All right, thanks so much. It's my pleasure to be moderating this panel with Amit Etkin, Founder, President, and CEO of Alto Neuroscience. Maybe, Amit, you can give us a quick snapshot of where you are with your various trials that are ongoing and reading out in 2025, and then we can do Q&A.
Sounds great. So we have a number of assets right now in the clinic for different drugs in four different phase 2 trials. We'll talk, I'm sure, about the ALTO-100 MDD readout, which just happened, but have a number of independent, mechanistically independent drugs targeting different subpopulations across a variety of psychiatric disorders. So all with a lens of defining who the right patients are and testing the drugs in those targeted populations, looking at two MDD readouts, one for ALTO-300, one for ALTO-203, first half of next year, ALTO-101, which is looking for cognition and improving cognition in schizophrenia, and the back half of the year, and then ALTO-100 in bipolar depression in 2026. So a lot going on in the clinic, a lot of different interesting biology and mechanisms we're after.
Yep. Okay. Maybe let's start, and you can just kind of go through what your learnings are from the ALTO-100 trial failure. You did some initial analyses, feel like there might be certain groups where there's a signal, but ultimately, what are the implications going forward?
Yeah. And I think that's really the most important part is what are we learning and how does that help us for our next trials. So just to bring everybody up to speed, for those who may not have looked at the details, ALTO-100 is a drug that we've been targeting for patients with poor cognition, the idea being low hippocampal neuroplasticity driving cognitive impairments and low mood in these depressed patients. But we looked at it in two populations of patients in that trial, and that was one population taking the drug adjunctively to a failed antidepressant and the other taking it as a monotherapy.
We reported negative top-line data combining the two, and that the monotherapy group had not succeeded, but recently reported that the signal that we see for the drug, in the magnitude of which of the drug-placebo difference was in the range of what we'd expected, is what you see in the adjunctive group. And it came down, as you said, to why, essentially. And the why for us, we try to be as simple and as much drawn by the ground truth as possible, turned out to be about compliance. When we looked at compliance with the drug in patients based on blood draws, we saw that the adjunctive population was fully compliant, and the effect size, therefore, is something that we trust.
The monotherapy group was far less compliant than we had anticipated, in fact, far less compliant than we'd anticipated based on doing the same PK checks in a prior Phase 2a study where we saw 90% compliance, and if you look at the compliant patients, whether monotherapy or adjunctive, we saw a consistent effect of the drug and a consistent effect of the biomarker in predicting better outcome with the drug.
Yep. Okay. So as it relates to that, what have you then done with your other ongoing studies to make sure this doesn't repeat itself?
So we've immediately pivoted in our focus to exactly that question. How do you take the ongoing studies and further strengthen them and, frankly, even address what level of risk any one study has? So that adjunctive population for us is a really key part of this, and that is people who are clearly real patients who are coming in on a medication that they're already taking daily and are far more likely to be the target population just from a clinical perspective alone. ALTO 300 is an adjunctive study, and the ALTO 100 bipolar study is an adjunctive study. But even for a trial like ALTO -203, which is a monotherapy earlier stage proof of concept trial, there we're giving patients the drug looking at single dose hours-level clinical effects in the clinic, and that is something we're measuring again by confirming by PK.
So at every point, the design is now focused on a more patient, more clearly patient population and, wherever possible, confirming compliance. The ALTO-101 study, likewise, being dosed in the clinic with these patches, and therefore we know that the compliance is in place there. And then around the edges, we're doing little things to strengthen each of these studies, but that was one of the biggest learnings, and I don't think you're going to see us anytime soon do a large-scale efficacy monotherapy study, just too much risk there.
Okay. Okay. So as it relates to the ALTO-300 study, do you want to just maybe take a step back, talk about the biomarker hypothesis there, and then maybe go through the Phase 2b design?
