Good afternoon, everyone. Thank you so much for joining our virtual fireside chat. My name is Yaz. I'm excited to have Marcio from Praxis Precision Medicines here. We have a full hour for this session. I know many of you are buried into earnings, as this morning has been very, very busy for all of you. I'm very grateful that everybody is able to join the fireside chat. Again, housekeeping rule is the same. If you have any questions during this fireside chat, just utilize the Q&A box at the bottom of your Zoom and submit your question. I'll make sure to direct them to Marcio. Marcio, I think that maybe a good place to start off is sort of big picture, right? Before Monday's readout, we had the phase one data, and we weren't really sure by looking at the phase one data how Vormatrigine would differentiate.
Now we've learned quite a bit around its differentiation with this data set, versus other therapies. Maybe walk us through what have you learned from the study? Question two, also broad, what are some of the, I guess, things and considerations that you think the street has not appreciated into the data? We'll go really deep into that RADIANT.
Sounds good. I appreciate the invite, first and foremost. I always knew you live in the future, Yaz, but you just confirmed it. I think it's still morning on every time zone in the U.S.
Yes.
Good morning, everyone, not afternoon. If you take a step back, just as you said, a big picture of Vormatrigine, what we are seeing, and I guess what you continue to see, carrying from great preclinical evidence to the PPR, to obviously all the safety and pharmacology and everything preclinical and clinical we've seen so far, is a drug that really can help patients with epilepsy a lot. That was a promise until this week. I think that's quite important. Maybe on the, I'm not sure if I say underappreciated. I'm not sure if there is a need for appreciation as much as just people paying attention to everything that's going on there in a world with very low and narrow attention span. When we had the PPR data last year, there was a lot of "pressure" to say, what does that mean? What does that mean?
You might recall we had some conversations, and we always, I would say rightly so, said what we know is this drug is in the brain. What we know is that's doing something for the epileptogenic formation that's happening there. We're measuring this. We're happy. We can't really translate to what's going to happen when we treat a focal seizure patient or a generalized patient. A lot of asks and pressure to try to translate. I think we stayed pretty serious about it because you really couldn't, right? That's where the science is, despite the fact we had 100% of the patients responding on that study. You fast forward to the place we are today, the data set we really wanted to have for multiple reasons. As a first data set, of course, is RADIANT, where we treated a significant number of patients, refractory, difficult to treat.
One could even argue the most difficult patients to treat, the ones that actually gravitate towards studies like this. It is undeniable how active the drug is. Transforming from a promise to a drug is a big deal in my view. That's what we have this week. Now we can go one by one, as I'm sure you're going to go, into the aspects of that and why we are so excited about it. In summary, when we look into the actual markets, looking to not people like you and I or all the investors on this call, but the real people living with epilepsy, it is very, very clear the issue out there. Number one, it is a massive market.
The idea that, as we go very quickly to this zero-sum game that there is a drug or two drugs or three drugs that are going to solve the problem of epilepsy, I think is a little bit delusional and far-fetched, right? Are there situations where you're looking to the market and we slice that differently and you say there are bigger winners throughout those slices? I think so, and we can talk a lot about that. When you have 3 million, and I'm rounding down, patients, and over 60% of them cannot tolerate or respond to the current treatment, I think we all would agree that there's a huge space here. Looking to the history of epilepsy drug launches that were safe and efficacious, they all did really well. Now imagine one of the profiles of Vormatrigine and then you can imagine what it can really do.
I think that's why we are so excited, right? We went through somewhat of a promise. Of course, we were excited about it to a lot more than a promise today with helping real patients with focal seizures reduce their seizures or resolve their seizures, as you've seen on a good number of them.
Thank you, Marcio. Let's now dig really deep into many elements of RADIANT, and then we'll talk about POWER1 and POWER2 and Three. Let's start, just confirm the trial size. RADIANT, when you reported out, you said you had 61 patients as of July 25th, and the total is going to be targeting 75.
Right.
You're going to read that out in 4Q.
Right.
Is that difference of 14 patients then generalized that we're going to get at, or are we going to get in the next 4Q a combination of both?
No, a mix. I would say we're pushing very hard for the generalized patients to be the majority of them, of course. That is the data set that is missing here for us to complement. There are still, I think we're very transparent on the disclosure, right? Talking about the patients that were screened, meaning they finished the screening. There are, of course, patients in screening, and some of them are generalized, some of them are focal, smaller number in focal. The patients that we dosed for more than a week, that's the 61, and between that and like multiple weeks. The patients that had completed, and that was the data set promised to the street, and we have here to discuss, that was the 37. It is actually, in a sense, it's a place of privilege, right?
To be with so many patients, number one, interested, and two, being able to recruit in such a fast study. It gives us a lot of safety data, gives us a lot of understanding on this patient, a lot of different background therapies, and things like that. We feel for the stage of the program, running like multiple phase threes on the next several months, it is a very good thing to have so much depth of data to learn what we can learn. We learn a fair bit about, number one, how to recruit these patients, how to follow up, and like we had incredible compliance to the seizure diary, for example, with something super important, but it's never a given compliance to drug. Everything else, management of AEs, as I'm sure we're going to talk about. It's a large number.
