Roche Holding AG (ROG) Earnings Call Transcript & Summary

Ladies and gentlemen, welcome to Roche virtual event of the 2022 ASCO Annual Meeting focusing on the oncology pipeline. My name is Henrik, and I'm the technical operator for today's call. Kindly note that the webinar is being recorded. [Operator Instructions] One last remark, if you would like to follow the presented slides in your end as well, please feel free to go to roche.com/investors to download the presentation. At this time, it's my pleasure to introduce you to Bruno Eschli, Head of Investor Relations. Bruno, the stage is yours.

Thanks, Henrik. So welcome to our ASCO '22 event focusing on our late-stage data presented at the conference and on our early oncology pipelines for both pRED and gRED. Let me quickly take you through the agenda of today. Next slide, please. Our first speaker, Elena Bernedo-Arzac, our Head of Oncology Therapeutics, will provide us with an update on our late-stage oncology pipeline. Our second speaker, Ginna Laport, Global Head of Hematology and Product Development in NHL and CLL, will take us through the pivotal Phase Ib results for glofitamab and third-line DLBCL, which we presented at ASCO. She also will provide a short update on some POLARIX data which we presented. And the second half of the event will then be dedicated to our early oncology pipelines. We have Johanna Bendell, Global Head of Oncology of pRED; and Andy Chen, Head of Research Biology at gRED with us, who will outline their respective portfolio strategies, overall, highlighting 14 different Phase I and Phase II drug candidates which span various technologies, including small molecules, monoclonal antibodies, novel bispecific antibodies, fusion proteins and also mRNA cancer vaccine. Next slide, please. To mention also, I think we have overall planned here for 1 hour of presentations, and then we will have a Q&A session of 30 minutes. I also wanted to use this occasion here to provide a quick comment on the remaining 22 news flow for Roche. So regarding the upcoming readouts, what you can see here, we still expect 5 Tecentriq readouts in the adjuvant setting. So we have head and neck HTC, TNBC, which is not shown here because it's an external study, and then renal and neoadjuvant non-small cell lung cancer to read out. We have 2 additional readouts for Tecentriq, tiragolumab combination to come in the second half. We have -- this is Phase II data in cervical cancer and the Phase III in esophageal cancer, which you can see here on the list. And for SKYSCRAPER-01, just let me say this again. This is our combination study in the PD-L1 high non-small cell lung cancer setting. There, we expect the next interim to happen latest in H1 2023. We also, as you can see, have additional Phase III data here for Venclexta in multiple myeloma, which is a decent opportunity; and then Susvimo in DME and diabetic retinopathy. And finally, in Q4, this is the big elephant in the room. As you know, we expect gantenerumab results in Alzheimer's disease. This will be on top line release of our GRADUATE 1/2 studies, which we will read out at the same time. Regarding the regulatory milestones shown here on the slide, I think all filings are on track. In addition to what is shown here in the table, we have now glofitamab to be filed globally, so in the U.S. and in the EU in '22. And with that, I would like to hand over to Elena for providing us with an update on our late-stage oncology pipeline. Elena, please?

