Roche Holding AG (ROG) Earnings Call Transcript & Summary

This meeting is being recorded. [Operator Instructions] At this time, it's my pleasure to introduce Karl Mahler, Head of Investor Relations and Group Planning. Karl, the stage is yours.

Yes. Thanks a lot, Henrik. Welcome, everybody, to our fifth science call this year. We have 1 hour reserved for you, half an hour for presentation and half an hour for the Q&A. And of course, I mean if you find difficulties or have difficulties to enter into the Zoom system, which I don't assume now, but you can also, of course, drop an e-mail directly to [email protected]. So I'm joined today by Kathryn Wagner. She is the Vice President of Neuromuscular Disorders, Clinical Development with Roche. Many of you may know her from her previous jobs and engagement. She was a Director of the Center of Genetic Muscle Disorders of the Kennedy Krieger Institute. She was Professor at the Johns Hopkins University, and she is well-known for her initiatives in the whole CNS field, but in particular, on the Duchenne disease research. I'm sure you will also may have some questions later on. We have during the Q&A session also Paulo Fontoura with us. He is the head of the development for CNS. Simona Skerjanec, Head of CPS CNS -- of the GPS CNS section, so basically the marketing part. So Paulo is doing the development part. We have also Kavita Patel with us. She is the life cycle leader for Evrysdi. Actually, we -- if you look at the key late-stage news flow, we had a really good year so far. I mean we are in a very good way to file faricimab by first half. So that is all on a good way for AMD and DME. Two good readouts for the dual antibody to fight COV 2, the outpatient, the beta and the post-exposure prophylaxis. Tecentriq was one of the focus of the last ASCO, the past ASCO, adjuvant non smaller lung cancer, actually received a very good feedback from ASCO. And today, we are focusing on every stage, type 1, 2 and 3 in SMA, a few data, which we will focus it on. Just to do a bit of an advertisement for the rest of the year, we still have the Pharma Day up coming on the 14th of September. An ESG event, we haven't done one for long, I have to say. I think it will be the first one actually this year. We do a lot in ESG. We haven't done a lot of public announcements on it. And I think this is -- but we are really, let's say, doing a lot, placing ourselves in the Dow Jones sustainability index every year now at the #1 position. So we felt that it's a good time to share some of those efforts with you. And we have ASH upcoming and also our digital event, which we do basically every year. If you look at the broad pipeline, which we have. In CNS, you can see that, I dare to say, we have the broadest and deepest pipeline in the end, also a portfolio in the industry. OCREVUS is launched, fenebrutinib is in a Phase III. We have Enspryng launched. This is in neuroimmunology, neurodegeneration. Alzheimer, I'm sure you will have some questions on that part later on, in Phase III reading out next year. Parkinson, Phase II, III. The neuromuscular disorders, Evrysdi, for both -- one today and one of the prior research also from DMD -- Duchenne muscular atrophy and disease. So here, we have also received some positive data, I have to say, from our partner recently and with an earlier stage neuro-development psychiatry syndrome which we just moved into Phase I in schizophrenia. So there is a lot to do. So if the focal point for today is basically on Evrysdi team. Meaningful evidence being generated across the broad program, spanning types from 1, 2 and 3 SMA, naive and pretreated patients, newborns to 60 years old. We have included also real-world spectrum SMA data, which is good because here, we have really an intensive research done over the past. And the focus today is on RAINBOWFISH. This is basically an entrant from both -- from birth to 6 weeks old. Smaller patient numbers, I have to say in the trial, but really, really nice data, very encouraging for patients. And the JEWELFISH first trial in patients from 1 to 60 years old, a wide range of disease severity being presented at this year's conference. And with this one, I wanted to hand over to Kathryn. Kathryn, over to you, please.

