so much. Day three, getting to the late afternoon of our Jefferies London Healthcare Conference. It's great to see everyone and the attendance this year. It's been fantastic. My name's Akash Dewari. I'm a pharma and biotech analyst here at Jefferies. I have the pleasure of hosting Richard from Alkermes. Rich, why don't I hand it off to you for some intro remarks, and we'll get started.
Good morning or afternoon.
Yeah, I know. It's good.
Day three of your conference afternoon is probably a good time for a split dose of an orexin agonist.
There we go. PRN.
We are, we're thrilled to be here, Akash, and particularly with you. You've been one of the first people to identify the pharmacology and the potential of the orexin class. It's just nice to see it gelling. A lot of the original hypotheses are becoming backed up now with actual data. I think the most significant amount of incremental data that has just happened is the large data set in NT2 that we just shared with you all, which seems like a month ago, but I think it was last week that we did that. There's a lot to talk about there because I think a lot of the original interpretations of it at the top line were people didn't really understand. There's so much new information that's been gained from running proper clinical research in large patient populations over extended periods of time.
There are new observations in those data sets because recall, no one's ever run an experiment like this before. No one's ever run an NT2 study alone, particularly with a range of orexin doses that span a whole range of therapeutic doses, all of which were deemed to be therapeutic at the outset. To get a sense of dose response, tolerability, variability, the time course of the pharmacology, the safety profile. There were a lot of myths that I think got slayed in that data set. We can talk about some of those. The two most prominent from our perspective were normal with safety. Remember before that study, people thought at higher doses, by definition, you were going to have intolerable side effects. That just wasn't true. It wasn't the pretest hypothesis, and it didn't turn out to be true in the data set.
We can talk about why that may be. The second was just how patients with orexin tone at baseline, how they adapt to an orexin agonist administered exogenously over a multi-week period of time. How it begins in one place and it ends up in another place, and how that informs dose selection for phase three. That is information that we would not have had had we not run over 90 patients for eight weeks at a time in three dose lanes on a consistent basis.
We feel like now the probability is quite high that Alixorexton will be used in NT1 and NT2, and it will be available to patients at a range of doses, and it will drive significant benefits, not just measured by the MWT, by the way, but by ESS, by cognition, by fatigue, by all these other domains that are just as important, if not more important than what Wall Street tends to focus on, which is the MWT number.
Right.
With that as an intro, I'll let you.
It's a provocative intro, and there's a lot to break on there. Let's actually, let's start with this. I think let's start with the dose response, right? And I think we've actually now seen multiple data points where there's really not that much of a clear dose response when you actually have normal baseline hypercretin. We saw that with the TAK-861 data with 994. We've seen that, I mean, look, even if you look at your SAD data, there wasn't really a big difference on MWT as we went from 12 to 25, right? That's a pretty big dose range difference, and there wasn't "a dose response." You know, when we catch up with your team afterwards, you're actually not necessarily agreeing with that paradigm. Even though MWT didn't have a dose response, I think you actually are confident there is one if we look at patient-level data.
Help us understand that.
The answer to that question reveals a foundational piece of understanding to understand the data set writ large, which is that in NT1s, the dose response that we saw on efficacy was a measurement of central tendency. Everybody with NT1 responds, while there's a lot of variability around each individual patient, generally the profile of the MWT was the same. Whereas in the NT2 patients, there's a distinct, on the MWT, there's a distinct cohort of patients who never get off the X-axis. They don't respond on MWT. Now, they might respond on ESS, by the way, but they just don't respond on MWT. Therefore, the simple example I've been giving is if you take a responder and a non-responder in the NT2 population, a non-responder has a, let's say, a two-minute MWT change, and the responder has a 30-minute MWT change. That's 30 plus 2 is 32.
Divided by 2 is 16. That's your average MWT. Nobody had a 16.
Right.
That's the number, and you use that number to compare two populations arithmetically, statistically, to see whether there's a difference in the groups. It's not the central tendency of each patient. What's so fascinating about this patient population is that I think if I gave you 10, 14, or 18 milligrams of Alixorexton on three consecutive days, my bet is that you would see a difference. Akash would see a difference across the board.
Sleep-deprived, healthy volunteers.
