Praxis Precision Medicines, Inc. (PRAX) Earnings Call Transcript & Summary

Good morning, everyone, and thanks for being with us here today in the office in Boston. I really appreciate it and everyone else in the webcast as well for our DEE portfolio update at Praxis. As you can see here at the beginning, the way we tend to think about everything at Praxis is daring for more. So we're there for more today as well. I think it's a great show here in not so sunny day in Boston and hopefully going to make things a little bit brighter and sunnier during the conversation. Steve and I are going to be doing most of the prepared remarks conversation, but we have other members of the team here with us as well. Tim Kelly, our CFO; and Brian Spar, our Head of the DEE program and many other members of Praxis team for going out afterwards. The way we thought about the event is really mostly a conversation about what we are doing, what the unmet need is and the commercial opportunity across the portfolio. So we're going to be structured that way. And we decided to deserve plenty of space at the end for conversation and Q&A and discussions and how we see all of those programs. We're going to be making certain forward-looking statements. So I strongly advise and recommend that you read the SEC filings, including our Form 10-Q that was posted this morning on the EDGAR system. Before we jump into the content, we thought it was very important to remind all of us here and elsewhere why we are doing this, right? It's not for us. It's for people that are living with those conditions and that needs good treatments, better treatments, safe treatments and that would impact themselves and their families and their loved ones and society at large. At Praxis, our mission is to do that multiple times and to keep doing that again and again until someone does better than us, and that's a great thing, and then we're going to do again and again better as well. So the focus is fundamentally on making CNS drug development and hopefully, very soon as well commercial opportunities here. As I mentioned, Steve and I are going to be leading this conversation. I'm going to go a little bit about the opportunity in DEE, what are DEEs in general. I'll cover some of the commercial opportunities for our molecules. And then we're going to jump into the relutrigine program, particularly how we are -- what we are seeing on the long-term extension with EMBOLD. So it's a little bit of new data there and updates. And Steve is going to talk a little bit about our EMERALD program as well in the rationale. And updates on the current stage of development for elsunersen or used to be called PRAX-222 in SCN2A as well. And what is probably the newest part of today is really our first foray into autism spectrum disorders with the PRAX-100 program. You might have heard about that before. But now we are at the point that we're going to be giving a little bit more in terms of the plans in general for us. Probably it is obvious why neurology, right? Neurology is an incredibly large therapeutic area. But maybe one of the biggest uniqueness of neurology is this white space that exists. And the white space differently from other therapeutic areas is mostly on areas that are fairly large. When you look into multiple indications, and of course, it's by no means like exhaustive in terms of all the indications in neurology, it's rare to see any therapeutic area where some of the largest indications are still the largest unmet need. So it makes it for an incredibly attractive market in general. It makes it even more exciting for us as a company because that's by chance and by design, where we focus on. So you can see here the focus, of course, as you know in epilepsy and autism spectrum disorders. Of course, it's still in essential tremor with the studies going on. And then adjacent to that, as we hinted last year, applications of some of those technologies in pain, in Parkinson's disease and so on. So not only the current areas of focus, but the potential future area of focus are quite interesting from a commercial standpoint. We're going to go back to the entire portfolio at the end. You're fairly familiar with all of that with our platforms, Cerebrum small molecule platform and with Solidus, our antisense oligonucleotide platform. But today, we're going to be focusing on relutrigine primarily and on the ASO platform as well. So maybe before we jump much further ahead, there's been a lot of discussion. We're incredibly happy with the fact that DEEs in general are like a topic of multiple conversations and a lot of interest in the industry, a lot more academic work, drug development, FDA, you name it, work here. There still seems to be more of an intent to divide than deunify DEEs when some of those conversations happen. So we thought it would be a good idea to bring them back to what they are, right? It's actually pretty simple. They -- it's in the name. They have an epileptic condition with development. Delays also happens at infancy. That is really the hallmarks, right? Infancy may be in a liberal way can go a little bit more sometimes into the teenage years, but really not beyond that. So that is the universe. And when you look into this universe, it's manifested by incredibly severe. I know all of you here in the room covers incredible number of diseases that are very devastating, pediatric, particularly. When you look very closely to these, these are amongst the most devastated conditions anyone can present, incredibly diverse, life-limiting, not only an impact from a comorbidities perspective, but reducing life quite significantly. Intractable seizures in the majority of the patients, not all, but in the majority of them. But maybe the saddest part so far that hopefully going to change, and you're all going to agree with us by the end of this conversation today is that more and more effective and safe drugs are being developed. And that is a huge change. Until