Hi everyone, thanks for joining us this afternoon for the last stretch of the TD Cowen novel mechanisms in neuropsychiatry and epilepsy conference. This is the MindMed Fireside Chat. I am TD Cowen Senior Analyst Ritu Bharal. I'm joined by my colleague Athena Chin, who works with me on our team Bharal neuropsych names, and the CEO of MindMed, Rob Barrow. Rob, thank you for joining us today. Really looking forward to going through your story and next steps. Obviously, the talk of the last year around MindMed has been around MM120 and your phase 2 data that was recently published in JAMA, but top line last year. Maybe we could start with the mechanism of that drug, which is LSD. That drug in generalized anxiety disorder, which is the first indication that you're going after. Why GAD over TRD? Is that mechanism related? Then we're going to hit the market. Yeah, first of all, thanks so much for having me and having Mind Medicine (MindMed) Inc. as a part of this. We're excited to be here. It's MM120. LSD was really the, is the most studied and the most prominent historically, most prominent psychedelic. I think everyone knows it by name. It was part of the reason why it really didn't take center stage when the resurgence in research started about a decade ago. Mechanistically, it is a serotonin 2A agonist. We know that it has similar activity as some of the other drugs in the category on the default mode network, on really probably working at a different level than just symptom suppression. When we talk about indication selection and generalized anxiety disorder (GAD) versus major depressive disorder (MDD) or treatment-resistant depression (TRD), we don't think it's selective for one indication over the other. Mechanistically, what seems to be happening with the psychological processing, which again is described by the network effects and turning off of the default mode network, it sort of depends on whether you use a lexicon of neuroscience or psychiatry or psychotherapy, but it seems to sort of reorient psychologically patients' relationship to the inputs from the external world, from internal narratives that are driving these sort of what have historically been called neurotic illnesses such as anxiety and depression. Mechanistically, there's sort of broad activity, of course, at a receptor level, again, 2A agonists, that drives these network effects and dramatically changes both internal and external perception for a period of time, with a resulting opening of the clinical learning period, a sort of reorientation of patient psychological processing that seems to drive that longer term durable clinical effect. Mm-hmm. As we think about the GAD market in the U.S., can you walk us through what the treatment paradigm is right now, what the unmet need remains? When we think of anxiety, and back to the first question you asked too, anxiety is an area that has been so severely overlooked. The last approval in GAD was Cymbalta in 2007. It was a label expansion beyond initial approval in depression. Historically, the focus of psychiatry has been on anxiety symptoms. Really, it was only the introduction of SRIs that shifted the focus over towards MDD and depressive symptoms. Over that period of time, since 2007, the most recent estimate we have for the prevalence of GAD in the U.S. is 10% of the U.S. adult population. That would be 26 million people. A lot of folks with GAD, even those who are diagnosed and actively treated, stop treatment. We have really two classes of drugs that are predominantly used, SRIs and benzodiazepines, which are no longer used chronically because of abuse and safety concerns. Mm-hmm. Right? We're left with less than a handful of approved SSRIs in generalized anxiety disorder that the treatment. What was the effect size for the SSRIs in GAD? Typical effect sizes are less than 0.4. We're talking about Cymbalta 0.36, 0.38 for both benzodiazepines and for SSRIs. These are sort of mild treatment effects. They're not particularly effective, and they have a pretty high AE burden, sexual dysfunction, weight gain, and just an overall, general sort of suppression of feeling, which is of course the goal of SSRIs to sort of blunt the bad feelings, whether it be anxious feelings or depressive feelings, and as a result try to improve the relative positioning. For really available therapies today, we just don't see a sort of long-term change in trajectory for folks who are living with GAD or MDD. Got it. Let's go through your data, the phase 2b GAD study. Can you talk to the effects I've seen, but also the rationale for selecting that 100 microgram dose? Yeah, so our phase 2b study was designed really to be a phase 3 study, but no one had looked at, and still to this day, really haven't seen any other studies looking at comprehensive exploration of dose response. We wanted to establish on clinical outcomes across a full kind of dynamic range of doses, what the optimal dose is on a set of clinical symptoms, not just on the magnitude of perceptual effects. That's why we designed a five-arm study, including placebo and four active arms of the drug. The primary analysis in that study, and this is reported in the JAMA paper, was a dose response methodology. We pre-specified dose response curves and then statistically fitted to those curves to both establish whether there is a dose response and to come away with a recommended dose to take forward. That's how we ultimately arrived at the 100 microgram dose. Of course, it helps when the observed data also