Yeah. So let's start even with what the drug is. ALTO-300 is agomelatine, which is a drug that's been approved in Europe and been in use for 15 years. The attributes that are particularly interesting about it, which then relate directly to the biomarker, are that it acts in a way that's different from other antidepressants. Here, we're developing it as an adjunctive treatment, so compare that to an antipsychotic. It's a very well-tolerated drug, stimulates melatonin receptors and blocks 5-HT2C receptors, which leads to increased dopamine. That pattern leads to efficacy and excellent tolerability, whether you're comparing to placebo or to active drugs.
What we've done then is find a biomarker here with EEG derived out of our machine learning analyses, which we've done and validated and published in various ways over a number of years back to my work at Stanford, and then prospectively replicated that this EEG biomarker predicts treatment response, and we hope soon to say more about how that mechanistically is linked to the way the drug works in the first place. So even though it was discovered in a data-driven way, we actually think that there's good mechanistic rationale there for that biomarker. That biomarker doesn't predict placebo. We've seen that in multiple data sets. It doesn't predict standard of care outcomes. It's specific for that drug. And so the Phase 2b prospectively selects for those patients with the biomarker.
We have a portion of patients who don't have the biomarker as a comparison group that we kind of bring in along the side. All of this is done at around 40 sites across the U.S., but blind to both patients and sites and their clinicians, as well as our own clinical operations staff, so nobody knows what their biomarker status is or why they're in that trial based on the biomarker. Randomize everybody to drug versus placebo, one-to-one randomization for six weeks, and then you have the option of continuing to an open-label treatment afterwards, so similar in that sense to ALTO-100, but a completely independent biomarker.
We also released some information about the implementation of that biomarker recently on our deck, noting that the biomarker itself is derived from a single electrode, so you can measure it quite simply, which makes scalability to the home, to the clinic, something that obviously we've been thinking about, but an important part to translation of what we hope will be a positive result.
Right. Okay. Makes sense. So as it relates to the biomarker itself, if you go back to the Phase 2a data and you compare it for 300 and 100, t
he error bars in the 300 data are a little bit wider. Does that mean that there's a greater signal-to-noise ratio with this biomarker or greater variability? What is that telling us?
It's just that the sample size is smaller.
It's all an artifact of sample size,
so the variability is relatively consistent.
Yeah. Exactly.
Okay. And what other work have you done with this biomarker to kind of disprove the notion that you're selecting for just better responders to anything, including placebo?
Yeah, so we actually look at placebo, right, just like we did with ALTO-100, where, by the way, placebo was very well controlled.
Yeah. I mean, historically.
Yeah. Just, yeah.
Yep. No, gotcha.
Okay.
We looked at placebo data where people had gotten EEG at baseline. We put that online a while ago. That was part of our kind of data release a long, long time ago when we started the Phase 2b. Even since, we've applied it to the 100 data, looked at that placebo arm, and again, found that it didn't predict placebo in that arm. All of that keeps kind of coming back to the same message, which is it's specific for that drug, predicts and prospectively replicates prediction for that drug, and it's not just either prediction of placebo or prediction of another antidepressant. We have 800 patients' data with SSRIs and SNRIs, no prediction by that biomarker.
That for us is really important because it's hard for us to advance a drug and a biomarker where you can just replace it with something else, right?
Sure. Sure. Okay. Makes sense. You know, one of the interesting things about the 300 study versus the 100 study is that agomelatine's already proven to work.
That's right.
What kind of delta in effect size do you think you need to show among the biomarker patients over the non-biomarker patients, assuming that the non-biomarker patients actually might show a real clinically significant response?
Yeah. So we haven't really thought too much about what a threshold there might be in terms of the degree of enrichment, how much bigger your biomarker-positive drug placebo difference is than your biomarker-negative. The study's really set up to test statistical significance in the biomarker-positive. There's only a small number of people who are biomarker-negative coming through. Our general bar is some degree of enrichment. So let's say you have, just to throw out a number, 0.4 effect size in your biomarker-positive. If you see something in the 0.2s, that might be a very meaningful difference if you kind of calculate how many people would have gotten better by virtue of that enrichment. Anything bigger than that, of course, would be great.