not that many years ago, there was absolutely nothing. And now there is a small timid, but growing number of treatments designed for DEE. From our approach perspective, we're looking into 2 major ways to address the issue here. So one, is a very focused genetic-driven, like a one-to-one relationship with the ASO platform. So there's a genetic defect, and we're addressing that with one of the genes -- one of the ASOs to the genes as Steve is going to discuss later today. And another one that is really looking into the manifestation of the etiology that can be genetic or nongenetic, as you all know, can be familiar or nonfamiliar in most cases. And it's incredibly exciting because you can really tackle like both sides or bit together or separately, as you're going to discuss in a minute as well. No matter how we cut it, right, the number of patients with DEE today in the United States never ends up being lower than 200,000. And I think that's quite fascinating because, again, there is not a lot of other conditions with this relatively large unmet need and with just a very small part of them with current treatments in general. But quite importantly as well, because the majority of the conditions are either environmental or de novo mutations in certain genes, the overall reduction in rates that we see in other conditions like this does not apply here. And when you look into very conservative estimates on the next 10 years as a planning period, we're looking for very, very significant growth, like 50% growth on the patient population in general. When you compound that, with the fact these patients should be living longer with better care, better treatments, should be identified earlier and therefore, have a chance of intervention earlier as well, it becomes even more interesting in general. In a few minutes, Steve is going to discuss our EMERALD program is one of the key objectives today that we discussed where we landed at. But one of the most exciting things for us is that EMERALD can really address all this like pie, right, the entire population. Now patients are not seizing. Of course, they're not going to be eligible. They're not necessarily the targets. But when you think about a large proportion of the overall market is where relutrigine is going to be focusing on. To give like a ballpark of what a commercial opportunity would be, if you use what I would call modest assumptions here, right? If you look into not necessarily high price points and even relatively small as I'm sure you're going to back-calculate market penetration rates here, we're talking about a very, very significant market opportunity that can be derived from not only relutrigine, but others as well. I think the more than [ earlier ] as people develop other therapies to address this market as well. Well, might sound counterintuitive at first is when you look into elsunersen, for example, right, which is a -- it's a relatively small number, not small for rare, but relatively small number here, about 5,000 patients eligible globally and by global in countries with well-established rare disease mechanisms to get access to patients and using analogs that are, I would say, I'm sure you'd agree, fairly reasonable for this kind of size and condition, maybe even on the low end. The global opportunity here is about $1 billion and the U.S. only opportunity is about half of that. So quite interesting, feasible from a commercial perspective as well. The newest part to introduce today from a commercial opportunity is really PRAX-100. We've been working for a while on the understanding that a very unique gene, right? SCN2A called for a Nav1.2 sodium channel and has this very interesting manifestation that when you have a lot of expression gain-of-function, patients have uncontrollable seizures at birth, and we know a lot what happens there. We are addressing them to relutrigine and elsunersen. But on the opposite direction, there's not enough insufficient phenotype. There is a very profound and one of the largest, if not the largest cause of genetic autism, not all autism is genetic, as we all know, but there is a fairly large part that is. So using conservative numbers once again, we're talking about at least 20,000 patients eligible and 0 drugs in development. We know increasing the expression of a gene is significantly harder than reducing the expression using those technologies. And I think you're going to be quite impressed with the work we're able to do here. So this is the next frontier on this portfolio as well, and we're going to remain quite focused on delivering on all this opportunity in general. So in the short term, if I bring us back to what's coming on the next 12 months or so. For relutrigine, we are fast and furious about in both cohort 2. We had some quite, in my humble view, phenomenal results last year, I think we're going to be impressed with the update we're giving today as well. We right away switched to cohort 2 here. It's enrolling incredibly well. And as you can see in our press release today in our disclosures, we are guiding for no later than the first half of next year for the readout. We are starting, as we speak, the EMERALD program, I'm going to discuss a little bit the design there and how we landed. And that would allow for, if all goes well and if it's all approved for an expansion of this market from the 2A and 8A into this broad 200,000, give or take total number. When you move to elsunersen, the EMBRAVE program, as Steve is going to describe as well, has a number of parts. We learned different things from the very first one. We're learning other things from Part A that is enrolling right now. And we are starting as well imminently EMBRAVE3, that is the registrational program for the U.S. So a lot going on. So with that in mind and with this, what I would call quite incredible opportunity in front of us, I'm going to hand over to Steve to talk a little bit about the details on some of those programs.

