Senior biotech analyst at Jefferies. Thanks for joining us today. Next up is Alto Neuroscience. To the left of me is Amit Etkin, CEO. Welcome, Amit.
Glad to be here.
Maybe spend a couple of minutes talking about the Alto story, what your angle is, why you're differentiated from other similar companies out there, and milestones over the next 6-12 months.
Yep. So what we're after is precision psychiatry, that is, precision medicine for psychiatric disorders, taking principles that we've seen work elsewhere in oncology, immunology, and so forth, in identifying what the right populations are for different interventions. So imagine, really, for the first time, being able to take a test, an objective test that tells you something about brain function, to either rule you in for a drug or rule you out. Right now, as Andrew's alluding to, that's very different from what we have available in care. I'm a psychiatrist, and forever, we've basically just rolled the die.
So what we've done through work initially in my lab at Stanford, where I was on faculty for a decade prior to starting Alto in 2019, is developed an approach through brain biomarkers, ways to assess brain function that are scalable and meaningful, things like behavioral tests of cognition, EEGs, that tell us something about a person, and through that, understand how to develop drugs with a variety of different mechanisms targeting disorders like depression and schizophrenia. All of them aim for a larger effect and therefore greater probability of success by understanding who the populations are. We critically do that by prospectively replicating every biomarker that we find with its ability to drive enriched and better outcomes prior to starting large-scale efficacy studies.
We're now in two of those studies for ALTO-100 and ALTO-300 as phase IIb's, and two other programs in the clinic in phase II proof-of-concept trials, again, different populations, different disorders. So it's really meant to be a broad portfolio with a generalizable and yet readily commercially scalable approach to finding drugs. So I think it's an exciting time now to be doing precision. I think there's really all the roads point to that being the path forward in neuroscience. But there's a lot of pages out of a lot of books to take out from oncology, immunology, metabolic disorders, and so forth that give us an idea of what the road forward looks like.
And then timing of those data readouts? You have four programs underway.
That's right. So first one of those will be ALTO-100, and it's phase IIb in depression with patients characterized by poor cognition. That's second half of this year, and that is tracking well. Then first half of next year will be the ALTO-300, phase IIb, depression characterized by a particular EEG signature, as well as ALTO-203, which is depression with high anhedonic symptoms. That's a phase II proof of concept trial. Finally, ALTO-101 in a phase II proof of concept trial in cognitive impairment associated with schizophrenia as well in 2025.
Got it. And who else, as we scour the landscape, who else is doing biomarkers for CNS and psychiatry? I know J&J just succeeded in their phase III depression study for their seltorexant, looking at a subset of patients who have insomnia. Is their approach different from yours, technically speaking?
Yeah. So their approach focuses on symptoms. In fact, that program was brought into phase III by our Chief Medical Officer when he was there, Adam Savitz, who came to us in order to do things differently, which is really through objective biomarkers, things that, unlike symptoms, are more scalable across disorders, across mechanisms of action. But conceptually, there are similarities, and that's an important element to draw on in terms of understanding how one thinks about these kinds of designs from a regulatory perspective, from a development perspective. There is an enrichment guideline framework that FDA published in 2019 that they followed, and so did we, that guides how you identify patients and design the studies to demonstrate efficacy in those targeted populations. And their success, which is really wonderful to see, I think, is a first step towards an ultimately much broader application of precision.
There's not a lot of people otherwise applying objective biomarkers, but we do see now a lot more messaging around precision and precision neuroscience, precision psychiatry. I think there's recognition that that is the path forward. Fortunately, we are hopefully doing the work to make that happen.
Right. Right. And so pioneering a new frontier or space can be like a double-edged sword in a sense. On one hand, it's a novel, but on the other hand, you might not have FDA aligned, maybe. But just curious, how much is the FDA aligned in this realm of space?
Yeah. So like I was saying, we're building on a fairly well-understood path in other areas and an enrichment framework that's already been well-defined by FDA and folks like J&J already having done work with FDA. So there's a lot that we have, as a field and us in our interactions, done. We'll continue, obviously, to interact with FDA. But this is a stepwise process, right? Get them to understand the approach, the nature of the biomarker, the validation of the biomarker, and then ultimately, it has to be the phase IIb results that will guide what a phase III program looks like.
Right. You mentioned the biomarkers can be scalable. So similar question, you're disrupting this treatment paradigm in the commercial side. Why would a doctor use this approach? Why would a payer reimburse?