very clearly suggested the 100 microgram dose was the best one to take forward because we maxed out clinical efficacy. We didn't see any increase in efficacy as we increased the dose beyond 100. We did see a higher AE burden, a longer time in the clinic on the day of dosing, so really sort of optimally balancing both the efficacy and the safety and tolerability. Rob, as I have discussions with clients about Mind Medicine (MindMed) Inc., and in fact in our KOL dinner last night, the focus around lysergic acid diethylamide (LSD) was essentially around that treatment burden, and the AE profile and the treatment window. Before we get into some of the other efficacy analysis aspects, the great remission rates, response, durability, et cetera, can you talk to that side effect profile seen, the most common AEs, time to resolution, et cetera? Absolutely. When we think of the AE incidents that happened, not all of them, of course, but the vast majority were limited to the day of dosing with any sort of incidents. These were the characteristic effects of taking LSD or a psychedelic: illusion, perceptual alterations, whatever specific form they took and however they were specifically coded, effectively a change in perception. This comes up as illusions, hallucinations, pseudohallucinations, things of this nature. In phase 2, we're using a capsule formulation. There was a pretty heavy GAD burden. Pseudohallucination? When we think about the category of things that sort of perceptual alteration, there's really sort of three different levels. These are talked about as hallucinogens and hallucinations. It's not really an appropriate characterization of what happens when someone takes a psychedelic. A hallucination means that there's a de novo synthesis of something without a sensory input. There is no radio in the room, but someone believes there is a voice speaking and they believe it to be real. On the other end of that spectrum is an illusion, which is looking at a picture and it, you know, taking a different form or some way, looking at a wall and the wall becoming a fractal or starting to spin or spiral or something. That would be an illusion. A pseudohallucination is sort of in the middle where there isn't a sensory input, so there is some, but a patient is able to determine that it isn't, in fact, real. Really what we see is more illusions than hallucinations. It kind of goes to a need for a high degree of focus on how those AEs are coded and observed and recorded. Exactly. It seems like the sort of thing that FDA would scrutinize closely because of the potential safety ramifications, which are certainly lower for illusions and higher for de novo hallucinations. Typically, we see all these fold together. When you look at, for instance, Cymbalta's label, all of those adverse events are captured as dissociation. Yeah, yeah. Whether they be illusion, hallucination, or actually coded as dissociation, when you have a lot of overlapping adverse events, it typically gets pulled when it comes down to labeling. Got it. Got it. Any severe AEs of note? What was the sort of patient experience and time course over the trial? Patient experience was remarkably benign. I think, you know, everyone has heard this sort of out in the real world stories of illicit use of psychedelics and these, you know, difficult experiences, these, you know, bad outcomes in some patients. Certainly those things happen out in the world with illicit drug use of any class. We were prepared for, you know, extensive interventions and for having folks with difficult experiences. We really didn't see much of that, particularly at the 100 microgram dose. I think as we get to higher doses, that experience can become more disorienting. When we look at our go forward dose of 100 micrograms, yeah, a pretty reliable, predictable experience. There's certainly some, these are anxious patients to begin with. There's some sort of transient anxiety, just kind of as part of a clinical trial, anticipatory anxiety around what's going to happen when I take the drug when I'm in this room. All of that mild, moderate, and transient. Really it was a, again, a fairly benign profile. Definite perceptual alterations. There was a degree of, again, a bit of a GI burden with the capsule formulation that we're seeing reduced. There was some nausea at higher doses, vomiting. Is that serotonergic, or was that formulation induced? We can't attribute it to one or the other, but we certainly are seeing less of a GI burden with the ODT formulation that we're looking for in phase 3 in our ClinFarm study. Again, a remarkably benign profile. I think the thing you mentioned is the time course. Yeah, I think there is a bit of a misunderstanding. We have taken a very specific and data-driven approach to how we try to document everything about the profile of the drug. Of course, the time course of perception alteration is a big factor with all of these drugs. In phase 2, we had an extraordinarily conservative set of criteria for how we assess when a patient was clear to leave. We've basically modeled around any perceptual alteration at all. Hallucinogen intoxication and DSM is sort of the model we used. As we gathered those data in phase 2, we both didn't see any sort of physiological burden that would suggest there's a need to monitor or release on blood pressure or heart rates like we see with some other products on the market. We also saw that, you know, we think a more real-world-like monitoring criteria and release criteria are