Hello. It's a pleasure to talk to you today about interim data of Roche's JEWELFISH AND RAINBOWFISH in -- Evrysdi for SMA. And these findings add to a growing body of evidence for Evrysdi in a wide range of ages and types of SMA and help lead us in our understanding of how we can continue to meet the ongoing unmet medical needs of the diverse SMA population. Next slide. So JEWELFISH is the first study of its kind. Next slide. This study is in a broad age range from 1 years old to 60 of types 1 through 3 and a diverse heterogeneous population in terms of disease severity with many more severely affected patients that are usually enrolled in a clinical trial in SMA. These patients were all previously treated with a disease-modifying therapy for SMA. And our primary analysis will be at 24 months, but we are presenting now some 12 months data on the safety, PD and exploratory efficacy data for the JEWELFISH population. Next slide. So here, you can see the study design and the primary objective of this trial, as I mentioned, is safety, but we will also obtain data on PK, PD relationships, and some exploratory functional endpoints such as the MFM32. Next slide. This slide shows the baseline characteristics of the JEWELFISH population. And I will just call your attention to the final column, where you can see, again, the age is ranged from 1 to 60, the types included types 1, 2 and 3, SMN copy numbers from 1 to 4. And the motor function was diversed and more severe, as I mentioned, than many SMA trials. So for example, 63% of patients had a baseline Hammersmith motor scale expanded score of less than 10. And that it's frequently an exclusion criteria for clinical trials, and again, high degrees of scoliosis, 83% and severe scoliosis in almost 40%. And so this represents the real world. This represents what physicians will be seeing in their clinics. Next slide. So why would a patient or a caregiver enroll in JEWELFISH to receive risdiplam? The reasons varied somewhat depending on what the previous treatment was. So for those who had previously received nusinersen, they chose to switch to -- primarily to treatment-related tolerability concerns such as intrathecal administration, which was followed by a lack of efficacy or a loss of efficacy. And those 2 comments on efficacy total 30% of the population. Other reasons were caregiver preference and patient preference. In terms of those who have previously been treated with Zolgensma, the primary reason to enroll in JEWELFISH and receive risdiplam was the hopes of an additional benefit in 57%, caregiver preference and a lack of efficacy of the gene therapy or other reasons. Next slide. So there were very low rates of discontinuation in the JEWELFISH study, only 9 out of 174 enrolled patients discontinued. This is 5%. And there were 76 patients who enrolled who had previously received nusinersen. Four withdrew, and you can see their reasons for withdrawal below. And then 14 patients had previously received Zolgensma. And there's been no withdrawals from that cohort. Next slide. There were no treatment-related safety findings that led to withdrawal in any JEWELFISH patients. And on this slide, you can see a total number of AEs and SAEs. There was only 1 treatment-related SAE. This was a child who had super ventricular tachycardia that the investigator felt might be related to the drug. However, it resolved despite, risdiplam dosing continuing. Next slide. So the SAEs and the AEs in the JEWELFISH population were very similar to what we have seen in previous SUNFISH trials and are just reflective of the underlying disease. So remember, these patients have had SMA and the manifestations of SMA for many years and up to 60 years. And so they have the morbidities associated with that, including among the most high upper and lower respiratory illnesses. It is worth pointing out that in the patients who were treated with Zolgensma and then subsequently treated with risk risdiplam, their AE profile was very similar to that of the other cohorts who have received different treatments and, again, very similar to naive SMA patients who received risdiplam. Next slide. So risdiplam led to a rapid and sustained increase in the SMN protein levels in blood. And here, you can see that SMN protein levels approximately doubled in the first month of treatment and then remain stable for the year. Next slide. As I mentioned, the primary objective of this study was to look at safety in previously treated SMA patients who now received risdiplam, but we did also look at some exploratory efficacy measures. And an interim data analysis demonstrated an overall stabilization in motor function at month 12 in patients who began to receive risdiplam following previous treatments. So this shows the change in the MFM32, which is a measure of both fine and gross muscle function. And there is essentially no change over 1 year. And we can't underestimate the importance of stabilization in this population, many of whom are older, many of whom, as I've mentioned, are severely affected. In fact, the SMA Europe recently published a paper in neuromuscular disorders, where of 1,500 responses, 97% of them said stabilization is indeed progress. Next slide. So in conclusion, the JEWELFISH population is broad and heterogeneous with a high degree of motor impairment at baseline, reflecting the real-world SMA population. The AEs and SAEs were reflective of underlying disease. Risdiplam treatment has shown a sustained greater than twofold increase in median SMN protein levels versus baseline. There