Thank you, Bruno. Hello, everyone. Glad to be here. I plan to cover a few highlights and progress of our pipeline that give us momentum of our oncology portfolio. Our mission at Roche Oncology has been for many years to lead the cure for cancer, and a key tenet of our strategy are our efforts in early disease. And this is a chance for cure for patients but also a significant business opportunity as we see an increase in the early detection and screening rates. Next slide. We have a significant investment in early disease across the different disease areas, more than 20 studies in late stage reading out in the near term. Starting by the landmark study of IMpower010 that was presented at ASCO last year, Tecentriq in non-small cell lung cancer. And I'll get back to it in my next slide. In lung cancer as well, we expect later this year results from the new adjuvant perioperative study IMpower030 that will complement the role of Tecentriq in the early lung cancer space. Previous slide, sorry. Yes. Other positive readouts include Tecentriq in the triple-negative space IMpassion031. And here as well, we expect later this year, data on the TNBC neoadjuvant adjuvant. In breast as well, as you are very familiar, we had a new adjuvant giredestrant and coopERA positive study. That was the first randomized study to show superior activity of an oral SERD over an aromatic inhibitor. And in the heme portfolio, another landmark study, POLARIX with Polivy in first-line DLBCL, which was a huge milestone for the field and for patients. The first Phase III study inoculative statin in more than 20 years. So we are looking forward, as Bruno said, to this upcoming data readouts. And you can see here what we expect in this year, in green. Moving on to the next slide now, please. So let's focus on Tecentriq, which is the first and only cancer immunotherapy to demonstrate positive Phase III results in the adjuvant lung -- non-small cell lung cancer. IMpower was presented at ASCO last year, and as I said, and was called practice-changing adjuvant treatment by the leading KOL for patients with PD-L1 greater than 1% stage II/IIIa non-small cell lung cancer. There is a big unmet need in this space as about 1/2 of all people diagnosed with early breast cancer, around 50% or more, will still recur at some point after surgery. So after Tecentriq was launched last year in the U.S. in October last year, it has also been approved in more than 20 countries, including China. And we expect the EMA approval any day now after the CHMP positive opinion. In terms of the launch in the U.S., we have seen a very strong launch. We're testing rates up from 40% when we launched. Up to 70% since Tecentriq was approved. And this is space overall, the early lung space, which is in full, I would say, evolution. And an important level here is going to be the increase -- it is being the increase in early diagnosis and early detection and screening. At the time of the approval of Tecentriq, so that you have -- right, an approximate figure here, only 30% of the patients with non-small cell lung cancer diagnosed with resectable disease. So we know this is critically important to improve patient outcomes, and we are fully committed here to improving screening rates and screening efficiency. And we have significant efforts, including with third parties to increase that. We also expect, in general, that the systemic therapy is going to increase with therapeutic options. And Tecentriq is set to become the standard of care for patients with PD-L1 positive patients. On the next slide, tiragolumab as you know, is an important asset in our pipeline, and we have a very comprehensive development program ongoing. You are well aware, we had a couple of Phase III studies, recent studies reading out. SKY 1 in the non-small cell lung cancer and SKY 2 in small cell lung cancer. Starting by the SKY 2 in small cell lung cancer, that was presented yesterday here at ASCO. This study evaluated tiragolumab in combination with Tecentriq and chemotherapy as first-line treatment for people with small cell lung cancer. You know the study did not meet the co-primary endpoint of progression-free survival, and the co-primary endpoint of overall survival was not met at this interim analysis. And it's unlikely to which statistical significant underplanned final analysis. This study, as we have previously shared was a calculated risk. It was well presented and discussed yesterday, I think, here at ASCO. We had hoped to improve on the outcomes of Tecentriq, right, mainly on IMpower133. It is a very hard-to-treat disease with very limited options. But however, we also know that this is a very different type of disease than other types of lung cancer. And that actually was noted as well yesterday by Dr. Rudin. This result shouldn't be extrapolated to other diseases. It was also noted yesterday that the study replicated the benefit of Tecentriq and -- so we saw both in PFS and OS in the control arm that Tecentriq really replicated the results of 133. So clearly reinforcing the role of Tecentriq as a standard of care in this disease. Moving on now to SKYSCRAPER-01, the non-small cell lung cancer. This study evaluated tiragolumab in combination with Tecentriq versus Tecentriq. As you know, the study did not meet its co-primary endpoint of progression-free survival. At this first analysis, the other co-primary endpoint of overall survival was immature, and the study will continue until the next planned analysis. Our expectations here are different than SKY-02 as we are encouraged that the numerical improvement was observed in both co-primary endpoints, PFS and OS, that confirm the biological activity of this combination. And we also know that the most meaningful endpoint in non-small cell lung cancer is overall survival, and even more so with this class of agents, cancer immunotherapy. So we are now looking forward to see more mature overall survival data for this study. We remain confident in the overall tiragolumab program and our strategy building on Tecentriq, providing new treatment options in advance and difficult-to-treat disease as well expanding early into earlier disease. We continue to believe that TIGIT has a role in cancer treatment and we'll share obviously additional results from our program as we -- the results emerge. Moving on now to the breast franchise, we have, as you know, a strong legacy with our HER2 portfolio that is delivering today $9 billion and will continue to grow in the next few years. We continue to invest with a personalized approach in the HER2-positive disease. But at the same time, we are actively expanding into the HER positive with 2 molecules that are potential best-in-class, giredestrant as a SERD and inavolisib as a PI3-kinase inhibitor. The next slide, looking at the HER2-positive early disease, we know that a high efficacy and safety bar has been established in the early setting here. Looking at our APHINITY study, the combination with Perjeta and Herceptin, and if we look at the 6 years of follow-up, we saw a 94% disease-free survival in the ITT population. So that shows you what a high bar we have achieved within this curative setting. The 8 years follow-up of APHINITY will be presented in 2022 as well. PHESGO here is bringing to the field an innovative subcutaneous coformulation of 2 monoclonal antibodies that significantly reduces health care cost and resource use, and is largely preferred by patients. We've seen a strong launch with 25% conversion rate in early launch countries, and including a very strong launch as well in the U.S. And as I said, we continue to build here on our deep understanding of the biology of this space of the HER2-positive disease. And we continue to expand with a personalized approach. Here, you see the heredERA study, where we are combining PHESGO with giredestrant, aiming to have a more comprehensive blockade of both pathways, the HER2-positive and ER positive. In the next slide, so in the ER positive space, the results from the coopERA study were presented at ASCO 2022. coopERA, as a reminder, is the first randomized study to show superior antiproliferative activity of SERD over an aromatase inhibitor. As you know, the study met its primary efficacy endpoint that was presented at San Antonio last year, demonstrated by a greater suppression of the Ki67 following 2 weeks of new adjuvant treatment. The final analysis show now that the improvement in suppression of Ki67, and therefore, superior antiproliferative activity, as shown here in the purple bar, with the combination of palbociclib was maintained with giredestrant at 81% of our anastrozole at 34%. And this is relevant because, as you know, CDK inhibitors are perceived as equalizers. So keeping the benefit we saw in the first window of opportunity study, despite the addition of palbociclib, is a strong signal of giredestrant. The data of coopERA clearly support our Phase III ongoing adjuvant study known as lidERA, that will assess the efficacy and safety of adjuvant giredestrant in 4,000 patients. We are going to know here on a head-to-head versus the physician choice. In the next slide, so we see that the high unmet need remains in the HR-positive disease. And we are targeting here both early and metastatic disease with 2 assets under development, giredestrant, I talked in the previous slide as a potential best-in-class SERD, and inavolisib PI3 kinase inhibitor as well potential best-in-class. With giredestrant, again, we recently disclosed as well the results from the acelERA study, which is a Phase II evaluating giredestrant versus physician choice of endocrine monotherapy. The study did not meet its primary endpoint of PFS. However, we showed a numerical improvement in the PFS hazard ratio, which was not statistically significant. However, we saw even a more pronounced improvement in the PFS in patients with higher dependence on the estrogen receptor activity. And the response rate was also greater in these patients comparing with the control arm. So the efficacy signals with the acelERA study, together with the positive outcome of coopERA. So both studies support our ongoing clinical development program, both in the first-line metastatic disease in the adjuvant setting. The second asset I wanted to highlight here is inavolisib which is a PI3-kinase inhibitor with a unique mode of action exhibiting high potency and specificity for the PIK alpha isoform. And the early evidence we see best suggest that inavolisib has a superior therapeutic window which is the differentiator here in this space versus class competition, including alpelisib with a better efficacy, safety and combinability. If we move to the next slide, and now switching gears into our hematology portfolio. We do have a very strong and broad portfolio in hematology, which is delivering today $8 billion at the moment and poised to continue to grow over the next 5 years. This is probably the first-in-industry portfolio. We are doubling down in areas of strength, particularly in non-Hodgkin's lymphoma, where we are advancing, as I mentioned earlier, in the first line with Polivy, but also introducing a new class of agents, the T cell bispecifics, mosun and glofit. And the pivotal data of glofit were presented here -- pivotal data in the third line plus DLBCL were presented at ASCO here, and you are going to hear more about that from my colleague, Ginna, in this call. At the same time, moving now into the multiple myeloma space, we are at multiple myeloma. We are actively expanding our footprint here. Multiple myeloma you know is significant market opportunity, $20 billion market. It remains a noncurable disease, which is characterized by multiple relapses, with an overall 5-year survival rate still in the 50%. So there is an opportunity to clearly do more here. And we are advancing to very exciting T-cell bispecifics into our early development program, cevostamab and GPRC5D TCB, and you are going to hear more about these 2 assets from my colleagues from gRED and pRED. Next slide, thank you, Lorraine. So Polivy, as I mentioned, with the POLARIX study, represented a huge milestone for patients on the field, right, the first Phase III study in 20 years, showing a clinically meaningful improvement in PFS. The study has been approved just recently by the EMA, and we are under -- and is under regulatory review and with other health authorities around the world, including the FDA. A very -- in Europe, we have -- there is approximately 40,000 patients diagnosed with DLBCL each year. And an unmet need clearly remains here. And as you know, this is a curative setting because around 4 out of 10 patients still relapse after the first-line treatment and the majority of patients that require subsequent lines of treatment have poor outcomes. Polivy offers an off-the-shelf fixed duration treatment with a well-tolerated safety profile, and physicians have already gained experience using Polivy in the relapsed refractory setting. So we believe that Polivy in combination with R-CHOP will become a new standard of care with patients with this aggressive type of lymphoma. And has been called out by key oncologists as a new standard of care. And this represents as well a multibillion market opportunity. The next slide is going to be my final slide on the CD20, CD3 bispecifics, we think that they will play a dominant role in non-Hodgkin's lymphoma in the next decade, and we believe that our 2 molecules give us the flexibility to address the different needs of payers, providers and practice setting and patients. We have a positive CHMP opinion for mosun in the third line class follicular lymphoma, and again, expecting any day now an EMA approval. And we're excited about the potential of this molecule as the only first-in-class bispecific in this setting that currently does not need hospitalization and fix course of therapy. And now this is what I wanted to share from my perspective. I'm going to hand over to my colleague Gina who's going to tell us more about the glofit data presented at ASCO.