appropriate. As we've gone forward into phase 3 and in our ClinFarm studies, we're using a more, again, sort of REMS-like, real-world-like set of criteria for how to release patients. That has not only allowed us, as we go forward in phase 3, based on the data we generated, to shorten the required monitoring time from 12 hours in phase 2 to 8 hours in phase 3, but we start monitoring for readiness to leave at hour 5. We're seeing a lot of patients, particularly in completed clinical pharmacology studies, that are ready, meeting that criteria, ready to go as early as hour 5, as early as we're measuring it. That is more reflective of something we see like with the release criteria for Spravato or you think about a PACU, a post-anesthetic, release criteria. Rob, when a patient, is there consistency to a patient's release time? Within a reasonably. No, I mean like intra-patient, like as patients get retreated, could a practice say, you know, I see patient X, they will have a release time of six hours, because certainly it would go towards predictability of practice logistics. That's exactly right. That is typically what we see is that, you know, on a given patient, it's not widely variable for an individual patient. That consistency, I think, is even further enhanced with the ODT formulation. You're getting to a really important aspect here, which is when we think about, I think at face value, there's been this temptation to say we want drugs that replicate exactly this Spravato model. That keeps getting referred to as this sort of reference point model out in the world. That has an hour window and, you know, whatnot. Yeah, yeah, two-hour observation. When we think about practice economics and practice operations, the predictability, the reliability of being able to bring a patient in, have them there for a workday, and then let them go at the end of the day, when the monitoring time is reimbursed on an hourly basis, leads to less requirements for room turnover, for patient turnover, to get the same source of revenue. When we look at those logistics, it sort of turns the assumptions on its head, but the kind of monitoring window is, let's say, it's six to eight hours ultimately. Obviously, we have to prove this up. That’s still an hourly business day. That’s not requiring overtime from practice employees. No, and in healthcare, you know, it's typically, you know, either eight or 10-hour days. Anything around eight hours would sort of optimally fit into a workday. You're filling up a bed for the day. There's not wasted time on the end where, you know, in healthcare, time without a patient is not reimbursed time. When we look at the practice operations, practice economics, this dynamic and what we're observing so far in our studies really supports what we think is a pretty attractive opportunity for these sites. That's also reflected in the data that we see out of Spravato. When we go look out in the world, there's a high, high concentration of Spravato prescribing and administration, right? It's only a small number of the REM certified centers that are actually driving Spravato scripts and volume. The remainder of that infrastructure isn't dosing 5, 10, 20 patients a day. Those sites are the ones that are, I think, well equipped for a lower throughput model with better economics and lower turnover requirements. They're the ones who don't have the Spravato scheduler. Many of these commercial sites, community sites have one person dedicated to scheduling all the in and out ketamine patients. Right. We just don't think that model applies to this entire category, whether it be our drug or others. We think there's certainly a different approach that's needed, and we feel really comfortable with the dynamics of our product. Let's go back to your efficacy for a second. Can you review for us the key points about the additional analysis you've generated, response, response in remission at week 12, and then durability, that you have, durability data that you have in hand? Yeah, so some really important findings from the phase 2b study. We saw that at 12 weeks, we had a 7.7 unit improvement in the 100 microgram arm over placebo. Mm-hmm. That is around more than double what we typically see from approved drugs. I said an effect size that is more than double the standard-of-care. Notably, that was over an abnormally large placebo. It had almost a 15 point improvement in the placebo group. To have that large of an improvement over a large placebo does two great things for us. One, it, you know, we think is exciting, impressive efficacy, but it also establishes the robustness of that efficacy because we're not comparing to a nocebo effect or a sort of, you know, paltry placebo response. We're really saying we've not only exceeded a placebo, but a very large placebo. As we go forward into the pivotal studies, assuming the placebo were to return back to a more historical standard improvement of like 9 or 10 points, it would potentially give us an even wider margin to work with when we think about phase 3 outcomes. In terms of other secondaries, a 48% remission rate in the 100 microgram arm at week 12. This is 12 weeks after a single administration with no concurrent therapy or no therapy that came thereafter, and about a two-thirds, 65% response rate at 12 weeks. We're seeing at least 12 weeks of durability. We didn't see any sort of regression back to baseline or loss of activity over that period of time. Are you continuing to follow these patients? Unfortunately, no. Of