This is not like 10, 15 patients as we had before. That allows us to make a lot of extrapolations as well.
I guess it's fair to say that when we get data in 4Q, it's going to include obviously generalized patients that we have not seen, and maybe a few additional focal epilepsy patients that maybe had not finished eight weeks at the time of the data cut of July 25th.
Correct, correct.
Okay.
Let me make a clarification here. All the difference between 37 and 61 are all focal. We only did focal numbers there. We can expect a few more, two or three more or something like that, focal, and then the rest generalized.
Okay. All right. Very helpful. Now I would like to spend some time on walking through exposure related to efficacy, safety, and discontinuation, because I think that's going to become really important as we think about this translation to the POWER1 study and POWER2. I guess the first question is, can you confirm that the PK in patients was comparable in healthy volunteers?
Yeah, beautiful thing. I love the fact you asked this because I don't think we discussed this enough. The number one thing we wanted to learn here, right, was exposure and then safety and then effectiveness. The exposure, there were a number of questions we wanted answered, and I think we answered all of them so far. The first one is exactly what you just asked. Is the PK profile outputs or PK comparable to our volunteers or not? We know that patients with epilepsy in general, but for concert seizures particularly, have a lot of other drugs, a lot of comorbid conditions. Things just happen, right, on these patients. They're not, quote unquote, healthy as the healthy volunteers are on a study. It's incredibly comparable. We're very, very happy with that. It is very reliable as well in terms of predicting.
You want to first know, like, can I compare with what I have for our volunteers? The second is, can I extrapolate from that?
Before you go there, I want to walk through each of them step by step, okay? We learned PK between healthy volunteers and patients was the same, check. Second check is, can you confirm that the 30 mg dose group was linear in AUC and Cmax?
Yeah, there was just like we had before, right? We were expecting, of course, we didn't have multiple doses here, but it was comparable to what we've seen with the previous multiple ascending dose for 30. The exposure spread was similar as well, which makes us believe that the way the drug is, the accumulation was similar, steady state duration was similar, that there is nothing to believe here that the dynamic that you've seen before of linearity wouldn't be kept on this.
Okay. Next question is, would you say that the level of variability is comparable?
Yes, fairly comparable. Yeah, surprisingly, right? You would expect a lot more variable here. That is another very good metric.
Okay. For example, if you previously disclosed that the coefficient of variable for 45 milligrams in healthy volunteers was 29%, you saw something very similar in focal epilepsy.
Correct.
Okay, perfect. Now I know that now with this data, you guys are going to do and you do extensive modeling, right? And think about sort of additional data over time. I guess the question that I want to know is like almost any exploratory analysis, you could look at quartile based on AUC, and then you look at the sort of seizure reduction, right? You can actually think of treatment response to AUC based on quartiles. Would you say it's fair to assume that the top quartile could actually generate 75% reduction? Like since it is linear and you can divide it up, is that, am I getting over myself as you think about that? Because that's going to be important when we think about the 40 mg dose group in POWER1.
I couldn't agree more that that's quite important. Let me take a step back.
Yes.
Right. We collected intense PK on day one and at the end of study on day 56, and then trough levels throughout the study. A number of predictions here, right? We use a POPPK model that's being developed based on the healthy volunteer data, and then we check with this and so on and so forth, as you know well. We ask a number of questions to that. It's like, how is the not only AUC, as you just mentioned, but to jump into AUC, you have to believe it's an AUC-based phenomenon. It can be both AUC and Cmax or other things based, and I think there's a combination of both here. I can tell you that that is a dose response or a concentration response because it's really not dose here, the dose was just one for both Cmax and AUC in terms of response.
Now, if you split that in four similar groups, as you just mentioned, right, and in quartiles, yes, the top group is better and better numerically. I always have to put the caveat, right? This is not that small of a study, but we have to be careful once we start breaking down things into smaller and smaller groups. The error definitely expands on a situation like that. It gave us confidence for two reasons. Without maybe getting off my skis and saying, oh, that's going to be a seven or an eight or a six handle, whatever it is on that number. I think there are a couple parameters we wanted to make sure were true for that dose of 40 to be selected for POWER2. The first one is, yeah, that number had to be bigger, right?
Meaning that is a reason to go there from an efficacy perspective. That alone wouldn't allow us to go there, right? If all the side effects were concentrated there, or there was a very direct relationship, then we would have to pause as well and say, then we go there and the safety profile. It's not the case, right? I think we have a lot more of a response in terms of efficacy than we have on safety from a concentration standpoint, possibly because there's a lot of other things influencing safety here, and efficacy is really being influenced by the drug. That might be why. This is amazing. This is almost like what you always want, right? You're giving more drug, you get more effect, but you're not having any worse safety profile.
That was the base on why we believe that 40 would be adequate here in POWER2. When you combine a time-based effect, which we haven't discussed, right? We're discussing concentration like independent of time, but you have both concentration and time. It's a longer period of time. We also see like a deepening over time of effect. Going to 40, it would not be proper or correct for me to say I don't expect a larger effect there because it wouldn't be true. That is what we were thinking.
