Roche Holding AG (ROG) Earnings Call Transcript

Ladies and gentlemen, welcome to the Roche live audio webcast on efficacy and safety of the risdiplam in patients with type 2/3 spinal muscular atrophy presented at SMA Europe 2020. I'm Andre, the Chorus Call operator. [Operator Instructions] The conference must not be recorded for publication or broadcast. At this time, it's my pleasure to hand over to Karl Mahler, Head of Investor Relations. Please go ahead, sir.

Yes. Thanks a lot. So good afternoon and good morning, ladies and gentlemen. A warm welcome from our side for everybody who is on the webcast and online here to review the SUNFISH data we just have presented at the SMA Europe in Paris. With us is Paulo Fontoura, he's the Global Head of Neuroscience and Rare Diseases Clinical Development, who will lead us through the data. And on the phone, we also have Michael Ostland, the Lifecycle Leader for the SMA program. You will have the opportunity to ask questions after the presentation and allow me also to thank Gerard Tobin and Sabine Borngraber from our team to take the initiative on this virtual pipeline call and helping prepare the presentation. On Slide #4. Many of you know that we had -- that we will have a strong year in terms of news flow. So we expect actually 2 new molecule entity. One we will discuss today is risdiplam and also satralizumab, both are under assessment currently with health authorities. So we are confident to launch these -- to these assets in 2020. We will also have data later this year in UC. For etrolizumab, there will be data in ophthalmology and there will also be data in oncology. So a strong year for news flow. If you go to the next slide, it's the Slide #5. This just gives you an overview of the SMA population by type and age. As you may know, there are 3 main types of SMAs, Type 1, Type 2 and Type 3. And the SUNFISH data will -- which we will show you today is basically focusing on the Type 2 and Type 3 patients, which is the main -- which are the main markets in EU and U.S., and they make almost 90% to 95% of the prevalent population. Additionally, adults make up about 2/3 of this opportunity. And we, unfortunately, have to say, that many of these patients go untreated today. So SUNFISH is the first and only disease-modifying therapy to study this patient population with SMA. And with this one, I would like to hand over to Paulo. Please?

