Roche Holding AG (ROG) Earnings Call Transcript & Summary

My name is Henrik, and I'm 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 Bruno Eschli, Head of Investor Relations. Bruno, the stage is yours.

Thanks, Henrik, for the introduction. Can we please go to the agenda slide, please? So just here for the agenda today for this neurology call focusing on the data we presented at CTAD, we have three speakers today with us. The first speaker will be Azad Bonni, our Senior Vice President and Global Head of Neuroscience and Rare Diseases at PRED. Azad will provide us an update on our Alzheimer's disease pipeline. The second speaker is Luka Kulic, our therapeutic area Leader, Dementia and expert Medical Director in Neuroscience and Rare Diseases at P-RED. Luka has presented the trontinemab data yesterday at CTAD and will again take us through this data today. And then the third speaker will be Margherita Carboni, indication Lead, Neurology and Clinical Development Lead in Neurology, Endocrinology and OB/GYN at Roche Diagnostics. She will provide us an update on the Elecsys biomarker data, which also were presented at CTAD. We have scheduled 40-45 minutes for these presentations. So we have 15 to 20 minutes for Q&A. And in case there would be more questions, I think the speakers just confirmed that they would have additional time available, so we could go a bit over. Can I please go to the next slide? Just to remind you a bit here on the building blocks of future growth for Roche and that neurology plays here an important part, but -- already with the existing portfolios, but then also with additional opportunities coming in the mid- and long term. You see on the left side with OCREVUS and Evrysdi, I think we still have significant opportunities. Elevidys, we have filed, as you know, in the European Union, expect approval next year. And then you see with the midterm opportunities, five molecules flagged here, Fenebrutinib and multiple sclerosis. Trontinemab, we will talk about today. GYM-329, our anti-latent-myostatin antibody, then Enspryng playing in two indications. MOGAD and autoimmune encephalitis and then prasinezumab in Parkinson's disease, where we expect the results in the fourth quarter. You also see on the box to the right, which is the potential launches 28 plus. The gamma secretase modulator. This is a molecule, I think, we have just taken into Phase II. I'm also excited about, and this is a molecule Azad will touch upon in more detail. Can we please go to the next slide? This is just maybe for the outlook into 2025, the key news flow and just highlighted here the neurology assets where we will have data and updates to be provided next year. Elevidys, I mentioned, the DMD is filed. The filing is progressing as expected, and we will get the -- hopefully, the approval next year. Then you see with -- in respect to the clinical results, a couple of readouts from neurology. Ocrevus High Dose here to be named first. If Ocrevus High Dose would be positive, I think this would be another significant opportunity, further strengthening the Ocrevus franchise. If positive, this would allow the possibility to develop high dose in [indiscernible] as a subcutaneous formulation as well, which would need an additional device. And in case Ocrevus high dose would be positive, we will also provide you an addition -- an update here in '25 on these additional development steps possible. Fenebrutinib, you see here RMS/PPMS. These are the MS studies, which we expect for the second half of next year. And if we go down, you see trontinemab here listed. This is the Phase Ib/II brainshuttle AD study we talk about today. We expect the final readout for this study with over 200 patients in the first half of next year. This will be safety and plaque reduction data. And in case these data would be positive, this would be the basis for our decision then to take this molecule into a pivotal Phase III program. That program would be scheduled to start in the second half of '25 with the first patient in and would have the final readout scheduled for 2028. Below, you see then two studies reading out for GYM-329, our anti-latent-myostatin antibody in two neuromuscular diseases in SMA and FSHD. And what is not shown here also, we have currently a third study ongoing. This is a Phase I dose finding in obese patients. This study was started with the first patient in, in the second quarter of '24. We expect, for this study, the data later this year. And we will -- we not plan to make this data public or share them. You will also not see this study anywhere listed, but the data which we will generate here will inform us of the next development steps, and we would expect that we kick off a Phase II development program in 2025 for the anti-latent-myostatin antibody in obesity. Can I have the next slide, please? Just to remind you, today, I think we have the neurology CTAD event, and then there's one more event to come here. This is the Digitalization Day, which is scheduled for November 25. This time, our Digitization Day will only focus on computational biology, AI, machine learning in early drug development. This will have a presentation given by Aviv Regev, our Head of GRED, and she will provide an update on Genentech's Lab and the loop concept. You will also have the opportunity at this event to ask questions. And with that, let me hand over to Azad Bonni for providing us an update on the Alzheimer's pipeline. Azad, please?

