as, I think, many in the field is that the path forward is going to be generating the appropriate agents such that they can be used in combination with one another in order to overcome or anticipate resistance. Next slide, please. The first agent I want to mention, there are 3 of them. As you can see on the previous slide, is our KRAS G12C inhibitor. This is certainly a molecule that should be well known to all of you due to the activity in the field. We believe that our molecule is at least comparable, if not better, than the 2 main competitor molecules that are out there. Obviously, we're much earlier in developmental stage. But the reason we're persisting and pushing this forward is because this pathway, as I've mentioned before, is nothing, if not complex; nothing, if not incredibly plastic. We have to anticipate that indeed there's going to be resistance, both primary and acquired resistance and feel that really the way forward with these types of agents will be to understand what the nature of that resistance is and be able to deal with it with both targeted and immunotherapeutic agents. Next slide, please. One thing that we've noticed over the years about the RAS/MAP kinase pathway, is the fact that when it's activated, it decreases the transcription of MHC Class I gene products, which means that cells that bear mutations in RAS or activation in MAP kinase or RAF will all show depressed MHC Class I peptide MHC complexes on the surface of these cells, rendering them less sensitive to T-cell attack. So a second molecule that we can use in conjunction with RAS, particularly for NRAS mutant as shown here, which is the pan-RAF inhibitor called belvarafenib together with cobimetinib, which is our MEK inhibitor, as you can see on the right, this agent on its own in the clinic is showing some promising activity. But again, I think, one of the main reasons for engaging in these BRAF for pan-RAF inhibitors as well as the KRAS inhibitors has to do with the fact that we anticipate they should provide excellent combination partners for atezolizumab as well as almost any other immune therapeutic that revives on T-cells recognizing peptide MHC complexes on the surface of tumor cells. So these again are relatively early stages, but based on our preclinical analysis, we're quite excited to see what's going to happen when these combinations do start to emerge. Next slide, please. Now let's switch to the cancer immune therapeutic pipeline more directly. As in previous years, we continue to believe that the best way to go after this opportunity for patients is to understand what cancer immunology is all about? What tumor immunity is all about? And as you know, we've broken down the various steps that are involved in generating T-cell responses to cancer into 3 basic immune phenotypes: One is the so-called immune desert, where there seems to be a failure of infiltration of T-cells and indeed, all other immune cells into tumors. And we're understanding what it is we need to do in order to overcome and reverse that situation. The immune excluded group, of course is the area where there does seem to be a T-cell response to tumors, but the T-cells cannot access the tumor nests or the tumor parenchyma because they are physically withheld from the tumor and/or immunosuppressed as a consequence of a stromal investment that is characteristic of these tumors. And so the T-cells remain on the outside. And finally, there's the inflamed group, which are those patients who exhibit already a T-cell infiltrate before therapy even begins. And it's this group almost universally throughout the field that's showing most of the reactivity to almost all immunotherapeutics that are being used to date. So this is a very active and a very promising group of patients to focus on. But I just want to point out that the bulk of the opportunity actually resides elsewhere in the cancer immunity cycle as important as the inflamed group may be. We believe and, in fact, we've seen that some patients will require only a single immunomodulator, such as anti-PD-L1 monotherapy to elicit benefit. Others that we'll need 2 or more. And as I said previously, what we are trying to do is expand the types and nature of combinations that we're doing to include also targeted pathway agents because many of these not only intrinsically exhibit positive effects on patients' tumor but also have now increasingly well understood immunomodulatory properties that we feel will synergize with these types of immunotherapeutics. Next slide, please. Now we're expanding multiple modes of how to go about doing this that involve both using immunomodulators that will boost endogenous immunity as well as other types of agents that will effectively provide synthetic immunity where there is none and the bispecific antibodies fall into the latter category. So there are monoclonal antibodies, bispecific cytokines to [ hear ] vaccines as well as our recent foray into the cell therapy arena. I won't talk about the cytokine activity very much. William already introduced you to pRED's IL-2 program, so we'll leave it predominantly at that, except to say that we do have a very active program in collaboration with Xencor on highly modified IL-15 molecules, which are now in patients and, in fact, are looking quite promising. But we'll wait to collect more data before saying anything further about those. In the next slide, I've outlined in a slide, I've used before. The fact that T-cells will exhibit a variety of co-stimulatory molecules for activating receptors as well as inhibitory receptors, such as PD-1 CTLA-4 that restrict T-cell activity. There's been a lot of interest in trying to see whether one can add to the ability of PD-1, PD-L1 blockade to work and yield a benefit for patients by targeting one or more of these others. Now of course, we decided to concentrate most of our efforts on the TIGIT CD226 pathway as shown here. Question is, why did we do that? Part of the reason, of course, was that TIGIT was really first discovered at Genentech some years ago. But as we studied more and more the -- how all of these systems work together, we realized that TIGIT does play an important role early on together with PD-1 and controlling T-cell responses in cancer. Now I would -- Levi will talk about the clinical work that we are pursuing with this agent, but I want to just present you with one important mechanistic insight that I believe helps provide us with additional confidence regarding how it is reusing this. As you can see on the next slide, just in the last year or so, we've been able to, using a series of functional and biochemical experiments, understand that there's really a remarkable convergence of the TIGIT CD226 pathway together with the PD-1 CD28 pathway that we've talked about and published on previously, namely that the co-stimulatory molecule that is normally thought to be under the sole control of TIGIT, namely CD226, is actually under the control of PD-1 to a large extent, as is CD28. And indeed, all of this seems as well to happen in these T stem cell memory cells that appear to be the major target for activation as a consequence of PD-1, PD-L1 blockade. TIGIT does itself play an important role, but not in regulating the activity, mainly the phosphorylation or activation of CD226, but rather by providing an effective competitor for a shared ligand between TIGIT and CD226, namely PVR. So what this says then is that both of these systems are unique and that they are present in the same important cell, type that TIGIT has a coordinated role in regulating CD226 together with PD-1. And rolling all this up together, it kind of shows you that you would almost anticipate that the only way you're going to see optimal activity with TIGIT is if it's in combination with PD-1, and that the combination of the 2 indeed will provide a mechanistic synergy that should result in better activity. Next slide, please. Now in addition to that, I think it's important to understand that TIGIT may be not a one-trick pony, but actually as a consequence of its activity in controlling CD226, which is going to emerge, I think, as a very, very important co-stimulatory molecule in T-cell and also NK cell biology. We are coming up with evidence as are others in the field that this interaction controls myeloid cell activity, it controls T-regulatory cell activity, namely CD226 activation will reprogram regulatory T-cells to actually become effectors. And finally, NK cell activity, which is also -- NK cells are also high in CD226, and as a consequence of its activation, one finds enhanced effector activity on the part of NK cells. So all these, all work in the appropriate and correct direction. And we anticipate they indeed will act together to boost patient responses. Next slide, please. Now our bispecific programs overlap, certainly with what you heard from William with respect to the CD20 CD3 agent, Mosun. And Levi will speak about that. Most of other activities really are concentrated on multiple myeloma. And I just want to take a minute to talk about cevostamab, which is our FcRH5 CD3 bispecific. And next slide, we'll show you our early clinical results on this. FcRH5 is a pan B-cell marker and also found in plasma cells and late-stage B cells. And as a result, we feel it's a very good target for eliminating multiple myeloma cells. And as you can see on the right, our Phase I step dosing dose escalation studies have yielded quite nice overall responses and certainly a healthy number of complete and very good partial responses, as is known in the field. So early days still for this, but we feel this is a program because of this early activity that will move quite rapidly. Next slide. Now just in the last couple of minutes, let me turn to the personalized programs because I think this is probably where the future lies for a lot of what we hope to accomplish for patients. As Karl mentioned, we've been working together now with BioNTech for some years on their RNA vaccine, adapting it for use in cancer. I think we all know how effective and transformational it's been in providing prophylactic immunity against COVID. And this is a very powerful platform and one that we are also learning more and more about to understand how we can use it for cancer patients and generate the same type of effective T-cell responses as we've seen for antibody responses with the COVID vaccine. Just figuring out the workflow and adopting the workflow to actually do this on a per patient basis has been no small task. And it is important to do, in fact, necessary to do because it turns out that, as you know, each patient bears their own unique profile of mutations. A fraction of those mutations are immunogenic. We need to be able to identify what mutations are expressed, cut, compute, which ones are likely to be immunogenic. and then very rapidly generate an RNA vaccine that is adapted to that particular patient's constellation of immunogenic mutations. To make a long story short, we figured out how to do this. I mean we can have very, very favorable turnaround times for this, which meet our goals. And our first studies with BioNTech in this vaccine, we found exceptional T-cell responses. but limited clinical activity. And I think we now believe the reason for that is that although the T-cell responses are very deep, they're not as broad as we need them to be. So we're actively working with BioNTech to actually increase the breadth of those T-cell responses and moving forward actively in the clinic and have 1 randomized Phase II study in melanoma underway at the moment. In the process of this though, next slide, we encountered a second platform, which is a DNA-based platform and partnered with another company in Oslo, Norway called Vaccibody. This is an approach that actually is very similar in its workflow to the BioNTech RNA vaccine, but one which uses DNA plasmid vectors, which can be injected with a needle-free injector into the muscle of the patient. And it's then the muscle cells that make the vaccine. And what's special about the vaccine is that although it contains all of the same type of preidentified patient-specific epitopes that we need to include in the vaccine, the [ immuno ] terminal or the front end of the vaccine shown here with these 2 green arrows comprises a chemokine called MIP-1-alpha, which is an attracted and activator for antigen-presenting cells, specifically dendritic cells. So what this means is the vaccine is targeted in the periphery specifically to these dendritic cells would become activated, carry the antigen back to draining lymph nodes and initiate what seem to be very deep and very broad tumor T-cell responses that are now being studied actively together in collaboration with Vaccibody in the clinic. We feel that this is an approach that has sufficient promise and sufficient importance that we really need to understand it and as a result, took on this second system. Now in the last slide, I just want to introduce you to an even earlier set of studies that we've begun in collaboration with Adaptive Biotechnologies in Seattle. Here, we're going to capture or are capturing all of the learnings that we've received from the vaccines in terms of being able to identify which mutations in patients are most likely to be immunogenic, which are most likely to be protective when a T-cell response can be elicited but get around the necessity for actually generating a T-cell response. So in these patients, what we do is, again, to biopsy the tumor sequence the mutations that are there, predict which ones are likely to be most protective or most immunogenic. And then we've adapted physically go back into those patients and isolate T-cell receptors that are matched to those predefined optimal immunogenic antigens, then take those T-cell receptors, insert them into autologous T-cells and then grow those T-cells up and introduce them back into the patients in a fashion that's similar to CAR-T therapy. But one that is much, much more specific and has actually already based on early results with TIL therapy, very, very promising activity with respect to being able to bring benefit to patients with solid tumor indications. So that's all I'd like to say. And now I should hand the microphone or whatever passes through the microphone to Levi.