When you look at it in the population, when we're looking in the population, 10, 14, 18, because it's so confounded by all these various responses, the average number, you won't see it. That's why when we say when we look at the patient-level data, we see more evidence of the dose response. We didn't see a corresponding dose response on the AE profile. We did not see, like we saw in NT2s, NT1, remember, four, six, eight, at 8 mg, we saw a distinct shift in the AE profile, albeit mild and transient, all the things we talked about. We didn't see that in the NT2 population.
To be fair, not, I mean, we've actually seen that even with 861. 861 did not have the same efficacy, but they actually did manage to have AEs even at lower doses. It is, you know, that's not a discordant finding to even what we've seen with Takeda. Rich, there's actually something else you're kind of alluding to, which is, you know, you're now exploring split dosing, and I think that's quite important. There does, again, you think about bifurcation of NT2 patients, responders and non-responders. When you think about split dosing or increased exposure, who does that help? Is that really helping the responders or you're like, you know what, they actually look like NT1 patients anyway? Or is it helping the non-responders? I think the nuance in that question is also how we think about wakefulness throughout the day.
I'd love your take on that.
If you could see the individual patient data, which we'll probably show at some point, I would imagine, we brought out that top-line result basically fresh off the press as we got the top-line result, and we're doing a lot more analysis. The efficacy signal that we saw so clearly at 14, 16, even at 10, was driven by the first two time points. It was driven by hour two and hour four. That's not the case in NT1. You really don't have to be a mathematician or a statistician. If you split the dose, if your goal is to pick up the six-hour and eight-hour time point, just split the dose. It's not particularly mysterious.
What becomes mysterious and why you need 93 patients' worth of data is to model the dose that will accommodate, let's say, 90% of the outcomes to get people above a threshold. That's all for competition. This drug can be approved with a P value QD at 18 milligrams, for example. We think we can actually do better and give patients more flexibility by splitting the dose. To answer your question, it will drive up the later time points on the responders. Our hypothesis, it'll pick up some non-responders and make them responders. I can't say that with as much certainty. Certainly, you look at the, take a responder, high two-hour, high four-hour, then it tapers off. Likelihood is that you're going to drive those later time points for them for sure. I'd be surprised if we don't pick up some non-responders.
Just a final nuance to complicate even further. Remember, we have patients who do not respond on MWT who report normal ESS.
Yeah, that's a good point. Okay, as we think about dose selection for phase three, and this was a question we had, you know, post your NT1 data. I think the thing about NT1, you saw this in SAD, and I think you even see this in MAD. There's like almost an eerily tight two-minute, two-minute, two-minute difference that you see with your data. I think you can model that population with more nuance than you can with NT2. The tricky question for you kind of becomes, you have this data internally. In NT1, you kind of said the range that we explored we're comfortable with, where you would kind of bookmark the ends. How are you thinking about just in terms of absolute doses? Do you feel like you've explored the range of doses that you want to look at?
When we think about splitting up dose, one of the things we saw with Takeda was they initially went five to two, then they went two to five right? They actually changed in terms of the actual dose they're giving and when. What are you going to do on PRN dosing to figure out not only the dose, but how do you split it up?
I think that's some of the benefit of making the huge investment in such a big phase two study is that we're going to keep a lot of that private for now because we have the ability to model it now with great precision. The problem that Takeda had, I think, is that they were absolutely limited on dose in investigating in phase two. They couldn't really, they didn't want to go beyond those doses for reasons you're well aware of. Whereas we've not run into a dose-limiting AE in the NT2s. That's a new observation, actually. We tested up to 18 milligrams in this particular study and saw a clear statistical significant result at 18 milligrams. We can go higher on an absolute mass basis.
We may in the split dose arm, but we're also not going to tell everybody what we're doing about that. One of the great observations is that you could have said a priori, we would run into a dose in the NT2s beyond which we couldn't go because of pick your favorite side effect. We didn't. Now with the information we have to model and knowing that we have headspace, we think we can actually come up with the most competitive package. The other nuance is think about it stacked against the NT1 doses. Fast forward, suspend disbelief and believe that these drugs are now approved. It's approved for NT1 and NT2. There's a range of doses that span from one to the other. They happen by fortune, they sit right on top of each other.