Thank you, Bruno, and good afternoon. As Bruno mentioned, I lead neuroscience and rare diseases in P-RED and this next 20 minutes or so, I'm going to tell you -- give you an update on the neurology pipeline in the early stage as well as the late stage and as Bruno mentioned, focusing on Alzheimer's disease. So as you know, Roche leads in the field among industry in the field of neurology. And to match that success that we have in R&D, we are taking advantage of advances in human genetics, disease biology understanding and technology in neuroscience therapeutics to establish an industry-leading portfolio. A number of examples are shown here in the disorders listed there, multiple sclerosis, neuromuscular disorders and neurodegenerative diseases. And I just want to say a couple of words about a few of these. So Bruno talked about OCREVUS, where, of course, it's established as treatment for relapsing MS and primary progressive MS, and we're doing more there. Building on that success, we are also, of course, interested in fenebrutinib, which is a BTK inhibitor that has potential to be best-in-class among BTK inhibitors. And three Phase III trials have completed recruitment in relapsing -- two in relapsing MS and one in primary progressive MS. In the neuromuscular disorders, I'm going to tell you a little bit later about GYM-329. Elevidys, of course, is a gene therapy approach that we have for Duchenne muscular dystrophy through our partner, Sarepta. It's approved in the U.S. and through Roche in 6 other countries. Bruno mentioned prasinezumab. I'm not going to say more about that. This is an antibody to aggregated alpha-synuclein and readout later this year is expected. And let's go to the next slide, which gives you our portfolio in the clinic. And here, I want to make three points. So we are in a number of different disorders that are shown -- that include the ones that I mentioned. The other point I'd like to make here is that we are using distinct modalities for neurological diseases in these projects, as shown in the left lower corner, the legend there, and that includes innovative modalities. And the third point and the final point - which is not shown on the slide - is that we actually have a large portfolio of research projects, so preclinical projects that, of course, feed to the clinic. I'm going to focus -- we're going to focus on Alzheimer's disease, of course, Luka will talk about trontinemab. Margherita is going to talk about -- on the diagnostic side, and I'm going to give you -- talk about gamma secretase modulator. Before I go into Alzheimer's disease, though, I do have one slide on GYM-329 in SMA. So as you know, in spinal muscular atrophy, we have Evrysdi, which targets the proximate cause of the disease, increases the levels of the protein SMN in this disorder. And what we're currently doing is to determine if we can get further improvement for patients with SMA by combining it with an agent that should lead to an increase in muscle size. And this -- we're doing this with a monoclonal antibody GYM-329, which deploys innovative sweeping and recycling technology that makes it more potent for factors that are in circulation, such as latent myostatin. We already have evidence preclinically that suggests this approach should work. And as you may have heard, Scholar Rock has recently come out with results of their SAPPHIRE study, suggesting that this approach of combining an agent that targets myostatin pathway should lead to benefits when combined with an agent that increases SMN protein. So we currently have the MANATEE trial, which is a Phase II/Phase III trial. The first part is to select an optimal dose, and this will be followed by a Phase III trial where we're going to test in addition to the safety, the efficacy. So let's move to the next slide, which now -- I'm going to focus now on disorders of aging and, in particular, Alzheimer's disease. And here, just to make the point that, of course, as most of you know, this is clearly one of the most important public health challenges of our time. With the aging population, the prevalence of AD is predicted to go to -- as shown on the right side of this slide to 150 million by the year 2050. The economic costs are obviously going -- predicted to be quite high, and enormous impact as shown in the middle slide. And what's not shown on the slide, of course, the burden on people and their families is going to be extremely substantial. And of course, the field has been focused on treating dementia and early Alzheimer's disease. I should also note that with the aging population, as people approach their 60s and beyond, some 40% of people have evidence of pathology of Alzheimer's disease that's not symptomatic, of course, but they're at risk for developing symptoms. So if we develop safe and effective drugs that can prevent that progression into symptom, of course, that would have even more impact on society. Now if you go to the next slide, I want to start out by saying that among -- of course, in AD, it's a complex disorder. There are many targets that we and the field are focused on. One that has received a lot of attention for decades is amyloid A-beta that is a peptide that leads to the formation of amyloid plaques and it's a key pathological feature of the disease. In the last few years, it's become clear that A-beta is the evidence suggests that it's a clinically validated target. Yet here as well, even for A-beta, there is room for improvement, both on the efficacy side -- so we need better clearance that's faster and deeper. And also in terms of safety and as shown on the third bullet point here, a common adverse event with anti-amyloid is ARIA, which stands for amyloid-related imaging abnormalities. So this comes in typically in the mid-teens to the 30% or so with the anti-amyloids. And so there's a need for lower rates of ARIA. And of course, beyond treating -- getting amyloid plaque clearance, as I mentioned, another sort of improvement that we're looking for is ways to prevent the accumulation of amyloid. And finally, on the last bullet point, convenience is going to be very important for the future. So agents that are going to be associated with less frequent monitoring, as well as more convenient administration, including potentially oral small molecule approaches. The other big picture that I want to tell you about is how we're thinking about this from the perspective of the patient journey. And here, we have -- the advantage at Roche that we have, of course, both the pharma and the Diet divisions and -- that collaborate. And so the way we're thinking about chronic diseases in general for neurology, and for Alzheimer's in particular, is how do we think about the patient journey and how do we develop diagnostic tools and therapeutics that take this into account where a person or a patient would be subjected to a test. And then, depending on the test, will be treated with the appropriate therapeutic and then monitored. So in diagnostics, you will hear later, but there's already an approved biomarker in CSF or cerebrospinal fluid for the diagnosis of Alzheimer's disease. And even more excitingly in development in DIA, there are -- there is work on the plasma biomarkers, and I'm sure you'll hear more about this. There is the p-Tau 181 and ApoE4 that is relevant for rule out of the diagnosis of Alzheimer's because it has a high negative predictive value and p-Tau 217 with a high positive predictive value allows a rule-in diagnosis. So these tests will be used across the journey, but especially early on, they will be relevant for a person coming in at risk for developing symptoms of Alzheimer's disease. And so then -- and then the idea would be then if they're positive, then they are paired with a therapeutic that we're still developing that would then prevent the progression. And so -- and similarly, when they are diagnosed later with Alzheimer's disease and they have dementia, then that would be paired with another therapeutic. And finally, for monitoring, we're interested in digital readouts and additional fluid biomarkers. Now you're going to hear about trontinemab from Luka. I want to actually tell you about one molecule, and that is the GSM modulator, which is really an innovative approach that targets not the amyloid clearance, but as I mentioned earlier, the prevention of amyloid accumulation. And to help explain how this works mechanistically, I actually have a video for you on GSM, which we can show, and then I'll explain how further. [Presentation] All right. So if you go to the next slide, before I actually go further into this, I want to, of course, tell you that this is a video and a depiction of what we're looking at. And it's early days, and it may well turn out if this molecule turns out to be -- to proceed further in development that it could be used in potentially in a number of stages, not just in early prevention, but also later for maintenance or combinations. But let me take you through what we currently have. So basically, on the left side is sort of a mechanistic understanding of GSM, and I won't spend a lot of time on that because you saw that in the video. So in the left lower corner, you see the APP protein and GSM is the blue circle -- blue field circle that is the molecule that's binding to GSM. So previous attempts -- people have known this for a long time that gamma secretase is important for the production of A-beta. But previous attempts were sort of where it was gamma secretase inhibitors were like a sludge hammer. They inhibited the production of A-beta and led to -- they were not beneficial. With GSM, what's really cool is because gamma secretase is a perceptive -- it acts -- it cuts APP at multiple sites, and that's why you get different species of A-beta. So what GSM does is it binds to gamma secretase and modulates it so that you can get less -- as shown on the right side of this slide, less of the aggregating amyloid A-beta peptides as was shown in the video, the yellow peptides, A-beta 42 and 40 and more of the non-aggregating peptides, A-beta 37 and 38. And if you go to the next slide, this was shown already in preclinic. We already also have shown in human and healthy volunteers, first of all, that this is safe and tolerable -- tolerated. And then in the CSF, in cerebrospinal fluid, we can see on the left side, in a dose-dependent manner, there is a reduction of A-beta 42 and A-beta 40 and a concomitant increase of A-beta 37 and 38 in these people. And we currently now have moved to a Phase II trial that is shown in the next slide. And here, in this Phase IIa trial, it's focused on -- again, this is in people who are accumulating amyloid. They are either cognitively normal or have mild cognitive impairment. And the focus here is on safety and tolerability, as well as on characterizing the pharmacological properties of this molecule in these people, including PK and PD looking at both imaging and non-imaging fluid biomarkers. So we're very excited about this. This is an ongoing recruitment; it's recruiting currently. And finally, I want to show -- this is my last slide, I want to give sort of an intro to Luka essentially, I want to here focus on trontinemab from the technology side. So Luka is going to give you an update on trontinemab. Here, what I want to tell you is that trontinemab is based on more than 15 years of innovation at Roche P-RED. And the idea here is to deploy the brainshuttle technology that's been developed here. So with a standard antibody, as shown in the slide sheet microscopy in mice, you see that it's -- a lot of it -- a lot of conventional antibodies that essentially 0.1% of the antibody gets into the brain. And when you give it intravenously, for example, systemically, a lot of it in the brain gets stuck in -- around the ventricles, which is what you see there shown on the left side. Whereas with the brainshuttle, because we think of the route of entry, it's a much broader distribution. So it's more than just that there's more of it in the brain. It's also the distribution is different. And trontinemab is our most advanced brainshuttle molecule. And of course, this project has implications to work beyond just A-beta and beyond Alzheimer's disease, but actually for a whole set of neurological disorders where the brain -- blood-brain barrier has been sort of the holy grail of neuroscience therapeutics. And with that, I'm going to turn it over to Luca.