Thank you, Karl, and good afternoon, good morning, everyone. So on this slide, we just like to highlight Roche's commitment to find new best or first-in-class medicines in areas of high unmet need in neuroscience, broadly, across neuroimmunology, neuromuscular, neurogenerative diseases and neurodevelopmental disorders. And neuromuscular disorders is a new area for growth for Roche, where we've recently also signed a deal with Sarepta to codevelop and co-commercialize their gene therapy, Duchenne muscular dystrophy. Now risdiplam this year listed as the first oral therapy for spinal muscular atrophy. And being an oral therapy has many advantages, but the key advantage of an oral daily therapy is actually that it has the ability to increase the expression of SMN protein throughout both the CNS and peripheral tissues, which is very important from a therapeutic standpoint. And also it will provide access for many patients who are more advanced and who cannot take other types of therapies. Next slide, please. So we are very proud of the broad clinical program that we've designed for risdiplam in collaboration with the SMA community. And our idea, really, here was to be able to generate the most amount of data across the broadest ranges of SMA patients in order to support access. So as you can see, the global program has studies both in Type 1 and Type 2 and 3 in SMA patients as FIREFISH and SUNFISH data, which are both positive in terms of their pivotal readouts. The JEWELFISH trial, which looks at patients who have been previously treated with other disease-modifying therapies, which is completed in its recruitment. And finally, the RAINBOWFISH, which looks at neonatal screen patients and is enrolling now. Now, of course, our objective here is to demonstrate a compelling benefit to risk ratio across the spectrum of the SMA types, not just in terms of ages, but also in terms of their real-world presentations and manifestations. So we deliberately included patients with more severe disability. And right now, we have over 400 patients worth of data to date. Next slide, please. Next slide. So what I'm going to show you are the results which were presented earlier today at the SMA Europe conference. And this one -- and in this slide here, basically, just to reiterate that the objective of the SUNFISH trial was to study risdiplam, which is a centrally and peripherally distributed oral SMN2 splicing modifier in a group of patients that is representative of non-ambulatory patients, which are typically seen in the clinics, including teenagers and adults and also patients with reduced motor function, which are patients, which, unfortunately, as Karl mentioned, have not really been studied thoroughly so far. Next slide, please. So SUNFISH was a randomized, placebo-controlled, double-blind study that looked at patients with Type 2 and 3 SMA, age 2 to 25 years old and who were treated for 12 months for the primary end point at a 2:1 randomization ratio placebo versus risdiplam. The primary end point was changed from baseline in MFM32 scale at month 12. And as key secondary end points, as you can see, we had responder analysis changed from baseline in the Revised Upper Limb Module scale at month 12, changed from baseline in the Hammersmith Motor Functional Score at month 12, and changed from baseline in the SMA Independence Scale in -- at month 12. Now an important point here is that because we designed this study to look across a really broad age range, from ages 2 to 25, and we were analyzing the data together not by age group, we have to select a primary end point, which was able to capture clinical change in all these different patient groups and subtypes. And because of that, we opted to use the MFM32 score, which is a broader motor scale than the Hammersmith. And therefore, we felt had a better chance of capturing those clinical changes. Next slide, please. So on this slide, we're showing you the overall baseline demographics in the risdiplam arm and the placebo arm. And as you can see, this is a very broad age 2 to 25 years old, but representative population in terms of its manifestations. The male to female ratios, SMA types, SMA copy numbers are as expected. And as you can see here as well in the final column, we had over -- we had about 38% of patients who were over 12 years old. And these are patients that have never been studied in a typical placebo-controlled trial of any therapy. Next slide, please. In terms of the disease baseline characteristics, as you can see in this slide, they're overall very well balanced between the risdiplam and the placebo groups, both in terms of the age of onset, the presence of scoliosis, and as you can see here, a very large percentage of patients, over 30% of them, have severe scoliosis, having had surgery for scoliosis. Again, we have 25% of patients who had surgery before. And then MFM, RULM and HFMSE baseline scores. And as you can see, there's a wide range of disability here, including a significant percentage of patients with very high degrees of disability as shown by either the Hammersmith or the MFM32. Next slide, please. So the primary end point, as I mentioned, was the change from baseline in the MFM32 scale in the overall population. And the graph here represents those data. As you can see, there's a clear separation between placebo and active after 1 year, with a delta of over 1.5 points at the last visit and a positive p-value of 0.0156. This change was driven mainly by improvements in the younger patients and stabilization of motor function in older patients, as I'll detail in subsequent slides. So can we move to the next one, please? So one key secondary end point was the responder analysis, just to orient you to this slide. On the left-hand side, you see the percentage of patients that had over 3 points change in the MFM32 score, and that is generally considered a responder in terms of improvement. Now importantly, especially for older patients, stabilization is the main treatment goal and this is because in this older population, first of all, the disease state is much more disabled. So there are disabilities that will not change. Being wheelchair-bound, having contractures or