In the real world, the physician will be able to start a patient on a dose and then move them on that dose depending on their physiologic set point as well as their own preference. We know from our sites that some patients want us to be able to stay up to go to the theater and stay up till midnight. Other patients want to go to bed at 8:30. That's not the same dose or the same PK. That's one of the virtues of introducing in the commercial sense the idea of splitting the dose, particularly if you have this bandwidth of tolerability that you're quite comfortable within.
Understood. Now, maybe you're just kind of stepping back. It's funny, my team and I, we were looking at Lumryz and some of the sodium oxybates. It's also very hard apples to apples to the sodium oxybate. I mean, you literally, in an act of Congress, it was almost an academic study for this RM data package. But sodium oxybate A takes a pretty long time to actually get clinical efficacy. In NT2, it's a three to four-minute MWT effect. Let's put it this way. The five to ten minutes that you are showing right now, do you believe if you stay within that range, because you're talking about upside, if you are in that five to ten range, is that clinically meaningful based on feedback you're hearing both from physicians and maybe strategics? Is there maybe a disconnect?
Because I think you'll hear sometimes from investors, "Oh, they need XY 20 minutes to be relevant in that population." Is that actually the reality?
Not even close. We have the benefit now of running these data through the participants in the study, the thought leaders in Europe and the US. I think it's being viewed as a landmark study. No one's ever shown. If you compare oxalase or anything else where they have a mixed population of NT1 and NT2 patients, the reported MWT deltas will be a synthesis of those two. Those labels are for narcolepsy with or without cataplexy. I think the apples to apples comparison might be to meld together NT1 and NT2 delta and compare that to what precedes it. Even putting aside the MWT gains, there are ESS, 70% of our patients at 18 milligram dose reported normalcy. They came in at 17.5, which is highly symptomatic. MWT is a tool that's used in clinical trials.
It's not actually used in the clinic. That's also why we introduced cognition and fatigue in the NT1 and NT2 studies, because when you talk to patients, the principal clinical presentation they talk about, NT1 patients talk about excessive daytime sleepiness, cataplexy, brain fog, and fatigue. That's what you hear people talk about. NT2, the same thing, but without the cataplexy. We think in the real world, these deltas on MWT objectively are excellent and clinically meaningful. It's more than that. If you just changed MWT and didn't move ESS or you didn't move, then that would be a Pyrrhic victory. I think we've got the dimensionality of this drug and Takeda's are beginning to reveal themselves. They're more than just keeping you awake in an MWT test.
I know you're saving this for a publication, but I'll always ask. In terms of the secondary endpoints like cognition, fatigue, clinical assessments, obviously you toplined the data, you didn't look at it. Any hints? Let's put it this way. The benefit we see on secondary endpoints in NT1, is that congruent with the way we're seeing it in NT2?
I think it'll turn out to be that way. Numerically, we hadn't run the analysis yet, and it's also relative to the baseline. Hard to say does an NT2 patient, again, given all the heterogeneity, we'll have difference in baselines. What does the average number mean? We are going to process all that data. I would be what we saw from the extension so far that we hinted at or even said explicitly, the ESS, for example, persists into the. That is what we saw in the NT1s as well. I'd be surprised if we don't see movement that is detectable all throughout the dosing period.
Understood. Out of curiosity, are you able to kind of prospectively identify patients who are responders and non-responders?
No.
How does baseline maybe fit into that?
Doesn't. So far, we'll continue to analyze this every which way we can, but there's no indication a priori of trying to figure out. The thing you want to do, of course, is you want to do lumbar punctures on everybody, look at baseline orexin levels, and see whether it correlates. That's a bridge too far right now.
Understood.
On that.
Yeah, go ahead.
It reminds you though, NT1 is a fairly reliable diagnosis in the right hands. It's not perfect, but in the right hands. NT2 is the presence of excessive daytime sleepiness without cataplexy. You get a whole lot of different phenotypes in that mix. We expect the same thing in IH, by the way. It is interesting that the pharmacology powers through all that variability, and you can still see the signal so clearly.
Understood. What is your confidence on IH, right? That data is going to be mid-year. There is another question about this. I can't help but think in terms of thinking about phase three, why not you're going to add a split dose arm for your IH study too. You are going to see if potentially you can move efficacy forward. Would Alkermes start a NT2 phase three study without getting at least the IH split dose data?