Marcio, maybe, you noted too, right, the PK and the exposure had no effect on safety. You just noted that. Did it have an impact on discontinuation rates?
Yeah, it did not. Yeah.
Okay.
You're going to need to check the box or to explain?
No, I think no, because we have so many questions that I want to kind of like team up in a way just as we think about it. Now that we're on the topic of discontinuation, let's just kind of close the loop as investors have asked a couple of things. One is when you think about the discontinuation rates that you reported of 23%, that 23% was based on the 61 that you got at the beginning of the study. Is that correct?
That's correct.
Okay. Second question was, was there a time by which these discontinuations occurred? What was that time cadence?
Yeah, the vast majority of those patients discontinued in the first three weeks of treatment, with a little bit more towards the first and second week there. It's interesting, yes, right? Like we are maybe to a fault, like we're saying, like we're not happy with this. It's not that we're not happy because the data is bad. It's normally like people are not happy because it's bad. It's actually very good from a discontinuation rate. Ideally, you wouldn't have any, but you got to understand, and we said this before, right? I think that's why I find it a little bit ironic that people are bringing it up as we haven't talked about this before. The setup of this study attracts patients that are a little bit more severe, they're a little bit more refractory, or also they know their own drug.
They are more likely to attribute anything to drug and potentially discontinue. I think what we failed to maybe act a little bit faster was when we informed the physicians, the PIs, and we trained them and reinforced the training with them. I would say it's never enough. Maybe if I'm going to call failure, maybe a small failure is to just reinforce even further. What we ended up seeing was it was probably more related to certain sites than to the overall effects. Now we can definitively say that, right? We'll keep running these studies. If the experience on POWER1 is of any information right now, and it's blinded, and of course you got to be careful with blinded data in general, the overall rate is significantly smaller.
I think we can overlap the fact that certain sites that had a little bit more discontinuation are not on POWER1, and they were the ones influencing. If it was a generalizable phenomenon, it would be seen across the board in other sites as well. Either the training that we obviously always reinforce, or the fact that there was really a little bit more PI driven to some extent, and I'm not throwing anyone under the bus, right? No one in this patient should discontinue. It just sometimes happens. That might have unduly influenced the rates here.
Marcio, since you noted that the POWER discontinuation rates on a blinded basis is less than what you saw in RADIANT, can you remind us what you have said around what your assumptions were in discontinuation rates?
Yeah. If you go back and look into the way we power this study, all the data we had to work with was historical data. You go through historical data on both drug and placebo throughout, like I'm going to call modern anti-seizure medication drugs. You have like X-TOLE and you have like the C013 and C017 from XCOPRI. Neurocrine study for their 1.6, actually Xenon's original molecule, but Neurocrine developed like 1.6 blocker and so on. Looking through all of that data, which is publicly available, it always tends to be around 20%. You go across the board and you look into that and it's like, yes, a little bit higher for the most part, but I think we can say that the central tendency there is around probably 25%, but then we are like 20%.
That was the assumption to begin with, that we are on the actual protocol and how we determine the power for that study. Of course, we believe that with more intense management, meaning really just talking to patients, right? I talked to investigators this morning. I talk to investigators all the time, and they're like, it's common sense to just call patients and check in on them in general, let alone on clinical studies. Sometimes it's just not done. We would, and we still believe that in the future we're going to have lower rates than that, but that's how it was defined. In a sense, almost every study in every disease has a base rate, and the base rate for epilepsy is around 20%.
Okay. Thank you, Marcio. Let's pause and take a couple of clarification questions since we are on discontinuation. I guess first question from our client is asking around what discontinuation and dose reduction rates were in the 30% of patients who were on background, so now I'm going to make patients versus the 70% that was not. Could you maybe talk about that?
Yeah, absolutely. If you look into those 30% or so, a little bit higher on an ongoing basis, now I'll make rates, right? I think the way it was portrayed to me, I don't like talking about patients like this, but I'm going to paraphrase. It's like the worst of the worst, right? It's like the patients we threw everything at them in general. It was no different than the overall. If we go and we look into two aspects of that, which is quite impressive in our view. One is to ask, and I don't know what you hear from doctors, but when you go and poll the market and we ask which dose of Cenobamate are you using, right? The most common number back and what claims data to the extent that triangulation tokenization works here is around 200 milligrams per day.
Now, those patients were around 300+ per day, right? They are way more prone to have all sorts of side effects and discontinuations, things like that. It basically just says it's not a hedge or anything like that. It's basically to say those patients were really on the top of what they could tolerate, the brink of that. The discontinuation was not any different, right, on this, which to me is quite positive.
I guess the one positive thing is that when we had done channel checks, a lot of docs don't really get comfortable to prescribe more than 200 mg. The fact that you just told me three, like majority of them being on 300, really highlights that these are really experienced docs that are part of the study because a normal doc would not feel comfortable to go there. Maybe another nuance here, just as you're talking through this, is sort of the integrity of the study and the quality of the sites.