Thank you, Elena. Hello. This is Ginna Laport, and I'm here to talk about glofit and we'll give an update on POLARIX in our subpopulation. So this past Friday at ASCO, we presented our pivotal Phase II data featuring glofitamab, our first-in-class CD20 x CD3 antibody for patients with relapsed refractory diffuse large B-cell lymphoma. This slide shows the key inclusion criteria, which included patients with aggressive DLBCL, diffused large B-cell lymphoma, and patients had to have received at least 2 prior therapies to proceed to this trial. As far as administration, it was IV administration for a maximum 12 cycles. Each cycle was 3 weeks. And looking at the dosing schema here shown, patients received 1 dose of obinutuzumab on day 1 of cycle 1 for CRS mitigation. One week later, they received their first dose step-up dosing of 2.5 milligrams of glofitamab followed 1 week later on day 15 with cycle 1 of 10 milligrams. This step-up dosing was also an additional measure for CRS mitigation. For subsequent cycles, patients received 1 dose, the target dose of 30 milligrams of glofitamab on day 1 of each subsequent cycle for up to 12 cycles. The primary endpoint was complete response rated by independent review committee. Next slide. This slide shows our baseline characteristics, and you will see this is a heavily pretreated, highly refractory population. The median age was 66 years old, the predominant NHL subtype was DLBCL. Looking at the evidence that reflects or supports our high-risk population, you can see that 60% of patients had at least 3 prior lines or more before entering the study, 1/3 had received prior CAR T therapy treatment, and 85% or 86% were refractory to their most immediate prior therapy. Nearly 60% were primary refractory and 30% were refractory to prior CAR-T therapy. So again, this reflects the high-risk nature of this population. Next slide. Regarding efficacy, with a median follow-up of 12.6 months, our CR rate was 39.4% with an overall response rate of nearly 52%. Responses were achieved fairly early with the median time to first response being 42 days. Next slide. Now when we look at complete response rates by independent review committee in our prespecified subgroups, the CR rates were consistent on most subgroups when we delineate by number of prior therapies, age, prior CAR-T therapy, NHL subtype. However, when we looked at refractory to last prior therapy or the post-transplant setting, the patients who were relapsed had higher CR rates compared to patients who were refractory, which again reflects the high-risk nature of this population. Next slide. We showed that our responses were durable even after cessation of therapy. So the median duration of follow-up of nearly 11 months, our median duration of response was 18 months and our median duration of complete response was not reached. Our responses were sustained as we show here because at the 12-month landmark, our duration of complete response was 77.6%, again, reinforcing the durability of our complete responses. Next slide. So to further analyze the outcomes and to support the durability of our responses, we actually look like what we call the supporting cohort. This is an earlier cohort of patients who are enrolled that fit the similar criteria of our efficacy population with patients with the same NHL subtype and patients who received glofitamab doses greater than 10 milligrams for at least 8 to 12 cycles. Using this criteria, we captured 101 patients from the earlier cohorts, and they achieved a CR rate of 35%. But what's notable our DoCR follow-up or duration of complete response follow-up was 2 years, and we showed that our actual median duration of complete response was nearly 3 years of 34.2 months. So again, this supporting cohort reinforces and shows the durability of our response even after recitation of therapy. Next slide. And looking at time dividend end points, our progression fee survival was 4.9 months with a median overall survival of 11.5 months. And you can see from these figures, there was an early trend for plateaus in both our PFS and OS curves. Next slide. And looking at our safety profile, glofitamab was well tolerated with a highly favorable safety profile and looking at treatment discontinuations. AEs leading to treatment discontinuations were low at only 3.2% for patients who had related AEs that led to discontinuation. And looking at the figure, our most common AE was CRS, cytokine release syndrome, at 63%, followed by neutropenia and anemia. Next slide. Now when we break down cytokine release syndrome, as I mentioned, the overall incidence was 63%. But the vast majority of that CRS was grade 1, demonstrated by fever of 47.4% and then subsequent lower incidences as the grade progresses. The median CRS duration was 30 months, and if it were to occur, it did occur early at a median 13 hours after the first dose. Looking at the figure, you can see that the vast majority of the CRS events happened with cycle 1, followed with the significant decrease in CRS incidents with subsequent doses. Next slide. And looking at other AEs of interest, again, I'd like to reinforce our lower rate of treatment discontinuations due to AEs. I'd like to point out here, ICANS neurotoxicity was fairly low at only 3% for any great ICANS greater than equal to Grade 3 and 2.6% for ICANS of all grades. And again, the take-home message from this slide is that treatment discontinuations were low that were related to AEs. So in conclusion, with our Phase II pivotal data with glofit for relapsed/refractory patients met our primary efficacy endpoint, the CRR of 39% in this highly refractory heavily pretreated population. The CRs were durable. We saw CRs in patients with prior CAR-T therapy. It was well tolerated at low rate of treatment discontinuations, low rate of ICANS. And again, glofit is the first, first-in-class T-cell engaging CD20 by CD3 bispecific antibody to show clinically meaningful outcomes for patients in this pivotal Phase II setting for this high-risk, heavily-pretreated population. Next slide. So I'd like to also briefly mention this patient cohort from our POLARIX trial. As Elena mentioned, and just to remind everyone, POLARIX was our pivotal Phase III global trial that showed that Polivy plus Rituxan and CHOP showed significantly meaningful improvement in outcomes compared to long-term standard of R-CHOP in patients with first-line DLBCL population. We took -- we performed this Asian subpopulation analysis that included patients from the global cohort as well as a China extension for the purposes of registration in China. As you can see from the slide, the 2-year PFS was 74% for the R-CHP Polivy arm compared to 66% for the R-CHOP arm. The overall safety profile was similar between both arms. So the take-home message or conclusion from this slide that in this Asian subpopulation analysis POLARIX results were consistent with the global study with clinically meaningful improvement in outcomes and a comparable safety profile. Thank you. I'd like to now hand this over to my colleague, Johanna Bendell, SVP of Pharma Research and Early Development.