Thanks very much, Azad, for this very nice intro. Hi, everybody. It's really my pleasure to walk you through our presentation here at CTAD. And yes, greetings from Madrid. So it's really a very exciting conference that we are attending here. So if we -- so what we presented at CTAD was essentially the latest update from our ongoing Phase Ib/IIa study of trontinemab in people with Alzheimer's disease. So updates, including safety and biomarkers as well as our recent efforts to expand the program and study trontinemab in a larger number of study participants. So this was the scope of the presentation. So these are our disclosures and Yes. So we also have a very nice brainshuttle video, which I'm not going to present here today. So this is just to give you an impression of the molecule. So trontinemab is indeed, as Azad nicely pointed out, a novel molecular entity. So it's a so-called brainshuttle A-beta antibody. It combines by recombinant fusion an anti-A-beta binder and an anti-beta binding moiety that targets aggregated forms of A-beta and amyloid plaques, with a brainshuttle module that specifically binds to human transferrin receptor 1. And this whole brainshuttle A-beta molecule then crosses the blood-brain barrier via an active transport mechanism that is called transferrin receptor-mediated transcytosis. And that then beats essentially to more of the therapeutic agent in the target tissue in the central nervous system to a better target engagement, and also in case of amyloid plaques, a more rapid engagement of plaques and also rapid clearance of plaques. This is a truly differentiated approach compared to conventional antibodies that really has a high potential to improve therapeutic outcomes for people living with Alzheimer's disease. If we go to the next slide, this is an overview of our ongoing Brainshuttle AD study of trontinemab. So this is, as I said, a Phase Ib/IIa study in people with MCI due to AD and people with mild to moderate Alzheimer's disease. The study has four different parts. In the initial dose escalation part or Part 1 of the study, there are four initial sequential dose cohorts. In each of these cohorts, approximately 15 study participants are randomized in a 4:1 ratio to receive either trontinemab or the placebo control once every 4 weeks, I.V. for a total of 7 doses. And then following review of the emerging data from this dose escalation part, we've got the option to investigate most promising dose levels further in the so-called dose expansion part, or Part 2 of the study. The goal here is to assess pharmacodynamics and safety in a larger number of study participants. So an additional 60 study participants per expanded dose cohort. And this is currently ongoing with Cohort 3, so the 1.8 milligram per kilogram dose level, and with Cohort 4, the 3.6 milligram per kilogram cohort. Without going too much into detail, there is also a Part 3 of this study, a so-called PK/PD part. So here, we are looking at the effects of different dosing regimens of trontinemab on PK/PD. This is an open-label part without a placebo control. And then lastly, there is also a Part 4 of the study, an open-label extension part that is offered to all study participants enrolled in Part 1, 2 and 3 of the study. And the goal and the objective here is really to collect long-term safety, PD/PK, immunogenicity data in all participants who had the chance to be enrolled in the first 3 parts of the study. If we go to the next slide. So the focus of the CTAD presentation was on the first two parts. So we provided an update on the most recent PD data, so biomarker data as well as safety data from the completed Part 1 of the study. And then for the first time, also shared the most recent interim safety results from the ongoing Part 2 of the study. We go to the next slide. This most recent interim analysis included in total 160 study participants who were enrolled in Part 1 and Part 2 of the study as of snapshot date on September 2, 2024. So we had in total, 60 study participants in the completed Part 1 and 100 study participants who were randomized to -- who were randomized in Cohort 3 of Part 2 and Cohort 4 of Part 2. You should note here that the two cohorts in Part 2, these two cohorts are still ongoing. So the mean number of doses received in Cohort 3 in Part 2 was 5.4 as a snapshot date in September, and 2.9 in Cohort 4 in Part 2. So these are still ongoing cohorts, and this is really interim data from these two ongoing cohorts. If we go to the next slide, yes. Or maybe one more back, apologies. So this is the baseline characteristics. And I should say here that, overall, Part 1 and Part 2 baseline demographics are very similar. So maybe a difference that should be mentioned here is that the cognitive status of the Part 2 participants tended to be slightly less impaired compared to Part 1. So if you see the mini mental scores in Part 1 were around 21 points-20 points. And in Part 2, we have slightly less impaired people, so 23, 23.5 points on MMSE. However, all the other baseline characteristics were very similar. So around 70, 71, 72-year old people, majority of them are female, the majority are ApoE4 carriers. So that's all very similar with the exception, as I said, of the cognitive status. We go to the next slide. So let's have a look at the blinded safety profile. And here, you should note that this is an ongoing study, which remains blinded to individual treatment assignments. So data from participants who received trontinemab or placebo. The data are presented together as cohort level data to protect the blind at an individual participant level. So compared to our previous interim results that we shared at AD/PD earlier this year, there were no new safety aspects in the completed Part 1 of the study. In the ongoing Part 2, we had two serious adverse events. One serious adverse event was a mild grade 1 infusion-related reaction in a participant who was hospitalized for precautionary reasons to monitor increased temperature in this participant. So that was the first SAE or serious adverse event. The other one was a fatal case of cerebral macrohemorrhage that unfortunately occurred in a 78-year-old female participant who was enrolled in Cohort 3 in Part 2 of the study. And in the following, I would like to briefly provide you an overview of this case and share with you some more insights and details that we gained from this case. So this participant experienced a right front lobar macrohemorrhage on study day 44 and then sadly passed away 2 days later. So approximately 3 weeks after administration of the second dose of the study drug. Subsequently, the participant was unblinded, and we performed a thorough examination of the case. And here, we included also external experts and advisers and also investigators. This thorough examination, this thorough assessment, revealed several important risk factors that likely contributed to this unfortunate outcome. The most relevant factor that was identified was a large area of occipital superficial siderosis. This is evidence of a prior brain bleed. And that was evident already at screening in this participant. And this finding was consistent with the diagnosis of probable cerebral amyloid angiopathy, so vascular amyloid deposition or what we call also CAA. In this participant, we found also additional factors, potential contributing factors, including additional findings on neuroimaging that I'm not going to go too much into detail. We can discuss this later as well as an ApoE2 genotype. An ApoE2 genotype, which is relatively rare in the general population. However, it is a known risk factor for cerebral amyloid angiopathy and also a vascular damage that is associated with vascular amyloid deposition in those who have this genotype. Overall, our in-depth assessment of the case triggered a protocol amendment and then led to the decision on our side to exclude participants with superficial siderosis. So the most relevant risk factor that was identified in this participant from the ongoing study. And these measures are in line with the recently published appropriate use recommendations for other approved anti-amyloid therapies, including lecanemab. And as we just learned yesterday here at CTA, this is also now the appropriate use recommendation for donanemab. So no superficial siderosis is allowed if you want to treat people with anti-amyloid therapies. So this is the recommendation. So we are fully in line with these appropriate use recommendations. If we go to the next slide. So moving on with the safety profile to the overview of the most relevant treatment-emergent adverse events. And the encouraging news is here that we are seeing positive trends regarding the two more common adverse events in this study, namely infusion-related reactions or what we call IRR and anemia. So at our previous -- or in our previous presentations earlier this year and last year, we already shared with