course, it would be great if we had two-year data on these patients, but based on the nature of that study, we didn't. We stopped observing at week 12. In the phase 3, we continue to observe patients for an additional nine months, so we'll be able to look at that durability well beyond the 12-week primary endpoint. Got it. What are your thoughts right now around the phase 3 program? Have you had all of your FDA discussions around what the final Voyage and Panorama design details will be? Yeah, so based on the phase 2 data, we were able to obtain a breakthrough therapy designation, which, very quickly, which of course also gave us a great opportunity to, and I've had the good fortune of working with multiple of these programs under the breakthrough program with FDA. There's an extraordinary degree of engagement we've seen with FDA, not only for us, just based on the nature of the program, but under the breakthrough programs, especially. Based on that, when we had an interface two meeting last year for our GAD development program, extensive back and forth and great conversation alignment with the agency, even beyond that interface two meeting. We continue to stay closely engaged with the division as we progress with the study. We feel very comfortable about the design of our phase 3 program, both for GAD and for MDD. Let's talk about MDD for a second. There were signals within your GAD study that informed your decision to start that 120 study in MDD. Can you walk us through those signals and that trial as well? Fundamentally, there's a pretty dramatic overlap between these indications. So something like a 60% to 80% construct overlap and certainly a similar diagnostic overlap between anxiety and depression. Anxiety often precedes a major depressive episode, and so while we, of course, didn't involve patients in the phase 2b study who are in a major depressive episode, there is intrinsically going to be some degree of depressive symptomatology in generalized anxiety disorder patients. The scales are just set up that way. The MADRS and the HAM-A are significantly overlapping, and so we started with something of a lower baseline than you would expect from a major depressive disorder dedicated study, right? Patients in a major depressive episode, we expect to have a little bit higher MADRS scores than we saw. We still saw a dramatic and statistically significant decrease in the MADRS 12 weeks after MM120. Really, again, we almost seem to have bottomed out the scale because we were starting from a bit of a lower point. We had almost two-thirds of patients who were in, well, we don't talk about remission because they weren't in a depressive episode, who would have met criteria for remission. That gave us a high degree of comfort to expand the program and pursue a major depressive disorder indication in parallel to the generalized anxiety disorder program. I think it's really worth noting that this kind of fits a larger approach, which is, you asked about generalized anxiety disorder versus TRD or some other indication. Our goal is to have a label that allows us to try to access patients and give them access to the drug as broadly as possible. There's, of course, going to be other barriers in place, prior authorizations, but to have a label for the two largest psychiatric indications in generalized anxiety disorder and major depressive disorder, you know, if we're successful in that, it gives us the largest possible market opportunity and the ability, ideally for a patient who walks in the door, to be most easily directed to MM120 over other available options. Got it. An inbound question from a client. As we've noted, you've mentioned the phase 2b was functionally unblinded. Why was there still such a large placebo effect, and why include placebo in Panorama and Voyage? There are sort of three points in there. For all of the active dose arms in the phase 2, 88% in the 25 microgram group, 90% in the 50 microgram group, and 100% in the 100 and 200 microgram groups of patients correctly guessed that they were on the drug. Yeah. Everyone was functionally unblinded. Yeah. Only the two high doses had a clinical response. Mm-hmm. It is a pretty strong argument to suggest that functional unblinding can't explain the clinical efficacy in the first place. Yeah. The presence of those lower doses, though, 25 and 50 micrograms, when we think about functional unblinding and controls in any capacity, they have a potential effect on not only the placebo group, but also on the active group. In our phase 2, a third of patients who received placebo asked that they were getting drug. Mm-hmm. That's probably more likely to happen when you have a low dose control than when you're saying in a head-to-head study, say 100 micrograms versus placebo. That in part explained, we think, a large placebo response. Uh-huh. We also know that just in any clinical trial, the higher likelihood that a patient is allocated to some dose of drug, the larger the placebo response tends to be. Mm-hmm. In our study, patients had an 80% likelihood of getting drug. I see. Okay. The third factor is a sort of in the weeds statistical one, which is that in patients who receive placebo and subsequently drop out of the study, their data from that point forward is imputed as if it is from the remaining placebo patients, imputed as if they're part of that placebo group moving forward. The patients, because we didn't offer anything beyond that initial randomized dose, the patients who did the worst were the most