Oh, no doubt.
Yeah, one more thing I want to drill down to is, and you said it in passing as we were just having the conversation, that the discontinuation rates were not like that they could have been run by few sites. Could you drill down because, I mean, did you get a chance to look at what the distribution was? That's another maybe aspect to keep in mind.
Yeah, yeah, no, we did. I think it's one of the first things we did on an ongoing basis. Our clinic operation seems quite on top of that and having conversations. They also happen to be some of the first ones to have patients, so no fault of their own in a sense. They are learning a little bit about this, and there was somewhat of a concentration, I would say. As numbers grow, we should keep asking, is this concentration real, right? Is everything else with you? I think now we can fairly confidently say, yeah, there was somewhat of a concentration as we have a significant number of patients dosed across the program, which translates into a good thing, right? Because it is very hard when you have no idea where it's coming from.
It becomes why, again, not necessarily a positive overall to have anyone discontinue, but when you know where it's coming from, there are actions you can take. I think the primary action here is just a reminder, like consenting of the patient, reminding of what the side effect is. There's always going to be a random effect, right, on what happens. I think we're trying to control the non-random effects like this one you just described.
Okay. Another question that a lot of people are asking us, since the majority of the discontinuations occurred between week one and week three, how is that impacting the efficacy? How did you calculate total seizure reduction for the ones that you missed the data for, both for baseline and end of treatment?
Yeah, absolutely. The most accurate way to calculate seizure rates in epilepsy studies, and maybe a little reminder there as well. I know it's obvious, but we use an electronic seizure diary where patients have to confirm their seizures or lack thereof every single day. We have that, and whether or not they took the drug as well, they confirm every single day on the diary. You have a number of days and a number of seizures, so you can create a rate there. That is pretty much how everyone does it. There are small variations here and there on the fringes. It does not really change the overall number at the end. You would imagine that if a patient discontinued earlier, they tend to have, number one, less opportunity to respond, as response depends over time.
The second is they sometimes are not the best responders because that's one of the factors on discontinuation. That's kind of a double negative in a sense. The data are used, if that was the question behind the question, their data is entirely used. One could hypothesize, while that's the proper way to analyze it, that it reduced the effects. When you reduce the number, you increase the effects, hypothetically speaking.
Okay. Another few quick, how many total patients discontinued out of that 37 that you looked at?
Yeah, we're going to put the exact numbers and numbers at risk at the IC post in like three weeks or so. It was similar in terms of the rates of discontinuation between the 37 and the remaining that are in the 61, numerically a little bit more on the first 37.
Thank you. That's helpful. Is it fair to say that at the final analysis, 32 out of the 37 finished eight weeks of treatment then?
That would be a little bit better rate, right? It would make the other ones to be less.
Okay. Yeah.
No, it's a little bit smaller. Yeah.
Okay. Got it. Okay. Marcio, would love to talk about, I mean, I think big picture here is if you look at other focal epilepsy studies, and I think it's in the appendix of your slide deck, comparing it to Xenon and Cenobamate studies, they were between 23 to 27. It's not really shocking that you ended up at 23. I think we get that. I appreciate you answering a lot of college around it. I guess another question that has come up is when you look at, I guess, one of the things is the total rate, the reduction of seizure, total seizure reduction was 56%. If you look at the time curve and you look at week one through week eight, you really see actually consistency. If you did calculate your seizure reduction from baseline to week eight, you would have ended up with 70%.
I think even in week five, you had 100% seizure reduction. Maybe kind of help us think about now that you learned this about your drug and POWER1 is going to be eight weeks, what do you predict to see by going longer and plus sort of, and then remind us, what is the calculation of seizure reduction, I guess, too, you know, in POWER1 versus RADIANT?
Yeah, absolutely. We do see, right, and I mentioned that on the PK discussion we had, both concentration response, which because there's some accumulation, there's a time response to it as well, a time-dependent response. Can we disentangle? No. They're both, and we can look into both. Maybe, again, too honest there on saying we really can't disentangle. We do see over time patients get better. I think the most obvious way to look into this is the spacing of the seizures, right? They get longer and longer in between, and eventually, I think we all wish they go to zero on the majority of the patients. Just by linearity alone, you would expect that you'll get four more weeks. That's the case for POWER1. It's the case for POWER2 as well. The number would continue to reduce.
It wouldn't be far-fetched if you actually look into our experience with RADIANT and EMBOLD, right? Every time we checked, and Ritalin is a great data set to look into that because there's barely, there is no discontinuation basically, right? We are really looking to the same cohort of patients without much of a survivor bias, which sometimes the biggest problem on longitudinal cohorts is that you're really not looking to the same number on the long run or even a similar number on the long run. I think when you take that learning here as well, and the learning we have for Vormatrigine, yes, it should. I'll go back to do we need it before people start creating patients, and it's like, oh, it's not good if. No, like very clearly no. Do we want it? Absolutely, right? Those are completely different things.