Great. Thanks, Ira. Hi, everyone. It's a pleasure to have the chance to give some highlights of our late-stage portfolio. And I'd also like to acknowledge and welcome Charlie Fuchs to Roche. He's our new Global Head of Oncology in Hematology, and he'll chime in during the Q&A session a little bit later. All right, next slide. So as you've heard from Karl, we continue to have one of the largest and most diverse oncology portfolios in the industry. You've also seen examples of this from Ira and William. And from this portfolio, we expect numerous data readouts that can support filings over the next couple of years. And this, of course, will start with Tecentriq in adjuvant non-small cell lung cancer based on the data we just presented at ASCO and that Heather Wakelee will walk you through shortly. And then later this year, we will see the results of the POLARIX study, which could support filing of Polivy in frontline diffuse large cell -- diffuse large B-cell lymphoma. And then we also expect to file on data for mosunetuzumab in the third line or greater follicular lymphoma setting. And also glofitamab in third-line or greater diffuse large cell lymphoma based on pivotal cohorts that we'll be reading out this year. And then in 2022, we'll see the first pivotal data for tiragolumab, which you just heard about from Ira, that will be initially in small cell lung cancer, and we'll also see additional adjuvant data for Tecentriq in the coming months. And then furthermore next year, we'll also see the first Phase II/III data for -- or Phase III data for giredestrant in second-line or third-line HR-positive breast cancer. And I'll say a bit more about each of those programs momentarily. Next slide. So as we move our portfolio forward, we see several strategic directions that are growing in importance. And one approach involves moving into earlier disease settings. And our data for Tecentriq in adjuvant non-small cell lung cancer is, of course, illustrative as are several upcoming readouts for both Tecentriq and actually other medicines in our portfolio. Now a second approach involves the pursuit of rational combinations and new indications as the requisite disease biology insights emerge. And here, our portfolio offers numerous opportunities for combinatorial plays as well as avenues into tumor types where we haven't been as active previously and those include multiple myeloma or which Ira alluded to that with cevostamab or of course, HR-positive breast cancer with the SERD. And then the last point, which I won't say too much about today, but we continue to explore new ways in which our portfolio can enable personalized health care through, for example, tumor-agnostic trial designs and increased use of advanced analytics to drive actionable insights from large-scale patient data. Next slide. So as you know, treating cancer early offers the best chance for cure and it could also help reduce the societal costs that are linked to management of more advanced disease over the course of time. And it's likely that the early cancer population, which is already large will continue to grow and in part, this is because of the increasingly powerful screening and early detection platforms that are coming of age. So clearly, there's an unmet need, and we see opportunities within our portfolio to pair both new treatments to the screening approaches and hopefully bring much more patient benefit in that setting. Next slide. And of course, our early cancer efforts already span multiple tumor types and approaches, and I'm not going to walk through each of these in great detail. But what you can see at a high level is that, not unexpectedly, Tecentriq is a key anchor across the board. In lung cancer, we also have studies of tiragolumab and Alecensa ongoing in the earlier cancer setting. We'll talk more about giredestrant in HR-positive breast cancer shortly, but we will be initiating -- we already have new adjuvant studies ongoing, but we'll be initiating an adjuvant study in that setting soon. And then several assets are being explored in earlier settings and heme malignancies as well. Next slide. So the unmet need in these earlier cancer settings is also endorsed by the fact that our data for Tecentriq in adjuvant non-small cell lung cancer was filed under the Real-Time Oncology Review, ROTR mechanism, as Karl mentioned at the beginning and project Orbis. And our filing strategy -- a fast-to-market strategy, given the impact of the data, was in PD-L1 positive patients with stage 1 to 3 disease. And Heather Wakelee will likely speak to this when she presents the data, but approximately 30% of non-small cell lung cancer patients present with resectable disease, and at present, chemotherapy is the only approved treatment option in this setting, but we expect that to change, particularly in molecularly defined subgroups or genetically defined subgroups. And of course, this includes PD-L1 positivity, but also, for example, EGFR mutations and other TKI mutations in the future. Next slide. All right. So now we'll talk about specific programs, starting with tiragolumab, which as you heard from Ira, is our anti-TIGIT antibody. And as Ira, very nicely laid out, there are several lines of scientific and now clinical evidence that support our tiragolumab development program. And first, I'll mention the results from the CITYSCAPE Phase II trial, which was presented at ASCO last year, and that showed that the addition of tiragolumab to Tecentriq was associated with an improved progression-free survival compared to Tecentriq alone in the PD-L1-high subgroup of non-small cell lung cancer. And so one of our Phase III studies, the SKYSCRAPER-01 study is essentially a confirmatory trial of those Phase II results. Now in addition, our Phase Ib expansion cohorts have provided some intriguing data for a possible benefit of adding tiragolumab to Tecentriq and esophageal squamous cell cancer. And so that provides the basis for a couple of other Phase II trials that we have ongoing. And then furthermore, we have a scientific rationale in small cell lung cancer, which, in addition to the fact that small cell has previously been shown in general to be an IO-responsive disease. Of course, Tecentriq is standard of care in that setting. But in addition to that, as you heard from Ira, small cell has elevated expression of the ligand for TIGIT, PVR. And so you could imagine that this could conceivably be analogous to sort of PD-L1 high tumors that often benefit greatly from other IO. So that data was part of the rationale for our SKYSCRAPER-02 Phase III study in small cell lung cancer that we expect to read out next year. Next slide. So our full tiragolumab development program is shown here. And in addition to the studies I just mentioned, we have a Phase III trial in stage 3 -- sorry, in non-small cell lung cancer and a couple of Phase II trials in other lung cancer settings. We have Phase II trials in cervical cancer and squamous cell head and neck cancer. And so those data will start to read out next year and then the following year. Next slide. Okay. Now let's move to breast cancer. And of course, we're making a substantial effort in HR-positive disease anchored by giredestrant, which is our oral SERD and HR-positive breast cancer. Of course, the estrogen receptor functions as an oncogene driver for that cancer, so endocrine therapy has been a mainstay for many years. However, a large portion of early breast cancer patients discontinue endocrine therapy within the 5-year interval because of safety or tolerability issues. And then even for those that complete endocrine therapy, as many as 30% may see their cancer recur. In other words, they develop either intrinsic or acquired resistance to this therapy. So our oral SERD giredestrant has best-in-class potential for actually several reasons. Its mechanism of action in addition to the well-known degrader activity involves nuclear immobilization and hence, there's suppression of the oncogenic transcription output that comes from estrogen receptor signaling. And so that combined mechanism, we think, will be important just given the importance of both ligand-driven and ligand-independent signaling in breast cancer. In vitro, the potency of giredestrant is 7 to 15-fold higher than other SERDs that are in development. And as I'll point out in a