Thanks, everybody. Super excited to moderate this very topical discussion with Richard Pops, Chairman and CEO of Alkermes. I'm assuming everyone here is well aware that Alkermes put out some data earlier this week in narcolepsy type 2 with the orexin compound. Maybe Richard, we'll just get right into that. I think I'd love to have you start with just one, what are your key takeaways? What have your conversations with investors been like? What have people been asking you? Then we can just kind of dive deeper into it.
First of all, thanks for having us.
Yeah, it's good seeing you, Matt.
The timing is perfect because that seems like a week ago we put out that data yesterday.
Is that right? That was yesterday?
Yesterday.
I just live in this casino in here with no clocks or lights.
I also hope you notice that we put out a lot of data. We gave you essentially a download from our top-line report, which is a regulatory document that gives you the data. I think that that's a really important point because we're trying to give you guys information to understand this evolving class of pharmacology, which we think is going to be incredibly important and we think is getting increasingly credentialed with each new slug of data that gets generated. I think also it's just an example of how what you guys do for a living is different than what we do for a living. We're developing drugs over a multi-year period, over multiple trains of development, with the eye of making a product that can address major unmet medical needs for a whole range of patients. In this particular case, it means NT1, NT2, IH.
What's different about those three designations is that the disease is actually quite different. Particularly NT1 versus NT2, the data set we just released reveals absolutely new knowledge about a patient population that had never been studied in such a large study as a single entity. Previous studies in narcolepsy had melded together NT1 and NT2, or in the old parlance, this idea of narcolepsy with or without cataplexy. The data that you see is an amalgam of those. This is a pure study in NT2 patients, which is a highly variable patient population. You can see that in the data. What's so powerful is that through all that variability, which we anticipated, the signal of the drug comes very clearly through. A, that would be the efficacy signal that just emerges clearly from this.
The second, of course, was the unknown question about tolerability at higher doses. We sat in many a meeting where people said the higher doses that you have to dose NT2 will by definition not be as well tolerated. That was never our hypothesis. What the data showed is exactly that. NT2 patients have a frame shift. They can tolerate higher doses, and they require higher doses to drive efficacy. I think the principal new discovery, the new finding from de novo clinical research in the NT2 patient population—remember, these are patients with a baseline amount of orexin in their brains—is that when you superimpose an exogenous orexin dose, we believe that sometime between the first dose and about four weeks into the study, you start seeing an accommodation. The threshold needed for efficacy rises.
The efficacy signal that we saw so clearly in this study that we showed yesterday on the MWT, which is only one way of looking at efficacy, but on that MWT measurement was driven almost entirely by the hour two and hour four time points. When you look at the individual patient plots, you can see that drop off at hour six and hour eight. The simple observation when we unblinded the data was, if we split the dose, we'll definitely raise those later six and eight hour time points for those patients who want that.
Yep.
Because the other observation was that we had patients who did not look like they had normalized on MWT, but normalized on Epworth Sleepiness Scale, and vice versa. The heterogeneity of the patient population was just self-evident, but the power of the drug to power through that was also self-evident. The last point I'll make, because I think over the last 24 hours, we spent a lot of time with investors and salespeople who just don't understand the following point. When we look at the maintenance of wakefulness test, the MWT by itself, in NT1 patients, because there's a fairly consistent response to the drug, the MWT moves in a fairly consistent uniform way with dose. In the NT2 patient population, you have a very clear population of non-responders on MWT.
To make the point simple, if you took two patients, a responder and a non-responder, let's say the responder has a 30-minute MWT change, non-responder has a 2-minute MWT change. Add them together, that's 32 ÷ 2, that's 16. That's your average. No patient had 16, but that's your average. When we look at the averages in the NT2, it's a mathematical construct that you use to compare two different groups statistically, but it's not a measurement of central tendency because of the difference across here. We do absolutely see dose response in the patients when you look at the individual patient levels. If I were to give you 10 mg, 14 mg, and 18 mg, we would hypothesize that you would see a difference between those three doses.