Do we want to help these patients more? Do we want ourselves or someone else to cure epilepsy? Absolutely. That's why we're doing what we do. We need to appreciate as well that this result we have, if replicated as we expect to replicate plus, is phenomenal to patients. Yes, the expectation is that it's longer, it's more time to respond, it's more time to get to like levels of concentrations that one would expect to be therapeutic, and patients just have less seizures. The median number, of course, gets a smaller number. Everyone gets happy and we move on. The ultimate goal of this drug is to be widely used. That is one step towards widely using, not the final step towards that.
Okay. Marcio, I would love to kind of think about POWER1. One of the things that on Monday was a lot of analysts, including myself, asked around timing. I stepped back and started doing the math in my head and saying, okay, let's say you do finish enrollment today, right? It's a 12-week study. By the time that your data could be done, we'll be sort of middle-ish of December, which then would mean the earliest, like would you want to put out data in the last two weeks of December? Probably not great for anybody. I guess, can I say that, is it fair to say that your hesitation to answer exactly pinpointing the timing of POWER1 is just that internal, you know what?
If everything goes perfect and it could be things are going to be done in mid-December, and that's also not a good time, so it could slip in Q1. Is that fair to say that is sort of what is keeping you to just say, yes, guys, the data's coming, I don't know, December X, Y, Z?
Yes, yeah.
Is that what's driving your hesitation and fear? Okay.
Absolutely.
I think you're going to announce enrollment completion, right? Everybody can do their math.
That is correct.
You feel very comfortable that very much the enrollment completion is very near term and you feel good about it. Nothing has changed. Yeah, we can.
Yes, since we're doing the math, let's do some math here as well. Yeah, so there were a little bit shy of 20 sites that were contributing to RADIANT. We see what happened, right? We gave them the opportunity to enroll more. They didn't enroll one or two. We're talking about having to tell people don't enroll more patients on RADIANT because I have too many. It's not a different problem with POWER. It's the same problem. The fact that I'm calling that a problem is so oxymoronic, right? Because it's really not an issue in terms of enrollments whatsoever. It is just a very long process for these patients. It's like four to eight weeks, the screening. Any patient that entered that period, it takes a very long time.
I think we learned a lot about how to screen these patients efficiently and the sites being phenomenal on helping with that, but it still takes several weeks. I think that's the uncertainty I would call versus fear on the final date of all of this.
Okay. Just a confirmation question from a client coming in. Can you confirm if you enrolled more than 50% of POWER1 patients?
That would be a material disclosure on this call, but I hope you can do the math there. I think it would be pretty obvious.
Okay. Got it. Next topic is going to be on, I think a lot of investors looked at the open label data and just argued the bear as well. It's open label. Patients knew they were in a drug, so maybe the treatment effect was exaggerated in some way. I think that if you do look at the Xenon OLE data, it really actually provides proof that open label treatment response versus being in a placebo-controlled study won't matter. Could you walk investors through that data? Because I think it's very telling in my view.
Yeah, I think you're right, right? It's a good recent example. I think that's why there are other examples in the past as well that show the same phenomenon, right? Maybe it is really because in our minds there's this embedded notion that people are just going to behave differently counting their seizures if they know versus they don't know, something like that, which there's no real evidence, right? I think when there is no evidence, people go into the universe of opinions, which is never productive in science. We do have like the X-TOLE extension. It's published now. It's been shown in multiple posters multiple times. I think it's pretty obvious, right? It's like you see the response, like you look into the switch on placebos. You look into the switch on the other dose arms.
The understanding of what happened, that is, of course, is very difficult to remove the survival bias when you get to one point. In a sense, there's an exaggeration there because of that survival bias, right? They have been exposed to something. The point I heard, and I haven't heard much of that, is you can't just move forward on anything in life and ignore that the baseline changes, right? The baseline is not the baseline at day zero, day one. Baseline is the baseline at day 56, at eight weeks.
When you look into that, there is no evidence whatsoever on that data set or in any other data sets, to be honest, that there is any difference between treating a patient in double blinds or on an open label fashion, even when there is full knowledge of the drug, as that was the case there in the trials being blinded on amplification. I think that's bogus. There are bare theses and there are bogus theses. This is a bogus thesis.
Okay. Marcio, what did you guys assume for the placebo response to be in the POWER1 study?
Yeah. Go back to when we don't have like a base rate ourselves. We need to go back and ask what happened to go to the same exact trials we discussed before, right? To define like effect or to define time to effect or whatever we want to define, you've got to ask. Trials that were run similarly, meaning both on the same pole, ideally contemporaneously, but there isn't really one fully contemporaneous to this, right? There is one that is the closest, and I would say the Neurocrine NaV1.6 is the closest of all of them from both the type of patients, the type of sites, and the overall design. You have, of course, Xenon and Cenobamate studies and so on and so forth.
When you look collectively on that, the numbers go from like a positive 5, meaning patients go worse on placebo, to like a negative 18. We conservatively think 20% is a reasonable number.
Okay. Marcio, when you looked at other studies, does the duration of extension matter? Like when you go from week eight to week 12, is there a deepening that occurs that could help us also think about the translation from RADIANT into POWER?