Do you have a sense of what the FDA is looking for?
It's kind of a similar bar, from what we understand, which is some difference, basically to show that you are enriching.
Yeah.
But remember, our goal here is not to necessarily have, because biology doesn't work this way, a hard and fast line. Everything we've seen in the biology for ALTO-100 or ALTO-300 has been a continuous relationship between biology and clinical outcome. And so what that means, we think from an FDA perspective, is that the way you get that drug and the biomarker labeled could be a hard and fast line or could be more of a gradation where you get it even labeled for everybody with an enriched response in those people who are biomarker-positive. Either way, for us, it commercially is what we're looking for and clinically allows people to have the ability to make a decision based on a biomarker outcome.
Yep. Makes sense. So when you go back and look through the history of agomelatine, it's obviously used and been used a lot in Europe and approved there. In the US, right, there's two things that come up when you look at the clinical data. One is the liver enzyme elevations at the higher dose, and the other is the nonlinear dose response. How do you kind of make sense of both of those nuances and how they relate back to your program?
Yeah. So I think both are kind of historical oddities more than real issues. So the nonlinear dose response, basically saying at 25 and 50, which are the doses that are approved in Europe, you get similar efficacy. That's pretty much the case for every single psychiatric drug, is that you don't really see much of a dose response. You see more of a threshold effect. The liver enzyme issue is also kind of a historical oddity itself as well. So you actually see liver enzyme elevations in other drugs like antipsychotics at similar or greater rates than the 50 milligram agomelatine dose. And that just hasn't been an issue historically. It's more of how agomelatine was developed back in the day.
But our logic for going for 25 was because of the fact that you get similar efficacy for 25 and 50, and you don't get the transient reversible liver enzyme elevations that on rare occasion, but sometimes you get in the 50, that you can just stick with a 25, eliminate the liver kind of overhang, and maintain the efficacy. But even in a worst case in the 50, you're talking about maybe 3% rate. It's entirely reversible, never led to liver failure in 15 years of use. So in some ways, it's sort of a historical part of the narrative, but not a practically meaningful part of the narrative now.
Okay. Makes sense. Let's see. So timing of that data?
First half of next year, remain on track.
Okay. Remain on track. And from a commercial scalability perspective, what could an EEG-based diagnostic look like in practice?
So we think that there's really two very important ways to think about that EEG. One is as a direct-to-patient tool. So think of a single electrode-based signal and something that a patient can put on themselves guided by software in their home. But the really interesting opportunity as well that clinicians have highlighted for us is what if as a clinician, and I'm a psychiatrist, so I understand this very much firsthand, you can actually get procedure code reimbursement for doing an EEG in the office if it's just that easy to do. Could that be a way to really incentivize clinicians to do more of these tests, prescribe more of the drug accordingly, and that work with their current inability to actually use procedure codes in day-to-day practice? That actually may be quite encouraging for uptake.
Ultimately, using biology, whether you're a patient or a clinician, we think will be quite important for how the field moves forward. Every person is frustrated by the current status of psychiatry clinically. Nobody wants trial and error prescription. Giving them a reason with an easy-to-do test that doesn't require any expertise to either administer or interpret because it's just software-based, that's what we've been wanting as a field for a very long time.
Yep. Yep. Okay. Makes sense. Anything else to add on 300?
I think we covered 300 pretty well.
Okay. Do you want to talk a little bit about 203? Maybe start with the background.
Yeah. So 203 is an H3 histamine H3 inverse agonist. The idea here is that what we're trying to go for is the ability to increase reward system dopamine, which we know is reduced in states like anhedonia, which is part of depression and various other conditions. The H3 receptor serves as a regulator of a number of neurotransmitters, dopamine being one. Interestingly, an H3 inverse agonist called pitolisant given for narcolepsy at higher doses causing wakefulness does not have that effect. We acquired a drug, ALTO-203, which does, but most interestingly, what was seen in a really well-done Phase 1 study by the originator is that in single dose in humans, so acute effect, you see the human analog of increasing reward system dopamine, which is people feel better. So subjective changes within hours.