scoliosis will not change. And therefore, the amount of change in the scale or in any scale is bound to be small. And secondly, because in terms of what's clinically meaningful for these patients, a recent survey from SMA Europe actually presented at this same conference show that, for these patients, the huge majority of them consider stabilization as the main treatment goal, almost 100%. So we're showing you both types of data. On the left-hand side, improvement as defined by over 3 points change on the MFM. On the right-hand side, stabilization or improvement as defined by any change above 0. As you can see in both of these responder analyses, we have a positive result favoring risdiplam with about a 15% difference between placebo and active and a p-value below 0.05. Next slide, please. Now when we detail these data according to the 4 main age groups, 2 to 5, 6 to 11, 12 to 17 or 18 and above, you see clearly that in the younger population, there's a much bigger percentage of patients who are significantly improving versus baseline. And as then I'll point to the top left-hand graph, which shows, for example, in the 2- to 5-year-old population, the percentage of patients that have over 3 points improvement from baseline is about 80%, which obviously translates to a mean change above 3. In the older populations, as you can see as, 18 and above, as well, there's less improvement from baseline. But importantly, there's a very significant increase in the percentage of patients who's stabilized or has improved. Importantly, as well, in all age groups, whether you define benefit by improvement from baseline or by stabilization, the data always favors risdiplam in both categories. Next slide, please. So what I'm showing you here now are data from the other key secondary end points, the RULM and the Hammersmith. Now to start on the left-hand side, the Revised Upper Limb Module is a scale that measures motor -- upper limb function in great detail. And we've always expected, particularly in this population, there is more disabled that we would be able to see more change on the RULM rather than on the global scale. And that's exactly what we see here. So we see clearly a great separation between these 2 curves across the whole range of this population, from 2 to 25 years old, with a very significant p-value. On the right-hand side, you see the data for the Hammersmith. Now clearly, the data is not positive, and the p-value was not met. But I'll point out a few things. First of all, there is a numerical difference favoring risdiplam across all the time points. Secondly, the placebo group here performed better than what's typically expected for a population of this type. Where you would usually see a decline over this 1 year, which would be very different from the slight change that we see here. Now there are many reasons potentially behind this. One of them being that trials right now are being conducted in very different populations than they were just a few years ago, given the availability of other therapies. Next slide, please. The final clinical equity end point I want to share with you is the SMA Independence Scale. And this is a novel scale, which was developed to assess the level of independence in activities of day to day living, things as you can see here, as eating a meal, brushing your teeth or writing. Now this is both scored by the caregiver and by the patient as long as they're over 12 years old, and that's the data you see here. So on the left-hand side, you see the caregiver reported outcome where you see clearly the majority of patients on risdiplam showing an improvement versus a deterioration of placebo-treated patients with a positive p-value, which is numerical. And on the right-hand side, for the patient, you see the same trend in the data. Next slide, please. Now moving on to the safety profile. Now importantly, the safety profile that we've observed in a SUNFISH study is essentially equal to what we've seen in the Part 1 of this study as well and in the FIREFISH study. So you see broadly that there is -- this is a severe population. So the majority of patients have adverse events as was expected. But importantly, there's no imbalances. And critically, no patient had to stop treatment due to adverse events or severe adverse events and all patients could continue therapy. Next slide, please. Here, you see details for the most prevalent adverse events and severe adverse effects, which were balanced, again, across arms and really reflective of the underlying disease. Regarding laboratory results, vital signs and ECG data, again, no findings were reported of significance and the data were comparable across arms and all the preclinical safety findings were not observed in any patient. Next slide, please. So in summary, first of all, we're very happy today to be able to show positive data in this really pioneering study where we looked at patients with Type 2 and 3 disease, including patients who are much more disabled than is usually studied and that we've really shown positive and clinically meaningful results across the MFM32 and the RULM scores after 12 months of treatment of risdiplam versus placebo. Secondly, that these results also apparently result in an improved independence in activities of daily living using the novel SMAIS scale. And finally, that, again, the favorable safety profile that we have observed so far continues to maintain and that no treatment-related safety findings have less withdrawal from SUNFISH Part 2. And therefore, we feel that this data really corroborates our view that risdiplam is the first treatment to have a positive benefit risk ratio and positive pivotal data in placebo-controlled trials in really the broadest range of population in the SMA patients, including children, teenagers and adults and that it preserves and potentially improves motor function and enables independence for these patients. Next slide, please. And finally, I'd just like to end by really thanking everyone who's participated in these studies, investigated, but most importantly of all, all the patients and families who contributed to this research. Thank you. I think I'll stop here, and happy to take questions.