moment, giredestrant is well tolerated both alone and in combination with other key breast cancer medicines. And we have a single dose, a 30-milligram dose that offers robust efficacy and tolerability, both as the monotherapy and in combination. Next slide. So let me just walk through some of the high points from our -- from the clinical profile that's been observed for giredestrant thus far. And while doing so, I just emphasize some specific areas that we think make actually a pretty strong case that giredestrant could be a best-in-class SERD. So let's start on the left. So on the left, you can see that the objective response rates and clinical benefit rate at our recommended 30-milligram dose really compares favorably to other surges at similar points in development. And we see robust responses in patients with ESR1 mutations and also in the setting of prior -- where there have been prior treatment with CDK inhibitors. And so really, that's the first point is that we have a single dose with robust efficacy. And really, every bid is good, if not better than competitor surge and we're carrying that single dose into all of our Phase III programs, monotherapy in combination. And in specific, we have seen no adverse PK effects or we see no drug-drug interactions when giredestrant is combined with, for example, palbociclib. And this is different from other surge in development where we've seen doses needing to be cut in half when combined with palbo because of exposure issues. And by the way, even then, the palbo exposure ends up being reduced sometimes by nearly 1/4 in that context. So that's an important distinction given, where obviously in the metastatic setting and prudentially beyond surges are going. And the second point is that giredestrant is well-tolerated. At the 30-milligram dose, we've seen a single grade 1 case of bradycardia out of 41 patients treated at that dose thus far. And in fact, that case was subsequently deemed unrelated to treatment by the investigator. And we've seen no ocular toxicity. And overall, our rates of treatment discontinuation are quite low. The incidence of Grade 3 or greater treatment-related adverse defense -- or adverse events is around 5% and serious adverse events is around 9% for giredestrant. And if you compare that to competitor SERDs, I mean, there's a competitor SERD that has SAE rates of 26% once you get up to the doses that they're using for combos and monotherapy. So overall, we have robust efficacy and a favorable safety, really that stack up quite well against the competition. And so now on the right, you can see results from a recent neoadjuvant window of opportunity study. And here, we're looking at the effect of giredestrant on Ki67 in tumor cells after a 2-week treatment window prior to surgery. And the ability of giredestrant to suppress Ki67 in tumor cells was quite robust at all 3 dose cohorts. And so this lends actually further support to our 30-milligram Phase III dose. And you can see here, on average, we saw a 78% reduction of Ki67 and actually 55% of cases showed complete cell cycle ORS after only 15 days of treatment. So we're pretty enthusiastic about this data package. And so we actually have another Phase II trial in the new adjuvant setting that we'll be presenting the data from that later this year. We have a Phase II -- we have a study in second-line and third-line atrial positive breast cancer reading on next year that potentially could support an initial filing. And we expect to launch an adjuvant study of giredestrant compared to an aromatase later this year. Okay. Next slide. All right. In the last few minutes, I'll discuss just a few highlights from our heme malignancy portfolio. And let's start with Polivy and the POLARIX study. And so Polivy was the first agent and actually remains the only agent to show an overall survival benefit compared to standard of care in the relapsed/refractory DLBCL setting. And that provided a key rationale for exploring this potential benefit in the frontline setting where, of course, the possibility of improving cure rates may exist. And just as a reminder, our Phase Ib data for Polivy certainly seem promising. Obviously, there's the usual caveats of cross-sell comparisons, but we had a 77% CR rate, which compares to sort of 58%. Our OR was 89%. And now we have 2-year PFS data that was 83% compared to a recent study control of 71%. So overall, our data, we remain bullish about these results. Obviously, R-CHOP in frontline DLBCL has proved difficult to unseat over the years. But POLARIS certainly offers an interesting opportunity. It's a well-powered study. We've got 875 patients with more than 2 years of follow-up. So we'll look for that readout later this year. Next slide. All right. So now I'll just say a few words about our CD20-CD3 bispecific antibody base. You've heard already about mosunetuzumab and glofitamab from Ira and William. And together, we have a blend of -- on the one hand, we have glofitamab, which has potential best-in-class efficacy, and we have mosunetuzumab, which also has robust efficacy but has a safety profile which could support use in particular settings where patients may need that kind of a therapy. And so given how important this mechanism of action could become in the management of B-cell malignancies, the idea for us is developing both of these bispecifics, each that has a distinct clinical profile could potentially meet a wide range of patient and provider needs, just given the spectrum of context, in which people with various types, indolent versus aggressive B-cell malignancies, community settings versus tertiary care settings may encounter the health care systems. So next slide. So we see our development plans for really both bispecifics playing out in 3 phases. And the first phase involves monotherapy in later line either follicular lymphoma for mosun or DLBCL for glofit and the filings for these, as I mentioned, are expected. Well, certainly, the data readouts are expected this year and then -- in a data-driven way, the filings will happen either this year or next year. And then -- so that's the first phase. And the next phase then involves rational combination. So for example, we have a Phase III combo trial of mosunetuzumab together with lenalidomide in relapsed/refractory follicular lymphoma, that's starting soon. And we already have a Phase III trial ongoing of glofitamab, plus GemOx in second line or greater DLBCL. And then finally, we aspire to move into frontline DLBCL. We think one or both bispecifics could have a path there. And to that end, we're exploring various combinations using both bispecifics, including combinations with Polivy, which I'll come back to in a moment. Next slide. All right. So just a couple of data slides to illustrate again why we're optimistic about both bispecifics. So here's data for glofitamab where we have a step-up dosing regimen that produces high and durable responses in aggressive non-Hodgkin's lymphoma. And these are patients who have failed several lines of therapy. And in this particular study, we saw a complete response rate of over 71% at the recommended Phase II dose for glofit. And here again, the CRS was generally manageable. We saw grades 1 and 2. It was combined -- confined to the first couple of cycles. So we think these data certainly endorsed the notion that glofitamab, may have best-in-class potential. Next slide. All right. Finally, I mentioned the idea of rational combinations between these bispecifics and Polivy and here are some data from a dose escalation cohort study between most done in Polivy and relapsed/refractory non-Hodgkin's lymphoma, including some patients who have progressed on CAR-T therapy. And you can see that we're seeing complete response rates that exceed 50% in some cases, together with a CRS rate of only 9%. And all of those events were grade 1 event. And so with this regimen, which is undergoing Phase II expansion now, there's no need for mandatory hospitalization and we're also looking at Polivy combinations with glofit. So certainly, these data are additional reason to be optimistic. So in summary, we continue to be bullish about our entire oncology portfolio as well as its ability to deliver meaningful patient benefits for years to come. And maybe with that, I'll stop and turn it over, I think, maybe to Heather next.