Yeah, that is very clearly, and of course it depends on the drug, right? There are drugs that tap out very quickly, but it's very clear it does get better. I think the phenomenon is, Steve, if he was in this call, he would say it's a world phenomenon. Less seizures lead to less seizures, and that's why we see this over time. If patients have benefits, they reduce. On trials that have no benefits, there is no change because, again, there's no benefits. In the case of having benefits, you do see a deepening. That's how we model. That's how we look into this. We conservatively modeled about 55% that's public, that is in the protocol for POWER, and about 20% placebo. That is more than plenty of POWER to be positive.
For all the reasons we just discussed, I think that is on both those metrics, there is a space to move on the proper direction.
Okay. When you look at the RADIANT and the POWER1 population and background meds, it seems like maybe your baseline seizures can end up the same, but is there a reason to believe that background therapies could be different between the two studies? Like how predictable is the two studies to one another, I guess?
Incredibly similar. The vast, vast majority of patients on RADIANT would be enrolled on POWER1 if that was the choice that was put in front of them. There was no reason to believe that those patients would be any different.
Okay. That's helpful. That helps. Are there, and there is significant overlap between the sites between the two studies as well?
Yeah, there is a good overlap. I would say one of the things we learn, and I think it's kind of obvious in a sense, in retrospect, everything in life is obvious. That like bigger sites, they just have a bigger machine to recruit and to screen. From a quality and speed perspective, it's just easier in general, bigger meaning like bigger available pool, not necessarily size. These quote unquote smaller sites, they just wouldn't be able to contribute to POWER1. We made a very deliberate choice to focus on some of those bigger sites. The overlap is not perfect, but it's very high in the U.S. The other decision we made is there was really not a lot more patients in Australia that would be able to participate.
Australia made a good contribution to RADIANT, but it was just not available in a lot more what I'm going to call hyper-refractory patients there as well that are on the sites. Not having Australia, it's in Europe, United States, Canada as well. I would say Europe, North America, and South America is the overall makeup for POWER1. We expect to be identical. If anything, virtually every site on POWER1 right now wants and likely will be selected to continue POWER2. It's going to be a very nice overlap as well.
Marcio, in the POWER1, you have 20 mg dose group and 30 mg dose group. The powering that you refer to, the 50% versus 20% placebo, was that for the assumption for the 30 or on a blended basis for both of the dose groups? What is the assumption based on the highest dose? Should we expect, since PK is linear, that we could actually therefore expect the dose response between 20 versus 30?
Yeah, so it's actually 20 and 30, right? Every patient's going through both of the doses. The effect is expected throughout the study on the 12 weeks, six weeks at 20, six weeks at 30. 20 doses maybe already got there, but we fully expect that to be fairly effective. We're not really expecting a major difference there. Obviously taking home with the last six weeks with 30.
We're going to.
Okay.
Since we're on that, and now that PK is like, what would you expect then the 40 mg dose group to do?
You can do this one of many ways, right? You can do this the most caveat way and add relatively small benefits there. There's no small benefit here. You add 5% to 10% more. It's massive. I think you can just expect that it's going to be somewhat more linear. Obviously, we are on numbers like the ones you're quoting at the beginning of the call. I think one way or another, we do have an expectation that this would create even bigger chance for higher benefits. I cannot hear you. Yes.
They're saying the question is for there is not to be a dose response between 20 and 30. I guess they're wondering why is there not, why is there a dose response between 30 and 40?
Yeah, no, it's like, and I don't believe that that's what I said, right? What I said is that is a multiple components of the response here. One is the dose you start, the concentration you get, and the other is your time you stay on that dose and on the following dose. What I said was we do expect very good response at 20. When I go back to the PK, when I go back to the exposures we had, that would be equivalent to 20 because there's always a range. Those were very good responses. You bring them in to not only maximize that response for six weeks, but also go for six more weeks at full on. Those are not like differently from baseline extensions. You don't have to reset at the week six, right?
They are already at a given level that is pretty high, that is much higher than a week one. You are adding something to an even higher level. That's where those models take into consideration.
Marcio, remind me how are you calculating seizure reduction in POWER1, like the time interval, and then also the seizure-free, and how does that compare to other phase IIb/phase III studies?
Yeah, so the two different methods, there are two different types of answers for the two parts of the question. The way seizure reduction is calculated tends to be fairly similar across studies. It's basically a rate per day multiplied by 28. That's what it really is. We talk per 28 days, but it's really per day times 28. Because most patients go to 28 days and they have very high compliance, it ends up being the same number, but that's how it's done. That is in a study with 12 weeks. It's the entire 12 weeks, but you express as 28 days. That's how we expect to show the number here, non-controversial. When you look into what one could call seizure-free, it really depends on a number of factors, right? You look into virtually every study that is being published in epilepsy.
