Good morning, everyone. Thanks so much for joining us. I'm Andrea Newkirk, one of the biotech analysts at Goldman Sachs, and I'm really pleased to be joined by Saurabh Saha, CEO of Centessa. Thanks so much for joining us.
Thank you, Andrea. Great to be here.
Maybe we can start at a high level. Over the last year, you've transitioned fully, or almost fully, to an orexin-focused company. Maybe talk us through your thinking as it relates to that, and why are you so excited about this space?
We're fantastically privileged to be working on this class of drugs, this target, the orexin 2 receptor. It's funny you asked that question. As I was driving in, I was thinking what it must have been like when people saw the first data sets for PD-1. I can't think of a better analogy than that, because there is rarely a moment in your career that you get to work on a target, a class that could be truly transformational for patients. I think you've already seen that with the data sets being released in this class of drugs.
It is also rare to be working in a space where you have a large effect size, you have a drug class that is very tolerable, you have a potential market opportunity that just in the rare hypersomnia space could be north of $15 billion, and that has also the potential to extend well beyond the rare hypersomnias to be able to treat excessive daytime sleepiness and other aspects of neuropsychiatric conditions like mood and cognition as potential upside. Very few general players are in the space and there is a large moat in terms of the IP and the ability to drug this class. At Centessa, we feel fortunate that we are one of those leading in this space. We have an extensive portfolio of orexin agonists as a company.
We've focused on it, much like my earlier statement, as when I think the original PD-1 data came out, people just dropped everything and decided to focus on that. We've taken that same mentality because we believe this is going to be not only transformational, but the ability to have a large repertoire of molecules that can be deployed in many different indications beyond the rare hypersomnias could be very special for our company. For Centessa specifically, in 2025, we've disclosed that we'll have data in NT1, NT2, and IH. We'll be the first company to do so with an extensive data set across all three indications, long duration of testing these drugs in those indications at many dose levels.
This is where we largely differentiate ourselves not only to be best in class from a tolerability standpoint, from an efficacy standpoint, but also the potential to be first in class in NT2 and IH, where the market potential for those indications, largely underserved with concomitant therapies, is so large.
Got it. Maybe let's dig in there on 750. To your point, you will have data coming this year. Maybe just taking a step back, as you think about 750, what underpins that best-in-class conviction that you have?
Yeah, it's a great question. The credit goes to Mario and his team, who have started this program many, many years ago, but from first principles that use structure-based biology, as you well know, to have a proprietary crystal structure to be able to generate the chemical matter that we have. That chemical matter is what largely differentiates us from everyone else in the space. We had the opportunity to see the other molecules play out and be able to optimize our metabolic profile, our properties of the molecule to be potent and selective. That hopefully will and has already translated into excellent data in the most difficult-to-treat population, which is those who have an intact orexin system, like healthy volunteers. We've shown dose-proportional increases all the way up to nearly maxing out the MWT.
The MWT is a test that we feel very, very strongly about and confident about to be a strong predictor of efficacy in this space across all the rare hypersomnia indications. We believe that we have a set of molecules now from 750, which is deep in the clinic, 142 soon to be in the clinic, and 489 that are well-positioned across a number of indications.
Got it. When you think about the healthy volunteer data, maybe just help us understand how translatable that is to a patient population. What gives you the conviction there that what you've seen will truly be replicated or have utility in IH patients?
Yeah, the beauty of having done the healthy volunteer data allowed us, number one, to pick out the doses that we would go into phase two into patients very quickly, in a matter of just three to four months. One milligram is a starting dose for NT1, two milligrams for NT2 and IH. So we started, what we feel, at a really good place to dose escalate and potentially de-escalate if necessary. I think most importantly, and we have not really spoken about this very often, is that that healthy volunteer data allowed us to do exceptional PKPD modeling. This is a class of drugs where you could predict down to the unit of nanogram per milliliter on a blood concentration curve of when a patient will be awake and when they will fall asleep.