were low rates of discontinuation. And interim exploratory efficacy data show that overall stabilization in motor function was observed in patients who began treatment with risdiplam following previous treatments. Again, the JEWELFISH study is ongoing, and primary analysis will be conducted at month 24. Next slide. So let's switch gears to talk about RAINBOWFISH. This is a study of risdiplam in individuals with presymptomatic spinal muscular atrophy. And as I'm sure you are well aware that risdi has been approved in 44 countries for individuals 2 months and older. However, ideally, we would really like to be able to treat babies before they had substantial motor neuron loss and before they manifest weakness. And this is becoming more and more possible as newborn screening is adopted. So RAINBOWFISH is looking to see what is the appropriate dose and the efficacy and safety of treating presymptomatic individuals. So we can go on to the next slide. So this is a multicenter, open-label, single-arm study of risdiplam in infants with genetically diagnosed and presymptomatic SMA. And they could enroll up to 6 weeks or 42 days at the time of their first dose. And the primary endpoint will be when the population that is being studied for the primary endpoint, which is able to hold one's head for greater than 5 seconds at month 12 has been reached. And there are a number of secondary endpoints, which include functional and PK/PD and safety. So we're going to report here the first time of the first 5 patients to reach this 12-month endpoint. Next slide. So the baseline characteristics of the 12 infants currently enrolled show that they had an age of 16 to 40, 4 -- 5 of them had 2 SMN2 copy numbers and 7 had greater than 2 copy numbers. And of the 5 infants that have been treated for greater than 12 months, 2 of these infants had 2 SMN copy numbers and 3 infants had greater than 2 SMN copy numbers. Next slide. So we are just really pleased to see that infants treated with risdiplam for at least 12 months achieved motor milestones on par with their healthy peers. So 100% have head control, 100% of sitting, rolling, crawling. 4 out of 5 can stand unaided, 4 out of 5 walk independently. One is standing with support and balancing, and we have every reason to hope that he will also achieve these motor milestones. So 80% of the infants scored a maximum on the HINE-2 with the total score of 26. And this included 1 infant with 2 SMN copies. And 1 infant with 2 SMN copies had HINE-2 score of 23. And just to put this in perspective, an individual with 2 SMN copy numbers would most likely manifest as the type 1 SMA baby who would have a HINE-2 score of 0. Okay. Next slide. So another way of looking at this data is by looking at motor milestones over time. And the gray bars are the WHO windows of achievement in healthy individuals. And the symbols are the achievement of motor milestones as observed at the study visit. So not necessarily when the motor milestone was first demonstrated, but it's when they were first demonstrated in the clinic. And some of the clinic visits were delayed because of COVID. But notwithstanding that, you can see that the majority of individuals reached some very complex motor function milestones such as crawling on the hands and knees, standing, walking independently within the range of normal healthy infants. Next slide. And the infants treated with risdiplam for at least 12 months, reached a near maximum CHOP-INTEND score by 4 to 5 months of age. 4 out of 5 infants scored the maximum score of 64 and 1 infant scored a score of 63. And again, to put this in context, babies with 2, 3, 4 SMN copy numbers rarely go above 40 on the CHOP-INTEND score at any time and then dramatically decline in function. So this is really showing us that risdiplam plan has rescued the SMA phenotype from these individuals. Next slide. So there were no treatment-related SAEs reported increase in presymptomatic individuals treated with risdiplam. Again, here, you can see the total number of AEs and SAEs. Next slide. And interestingly, the AEs now were reflective of symptoms and signs that normal individuals under the age of 1 have, so not reflective of underlying SMA. So nasal congestion cough, teething, vomiting, et cetera. Next slide. So in summary, RAINBOWFISH, most of the infants treated for greater than 12 months achieved motor milestones within the WHO windows for healthy children. And as of the data cut, all 5 who had received risdiplam for 12 months reached the maximum score of 64 on the CHOP-INTEND. There were no treatment-related SAEs reported in presymptomatic infants treated with risdiplam now for up to 18.1 months. And so I think the risdiplam story is really the future of SMA, where we treat individuals prior to their manifesting disease and allow them to grow up as normal children. Last slide, I think. So just to summarize, JEWELFISH was the first trial in a diverse SMA population ages from 1 to 60, who received prior treatment which showed a consistent safety profile and a greater than twofold increase in SMN protein levels and stabilization in motor function. In RAINBOWFISH, presymptomatic babies with SMA treated with Evrysdi for at least 1 year were able to sit, stand and walk in preliminary data, achieving motor milestones within the WHO windows for healthy children. And so we are looking forward to completing these trials, JEWELFISH and RAINBOWFISH, with speed and with quality and fulfilling our commitment to the SMA community to provide this drug to a diverse population globally.