Absolutely. We can model it quite reliably right now. We do not see much incremental risk. Particularly, the efficacy modeling is pretty straightforward, but you have the tolerability bandwidth now. If we were running into severe AEs at the higher doses, and you had to really think about it, but we are not. So we feel quite comfortable. With that said, as you indicated, we have not made the formal decision yet, but I think we are thinking about feathering in a split dose into the IH protocol that is underway right now. So we just get some operational experience with it. We will light off the phase three before that.
Understood. You just mentioned in NT1 and NT2, we're very confident that we're going to have a commercial product. I don't know, and I feel like we thought, okay, once NT2 works, then it's going to quote unquote de-risk IH. I feel like for IH, we're all, even me, I think we haven't done the work necessary. It seems even more distinct from NT2, perhaps even with more baseline variability. What is your team's internal confidence that you can deliver on MWT for all its limitations, a stat-sig response there? How do you want to frame expectations for investors with that data mid-year?
First of all, I should give the evergreen admonishment that I try to give all the time, which is beware of any CEO or anybody in a company saying that they're confident any study is going to work ever. That's why you do the experiment. There are all kinds of gremlins that come into clinical trials that can cause positive drugs not to. That's why we do the experiment and we look at the data. You try to design your studies with enough rigor so that there's enough redundancy, there's enough power that you're going to see the signal no matter what. That certainly has played out so far in this program. The Vibrance Program is about well-powered studies run pursuant to FDA input in a new therapeutic category.
To answer the question, in IH, our confidence is higher in IH now that we have NT2 data. Because we were able to see signal through the variability. I'll also say we applied a phase three level statistical plan to a phase two study where we had no way to really estimate effect size. Yet it still worked. All that allocation of alpha stuff, you've been through the hierarchical, that was all done based on modeling, not based on any name. Now we have the data in NT2 to be able to model effect sizes and variability. IH will be a different kettle of fish. It'll be equally variable, we believe. There's no pretest hypothesis about the difference of a direct and agonist in an IH population versus an NT2.
I can tell you talking to all the experts, nobody says, in advance, I predict the following. Yeah. And often the diagnosis is overlapping. Interestingly, just in terms of the design, MWT is not the primary endpoint in IH, yet it is in NT2. We will use ESS, and then we'll use the IHSS. Now we're going to measure MWT because we're interested in that variability, see how it compares across the populations. It is different.
Do you have the ability to change your statistical plan and your hierarchy now that you are kind of processing the NT2 data, maybe even increase powering for IH, or at this point the study is kind of set up?
It's a phase two study. Phase two studies are about dose ranging, trying to see. We actually cared less about the multiplicity adjusted statistics. We wanted to see what the data showed. The multiplicity corrections are basically based on the allocation of alpha in advance. As I said, we didn't have any experience with that. Notwithstanding that, we saw a multiplicity adjusted statistically significant signal. When you just look at the data itself, just the nominal P values, you can see how the drug is behaving.
Understood. Maybe moving to Avidel. We got an announcement this morning on your valuation for that asset, and you increased your offer. You know, it's funny, we always speculate. How much does a company know? Why did someone do this? The question that I think we were all thinking when that deal announcement got made, do you buy Avidel if you are confident that your NT2 data is going to work, or do you buy it because you're actually worried, and this is a way to kind of preserve a floor valuation? The question maybe for you is, what is your assessment for the value of Avidel strategically for Alkermes before you had NT2 data? Before you, again, you hadn't answered that scientific question. Now, how does the strategic value change? Maybe what are the limits of that strategic value in your eyes?
I should say where we are now is governed by the Irish takeover panel rules, which means that we can't say anything other than what's in the 2.7 press release and the most recent press release. If I'm not quite as forthcoming as I usually am about these things, it's because of that. We announced that Avidel deal before we had the NT2 data. It was based on a tremendous amount of confidence that we were going to find ourselves moving into the excessive daytime sleepiness market based on our NT1 data. We started the discussions about it before we had the NT1 data, but we had a strong sense in the clinic that NT1 was going to be strong.