the community that infusion-related reactions were relatively common with trontinemab treatment, and they appear to occur in a dose-dependent manner. Now with the partial implementation of premedication here in Cohort 4 in Part 1, if you look at the table on the left-hand side, we already see a significant drop in infusion-related reactions. And with further implementation and then optimization of the premedication regimen in the dose expansion part in Part 2, these numbers continue to decrease. Regarding anemia, we can say that we have seen only transient, mostly mild and limited anemia so far. For example, I mean, if you look at the Cohort 4 in Part 1, a doubling of the dose from Cohort 3 to Cohort 4 did not result in more anemia cases. On the contrary, we even had less cases or fewer cases of anemia in Cohort 4. And in Part 2, what we have seen so far has been very limited or fairly limited anemia, which we also explained with the fact that we have significantly reduced the total blood volume drawn in Part 2. I should say here, and this is what we also shared last year at CTAD, we believe that the frequent blood draw, and we have some data now also to confirm this or to support this belief is that the frequent blood draw is an important confounder when it comes to anemia in this study. So if we move on, last but not least, let's have a look at the amyloid-related imaging abnormalities that Azad also mentioned or ARIA. And we are really pleased to report here that the ARIA incidence overall remains very low with trontinemab treatment. So our initial encouraging results that we shared also earlier this year from Part 1, where we had essentially only one ARIA-E case in Cohort 3 in Part 1 and no ARIA-E cases in the meanwhile, completed Cohort 4. These encouraging results appear to be confirmed now that we are dealing with a much larger sample size in Part 2 with almost 100 randomized participants. In Part 2, we had only two additional cases of ARIA-E. Both cases were mild or mild plus in severity. So this is the kind of the radiographic severity on MRI. Both were asymptomatic, not associated with any clinical symptoms, and both occurred in Cohort 3 in Part 2. So in our opinion, these are really remarkable findings, especially if we put them in relation to the very robust and rapid amyloid lowering that we have seen with trontinemab, especially at these higher dose levels of 1.8 milligram per kilogram and 3.6 milligram per kilogram. If we go to the next slide, this is the data that we shared at AD/PD earlier this year. You have seen this already. So there is a clear dose-dependent effect on amyloid PET and a very rapid and robust amyloid plaque depletion, especially at this highest dose of 3.6 milligram per kilogram, where we saw a change from baseline to week 12 of roughly minus 90 centiloids. If we go to the next slide, in the now complete Part 1 data set, we were able to confirm these impressive results at 12 weeks. And at 28 weeks, we see that there is a further reduction of the amyloid PET signal, now minus 107 centiloids versus baseline in this fourth dose cohort. If we go to the next slide, just briefly, most of the study participants in these two higher dose cohorts at 1.8 milligram per kilogram and 3.6 milligram per kilogram have formally become amyloid PET negative at week 28. And please note here also the very high amyloid PET burden in the 3.6 milligram per kilogram group, where we started essentially with almost 120 centiloids. Despite these high amyloid levels in the fourth cohort, there was a very strong reduction of the amyloid PET signal, and we were still able to achieve amyloid PET negativity in the vast majority of the participants, even in that Cohort 4 that started very -- started out very high. So overall, these are really impressive dramatic pharmacodynamic effects, robust effects. They occur rapidly, literally within weeks. And the big question that we are now investigating further is what about other aspects of the AD pathophysiology or other biomarkers of AD. I should say here that we have just -- if we go back again, I should say here that we have just recently started to look at these markers. And so markers beyond amyloid PET. And these include markers both in blood, so in plasma and in cerebrospinal fluid. And now we can go to the next slide. So we have just recently started to look at these markers and the very first markers we have looked at are so-called core downstream AD biomarkers in cerebrospinal fluid or CSF, so including total tau, phospho-tau 181 and neurogranin, a synaptic marker. And what we see here is really encouraging. So we see dose-dependent large magnitude effects of trontinemab treatment on these three key markers, key downstream markers, of AD pathophysiology. The percent changes that we are seeing, of course, this is very preliminary data in a very limited number of participants, but the effects that we are seeing are already so large that we can see them already in a relatively small sample. The effect sizes that we are seeing there, the percent changes that we are seeing with this marker, especially at this high dose of 3.6 milligram per kilogram are quite pronounced, especially for this early time point of 25 weeks when we did the lumber puncture in Part 1 of the study. In our opinion, these results highlight the potential of the molecule of trontinemab to not only rapidly deplete amyloid plaques, but also have an early and also a very pronounced effect on key markers of AD pathology that are really closely linked to the AD pathophysiological process. With this, I would like to summarize. And yes, what we have shown at CTAD was an update on safety. And here, we can say that trontinemab has an overall favorable safety profile with a very limited ARIA incidence that we have observed so far. We discussed in detail the case -- a case of lobar macrohemorrhage that happened in the expansion part of the study. This event occurred in a participant with significant risk factors and probable cerebral amyloid angiopathy. This event triggered a protocol amendment and led to the decision to exclude participants with superficial siderosis from the study. And as I said, in line with the appropriate use recommendations for other molecules that have recently been published. The results from the complete -- the amyloid PET results from the completed Part 1 of the study demonstrated rapid and robust amyloid plaque depletion at 1.8 milligram per kilogram and 3.6 milligram per kilogram after 12 to 28 weeks of treatment. And these rapid effects on amyloid were accompanied by large magnitude changes in relevant downstream biomarkers in cerebrospinal fluid, including total tau, phospho-tau 181 and neurogranin in CSF. Overall, these most recent interim results in now more than 160 randomized participants, we believe that these results clearly support the continued development and also the expansion of the trontinemab program. And maybe I can -- Richard, you wanted me to speak also to the extra slide or we can keep this for the -- yes, I can show this. So this is just to -- again, the amyloid PET. And I should say here that -- so this is not based on any head-to-head data or so. So we have simply plotted published data from other molecules to just put these dramatic and robust effects on amyloid plaques that we are seeing with trontinemab into perspective. And here, you see in yellow, what has been published for lecanemab in the CLARITY AD Phase III study and also for donanemab. So these are the results from the TRAILBLAZER-ALZ study. So on the Y-axis, you see the amyloid PET burden measured in centiloid units that I touched upon. And on the x-axis, you see time. And I think it becomes very soon very clear that the speed and the magnitude of amyloid lowering at these two higher dose levels, but particularly at the fourth dose level is very, very fast, and we have a very steep trajectory of amyloid lowering with trontinemab, which gives us really confidence because what we have learned over the past few years in the field, and these are learnings from -- this is converging evidence from multiple large trials, right, is that rapid and robust amyloid plaque depletion is necessary for clinical efficacy. And the more people get below a certain threshold earlier during the treatment, for example, a double-blind treatment period of a trial, the more or the larger likely -- the more likely it is that the effects on clinical efficacy will be larger. So this is the relationship, and this is why we are particularly excited about this molecule. In addition, as I said, we have shared also the preliminary data from other non-amyloid markers, the downstream biomarkers, which seem to also behave in a similar way. So we see also early and pronounced effects on these so-called downstream processes, which further reinforces our confidence in the approach. With this, yes, I would then pass it to you, Margherita.