Well, thanks. That's great. It's always good for me in my lung world to be able to see all the other exciting things that are going on. So I'm going to bring everybody back to lung cancer and talk about this really exciting study that we presented at ASCO just over the weekend. So this is IMpower010 and this was a randomized Phase III study looking at adding adjuvant atezolizumab after surgery and chemotherapy for resected early-stage lung cancer. Let's go to the next slide. Just to set the background, when we think about lung cancer, of course, it's the second most common malignancy, #1 cancer killer. So still a big deal. And we're finding about 30% or so of that in receptible early stage, trying to shift more of the total lung cancer population to early stage with screening, but that's not the focus of today's talk, but we do expect that there will be more early-stage patients identified in the future. And the standard of care when I started practice was just surgery. We then learned about 15 years ago from randomized trials that if you added chemotherapy, given over a period of about 3 months, you could improve disease-free survival. And disease-free survival hazard ratio when you pooled all of those chemotherapy trials was about 0.84. I'm mentioning that to kind of put it in context if we talk about what we're seeing in this trial. So disease-free survival hazard ratio of 0.84, that did translate into an overall survival benefit. And that changed the standard of care and for patients with resected Stage II and IIIa disease and some 1b, we offer chemotherapy after surgery. The only other significant outcome that has happened in this patient population is in the approximately 10% of patients will have an EGFR mutation in their tumors. And in those patients, we've been studying giving EGFR tyrosine kinase inhibitors, the oral medications for usually 2 years in some of the trials, 1 year or 3 years in the case of ADAURA with osimertinib. And all of those trials have shown significant disease-free survival benefit with osimertinib very profound disease-free survival benefit. However, none of the studies have shown an overall survival benefit. And when you look at the curves for overall survival, for any of the studies, you're not seeing that separation. So we got some new updates at ASCO on some of the trials done in Asia with other agents. Jury remains out on ADAURA but it's a different mechanism when we're dealing with the TKI suppressing in the adjuvant setting versus chemotherapy or immune therapy where that longer-term benefit perhaps cure seems a bit more probable. So I just again, to set some framework. So bringing back to this study, IMpower010 was a randomized Phase III trial looking at atezolizumab or not after surgery and chemotherapy. And this is the first trial looking at an immune checkpoint inhibitor in lung cancer, in the adjuvant setting, with the disease-free survival endpoint. So next slide. So this is the scheme. I'm not going to go into too much detail. It's pretty straightforward. Basically, patients had 1B and if they were Stage 1B, meaning no lymph node involvement, the tumors had to be at least 4 centimeters in size. That's a historical cut point based on some of the data from the earlier chemotherapy trials, when we look at who's at highest risk of recurrence and where we're perhaps doing more benefit. And then Stage 2 with lymph node involvement and 3A. All patients received chemotherapy. There's a listing of the different chemotherapy regimens we used and then were randomized. Now not all patients who started chemo were randomized, some decided to drop out of the trial after they've been through some of the chemotherapy. So we randomized just over 1,000 patients, 1,005, 1:1 to get either atezolizumab for up to 1-year or to receive best supportive care. A couple of nuances about the study. When the study was opened, we stratified by PD-L1, but we used the SP142 assay. That is an assay that uses both the PD-L1 on the tumor as well as on the immune infiltrating cells. As time went on, realized that, that assay, the SP142 was not in line with most of the kind of the way that the whole field was moving, which was to look specifically at the tumor-based assays. And so the analysis of the trial was based on the SP263, which again kind of correlates with what we are doing much more often in practice in the metastatic setting. So just to help you all understand that. The study was designed to look hierarchically at a couple of end points. So first, looking at the PD-L1 expressing stage 2 to 3. Those are the patients who, by nature of Stage 2 to 3A more likely to have a recurrence and more likely to have that happen earlier, so endpoint. But also by PD-L1 expression, not knowing at the time we designed the study, whether the PD-L1 biomarker, which is important in many studies in metastatic disease, but not all studies in metastatic disease for checkpoint inhibitors. We didn't know if that was going to be an important factor or not, but we chose to look at that first, and it turns out it is important. The way that it's set up then is potentially looking at intention to treat population and overall survival. So next slide. This goes into that testing plan a bit more. So again, first population, PD-L1 expressing Stage 2 to 3A. Next population, all comers regardless of PD-L1 expression with Stage 2 to 3A. Next is all comers in total, that 1,005 patients who were randomized and then going to overall survival. And of course, an overall survival endpoint takes longer to get to because you have to have those events versus the recurrence events for disease-free survival Next. Next. Okay. And none of you are going to be able to read this. You probably all looked at the slides that came out at ASCO, but this is the patient population. So this again was a global study, but predominant enrollment was in Europe, North America, some in Asia. When you look at the ethnic background of the patient population, it fits with where the enrollments happen. Median age was in the 60s as expected. I think the key points to focus on, for the stage, only 12% of patients were Stage 1B, majority Stage 2, 40% Stage 3. When we look at PD-L1 expression, 55% of the patient population expressed PD-L1 that was across the stages fairly similar. We also looked at EGFR mutations and ALK translocations but only in the patients with non-squamous histology. So we looked on the study-enrolled patients with both squamous histology, which is about 1/3 of the patient population and non-squamous. The driver mutations tend to only happen in the non-squamous, so that's where we mandated having the EGFR and ALK populations analyzed and the unknowns or the squamous where that wasn't required. Next. Okay. So this is the key take-home. This is a very, very good disease-free survival hazard ratio curve. And you can see what this means is that by giving the atezolizumab, we reduced the risk of disease recurrence or death by 34%. That's pretty important. If you remember, the disease-free survival hazard ratio with chemo was 0.84. Here, we're dealing with 0.66 in this patient population. Again, this is selected by PD-L1, 55% of the patients with Stage 2 to 3A disease. What's nice about this curve as it separates early and it stays separated. Recall the treatment only was given for up to 1 year. And so when we look at these curves, a large section of