I understand.
When we put that into a whole group where some people respond, some people do not respond, you look at the average, that does not present itself. What we did not see dose response on as much was on adverse events. The overall adverse events profile was very generally well tolerated. The most commonly reported AEs were the ones you would expect of polyuria, urinary urgency, insomnia, and visual disturbances did not rise to the level of the most commonly reported AEs, which I think should surprise most people. I think so taken in the aggregate, we were quite confident this drug would be approved for NT1 based on the Vibrance One data. We are quite confident that this drug will be approved in NT2. Let me blanket that with the overall admonishment in drug development that things can go wrong and there is always a risk in this thing.
I think this class is incredibly, it's being advantaged by the fact between ourselves and Takeda, we're populating with a lot of data supportive of the use of these medicines.
Okay. I have five follow-up questions from that great opening remark. On the accommodation side, can you talk a little bit about this, about the MWT over time? What is your level of confidence that this isn't tachyphylaxis that over six months is going to essentially evaporate the drug effect for this population?
The data. That's what we believe this accommodation, tachyphylaxis, sensitivity, tolerance, whatever you want to call it, is happening between dose one and about four weeks into it.
Was your four-week effect size relatively similar to the eight-week effect size?
Yes. When we look at the ESS data, you can see that plot, which we provided, it very much like the NT1s, drops very quickly at week two and is sustained. We know from the open label extension, which is already ongoing, it's continuing into the OLE. We see no diminution of signal. Whatever that reaccommodation occurs, remember in our phase I-B study, we saw no such drop-off in that short exposure. By week four in our hands, on MWT I'm talking about now, you see that drop-off at those later time points. It's not subtle. You can see it very, very clearly in the data.
Yep. Okay. On this idea of taking the doses and splitting them, is that going to be good enough? Couldn't you lose efficacy in your morning time points? I mean, how do you know you don't need to just give the same QD dose BID?
What we have now that nobody else in the world has now is 93 patients of data over eight weeks, plus another five weeks looking at variability and dosimetry in these patients. We can model PK, we can model response. The answer to the question is, there are two elements to what we're processing right now in the company as we prepare for phase three. In the NT2 patients, clearly we can go higher on dose if we need to. We will continue with a once daily dose in NT2 because we think they're patients who are really benefiting.
Maybe you go up to 25 or something.
Could go higher if we want to. We have not decided that yet. Then we can split some dose between 0 and 25.
Could you be splitting 25?
We could be splitting 25.
Instead of doing 18 QD?
Exactly. It isn't necessary. Take that as a number. Would it be 12 and a half, 12 and a half?
I made that up, but just like something that's slightly higher is a little bit more efficacious.
That's right. How do you divide over the course of the day? We can model that.
Would you do 2/3, 1/3?
Exactly. Or one-third, two-thirds.
I mean, I think you might want a better MWT at 8:00 A.M. versus 7:00 P.M.
You know who has the best data to model that? We do now.
Yep. Yep.
We think that you need a large cohort over multiple weeks to be able to power that analysis.
Is there an irony here where maybe Takeda doing the BID thing was kind of right all along?
I think they did it for different reasons, though.
Okay. Yeah. Do you want to talk about that?
You know, I'm a fan of Takeda because I think they're doing this correctly. I think they're developing their drug well. I think their drug is what it is. It's a twice-a-day drug at one dose in one indication. I think it's a major contribution for patients with NT1. I think it opens the door for all the rest of us to come behind with improved and even themselves with better drugs. What do we want? At the end of the day, what we want is a drug that can be used in diseases of hypersomnolence, NT1, NT2, and IH. Why? Because the diagnosis is very fluid between all those things. We want a range of doses. It doesn't matter. You shouldn't just be pigeonholed that you get one dose because you're NT1. Maybe there's doses available 4, 6, 8, 10, 12, 14.