likely to leave. Right? If you have a placebo patient who most likely did not get better or even got worse, they left the study at a higher rate than anyone. That also numerically is going to likely drag down that placebo response. All of these, which we think could potentially be more favorable in terms of mitigating placebo response in phase three because we offer the extension period where patients can get open label drug. We have a lower allocation ratio and we don't have multiple very low dose controls in the study. To talk about the phase threes and why I included placebo, functional unblinding is sort of a spotlight has been shown on it for this category of drugs. I was talking to a KOL a few weeks ago, this is true. Functional unblinding happens with every drug in psychiatry. Let's look at benzodiazepines, you know, one or two milligrams of Xanax, it is clearly functionally unblinding. The gold standard for all of clinical research has been to compare it against a placebo. In one of, in Panorama, we're including the secondary control, right, of a 50 microgram arm. That's there to basically say to the patient in the informed consent process, if you feel the effects of a drug on dosing day, you can't assume it's a clinically active dosing drug. We've actually shown that one of the doses being tested didn't have any clinical improvement over placebo in a prior study. It's there as a confounding dose to try to sort of functionally blind that. To compare against anything but a placebo for us departs from a long established gold standard in all of clinical research. We've been very clear and we think very aligned with FDA, around our logic for doing so and the approach in doing that in phase three. In part B of Panorama and Voyage, you're allowing re-dosing. Can you talk about the parameters where that's allowed? Absolutely. Patients who are enrolling in the study have to have a HAM-A score of 20 or greater. Mm-hmm. After week 12, patients who continue the study, if they have a HAM-A of 16 or greater, become eligible for open label treatment with drugs. In most open label studies, a daily drug, an SSRI that has an open label extension, patients immediately go on to open label drug and therefore are unblinded. This design allows us to do two things. One allows us to look at the durability of the initial response in a blinded status well beyond 12 weeks. A patient who takes the randomized dose of drug or placebo and is in remission for the entire year, we can look at the rate of that and the sort of survival curves over that entire year to see how long a single first randomized dose lasts in comparison to a placebo. The only thing it does, though, is then for all of the patients in the study, regardless of what they're randomized to first, we can look at the characterization of retreatment over the course of the year. Like real world, like what on average, how many doses a year does a patient need? What is the dynamics of retreatment? How often is that needed? What is the interval? Does that interval change over time? All of these very interesting and important questions that could be useful in certainly informing regulatory discussions, but also I think as importantly informing clinicians and our meta-analyses and publication strategy over time. Athena, do you want to take the last set of questions? With the last minute that we do have, how do you think this would fit within the real world paradigm? How many monitors do you think this would take, and how quickly do you think these patients would be discharged upon the ending of the treatment? For the first question, our approach in development is to make every argument based on data. We go to great length in the trials to document what monitors, we call them dosing session monitors or DSMs, do during treatment sessions so that we have data to articulate. If it turns out that no one has, no DSM has had to intervene at any point in time, it's exactly clear what they're doing in the room. That's what we hear from past studies. A lot of times, we've had investigators say, look, I sat in the corner of the room and caught up on notes or read a book. I didn't do anything during the day. We want to be data-driven in the decisions. If it turns out that multiple people need to intervene, then that's going to be important data that everyone should be collecting as well. We feel really confident about that. So far, we've been really encouraged that, again, there's no concurrent therapy and there's not a lot of intervening that happens by the dosing session monitors. When we think about monitoring time and release and when patients will be ready to go, again, it's a data-driven approach. We'll have a structured, granular assessment on an individual patient basis. Of course, we will do summary statistics around when we come out of the phase 3 studies to argue or to establish appropriate data-driven conclusions for what appropriate monitoring periods would be. We so far are really encouraged by what we're seeing with the ODT formulation. In our clinical pharmacology studies, we've seen a pretty dramatically shortened period of time before patients are ready to go. I think it would be predictably under that eight hour or within that eight hour kind of workday period. Understood. Thank you so much. We're really looking forward to that phase 3 data, both from GAD and MDD programs. With that, I will wrap up and we will be hosting Acadia for our next fireside chat. I hope to see you there. Thanks, Rob. Thanks, everyone.