Good morning. Welcome back to the next session at the Leerink Global Healthcare Conference. We're lucky enough to have Alto Neuroscience here and Amit Etkin, who is the CEO and founder of the company, is going to give a presentation. Amit, take it away, please.
Great. Thank you, Marc, and thank you for having us here.
Yeah.
This is the disclosure regarding forward-looking statements. Before going into specific data sets, I want to give a broader picture of where we are with respect to our efforts across the board and really contextualize what we're doing here. All of this is around a precision psychiatry focus that is developing drugs along with tests that identify for whom those drugs are best suited. We've already put that into play with a number of programs in the clinic, having dosed already in trials over 800 people across different mechanisms, different populations, different biomarkers, addressing large unmet need across depression, bipolar disorder, and schizophrenia, now with four Phase 2 readouts over the next 2 years, and importantly, cash well beyond that into 2028. That's a lot of activity, and I'm going to try to take that apart for you for individual programs.
To start with a general framing of the way we use biomarkers, thinking of it as two different and complementary approaches. One is at the earliest stage, understanding what it is that a drug does in the brain. So how do we understand target engagement relevant to psychiatry from a biomarker perspective? That's something that's typically not done, or at least not done in like a well-powered, scalable, and interpretable way. Then the other, even more important thing is to be able to use biomarkers for patient selection. That is to understand who responds to a particular agent and why. As you'll see, some of that can even lead to discoveries around mechanisms that the approach to finding the biomarker itself didn't, in fact, require, but comes out of our learnings from doing that.
The key part there is that we've set up our data science around the concept of prospective replication within the context of these trials. With every point that we pass using a biomarker, either early or later stage, there is a stage gate that you need to go through centered around biomarkers that should then decrease risk for the next stage. We've now seen this play out across our pipeline and across our efforts, including on drugs that are out there in the market that we just need to understand how they work, if only as a way to sharpen our tools and contrast with the drugs that we're developing. We've found biomarkers for ALTO-100 and 300, prospectively replicated them. We've found biomarkers for standard of care treatments like SSRI, ketamine, even things like psychotherapy have biomarkers. We've conducted large studies.
We've set up a decentralized clinical trial infrastructure. A lot of the piping, if you will, to being able to run these biomarker studies has now been done. We're now seeing the fruits of that labor. We'll talk about target engagement markers for ALTO-100, but then our efforts with ALTO-100 and 300. This is the pipeline. As you can see, multiple different assets, all in Phase 2 , but different stages of Phase 2. Two of them that I'll spend the most time on today, ALTO-100 and 300, are both in Phase 2b for different populations, bipolar depression for ALTO-100, major depression for ALTO-300, different biomarkers, different mechanisms of drugs. ALTO-203 and 101 are earlier stage where it's really more about a pharmacodynamic and target engagement measure using biomarkers, all of them though reading out by the end of 2026.
Let me start with ALTO-300. To start with, as we think about what the development plan is, you have to start with the fact that this is an approved antidepressant in Europe and Australia. It has known efficacy, and that efficacy has been demonstrated in multiple studies. It's used consistently in the clinic, and people like to use it because of its unique mechanism. That unique mechanism then begets a unique biomarker. The mechanism of the drug here is as one molecule stimulating the melatonin receptors, and that helps resynchronize circadian rhythms and effects on mood, and blocking the 5-HT2C receptor, which regulates the activity of dopamine and norepinephrine. When you block 5-HT2C, you lead to an increase in dopamine and norepinephrine. It's a multimodal mechanism, as you will.
You can see on the right the various things that each component does, but often it's actually synergistic action at both of these receptors, even on something like dopamine. This is a summary of the development plan or of the development that we've taken to date with the drug. I'll focus here really on the biomarker component, because that's really the part that tells you both how we select patients, but then also the why and what it means. On the top, what you can see is the scheme that we followed for our machine learning approach, which is to find in a data agnostic manner biomarkers out of EEG data collected pre-treatment, something we've done over and over in different indications, and from that derive a signal that helps divide the populations into responders and non-responders.