It is when did we expect to reach steady states for this drug, and then we can count from there, or we can count from after this first disturbance of the system happens in the first few weeks. You look into Cenobamate, for example, which a lot of people quote as a very high number of seizure-free from them. The entire first six weeks, right? There was an 18-week study. The titration can no longer be the one that was done there. As you know, that titration is not in the label in the U.S. or elsewhere, but originally was. Six weeks, the data was completely removed, like literally completely removed. All the response, not only the 100% that would be seizure-free, as you call, but all the other response were calculated based on the last 12 weeks. I think that's actually fair for that drug.
You see that on other studies on clinical trials of AVV as well or the publication in general. I think the question we asked ourselves is there are kind of three major components here, right, for seizure-freedom from RADIANT, and we're going to be looking the same for POWER1. One is at the time these patients got to stabilize on the drug, like around two weeks to three weeks here, what happens to their seizures moving forward? That rate we showed for RADIANT is like 22% or so on the last four weeks. The second question is, was that only on the last four weeks? That is very important. We go back and it was not. The rates of seizure resolution or 100%, whatever you want to call, was double digit throughout RADIANT.
Now we really believe that you got to get the data out for the most important for the ultimate user of the data, that is the physician, and that last four weeks are quite important, right? Obviously the entire study as well. If I pick the entire study, the first period, the second period, and by period is four weeks, they're all double digits. They're all more than two times what is the rate on other studies that were done as. They're incredible if you want to ask my opinion about that. For POWER1, we expect the same. The last six weeks is a very important time to measure that, but the entire study is an important time to measure that as well, right? The important part is to give a very good benefit for those patients, and we have no doubts that that's the case here.
Okay. Marcio, what do you fast forward? We see the POWER1 result. What do you need to see from POWER1 in terms of dose selection and patient criteria to kind of fine-tune your plans for POWER3 that you unveiled on Monday?
Yeah, no, POWER3, it was, as we talked to a lot of people, so I'm going to concentrate that if that's okay, because we haven't talked about POWER3. You're like, okay, this drug is very competitive on this market that is, well, the rotating 30, right? It's the people who are always switching drugs. Is there space for everyone in the market? You could have 10 drugs, they all do well. They all would be billion-dollar drugs for that population. I think we checked the box and we're like, okay, this can do really well here. When you look into the overall safety, for example, in like 20, 25% less patients are having any side effects when you have the overall efficacy on the potential to go to a higher dose, potential higher response. Other things, right?
The first question I get from physicians is none that we're discussing today. It's like, is this teratogenic? There are multiple drugs that are, and absolutely not. Does this have any meaningful drug-drug interactions? The answer is not, because that's what a lot of them care about. We know epilepsy primarily affects females, right? A lot of reproductive age. Does it interfere meaningfully or at all with oral contraceptive medications? The answer is not. When you lined all of those things, I was like, and when you give this drug, like in combination, it's fairly manageable side effects, it's a good response, and likely going to be even better if you reduce the numbers and so on. The place for this drug is to be earlier and earlier in the treatments. The ultimate place, not the only place, right?
We're going from single digits billion dollars to potentially even double as we go to this market. That's what POWER3 is aspiring to show, right? What we're planning there is like there has not been a drug in modern history of epilepsy treatments that had the confidence to do that because no one so far had that entire profile that allowed it to do. What you're looking to POWER3 is, of course, we're still finalizing the design and there's much water to go under that bridge, but it's really to go with what we would expect, for example, 40 milligrams. From that point on, see the response. We expect high response. Because we are switched to monotherapy, really not a lot of liability associated with that. That would be quite complimentary. We don't need that.
We don't intend to have that done at the time that we potentially would be filing an NDA, but that would be a very nice compliment to increase overall perception in the market, number one, but then shortly thereafter to file as an SNDA.
Okay. A couple of questions on sort of the RADIANT data that's coming in 4Q, right? The generalized and maybe the additional patients. I guess based on just the timing, we would expect the additional RADIANT data is in 4Q, but probably ahead of POWER1, right? That's fair to say.
I think yes, yeah.
Okay. Second question is people are just trying to do the math. They're wondering how many, it sounded like when we were on that topic at the beginning of the call, you said few patients have now finished week three and four. To be cut, I guess what people are trying to figure out is how many additional, you know, focal onset could we get. We could get a few generalized as well. I don't know.
The six to one total minus 25 that already showed the data, they're all focal.
Okay. Yeah.
They are all well underway in this study.
Yeah.
The vast majority of them are towards the end of the study, are going to be finishing relatively soon. We're very good there. There are a few patients that were done with screening, having those yet. We didn't count that on the 61, both focal and generalized. Forward to guess right now, we're going to have a little bit higher than 61 on focal and then all the generalized. We didn't tell you how many generalized we had.
Yeah.
You are going to have that generalized cohort as well that you haven't seen anything about.
Yeah.
The premise for generalized epilepsy is that the drug would not work.
Yeah.
We're talking about a premise here that the drug would work, right?
Yeah.
Common sense would say, shouldn't expect much from this, like that is not other drugs one could call similar. They're very dissimilar, but no, what I'm telling you right now is so far that's not the case.
Yeah.
We're actually pretty positive about the impact on generalized epilepsy. You're going to see by the end of the year whether or not that's confirmed.