Thank you so much, Luka, for sharing all of this. And greetings from Madrid from my side as well. I'm Margherita Carboni. I'm the indication lead for neurology at Roche Diagnostics. And on the next slide, actually, I would like really to show to you what are we doing at Roche Diagnostics to really make sure that we can allow the access to those really interesting disease-modifying therapies that our colleagues at Roche Pharma as well as across the world are actually developing. So I'm really going back to the patient pathway that actually Bruno mentioned already. And where I would like really to focus the attention is on the mid part of this slide. We, of course, have already on market p-tau181 over A-beta 42 as well as total tau over A-beta 42 as a ratio in CSF for the determination of amyloid pathology. As you have heard now multiple times from Luka, really this detection of amyloid pathology at the early stage is of crucial importance, both from a diagnostic perspective to really select the patient to give them reassurance about what is going on in their brain, but also for the best decision on treatment. This is currently done in standard of care in different geography via CSF, as I mentioned. But the field has been moving really fast. It has been really exciting, and we have been moving into blood-based biomarker, as, again, a lot of you already know. And at Roche, we are -- we got actually breakthrough device designation on two products. One is the EAPP, the Elecsys amyloid plasma panel, which is formed by two main biomarkers, phospho-related tau pTau-181 and APOE. And then we got a second breakthrough device designation around Elecsys-P-Tau-217. Those two tests for us are really at the core of what we call the triage part of the patient journey, really bringing to the patient and to the clinician a possibility of giving an answer to those patients that are showing up with cognitive impairment. And on the next slide, I really want to go into the details of the Elecsys amyloid plasma panel because we have been conducted and we are really proud of our interim analysis. We have been running a really big -- one of the biggest study of its kind for the validation of the Elecsys amyloid plasma panel and for p-Tau-181 as well. This big prospective multicenter study was actually run across the globe. It was run in the U.S. It was run in Europe. It was run in Australia. And what we were really looking for was actually understanding what we call the clinical performance of our test. And as you can really see both from the AUC as well as on the negative predictive value, we really read it above 90% clinical performance. We even hit 96.2% for EATP and 97.6% for p-Tau 181, really showing how those blood tests have the possibility of ruling out of being sure about those patients that are actually negative and really give confidence in for clinicians. The other part I really want to draw your attention on was really this big clinical trial. As I mentioned, it was really run across the board and with really, really specific attention on mimicking the real-world setting. We really wanted to make sure that our results were really, really true for different comorbidity for different demographic to make sure we are bringing something to market which is reliable and that it is clinically validated. So yes, we were really proud of seeing such a strong rule-out performance well above what we expected and what the field expected as well. And then on the next slide, actually, those are also a method comparison which we have done in collaboration with our colleagues at Roche Pharma of our Elecsys pTau217 to Lumipulse pTau217. So what, again, we show here is a high clinical performance with a significantly smaller gray zone. Let me tell you about this gray zone. So what the field is really looking at in the diagnostic pathway is really to have a test that is able to say for the patients that are most likely to carry amyloid pathology, those patients that are most likely to not carry amyloid pathology to inform on the Alzheimer's disease diagnosis. But then we are also looking at really reducing the gray zone. So those patients for which the level of pTau217 are unfortunately not conclusive. What everybody in the field is speaking about is actually in order to have such a test in a meaningful way, you need to have a gray zone, which is below 30%, 3-0. And what we show here is that in this specific population, Elecsys-pTau217 actually show a significantly smaller gray zone with 10.2% of patients ending there. Meaning that, again, when we think about in future, the clinical use of this, we will have the majority of our population classified as either positive and negative, leaving only a reduced part of the population with an indeterminate result, which was, again, significantly smaller than the results found in Lumipulse. So on the next slide, we are proud of these results. We are really happy with those, and we also think we are really, really well positioned to actually drive access to those results and those tests in the future. Of course, we are collaborating on a daily basis with our colleagues in Roche Pharma, but we also have a really big Elecsys instrument coverage, right? So we really believe that bringing those tests to all those Elecsys instruments will really allow us for broad acceptance. And with that, on the next slide, I actually hand it over back to Bruno for Q&A.