That is what we're looking for as the outcome in our ongoing MDD with anhedonia POC: to get a single dose of this drug and you see an hours-level change. We'll have exploratory outcomes over weeks of treatment, but the primary outcome here is single dose. So what you think about when you think about that, what that means is an oral, well-tolerated treatment hitting a neural system we understand well and will develop that biomarker kind of support further as that work continues, where you see an acute response. That's a really unique profile, especially with the increased focus we have in the field around anhedonia. Really, again, driven like all of our drugs by understanding the biology, the brain circuitry involved, and then ultimately identifying the right patients based on a biomarker.
Yep. Okay. Makes sense. So mechanistically, has anything similar to this really been tried before?
Yeah. So the signals that have been kind of encouraging in this space is actually around stimulants like modafinil and armodafinil, drugs that act to increase reward system dopamine. In fact, modafinil was used as an active control arm in that Phase 1 study, and ALTO-203 had a similar to or greater subjective effect than modafinil. Modafinil, when given over, let's say, four weeks, leads to an antidepressant response in clinical trials. So all of that mechanistically suggests that going after reward system dopamine, which you can almost think of as a pharmacodynamic outcome on a subjective response acutely, will be something that leads to an antidepressant response in a unique way. Temporal kinetics that are very different. And again, you'd expect modafinil is that kind of timeline to be a guide for how we think of ALTO-203.
Yep. Yep. Okay. And maybe just a little bit more on the design of this proof of concept study and the rationale. I believe the double-blind portion's around 60 patients. Is that right?
Yeah. Crossover.
Crossover. Yeah. So what does that give you in terms of powering to really get a definitive answer here?
Yeah. So your powering is around, depending on how you do the modeling in the 0.3-0.4. And what you're looking for here is essentially, as I was saying, a pharmacodynamic outcome. So the effect of the drug acutely is a change in subjective emotion on a scale that is designed to be sensitive to momentary change. The MADRS is not. People have sort of backfitted the MADRS to try to do that for rapid-acting treatments. Our view is use a measure, in this case, a visual analog scale. It's been used since the 1970s in a lot of people's trials to index acute changes in positive emotion that correlate with depression symptoms over time. And so the design here is that single dose administration, but each person is their own control. It's a crossover design.
Each person, these are completers of all three conditions, will get two different drug doses and placebo. And then you get about a week to 10 days of washout in between. So we measure the response to each. We get biomarkers with that. And then everybody goes into a daily dosing for four weeks, mainly to build the PK and safety database to support what would then be an adjunctive MDD Phase 2b trial, but gives us an early view of what the clinical outcome would look like. Not powered in that part. It's powered just in the primary single dose outcome. But you can imagine, right, if that's positive, how exciting would it be to have an oral, well-tolerated drug with a rapid onset of action.
Yep. Okay. Great. So timing there of that readout?
Also, first half and on track.
Also, first half.
Okay. Great. You want to round it out and talk about the PDE4?
Yeah. So PDE4, ALTO-101, is a drug that taps into a mechanism that, of course, we know works in the immune system. Otezla and the like are PDE4 inhibitors, but has been long of interest in the brain. The idea is there you increase cyclic AMP that enhances plasticity and cognition. We're targeting it for cognitive impairment associated with schizophrenia. That's where the proof of concept trial is right now that is on track to read out end of first half and sorry, end of in the second half of 2025. The important part here is unlike in depression, where you're asking people how they feel, and that's the core definition of the outcome, here we can actually measure it directly.