Thanks, Marcio. So what I'll do is talk about a little bit of the science and the development for relutrigine and then elsunersen and then touch on our further pipeline for the Solidus or ASO program and highlighting the PRAX-100. So start from a sort of a fairly theoretical point of view. And when you're faced with the challenge of how do you make a drug that can give you as much efficacy as you can possibly extract without bumping into tolerability. And that's always the refrain you hear with current antiepileptic drugs is that well, we're always limited by toxicity, and we're leaving a lot of efficacy on the table. Now we're a precision medicine company and precision takes many different forms, whether it's genetic, whether it's precisely how you understand the neurophysiology and how you think about delivering medicines. In this example, what I'm trying to show is that in the blue line, there is what healthy neuron. This is how you understand a neuron, you stimulate it, which is the bottom axis, and it gives a response. So input, output, stimulus-response, however, you want to look at it. Generally, you follow that curve for the healthy neuron. In epilepsy, what happens is that neuron for the same stimulation that normally underlies normal function that we need to operate every day pushes into that epileptic range. And that's something that all drugs are trying to stop from happening because a neuron has to fire and that underlies all the paroxysms and the seizure manifestation. An ideal drug would push that back down onto the normal curve. And what doesn't happen, though, most drugs today or all drugs to do this push to stop you getting into that epileptic zone, but also impair function at the lower side of that curve. And that's where all the normal things that we do has to happen, the cognition, the movement, speaking, everything happens down in that zone. And if you're impairing that, you're going to have tolerability concerns. The ideal drug would just take that red curve and pop it straight back onto the healthy neuron curve. And then if you're trying to do that, how do you think and what are the levers that we've got for developing a drug against the sodium channel to enable that? And if we go to the -- if we look how we look at relutrigine on the left compared to mainstays of antiseizure medicine treatment, carbamazepine and lamotrigine. I'll just walk you through, there are 4 properties there shown with different symbols. The diamond is that property called tonic current, which is the standard operating mode that a sodium channel supports an action potential in a neuron that then supports normal bodily function. Totally critical to let that happen, leave that alone. If you start to interfere with the function of that, you start to get some tolerability concerns. The other currents, persistent current, voltage-dependent current, use-dependent shown in the black, the purple and the square symbols, these are all the currents that sodium channels can support to make a neuron fire into high frequencies. And that's what happens in [ epilepsy ]. Those currents are what support that epileptic firing. So the challenge for a drug is to -- when a neuron starts to do that, to selectively impair those, so they don't get out of control, but leave the tonic current alone. And that slide for [indiscernible] that shows the pharmacological separation we've been able to achieve between the tonic current and that triumvirate of persistent voltage and use-dependent current. No other drug on the market is able to separate those 3 from tonic to this extent. And this is what -- this is the profile that you would need to support the previous graph that I showed you where you block activity at the high level, you spare it at the low level. It's exactly got the right pharmacological profile. And this is the work of a lot of computation on neurobiology, a lot of experimentation that we have done and the field has done over the years is bringing that together. And it really is the next step. You've probably all heard about use-dependent and state-dependent channel blockers. That's from 20 years ago, but this really is the next evolution of that to bring it to this level. What happens? What are the advantages if you achieve that with a mechanism? One really important advantage, and Marcio alluded to this, that network excitability, hyperexcitability being at the core of all epilepsies, whether they're a DEE or anything. And that's because everything has the funnel in the end through a sodium channel mechanism in order for a neuron to fire an action potential. Unequivocal, you cannot fire an action potential without a sodium channel. Everyone that's the fundamental property of neurophysiology. What that lets us take advantage of, irrespective of the upstream etiology, you now have a choke point, all that etiology has to flow through in order to manifest as hyperexcitability, it affects the neuron, the channel, affects the neuron and then affects the network.