That signal, as it turned out through the course of how we set up the machine learning, identified a single signal, and that was around neural variability. You can see in the top trace where it says positive there, that's biomarker positive. People respond better to the drug. They have greater neural variability in this particular measure called sample entropy, and that greater neural variability is more patient-like. Lower neural variability you see in the negative trace, you respond less to the drug. We know that that prospectively replicated because we had a separate and independent group of patients that we tested the biomarker on, and you can see this in the bottom, and those who are biomarker positive responded better. The question here is not just, can we develop a drug using the biomarker?
That path is relatively well understood through the FDA's enrichment guidelines, and that's exactly what we're following. We'll talk about the Phase 2b program momentarily. There's always this question of why. Now, you don't really need the why for developing the drug. You just need it to help patients and lead to a positive effect. That why is still important for us to understand further. Ultimately, we think clinicians are going to want to understand more of the why behind these tests. If you remember the mechanism of the drug here, around 5-HT2C antagonism decreasing dopamine, dopamine is important for signal to noise in the cortex. When you have low dopamine, you have worse signal to noise, and that would be expected to look more like that positive trace. We tested that hypothesis in animals.
Even with our first effort, which you can see to the right over there, came up with quite striking results, which is that if you stimulate 5-HT2C and you cause an anhedonic depression-like phenotype in animals, you actually see in a dose-dependent manner that you increase this biomarker. In other words, we've been able to discover a biomarker entirely through human data-driven, agnostic machine learning methods, and yet come out with something that is mechanistically directly related to the action of the drug. We're now looking at this biomarker further to understand its mechanisms both in animals and in humans, and that's been quite rewarding as we now look at the clinical program. The design of the Phase 2b is relatively straightforward from a clinical trial perspective. That is a one-to-one randomization, six-week double-blind placebo-controlled period.
Everybody gets the option of an open-label period afterwards. The important part is how we select patients upfront, which is here done blindly with respect to both patients and sites. Trying to manage expectations on the basis of this EEG biomarker. We have patients who are positive for the biomarker. That's our primary efficacy population. That's the population we've powered and the sample size that's driven by that. Then a smaller group of patients who are negative for the biomarker, that will start to tell us about enrichment. You see a bigger effect in the biomarker positive, but it will also start to tell us about how we're managing patient expectations, because a big part of this is not having a patient come into a trial and say, "Well, I'm in the trial because I have the biomarker," which would inflate expectations and with that placebo response.
Our ALTO-100 study suggests that that approach was successful in controlling placebo response, which has been a big problem for the field historically. We'll continue to learn about that as we go through these trials. There's a lot that we're doing here that I'll touch on in a second to learn from our ALTO-100 trial results this fall, but that's kind of the core design we followed. I'll go into the ALTO-100 results in more specific terms momentarily. The context here for the interim analysis that I'll go through that we did recently here comes out of the ALTO-100 Phase 2b program, which followed a similar approach in MDD using a cognitive biomarker to select patients. We failed to hit the primary outcome in that trial, and there were a lot of learnings that came from that.
I'll dive into them in a moment going through the 100 data. We wanted on the back of that finding to immediately strengthen our other trials, because the issue that really we think drove the 100 result was a professional patient site-level conduct issue. That represents a risk, frankly, a risk that we see across the board in our field right now, and we've seen this across study after study. We wanted to make sure we were in a good position to address that as quickly as possible and make sure that the ALTO-300 program is in the right path, having corrected those issues.