that is after completion of all treatment. So this is an effect that persists despite stopping the treatment, and that fits with the mechanism when we're talking about an immune checkpoint inhibitor. The median follow-up here is 33 months. Next slide. This is a first spot looking at all of the different populations. And you can see that pretty much all of them had that benefit. A couple of outliers, patients currently smoking. We don't have an answer for that. We'll look into it more. It's a small group. And then not unexpectedly, patients without translocations didn't seem to have any benefit. Next. Okay. So then this is when we bring in all-comers. So this is bringing in that 45% who did not have PD-L1 expression Stage 2 to 3A into the whole mix, including those with the PD-L1 expression. We still have a statistically significant endpoint hazard ratio of 0.79. But the significance of this, it's still statistically significant. The total magnitude, though, has decreased. And this to us is saying that PD-L1 expression is really important. Next. Okay. This has some really critical information when we look at this subpopulations. And the subpopulation, I'm drawing your attention to with the red box is based on the PD-L1 expression. So you can see on that the patients who had the highest PD-L1 expression greater than 50%, this was a prespecified cut point. We did not prespecify in the statistical analysis to look at other cut points, but of course, questions are coming up about that. But for those with greater than 50%, the disease-free survival hazard ratio is 0.43. That's great and a lot of excitement around that. When we look at the greater than 1%, again, that's 0.66, still quite a significant benefit in reducing disease recurrence. That's perhaps going to translate into an overall survival benefit, we don't know. But I, at least looking at these numbers, am pretty enthusiastic about that. But what we also see is that for patients with tumors that do not express PD-L1, we're not seeing a benefit with atezolizumab. So this does seem to be biomarker-driven by that PD-L1 expression. The other parts of this slide to note, we're not looking for the EGFR and ALK patient populations where the tumors are expressing those biomarkers, they're not seeming to have the benefit. So we're not really focusing in there. Next slide. Okay. So reminding you about the statistical analysis. First, PD-L1 expressing Stage 2 to 3A hazard ratio of 0.66. Next, all-comers, still statistically significant in the Stage 2 to 3A population. Next, we're looking at the Stage 1B, bringing that 12% of patients in to this mix. So that's the entire 1,005 patients. And then once we get final results from that, then we get to formally look at overall survival once enough events have occurred. We have not yet had those events. When we look at this all-comer patient population, the significance boundary was not crossed. So what that means is there haven't yet been enough events; however, we have some preliminary information, which is on the next slide. You can see here, again, this is not formal yet. We haven't crossed the significance boundary. That's going to take a few more events to get there. We're close, but wasn't completed. Here, that disease-free survival hazard ratio is 0.81. So not that different from the 0.79. Again, we don't have the formal testing yet. And we'll have to see that, that result should likely be available in a relatively short time with the next interim analysis, but I don't have the exact timing on that yet. Next slide. So these are the very, very, very early, very preliminary overall survival. I'm going to draw your attention to the far left, and that is the patients with PD-L1 expression on their tumors Stage 2 to 3A. It's early, but we are starting to see a hint of separation. That is very encouraging. Of course, time will tell. When we look at the other patient populations, it's difficult to claim any clear separation at this time. Again, these are all very, very early. You can see lots of hashes, very preliminary. Next. Okay. Of course, you can never talk about our treatment without talking about the toxicity. We did not have any surprises, but atezolizumab does bring in toxicity as any cancer therapeutic does. So these were all patients who had had surgery. They all had chemotherapy. And so when we look at adverse events, 92.7% of patients receiving atezolizumab had some adverse event but 70.7% of those on the best supportive care. When we look at the more significant adverse events, that number is 21.8% for atezo and 11.5% for best supportive care. When we look at those that are considered serious, that's about 17 -- it's 17.6% versus 8.5%. Most of the additional toxicities with atezolizumab are immune-mediated and this is a class effect. We have seen this across all immune checkpoint inhibitors, PD-1, PD-L1 regardless of disease, regardless of stage. So no surprises, but it does mean that we do want to be careful in selecting patients who are the most likely to benefit given that there is some risk of harm. And again, that's true of chemotherapy of any of our other therapeutics as well. When we look at the Grade 5, our patients who died on -- during the trial conduct, there were 8 patients who died on the atezolizumab ARM 4, were felt to be treatment-related, 3 patients on best supportive care during this time after the randomization. And those were, for them again, were felt to be immune-related adverse events. So on the next slide, we'll go into a bit more detail. This is a long list, but essentially, what we know about the immune checkpoint inhibitors is that if an autoimmune toxicity develops, it can basically affect any organ system. Usually, these can be treated with steroids and other immune modulatory agents. But it is something that clinicians are mindful of the need to continue to be very vigilant about. There was no 1 outlier here and really no surprises as far as the toxicities that we're seeing. Next slide. Okay. So wrapping it up, so we have time for some questions. So IMpower010 is the first Phase III study of cancer immune therapy to demonstrate disease-free survival improvement in the adjuvant non-small cell lung cancer setting after platinum-based chemotherapy. For patients who had had complete resection and had their adjuvant chemotherapy, the addition of atezolizumab led to a statistically significant disease-free survival benefit. That hazard ratio was 0.79 in all patients with Stage 2 to 3A, 0.66 when you look specifically at patients whose tumors expressed any level of PD-L1. So we do believe that, that biomarker is very important for finding the patients most likely to benefit. We are going to continue analysis. We do not yet -- we cannot yet say anything about the formal statistical outcome, disease-free survival and the intention to treat population. We have not crossed that boundary at this time. And the same for overall survival, it's still very premature but trending in the right direction. So -- and the safety profile, of course, we just talked about, no surprises. Do need to be mindful about safety with any cancer therapeutic. We believe -- I believe that atezolizumab is likely going to be practice changing. We have seen other agents approved in the adjuvant setting based on disease-free survival outcomes. So with that, I will stop, and I think we're going on to questions. So thanks.