Yeah. So in many ways, the question is, why would they not when it's available, right? So imagine you get this test. It's a 15-minute, for ALTO-100, 15-minute web-based self-administered cognitive test. You can do it on the web attached to a direct-to-consumer ad so the patient can find out and drive the prescription, or you can do it in the traditional education of doctors, and they can ask the patient to do this test. If you're positive for this test, why would you take any other drug but the one that you're predicted to respond better to, right? The mindset actually shifts at that point. Why would you accept a generic that's going to take a while to maybe respond just because it's cheaper? I understand, obviously, insurance might want to manage costs.
But even there, the proposition is a value-based care proposition that we'd expect reimbursement likely goes with positivity and maybe at a premium and not for the people who aren't positive. So in some ways, it actually better aligns incentives. And because the test is cheap, it's just a web-based test, it's not like we have to worry about reimbursement for the test per se because that part doesn't really cost much to implement. It's really part of your patient identification costs.
And so the alignment of interests in faster outcomes to response on a drug that people, by the way, can take for a long time because it's well tolerated, doctors feeling like they actually have information and "know what they're doing" in the clinic, whereas historically, we've just basically been telling patients stories about how we select one drug or the other or really based on side effects alone, patients having faith in the prescriptions they're getting. 50% of patients with depression don't even get treatment. All of those things get aligned along with the interests of payers.
Makes sense. How have you exactly chosen these biomarkers? What data or what work went into choosing each of these for these respective compounds? They're all different.
They're all different. Yeah.
How do you know you've chosen the correct and best one for each?
So each one of them has their own different story. I won't go into a lot of detail in each. But to start at the top level, so with ALTO-100, this is a drug where we'd already selected it based on the idea that patients with poor cognition have a reduction in neuroplasticity within the hippocampus, a brain structure important for mood and cognition. And so enhancing that area of dysfunction as identifiable with a cognitive test would be a novel way to intervene in those folks' cognition. And that then proved out as we ran our own phase IIa and prospectively replicated a memory test-based biomarker that predicted response.
ALTO-300, we used EEG, and we've done a lot of work around machine learning understanding of how to mine that kind of data for a scalable and predictive signal that's also highly test-retest reliable, so stable over time. And that built on a lot of years of work just with that methodology alone. So each one of them is distinct. But in each case, we prospectively replicate that we can identify the right people. We also make sure that those people don't respond better nonspecifically. So to placebo is one example. We have placebo data with all of our biomarkers. Standard of care treatments, another example. We want to make sure that you can't just replace it with Zoloft or what have you and that these are targeting different populations.
ALTO-100 and ALTO-300 are two different biomarkers together covering about three-quarters of the depressed population with something that should be better than standard of care.
I see. And maybe walk us through the data you've seen for ALTO-100. I think we'll spend most of the time here today because you have some big data coming up in the second half of 2024. And so walk us through the data from the prior, from the predecessor company, and then what you've done and what you exactly saw, and how does that efficacy compare to existing treatments out there? Why is it better?
So the rationale for this drug initially and its identification is a drug that enhances neurogenesis, the proliferation differentiation of neurons, which takes a few weeks to happen. This was then found on a test of neurogenesis in vitro through a functional assay found to enhance synaptic plasticity in a much faster minutes to hours timescale. So as a drug, it was really well tuned for that profile. Problem is that the originator, Neuralstem, that had done really lovely work in developing the drug didn't have the experience in running depression studies and run it in an all-comer study. And that didn't succeed on the primary outcome for which it was underpowered.
But even in starting to work with a drug, we looked at their data for cognition at baseline as a way to define patients and already found results consistent with our hypothesis that this drug, in a dose-related manner, improved depression in those patients with poor cognition. We then designed and ran our prospective study. The reason for doing that is that we didn't want to just go on the back of retrospective data. We wanted to prospectively replicate and make sure that before we put the time and money into a definitive efficacy study, whose readout is coming now second half of this year, we really had nailed this question of what the biomarker is and what enrichment means.
So we divided our phase IIa sample into two components, one to lock down what that biomarker is, ultimately settling on memory, and the second to prospectively replicate in a blinded way that, in fact, we do get a better response. And the magnitude of that better response in the patients with this poor cognition profile is around a Cohen's d of 0.6 or roughly 6 points delta on the MADRS, which is, of course, the approval type endpoint used in these studies. That's roughly double the all-comer drug placebo difference in depression. As I mentioned, we also verified that placebo in two different data sets where memory was also collected was not enriched, in fact, maybe slightly worse in these poor cognition patients. So that gives us an effect size we estimated to be 0.5, maybe a little bit greater in those people who have this biomarker.