I'm just making that up. There's a range where the differential diagnosis doesn't matter. The dose, you can adjust like many other psychiatric or neurological.
Right.
Now, we heard from investigators in Europe, for example, some patients would like to stay up until midnight. Right? That's a different PK profile or a different dose than somebody who wants to go to bed every day at 8:00 P.M. It's, why would one dose accommodate both of those? Guess what? It won't. What we've learned in the NT2 patients is that splitting the dose probably responds to the biology in a way that we never would have known before.
It's interesting. It sounded like you were alluding to in your open remarks that the response curve here could be like, I mean, maybe bimodal is an exaggeration, but something that more looks like that versus like a bell shape. Is that right?
Is it the distribution of responses?
Distribution of efficacy. Like, it's like you've got some big responders and some non-responders.
For sure.
Do you have any?
That varies by MWT assay or the ESS assay.
Can you actually phenotypically discern who these people are?
Yeah. But now we just got this data set. First question, can you look at the baseline characteristics or any other thing to discern what camp they're going to fall into? So far, the answer is no. Ideally, what you would want to do is you want to do lumbar punctures in all these patients, see their baseline level of orexin, and then match the dose and see whether there's correlations. I think that research will all happen over time now that this field is getting so exciting. Right now, I think that's why we're so encouraged that through that variability, we can drive a P value even without an unoptimized stat plan in a first-time study to see that efficacy.
Yep. Okay. It feels like the street for NT2 kind of got emotionally benchmarked to some of the single dose data from you guys and the old Takeda IV data. I think it's almost like some of the people who thought this was going to work thought the effect might have been bigger. How do you put a 7 minute-10 minute MWT effect into context for the competitive landscape in these patients?
There is no competitive landscape because nobody's there.
For NT2?
We're the first patient, the first company to show randomized double-blind.
Yeah. I actually don't even mean for just orexins. I mean in general for the drugs that are used, like there's PROVIGIL, NUVIGIL, XYREM, WAKIX. There's other stuff here.
Good question. To that point, what's interesting about those data sets is they are an amalgam of NT1 and NT2 patients.
Yep. They are. All of those data.
Yeah. There's not a pure NT2 cohort in those. I think the apples to apples might be to blend our NT1, NT2 results and show the difference.
Interesting. Like with the XYREM, XYWAV data that's showing eight or nine minutes or six minutes. It depends on the study.
Those studies, and anybody here, correct me if I'm wrong, but I believe they all enroll narcolepsy patients with or without cataplexy.
Interesting. Okay. Okay. Yeah. Okay. Fair enough. Do you think this is the best result for a pure NT2 population that we've seen? Do you guys know of something that's better?
I think it has a couple of virtues. One, it's a new mechanism, and it's not a scheduled drug like an oxybate would be. I think that we're just beginning to tune the maximum efficacy that we're going to see in phase three. For a first foray, it's funny, we applied a phase three level stat plan to a phase two population where we had no experience with the magnitude of the changes on the endpoints. We set up an architecture of splitting the alpha and cascading through various doses, completely just based on instinct. We had a positive result on that.
Yep. Yep. Okay. Can you clarify some of the?
I was going to say it wasn't based on instinct. It was based on things other than instinct. That's correct.
Can you clarify some of the stat stuff? I think there's some confusion around some of the lower doses. Would X or Y have been stat-sig if there was not a hierarchy? Maybe just walk through the pre-Hock hypotheses and yeah, that would be helpful.
Like I said, this is a phase three structure applied to a phase II study. In phase two, you're looking at your dose ranges and looking for the quality of the efficacy signal that you can derive from the study. In this case, with both the Maintenance of Wakefulness Test and the Epworth Sleepiness Scale as dual primary endpoints, this level of statistical significance begins at the highest dose at a level of 0.25. Right? You split the alpha of the 0.5. If you win on both of those, you can move to the next dose and so on. That's why at the 10 mg dose, even though we had a nominal P value well below 0.05, technically we couldn't analyze it because.
It's technically exploratory.