Many thanks. And maybe Henrik, we can -- very good. So we have now all of us here on stage in brackets. And I have already seen a lot of interest in Q&A. Let's start with Emmanuel Papadakis.

Emmanuel Papadakis, Deutsche Bank. Perhaps just a question on JEWELFISH. You hopefully gave that one -- presented that one chart with functional benefit, MFM32. You didn't provide the split by patients previously with Zolgensma versus Spinraza. So could you just give us any kind of qualitative direction? Did it differ or was it very similar regardless of what therapy patients have previously received? And then just your thoughts on the implications of that for justifying, I don't know what you call it, combination use or subsequent use of Evrysdi patients that have previously received Zolgensma, either lack of apparent improvement in functional status. Does that influence your thinking about whether that will be justified? And then second question. We saw some very interesting data at conference from apitegromab combination with Spinraza. I would love to hear your thoughts on the potential implications of that. And indeed, whether you will now consider collaborating, supporting a trial combination of risdiplam with apitegromab.

So I think your question had 3 parts. The first question was whether we have seen differences in improvement or stabilization of function depending on which pretreatment or previous treatment the individuals have had. And as you saw, we have some fairly small subgroups, such as those who had Zolgensma were 14 patients. And we really haven't done that type of subgroup analysis on the preliminary efficacy measure. Your second question, I believe, was to comment on the combined use of Zolgensma and risdiplam. And I would first say that the wide range of Evrysdi studies across ages, types and disease severities has shown efficacy as a monotherapy of risdiplam. That said, there are currently families and physicians who are choosing to put children who have received gene therapy also on risdiplam. So that is part of the real-world that is out there. We have hoped that JEWELFISH -- we do hope that JEWELFISH will continue to provide us data on the efficacy and safety of doing so. We believe at this point that, that is safe. And as I've mentioned, we have just preliminary data on efficacy. I think the third question of Spinraza and risdiplam, I'm going to see whether my colleague, Kavita, would like to answer.

So with Spinraza's and risdiplam, we've seen that currently in the market that 2/3 of patients that have been previously treated are being switched from Spinraza for risdiplam today. So that does exist, as Kathryn said that we do see previously treated patients being moved to risdiplam plan.

And the potential agent, which is here to do any kind of trial. So I have to admit, I have to skip this one, maybe, Paulo, you have any idea or I'm not sure if there's anything planned here in terms of combination studies.

What agent?

Apitegromab, if I heard it correctly. Emmanuel, if you could...

Yes, sorry. I mean, you can skip it if you don't have an answer. It was apitegromab.

Obviously, we don't have an answer to this one. So that means also most likely that we don't have anything planned at this point in time. I think the answer is given by a nonanswer, I guess. Okay. Did we address your question?

Yes. You can just follow-up with [indiscernible] Yes, sorry. Sorry, I just -- I wasn't quite aware they had an INN yet. But yes, so this is an interesting mechanism. We do have an internal program targeting myostatin, actually a really good molecule which is in Phase I trials right now. I mean there's no firm plans yet. But certainly, it's a mechanism that we think might be interesting as a potential combination partner for any of these therapies because it works by a different mode of action. So we're looking at those there. It's still pretty early. And I don't know, Kathryn, if you want to add anything on that.

So, I would just reiterate what you said that it is an interesting possibility to be able to continue to grow the atrophy muscle of SMA patients after the motor neuron dysfunction has been addressed.

Richard, Richard Vosser from JPMorgan.

So just a question, firstly, looking at the JEWELFISH data by age. Was there any benefit in younger patients? So stabilization in the overall population, any benefit if the patients were younger? Second question, on the stabilization of disease, just any sort of payer feedback on their positivity about that sort of level of benefit? Clearly, patients like it, as you say. But any level of feedback there? And then just finally, on the motor milestone benefits in past 12 months, are you continuing to see beyond 12 months? I know it's early, but continuation of motor milestone gains in the RAINBOWFISH study?

So I can dispensed with the first and the last pretty quickly. And then the payer feedback, maybe Simona can address. So again, we have not subdivided or subgroup analysis on the efficacy, the exploratory efficacy of the JEWELFISH. And in the RAINBOWFISH, we do not have the data yet on post 12 months.

Payer feedback on Evrysdi, right?

The feedback on Evrysdi for the stabilization of the disease, yes.

I think Kavita has probably the best handle on these things. So I just would suggest, maybe she could best answer that.

Kavita, please. Yes.

Yes. Yes. No, for sure. And I would say that you know from SMA Europe that patients are really excited about stabilization. 96% 97% of patients see stabilization as an outcome that they're very excited about. And as it relates to payers, as these are early data, we haven't approached payers extensively, but we do see coverage now in the U.S. where we're launched. We do see coverage in patients that are previously treated amongst U.S. payers.

Okay. So it's moving. Does did that address your questions, Richard?

Yes.

Next one would be Simon Baker.

A couple of questions on JEWELFISH, really going back to Richard and Emmanuel's questions. Do you have any data at this stage on efficacy by SMN2 copy number? And will that be -- if you don't have it now, will that data be presented when the trial is concluded? And just going back to Slide 18 with the -- just looking at the patient numbers there, I'm assuming those patient numbers have been impacted by missed appointments rather than dropouts. So can you confirm that there is still 165 patients in that study, i.e., is it 174 minus the 9 that have dropped out?

So your first question about SMN copy number. Was that in reference to JEWELFISH or RAINBOWFISH?

JEWELFISH, Yes.

No. So we have not have broken that down by copy number. In terms of the assessment of patients in the JEWELFISH, there were only 9 withdrawals. However, you are right that there were many study visits because of -- missed study visits because of COVID. So for the manual -- sorry, the MFM32, for example, there were 134 patients for that.

Michael Leuchten from UBS.

Taking the same question backwards. Can you hear me, sorry?

We can hear you, yes.

Taking the same question backwards. If you look at a doubling of the SMN but you do not see a change in functional score, at what level would you have expected there to be a change in functional score? Is that doubling a number that you thought might have triggered that? And if not, why not? And then just because Karl invited a question on Alzheimer's, Paulo, for you, given the change in the FDA stance to potential future approvals, does that make any difference in the way you think about trial programs from your perspective?

