All right. Welcome, everyone. I'm Gavin Clark-Gartner, one of the senior biotech analysts here at Evercore ISI, and I'm really happy to be joined with the COMPASS Pathways team. In order, we have Teri Loxam, who is the CFO, Kabir Nath, who is the CEO, and Mike Gold at the end, who is the Chief R&D Officer. Thank you all for joining, and by the way, thanks for a really great dinner last night in addition.
Pleasure.
So maybe just open us off with an overview of the company and where things stand today.
So great. Thanks, Gavin. And just a reminder, we will be making forward-looking statements, and you should refer to our risk disclosures. So no, thanks, Gavin. Thanks for the opportunity. So we are firmly executing on phase III for treatment-resistant depression. As a reminder, this is a very significant unmet need with only one currently marketed drug, esketamine, which we might talk about more later. We generated very robust phase IIb results that were published three years ago and in the New England Journal two years ago, showing really a very significant efficacy signal in this hard-to-treat population with a balanced safety profile. Based on that, designed and are currently executing a very robust phase III program. The first study, which we call 005, a single dose of COMP360 against a true placebo. We will see the six-week primary endpoint data for that in the second quarter of next year.
And the second study, a three-armed study with an active dose of 25, another active dose of 10, and an inactive control of one milligram with a repeated fixed dose after three weeks, the same three doses we used in the phase IIb. That study, we will see the 26-week data in the second half of 2026. So while the core focus is on execution of those trials, in parallel, we have foundational work around how to build a commercial business that will be scalable in time for launch. And a lot of that is building on some of the lessons, both positive and things that need to be done differently from Spravato.
All right, great. Let's start off with the first phase III study for 005. What specifically has been causing the delays, and why are you really confident this will not spill over into the commercial setting?
So I'll ask Mike to comment on the delays, and then I'll come back on the commercial piece.
So let me kind of be clear. These are best-in-class studies for psychedelic and TRD. They're complicated, demanding trials, right? It requires a lot from the sites. It requires a lot from the participants. There is a very stringent criteria in terms of documentation to make sure that people actually do meet sort of current working definitions of TRD. And that means that we spend a lot of time getting medical records to make sure that they've been on, they had the appropriate therapeutic trials. So getting access to those records had become somewhat logistically challenging. So we had to block and tackle to find a solution for that. Some of the sites had delays getting their DEA licenses because it varies on a state-by-state, almost local jurisdiction level. So that took a little bit longer.
Figuring out how to do the therapist for the psychological support model in terms of identifying the people, training them, retaining them, figuring out the schedule, that also generated some logistical challenges. So what we've seen over time is essentially peeling off the layers, blocking and tackling these independent things. The one thing that we also, and it caught us a little bit by surprise, was that a lot of the sites, because this is a very different way of administering a drug in general compared to other depression studies, I mean, you don't just give a little pill and somebody goes away and comes back. This is actually resource-intensive. Somebody sits with a subject for the duration of the psychedelic experience.
A number of the sites wanted to actually kick the tires and get one or two subjects through the process, figure out how to get it done right, the patient or the subject flow. And once they did those one or two subjects in series, then they started to open up and be able to treat and bring patients in more than one at a time. So we're seeing all that kind of come to play. It took a little bit longer, but certainly the recruitment is accelerated. It keeps on accelerating. All the sites are active. So I think we're doing a stand-up job of addressing sort of the individual site issues. And I guess maybe it's stating the obvious, right? Psychedelic trials in Florida are probably not going to run the same way they will run in Chicago, certainly not in Missouri.
So we have to flex and we have to adapt to what works best for the individual sites and the culture in which they operate.
Kind of three key reasons why we don't think that translates to the commercial setting. First is in terms of the selection criteria. As Mike said, we have extremely strict inclusion criteria that require complete documentation of at least two failed episodes. A typical prior auth, which a commercial payer will likely impose, we almost certainly will see two steps. That's typically just physician attestation. That's all that's required. So it's a completely different level of proof that's required for a patient to be treated commercially. Second, as Mike said, per FDA requirements right now, the people in the room providing psychological safety and support have to be postgraduate licensed psychotherapists, even though they are not offering any form of therapy. They are literally there for support and safety only.
It's clear that in the real world, it will be a much wider pool of licensed professionals who can be in that room. It could be nurses, social workers, PAs, as well, of course, as therapists themselves who might choose to participate. So again, that bottleneck around accessing specifically a therapist who's been trained by COMPASS will go away. And the final thing I would say is these are very complex trials. And the whole informed consent is, of course, not about getting people better. It's about presenting as neutrally as possible all the things that could happen in a trial context. In the real-world context, it's all about getting the patient in front of you better. So the psychiatrist, the office staff, and the patient are all in a very different case in terms of that.
The final thing I would say is none of this is around the top of the funnel. Unfortunately, the unmet need and the number of patients out there with treatment-resistant depression who are already somewhere in the healthcare system is huge.