Yes. Thanks a lot. Just for your information, we have about 100 people on the webcast and 70 online so quite an interest in the call. So thanks for your interest. And with this one, we would like to open for Q&A.

[Operator Instructions] The first question comes from the line of Jain Sachin (sic) [ Sachin Jain ] from Bank of America.

Sachin Jain from Bank of America. Three questions, if I could. Firstly, on the primary end point, the 1.55 improvement from baseline. It's less than roughly 2.7 you saw in Part 1 of the study and below the 3 improvement that generally viewed as clinically relevant. Can you just comment on that? Is that just driven by the age mix and baseline characteristics of these patients? Or is there anything else that we should think about? The second is, if you had any more details on the Hammersmith end point. You've given it for the overall population, I wondered if you have that by subgroup, particularly in the 2- to 9-year-old patients to allow some comparison with the existing agents. And then the final question is on the baseline characteristics. You specifically called out percentage of patients with scoliosis or surgery for scoliosis. Wondered if you just could give some color as to what proportion of those patients would have difficulty or wouldn't be eligible to receive intrathecal therapy with either gene therapy or SPINRAZA.

Thank you. So starting with your first question, the 1.55 improvement. So first -- there are important differences, first of all, between Part 1 and Part 2 of the study. Part 1 was not placebo-controlled, and therefore, the comparison there was to natural history. And of course, there, we are looking at data from a different population in this study. As I mentioned, the patients which are being recruited into trials now are very different patients because there is potentially a selection bias for, let's say, more stable patients now because all of the other ones are being treated. So it's natural that you will see differences. The second thing is that I think you pointed to the main factor that explains, let's say, the 1.55, which is we're really looking at a broad range of patients. Now as I mentioned, in the younger patients, when you see the percentage of patients who improved by 3 or more points, reaching like 80%. Obviously, the mean change will be above 3 and it is. In the older patients, that change is smaller just by the fact that, again, the scale sensitivity to change is going to be much different. So there's nothing really at this point that we can see that will explain that difference. Regarding the details of the Hammersmith. We really don't have those data right now. We are analyzing the data still and we will be presenting it later. And your final point about scoliosis, it's obviously hard to go case-by-case and say whether the patient would be eligible or not. Broadly speaking, though, scoliosis and spinal fusion surgery are contraindications to intrathecal therapies. So a lot of these patients are simply not being treated. And even if they can be treated once, compliance is really difficult just because of that. Also secondly, regarding gene therapy, again, the older the population is -- first of all, we really don't have data right now. Secondly, the percentage of patients with anti-AAV9 antibodies becomes higher with increasing age. So again, we're talking about the most prevalent population here, Type 2 and 3 patients, will either by nature of their disability or their preexisting immune status really don't have other options right now.

There's a question from the back from Marietta Miemietz. I'll just read it to you, Paulo. Could you please elaborate a little more on the drivers behind the high placebo responses, stabilization rates on MFM? So Karl, it seems counterintuitive as untreated patients should gradually get worse. So what happened here?

Right. So again, broadly speaking, there's -- so there's a number of reasons that explain these different placebo rates. So first of all, we are talking about a population, which is much more heterogeneous than has been seen in other studies, and that should be really called out. So until we -- looking to those subgroups and younger versus older, it's hard to deconstruct those data. I can just tell you that, at baseline, for example, we have over 41% of patients that had a Hammersmith score below 10, and these are patients that simply do not move. They're at the bottom, right? So if they're already almost like at a stage where the scale does not move, you don't really expect them to deteriorate further. The second point, which I think is relevant here is that we are, as I said, recruiting patients now for these trials, which essentially have other treatment options. And the availability of SPINRAZA and, more recently, gene therapy has skewed the type of patients who get into studies. So the most severe patients, patients who are at risk of declining faster, will naturally seek treatment options more aggressively. So it's obviously impossible to prove these points, but we do suspect that there might be some selection bias happening.

The next question comes from the line of Tim Anderson from Wolfe Research.

This is [ Richard Wagner ], speaking for Tim Anderson at Wolfe Research. Just one question. Engaging with the data as presented, which is to say, across the entire cohort of patients at different ages and degrees of disability, what I see is that the clearest benefit of risdiplam is a stabilization and that the effects in terms of improvement take time to manifest. Its significance is only reached after a year. Roche has described an advantage, a key advantage of risdiplam, is the ability to have treatment immediately. And that treatment is essential in the disease -- early treatment is essential in the disease. I guess my -- while investors have thought about risdiplam as being a continuous therapy, given the time to effect and its principal benefit as being stabilization, should investors instead think of risdiplam perhaps as being used as a bridging therapy prior to use of a potentially curative gene therapy?