Yes. Thanks a lot, Heather. So we have a long list of questions. And maybe we can just try to keep the questions per person to ask a question to one, if that is possible, because we have really a long list. I wanted to start off with one to you, Heather, from John Murphy from Bloomberg. He was asking what would you use this regime, in which patients? That was his question.

Right. So of course, we have to wait for approval. But I believe this would be something appropriate for patients with resected. At this time, I would focus on the Stage 2 to 3A though I believe there would be some 1B patients I would talk about with it, if it was approved. And I would use the biomarker, the PD-L1 biomarker with that 1% threshold cut point. I would also make sure it was excluding patients with EGFR, where we have other treatment options, at least until we have further data, I think there'll be some questions to be asked there.

Thank you. Sachin, I'll open your line.

Sachin Jain from Bank of America. Just a few questions, if I may. So first on Tecentriq adjuvant and the RTOR process. Could you clarify how far into that process you are and how you're thinking about timing of approval? That's question one. Question two, could you just clarify the population you filed in. Levi, I think in your introductory comments you said you filed in 1 to 3A PD-L1 positive, which wasn't one of the prespecified groups. So if you could just clarify that? And if I may, just one on the news from the FDA yesterday. Levi, while we've got the opportunity, big picture. Obviously, it's a bit of a surprise from the FDA. But just any early thoughts as to how you think about changing development program, if at all? And then given lots of questions, any updates on whether you open the box for graduate any earlier? And apologies for that question. I know it's not oncology-related.

Sure. Maybe I'll ask Charlie to comment on the filing -- high level comment on the filing strategy for Tecentriq, IMpower110 data, and then I'll come back and take the Alzheimer's data.

Sure. Well, as you said, we were successful in submitting under the RTOR process. And we're looking forward to a faster market approval process with the FDA. Specifically, just to clarify, the indication we're looking for is Stage 2 to 3A. And given the data that are available today, that would be in PD-L1 positive patients. So I think, obviously, the trial included 1B, but specifically, our submission is 2 to 3A.

Thanks, Charlie. And then yes, on the Alzheimer's, certainly, as has been mentioned, milestone day for Alzheimer's disease yesterday. From our standpoint, basically, we are -- as you know, we're already -- our graduate program with gantenerumab, 2 Phase III studies, fairly far along. At the moment, we don't have any specific plans to change the time frame of the readout. Certainly, we continue to be bullish, especially in light of the FDA action yesterday. We know, of course, that our program, it's a subcutaneous antibody. We see robust clearance of beta amyloid. We've dosed -- we have a titration-based dosing mechanism that ensures that all patients get to the maximum dose of gantenerumab. So certainly, we think that those characteristics bode well for a good outcome. But of course, we have to -- Alzheimer's is challenging. This is a tricky -- very tricky setting, but we certainly like our program at least as well today as we did yesterday.

Yes. So the team went for a fast fighting strategy. And as the data evolves, we'll keep you updated then about the next steps as this was an interim analysis and then we -- as Heather pointed out, we will have more data, let's say, along the way. Luisa, you would be next. Luisa Hector, I opened your line.

I wonder if I could ask Dr. Wakelee a couple of questions on the data. So the first one, I mean, not to distract from the amazing progress that we have from this study, but I'm just curious about what you would like to see next? And could you see perhaps a role for Tecentriq plus TIGIT in the adjuvant setting? And any thoughts about those patients not responding quite so well? So -- and, therefore, would a combination therapy look good there? And do you have a view on the duration of therapy? I know it was 1 year, but could you go longer? Would you like to see longer duration of therapy? So that's Tecentriq. And then I just wondered with William, you did make a comment about the bispecific PD-1 approach. So you said that the advantage of the bispecific rather than the individual antibodies, could you just remind us why that is superior? And it may be a slight stupid question, but just to check the reason for PD-1 because obviously, you have atezo PD-L1. So is it the sort of the binding the receptor situation that you want to have the PD-1 there rather than PD-L1?

I guess, I'll go first.

Yes. Heather, yes. Absolutely, please. Yes.

Okay. So I think, so to get to your question about duration first and how do we go forward, there are many questions in that. We are looking and -- we, meaning the lung cancer global community, looking at giving these immune checkpoint inhibitors prior to surgery, so in a neoadjuvant approach, those studies that are looking at that also include giving the adjuvant. So the adjuvant is the pure question for patients who go straight to surgery, but there is going to be a lot of investigation in giving the drugs early as well. Within that context, we can also then see which patients are benefiting upfront versus just those who are having this benefit. You can't see until time has gone by. Tiragolumab is being investigated and that's sort of an approach, bringing this into the adjuvant setting. As far as that duration of treatment, when we're trying to plan these studies, it is somewhat arbitrary, whether it's 1, 2 or 3. 1-year does seem to be appropriate for the adjuvant setting, at least if you look at the disease-free survival curves I showed you where when we look out at the year 2 and year 3 marks, granted, not all patients are getting to that point, but our median follow-up was close to 3 years. We're not seeing those curves come together. So at least from that perspective, that 1 year did seem to be an appropriate time period. It's also difficult for the patients to go much beyond that. It's a lot of discussion when someone's like, okay, surgery is done. I've had my chemo. I want to be done with this treatment. That 1 year is doable, longer than that becomes very difficult when we're coming in for infusions every 3 weeks. So hopefully, that answered both parts of that question.

And...

William, please? Yes.

Thanks, Karl. In regards to the bispecific versus the monotherapy. For example, LAG-3 is expressed on T-regulatory cells and TIM-3 is expressed on myeloid cells. So there's possibly a sync by giving the monotherapy agents as well as potential toxicity that you can get. So the way we've engineered our bispecifics is, they actually have a higher affinity for PD-1. And then we are targeting basically the PD-1 positive cells. So that's the other rationale for why we use PD-1. So basically, the antibody should bind a PD-1 first and then inhibit either a LAG-3 or TIM-3. And we see potential advantages with that in terms of potential efficacy but as well as side effects.

Yes. Did we address the question of the combination of TIGIT and Tecentriq in adjuvant? I think this was also one of your questions, Luisa.

I touched on that. So yes. So we are looking at that, yes.

Okay. Good. So I was not very successful in expressing myself that we have maybe one question because, otherwise, we are going overboard. Maybe we can do 2 questions per person so that we have a chance to go through the question. I know that, Heather, you have a hard stop in about 10 minutes from now. So maybe we could also, let's say, direct any kind of questions to you for the next, let's say, time for the next 10 minutes. Matthew. Matthew Weston from Credit Suisse.