To be more precise, you could be higher than this concentration by two units and have an MWT near 40. You can be a few units below this point, and you could be asleep in three minutes. That is the level of precision one can predict how efficacy will occur in patients. That curve can be plotted over the course of the night, for example, where sleep pressure increases in the early hours of the morning in our healthy volunteer studies. You can get a very nice curve of PKPD. Now, why do I mention this?
Because you can take that large data set we've generated across doses, and then you can start to extrapolate from there, as we're seeing our patient data, what precisely is the dose we want to be at in terms of the exposure levels to stay above those wakefulness threshold points in the curve during the day, between the hours of 8:00 A.M. I think people get the notion that it's okay to just stop at 4:00 P.M. I think patients would want to stay awake and have the flexibility to stay awake till 10:00 P.M. at night if they wanted to. That's where that PKPD modeling helps us quite a bit in terms of looking at the MWTs at each of those two-hour time points between 8:00 A.M.
and 4:00 P.M., and even beyond 4:00 P.M., to understand, okay, this is when a patient's going to be able to start falling asleep in the evening, and to ensure, most importantly, that they actually can go to sleep and not have insomnia. The science behind this is just truly a joy to behold.
When you think about MWT, and we'll get into the upcoming data, but when you think about the thresholds that you would like to achieve here, what is that? Is that 20 minutes? Is it 30 minutes? Are you trying to get close to 40?
Yeah. We strongly believe, as I mentioned before, that MWT is an excellent, excellent predictor of how patients will do in terms of staying awake during the day on drug. What do I mean by that? Let's say they take the drug at 8:00 A.M. You measure MWTs at 10:00 A.M., 12:00 P.M., 2:00 P.M., and 4:00 P.M. Now, if you have an MWT, let's say, of 28, that means you may be scoring up to the max of 40 minutes at 10:00, 40 minutes at 12:00 P.M. I probably have to do my math right. 25 minutes at 2:00 P.M., and 12 minutes at 4:00 P.M. Now, if I'm a patient and I'm at 12 minutes at 4:00 P.M., that's not great. The MWT looks good. It's 28, right? It looks great. Everyone's happy.
If you're a patient, you're not so happy because you're falling asleep at 4:00 P.M. Our view is a long game here, short game and long game, is commercially we want to be the drug that everyone takes because they have the flexibility to stay awake between 4:00 and 10:00 P.M. I think that's where the game is won. I think people aren't talking about that, but they should be talking about that. Because yeah, you can get approved, you can get an MWT of 28, say, oh, if you're above 20, you're normal. The difference between 28 and, let's say, somewhere in the 30s is remarkably different because it means that that fourth time point at 4:00 P.M., you are staying awake above 25, above 30. You're in a good place. That also means looking at the curve at 6:00 P.M.
At 8:00 P.M., you're probably still staying awake, but now you're starting to fall asleep. That's exactly the type of PK curve you want. That's what our goal is as a company, to make sure that when this drug hopefully hits the marketplace, patients have that flexibility to be able to stay awake as long as they want. If they want to go to sleep at 6:00 P.M., they can take Dosex. If they want to go to sleep at 10:00 P.M., they can take Dosex plus a little bit more. Now, if you're capping your doses in a phase two or phase three setting and you can't go beyond that, then they don't have that alternative. With our design, hopefully we'll get to talk about it, we can offer that flexibility.
Perfect segue. Talk to us about the phase two design. Pretty innovative here with the cross-trial or the crossover design. How did you come up with this? How much do you have that the FDA looks at this trial design and says, all right, good results, you're good to go for the phase three?
Yeah. So let's take a step back here. The phase two is designed, and obviously we're doing the phase two. The FDA's seen our protocol. We're doing the phase two to inform one important aspect going forward, which is what is the dose range we want to take into registration studies? That is our primary goal, to establish a therapeutic index. As I mentioned before, to ensure that patients have the ultimate flexibility and be able to dose up and down based on whatever ADLs, whatever physiology they have. Our phase two design is very innovative. It's very different from what's been done previously. We work backwards from what is the goal? The goal is to define a dose range in our registration studies. How do we do that in the fastest way possible?