What we did is, as quickly as we could, we froze the sample, which is at that point 171 people enrolled in the ALTO-300 program, did a retrospective case review on the participants blinded to outcome to understand, are there site-level execution issues that present risk to the quality of the data in the trial? In doing that, blindly then removed data from four sites where we found evidence of potential site-level misconduct, things like data manipulation, things like coaching patients, the kinds of things that are happening all the time in these trials, and we really have to root them out. We have taken that extremely seriously and further strengthened all of our methods to catch this, root it out prospectively for all of our trials. We then conducted an interim analysis on the population with a biomarker.
Again, that's that powered primary outcome population, and had 3 different outcomes to this interim analysis. Those included stop the trial early for futility, stop the trial early for success, both of which we gathered would be least likely outcomes. The most likely outcome was continue the trial with the option of sample size re-estimation. That's exactly what came out. We have continued the trial supported by these results, suggesting that there's drug-like signal in there. Again, this is a drug also that's approved and on the market in Europe and Australia, so perhaps that shouldn't come as a surprise. It's a good thing to see, especially nowadays in these trials. We also upsized the size of the trial from 150 biomarker positive to 200 biomarker positive to put ourselves in the best position for success at the end of the trial.
That 200 biomarker positive population, I should mention, is the exact same size as the trial for ALTO-100. All very consistent with what we've done before and very excited for where that puts us now in mid-2026 as a readout for this trial, but in a good position having looked at the cases, the patients that have come in and make sure we have the right patients and testing this drug in the right way and going forward into that top-line readout. Other things I should mention along the way here with respect to what we've further done to strengthen execution in this trial, we've instituted a sponsor eligibility review. In other words, sites can't enroll patients without our say-so. We're measuring urine antidepressant levels. Medical and pharmacy records are required. The level of rigor and control over what's going on at the site has been intensified.
On top of that, we're also working on essentially diversifying our clinical trial sites and patient sources away from the common sources of patient risk, like recruitment through social media, that has driven a lot of problems across trials to date. Transitioning now to ALTO-100, which is in a bipolar depression Phase 2b , I want to very briefly go through the history of the molecule and its development and then focus a bit more on the learnings that have come out of the results this past fall. The core hypothesis here, whether you're talking about unipolar or bipolar depression, is that supported by a wealth of work at the cellular and molecular and brain imaging level, that there is a reduction in hippocampal neuroplasticity in depression. By depression here, I'm including both unipolar and bipolar.
That means reductions in signaling with genes such as BDNF, AKT, MEK, all of these downstream effectors of neuroplasticity, reductions in cell numbers or dendritic complexity, and ultimately reduction in the size of the hippocampus and impairments in hippocampus-dependent plasticity functions like learning and memory. You see this across populations, but you also importantly see it in individual people much more than other individual people. In other words, it's a way to characterize a subgroup of depression as those that are deficient or have impairments in hippocampal plasticity. Framed in that way, the mechanism that would then be of direct relevance would be one that would enhance hippocampal plasticity. That's exactly what ALTO-100 is.
ALTO-100 was discovered based on a screen for neurogenesis, so an element of hippocampal plasticity, and was found by us and others later to stimulate release of BDNF, brain-derived neurotrophic factor, which is a core molecule important for learning and memory and mood, and to lead to all these plasticity-related signals that are abnormal in these patients with neuroplasticity impairments in the hippocampus. We found what the direct target of the drug is, which is the G protein-coupled receptor that regulates things like BDNF activity. We then in humans conduct a large Phase 2a study, 239 people, where we found essentially what I just laid out as the hypothesis, which is those people who have evidence of hippocampal neuroplasticity impairments in the form of poor memory, verbal memory in particular here, doing better with respect to clinical change on antidepressant measures.
We showed that in a reproducible and prospectively replicated way. That guided our design of the Phase 2b trial, which looked very much like the ALTO-300 trial I showed you, one-to-one randomization, six-week double-blind period, selecting patients for the verbal memory biomarker, but also including people with good verbal memory, again, to manage placebo expectations and to understand enrichment. Unfortunately, the top-line results for the trial were negative, and that top-line included two different populations. One population, which by design was about 70% of the sample, was a monotherapy population taking the drug alone. The rest of the sample was an adjunctive population taking the drug on top of a failed antidepressant that had continued on at that dose throughout the trial. That was about 30% of the sample. All of that by design in terms of the split.