The logic of it was getting ourselves into those sleep centers with the top practitioners and their nurses and their pharmacists and their payers for a couple of years prior to launching Alixorexton and shifted our launch curve significantly on Alixorexton, coupled with the DCF value of what we think is an attractive product. In our modeling, we can model what we think is going to happen in the oxybate market, assuming the entrance of orexin agonists. They are going to have an effect. We also think one of the things that is going to happen is the overall market is going to grow for the treatment of narcolepsy with agents like ours in the marketplace. There is a major disparity between the prevalence rate of around 200,000 in the U.S. and the treatment rate, which is about 80,000.
What you've seen in other therapeutic categories is new, well-tolerated disease modifying come into the market, the diagnosis rate goes up, the treatment rate goes up, and that wedge that the oxybates satisfy, that wedge will continue to persist. We model all that as we go into our calculation. That yields then a very quantitative assessment of what the value of it is to us. We won't go outside of that.
Now, when we think about you talked about strategic value, I think from a commercial perspective. Let's talk about it maybe from a data perspective, right? Do you feel like it's worth exploring a phase three study where you combine a once-nightly sodium oxybate along with an orexin? Do you feel like that could actually?
When we call it a phase three, I call it a phase four study. I think the clinicians, and you've done this work, we certainly have too, as you talk to clinicians, you realize that the oxybates serve a different clinical purpose for the patients. It's that consolidation of sleep at night. So far, the orexin data doesn't say that it fundamentally changed that. It doesn't consolidate sleep in the same way. Most patients aren't on oxybates. What we've learned, the ones that are appreciate them very much. I think they'll continue to appreciate them. This question, does the zenith of treating this disease comprise of driving wakefulness during the day with an orexin and for the patients who want to go through being on oxybate, consolidating that sleep at night? I think that's an experiment worth doing.
Understood. When you think about for a payer discussion, right, you have both drugs. I mean, typically, these are both branded drugs. Narcolepsy is absolutely a pretty big disease burden. You think, I might be paying a Ferrari to control my sleepiness. I mean, that is a bit surreal. What flexibility does that give you if you're talking to payers where you can say, hey, I can offer both treatments and I can maybe bundle payments here so that you can really treat some of your more refractory patients? I mean, how much of that strategic value played into your decision to look into Avidel?
It does not go into the valuation per se, but it goes into optionality, I'd say more so. Just to correct the semantics, we are not treating people's sleepiness. We are treating their disease. People with NT1 and NT2 have a serious disease and it affects their life in a profound way. It is not just measured in MWT minutes. It is measured in all kinds of things that ripple through their lives. We will see where Takeda prices the first product. That is an unknown for all of us at this point. There are plenty of orphan diseases that are being well reimbursed for on-label use at very high prices if they really have a profound impact on people's lives. I think there is plenty of room to argue, not for every patient, but for the patients who really derive that benefit from both branded medications.
I absolutely think there's a role for it. You need to generate the data proving that.
Last question. Your team had been looking at this. I think you were one of the only companies that looked at EEG and looking at the impact of hypercretin and orexin biology on the hypothalamus. That actually led to you being able to kind of figure out your NT1 dose without actually running sleep-deprived healthy volunteer studies. You have been kind of fascinated with this relationship since the beginning. You have two backup candidates that are going in orthogonal kind of areas where orexin biology could make sense. You feel a little more confident on IH. Tell me what you have learned for, let's say, ADHD, adjunct Alzheimer's, all these other indications. Do you feel more or less confident that an orexin could have a clinically meaningful effect there?
I think we all have to be more confident with the NT2 data now, showing that in patients with baseline orexin tone, that you can drive wakefulness in these patients. The absolute paucity of orexin is not a precondition to having a response to it. It's the other way around. You can drive it with what you would call super physiologic doses of the orexin. I think our confidence in that, we bet on that a couple of years ago, and I think that bet is paying off. That's why it's good to have these other two in the clinic. Interesting, the quantitative EEG is interesting. When you talk to practitioners in the world, most people focus on sleep EEG rather than wakeful EEG. It's a new area, actually. We think that there's definitely a signal.
We are looking at that in each of the programs now as we go.
Can you remind me when we're going to get human data for both of those programs?
I can't remind you because I never told you. I think that what we're going to do is we'll finish the SAD and MAD. If those finish completely, we'll move into the translational studies in 2026. I think next year should be the time you'll get more information on that.
Understood. Thanks so much. I really do appreciate it. Thanks everyone for joining us here.