We know, number one, there's three areas: number of patients, duration of dosing, and the ability to have flexible dosing to define that dose range. In terms of number of patients, we've defined the number of patients to be a minimum number, as we've already publicly released for NT1, NT2, and IH. That basically is the minimum number needed to be well-powered to determine whether we have a statistically significant effect size. That gives us confidence as a company that we have enough data to make a call on going forward with that particular dose level. That doesn't imply that that's all the patients that we're going to dose at that dose level. We can dose many more patients. That's just the minimum number.
From a safety and tolerability standpoint, we can certainly augment those numbers if we need to in NT1, NT2, and IH. Second is duration. What we've seen, and it's great to be coming just slightly from behind, is what we've seen is there's a first dose effect and efficacy in those who have a deficiency in orexin signaling, where if you just do a single day of dosing in NT1 or NT2, you could overshoot in terms of thinking, well, I have a great efficacious drug with MWT. This is great. No, you have to discount that by 20%-30%-40% going forward because we know that that kind of asymptotically approaches a steady state after the patient kind of equilibrates to the drug into orexin signaling over the course of two weeks and then hit steady state in four weeks.
Number two, duration on drug, four weeks for us, allows us to assess both in NT1, NT2, as well as in IH, what the steady state effect size will be after four weeks on drugs. That's here. Third, our clinical trial allows for what I think is real-time drug development. This is 2025. This is what we should be doing, frankly, instead of predicting doses with a crystal ball ahead of time, is what is the dose that is going to achieve the best therapeutic, optimal therapeutic index? I don't have a crystal ball. I don't know what that is. We have a good idea from our healthy volunteer studies what that might be. But we're not a priori setting the doses. We are looking at our data set after every cohort, NT1, NT2, IH, after every dose level, after they've been on drug for four weeks.
We're assessing, OK, where are we with MWT? Where are we with ESS? Where are we with safety and tolerability? Based on that, we can say, OK, let's go up half a milligram. Let's go up a milligram. Let's go up two milligrams if we need to. Based on that, we are not only accumulating a good amount of data, but we're also then opening ourselves up to the possibility of benchmarking and pegging against the competitor data sets. As they release data, we'll know, oh, we're there. We're done. Or OK, maybe we need to go a little bit higher and establish a therapeutic index with a larger effect size than everyone else. That's why this phase two design is great. I mean, there was an article recently that Eli Lilly doesn't even do phase twos for oncology anymore.
You probably saw that, which I thought was interesting. I think there is this notion where phase twos have to be hundreds and hundreds and hundreds of patients before you can get to phase three. I just think that is an old school way of thinking about things.
Moving on to your first point about the reduced number of patients that might need to be studied in this trial. In NT1, it's a much more homogeneous patient population. As you think about NT2 or IH, I guess how smart is that trial design then when you think about how heterogeneous that may be? Can you really capture or pull out a signal?
Yeah. Happy you asked that question. You can make it as heterogeneous as you want if you do not have good inclusion/exclusion criteria. We do have good inclusion/exclusion criteria. We have defined at baseline an MWT that homogenizes that patient population, that basically gives us what we think is a fairly low baseline MWT for NT2 and IH. These are lessons learned from what others have seen going forward. We do not want to be enrolling subjects who have baseline MWTs of 30. You only have 10 minutes to play with, which is not going to be very useful since you are capped at 40 on that test. That is number one. Number one is we are basically bringing in a set of patients that we can actually interpret the data with a large effect size to play with and therefore define efficacy well in NT2 and IH.
We strongly believe that these orexin molecules will be very effective in NT2 and IH. That is why we feel very strongly that as a company, we have the potential to be best in class and first in class in both those indications. Those are huge markets. MWT is going to be a strong predictor. I think with MWT, ESS and patient-reported outcomes will follow. It is not like they are decoupled from one. If you have a low MWT, are you going to have a good patient report? Maybe. That is patient-reported. MWT is objective. I think objectively, if people are staying awake, they are going to feel better.
Got it. Maybe to your point on having dosed multiple cohorts and you're able to unblind after each cohort and kind of see the profile that's emerging there, what is your ideal TPP here?