Just a quick clarification question. You mentioned -- I believe the word you used triumvirate of sodium channel. It was tonic, it was excitability...

The ones on the bottom left, the pathological excitability currents, persistent-voltage and use-dependent if you draw a circle around that, that triumvirate separate from the tonic exactly.

And just to clarify, did you mention that PRAX-100 was also a sodium channel drug?

We'll get there.

Opposite direction.

There's some nuances to that, that we can explain when we get there. So as I was mentioning, irrespective of the upstream etiology or the upstream seizure driver, however you want to look at it, you pass through this choke point of sodium channels. So if you can modulate the sodium channels, you're going to modulate the output of the neuron, you're going to modulate the emergence of pathological firing. And again, relutrigine is perfectly positioned to serve that role as a modulator of that choke point. Now that gives us the opportunity to say, well, we should be broad now because we've gone from a computational understanding to a pharmacological and biophysical profile for a drug that we think is going to meet that computational and theoretical understanding. And now this is where we first test it. What happens when you put relutrigine on animal models that are harboring different etiologies. On the left, it's the obvious stuff. It's a sodium channel etiology. This is where you would expect it to work in its best state because the sodium channels themselves are mutated and are not just driving the seizures, but they're the source of the pathology. And that's why you will see our first study EMBOLD asked that question of patients that have got gain-of-function mutations in 2A and 8A, what happens when you give them relutrigine? And we saw the remarkable data come out from that last year, and we'll give you an update from the extension today showing you how that's panned out. But now the next step was what happens in the potassium channel. This is no longer the same mechanism. It's not a broken sodium channel. It's a broken potassium channel that's now making that neuron fire higher. The sodium channel is participating in this process because that's what it does. There's an incredible synergy between activity in the neuron and what a sodium channel does. But remarkably and consistent with everything we've been saying, you give relutrigine, you completely abolish the seizures in a potassium channel etiology. That was the real first assay test that is this going to work with different mechanisms. And if you look at where else we've taken this, and this is really a little bit of something that's a bit of a disruption, I would say. You probably heard a million times from all the KOLs you've spoken to and things that you've read, you don't give a sodium channel blocker to a person with Dravet because the disorder is already due to an insufficiency of sodium channel function in the interneurons. Interneurons are the brakes, the excitatory neurons are the accelerator in the brain. Not enough braking. And that's why so why would you do something that impairs the braking further? And there are cases where you do that with some of the sodium channel molecules that are on the market, and you can see exacerbation. This slide shows that in the zebrafish model that if you give relutrigine, you see essentially a complete reduction in seizure activity all the way to 0 and at a very, very potent 0.04 micromolar EC50. And as I did mention in the previous slide, relutrigine is orders of magnitude more potent against sodium channels than anything else in the market. That's great because all the off-target stuff is also going to be limited because of that, very, very potent. And now if you compare it to the molecules that are approved for Dravet, fenfluramine, most recently, it's as efficacious but more potent than fenfluramine. You compare it to bexicaserin, which is the upcoming molecule that's now with Lundbeck, hugely more potent. And in this model, it couldn't even achieve full efficacy. So it really, really compares well with molecules that are out there for Dravet. And this is a bit of a surprise. And we think, again, it's another manifestation of the biophysical properties of relutrigine because an inter neuron that's not firing hard enough is not going to be subject to interaction by relutrigine because it's not the sort of state of activation of a sodium channel that's going to favor finding and inhibition. And we think it's a remarkable -- another remarkable tick against what we expect for the profile to look like from an in vivo perspective.

The results here in the previous slide, quite mind-blowing because a lot of KOLs that we spoke at AES said that sodium channel inhibitors are contraindicated for Dravet because it's a disease of half insufficiency. You don't want to further reduce the activity of sodium channel, but you're showing efficacy here, again, mind-blowing. But in this model, what happens if you apply a standard regular sodium channel inhibitor, does it get worse or better?

Well, we can we can show some of that data. Look, in many cases, you see exacerbation in Dravet models...

Then why it is contradicted...

Which drives the contraindicated. So that has been shown in the literature. So that's something that -- so the model doesn't just show benefit. You can see exacerbation in some of these models, and we're happy to try and dig some data up to talk about that.