Kathryn?

Yes. So I think I need a little clarification on the first question. So is the question if we doubled the SMN protein level, would we have expected an improvement in function. Is that the question?

Michael? Maybe what we could do is we could maybe go for the Alzheimer one, and I could reopen your question later on, Michael, when the line is a bit more stable. Maybe, Paulo, you can go with Alzheimer one, please.

Yes, of course. I mean, it is an interesting question for speculation, I would say. I mean,the decision by the FDA around aducanumab is obviously a precedent setting and opens up all sorts of new possibilities. It's the first time that the medicine is approved for a neurodegenerative condition-based on a serving endpoint, essentially on change on the biomarker with supported clinical evidence. And of course, up until now, this has been a little bit the dream of a lot of us who've been working in this area that you might be able...

We seem to have an issue with the line for -- yes.

Maybe while Paulo is...

Maybe you could take it ...

So maybe I can just take a little bit since Paulo and I tend to talk about this things quite a bit, right? So maybe just, I think where Paulo had left, it's we've -- this is an interesting point in time where potentially in neuroscience the ways of getting approval can potentially be based on surrogate, which is not an unknown path from the FDA, as we all know, right? It just has not happened in neuroscience so much. I think from our perspective, we're excited about this because that brings a new potential medicines to patients faster. But it is going to be important to make sure that the circuits are clearly correlated with clinical outcomes so that we end up generating the right data short term and long term and be able to have the sustainable benefits to patients. We'll definitely be looking at these things ourselves in programs that make sense. I don't know if that answers the question, but it is an exciting time in neuroscience, I will say.

Yes. I mean it's also a bit difficult to comment on those things at the moment as we didn't have any contact with the FDA since the approval of the aducanumab data. And we take it now -- let's say, take it from there. And we will keep you updated as discussions evolve. But I mean what we have shared with investors in the meantime, I mean, unless we have now any other information for you, we would advise the market, let's say, to just go for the full line of the readout. At this point in time, that would be, let's say, prudent to do. And then if there is something else already and whatsoever, we'll keep you updated. Yes. So that's -- and our gantenerumab will read out in 2022. So this is anyway not too long time out from here. But just to give you a bit of perspective where we stand at the moment.

Can I try to answer Michael's first part of his question? There may be a nuance to that, that I missed. But I think what he was asking with the doubling of SMN protein, would we have expected an improvement in strength rather just stabilization. And so we saw a doubling of SMN protein levels in all the FISH studies. And what is different about the JEWELFISH is that there are patients who have severe weakness, low muscle mass contractors, severe scoliosis. And those all play a role in the response to treatment. And so no, I don't necessarily think that a very severe individual who is treated with risk discipline will necessarily have the same response as a younger, less affected individual.

Hi, Karl. Yes, sorry. I'm back. My laptop decided to die on me suddenly.

Gosh. Yes. That happens sometimes in this new world. Michael, I hope you could -- yes, I see that you are at a different place now. That's fine, no worries. Michael, I hope we could address your questions. And I would go on to with Jo, Jo Walton.

My guess is -- Jo Walton from Credit Suisse. I guess my question is along the same lines, and I'm looking at Slide 17. Forgive my ignorance here, but you've managed to double the level of SMA protein levels. How close do you get then to a normal, healthy child? And secondly, given that you appear to be able to double the protein levels in those patients who've had prior treatment with Zolgensma, isn't this the ideal slide to show that you should be using dual treatment? And that if you can take Zolgensma -- because presumably, these were only the younger end of the range who were able to take that. You can -- that in itself must improve the protein. And then on top of that, you're doing it again. Isn't that an absolute shoo-in for dual treatment?

Unfortunately, it's a little more complicated than that in that our SMN2 protein level is measured in the blood because risdiplam is delivered systemically and affects multiple organs. However, other treatment options, such as Zolgensma, nusinersen have either targeted to CNS or muscle or a different route of administration that would not -- neither one of those would show up as SMN2 protein levels in the blood. The reason that we use SMN protein level in the blood is that it is a noninvasive way for us to measure, to have a biomarker of efficacy of risdiplam. So we don't need to do a lumber puncture, for example, to look at SMN protein levels in the cerebral spinal fluid. However, our preclinical data has suggested that SMN levels in the blood are highly correlated to that in the CSF to brain, to muscle.

And can I just ask if there's any view from the regulators or the payers about the concept of -- if you've taken the Zolgensma approach, are you finding that patients are finding it difficult to then add this on top or such a devastating disease that the payers are likely to accept dual therapy?