Yeah, that's very clear. All right, and turning to the second quarter of next year, when you'll be announcing the top-line data, you won't be showing all the trial details at six weeks as to not break the blind. So maybe just help us understand specifically what data you're showing. Just remind us of that, but also how you decided that this amount of data did not compromise the rest of the trial.
Yeah. So what we will be showing is the primary endpoint, which is the difference in change from baseline on MADRS between the 25 mg arm and the placebo arm and the associated p-value and confidence interval with that. We will also have a statement from our independent DSMB around safety. And if you like, we can get into a little more about what that will say. Clearly, the trials, as you rightly said, this is the six-week primary endpoint. They are running blinded for 26 weeks. The decision to announce some six-week data is like all these disclosure decisions.
It's a trade-off around risk, around the desire to maintain the blind as far as possible, but also a recognition that actually demonstrating that we can replicate the efficacy we saw in phase IIb in our first phase III study is a really, really important signal for the market and for investors. So that was kind of the trade-off. Anything you want to add, Mike?
Nope. Good.
That's great. All right. Let's zoom into a little more detail on the safety side with suicidal ideation. Maybe just kind of frame why you're not particularly worried about this side effect, but also with the DSMB statement, how did those conversations go? Do they have any criteria around what determines an imbalance in suicidal ideation and how we should kind of interpret the statement that comes out?
No, that's great. So just for context, suicidal behavior, ideation, suicidality is part and parcel of TRD. It's part of the reason that the unmet medical need is so critical, right? So there is a background rate of suicidality ranging from just kind of ideations to attempts to completed suicides in TRD. And so nobody should be surprised that in any clinical trial, we will have some of these events, right? What we discussed with the DSMB is that as we present efficacy on the one side, obviously, I mean, I'm a clinician, so I want to know benefit-risk. Always want to understand what the benefit is and is there a particular risk. And I think all of you are probably in the same place.
What we did is we actually approached the chair of the DSMB, who's a very experienced psychiatrist who's done a lot of these clinical trials, said, "Would the DSMB be willing to put a statement out about either our latest review of the data because they see both blinded and unblinded? And then specifically, would the DSMB be willing to put a statement out about whether they see any meaningful imbalance in suicidality between the treatment arms?" That gets to sort of the last part of your question about how do we define meaningful. Here, we've taken really kind of, we've deferred to the expertise of the DSMB. We do not have specific quantitative criteria that say if it's three cases versus two cases or four cases versus two cases.
We haven't established that because every single one of those events has to be interpreted on its own. It has to be contextualized. So we've actually left it to the DSMB to tell us whether they believe that whatever data they're seeing crosses their threshold of meaningfulness, right? And so the expectation based on the latest review of the DSMB is that the comment to us continues as is. They did not identify a signal, and we expect that to be the continuing sort of message.
I think it's important to clarify for those who are not in the details of our phase III trials. Both of our phase III trials, the primary endpoint is at six weeks. However, both those trials remain blinded through 26 weeks. The reason we're having some of these conversations and a proxy of the DSMB for safety is that at six weeks, we are not fully unblinding the data set. We're only providing a limited set of data, even though it is the primary endpoint, to maintain as much blinding as possible in that trial. For the second phase III trial, 006, we're now going to be only disclosing at the end of that blinded period of 26 weeks to avoid any of this complexity and preserving that full set of data.
Yeah, that's great. And I know you're not giving explicit numbers for a lot of this, but maybe any qualitative comments on what you're seeing for the 005 trial that kind of inform trial integrity, thinking about number, concentration of patients coming in per sites, discontinuation rates you're seeing, patients going on to receive subsequent doses, all of those measures?
So what I can say is that the discontinuation rate is running below what we actually planned. So subjects are staying in the study as planned. We're seeing a good number of subjects come into part B now, right? We may have the first couple of subjects come into part C shortly, right? So we're seeing good flow. We have a very extensive data quality process. It's actually three parts. It's the work that the CRO does for us. We have a central rater also that looks at data, and internally, we're looking at data. So in terms of the nature and number of queries that we're raising, there is nothing atypical about the nature of the queries. The data quality looks actually extraordinarily clean.
We have additional other metrics that we look, for example, part of the rater training program to look at how well that psychological support is actually being provided, so we have some fidelity numbers, and those fidelity numbers are well above what the industry would consider a reasonable cutoff in terms of the provision, so across the board, safety looks pretty good. Adherence to the protocol looks good. We're not having any problems with any particular sites. We are in the process of culling sites that are just not able to recruit. We're not going to carry nonproductive sites, so we're always watching that, but no GCP issues to speak of, so quality looks good.
No single site will be more than 10% in practice. It's likely to be a little bit lower than that, the maximum number that will come from any single site.
All right. Great. And thinking about the MADRS effect size, which will definitely be a big focus area for the top line, what do you think is the minimum bar to be commercially viable and/or from a trial statistics perspective?