Thank you so much, Ginna, and thank you all for being here today as I share our early pipeline programs for pRED. Next slide. Here is our high-level approach to our pRED oncology strategy. First, we're bringing the T cells to the cancer via redirection with the T cell bispecific antibodies. Our approach is using a 2:1 format, which results in improved target binding and our coming generation of T-cell bispecifics will be using tumor-activated approaches to more target that immune reaction to the cancer and minimize the on-target but off-tumor toxicity. Next, we're working to reinvigorate and/or enhance T-cell attack via cytokines, co-stimulators or counteracting immune inhibition with both large and small molecule approaches. You'll see some examples and molecules discussed later in this talk. And our third approach is generating cancer-specific T-cells in situ with targeted myeloid cell agonists. Moving now to our cancer targeted therapies. We're focusing on 2 main areas, looking at inhibitors and modulators of oncogenic drivers, examples in our portfolio such as our brain penetrate BRAF inhibitor, and then synthetic lethality, namely DNA damage response, with molecules coming to the clinic later this year and into next year. Next slide. This is our early clinical portfolio. Notably, we have at least 3 new molecules due to enter the clinic towards the end of this year and early next year, including a small molecule immunotherapy asset. We also have our Phase II PD-1 LAG-3 molecule, which I'll discuss later in this presentation. This molecule, really proud to say, is an example of closer collaboration between our early and late stage as we optimize the seamless clinical development plan and transition from early to late stage. So next, I'm going to highlight some key molecules from our early clinical portfolio. Next slide. So the first molecule I'd like to show is our G protein coupled receptor family C Group 5 or GPRC5D CD3 T-cell bispecific antibody. GPRC5D is an orphan receptor that has no known function in humans and is expressed on plasma cells and is the most specific target yet known in multiple myeloma. High levels of GPRC5D can be associated with poor prognosis as well as higher tumor load in multiple myeloma patients. This molecule has, as I mentioned before, the specific 2:1 format that confers a higher avidity to its target on plasma cells, and the head-to-tail geometry of the molecule creates a higher potency than other formats. Here, us the preclinical data comparing our novel 2:1 format GPRC5D bispecific with another 1:1 format GPRC5D bispecific. On the bottom right, you can see regression of tumor volume consistently for our 2:1 GPRC5D at higher and lower doses compared to the 1:1 format bispecific, really speaking to its potency. This potency is also reflected in clinical data that's going to be presented at an oral session at the European Hematology Association conference later this month. On the upper left, you see the 2:1 GPRC5D bispecific increases T-cell activation and tumor cell lysis compared to the 1:1 format. And then on the bottom left graph, you see the potential combination efficacy of this molecule with daratumumab and lenalidomide. Next slide. Here are the clinical data that will be part of the oral presentation at EHA 2022. We started the clinical Phase I study with this molecule in November of 2020 in heavily pretreated patients with refractory multiple myeloma using both IV and subcutaneous dosing. To this date, we've treated over 100 patients. The data that will be shown at EHA are from the IV cohort with response rates, including patients that were treated at all of the dose levels at 68% with good durability. We've also seen responses in patients post anti-BCMA therapy, and we've also not yet defined our optimal dose. Toxicities are manageable, and we're working on defining our dose as well as beginning combination studies with standard-of-care agents. We're really hoping to show you more clinical data at a conference in the later part of 2022. Next slide. So while we've seen the efficacy of T-cell bispecifics in patients with lymphoma, can we actually improve on this efficacy by adding co-stimulation similar to what we see with second-generation CAR-T approaches? So here, we're providing Signal 1, which is NK or T cell activation via the CD20 CD3 T cell bispecific. And then Signal 2, which is enhancing NK and T cell activation and promoting a durable immune response by co-stimulation with our CD19 4-1BBL molecule. The addition of 4-1BB co-stimulation and CAR T approaches has actually shown prolonged and/or increased duration of responses. Our CD19 4-1BBL has a first-in-class potential for B-cell targeted co-stimulation of T cells. And the combination with the CD20 CD3 T cell bispecific is an off-the-shelf first-in-class potential to approach -- approach to deliver both the Signal 1 and Signal 2 as an alternative to second-generation CAR T cell therapy. As seen in the graph in the middle of this slide, the preclinical antitumor activity of the combination of glofitamab plus the CD19 4-1BBL molecule is actually superior to glofitamab alone. And in pictures on the right, you can see this combination shows improved in vivo T-cell effector function compared to either molecule alone. CD19 4-1BBL entered the clinic in August 2019 with dose escalation in combination with Gazyva and glofitamab in patients with mixed relapsed/refractory non-Hodgkin's lymphoma. We've enrolled 35 patients in the combination with Gazyva and 59 patients in the combination with Gazyva and glofitamab. We're expecting to see dose escalation to complete in quarter 3 of '22, and we hope to present initial clinical data at a conference at the end of this year. Next slide. So co-stimulation with CD28 is another approach we're taking to providing Signal 2. Both 4-1BB and CD28 are co-stimulators but are differentiated. For instance, CD28 is constitutively expressed while 4-1BBL needs to be activated by SIGNAL 1. This allows for co-stimulation with CD28 to be an alternative co-stimulation approach as well as opening the possibility for combining both types of co-stimulation with both 4-1BBL and CD28. This approach is also being studied in CAR T cells. On the right of the slide, we seen this orthotopic model that the combination of CD19 CD28 with a CD3 T cell bispecific shows a strong combination effect over the CD20 CD3 T cell bispecific alone. Our CD19 CD28 molecule entered the clinic at the beginning of the year, and we're hoping to see data as it emerges. Next slide. Now moving to our PD-1 LAG-3 bispecific, it's a molecule that binds to both of these co-inhibitory checkpoint receptors and was designed to improve efficacy over given the combination of 2 separate PD-1 and LAG-3 antibodies given together. Our bispecific was designed with high affinity to PD-1 and lower affinity to LAG-3. So in general, antitumor T cells have higher levels of PD-1 versus T regulatory cells. So by directing that high affinity binding to PD-1, our molecule preferentially targets and reinvigorates the tumor-reactive infiltrating T lymphocytes rather than T regulatory cells. In addition, because of binding to the tumor-infiltrating T cells rather than T regulatory cells, there's an avoidance of the immunosuppressive effect that occurs by binding LAG-3 to the T regulatory cells. So you can see on the graph on the right, we can see preclinical evidence of improved antitumor efficacy with our bispecific molecule to either PD-1 or LAG-3 antibodies alone or the combination of the 2 separate antibodies. We've treated over 150 patients with our PD-1/LAG-3 bispecific and are excited to show the clinical data at a conference later this year. We've also initiated Phase Ib/II studies in patients with esophageal cancer hepatocellular cancer and neoadjuvant melanoma. And we have studies that are looking at other tumor types that are coming shortly. Next slide. Here, you can see a scan from one of our patients with primary checkpoint inhibitor refractory melanoma, who had a confirmed partial response and the duration of response is continuing to on-go at over 60 weeks. Next slide. So IL-2 is a pro-inflammatory cytokine, which could be a much more transformative therapeutic if its side effects can be harnessed. A recent approach is second-generation IL-2 molecules has been to reduce or eliminate IL-2 binding to CD25, the alpha chain of the IL-2 receptor, which can then reduce systemic toxicity and decrease T cell regulatory stimulation. But with this approach, we also see a significant amount of activity on the tumor-specific T cells being lost. Our PD-1/IL-2B molecule is a third-generation approach, using a high affinity binding to PD-1 to localize this modified IL-2 variant to PD-1 positive T cells in tumors. Specifically with this molecule, we target the resource T cells, which are PD-1-positive -- stem-like CD8-positive T cells. So you can see in the cartoon there on the top left, when these cells are exposed to PD-1 or PD-L1 antibodies and you follow the bottom arrow here in the picture, these cells become transitory effectors. But when you're continuously exposed to PD-1 or PD-L1 antibodies, you eventually have exhausted T cells. However, if you add IL-2 effect, and when you -- like for by being exposed to the PD-1 IL-2 bispecific, so you follow along with the top arrow in this cartoon, they actually become better effector T cells, acting more like T cells in the setting of acute inflammation with a better proimmune effect. Preclinically, we see evidence of this mechanism of action for our molecule, here shown in PANGO-2 tumor models. On the top graph, you see 100% survival at the 1 milligram per kilogram dose compared to our second generation fat targeted IL2v with or without anti-PD-1. To the right and to the bottom, we see that PD-1 IL-2v increases better effector T cells without signs of T cell exhaustion, while anti-PD-1 and our second-generation FAP-IL2v, either a single agent or in combination, do not. And instead, you see here, they produce higher levels of exhausted T cells. Next slide. So we entered the clinic with PD-1/IL-2V in May of 2020. We've treated over 150 patients so far using both IV and subcutaneous dose escalation cohorts as well as biopsy cohorts. We're currently also evaluating dosing in combination with atezolizumab. These scans show a confirmed partial response in a subcutaneous lesion and a gastric wall lesion of a patient with advanced melanoma and who is primarily refractory to PD-1 as well as to anti-CTLA-4 antibodies. The duration of this patient's response is over 60 weeks. So we're hoping to show you more of the clinical data at a conference once we've completed our combination cohort. Next slide. HLA-A2 WT1 CD3 T cell bispecific uses high-affinity binding to an HLA peptide complex to target the intracellular protein WT1. So WT1 is overexpressed in AML as well as ovarian cancers. Because this molecule targets the HLA peptide complex, there is a restriction to patients who have HLA-2, but this molecule also has a very high specificity for tumor cells compared to normal healthy cells. So here, we see the preclinical activity of HLA-A2 WT1 TCB in the bar graph, which shows an antiantibody mediated T cell site toxicity against patient-derived AML cells using both healthy donor PBMCs and patient-derived PBMCs. The graph also shows the antitumor efficacy across doses in xenograft and PDX models. We started a clinical trial of this molecule in patients with AML, with 38 patients treated so far with relapsed/refractory AML and patients with minimal residual disease. We're testing also both IV and subcutaneous formulations and are excited to show you the initial clinical data at a conference later this year. Next slide. So we're also looking at the co-stimulatory molecule approach targeting solid tumors. So FAP-CD40 is one example of this approach, where FAP binding localizes the molecule to FAP-expressing solid tumor stroma. This allows for cross-linking of the CD40 agonist to the CD40 receptor on tumor antigen-presenting cells resulting in their activation. This then leads to antigen presentation specifically within the tumor. The FAP binding is required for cross-linking for the CD40 to occur, so there's really no systemic CD40-induced myeloid cell activation. Previous CD40 molecules given systemically showed primarily liver toxicity, which limited dosing and then, therefore, potentially clinical effect. The bar graph shows the preclinical tumor-specific CD40 activation with this molecule as compared to systemic CD40 activation. And then these PET scans are from our imaging substudy of FAP CD40 showing tumor-specific uptake of the labeled FAP CD40. We've dosed 44 patients so far with this molecule, and are still in our biomarker cohort. We hope to share our clinical data at a conference in 2023. And of note, we have other FAP targeted co-stimulatory programs in the clinic, including a FAP-4-1BBL which is currently in dose escalation in combination with our CEA CD3 T cell bispecific sovesatimab. Next slide. So this presentation has just been a piece of the work we're doing within pRED Oncology. These molecules are examples of drugs that are being developed within our committed focus areas that we discussed at the beginning of our presentation. Emerging treatments in this area represent the bulk of our portfolio. However, to remain agile and continuing to assess innovative new modalities, we also have designated a smaller part of our portfolio to both exploratory as well as opportunistic areas, primarily through collaborations and external partnerships. We look forward to sharing the clinical data from these and more of our molecules together with you in the near future as well as our next generation of molecules coming to clinic. So with that, I turn to Andy Chan, who is going to discuss our gRED portfolio. So thank you very much.