I'm happy to address that, Jo. And I think it's a great question from a payer lens. And as I said, previously, I mean, in the U.S., the first market that we've launched, we have seen patients that were given Zolgensma or gene therapy being treated with risdiplam. So there is some access in the U.S. for this patient population. And as we see more data and now hopefully with JEWELFISH, there's even greater confidence in the safety that risdiplam can provide in these populations.

Thank you. So there is a question from the chat, which is just with your biggest -- I think Kavita, this is for you. The question is from [ Diana Na ], and she was asking about the Evrysdi uptake so far in the United States. Could you perhaps provide more color where patients are mainly switching from? Are they more coming Spinraza? Are they coming more from Zolgensma? Which subtype are you seeing Evrysdi largely being prescribed? So basically, if you could kind of give us an update on the U.S. market dynamics? And she was also asking about the Europe approval, recent approval, if you could give us an update on where we stand at the moment with the first signs of the launch.

For sure. So I'll speak to the U.S. first, and we're really happy with what's happening in the U.S. We see a really strong uptake of risdiplam across a broad range of patients. So similar to -- in our trials, what we've been studying in a broad range of patients, we're seeing that in the market play out as well. And so what we're seeing is similar to what you see in the type I, II, III. 25% of our patients are type I. 50% are type II, and the other 25% are type III. You asked about what does it look like in terms of treatment naive versus previously treated, we see about 1/3 of our patients are treatment naive. We see about 2/3 that are previously treated with both Spinraza as well as Zolgensma, more switch on the Spinraza side, as Zolgensma is still early in terms of the number of patients that are on gene therapy today in the U.S. So overall, a very strong uptake. We are the fastest, I would say, DMT post -- in this market -- the fastest uptake we've seen thus far with the DMT in the U.S. market. And we're seeing very similar performance now in Europe as well. So the most recent approval in March, we -- our first country to launch is Germany. And what we're seeing there is also a very strong uptake in Germany, similar as we're seeing in the U.S. Thus far, over 200 patients have been treated in Germany. Within 24 hours of approval, we got a patient on product, which is fantastic and shows you that it's accessible to those patients in that market. And now we're seeing -- 6 weeks out, we're already over 15% market share. So again, very strong performance in Germany. And as the other markets get up and running, we'll be sure to share more.

Very good. Thanks a lot for the update. Next one would be Stephen Scala from Cowen.

Can you hear me?

Yes, we can hear you. Yes.

Two questions. And actually, they're both on Alzheimer's. But the answer to the gantenerumab question had all the appropriate conservatism. But there has been a major change in the landscape, and Roche is best positioned to take advantage. So perhaps you can lay out the range of possible alternative pathways to approval that Roche is now exploring. And would one of them include a possible approval of gantenerumab this year? And secondly, initial Roche data on tau was disappointing, but Roche has another candidate in development. Several competitors are following. Does Roche think tau is a failed target? Or are you optimistic that different agents with different target engagement could still show benefit?

The good news is, Paulo, that you are back up online. So I would say we give both to you.

That sounds good, Karl. And it's a good challenging question, I don't want to debate it. We are certainly looking at the full range of options right now, okay? And the extremes, if you want, one of them is obviously a wait for the data readout from our graduate program, which should be coming second half of 2022. And that obviously is the full data package. It includes all the safety, all the efficacy, all the biomarkers, everything else. And that is still our base plan, by the way. That's still what we're doing. But we are looking at all the other options. And the other extreme range of the options is considering something like an accelerated approval based on biomarker evidence. Of course, we don't know what that looks like. And this would require some sort of dialogue with the FDA, with the agencies to find out what they, say, standards would they want to have. And it's something that, yes, we're keeping up as an option for sure. But again, our base case continues to be that we want to have a full robust data set that includes the clinical data because that is what in the long term will matter to physicians and to patients. And of course, we understand there's -- it's a brave new world since last week. And we're not blind to the multiple alternatives here. Regarding your question -- sorry, Karl?

On the tau, that was the question.

Regarding the question on tau, yes, I mean, we reported out the data from our TAURIEL study, which was the first-ever attempt of using a monoclonal antibody against tau to try and have an impact on cognition and function. And as you mentioned, the results were disappointing. There are a number of open questions even regarding that trial and also regarding another trial with the same antibody called LAURIET, which is being studied in moderate Alzheimer's patients. And the questions are around target engagement. We're still looking at our biomarker data to see whether we had enough drug on to target and if there's any correlation with that. Secondly, we are -- we really don't know what the best population is to try these drugs in. And of course, with amyloid, the general mantra has been that the earlier, the better. And -- but it is really something we've learned along the years. And for tau, we're just in the beginning of that journey. And regarding it being a failed target, I think it'd be incredibly premature to say that based on the evidence of one single Phase II study with one monoclonal antibody targeting one of the epitopes. We are looking at other epitopes on tau. We have other programs looking at tau. I think, generally, to answer your question, tau, is without a doubt, as far as I'm concerned, a super important target for Alzheimer's first and neurodegenerative conditions broadly. So we're still very, very hopeful about that.