What we saw when we looked at the data is that the top-line was negative. The monotherapy group was negative, but the adjunctive group showed an evidence of signal here that I'll walk you through and really is what drove the learnings that came out of this. On the left, you can see the curve for the adjunctive population. Even though this is a 30% sample of the whole population, and that's what we'd expected to see and was therefore pre-specified, it came nearly significant with a delta here of 4.2 points on the moderates. That suggests that there's a signal there and doesn't show it definitively. It's not statistically significant. Of course, incredibly frustrating to have a negative top-line readout here, especially after all the work that had gone into setting up this hypothesis.
There is a learning that comes out of this, and that's really the critical part here is what we think happened and then how we've strengthened our studies, part of which I've already outlined to respond to that. One could ask oneself, is there a difference fundamentally in biology between monotherapy and adjunctive patients? Is there something we didn't understand? Turns out there wasn't. If you look at the different populations, if you look at the Phase 2a , it came down to something much more simple, which is non-compliance with drug. That clustered at the site level, so it wasn't just the type of patients or adverse events leading to non-compliance or anything of that sort.
In our Phase 2a , we saw about 90% compliance, and that's measured by PK in either the monotherapy or adjunctive group, and we saw enrichment in both groups for the biomarker. What you see on the right is a plot of the PK positive compliance measures in about a third of the patients who were able to get PK on just due to site availability, and the results are quite striking, which is that all of the patients on adjunctive treatment were compliant, whereas only 56% of the patients in monotherapy were compliant. That helps interpret because you obviously need to have drug in your blood to benefit right from the effects of the drug. That helps interpret the adjunctive result, but the monotherapy result, furthermore, as I was alluding to, isn't just a random here and there non-compliance that's equally distributed across sites.
It was heavily clustered to certain sites, and those certain sites brought in a fair number of patients, but with a very high rate of non-compliance. That's this professional patient concern. That's the site-level conduct concern that we've then taken extremely seriously for all of our programs, including the ALTO-300 program, but also likewise our ALTO-100 bipolar depression program. It's this result that has encouraged us to continue that ALTO-100 depression program, given the signal here that's suggested by the adjunctive result, but also looking at those patients now gated by having drug in their blood, so treatment compliant independent of whether they're adjunctive or monotherapy. You see that there is both a signal of drug versus placebo difference here, and on the left, and on the right, you can see that there is a difference as predicted with respect to stratification.
Those patients who have the biomarker, in fact, respond better, just like we'd seen before. You had to be taking the drug, and that's really the control we're trying to institute across the trial is make sure we have the right patients in. We reduce the chance of site-level misconduct. We have measures of compliance in this case, rather than at the end of the study for the bipolar depression program, we're doing it in small batches throughout so we can monitor individual patients and sites and have literally as much visibility as we can possibly get, bringing in a variety of additional technology tools to help understand exactly the patients in these trials.
This has therefore not only encouraged us to go and continue the program for bipolar depression, which is furthermore supported by Wellcome Trust, but to understand, just given the tremendous unmet medical need, there the only treatments that we have for bipolar depression are antipsychotics, the opportunity afforded by a potential new treatment. Excited for where that puts us, and that will be a 2026 readout. The last two things I'll discuss more briefly are the two proof of concept trials, Phase 2 POC trials that are going on, one in ALTO-100 and one in ALTO-203. Both of them are 2025 events. ALTO-100 is a PDE4 inhibitor. We know about the efficacy of the PDE4 inhibitor class in the immune system, Otezla and the like over many different indications in many years.