So thank you, Johanna. So I'm delighted to continue our discussion this morning and share with you an update on the gRED strategy in oncology. Next slide. Our strategy entails integration of 5 biological pillars, targeting oncogenic pathways, synthetic immunity to directly harness T cell immunity, enhancing adaptive and innate immune functions to particularly target cancer neoantigens, overcoming the immunosuppressive tumor micro environment, and identifying and targeting dysregulated survival lineage development of programs that have been hijacked to promote oncogenesis. In addition, using large data sets generated from both our preclinical and clinical studies, we're applying advanced computation analytics to generate new insights for target discovery, drug discovery and clinical development. Next slide, please. So let's begin with the longest-standing pillar, targeting oncogenic receptor tyrosine kinases in the downstream oncogenic signaling nodes. Next slide. Roche has a broad and deep portfolio targeting these major oncogenic drivers. In addition to inhibition of driver receptor tyrosine kinases, we have developed a large armamentarium of small molecule inhibitors directed against each of the intracellular RAS-activated signaling nodes. This armamentarium enables identification of the optimal combination strategies to target crosstalk, feedback and feed-forward signaling as well as resistant mechanisms. And today, I'd like to share with you 2 examples of how combinations may provide greater patient benefit. Next slide. So let's begin with our KRAS G12C inhibitor, GDC-6036. 6036 represents a potential best-in-class covalent inhibitor of mutant G12C KRAS. Its preclinical profile includes greater potency and selectivity than other G12C inhibitors. 6036 has wider preclinical safety margins, and we are encouraged by our Phase I experience, which we plan to present, I believe, in -- later this year, I believe, it's in August of 2022. 6036 has advanced to our late-stage colleagues. And with Elena, we will be initiating a pivotal Phase III trial in second-line non-small cell lung cancer later this year. In addition, we're initiating combinations with checkpoint inhibition, multiple inhibitors of additional oncogenic pathways as well as with anti-angiogenic therapies. Next slide. Another major oncogenic node is the SHP2 protein tyrosine phosphatase upstream of RAS. Inhibition of SHP2 attenuate RAS is signaling and can indirectly augment antitumor immunity. In collaboration with Relay, we are advancing GDC-1971, an allosteric small molecule inhibitor of SHP2. 1971 attenuates MAP kinase signaling and cellular proliferation. And since it maintains KRAS in the GDP-found form, 1971 can also enhance KRAS occupancy by 6036. And the combination of 1971 and 6036 cooperate to synergistically enhance antitumor efficacy. Our Phase I combination trial was started in 2021, and we look forward to the clinical data generated with this combination. Next slide. A third oncogenic node is RAF. In combination with Hanmi, belvarafenib is a type-2 RAF inhibitor. Belva inhibits both mutant RAF and mutant RAS-driven RAF dimers and mitigates the paradoxical activation observed with Type 1 or Type 1.5 RAF inhibitors. In our Phase Ib expansion of patients with mutant NRAS melanoma, the combination of belva and cobimetinib, our MEK inhibitor, resulted in responses in 5 of 13 patients, including 5 of the 11 checkpoint inhibitor experienced patients. And the combination to date has been well tolerated, and we're continuing studies in mutant NRAS melanoma, as well as further signal seeking using our TAPISTRY trial platform. Next slide. Our second scientific pillar focuses on synthetic immunity and is based predominantly earlier on, on our bispecific antibody platform to engage T cells directly to tumors. And we have recently also initiated efforts to engineer T cells to target cancer neoantigens, which I will discuss later. Next slide. We're encouraged by our promising clinical data with mosun and look forward to bringing this new drug to patients with B cell lymphomas and leukemias. We're also advancing cevostamab, our FcRH5-targeted T cell bispecific in multiple myeloma. And to date, we have a 56.7% observed response rate in patients with relapsed or refractory multiple myeloma at our 132- to 196-milligram dose level, and we're also currently exploring subcutaneous formulations and combinations. In collaboration with Immunocore, we have generated interesting early data with our MAGE-A4 T cell receptor-targeted bispecific in ovarian cancer. And we will continue to explore the potential of this platform that has the interesting capability of targeting intracellular tumor-specific antigens. And finally, we're advancing bispecific antibodies in solid tumors, including HER2-positive breast and gastric cancers as well as a new agent for colorectal cancer. Next slide. Our third pillar to augment productive adaptive and innate immune cell functions includes our major efforts to target cancer mutant neoepitopes. Next slide. Here, we have taken a broad multimodal potentially complementary approach to generate robust T cell responses to neoantigens. In collaboration with BioNTech, we're advancing an RNA-based neoantigen vaccine. In collaboration with Nykode, we're advancing a DNA-based neoantigen vaccine and we have also begun a cell-based synthetic immunity approach for neoantigens in collaboration with Adaptive and Adaptimmune. We're identifying T cell receptors specific for neoantigens in collaboration with Adaptive and are developing an allogeneic iPSC-derived T cell platform in collaboration with Adaptimmune. And this off-the-shelf approach may greatly enable personalized T cell therapy in a time-efficient manner. Next slide. Now returning back to our RNA vaccine collaboration with BioNTech, we're quite excited about the Phase I study that I believe was just discussed yesterday by Dr. Balachandran from Memorial Sloan Kettering. In the single-center study in adjuvant PDAC of 19 vaccinated patients also receiving atezolizumab, followed by mFOLFIRINOX regiment, 50% of patients generated a polyclonal gamma interferon-producing neoantigen-specific CD8 T cell response. And in an early median follow-up of 15 months, vaccine-induced immunity appears to correlate with improved relapse-free survival. And we're now planning a randomized Phase II trial to extend these observations. Next slide. Since the success for personalized cancer immunotherapy requires accurate prediction of neoantigen immunogenicity, we are fortunate to have assembled a large experimental data set of immunogenic peptides presented at MHC from a number of different cell lines. And using machine learning, we have improved the performance of immunogenicity predictions using our Genentech Recurrent Attention Framework algorithm, or GRAF. GRAF outperforms 2 state-of-the-art immunogenicity prediction algorithms assessed by 3 different metrics, and we are currently utilizing GRAF to support our present Phase II clinical trials. This is one of many examples where data and computation is accelerating our predictive efforts in both drug discovery and development. Next slide. And the last of our biological pillars that I will share with you today is targeting the immunosuppressive tumor microenvironment. Next slide. Through Ira Mellman's pioneering work, we have described 3 tumor immunophenotypes, and most clinical success has been achieved in the immune-inflamed tumors. However, half of tumors are immune-excluded where CD8 positive T cells predominantly reside in stroma and unable to efficiently infiltrate tumors. And through our preclinical work, we've discovered that one of the mechanisms by which the stroma contributes to this immunosuppressive phenotype is through the TGF-beta family of cytokines. Next slide. TGF-beta 1 is one of 3 TGF-beta isoforms and is regulated by both proteolytic and integrin-mediated activation that give rise to its mature active form. In collaboration with our Chugai colleagues, we are advancing an anti-TGF antibody that selectively binds and neutralizes an inactive latent form of TGF-beta 1. And in preclinical models, blockade of latent TGF-beta 1, in combination with anti-PD-L1, reduces both tumorigenic and immunosuppressive functions of TGF-beta 1, increases T cell infiltration in tumors, as you can see in the right panels here, and finally, enhances anti-tumor immunity and induces tumor progression. And we're looking forward to our data from our ongoing Phase I clinical trial. Next slide. So in conclusion, thank you for letting me share with you a glimpse of 4 of the 5 major approaches that we are undertaking in our gRED early oncology portfolio. Importantly, our ability to combine not only within each biological axis but across the biological axes provides multiple opportunities for us to provide the greatest clinical benefit for patients. And with that, I'm going to return the podium back to Bruno.