Yeah. Our ideal TPP, number one, working backwards, is if there are a choice of orexin drugs ultimately on the market, will the Centessa drugs be the ones that patients and physicians will want to use? It goes back to, number one, do you have a wide therapeutic index? Number two, do you have the ability to have the flexibility to take a dose that basically works for you on a given day, especially between that 4:00 P.M. and 10:00 P.M. range in the evening? Number three, do you have a tolerable safety profile? That is super important. That generally is the TPP. I mean, if you look at the class in general, drugs are very tolerable. They're efficacious.
I think the devil's in the details, as I've mentioned, on I think it's incumbent for everyone to look carefully at what efficacious really means as these data sets come out between the 4:00 and 10:00 P.M. hours. Most of all, I think if you look at where we are compared to standard of care, I think the class has already established itself as dominant. This is why I go back to the analogy of this is a step change in neuroscience. It's a step change, not just in the rare hypersomnias, but the potential for beyond that.
I guess maybe just following up there, it sounds like, and correct me if I'm wrong on this characterization, but it sounds like the TPP then is maybe in some ways being dictated by what the competitors are going to be able to show.
We have our TPP internally, which is kind of the ideal product profile, which it should be. Our goal at the end of the day is to make sure we have a best-in-class drug. If you're going to be able to say that, then you have to be able to back that up.
Got it. How many cohorts have you dosed?
We haven't disclosed any details on efficacy, safety, or tolerability from our current phase two study. Stay tuned.
OK. Fair enough. Maybe to ensure that you are best in class here, maybe let's talk a little bit about the AE profile that sets 750 apart from your competitors. Where do you see the biggest point of differentiation?
Yeah. The biggest point of differentiation from us, I mean, you've seen our healthy volunteer data. We've been very pleased with the tolerability profile that we've seen while still able to achieve, I think it was what, 38 minutes, I think, on our 5 mg dose. So we feel very good about the tolerability profile in general. I think what will be interesting to see is the tolerability profile of the others. Are they mild? Are they moderate? Are patients dropping out? These are the things that we will want to benchmark against and differentiate against.
Got it. I guess maybe if those AEs are mild and moderate, I guess how concerning are those to a patient, to a physician? Is that going to be a reason enough, to your point, to pick 750 over Takeda's asset, over Alkermes' asset?
Yeah. We certainly hope to be differentiated on tolerability and safety. Of course, that's part of the best-in-class profile. If in general the class has tolerable AEs and no one's dropping out, I think that's great for patients. It's great for physicians. It's great for all of us in the field. It's a big market. I think that's wonderful. I go back again and say, how are you doing between 4:00 P.M. and 10:00 P.M.? How are you going to be able to dose higher if you're capped at a dose and you need to take more of the drug that day? Your physiology is different. Your needs are different that day. Do you have that flexibility to go higher? This is where we aim to differentiate. We have that adaptive design to be able to do that.
Certainly in our next step in registration studies is to be able to perform those that allow for that dosing flexibility.
What is the regulatory path post this phase two?
The standard regulatory path that I think everyone takes, which is you have your end of phase two meeting. You talk to the FDA. You align to the best possible extent on the duration that you're dosing, number of patients, your safety database, and the overall design of the study to be adequately and well-controlled.
Do you think there's scope to be innovative in the registrational trial similar to the phase two that you're running right now? Or do you think you'll have to follow essentially what Takeda?
We would like nothing more. As you know, in our phase one, we did that probably two to three times faster than everyone else. Our phase two is going to be as innovative as you possibly can get in real-time drug development. Our phase three, we're certainly going to do everything in our power to not only leapfrog ahead but establish ourselves, as I mentioned, as being best in class.
Great. If you are best in class, let's talk about what the commercial opportunity could look like. What does that imply for 750 and where it sits in the treatment paradigm?