It didn't get past the cascade. We have a nominal P value that's significant on ESS at the 14 mg, but with the alpha being split, it doesn't meet the prespecified. We tend to look at more just like what are the data showing us rather than what's the stat plan.
Yeah. I guess going back to the whole dose selection question, I can totally understand the idea of BID dosing. Maybe the total drug volume is a little bit higher. Is there a reason to think that a higher QD dose is going to get you that much more efficacy?
Only because I think the easiest way, if you really want to maximize MWT, if that's the assay you said I want to active on, the easiest way to do it is split the dose. We do think we do see dose response on duration, which sort of stands to reason with proportionality of it.
Is splitting the dose based on the way the other scales are captured going to help you with the other scales, or could it hurt you?
I don't think it'll hurt us, no. I think.
No? Like when is Epworth taken? Or is it taken all day?
Epworth is a patient-reported measure that integrates their experience over a multi-day period of time.
Got it. Okay. So it's kind of how you've been feeling.
How you feel. What is your likelihood of falling asleep in the following circumstances? And our baseline.
Splitting the dose would help, probably. If one of those circumstances is that dinner. If they're describing different scenes, right?
Exactly right. What you would worry on a split dose notionally, conceptually, is if you have split the dose and you give too much too late, could you drive more insomnia? Could you drive any other thing? That is why the modeling and the experience with so many patients gives you the information to more accurately model that.
Yep. Yep. Okay. How are you comparing your data with the Synthessa data?
I don't. I think there's literally no way to compare it.
You haven't made like a PowerPoint deck comparing it yet?
That's your job.
It's actually publishing at 4:01. Before then, I'll take any notes. Like what are the, you know, you look at their data, like what do you think?
I don't see any data. I see your press release. I think you need to present data. What I'll say is I think that what we've learned is that small cohorts of patients at two weeks is not comparable to large cohorts at eight weeks. I think that's without editorializing. It's just a fact. To compare the data sets, you need two data sets.
Okay. You are confident based on all the data you have that this tolerance dynamic is kind of a one-month adaptation, then we're likely done.
That's what the data is telling us.
Okay. Okay. So what are you going to, what else do you have from the study that you might be able to share with us versus what might you keep in-house competitively?
Certainly our choice of doses, we're going to keep that competitive.
For the phase three?
As long as we can. Because we think there's real learning.
I mean, eventually that's going to have to be out there when there's like a protocol, right? Or is there a way to keep that confidential?
Both. I do not think we have to be cloak and dagger about it forever, but I think for a while, it is really interesting. It is new and it is important. It is insight. It is how I got to this idea that this is why you do clinical research, because you learn new things. NT1, we had a pretty good sense of how to model effect size from our study and from Takeda's study as well. NT2 was completely new turf. There was no precedent. There was no way of knowing. When we elevated ESS and MWT, it was really based on this idea that both are important and we really do not know the power calculations of the two. Interestingly, in the idiopathic hypersomnia study, MWT is not a primary endpoint. It is not used, even though those patient populations overlap greatly with the NT2s.
In that study, ESS is the primary endpoint. You use IHSS as the secondary. This is new clinical research because these are the first agents in this field that drive wakefulness. Think about oxybates drive sleep at night. These are the first agents that are not stimulants that drive the brain circuitry that affects wakefulness.
Do you think IH is going to be easier or more challenging for orexins than NT2?
Our belief right now, not seeing it, it should be very similar for two reasons. One is that we were quite encouraged to see with all the variability in NT2 that the drug signal clearly emerges. We expect a similar amount of variability in the IH population. Again, different endpoints, so we'll see how.
Yeah. Can you review people those endpoints and how you're thinking about efficacy there?