Thanks, Andy. And with that, we come already to our Q&A session. For the Q&A session, we are joined by a few additional colleagues and experts in the room or online, which I would like briefly quickly to introduce. So we have with us today, Charlie Fuchs, Global Head of Hematology and Oncology Product Development; we are joined by Peter Ahnesorg, Franchise Head, Hematology and Global Product Strategy; by Heather Stevens, which is -- she's our life cycle leader for tiragolumab; by Pablo Perez Moreno, which is our global development lead for giredestrant; and then we are joined for the early development side by Ira Mellman, Vice President, Cancer Immunology at gRED; and by Pablo Umaña, Head of Cancer Immunotherapy Discovery at pRED. And I think with that, we are set to take the first question. The first questions would come from Wimal Kapadia from Bernstein. Wimal, please?

Great. I'm Wimal Kapadia from Bernstein. So can I first just ask about inavolisib, please. Firstly, what gives you confidence that you have a superior molecule to PIK3? I think one of the challenges for the product in terms of uptake has been safety. So I'm just curious how you think about your safety profile with respect to PIK3. And then just tied to that, first-line usage, is that a realistic outcome for this product? Or will the product will maybe be used to mainly mop up resistance in later lines? And then my second set of questions around the strategy for early breast cancer for giredestrant and the oral SERDs. So you seem to be running the adjuvant trial, excluding CDK4/6, but then in the neoadjuvant setting, you've run a combination with palbo . So I'm just curious a bit to understand the rationale for this. And just tied to that, if we did see success for the CDK4/6 in the broad adjuvant population, i.e., the Kisqali NATALEE study, would that change your thinking about the trial design in the adjuvant setting?

Thanks, Wimal, for the questions. I think the first question -- actually, both questions, I think would go to Charlie. And maybe for the second question on giredestrant, Pablo, you could join in.

Yes, I can certainly start with inavolisib. The preclinical data for this molecule have really convincingly shown it to be more specific, and frankly, more potent than the other available molecules on the order of more than 50x more potent. And we actually -- the data for its specificity are, I think, both impressive preclinically and clinically. As you know, in this class, the major problem has been tolerability and the rates of treatment discontinuation with other molecules in this space have been high on the order of 20%. So it's very difficult to continue treatment and we believe the duration of suppression is critical. Our early data would suggest that our treatment discontinuation rates are really quite low in the low single digits. And so we think both its increasing potency and specificity make it a best-in-class molecule, and that's why we're advancing it in multiple settings. In terms of the ability to really gain traction clinically in the community, I think the issue has been tolerability among practitioners and patients. Given the tolerability of our molecule, we're pretty confident that actually will be embraced by patients and practitioners. And I think given the ability to target PIK3CA and given the increasing utilization of tumor sequencing, we think that sort of all efforts are aligned with our development program to really make it a standard of care.

So for the giredestrant question, I'm going to start with the question on coopERA. So coopERA had 2 components. It had a window component, which was monotherapy for 2 weeks, which was the primary end point. And then it had the combination component, which I think you were referring to in combination with palbociclib. So in both components, we've seen a superior K67 reduction. And I'm also going to link to the question on the adjuvant setting, why we're doing a monotherapy study while the landscape is evolving with the advent of CDK4/6 inhibitors, as we know, for Monarch-E and the upcoming potential data for NATALEE. So our vision as a molecule is to become a new backbone of therapy. We know even with the advent of CDK4/6, that majority of patients are still going to receive monotherapy, and we believe that the future is still going to be mainly dominated by monotherapy. So by that, we believe that by bringing a new and a more potent and more tolerable option, that's going to impact the majority of patients. We can't ignore the future, and we're also thinking and considering what other options and combinations, and those are the things we might incorporate in our future development program.

Thanks, Pablo. Wimal, did we answer all your questions? I hope so. I think then we would move on to the next one in the row, which would be Matthew Weston from Credit Suisse. Matthew, please.

Two questions, please. The first on CD3, CD20s, there's obviously a huge amount of interest in what's a competitive market, where there's a lot of focus at the moment in late-line data. But the real commercial opportunity, I'm sure you'll agree, is in earlier lines. So I would love if the team could remind us of the strategy for Glofit in earlier lines of DLBCL and what combination partners you're thinking of using. And then the second question is around second-line lung cancer. There was obviously a great deal of interest in tiragolumab ahead of this meeting, and we're all looking forward to seeing the numerical benefit at ESMO and the OS data next year. But you do have a relationship with Exelixis on cabozantinib, where the CONTACT-01 Phase III study is due to report later this year. And I think there was a relatively encouraging Phase II cohort from COSMIC-021 that were presented at the ASCO meeting. I'm just curious, why don't Roche ever talk about it?

Thanks, Matthew, for the questions. The first questions, I think, goes to our heme specialists.

Yes. This is Ginna Laport. Thank you for your question. Regarding Glofit in the earlier lines of therapy, yes, we are currently evaluating it in second line and plans to do in the first line. We have a confirmatory trial in the second line with [indiscernible] R-GemOx versus Glofit-GemOx for second-line DLBCL, and we are planning to take it the first-line study for first-line DLBCL using the POLARIX backbone, and that will soon to come.

Thanks. And the second question on our collaboration with Exelixis, and you are right, Matthew, by the way. I got the question several times in the last week. This is maybe a question for Charlie to give -- provide a bit more details?

Yes, absolutely. No, I think we're very encouraged by the possibility of this combination being the utility for patients in second-line lung cancer. So the notion that we're underplaying it, certainly, it's not by intent, and we anticipate getting data from that trial hopefully later this year. So it's something we really think has the potential to be important in the second line treatment of lung cancer. Clearly, patients with lung cancer who progress beyond the first-line need more options, and we're looking forward to sharing those results when they become available.

So thanks, Charlie. So no downplaying here and another potential upside. Matthew, did we answer your questions or any follow-on questions?

Perfect. It's great.

Next one would be Luisa Hector from Berenberg.

Bruno, thank you, and thank you for all the details today. I guess I'll ask two. On Glofit, maybe the issue around safety to CRS. Are there any particular patients most at risk? And do you see -- are you worried that as you move into a more real-world, larger population that the CRS could increase? Or do you think that your dose titration strategy really has this under control now? And then perhaps to go to the multiple myeloma. So the GPRC5D what toxicities, in particular, should we look out for? You talked about them being manageable? And how soon could you have pivotal data that you could file on? And maybe I just wonder how this one fits together with cevostamab as you have the 2 approaches there. So just how do you see those being positioned together.

Thanks, Luisa. Who wants to take the Glofit CRS question?

This is Ginna Laport again. Regarding safety and CRS for glofitamab, your question about high risk, who are the patients at risk for CRS. Not surprisingly, we have seen data where patients who have the high tumor burden. But we have -- we feel we have adequate mitigation with our dexamethasone, our Solu Medrol pre-medication as well as step-up dosing as we only require monitoring for the first dose. And as we showed in our data set that CRS is highly predictable. It comes with a first dose, and it's a fairly short duration. As far as -- so again, between the dose titration, and we also have data using specifically dexamethasone, that appears also to better mitigate CRS compared to methylprednisolone. So we're not concerned about our safety profile of Glofit, and we think it's highly promising for the outpatient setting.

Thanks, Ginna. And then we have the question on multiple myeloma. And I think the question was on the development pipelines. Johanna?

So the first was the GPRC5D toxicity, I believe. So I think what we've seen is very much in line with what other GPRC5D CD3 bispecifics have shown, which is an on-target off-tumor-related mild skin rash. We've also seen some cytokine release syndrome, which is actually very well tolerated in regards to us using a step-up as well as subcutaneous dosing. And so hopefully, you'll see some more details on that. But nothing to suggest a significant difference from what's been reported with other molecules in that same setting. And I'll turn to Peter now for the regulatory.

Yes. Thanks, Johanna. Peter Ahnesorg, Global Franchise Head for Hematology. So I mean, we can't comment on any specific time lines or pivotal data yet. But you asked a question about having both cevostamab and the GPRC5D in our portfolio, and I think we believe that's a real strength. Multiple myeloma is a big space, as Elena alluded to in the beginning, and we believe there's much more molecules needed in order to address patient needs. And these two, we believe, are incredibly promising. And we look forward to presenting data next week at EHA on the GPRC5D for the first time.