Yeah. The commercial opportunity is large. It's large for the orexin agonist space. That's why I equate it to the PD-1s and the GLPs. I think you're starting in the equivalent of melanoma and diabetes, and it's the equivalent of the rare hypersomnias. Where this ultimately ends up is anyone's guess. If you're helping improve excessive daytime sleepiness, you're helping improve mood, you're helping improve cognition, you're helping improve executive function, you're talking about untapped potential that I can't sit here and quantitate. Now, it's been honestly a bit of a head-scratcher for me as our teams have gone and done a ton of commercial work with external parties internally. We've done enough to be able to say and confirm, I think, what others have said in this space is that NT1 alone could be a $5 billion-$6 billion-plus opportunity.
When you translate those number of patients from just like the Epi data and then obviously talking to physicians in NT2 and IH, you're looking at north of a $15 billion opportunity. Now, I still can't get my head around that. I'm as conservative as you'd want to come. I certainly hope that is the case. It certainly gives a lot of opportunity for anyone, the limited people who are playing in this space, companies who are playing in this space. I think what we're seeing here is incredible excitement being reflected by physicians and patients clamoring to get on orexin agonists because it is essentially a functional cure for those who have orexin deficiency. It's a life-altering drug for those even who don't in NT2 and IH.
We have an opportunity here, along with the few others in this space, to really offer something special. When you do, the market follows. The dollars follow. It is a double bottom line business. Patients do well. Companies do well.
How does that market split? If you're thinking about NT1 and presumably at least three players will be on the market, is this a one or takes all market? Or is there going to be segmentation across 750, Alkermes' product, and Takeda's product?
Oh, certainly. As I've mentioned, we hope to be aimed to be best in class and first in NT2 and IH. We hope to get a significant amount of market share. You have to go back on precedent just in general in drug development and ultimately commercialization is that if you have one, two, three, four players, and the market size is very large, it is very significant as an opportunity, a dollar opportunity for everyone who plays in this space. That is just the rare hypersomnia space I am referring to. Now, if you look at 142 and 489, and we have not disclosed yet on all the indications we plan to pursue there, there is just so much white space and so much opportunity. We could be the only ones in that part of the indication universe, let's say with 142.
That could be very significant for us. I think when the pie is big, there's plenty. There's plenty of opportunity for everyone here.
When you think about the pie, is that pie expanding with the introduction of orexin agonists? Or should investors look at what is the Oxybate market? That is the market that can then be captured.
Yeah. I mean, that's like what's the market for drugs that are not before there's been a redefining of standard of care and you're pegging to the incumbent market, I don't think that's the right way to extrapolate. The right way to extrapolate is if you had a functional cure for a disease where, in this case, there's deficiency of orexin, who would want to go on that drug? Who would want to prescribe that drug? I think you have your answer there. If you have 50,000 diagnosed and treated NT1 patients who are on modafinil, who are on these stimulants, who are on Oxybate, and a vast majority of them either come off the drugs or are not tolerating them well enough or are just not efficacious for them, then you just look at that 50,000 number.
That is not even the total prevalence of the population, which you can see in our corporate deck what those numbers are. I think top down, looking at what that market looks like and what the opportunity looks like is probably the right way to do it. Now, if you extrapolate based on claims data sets and looking at the branded products today and kind of what commercially they bring in, I think that is just not the right way to approach it. That is a way of approaching it. Even if you do it that way, you are still going to end up at very, very large numbers. Either way you look at it, it is going to be very large.
Maybe talk to us about that work that you've done to understand the prevalence. Because I think the numbers that are in your corporate deck are higher than many have put out before. What work underpins the new numbers?
Lots of work. There's been updated claims data sets. There's been a lot of effort on our part with obviously external help to talk to 90-some physicians who treat about 4% or 5% of the entire population of rare hypersomnias. It's a good way to proxy for what the true prevalence is and who's getting what drug today and what would they want to use as first-line agent, for example. Based on that work is how we've arrived at these numbers where the 50,000, the 120,000, 100,000 numbers that we've come to in NT1, NT2, and IH. Those are significant, but they're now more well-founded, if you will. I think when you look at the older data sets, they didn't take into account that a large portion of this population of patients are on drugs that are generics and they're not well-coded, they're not well-captured.
There's a large swath of patients that are just missing that aren't accounted for. I think only if you look at the Oxybates and some of these branded drugs, you get a sense of, OK, who has narcolepsy in NT1, NT2, and IH. If you really look carefully, there are quite a few patients who are trying to get by on drugs that are generic or not effective.