What we just did in the NT2 study had two endpoints. The MWT we've been talking about, which is this measurement of how long you stay awake in these 40-minute bouts over eight hours. Then what's called the Epworth Sleepiness Scale, which is eight questions scored up to three points of measuring how sleepy you are in various circumstances. It's a very simple assay, but it's used widely clinically, whereas MWT is not used clinically at all. It's like that first patient screen in the office to determine whether somebody has a disease of hypersomnolence, and it's widely used. It's also been a regulatory endpoint for approval of drugs. As I said, NT2, it's ESS, Epworth plus MWT. In IH, it's the ESS primary. Key secondary is what's called the Idiopathic Hypersomnia Severity Scale, IHSS.
Those will be our two primary or first and key secondary endpoints. We are also going to do MWT in that patient population as a secondary just to see what the variability is, see how it compares with the NT2. What we know from the real world is that many patients that we've talked to were first diagnosed as NT2 and then as IH. It is a very fluid diagnosis. Often the pivot is just the number of sleep onset REM events you might have in one overnight stay in a sleep lab. If you have two, you have one disease. If you have one, you have another disease. It is not the most crisp demarcation between them.
No, it's not.
We expect the patient population is to overlap.
Any questions for Rich? On narcolepsy type 1, I think there's still a wide variance in how people are thinking about the market opportunity. On the one hand, you could paint this very cautious picture that the oxybates sell X. Those drugs are going to become extraordinarily cheap at some point, or some of them will. And so you're going to be kind of fighting over the same patients. There's going to be payer hurdles, right? The other hand, you can kind of talk about all these patients that have fallen off oxybates. How are you guys mapping out the different cohorts of patients that are likely adopters of orexins?
To our mind, it's quite clear in NT1 at this point, which is the wrong way to look at it, we would say would be to start with the oxybate market and figure out what fraction of that you're going to get. The simple reason is most patients aren't on oxybates. Oxybates are not easy to use. They're scheduled drugs. They're nighttime. It's a real commitment to be on oxybate. Many patients have tried them and end up not staying on them. There is a dedicated cadre of patients who really do like the oxybates, but they're not disease modifying. In NT1, our belief is that the addressable population is the entire population of NT1. This is confirmed by talking to practitioners.
With a disease modifying agent whose efficacy data, if they continue to bear out like we've seen in phase two and like Takeda has seen, these are much, much more efficacious drugs than current therapy. They're quite well tolerated now. We expect it will become the foundational treatment in NT1. NT2, I think it really depends on that distribution. I think there are NT2 patients who look very much clinically like NT1 patients. I think there's also NT2 patients in that designation who will do just fine on modafinil or some inexpensive drug. Then everything in between. I think from a modeling perspective, we would model various degrees of penetration in NT2 coupled with a very deep penetration in NT1.
Do you have a good idea of how many diagnosed NT2 and IH patients are in a specialist care right now?
We know that about 80,000 patients are being treated right now for narcolepsy. Only about 16 or 17,000 are getting oxybates. The epidemiology has shifted over time. We showed that slide last year at our analyst day. In the old days, the estimate was it was two to one NT1 to NT2. Now it is almost flip-flopped. There is probably twice as many patients with NT2 as NT1. As you talk to practitioners, you will hear more and more with the orexins coming, many patients with what have been called NT2 are actually NT1 because their cataplexy is very subtle. We do not do lumbar punctures for the differential diagnosis in the U.S. I think both those diagnoses are, I would not fixate on the number specifically, but the basic math looks like there are more NT2s than NT1s. They are both big opportunities for better treatments.
Makes sense. Are you able to talk about Avadel or not yet?
Can't really. That is only because we're in the midst of this merger. It is governed under Irish takeover rules. We filed a big press release called a 2.7 disclosure. It has a lot of information. I think the preliminary proxy is out today that has additional information on it. That is priority to us.
Okay. So you can't talk about your view on the durability of oxybates?
No, I shouldn't. I shouldn't. But I'd love to. But it's obviously foundational to our thesis. But we can talk about it.
Yeah. Yeah. Yeah. Okay. Okay. So what else should we cover with alixorexton? Maybe just talk about your plans with engaging the FDA and what the timeline is there.