Thanks, Peter. Luisa, did we cover your questions?

All clear.

Then the next one in the row would be Simon Baker from Redburn.

Two questions, if I may. Firstly, going back to the SKYSCRAPER-02 presentation. The discussion mentioned the role of ADCC with tiragolumab. Was questioning whether it was a help or a hindrance. So I just wanted we'd get your perspectives on that. And then secondly, moving on to the TGF-beta. Merck KGaA on Friday, where I asked them about their own experiences with bintrafusp setting. And they said that with the benefit of hindsight, the biology of TGF-beta was probably not as settled and understood as they had originally thought. I just wondered if you could get your perspectives on the -- your handle on the biology and the challenges of the biology of TGF-beta in this setting.

Thanks, Simon. I think the first question goes to Ira, our expert on tiragolumab.

Well, thank you for that title. We do find a dependency on having an Fc domain present. There's some preclinical studies that clinical data are obviously not available yet. But we don't think that, that really reflects the contribution of ADCC either to deplete tumor cells or to deplete T cells, but rather they're playing a role in activation of the myeloid cell compartment [indiscernible] to an enhanced T cell response. So if there is a contribution of ADCC, we have not discovered that as yet.

And then we had the question on TGF-beta and what do we know about the biology.

There -- I mean, certainly, Merck KgaA is correct, the biology, immunology, cell biology TGF-beta is nothing, if not complex. But I do think we know enough actually to proceed. So for example, one of the big differences between the approach that Andy Chan described versus what Merck KgaA did was to go after in our case TGF-beta at its site of generation because one thing that has been learned over the last couple of years is that this particular cytokine can be active at its site of generation as opposed to just cytokine that had been previously generated and is circulating in blood. Predominantly what agents, such as the Merck KgaA agent did was to go after preformed soluble cytokine. And we believe that may not be the optimal pool. So they may have done a great job at sequestering soluble cytokine, but not so much given the other characteristics of that molecule to address TGF-beta at its site of action, which is in the tumor.

Okay. The next one in the row would be Keyur Parekh from Goldman Sachs.

Two, if I may, please. The first one is you guys have spoken twice or thrice in this presentation about seeing biological activity when you were talking about for giredestrant and then separately on the SKYSCRAPER-01 study as it relates to both PFS and OS. Just wondering kind of does it not historically done that Roche or Genentech have used either as studies worked or it's failed. So just wondering kind of why you guys are using those terms to describe these studies? And what is the inherent message that you are trying to tell us there? That's question number one. Question number two is there was a presentation yesterday with the statistical plan for SKYSCRAPER-02. Is there a reason why the statistical plan for SKYSCRAPER-01 should be meaningfully different to the one that we saw for SKYSCRAPER-02 in terms of the alpha spend on the various end points?

I think the question on tiragolumab will go to Heather Stevens. Why are we still optimistic and why is every positive outcome for tiragolumab for where we are right now upside?

As we have previously stated, in frontline non-small cell lung cancer, we do think OS is really the most important end point. And in fact, just this morning at ASCO, there was a pooled analysis presented by FDA that shows that neither at the patient level, nor at the study level, PFS is not a great correlate for OS. And so with SKYSCRAPER-01, we do have the study optimized to detect OS. The language that we've used in seeing the numerical difference between PFS and OS does give us some comments in the biological activity of tiragolumab, coupled with data that we've seen across the Phase I study, which is ongoing as well as the CITYSCAPE study.

And a similar question here on giredestrant. Why do we believe any biological activity from the neoadjuvant study and the second third-line study as well?

On the -- as we mentioned before, in coopERA, we've seen a clear biological signal. And this is also supported by some historical data where we see there's a correlation between the degree of Ki67 suppression with outcomes in the adjuvant setting. So that is a strong reason to conduct the lidERA study. When it comes to acelERA, we cannot be too specific on the answer. We're going to present the data later this year. But it's a very different patient population. It's very heterogenous. The majority of patients end up in [indiscernible]. But in those patients who we believe are still endocrine sensitive, we see a clear signal of superiority of giredestrant versus standard of care. Although as we mentioned before, in the ITT patient population, we haven't met the primary endpoint, and this is a signal that validates our support and the conduct of our [indiscernible] studies.

So for both molecules, the race is open. I don't know, Keyur, do you have any follow-on questions?

No, that's great.

The next one would be Stephen Scala from Cowen.

I'd like to follow on Keyur's question, and that is, can you qualitatively describe in a bit more detail the positive PFS trend in SKYSCRAPER-01. For instance, would you describe it as modest? Would you describe it as impressive despite missing or somewhere in between? And then second, how is Roche positioning Tecentriq versus Opdivo in resectable lung cancer, are you seeing adjuvant Tecentriq used, for instance, after neoadjuvant Opdivo?

Given the first question was a good one, I think we cannot really provide an answer here according to this classification. I think there is not much more we can share with what Heather just basically said on the PFS. So numerical benefit, the curves are separating and you know the story about the alpha spend. On the second question, I think this would go to Charlie about the evolving treatment paradigm in the adjuvant in non-small cell lung cancer setting as -- yes, what we know today.

Of course. Well, obviously, as you know, in parallel and always the first study to show the benefit of immunotherapy, cancer immunotherapy in the curative setting of lung cancer and recognizing that other studies emerging, including the one you alluded to in the neoadjuvant space. We have our own study, Elena mentioned it, IMpower030 looking at the neoadjuvant therapy. And albeit at the moment, I think postoperative therapy is really principally the approach that most practitioners are taking. We are keen to look at the opportunity of providing the neoadjuvant option to providers and patients. And we anticipate, as Elena mentioned, that additional study with Tecentriq in the neoadjuvant setting reading out later this year. So it's difficult to compare our study comparisons, but obviously, to build on the case for Tecentriq postoperative, we're looking forward to providing a neoadjuvant data set to follow soon.

Stephen, did this answer your second question?

Yes.

Next one would be Michael Leuchten from UBS. Michael?

So just one question going back to SKY 01. We obviously know about pseudoprogression in IO. Why take PFS into the study as a co-primary endpoint? Why not just go with OS and spend any alpha on PFS? And then a broader question, if I could. You talked about the complexity of biology for TGF-beta. I guess, the tiragolumab trial results also may suggest that the biology is not that well understood either. I talk about bispecific and bispecific combinations, the permutation of the risk factors here are quite extraordinary. What does Roche do to make sure you're not taking too much risk into later stage settings with bispecifics that depend on antibody constructs, ratios and the biology?

Thanks, Michael. I think on the first one, SKY 01 pseudoprogression, whether including PFS, was the right choice. I don't know whether we can add anything here. Heather?

I guess, just maybe one comment. PFS was included in this study as it can sometimes be used for a faster access for patients and a path to accelerated approval. And so we do include it in many of our studies for that reason.

And the second question, I think, is a bit more maybe of a philosophical nature. The broader question whether we -- when we create now even more complex fusion proteins and bispecifics and sometimes even not fully understand the underlying biology of disease pathways, does this change somehow our chances of success? I don't know any one of the pRED, gRED people here in the room might want to start or comment. Pablo?

Yes, going to start just with the ones that we have more advance. In pRED, what we have taken is basically a more mechanistic approach on what is already understood and what are the key signals that drive T cell activity and durability of T cell activity, which have been kind of indirectly in a way, started already before by second-generation CAR T-cells. So combining those 2 pathways [indiscernible] first signal 1 with a bispecific and then signal 2 with a costimulatory agonist. And there have been previous costimulatory agonists in the clinic before the first generation ones that were not targeted, and they had a lot of issues of side effects because of their systemic activation. So we have learned from that, and so it's kind of a stepwise approach where we have now removed that systemic activation and we have designed the molecule to be active only at the site where it is needed. So it's -- I think we're -- that's how we are doing this, and we are testing it now in Phase Ib, both in hematological tumors and in solid tumors.

And I think in a way, we have a similar story to tell with the Interleukin 2, where we have basically now a third-generation molecule.