But then making sure that we don't just set this up to require perfection as an outcome, but set it up to hit a clinically meaningful target, we set up and ran and powered the phase IIb to hit a Cohen's d of 0.4, which puts it, back to your earlier question, at just above standard of care treatment. So that would be the kind of thing that would make it compelling as a clinical case beyond being a first-in-class medication mechanistically. That's the study started in January. That is a 6-week double-blind treatment period with an open label afterwards, trying to do all the things that we would also consider to be best practices for mitigating placebo response and now on track to read out second half of this year.
So if you zoom out, there's now a huge amount of information known about this drug and the biomarker in the population before starting the phase IIb and then all the learnings on execution being carried through. We know the drug is well tolerated. Over 400 people have been exposed to the drug therapeutically prior to the phase IIb. We found and replicated in multiple different populations the ability to enrich for these poor cognition patients having a substantially better clinical response. They respond less well to standard of care. So that positions this high-need population with a novel drug. And then we run these trials ourselves. We don't use CROs.
We have an internal clinical operations team, which allows us to make sure we have the right sites, are bringing in the right patients who are consistent between the phase IIa and the phase IIb, all run within the United States, and that the design of the phase IIb, in turn, leaves little gap to the phase III. It's designed to look like a phase III as well. So trying to give ourselves every chance of success and ultimately trying to avoid the pitfalls often in biotech, which is where you just take that big swing prematurely before you actually know enough about your drug. That's true really for all of our programs.
Thanks. We're in June now, the guidance at second half 2024. So is it coming earlier or later in the second half?
It's coming. We'll let people know when the enrollment finishes, and then you can expect about 3-4 months afterwards is when the readout will happen. But confident at this point, it will be second half.
Okay. The study is enrolling both biomarker-positive and biomarker-negative patients. The primary endpoint will be looking within the biomarker-positive patients, randomized drug to placebo. So what percentage of the total number, 266 patients.
That's right.
Will be biomarker-positive? Is there a sufficient amount of sample size for us to draw a conclusion within that subpopulation?
Yeah. So the powering, again, is to 0.4. There's a couple of reasons for including both positive and negative patients here. The positive patients, that will be the on-label population, right? So that goes without saying. The negative patients, though, we're looking for some measure of enrichment. The guidelines from FDA don't specify how much or that it even needs to be anything but qualitative, but something to, again, build on our knowledge. But another important element to including biomarker-negative patients is to manage expectations from the patient side. So we keep everybody blinded to do the most rigorous study we possibly can. Even though we run the trial, our staff don't know biomarker status for a patient. Sites and patients don't. And they don't know the ratio, which we've not disclosed, of biomarker positive to negative.
We've not disclosed it precisely for this reason so that people are running the trial as a trial. They don't start to build up expectation of, "I must be in a trial because I have the biomarker. Therefore, I'm going to respond," and you start to inflate placebo rates. We tried to think of all of the pitfalls to doing things differently, again, taking some of Adam's learnings from doing that enrichment work on seltorexant at J&J and put them into practice here. The powering here at 0.4 ensures we have enough in our primary efficacy population out of that total sample.
Got it. Within the biomarker-negative population, what do we want to see? Drug versus placebo? No separation, some separation? What would it mean anyway?
It's hard to know exactly what it means in the sense that it's not a powered population. So we're not powering the contrast, for example, between biomarker positive or negative. And of course, if you're expecting less of an effect and you have fewer of those people, you're certainly not powering a difference from placebo, but looking for some measure of enrichment. Ultimately, the line here is, do we hit the primary outcome in the powered primary efficacy population?
Right. And I know it's powered to show 0.4. Let's just say the standard deviation was similar to the other traditional studies. What would that equate to on the MADRS?
Usually, more or less, standard deviation is about 10. Standard deviation, by the way, relates to how tightly you run the study and assess people. A 10, that means a 4-point delta in terms of the magnitude of difference between drug and placebo that it's powered to.
Is it safe to say that within the biomarker-positive patients, if you see 2-3 points in the study, which fundamentally is technically approval if this was a phase III study, you would not move forward with the program because it's not robust enough? Or how do you decide from there?
Yeah. I mean, those will be game-time decisions as we actually understand our data. Obviously, it's hard to prognosticate now. It depends obviously also on the standard deviation and what effect size that means. It matters what else comes along with it. Our focus is really at this point making sure that we put ourselves in the best position to hit that primary outcome.
Makes sense. How are you controlling for placebo? I know you mentioned you have your own internal team, no CRO. Anything else you'd mention?