Got it. Just to make sure, it's like you're not enrolling more patients beyond those 61 on focal. You're done with focal. Anybody who's coming into focal is going to obviously go to POWER1. There's no reason for more RADIANT. The nice part is 61 minus 37, that gives you 24 patients that haven't finished the total treatment duration for which we could potentially see data. Then another 14, right? 75 minus 14. So 14 patients in generalized that we could get.
Might have one or two more in focal that were in this mix.
Yeah.
That haven't finished screening.
Yeah.
Disclosed, but it's not going to be anything materially different.
Okay, you haven't disclosed how many of those 24 that's in the queue are on background, Cenobamate, or any other background therapy that are being covered.
Yeah, the rate was fairly consistent throughout.
Yeah.
We did talk about the 30% on the first patients. The second quarter is smaller. The rate is similar. It's a little bit numerically, a little bit higher as a percent, but it's in the 30s as well.
Okay. That's helpful. I just want to make sure there's a lot of questions coming in, because I have a long list of mine. Also, since we love discontinuation there, the question is asking, does that mean the discontinuation rate should not change in the 14 out of 61 and 6 out of 61 dose-reduced focal patients?
Yeah, that's absolutely the intent, and that's what we are seeing so far.
Okay. Good. That's helpful. I guess if I had to like think about sort of what we learned today that I think didn't get appreciated was like the checklist of we're going to get potentially 24 more patient data, 14 generalized patients later in the year. We should actually, since we're on that topic, what is your thought on as you add more patients in focal in the new generalized? How do you think, what do you want to message to the street? What do we should see in 4Q, both in both populations?
One more patient is always refine the estimates, right?
Yeah.
It's going to be more precise at the end. Of course, we monitor these patients from the exact, exact. If these patients are behaving any differently than the first cohort of patients, I would say it.
Yeah.
That is not the case. I think we should expect something very similar to what we're seeing here. Of course, these patients all have to finish, and until they finish, they haven't. No formal statement about what's going to happen, more like a forward-looking in terms of what do we expect. Nothing problematic. The general background and the general seizure counts, they're very, very similar to the first.
Okay. Marcio, now I'm going to run a checklist by you really quickly based on what I learned from this call. If I missed something, you add it. If I misspecified, you also explain. Okay. All right. Let's start. RADIANT, right? We're going to get more data, potentially 24 patients and 14 generalized. It's going to be a similar treatment response you expect, but no surprises on 4Q. We also learned timing of POWER1 could be pushed into early 1Q because of the dynamics of just in case getting the data ready in December. That's why you're being quite quiet around it and not saying, yes, it's on track for December XYZ because of those challenges.
Third point is we learned that based on your PK analysis, that the PK is very linear, both across AUC and Cmax, which therefore should make you very comfortable around deepening of a response when you go to the 40 mg dose group in your POWER2 study. You also talked about that your coefficient of variability is also very consistent. You also noted that your exposures have no effect on efficacy and no effect on discontinuation. You also talked about, yeah.
That is not correct. Yeah, no effect on discontinuation or side effects, but an effect.
Yeah. On efficacy, that's right. That's what I meant. Yes. On discontinuation, you noted most of them occurred between week one and week three. You also talked about the imputation of those into your efficacy analysis. You also talked about on a blinded basis of POWER1, which you assume 20% to be the case, you don't see it. You see it to be, you said something along the lines, substantially lower. I heard that. That's good. You talked about POWER1 being powered for 50% and 20% placebo responses. We talked about the Xenon data that if you look into that, and anybody who's on this call wants to send that to me, feel free, or needs the publication, I'll send it to you because I think it's a very important analysis to figure out the chain that it doesn't inflate the treatment response, right? That's another takeaway.
We talked about Cenobamate patients were highly treated. 30% that were treated were in the 300 mg dose group, which is a very high dose, which most physicians, based on my channel checks, do not feel comfortable. That tells you these are really well-experienced docs. On discontinuation, under your breath, you said something that some of those are not run by few centers rather than generally seen. I feel like we tackled a lot. Was there something we missed?
I do believe we tackled all of that in our work.
Yes. Okay. All right. That was super helpful. I think we're very much looking forward to what is coming at the scientific conference here. What are we going to see?
are a lot more details about all these questions you asked. I think what is super interesting as well, I'm very interested on is there are different seizure types.
Yeah.
I think when we look into the different seizure types, it tells a further story about the drug, right? We're going to be showing that data as well. I think we are all very excited. The presenting investigator is very excited about it. I'm not sure if I'm going to see everyone in Portugal next week. I hope so. I'm going to be there, so please join me. If not, we'll share the posters we always put on the resource page of our websites.
Great. Marcio, I do want to congratulate you on the data. I think a lot of the supporters of the company see the data. I think this fireside chat was helpful to really dig into some of these topics. Again, thank you. I know you're extremely busy, and thank you for all of our audience. Obviously, many of you are buried in earnings today. Thank you for spending time with us.
Thank you so much, Yaz. I appreciate the invite.
Thank you so much.
A lot of fun.
Thanks.