So does that particular sodium channel in the interneurons not have a persistent current?

It doesn't have a persistent current. The normal persistent current is driven by pathologies that cause gain-of-function and it's a property you see a normal physiological property you see in the dendritic tree of normal sodium channels also have that, but not 1A, doesn't have a big persistent current. And then that's why -- and also then it's not available because it's not that active in those neurons and not available to interact with this drug as well. But this is perhaps the most remarkable thing when you look against 10 different disease models. And some of this is from the literature and some of this is work that we've commissioned. And what you can see, if you look at Dravet first, there was the fish data that we showed you, scn1Lab (fish), the data we just showed you, the two molecule, very similar pharmacology to Dravet was also applied in the mouse model of Dravet and that independent validation of that efficacy, unexpected efficacy in Dravet, again, telling you about this biophysical profile of how it works. 2As, 8A I showed you, but a whole range of potassium channel etiologies, full efficacy. A hyperpolarization-activated [ cation ] channel HCN, also efficacy with this model because that whole idea of the choke point being the sodium channel, has to manifest through that in order to have a seizure. Well, we're modulating that behavior. None of these things can get through that choke point when relutrigine is around. So...

It might be interesting to say, right, because HCN1, the very last model here, we cannot find any other drug that has this property. It's not like -- we know there are approved drugs for Dravet. We know that other mechanisms work for Dravet. We know there are ways to modulate potassium channels, right, and certain potassium channels. But there is no way to modulate that pathology. We used to have a table that has like multiple ways that is the one that is red, meaning proconvulsant in any other molecule. So it really reinforces proof that things are going to work across every possible model, but it definitely maybe is one of the most compelling case we have to really understand how this drug can benefit patients globally.

So what's the profile of the relutrigine modulation that hits -- that prevents, I guess, this is the interneuron activation that's...

Interneuron inhibition. Further inhibition in Dravet.

Not for red eye, forgive me. What aspect of that modulation allows for that when -- that the other compounds don't have?

Yes. So most compounds just bind the sodium channel and block it. They don't care what the sodium -- what state it's in. They don't care that it's in a resting state or an active state, they interact with it. And because they do that, essentially, that sodium channel is out of action all the time. But if you can stop that from happening, only interact with the sodium channel when it's in that pathological state -- when it's supporting hyperexcitability, right?

And there is no other drug that you believe modulates only the hyperexcitability state?

There's some that will modulate a little bit of persistent current, but none of them that take all of those 3. By taking all 3, you're blocking all of the drivers of pathology.

And they are anatomically not on the same place in the brain, right? So I think that's on the neurons. I think that's important as well, maybe a little disservice that we call them like 1A, 2A -- like it feels like they are anatomically on the same place, but they're not, right? So maybe you can talk...

2A is only an excitatory neurons and 1A is only in inhibitory neurons. And that's why Dravet is a disease of interneurons and 2A is a disease of pyramidal neurons. So they do anatomically separate and drive different behaviors in those. Our drug, relutrigine will interact with all of them under the right circumstances, but it's the circumstances, it's the manner in which you interact that matters so much and that differentiates us.

So maybe what was sort of the pharmacology with which you were able to achieve that? Maybe give us some color on that. And then you showed us a lot of different models, and I understand some of it is literature, some of it is work that you've done. But is the sort of impact the same across the different models that you're seeing, the effect and kind of the safety?

So the safety profile is the safety profile. And so when you give the same concentration of the drug to the different models, there isn't a different safety profile that emerges from that. And the inhibition is universal across all of these that you can either stop or have a massive effect on the seizures in these animals. The pharmacology, when you do discovery on sodium channels, you can separate those tonic current, the use-dependent, the persistent, the voltage-dependent from a protocol perspective, when you're manipulating the channels in a cell, so you can isolate them and then you can look at the specific pharmacology against each of them individually and in a separate fashion. And then you can say, okay, we profile this drug, we see ones that hit these 3 and spare that. So it's very easy to do that. And there are multiple ways of achieving that with patch clamp electrophysiology in the assay.

Is it fair to say binding safe is what's unique...

Certain binding sites becomes available...

That's either a genetically determined mutation that causes that binding site to be more prevalent or the temporary state of the channel has that...