Correct. Correct. Equally. So as Johanna explained, for the IL-2, it has been a long learning process, but I think we are steadily making progress. We have learned what are the things that we want to remove from the things we want to add, and that's what we're testing now in the Phase I with the third-generation PD-1.

I don't know any one of the other scientists in the room want to add some thoughts. Andy, maybe?

Yes, I'm sure I'm not a hematoscientist. But nonetheless, I think in addition to what Pablo said regarding the mechanisms, I think it's also extremely important to understand the identity of the molecules that are within the tumor bed. So this is where a significant amount of resources from both RED as well as collaboration with late stage is trying to understand what is the nature of the cellular infiltrates that are within the tumor in the stroma because in the end, it is an issue of real estate. It's very complex as the question alludes to, but it is -- we have the tools to be able to understand who the players are. And these are areas that we are all expanding a significant amount of investment in.

And I think also it's probably fair to say many new technologies which come here to play. Okay. Thanks, Andy. I don't know, Michael, did we cover your questions?

Yes.

So then we would move on to Andrew Baum. Andrew from Citi?

So the first one is while PFS is not a terribly good surrogate as noted, response rate historically has been. You have looked at response rates within SKYSCRAPER-01, but yet you have declined to comment on qualitatively a numerical difference and whether material despite having the CITYSCAPE data where it was very impressive. I understand that there's a secondary endpoint. But given its value as a surrogate, it would be helpful if you could share any color on response rates. And then second, I know there's a 100-patient Chinese tail that was not included in the ITT sample population. Obviously, therefore, that would reduce the powering compared to what was posted on clinicaltrials.gov. I'm just curious as to design, why not include them as part of the ITT in order to increase the sample size I assume it's part of the bridging strategy. Any thoughts?

Thanks, Andrew, for the questions. I think, Heather, they would probably both go to you. Anything to say on response rates?

Yes. As you know, response rate is a secondary endpoint. And so at this interim analysis, we are not reporting it out. I would just refer back to, again, the FDA-pooled analysis that's being shared today, which actually shows that ORR and CIT in frontline non-small cell lung cancer, there is not a close correlation with OS looking at the data set that they're sharing. With regard to your second question about the Chinese extension cohort. That particular patient sample has a separate statistical analysis plan. And so what we're presenting in our primary analysis set, which was the subject of this first readout for the final PFS analysis and first interim OS analysis, we feel that we are powered appropriately.

Andrew, did this answer your questions?

Yes.

Next one would be Peter Welford from Jefferies. Peter, please?

I just wanted to go back to glofitamab, if I can, please. Just firstly, with regards to the duration of the complete response, obviously, still ongoing. I wonder if you can comment in terms of your thinking behind the fix duration of dosing with the drug? And I guess, why necessarily to show to choose 12 months and how that could potentially impact the duration of complete response? And can you [indiscernible].

You unmuted yourself -- you muted yourself for a few seconds.

Sorry. I think you've got it in terms of also if you could possibly give what proportion of patients did actually complete if you have the full cycles of therapy. So obviously, presuming not every patient, meaning of those who in the complete responses or did all those patients complete this fixed dose, this fixed course of therapy? And then secondly, on Glofit, wondering potential use beyond the diffuse B-cell and follicular, particularly CLL, but also any other potential leukemias lymphomas and any thoughts perhaps on the CD20 bispecifics in those indications.

Ginna?

This is Ginna Laport. Regarding fixed duration, we strongly believe in fixed duration. Fixed duration extends the treatment-free period for patients, which reduces its risk -- their patients' risk for long-term side effects. Based on our data with fixed duration, our data is strong. So we believe that fixed duration is beneficial for both patients and for society. As far as CLL, other indications. You also mentioned -- you specifically asked for follicular lymphoma. We do have that and are exploring -- have a study ongoing with Glofit and follicular lymphoma. But we're still waiting for results to read out. And right now, that's not our priority for glofitamab.

Sorry, CLL, any thoughts on CLL, chronic lymphocytic leukemia, is that possibly an indication you're considering?

Currently not under consideration, no. Not yet. We are actually exploring most inotuzumab right now for CLL or other CD20xCD3 bispecific and the subcutaneous formulation.

Peter, any more questions? Okay. Then I would move on. The next one would be Emmanuel Papadakis from Deutsche Bank.

Apologies, maybe I could take a follow-up on SKYSCRAPER-02. I think we get the point on the lack of PFS/OS correlation, but presumably, if PFS should be directional plus there was a total lack of OS curve separation SKYSCRAPER-02 as we saw this weekend. So is the tumor biology so different in non-small cell, we should expect to see some meaningful synergistic activity there, and that clearly was not the case in small cell. And then perhaps you could just remind us the time lines on the SKYSCRAPER-06 study, the timing for an interim decision and potential Phase III expansion decision. That would be helpful. And then on the CD20 bispecifics, you didn't mention subcutaneous time lines for glofitamab. Perhaps you could just remind us where you're at on that? And are you expecting that will effectively solve the step-up dosing issue? Or will you need titrated dosing with the subcutaneous format as well? And perhaps a final one on the neoantigen [indiscernible] data. You saw pretty durable responses, but only in the half of patients that responded. Do you have any idea of how you will be able to identify those patients in advance? And what we should expect indeed to see with the melanoma data later this year?

Thanks, Emmanuel, 4 great questions. I would give the first one maybe to Ira, the biological differences between small cell lung cancer and non-small cell lung cancer.

Those 2 disease indications, they both -- all they have in common is that they take place in the lung. I think the cell types of origin are quite different. We classified small cell as effectively a neuroendocrine type tumor which is -- has an entirely different pattern of immunosuppression from almost all [indiscernible] tumors, definitely non-small cell lung cancer. So I don't think one biologically or immunologically reads out on the other.

And we had a question. I think, let's do the science part. First, maybe on the neoantigen data of the RNA vaccine. Andy?

Yes. So I think the data set is still quite small. We are -- we don't have any smoking gun explanation pretreatment as to how can one predict response, but the data so far does indicate that there's a very good correlation between the generation of a neoantigen immune response with the clinical response. So I think the other way to look at the 50-50 result is how do we enhance the immunogenicity predictions, enhance the immune response so we can have a larger percentage of individuals that will get that response.

Then we had questions in hematology. Peter, do you want to take the first?

Yes. So subcutaneous, obviously, is an important way to administer drugs. And we're looking at subcutaneous administration for both mosunetuzumab and glofitamab. However, what's really the point here is to try and get these drugs into the outpatient setting and to get away from excessive hospitalization. And we feel we're -- the data we just presented here at ASCO, for example, are extremely encouraging, and they're showing that CRS is easily managed even with the IV administration of glofitamab. And if you've seen our data, for example, in combination with R-CHOP in the first-line DLBCL setting, you will see that the CRS rates are even lower than what we're seeing in these refractory patients. So we think the key here is really going to be to find ways of giving these drugs that will allow patients to be in the outpatient setting. And then down the road, if subcutaneous might further add to that, I think that's great, and we're exploring that.

Do we have one more -- was there another question? No. Okay. All answered, Emmanuel?

Yes. The only question was expectations with the neoantigen melanoma data we'll see later this year, if you can provide any comment. Otherwise, all done.

Andy, maybe on the timing for the melanoma data?

I believe it's later this year.

Okay. So we can confirm, Emmanuel. So I think with that, we are already at the end of our session here. I would also like to thank again all the speakers for their time and commitment, and also the IR team members who have worked on the individual slide decks. First to mention Sabine Borngräber, who had the overall event lead and also had the lead on working on Elena's part. We have Gerry Tobin, who was responsible for Ginna's part, and then [ Lillet Eser ], who supported Johanna. And [ Lauren, Carmen and Ditang ], who supported Andy. And last but not least, I would also like to thank Melanie Wolf who is responsible for organizing the event and setting up everything here at ASCO. I hope the event was helpful in summarizing our oncology portfolio, especially providing you some new insights into our exciting early late-stage pipeline. If you have any remaining questions, please reach out to the IR team any time. And I wish you a good day, and talk to you soon. Bye.