Thanks, Margarita. And with that, we jump into the Q&A session. And let me also remind you that we have an additional panelist here with us, which is [indiscernible], the lifecycle leader for Trontinemab. The first question, they would go to Richard Parkes from BNP Paribas.

Yes, just a couple of questions. First one is more of a sort of thought experiment. It's just about the one thing that I kind of worry about when you're thinking about the risk of ARIA with a brainshuttle antibody. And I'm just thinking, is there -- I'm thinking if there's a risk that we might see a lower instance of incidental ARIA findings, but a higher frequency of more severe events given the deep brain penetration of the product? So I'm just wondering, is there anything that might link to the patient that died with effects that might implicate that? And what's your thinking about that risk? And is there a biological reason why I should worry about that less? And then the second question is on the Phase III design and the time lines that you outlined because obviously, the FDA now will accept beta amyloid reduction as a surrogate endpoint. Can you talk about how you might factor that into the trial design? And does the 2028 readout assume demonstration of clinical outcomes? Or is there potential for an interim to accelerate that?

Shall I take it? And then maybe I will pass it on to you, [ Hanno ], for the second question. I think that will be probably appropriate. So thank you very much, first of all, for the question. So the question was about the potential risk of kind of isolated severe events, right, with the treatment. So what we can say is based on our preclinical evidence and evidence so far that we do not have indication for that, right? Any indication that this would be something we would expect. This event is a single event, right, that occurred in a participant with significant risk factors. So we mentioned the superficial siderosis, which was large in size. However, there were also additional factors, including also the genetic constellation, prominent white matter lesions that are very characteristic also for cerebral amyloid angiopathy and other aspects. So based on this one case, yes, I would not be able to -- or wouldn't want to establish any risks or potential risks that are associated with. What is important to note is that in this participant, we had one post-baseline MRI scan that we did approximately 3 weeks prior to the incident, and that didn't show any ARIA, any changes from baseline. Like it was identical to the screening MRI. So there were also no changes that would kind of suggest that something precipitated that event. So that's important to note here.

Let me -- I'd like to add actually to -- Richard, to your question from the biology side because you asked, can I provide you reassurance? And I would say, yes, because on the biology side, what the deep penetration tells us, it's sort of the brainshuttle molecule is getting into the brain parenchyma and sort of getting a broad distribution that I showed, at least is what we think is happening based on what we know on the biology. And that has -- that is really very distinct, very different, and we haven't seen anything that would suggest at this point that there would be sort of an adverse effect because of that. And with the case we're talking about, that's a vascular event. It's -- as Luka mentioned, it's basically their cerebral amyloid angiopathy. So that's a vascular issue. Whereas with the brainshuttle, what we're talking about is a molecule that the through the endothelial cells on the capillaries, the very, very tiny vessels that are deep in the brain get on to the other side, as it were, of the endothelial cells into the brain parenchyma. So from a biological perspective, I can reassure you. And I hope that helps you.

Yes. Let me take the Phase III question. So I mean, obviously, we are looking to differentiate ourselves on efficacy and safety for the molecule. And therefore, it will be very important to collect these types of data in a Phase III study. But we are well aware of the regulation and the discussions from regulators, and we will keep this in mind and look for opportunity also to get into the market earlier.