They've been of interest for a long time for the brain for cognition, and this goes back decades. There have been two fundamental challenges with PDE4s, especially in the brain. One is not knowing how much to give and against which targets for a variety of reasons that all tie back to the same biomarker motivations that we have in the first place. The second is that the more you give on the hypothesis that more is better, the more you get intolerance due to adverse events, and those adverse events are driven by nausea, vomiting, and diarrhea that are class-wide. We tried to see if we can solve both of those problems as we develop this drug, and it's targeted for cognitive impairment in schizophrenia, an area where there are no treatments.
It is the core basis of schizophrenia, starts early, persists through life, and is the biggest predictor by far of disability. In that framing, we first looked at measures of biomarkers, of EEG biomarkers that are well characterized within schizophrenia and help us understand cognitive impairment. This one in particular here, we have linked in a prospectively replicated manner as to the best way to differentiate cases from controls and the best correlated with that core CIAS pathophysiology, and that's the theta response or low-frequency response to auditory stimuli. Low-level sensory processing measure that's relevant to the excitation inhibition imbalance in these patients. You can see on the left that we saw a dose-dependent improvement in this measure, and this is in a large 40-person phase one crossover study, two different doses of drug and placebo.
You can see in the middle here, processing speed, a measure of cognition, is acutely improved. This is all placebo-adjusted, fairly large effect sizes even with single doses. On the bottom left, the degree of EEG change correlates with degree of cognitive change. It gives you really a target engagement measure to continue to follow, and that sets up our POC study, which looks both at the EEG measure as well as cognition across 10 days of treatment in patients as a crossover design here between drug and placebo. The other important element is how we're delivering the drug, and that here is in the form of a transdermal patch, which you can see to the right. The logic here was that if you slow down absorption, you might actually be able to get around the adverse events, which happen in a dose-related manner at CMAX.
In blue is the immediate release oral, and in pink is a transdermal formulation. As you can see, we can actually triple the amount in the blood by AUC. We reach steady-state levels that are similar to CMAX with the oral, yet we dramatically reduce adverse events. We think we solved the two core challenges behind PDE4s, and if positive, that POC study can then directly go into a longer-term treatment study in schizophrenia that would have a pivotal-like or registration-like design. The final drug here is ALTO-203, which is an H3 histamine H3 inverse agonist. The motivation behind this is that the H3 system regulates a number of other neurotransmitters, dopamine being our greatest interest here, and particularly dopamine within the reward system, which regulates things like motivation, how people feel, and then things like cognition, so motivation driving a lot of other important functions.
What got us particularly excited about this molecule is that in a phase one study that was placebo-controlled and in fact active drug-controlled as well, you can see on the right that there was an immediate reduction in mood and alertness symptoms with several different doses of the drug, similar to or greater than what modafinil does, which is another drug that leads to release of dopamine, especially within the reward system. This design here is, if you will, an even earlier stage, the ALTO-100. It's a single-dose pharmacodynamic design in patients with depression and elevated anhedonia symptoms motivated by the at least theoretical alignment between anhedonia and reward system dysfunction, but really looking at pharmacodynamic readouts, be they on the right, the measure of immediate subjective response, but also things like cognition, measures of motivation, wearables.
This is still an H3 inverse agonist, so we're looking at effects on the sleep-wake cycle and EEG to understand what does the drug do and how to best direct it in the clinic. That is on track to readout first half of this year. Where we are right now is looking at four different readouts over the next two years. Cash puts us well beyond that into 2028. Any of these readouts have the potential to kick off late-stage efficacy programs in the case of ALTO-100 and ALTO-300 to kick off Phase 3 programs. Really, really excited for where that puts us now.
We have a lot of different proof points coming across our platform and our pipeline, and frankly, a lot of opportunities here to move the needle in an area of psychiatry, which has seen very little progress and which even just the past 6 months has brought a lot of headwinds, and the need persists and in fact been underlined by the headwinds that we've seen across the field. I thank you very much and look forward to seeing what the next couple months and years bring. Thank you.
All right. Thank you.