Sure. No, thank you for the question and thank you for the opportunity, actually, of clarifying that because I think there's potentially some confusion or some misreading of the data here, right? So when you look at the primary end point data, I guess, what you're pointing to is that significance is only achieved at 12 months. Again, let me point out that the data we're seeing at here is data for the ages 2 to 25 years old, right? So naturally, the effects are, let's say, diluted, if you want, by the significant numbers of patients in this data set, which are older and therefore, did not improve as fast or did not improve, but just stabilized. When you look at the younger patients, and that's the data I tried to show you for the responders. And I'm sure, again, more data will be presented later. You do see a fast pickup in terms of change, that patients do improve faster and improve more significantly. And actually, if you look at FIREFISH, the data we've presented already and will present again soon, you see, let's say, the effects of -- that risdiplam has very, very early on. So this manifest itself within weeks of starting treatment, not within years. So the speed of onset is very dependent upon age. And that is, let's say, a pattern that we see across all therapies now, which is the earlier you treat across all therapies, gene therapy, intrathecal, risdiplam, you see that the younger patients are, the faster the onset of action is, which is probably due to the fact as well that, let's say, first of all, the body have -- is in the growth stage anyway. So it will pick up more motor function. And also secondly, there's less entrenched disability. So there's more of an opportunity to see change. So we certainly see risdiplam fitting that same pattern, is that you should start treatment as soon as possible because the earlier you treat, the better the outcomes are. The point about risdiplam being available or being, let's say, more easier to access more sooner has to do with the fact that I mentioned in the beginning, which is being an oral therapy that does not require an intrathecal delivery, does not require antibody testing. It's something that essentially, after diagnosis, you can start the next day. And again, every day counts for these babies. Everyday counts because every day is an opportunity to regain motor function, okay? So I hope that kind of clarifies this point. Even though the broader population, of course, in older patients, stabilization is mostly what you see, in the younger patients, that is not true. You see very significant and definite improvements in motor function.

Does that help you, [ Richard ], the answer?

Yes.

There's another question from the back, from [ Mike -- from Colin ]. He asked about, is there any chance that the improvements in the risdiplam group could continue past week 52 and increase separation from the placebo?

Thank you. I mean as you probably saw from the trial design slide, we are continuing to follow these patients for another 12 months now, so we certainly will be looking at those data. There's no reason to believe why improvement should not continue. Again, these patients are getting older. Some of the patients who are getting older get more to the stable phase. But of course, patients who are still in their growth phase for the younger patients, we have every reason to suspect that they will continue to get benefit. Yes.

The next question comes from the line of Sarita Kapila from JPMorgan.

This is Sarita from JPMorgan. So could you just let us know what some of the early feedback you've been receiving from focus groups and physicians are on potential SPINRAZA switches, particularly in older treatment-naive patients? And then is there potential for risdiplam to be approved in the U.S. earlier than the PDUFA in May?

Thank you. Well, regarding your first question between switchers from SPINRAZA to risdiplam, we -- so the JEWELFISH study, as I mentioned, is looking at that precisely and we really don't have clinical data yet. We hope to have data later this year. The data we do have is from safety and from biomarkers from earlier patients that we do see essentially the same profile, which is it's very, very safe, very well tolerated, and we do see this, roughly, the same levels of protein increase that we see in SUNFISH and FIREFISH. And so that being said, again, our expectation is that the clinical benefits are likely to be similar. Of course, I want to comment on anecdotal data from individual investigators or individual patients because it's very hard, first of all, to make sense of those data without any control. So right now, I really don't want to speculate there. Regarding earlier approval, I mean our PDUFA date is in May 24, I believe. We are having ongoing conversations with the FDA and we sense that there is openness to look at these data. I mean obviously, I think everyone wants to make the right decision here. And we'd be looking for every opportunity to work closely with them to try and make risdiplam available for patients as soon as possible, but it's clearly like an FDA decision to do that.

The next question comes from the line of Rajan Sharma from Deutsche Bank.

So it's clear that the benefit of the drug is kind of more modest in older patient population, and it seems like generating long-term data supporting the stabilization will be to pricing in this asset. So could you just kind of discuss how you might go about collecting that data, given that it's difficult to kind of conduct long-term placebo-controlled trials? And then secondly, just on pneumonia signal. Could you just discuss the severities of this and potentially discuss what mechanistic explanation could be?