Okay. The next question would go to Sachin Jain from Bank of America.

Sachin Jain from Bank of America. Just a couple more on Trontin(sic) [ Trontinemab ], if I may. So first, you talked about the fact the [indiscernible] amyloid reduction having benefits on cognition. I wonder if you've got any early modeling data that points to a target profile relative to existing MABs on traditional CDRs in the boxes. The slides you put up on speed and death amyloid reduction was helpful, but I wonder if you could translate that to any efficacy data from your modeling. And the second question was just to understand what additional data you want before deciding on a Phase III. So I think the commentary has been you want data in roughly 200 patients. What is it that you're specifically looking for in the remaining patients to give you comfort? And I just want to check, if the death occurred late in the study something that you need to get more comfort with? Or is it other factors?

Good. Then I'll start. So the modeling. So we shared at CTAD a year ago the results of a meta regression analysis, which is not -- I mean, it is kind of a modeling word. So it is meta regression work that looked at different available data sets from anti-amyloid trials across molecules also from other molecules, donanemab, lecanemab and so on. And what this -- the results of that meta regression analysis suggests, of course, under the assumption -- it's based on data, right? And this is kind of an extrapolation of what would happen if we would achieve amyloid plaque depletion at an early time point during a trial, what would happen then at trial endpoint a year later, right? So this was kind of the objective. And the projection is the more people we get below the threshold of amyloid positivity at, let's say, 6 months within an 18-month trial, the higher the effect on CDR sum of boxes will be. And this is based -- this is meta regression data based on already available data from different trials, including our gantenerumab trials, but also trials from our colleagues from other companies. So based on this assumption and early effect, for example, at 6 months, if we would achieve 70% or 80% plaque clearance at 6 months, there is based on this work, some indication that the effects on the clinical efficacy endpoint on CDR Sum of Boxes at 18 months may be significantly more pronounced than what we have seen so far. So beyond actually 40% or even more percent of slowing of decline or up to 50%. But this is, of course, essentially a meta regression and an extrapolation of currently available data. So that's -- in terms of what I'm aware at least of available data from modeling activities so far. And the second question was related to the data that would be required from the current trial. Is this -- was this the question? Yes. So I mean, clearly, the primary objective of the current trial is safety, right? And what we are doing, especially in the expansion part, I mean, we have designed the expansion part, the Part 2 in a way that we really collect robust data, especially on the safety side. And with regards to actually the ARIA -- so the sample size was determined with the goal to predict the ARIA rate -- to predict with a certain precision that the ARIA rate is indeed as low as we believe it will be, right? And so for that, we need approximately, well, 60 plus then an additional 15 study participants from Part 1. So this is the basis of the sample size that we are using here. So we will be looking at safety. So this is the most important, most relevant outcome. But of course, then the key secondary is the amyloid PET effect. And there, we will be looking at, obviously, what is achievable with this molecule. And as I said, I mean, we do have some indication that this early plaque clearance in a substantial number of people at an early time point will result in larger effects. So yes, Hanno, maybe you would like to comment?

Yes. So the only -- Yes, so what is really driving us is to have sufficient safety on the dose that we are selecting for the Phase III as well as confidence that we achieve the outcomes in terms of amyloid PET in the induction phase as well as biomarkers. As well as think then about what do we want to do in terms of a patient-friendly maintenance scheme as well in this trial. And for that, we need a bit more data than from 15 participants that we have shared with you today.

Next question would go to Colin White from UBS.

Yes, two questions from me, please. It's Colin White from UBS. The first question I had was on when you look at the data for the two highest dose cohorts and you see the rate of are patients reaching amyloid negativity, what does it tell you about what you could possibly achieve if the dose was higher? Do you think you could achieve meaningfully more efficacy or not? And then my second question is on the brain -- blood-brain barrier crossing technology. Obviously, we've seen recent news of a competitor buying a company related to this. And so how should we think about this in terms of if this stuff works, just that you are meaningfully ahead in terms of the race to develop blood-brain barrier crossing technology? Or is there some reason to think that your technology will be differentiated from others that are being developed?

Very good questions. So in terms of going higher with the dose and achieving even faster plaque clearance, right, whether this is going to translate into more efficacy. As I said, it is based on certain assumptions and also some work also that is based on clinical data, the meta regression analysis I mentioned that we shared at CTAD earlier this year. I think we don't know yet, right? We know, however, and this is actually also the exciting data that we shared at CTAD, right? I mean, at this higher dose of 3.6 milligram per kilogram, we see that there is an additional significant benefit when it comes to these downstream biomarkers and a really early and pronounced effect in these markers. Which are essentially beyond amyloid PET, so total tau, which is a classical actually neurodegeneration marker, then phospho-tau, which is closely related to amyloid deposition. But then also a synaptic marker, neurogranin, which also are significantly reduced to an extent, which goes beyond what we have seen so far with other molecules and especially not at such an early time point. So we are -- based on this data, and it's a Phase Ib/IIa trial, right? So we're exploring still the dose response relationship. The data tells us that, yes, I mean, we are potentially not yet -- have not reached the dose response. And yes, it makes sense to go higher. How this will translate into efficacy? I think the more evidence we have from these different disease-related biomarkers, the bigger the confidence right? And as I said, there is evidence from the field that a deep clearance in the majority of people at an early time point may or will translate into clinical efficacy and into larger effects.

I can take the second one if you'd like. The question about the technology. Would you like to address it or -- I can start.

With the other technologies, I think it's -- well, we do not have any data to compare, right? So it's really hard -- or we shouldn't be doing this without data, obviously. I mean we are quite excited about our molecule, which, as I said also at the beginning, it has a -- or I didn't mention it, but I mentioned it in previous presentations. It has a fully preserved FC domain. So it has an effect of function that is fully preserved in this molecule. And this is important for target engagement and also efficient clearance of amyloid plaques. We think that this is really an important aspect of trontinemab and really drives also contributes to this strong effect on amyloid. So we think that, yes, this is clearly a differentiation. But Azad, maybe you would add?