This group of drugs is under the authority of the Division of Psychiatry at FDA. They have been very clear with us and I think with Takeda what their expectations were for the phase two design. That is why you see the phase two design that we have, which we have now satisfied through two different studies. What we are doing right now is pulling together the information to drop into the meeting request for the type B meeting, which is usually granted within 60 days of the request. We expect to meet with FDA in early Q1. We hope to light off the phase three program in Q1. We are designing all those protocols now. The actual architecture of the studies is not particularly mysterious. It is what you have seen before. It is like our phase two and like Takeda's phase two.
Parallel design, three-month group sizes of 40, 50 patients per arm using NT1, cataplexy, MWT, ESS, NT2. We would expect ESS and MWT. There is not a whole lot of controversy about how you design the studies. It is more about what doses we are going to use and what are the requirements. We are really interested in these other outcomes, namely cognition and fatigue. We were the first company to show a real powerful effect on cognition and fatigue. We chose those two domains because in patient research, like I said, people do not talk about MWTs. They talk about brain fog and fatigue. Fatigue being different than sleepiness. You can get a good night's sleep and still feel exhausted. This is what NT1 patients talk about a lot and NT2 patients. We put these measurements into our phase two study of measuring cognition and fatigue.
What we saw in the NT1 study, normalization across these domains. Really exciting stuff. What we'd like to talk to FDA also about is how do we incorporate those. Whether they're label enabling or not remains to be seen. We're certainly going to capture them in the study and publish on them and educate on those.
If IH is positive, assuming you'd advance that, but are there other indications that could be coming soon after?
For 2680 or alixorexton, possibly. The core for alixorexton will be these diseases of hypersomnolence. We have two other rexins in the clinic right now. One is nearing the completion of its single ascending and multiple ascending dose studies in healthies. The other one is in the midst of those right now. We are thinking that in 2026 we will move into some of these new areas that are not narcolepsy or hypersomnolence, but where there is a really strong biological rationale in order to justify treatment in those areas.
Would you advance any of those for narcolepsy or IH or no?
No. No.
Okay. No need.
No, we think we've got the right configuration in alixorexton for those diseases.
Okay. Okay. Anything on the commercial business you want to highlight or you feel like is specifically being overlooked right now?
We were pleased with the quarter. We are pleased with the year. Selling drugs in the U.S. is complicated. With all the dynamics affecting these markets, we are really pleased to see continuous growth. The growth is driven by demand. I think that is probably the healthiest sign for us, the growth in demand. Thankfully, there is not a lot of drama there. I think it is a billion-plus dollar top line that generates significant profitability that could fund the expansion of this pipeline in a non-dilutive way. I think even the Avadel acquisition that we talk about, it is not using stock. That is using cash that derives from a strong economic underpinning of the business, which is rare.
Awesome. All right. Any last questions for Rich? Okay. Mark.
Rich, you followed the FDA closely for 35 years. We know each other. This has been an unusual time at the FDA. What's your view on some of the FDA predictability and some of the, I know your involvement in bio over the years, this is something you've got a very good handle on. You've seen a lot of crazy stuff.
The thing I'm most focused on, Mark, right now is PDUFA 8 reauthorization. At the beginning of this new administration, there were some questions about regulatory capture and the viability of user fees, yes, no. I'm quite pleased that PDUFA 8, which negotiations just kicked off last week actually, looks like it's going to be normal course. As you may or may not know, these user fees fund approximately 80% of the new drug review in the U.S. To remove them for whatever reason would have been catastrophic. It's really nice to see that so far appears to be moving a pace. I won't comment on the various things that happen from time to time.
I think that in our little world, looking at the Division of Psychiatry and the regulatory pathway for a drug like alixorexton, because the safety appears to be so acceptable and the efficacy appears to be so robust and it's doing new things for patients, those lanes tend to remain undisturbed as we develop new medicines. They're not politicized in any way. I think I'm hopeful that they just run normal course.