Got it.
Or just do not want to be on those drugs.
Yeah. Yeah. Makes sense. Maybe in the last five minutes, we can touch on 142 and 489. Where do you see these assets truly having utility? What types of indications? Are you looking at other rare neurodegenerative indications or more the larger opportunities?
Yeah. I mean, this is a real fun place to be. I mean, we've already shown that ORX750 works well in healthy volunteers, which is actually a really good proxy for some of these other indications where the orexin system is intact. Certainly, there are indications where there's already orexin loss, let's say Parkinson's, for example, where there's just lesions in the brain which have reduced orexin signaling. There, I think certainly there might be an opportunity. There's opportunity across the board in other neurodegenerative diseases and certainly neuropsych diseases where mood, cognition, attention are important to help these patients feel better but also stay awake, where EDS is a component. If you look at some of these endpoints on these scales, being able to stay awake and not feel sleepy or fatigued is a major component.
For us, it's a really exciting time to be able to have two molecules that we think are well-credentialed, like 750, for example. We have the ability now to kind of play with them now and see where they could be helpful for patients. I mean, it's fantastic to be able to get up in the morning every day and just dream and dream big on these.
How do 142 and 489 differ from 750? Maybe speak to us a little bit about the preclinical data that you've generated for both of these next-gen assets.
Yeah. These next-gen assets come from the same structure-based drug design that we've done because why fix something that's not broken with 750? There have been no liabilities of 750 that we've identified to date preclinically. Our GLP tox studies were clean on 750 across the board. With 142 and 489, the molecules are even more selective and even more potent. That portends to potentially even lower doses, which is important from a DDI perspective because some of these disorders that we're thinking about, there's going to be con meds associated with therapy to treat the underlying cause or more the symptoms, I should say, of some of these other diseases. We want to make sure that we minimize any type of drug-drug interactions that occur. Lower dosing is always a good thing.
These molecules are, as we've developed 750, as we've gone through the preclinical studies of 750, we've learned so much that's helped with the selection of 142 and 489. When we think about these larger populations to go into where they're going to be taking more different drugs on top of 142 and 489 potentially, we believe we're well-positioned to be hopefully effective there.
Got it. I guess maybe characterize the extent of the preclinical data that you've done here. Is it related to MWT? I think you've had some data for 142 in non-human primates. One, talk about the translatability. I mean, I think clearly from 750, we should feel pretty comfortable that there is translation there. Do you feel, I guess maybe have you done any work to look at cognition, mood, symptoms related to some of these potential indications you might go into?
Yes, exactly. With 142, we've shown some SINO data, which has been very exciting. Obviously, you want your drug to enter the brain and work. We've shown that pretty effectively. We've run it through the same battery of tests that we've run 750 through. We also have and are constantly exploring both molecules as much as we can preclinically in these other indications that could help with cognition, mood, and executive function. That is helping us gain even further confidence that we feel 142 and 489 will be well-positioned in those indications.
Great. Maybe last question here, just on your cash runway. What does that help support?
Yeah. Cash runway into mid-2027, that basically gets us through a number of catalysts, a number of clinical milestones. Data is coming. A lot of data is coming, obviously, this year for 750 with NT1, NT2, and IH. Again, we'll be the first company to have substantial data sets in all three indications in 2025. With 142, we'll have our healthy volunteer sleep deprivation data like we did with 750 this year. Obviously, we hope to be taking our third molecule into the clinic soon. The cash runway into mid-2027 gets us pretty far along with 750. It gets us pretty far along with 142 as well in 489. We feel very strong when it comes to our balance sheet and what it can do for us in terms of hitting points of inflection.
Maybe one final question just on your data disclosure, which I know you're not saying very much. Can we expect that data from all three indications will come together this year? Should we expect them to be phased out?
We have the option to do all three at the same time or subsets of those at the same time.
Is there a preference?
Right now, we haven't decided.
OK. Fair enough. We'll stay tuned. Thank you so much for.
Thank you, Andrew.
Thanks so much.