Right. You're coming across a little bit muffled. I think I understood your first question to say how we're going to collect longer-term data, right? Especially in the older population who is more stable. So as I mentioned within the SUNFISH study, we're now at the second year of this trial. And after that, we have a long-term extension study. Well, we hope to keep most patients on drug. And obviously, we'll be collecting data continuously for a few years. I mean that's the typical thing that we do in these chronic conditions. So besides that, we have the JEWELFISH study as well. And therefore, we hope to be able to keep generating longer-term data that will help support the benefits of risdiplam access for these older patients. Regarding your second question about the imbalances, which we're seeing in pneumonia. Again, it's a numerical imbalance between both arms. We're talking about relatively small numbers. None of these were, let's say, particularly concerning in terms of their severity. I mean there are serious adverse events. But again, this is a sick population that unfortunately have a lot of respiratory infections. And there's no reason, again, either from preclinical data or from clinical data to assume that this is in any way related to risdiplam. So actually, none of these were related adds to the drug. So in all, I think, again, this is the data that we have now. Obviously, we care about patient safety, first and foremost, but we have no reason to believe right now that this is in any way a significant red flag for us.

And we also have seen the safety profile from the other trial. So I mean if you put everything together...

Yes. And when you look at it again -- that's a really good point. When you look at the older 400 patients we have in all of our studies, this is essentially not a signal that's benign.

There's another question from...

Could I -- sorry, Karl. Karl, it's Michael Ostland here. Could I interject one point on pneumonia, just from the podium today. Dr. [ McCurry ], the lead investigator also added his comment. He was actually surprised more about the low 2% rate in the placebo arm compared to the observed 7% rate that we saw in the risdiplam arm. So he said these were typical manifestations that you'd see in SMA. And if anything, he would have expected a much higher rate in placebo. Just to add some color to that. Thanks.

Yes. Thank you, Michael. So this was Michael Ostland from -- joining us from Paris. There's another question basically going into the same line of questions, which you already had from Mark Purcell from Morgan Stanley. He asked, "How can we compare these data to the SPINRAZA data ages 2 to 9 years core from the label?" Where he talks about the 3% change from 56% versus 26% at 15 months. So there is an absolute change of plus 3.9%, but there was -- versus a minus 1% in the placebo arm, and we have a plus in the placebo arm. So he was also wondering where does that come from and how do you explain these things according to the age core. And maybe it's growing a bit in the same direction what we had before. So why do we have a higher, let's say, in favorable placebo than other trials?

Right. So I think the question I guess behind the question is how can we compare across therapies, right? And clearly, I see that this is a point of interest, but unfortunately, we really don't have data to make those types of comparisons. There's -- the only true way of doing this is a head-to-head study. And the reason for that is that there's numerous differences between patient characteristics, baseline, demographics, et cetera, that influenced those results. So as I said in the beginning, our objective was really not to do a comparative study. We wanted to look at the real-world SMA population, including these older patients, which had -- were not studied with CHERISH or SPINRAZA, generally. And just to give you some example of how different those populations are, I mean the population we have in the SUNFISH study is much broader in terms of age, the percentage of patients with scoliosis. The baseline Hammersmith scores where we have 41% below 10, all of these patients would have been excluded from that CHERISH study. So it's really very different populations. Also the duration of disease at baseline, we're talking about very different ones here again because CHERISH was conducted at the time where there were no other options, so patients will go into trials really, really fast. And so we really don't feel actually that it may be possible in the future to conduct head-to-head studies because those -- population of those comparisons are going to be very, very different. So I think it's just really hard to make any bridges across these 2 studies. What we do see is that, in the SUNFISH data, as I've pointed out, even in a very different population, which arguably is much more severe because it not only includes a high percentage of older patients, a high percentage of patients with severe scoliosis or surgery, a high percentage of patients with very low baseline scores, we still see a significant result even in the overall population. She's like the worst thing, like the worst-case scenario, the most stringent analysis. Even then, it's statistically positive and it's certainly very, very positive in the younger patients. And in the older patients, you see again the stabilization of disease, which is the thing that really matters. So overall, I appreciate the need, the desire to do comparisons, I'd say that these data stand on their own.