Exactly. The mutation might make the channel more active, therefore, more exposing that binding. So not necessarily at the binding site where the mutation is, but it makes that site more prevalent because the channel is doing more and it's making the whole neuron and the voltage changes. These channels are exquisitely sensitive to membrane voltage. And that's what causes them to change shape to open to let sodium in, but also to then expose novel binding sites that drugs like relutrigine can interact with. And then -- and so you can favor that binding site over the binding sites that are available when the channel is just sitting in minding its own business.

So it allosteric and that it's changing combination of the channel back to normal? Or is it just binding and enabling...

Binding and inhibiting.

I think it's your question, right, the binding dynamics are quite important as well. So that is a kon and off ratio that we believe to be ideal. Are we at ideal? We don't know, right? By definition, we'll have to test all possible combinations. But what we know right now is that binding very fast is important. And it's staying there for a while but not too long is important as well. And when you look -- I believe a few years back, we showed the profile, the binding profile of this molecule. People normally don't pay a lot of attention to binding profiles. That is incredibly important as well, right? So when that site is exposed, binding fast and inhibiting is maybe the reason why other drugs miss the mark and are not able to help here because it is incredibly fast that it binds to the sites, right?

[indiscernible] important at all here?

Say it again, sorry.

The koff, which is that you...

It's important yes. So the ratio is very...

That is your affinity.

Once the current is gone? Or is the off independent?

The off is independent of the -- the on is dependent on the amount of drug and the biting side, the off is independent of that, and it just falls off.

Itself, like once current is gone, the confirmation changes, does the drug come off or is the drug...

No, it comes off because the residing time on that binding site is governed by a statistical process and it just drops off [indiscernible] put mark of process, how it drops off.

Any comment at all on how relutrigine differ structurally and from a binding profile from lamotrigine?

We have not, but I will, Laura. When we first designed -- so relutrigine came first, right? They are from the same series. You're going to be able to find their structures right now. They're both published. So we might as well buy the -- they're incredibly similar. And to the point that I think most untrained eyes would say these molecules are going to do something very, very, very similar. And they do from certain properties, right? But maybe the oversimplification is those channels are incredibly sensitive to small changes on the -- what I'm going to call the electro fields around the molecule, right? I think there's no chemist who would call that something a little bit different. And just that one nitrogen change between those 2 molecules significantly change that interaction and the metabolism of these drugs. It's completely unexpected. The beauty of that, not to expand way beyond the scope of today's conversation, that there are many more of these molecules that we have patented. You can imagine certain molecules that are -- would be having a better either primary or secondary pharmacology for other neurological conditions that can use that property as well. Neurons are not made to fire and create seizures, right? We have all of them for other functions as well and modulating this fundamental process with molecules is definitely incredibly important. We know that sodium channel blockers are used for mood stabilization, for pain, for other indications that not necessarily today's discussion, but you can imagine how even understanding those small subtleties between the interactions can be important for future discussions and indications as well.

Maybe a related question. How would you modulate or how are they different? Is it the binding site that's different? Or is it sort of the kon/koff ratio?

And the clinical pharmacology is different as well. So the primary pharmacology is relatively similar, meaning the way by which they inhibit different occurrence, but not the amount of drug, the persistence of the drug, right? One of the concepts we discussed quite often is when you look into the half-life, for example, of each one of those molecules, they're incredibly different. There are molecules you want shorter half-life. There are molecules you want longer half-life. Once we discussed the profile for relutrigine, it was important to be that long, to be as long as we possibly could because there are tons of intercurrent events that happens with patients with DEEs, want to make sure that when they take the drug, it doesn't matter for the effect of the drug. Like we know, for example, there are cyclic vomiting in a lot of these patients, right? So independent of when they are taking the -- it goes through like a lot of vomiting 1 day, several days without, a lot again. No one knows exactly why it happens on some of the DEE patients actually very common manifestation. They get a lot of infections. So they might need to actually treat the infection before. And imagine if every time that happens, the drug concentration goes to 0, now the seizures go all the way to [indiscernible], it's terrible for these patients. It might be one of the issues why there is uncontrollable seizures with other medications. So we wanted to make sure that small chains don't affect fundamentally the ability of the drug to control the channel, and we're able to do that with relutrigine. It's less important to know important to lamotrigine. So I think when you're looking to these drugs, right, we spend a lot of time on the mechanism and whether or not they work, but then real people like us end up using those drugs, and we have realized with real complications and interference events and so on. So we try to match those drugs on the clinical pharmacology, not necessarily the primary pharmacology. The kon/koff is slightly different. I mean we're more than happy to post this afterwards as well for those molecules as well. I believe Yas has a question as well who is on Zoom or maybe not. We didn't design an interactive, but it works.