Okay. If we could just continue here. I have one more from Mark Purcell, which is going to the same direction. If you could maybe talk about the advantages of any -- or potentia, the challenges of a subcutaneous administration approach with gantenerumab versus, let's say, IV aducanumab or with this compound.

Yes. I'm wondering if maybe Simona and I can share this one because I can speak from it from a physician patient standpoint. I mean it's building in the [ 5G ] capacity is going to be a challenge necessarily. And obviously, in terms of convenience for patients, especially patients who are, yes, usually older and more incapacitated and need to be accompanied to visit, et cetera, et cetera, being able to administer at home whether self-administration or administration by a caregiver or by a nurse, I think it's going to be very, very relevant clinically, right? And we do have data showing -- actually have a lot of data showing that it's safe, it's well tolerated, you get the same exposure. So provided the data looks robust, I'd see it as a big advantage, especially for such a big population. We're not talking about the rare disease with a few hundred patients. We're talking about millions of patients potentially globally. So yes, I'd say that's a big advantage. But maybe, Simona, you want to add comments from your side.

Yes. Just continuing where you are. So to me, for me, it's in this particular disease because the prevalence is so hard that I don't think finding ways to bring these patients out of the hospital for treatment, I think it's really, really important. Also, from -- given the disease and the caregiver burden here, I think that's the other aspect that brings value to alternative ways to administer and to continue -- to have appropriate compliance. So I think this is -- to me, the differentiation on the administration side, I think, it's really, really important and for patients and for the family. So I see that as a big advantage for us.

Thank you. We do have 5 more minutes, and we have 2 more questions in the line. One is from Peter Welford. Peter, I open your line from Jefferies.

I've really got just 2 albeit it's still on the broader neurodegenerative area. Firstly, I think it was mentioned by Paulo or I think Simona, if I remember, that there are a number of surrogates that are well established. Curious which surrogates in your view are well-established in neuro as being correlated to clinical benefit and, therefore, could now be considered surrogates for an expedited approval. And then secondly, I guess, I'm looking for a comment that Roche may have on PET scans for Alzheimer's with regards to -- not so much thinking in clinical trials, but more your view on how that is evolving and availability of that in the real world.

Before you answer the question, Paulo, I guess, this come to you. Michael Leuchten just was asking a similar question via the chat. I just wanted to complete the line of questions. Paulo, please. Paulo, you're on mute?

I'm sorry. So surrogacy is -- it is -- okay, so there's a regulatory concept of surrogacy, which requires a certain level of evidence and obviously that the health authority declares it as a circuit. And amyloid is the only one so far that has been declared as a surrogate, meaning that you can take that to substitute for a clinical endpoint and get approval on it, accelerated approval on, okay? Scientifically, there are many other surrogates that have quite extensive evidence. For example, tau levels in neurodegeneration are well-known marker of axonal loss. Neurofilament light chain in MS and in other conditions are well-known marker -- it is again associated with progression with neuronal loss. Some of the imaging markers that we see, whether those are brain atrophy or [ hypo ] volume or ventricular volume, they've never migrated, if you want, to the level of surrogacy that the health authority would say, "I'll give you approval based on this evidence." But for example, even in neuromuscular disorders, the precedent of using the expression of dystrophin in Duchenne for the exon skippers is essentially a surrogate, which has been used for an approval, at least in the U.S. Now one can argue, and I think there are legitimate doubts and questions about the value of all these surrogates and how much they do correlate with clinical evidence. But I think the general point for neurodegenerative conditions has been that one of the key problems that we've had as a field in terms of making progress is the fact that we need very long, very large studies, using endpoints, which have a big signal-to-noise -- sorry, very small signal to noise ratio because just the heterogeneity in these patients. Having surrogates that one can base, let's say, either an approval on or a development on that are recognized by health authorities is or could be a really big development for all of us for this field, okay? The second question around PET scanning. Well, it's certainly becoming more prevalent or more available for a variety of patients. It's still pretty much reserved, though, to tertiary health care centers because it requires pretty advanced technology, right? So it's never going to be as widespread, I would imagine, at least in the foreseeable future as CT scanning or even MRI scanning. So it's always going to be limited just because you need to have the ability to manufacture and distribute new [ clicks ] pretty fast. So for, again, millions of patients needing a diagnosis or qualification view via PET, that is going to be a challenge. That's actually one of the reasons why from the beginning, we developed a CSF diagnostic that could be used and is being used in our graduate program as the eligibility criteria for treatment because, again, CSF testing is a much more widely spread technology that can be easily done. And now we have actually a diagnostic, which is -- got breakthrough designation by the FDA and is approved as a stand-alone diagnostics. So anybody can use that. So I do think for a broader-based population screening, those types of tests are going to be more and more relevant.