And I'll follow instructions. So 2 questions for Dr. Wakelee, please. I'd be very interested as to what you would do. You were asked earlier on what -- would you use it and what patients would you use adjuvant Tecentriq in? And you answered very clearly. I'd be very interested what you would do with a patient who fails adjuvant Tecentriq in lung? Would you look to move to a different checkpoint inhibitor plus chemo? Or where would you go? Because it's going to become a very relevant discussion item in time. And then secondly, If I look at the [ farthest part, ] then the 2A patients did significantly better than the 3 patients. The 2Bs are a little bit different, but it's a much smaller group. So should we be hopeful that the Stage 1 patients are actually going to show efficacy and all we need is a little bit more time?

Great. So in the setting of recurrence after adjuvant treatment, of course, there's a lot of factors in that as to if it's localized or not and what to do. But for many of those patients, it would be a combination of chemotherapy plus checkpoint inhibitor. I don't think I would necessarily switch to a different one. It would be depending on what chemos to combine. So I don't necessarily believe that switching from one checkpoint inhibitor to a different one is all that important of a question just because we can rechallenge adding in different things and maintaining the same drug and see responses in metastatic. But that's kind of getting into a different conversation. As far as trying to make too much out of the subpopulations by stage, it's biologically very difficult to come up with a rationale now for that. So I'm not ready to really try to respond to that until we've seen a bit more. And I am optimistic that we are going to see that benefit in the 1Bs as well. It's not surprising that it's taking longer to see that because, again, it's a smaller population that recurs. And so if you have a smaller likelihood of recurrence, it's going to take a bit longer to see a significant difference in those who do recur based on any sort of a therapy. So I don't think we're going to necessarily see a stage dependent benefit once we get to the end of all of this.

Wimal, you will be the next.

Wimal Kapadia from Bernstein. I'm just going to stick with one with a couple of little parts to it. So maybe for Ira actually, and the vaccine and the NeoT approaches. So how do you actually go about improving the neoantigen target selection? How do you determine which ones make a difference? And is there actually a maximum number of epitope targets before it becomes ineffective? And then just tied to that, would you use adjuvants and can this ever be a monotherapy?

To answer your second question first, I don't think it's a monotherapy. I think probably one would always have to combine at least with the checkpoint inhibitor because you will otherwise probably not be able to squeeze out as an optimal T-cell response and just wind up with large populations of exhausted cells to otherwise good antigens. In terms of identifying the best antigens, this is a work in progress. We now have changed our approach. So far, most of what you do is to see which mutant antigens bind best MHC molecules. But that does not allow you to distinguish between those that bind and those that actually exist. So over the past 2 years or so, we've generated an AI based on a machine learning model, just looking at all those peptides that are possible to be presented. And we do our predictions now increasingly using this AI, which we can find objectively using patient data is about 50% better than the original model. So we're hoping to have a much more accurate selection scheme than we've used in the past.

Okay. Great. And just in terms of the maximum number of epitope targets?

That's unknown. I think it's not unlimited simply because there's a phenomenon of immune dominance, which I think is a problem in these kinds of vaccines, so you have to choose an immunization scheme as a set of adjuvants that provide enough stimulation to get a good range of responses, but not so much simulation that one by itself predominates. So I would say somewhere between 5% and 10% at the most.

This gives me the opportunity to say thank you to Heather. I know that you have to head off to another important meeting at ASCO. Thanks for your participation. I think we could address actually all the important questions to the 010 study. Thanks for all to help on this one. Thanks for your help today. I wish you all the best, and bye-bye, Heather. The rest will stay on. Bye-bye.

Thank you, all. Bye.

Bye, Heather. Bye-bye. Thank you. So next one would be Emmanuel Papadakis.

I'm just going to ask Dr. Wakelee a question but you just released her from duty. So maybe I'll take one on SHP2, and you're pretty late. The early monotherapy data has been disappointing. Do you believe your molecule has any particular points of clinical differentiation? And how constructive are you on the various combination opportunities, in particular, KRAS G12C? And then maybe a quick one for Ira. IL-2, a lot of industry energy continues to go into that target. You just added a new PD-1 conjugate yourselves. We discussed in the past, center for IL-15, which you alluded to as an equally, if not more interesting and safer target to go after a similar effect. If you could just give us your latest thoughts there, that would be helpful.

Who goes first?

Yes. Please start off, Ira.

Well, IL-15 and IL-2 are related, but they're different. And I think the difference is, particularly when using an intact IL-2 molecule -- IL-15 molecule rather that has the high-affinity alpha domain associated with it. Its distribution seems to be quite distinct from what one finds or what one couldn't find even in the literature with respect to IL-2. So they have different characteristics which will map out, I think, the different activities and also different safety profiles, which I think were raising in terms of which one turns out to be more efficacious. We have to wait for the data depending, but they're related to non organical. And was the last question for me as well? Or is that for Levi? I'm sorry.

I can certainly take it at a high level unless you want to take that. At a high level that the -- we think that -- so first, our SHP2 inhibitor, there are important differences in the molecule. We think the first-generation SHP2 inhibitors may have been constrained. There were dosing constraints that could certainly have dampened the ability to see monotherapy responses, the time on target issues and toxicity concerns -- so there's an opportunity to improve there for sure. And then absolutely, there's quite robust preclinical data to support the combination of SHP2 together with KRAS inhibition. And actually, there are other SHP2 combinations that are supported in oncology as well. So we think that there's opportunity both from the standpoint of the molecule itself, but also in terms of the combinatory regimens that we can build with that inhibitor.

Yes. Let me just add one further comment to that, which I neglected to say, which is, it's important when choosing targeted agents, the best ones would be ones that also don't interfere with T-cell activity or in fact, may help T-cell activity. In the case of SHP2, remember that this is the negative phosphatases associated with restricting T-cell activation via both CD28 to CD226, so we think all these things add up actually in the right direction.

Yes. Does it address your -- all your questions, Emmanuel?

Yes. Thank you.

Yes, thanks a lot. Mark Purcell would be the next. Mark, I opened your line.

Thank you, Karl, can you hear me?

Yes, clearly.

Great. There was a question, I mean you've got a very broad neoadjuvant and adjuvant program across multiple cancers. I was just trying to understand your perspective at the moment, where we are in terms of regulatory acceptance and physician acceptance of surrogates to overall survival. So across your programs, you've got pathological complete response, EFS, DFS and obviously, OS. And the reason why I ask, obviously, there's a debate in breast cancer at the moment. But if I sort of take 2 points: Firstly, you're the only company that has a HER2-positive neoadjuvant trial ongoing, the IMpassion050 trial versus PCR endpoint. So will you need overall survival for that trial to be accepted? And then secondly, when I look at IMpower010, in the great than 1% patients, the hazard ratio of 0.66 on DFS. But when you think about greater than 50 and 1 to 50, the hazard ratios are 0.43 and 0.87. So in isolation, it looks like the vast majority of any OS benefits can be driven by the PD-L1 greater than 50%. And you may not see that in the patients that are 1 to 50. I know that's not how you've segmented the patients based on PD-1 status, but that looks what the data suggests.