100%. So we tried to incorporate, even though we studied placebo as an intervention, right? We have to remember placebo is an intervention that merits understanding on its own. We put all the things that we would consider to be best practices into all of our studies. So things like 1:1 randomization, so not playing with expectations around what the likelihood of getting drug versus placebo is, not assessing patients too frequently, having a third party assess symptoms at baseline so you don't get score inflation from sites, having everybody go into an open label extension afterwards, which decreases anxiety about getting placebo in the first place. Actually, that's more like a phase III than a phase II. So trying to do all of those things that best position the study while also watching every site and every patient like a hawk.
So every single clinical and biomarker dataset gets reviewed prior to randomization at Alto, as well as by the site, as well as by the third party and rating vendors. So a lot of eyes on everything, constantly monitoring sites, making sure that sites don't get that any one site doesn't go and recruit that many more people than other sites. So we have representation broadly. And so far, so good. I think actually executing it through an internal team also builds a lot of ownership in what we're doing.
Right. Would you expect the placebo within biomarker-positive to behave differently from a traditional placebo?
There's some evidence that placebo response might actually be reduced in those biomarker-positive patients. We saw suggestions of that in our analyses of placebo data split by this biomarker. But we're actually not counting on that at all. We're really focused on the enrichment of drug response. Anything else is great to have on top.
Yeah, icing on the cake, right? What do we know about the safety profile of ALTO-100? Are you seeing anything on a blinded basis, SAEs?
Yeah. So there's been no issues that have come up in the phase IIb that are a surprise. The drug, as I mentioned, is given to over 400 people prior to the phase IIb. Very well-tolerated drug, similar adverse event profile to placebo, slightly increased rate of mild headaches. It's not been a driver of discontinuation. In fact, discontinuation has been at a very low rate. And in the prior phase II, it was statistically significantly less based on AEs than placebo. We actually see fewer of these mild AEs in the patients positive for the biomarker in the phase IIa, also indicative of efficacy. People who feel better tend to be less bothered by these mild adverse events. But nothing new has come up. And that's part of the rationale for making sure you really understand the drug before going into the phase IIb.
Okay. And as I think out loud, most of these depression drugs approved today are once daily. Yours is technically twice daily. So are there examples of commercially successful twice daily drugs?
Yeah. Well, Wellbutrin when it came out was twice daily. The half-life is such that it could be once daily. But where, again, the principle is change as few things as possible as you ramp to that efficacy study. It started out as twice daily. We continued that. So let's have as few variables changing as we focus on first demonstrating efficacy. And then we can solve problems from there.
If this trial is positive, obviously, it's great. The biomarker approach is valid. I'm really hoping that's the case.
Likewise.
If it fails, what would be your messaging? Why would it not read across to your other three programs?
Yeah. So there's a lot of reasons a drug could fail. Some are idiosyncratic for the drug. Some are related to the study. And if you look at that next trial coming up, the next big phase IIb is ALTO-300. That's a drug, agomelatine, that we know works. It's approved in Europe and Australia, has the same all-comer efficacy as other antidepressants that are out there day to day. That just needs a little bit of a boost through the biomarker. So there's reasons why that would work if ALTO-100 doesn't work. Hopefully, both work or we've learned along the way what makes one of these work or not.
Right. Okay. And so maybe one more question is just summarize to us why you're confident this study can work. I know we walked through a lot of things, but if you can summarize.
Yeah. So it's a very directed, hypothesis-driven approach. We knew the population we were going for. Mechanistically, that told us the drug to find. Then multiple replications, including prospective replications of the biomarker that enriches response, which is specific for the drug in a study format where we know we can pull the right patients out consistently. And then we know through execution that we've been able to carry forward our processes very smoothly from phase IIa to phase IIb. And everything has gone so far very well in execution of the trial. And we're around the corner from the results themselves.
Okay. Do you want to last 20 seconds? Is there anything else you want to highlight? I don't think we have time for the other programs, but.
Yeah. I mean, maybe just very briefly mention ALTO-300. As I mentioned, it's agomelatine, which is a known antidepressant. Here, it's an EEG biomarker. Defines about 50% of the population that's responsive. We see it's against specific versus either placebo or standard of care treatment. That trial is another phase IIb, 200 people. That one is adjunctive treatment and will read out first half of next year.
Right. And then followed by the other two programs in 2025.
That's right.
Rich news flow. Okay. Thank you so much for the discussion. Appreciate it.
My pleasure.
Look forward to the data and.