I would say anything north of two-point difference between control arm and active in a TRD population would probably be of interest to clinicians and patients. We're aiming for higher than that, right? So the studies are generally powered for effect size. So we're about three and a half, four points, right? We saw larger effect size in the phase II, but remember the primary was at three weeks, right? At six weeks, you still saw a pretty big effect size. But these are larger studies with a more heterogeneous collection of sites. So we took a bit of a haircut in terms of what the effect size were powered for. But it's still very comfortably in the clinically meaningful effect size arena.
Yeah. Great, and similarly, kind of on the same topic, for durability, what do you think is kind of the optimal or the minimum maybe interval between treatments in the real world? What's kind of viable?
Yeah. Well, I mean, I think we should start by contextualizing that because the options patients have today are daily orals, which essentially aren't working for them, which is why they are resistant to treatment with those, or conceivably esketamine. And esketamine has been successful precisely because of the rapidity of onset. But remember, it requires, per protocol, eight visits in the first month, four in the second month, and thereafter once or twice a month ad infinitum. So in that context, I would argue even six weeks, robust data at six weeks would be pretty impressive. What we saw in the phase IIb was we had durable response out to 12 weeks for about 22%-23% of the population. Through part B of the two trials, we will get a range of durability for either one or two doses somewhere, hopefully in that 6 to 26 weeks.
I think in the past, we have said this might look like two to four administrations per year. But again, we really have to see the data and really understand and recognize there's no such thing as an average patient either for TRD.
Just a little color around this. Remember that this is intermittent dosing with 100% adherence. We don't have the problem of losing, right? So supervised 100% adherence, and it's intermittent quarterly or semi-annually. It's much easier on patients, a lot of whom don't work or some who work taking time off of work to come into the office. So this is much lower burden.
Yeah. Great. And I'm thinking maybe we have a separate conversation on commercial since we could spend a whole hour there. And maybe what we do is we wrap up here on the regulatory side. It's probably worth just framing the Lykos AdCom, the outcomes and the learnings. I mean, the way I distilled it, there were a bunch of points, maybe four main ones as one. There was maybe poor coaching and misunderstanding on the functional unblinding component. The monitoring versus counseling is an important dynamic. I mean, safety monitoring is obviously important. And then there's trial population consideration. So anything you want to frame and how that doesn't really read through to you guys.
Yeah. So look, I mean, that was not a fun AdCom to watch, right? And so I think you nailed the main points of difference. A, we don't do psychotherapy. I've listened to transcripts of actual recording of sessions. And in a six-hour session, what you hear is five hours and 55 minutes of the subject breathing, music playing. There is no conversation going on other than, "Do you need a blanket? Do you need to go to the bathroom?" So right then and there, there's a big difference between the problem that the agency had about how much psychotherapy counted for the benefit versus MDMA. We also were very careful about not biasing our sites in terms of prior experience with psychedelics.
Our therapists, we don't require or insist or care if our therapists have. We actually don't want them to be biased towards a psychedelic. So we don't have that problem either, okay? You didn't get to it, but the other part that was also missing out of the Lykos submission is they had a very incomplete ClinPharm package, right? Safety, drug-drug interactions, cardiovascular safety. And these are vasoactive compounds. We're going to be coming in with an absolutely complete, very vanilla, right? Very vanilla ClinPharm program, but it's exactly what the regulators, FDA and other regulators are asking for to build confidence that we have an appropriate package for that. So across the board, I think we're very different. I can't help but say we'll be much better prepared for an AdCom than our predecessors.
Yeah. And the one area where we clearly have revised at least our disclosure plans was around the disclosure of 006 and the determination to keep that no disclosure until week 26. And that was to your point. I think you said it well. I mean, there's a lack of clarity. There's a lot of confusion about unblinding, about expectation, and so on. So in the light of that and the confusion that we saw, that was the decision around 006.
Yeah. That makes sense. And maybe I'll ask one closing question, which we've started to get more increasingly, which is with the RFK nomination. What do you think that means from a regulatory perspective?
So the easy answer is who knows because, yes, we've had lots of nominations. We have no idea what's actually going to happen come January or thereafter and so on. But I think a couple of things. I mean, first, to build on what Mike said, hitherto, we have breakthrough designation. Our interactions with the agency have been nothing but positive. They clearly are supportive of the work we're doing. Tiffany has publicly been supportive of psychedelics and the fact that she would like to see some of these be made available to patients. No shortcuts, which is also absolutely fine. So that's the status quo. I think clearly there is a possibility we've also always guided clearly that if you look at the FDA's guidance, we need both COMP005 and COMP006 for filing. That's clear from their guidance of what constitutes a well-designed phase III.
At the same time, clearly COMP005 reads out earlier. So if we have really great data from COMP005, including durability data from the 26 weeks, with our breakthrough designation, we will absolutely be having a dialogue with the agency. That is not to suggest that that's the base case or likely, but to me, that's the one area where you could see a potential. But I don't think it serves anyone for there suddenly to be a one way that says psychedelics can now get approved more easily than any other psychiatric drug. That actually is not something that we would support.
Yeah. All right. Super helpful framing. So we'll be back for many discussions in what's going to be an exciting 2025. Thanks for joining us.
Thank you.
Thanks very much for hosting, Gavin. Thank you.