Simona, anything to add from your side? Or...

No, no. I think I just said earlier, I think that was probably me when I said that we are excited about the possibility, right, that this approach opens for -- not only for -- in Alzheimer's disease, but in other diseases. And since we are committed to neuroscience at Roche, I think that obviously is something that we're going to be looking at across, at the same time, making sure that we have the right science supporting it, right, which ends up being important at the end to show the clinical outcomes. That's what matters to patients and to doctors and to payers. So I think it's important to do the whole package at some point or other way, but really do take advantage of this potential opportunity because it's not so easy to get to that clinical data, it takes a long time.

Absolutely. Last one will be Sachin, I'll open your line, from Merrill Lynch.

I have a few just quick ones on Alzheimer's again, if I may. So you've mentioned you're pending a meeting with the FDA. Have you requested that meeting as yet? And do you have any time lines in mind? Secondly, Paulo, you mentioned one range is a potentially accelerated file on biomarkers. So I wonder if I could just explore that further. What biomarker data would that be? Is that the open road data, which contains some amyloid reduction at the highest dose rather than the Phase III Scarlet and Margeurite data? And if you could just comment on how you look at the quality of that data versus the analog reduction data that Biogen has in. The reason I asked the question is some of the physician feedback that we've had on that data is very strong given that you're getting amyloid negativity in a high percentage of patients. And then the third question is, as you discuss with the FDA, one assumes that they're open to some sort of proposal. So I ask the question as you think about the Brain Shuttle gante program, which is in the sort of sweet spot of potentially benefiting here. So just -- I'm sure you've discussed this internally, but what sort of proposal or thoughts do you have around an accelerated sort of Phase II/III program when you think about level of amyloid reduction, number of patients and what cognition data would be given at this accelerated stage? So apologies for those crushers, but I would love the thoughts.

Yes. Thanks, Sachin. And I think you may be overinterpreting a little bit what I said because I said we are thinking about having meetings with the FDA, right? So we've not booked anything. We're not on the flight to Washington. There's nothing like that. And of course, these are delicate conversations that we want to make sure to respect the regulator's time and viewpoint as well. So we are considering right now, okay? We have the range of options, we are thinking about them. So anything else is speculation right now, okay? Now regarding biomarker data, I mean there's lots of data that we've been showing throughout the years and which is public, as you mentioned, Margeurite RoAD, Scarlet RoAD data, now open RoAD data. We think there's very interesting data there showing amyloid negativity after not so long of a treatment period and after 3 years about 80% of people being amyloid negative. So -- and there's a dose response relationship. We've shown a lot of our modeling data that underlies our dose predictions for the GRADUATE studies. So there's quite some data, which is robust and public. We've not had any conversations with the FDA. And therefore, we don't really know what type of data, if we were to explore that option, they would be interested in. I would imagine that data, which is probably might be a part of it. I don't know if it would be enough or if there would be requirements for, let's say, more safety data because it's a new material. It's G4 and the RoAD studies were done with G3. So there's a number of technical complications, right? And then again, if we were to approach the agencies with this, we would be approaching it in the spirit of full openness and collaboration because, obviously, we appreciate that this -- it's a great new world for them as well. And therefore, we want essentially to explore the option of working together to see what could be possible, right? But I do think that it is a very interesting time right now because of the acceptance of amyloid biomarkers as a surrogate potentially for clinical outcomes. I mean does beg the question, what type of data would be required, right? So again, we're just -- we're entering into this conversation with open hearts and open minds, knowing that this is going to be something, if we explore, that will take a lot of collaboration between the agency and ourselves and any sponsor for that matter.

Yes. Thanks a lot. Yes, thanks for your interest in Roche. I think what we could present to you today is really a very exciting new data for risdiplam. Making good inroad in all markets where it's already a bit established, like in the United States, but also the first signs and signals in Europe with a 15% market share in a very, very short period of time actually speaks for the drug, speaks for the acceptance in the market. There is a clear evidence of best-in-class for type 1, type 2 and type 3 data building here more and more. You will hear more about the next weeks and months on it. I'm not surprised that we had a lot of questions on Alzheimer, I have to say. As I say before, there is so much we can't say at this point in time. There's a bit of a limitation on our side as well. I think you will understand it. But clearly, the field is moving more fast than we would have expected 2, 3 weeks ago. So that is actually encouraging. And that brings me to a big thanks to Kathryn for presenting the risdiplam data. Thanks to Kavita. Thanks to Simona. Thanks to Paulo. I also wanted to thank Gerard from our team for taking the lead on that call, putting the Q&A together and the slides and the story line and also Melanie for helping with the organization. Wishing all of you a nice day. It was a pleasure having you heard. Thanks for your interest in Roche, and all the best to you. Bye-bye.