Maybe we'll give Charlie a chance to kind of speak. There's a high-level question there, and then there's a very detailed question, so maybe we'll -- Charlie, you want to start?

Of course. Why don't I start with the second question first. Namely, the data, as you saw, it was predefined show DFS benefit. And what we think is clinically meaningful in individuals who are 1% higher and in our discussions with regulators and based on our submission for HER2, our submission will be in PD-L1 positive, which I think Heather very nicely, Dr. Wakelee, very nicely spoke to. And as more data emerged from this norm analysis, we'll continue to look at these subsets, including the PD-L1 negatives. With regard to your broader question, as you know, this is an evolving field in terms of what is an acceptable surrogate endpoint for regulatory agencies, including the FDA. In certain circumstances, certainly, disease-free survival is an acceptable endpoint. And I think in most cases, in adjuvant therapy, that is the case. With regard to pathologic complete response, I think, what the agency has been clear is that there needs to be considerable validation of past CR with respect to its ability to predict DFS or OS. So I think it's very context and disease-specific. So it's difficult to answer any particular question broadly.

Yes, excellent. Thank you, Mark, for your questions. Hope we you could address all your reference questions in a good way for filing and for readouts, I have to say. Richard Vosser would be next.

Two questions, please. Just on the CD19-4-1BB combo with glofit, when can we see data on that? And would you also look to do combinations with mosun as well? I understand maybe more DLBCL directed for this, but just thoughts there. And also, you went through the program and the filing with glofit and mosun. We saw at ASH probably better safety profiles with subcutaneous, but I think these are all IV. So when can subcutaneous be brought to the fore here for these products?

Karl, do you want to start or do you want me to start? Or William, do you want to start?

Yes. I'll just start with the CD19-4-1BB ligand combination. So we're currently in dose escalation in the combination. So I can't give you a time frame of when we'll be able to release the data. And yes, we would be open to combining with mosunetuzumab as well.

I can just add, we're really excited about the combinations for both bispecifics and beyond their activity as monotherapies with regard to most on, you've obviously heard from Levi with a combination with Polivy. And similarly, we're interested in combining it with potentially other therapies, as you've heard within our portfolio. And I'm just -- I think you asked about the subcu formulation. And we actually are moving that forward. We see that as really a unique opportunity and a very patient-centric opportunity to simplify care delivery. As you know, it's -- the demands on infusion centers are considerable as well as the time for patients and providers. So we think that the subcu formulation is a really important opportunity to deliver the benefit of mosun and potentially even improve its ability to be delivered easily, maybe even with potentially even less toxicity. So we're moving forward on that and anticipate data, hopefully, some data next year.

Thanks a lot. Simon Baker -- thank you, Richard, for your questions. Simon Baker, I opened your line.

Just the one because one has already been asked. And going back to Luisa's question, I wonder if you could give us any more details on the Phase II study of TIGIT plus Tecentriq plus chemo that started in April. That seems to overlap with the IMpower010 patient population but also includes neoadjuvant. So is that potentially the future if that works? And is there any timing yet on when we'll see data? I appreciate it's extremely early, but anything there would be very useful.

Charlie, do you want to start on that? .

Yes, sure. Well, obviously, in the immediate term, the benefit of 010, we hopefully will get approved and be of help to the patients in the near term. With regard to the ability of adding tira in the early settings, we're -- as I think Ira really shared, we're very excited about the combination of Tecentriq with tiragolumab. And albeit we are really pleased with the benefit we can offer patients with adjuvant Tecentriq. We want to continue to build on that platform. So in some respects, we think it's -- if indeed the data with tira hold up, we think it will be a logical progression to move beyond 010.

Okay. So we have 5 more minutes. Thank you for your question, Simon. 5 more minutes, 2 more people in the telephone queue. I would say, Stephen Scala from Cowen, I'll open your line.

Can you hear me?

Yes, we can hear you.

Two questions. In the past few months, it appears that the SERD filing has been moved up from filing in 2023-plus to 2022. I'm curious what enabled that acceleration? And then secondly, a few immune-targeted bispecifics were noted today. Curious why is Roche not pursuing a PD-1 TIGIT bispecific?

Yes. Maybe I'll start on the SERD and, certainly, we'll open it up to Charlie or others who want to answer. So we -- of course, our big picture, of course, is that this is all data-driven, but we have data spanning the later-line metastatic setting, the frontline metastatic setting, and of course, now moving into the earlier setting. In the later-line setting, I mean, as we -- with our Phase III dose, obviously, we see the data, the robust data, including in ESR1 mutant patients and after additional therapy, we think the design of the study could enable filing if the efficacy results support. And so that, in principle, would allow us to file as early as next year. But obviously, it's data-driven. What was the second question?

PD-1 TIGIT call.

Bispecific.

Yes, maybe William or Ira can speak to that?

Ira, you want to take that one?

Yes. I mean, initially, the reason is, we just -- we didn't see an obvious advantage, but I think more importantly, when combining 2 agents like this, we didn't know what the optimal dosing or dosing schedule is going to be. And if you have more -- you retain more flexibility by doing them separately. Conceivably, we could go back to it and do it, although there are some potential problems associated with it, whereas we believe that you need an intact Fc domain on TIGIT. It's not obviously the case that, that is true also for PD-1, which may, in fact, contribute to the depletion of effector T-cells.

Last but not least, Peter Welford from Jefferies.

Fairly brief. Firstly, just on the glofit and mosun regulatory pathway. Curious like at one point in time, you said this year filing for glofit. Then I think later on, you said next. Can you just clarify exactly what potentially perhaps is needed for mosun and glofit in those expansion cohorts to enable filing. Is it purely patients for safety? Or is there a certain hurdle in terms of response rate or follow-up that you think is needed before we can do those filings? And if I could just ask a clarity on Tecentriq, perhaps to Charlie on a regulatory standpoint. Can you just clarify, given you've used RTOR, can another company pursue a filing for adjuvant lung cancer with a PD-1 also using the RTOR pathway? Or are there some sort of FDA rules around this?

Yes. Maybe I'll take the bispecific one and let Charlie answer the RTOR question. So essentially, the readouts are this year. This is all just about timing. The readouts are this year. And based on where we are in the studies, we think that the filing of mosun could happen later this year, the filing for glofit will kick into next year. But it just -- there's nothing beyond that. The studies are on track. And so it's really just about the timing of when the data is available. Charlie?

Yes, with the question, to my understanding, a subsequent company could submit an RTOR following, hopefully, our approval. But I want to -- we probably should confirm that with the regulatory expert. My understanding is you're not restricted from applying for an RTOR review.

Okay. So we will check it, and get back to you, Peter. So with this one, many, many thanks to all of you, to our speakers, to the panel. Sabine, to you, for taking the lead in organizing this event. Loren Kalm from our U.S. team being instrumental also for 2 of the presentations. Yes. And Gerard, for the Q&A, for the preparations for you, for your interest. Hopefully, we see each other next year at ASCO in person and I sincerely hope that. I really miss our dinner event and the red wine in the evening. Having said that, wish you all the best, and have a nice day. Thank you. Have a nice day and a nice evening. Bye-bye.