I'd like to add a little bit to that on the technology side. So without sort of going into the details of the other approaches, it's -- of course, first, it's absolutely something that is of great interest. So there are a lot of people interested in this. But a lot of protein engineering went into this for the brainshuttle that we have. That essentially leads it so that it becomes transcytos efficiently and allows for a molecule to work on the other side of it. And sort of where we are, of course, trontinemab is the most advanced brainshuttle molecule in Alzheimer's disease. And so I think that, of course, there's a lot of interest, but -- and which is good, but that's where we're at.

Okay. Then we go on with the Q&A. And the next one in the row would be David Evans from Kepler.

Hopefully, you can hear me. So yes, really unprecedented amyloid clearing. But I guess as other questions have alluded to for people to really believe in the potential, how confident can we be that, that correlates with slowing progression. So I'd like to ask on the biomarker data. As you mentioned, it does seem to much greater reduction in those markers than with donanemab or lecanemab. Am I reading it right that you're kind of in the region of roughly, roughly kind of double the lowering that you've seen in those other molecules at the highest dose? And secondly, probably more relevantly, how strong is the evidence for correlation of these biomarkers with future cognitive decline? Especially for neurogranin, because I have to say I'm not totally familiar with that. But could you just outline the kind of level of confidence that, that is actually predictive biomarker?

Yes, it's a very good question. So we do not have -- I mean, obviously, this is a preliminary data set, right, in a handful of patients from Part 1, but the data is already -- I mean, the results based on the magnitude of the effect already, we see clear differences. And however, this needs to be confirmed, obviously, in the much larger sample size in Part 2, and this is ongoing and it will be actually also one of the important readouts in the first half of next year. So the magnitude of the effect, I mean, yes, I mean, these reductions that we are seeing are probably close to maybe twofold. But of course, I don't want to like -- because it would not be appropriate. We have not -- it's not a head-to-head comparison. But what I think is exciting is it is evident at 25 weeks right? So there is a normalized -- well, normalization. We are moving towards normal at an early time point. And that suggests that multiple processes of AD pathophysiology are affected in a way that things are beginning to normalize earlier, right? So for me, this is encouraging and could suggest then the potential for a benefit -- clinical benefit at a later time point. In terms of the predictive value, we do not have really data from interventional trials that would suggest you need to lower neurogranin or total tau by, I don't know, so much to have clinical efficacy. I think it's too early to give you there any numbers, but we know that all these markers in observational trials are important markers that predict clinical progression. So that's really an important aspect. So we are -- essentially, we have an early and pronounced effect on markers that in an observational cohort are highly predictive of clinical decline and progression.

If I may add just very quickly, of course, ultimately, we're going to need to have a Phase III and that kind of trial to look at the effects on the clinical efficacy. That said, what is very clear from the field when you look with meta-analysis is that the deeper that you reduce amyloid and the faster that you reduce amyloid, the more likely that there will be a beneficial effect. So all indications are sort of going in that direction. We don't have something that tells us now, okay, well, it's going to work, and that's why you need to do the study. But the biomarker data actually are also quite impressive in my view. So I think things are looking encouraging. But ultimately, for the clinical efficacy, we need to do the studies.

Very good. Next question, go to Luisa Hector from Berenberg.

As you think about Phase III design, would you consider treatment to amyloid negativity and then stopping as an option? And could you comment on what you think the reason is for the low ARIA rates given the really impressive amyloid reduction?

Thank you very much. Yes. I think it's too early to comment on the exact design or dosing regimen that we would take forward to potential Phase III. Obviously, there will be learnings from the current study. I mean, what is exciting about this molecule is that really a vast majority of people becomes amyloid negative at an early time point, which, of course, might offer even the potential to not have a treat-to-deplete regimen, but really like a fixed dosing where you could say, okay, after that time in principle, you can be sure that a significant number of people will already be negative. But yes, it's simply -- it's too early to really comment on these aspects. I don't know, [ Hanno ], whether you would like to also chime in here?

Yes. So I mean, as we said earlier, like why do we need to collect more data? I think this is exactly the type of data that will inform the right dosing regimen for Phase III. And we also, of course, look what is evolving in the field. We are very interesting in the biomarker data developing as well. But I think we can say today that we have actually a very good profile in place that gives us options. So a rapid induction, but it also gives us opportunity for like a low-frequency patient-friendly maintenance regimens, and we need to decide how we are going to structure the Phase III exactly. And maybe chime in on the ARIA.

Yes. Regarding ARIA. So I mean -- so the hypothesis that we are currently also looking at actually also in preclinical experiments is that the more direct access of this molecule into the brain via capillaries likely makes the difference. So standard antibodies access the brain in a different way. So the concentrations, you need much higher systemic doses. The doses -- there is a hypothesis that the standard antibodies also enter the brain via the blood cerebrospinal fluid barrier. So there are higher concentrations also in the cerebrospinal fluid. And then from there, there is a diffusion around perivascular spaces where the antibodies then engage also with vascular amyloid, which then leads to ARIA. So I think this is one of the probably most favored hypothesis, especially with Cynthia Lamere's recent work in the field. Trontinemab enters the brain, as also Azad nicely showed directly via the capillary. So it kind of bypasses potentially -- This is one of the hypothesis, the vascular amyloid, which is largely in the bigger vessels in the arteries and not necessarily at the capillary level and has a much more direct way to access its target, the amyloid plaques in the brain. In addition, there is also -- there are much lower doses that we are administering systemically. And it is hypothesized that this also contributes or could be one factor that could explain these lower ARIA incidences. There is recent also research work from others in the field who have also similar transferin-based approaches also that have been tested in preclinical models that seem to support this hypothesis that the co-localization -- essentially that the access to the brain via the capillaries might result in a different engagement with vascular amyloid, or less of it, compared to a standard antibody, which could make really a difference when it comes to ARIA. But of course, this is something that we are also looking at in our own preclinical models. And yes, we hope to be able to share some updates early next year on that.

Thanks Luka, and next questions would come from Richard Vosser from JPMorgan.