We can measure it through EEG changes that we've directly linked to the pathophysiology of the disease and to cognition, which we saw move even with single doses of the drug in healthy people in a Phase 1 study before. And moreover, what we've done here is formulate this drug as a transdermal patch, which slows down absorption and in so doing largely eliminates the PDE4 class-wide adverse events that we know from other drugs like nausea, vomiting, and diarrhea. So the patch has been very well tolerated and delivering the amount of drug that we think we need into the brain based on the biology that we're able to track with EEG and cognition.
Yep. Okay. And who are the specific patients you want to target with CIAS? Are these people who are actively psychotic, stable, and still having residual cognitive symptoms? There's that whole pseudo-specificity question. How are you thinking about that?
Yeah. So pseudo-specificity is less of an issue with schizophrenia. It's been kind of a thing people have talked with more in depression. But the important thing is you can't confound your cognitive change with psychosis. So all of these CIAS or cognitive impairment trials are in people with stable and well-controlled positive symptoms. So they're already on an antipsychotic. And then you add this drug on top, and you're looking for both cognitive change and functional change, their ability to actually engage in relationships and work and all of these kind of basic things that are impaired in schizophrenia.
Yep. Okay. Okay. So, maybe take a step back and talk about your engagement with the FDA that you had that you talked about your analyst day before the ALTO-100 data because broadly, a lot of what you've sort of aligned on, I think, just applies to the whole approach.
The whole thing.
Yeah.
Yeah. So I alluded to it a little bit earlier, which is what the nature of the relationship between the biomarker to define people, which is continuous, and the difference in clinical outcome looks like, essentially guides the regulatory path. So they understand in general terms that we're using the enrichment guidelines, which they published in 2019, to guide our work and that we're within our trials, including patients positive and negative for the biomarker, that we have a reason biologically and in terms of the data for any one program to select a biomarker, which for us involves prospective replication, that that's the right biomarker, that it stratifies populations.
And so if you put all those pieces together, the translation of that to what a phase 3 study would look like and ultimately what the label looks like, I'm never going to say is straightforward, but it's something you can understand and get your arms around because you have the pieces to define the patients, understand the nature of the relationship, and understand from there where your inclusion/exclusion criteria should be. And those then tend to map to the label itself.
Yep. Makes sense.
But obviously, data rules, right? We have to let these trials read out. And as they read out, we'll know a lot more in terms of what any particular program would look like.
Right. Any questions from the audience? Nick, do you have any questions for Amit?
Always.
Yeah. I mean, I think, look, I think the only other question is just I'm sure you've heard it from investors today. Why doesn't the ALTO-100 failure reduce the probability of success for everything else?
Yeah.
And I know you guys feel like it's the trial that failed the drug, but could it just be that the mechanism doesn't work the way you thought? How do you kind of make sense of that?
Yeah. Look, we obviously did not show that the drug works in the trial, right?
Right.
Let's be very clear about that. We're trying to understand what the data show us, and we think that that's mainly a compliance issue. But one of the things that we did see, which we mentioned in our earnings recently, is that when you do look at the patients who are taking the drug, so people with PK confirmed drug in their blood were assigned to drug, and you look to see whether the biomarker replicates. That is, do we see that patients with poor memory respond better than patients with good memory? We do, in fact, see that. So actually, the biomarker thesis not only holds, but in the results that we saw, we become more confident in the ability to find a population and prospectively replicate that they respond better. That is the core to the approach we're taking throughout.
It's incredibly frustrating for us and for everybody else that we saw the outcome we saw. The compliance was a factor. That particular patient profiles that lead to that were a factor. One we're eager not to see repeat. At least the biomarker approach has been supported. Importantly, each of these drugs are unique in terms of their mechanism and target population. We tried to be explicit in taking out correlated sources of risk in biotech. It's not the same molecular target. It's not the same chemistry platform. It's a data science approach that's put it together. That data science approach has seen further support.
Yep. Yep. Okay. Makes sense. All right. Look forward to the next events. Thank you. We appreciate it.
Yeah. As are we. My pleasure.
Thanks.