So sorry, I wish I was there with you in person. On this really amazing preclinical table that you guys put together, I assume -- I guess, have you also looked in the predictability of these broad DEE models the translation to human data? Like could one say that the translation of one preclinical model is stronger than another? And then maybe you probably will talk later about it, but I assume you dose in the preclinical model sort of uniformly. I guess the question becomes how do we think about dosing and figure out the optimal as you contemplate like in the preclinicals, I guess, do you use the same doses or different doses based on your modeling, and therefore, it could warrant different doses as you open it up into broader indications?

The fundamental question on the overall dose at the end, right, for all patients, I think we have to be incredibly honest. It's going to be an open question for a while, right? So we are taking the approach that we have a dose that is effective or we expect to be effective for all those patients, knowing full well that some of them might need more and some of them might need less. But when you look into the predictability to your first question of the exposures on those models and correcting for protein binding and properties that humans have from the animals, they are all within the therapeutic dose that we've been giving to kids and adults on the EMERALD study. So there is none of them require a lot less or a lot more. But once again, we are all different. Some of us have full response at full dose and some of us don't have for any approved drug. So it's not a question as much for the clinical development, what things should be standards, but a lot more for after market. I can tell you without giving much details that we receive a fair bit of emergency requests for this drug because of -- and some of them are on certain conditions that would not expect effects. So some of the things we're saying here is a little bit rhetorical because we actually know already has effects in some of those conditions in humans and that those are very similar to the ones we are using right now. And I'll stop there and I'll not entertain questions on that specific topic. And Doug?

Similar to Yas. Sorry, I can't be there. I'm just curious, in terms of this process of identifying unique binding sites and binding sites that are sort of available only when you want to interact with them, does this -- in terms of your process for identifying those, is that something sort of proprietary to the company? Or is it just sort of like an orthogonal method that -- in sort of knowing how and when to sort of put together somewhat sort of available information and work that's been done broadly in the community?

I'll let Steve comment on most of this, but maybe because he's not going to do himself because he's very humble man. The most complex interaction of neuron system in the world was developed at Steve's lab when he was a professor at University of Melbourne. And we leverage a lot of that work to understand maybe a fallacy of a lot of those models is we plan like one of the cells, as you know, when you look into the behavior on that given system. But our brain is not like that. It's not -- there's not a single neuron or a single part of a neuron firing that is a system. And what we don't know completely if that is the full secret sauce is definitely a lot of why we are here today. But Steve, go for it.

We just talk a little bit about it. It's a way of bringing a computer and a neuron together, computer and a cell together with the channel, like I said, channel in it, and you can produce a hybrid thing that in real time, you can investigate the way drugs and the behavior of the system works. So it's called dynamic-clamp, and it really lets us get incredible insights into how not just the neuron affects the sodium channel current, but how it would affect the behavior of an actual cell of a real neuron. And so how the drug affects actual hybrid neuron. So that really compresses the time it takes to understand mechanisms to have predictions around what might happen to a real neuron using pretty much the same method. You just throw a computer -- a real-time computer in the loop. And now you've got a virtual neuron to play where it normally would require a mouse model and slow data. So that method has been very helpful in exposing some of the fundamental properties of the system.

But I guess as a follow-up, Steve, does that help you identify the specific binding sites? Or is it just simply the sort of neuron functionality and the firing that you're...

Yes. So without going into too much detail, Doug, you don't necessarily have to know the binding site. You need to understand what state you're moving the channel, the neuron into and then look at the sensitivity of that state to the drug because that's what you really want to -- if you might know the binding site, it doesn't prove that you're going to be able to have the interaction you need. This is the downstream step, okay? We don't want -- we want channels that are being activated too much to be inhibited. So we can directly look at that functional consequence without -- then we can go back and deconvolve and say, well, what part of the channel is probably going to be exposed here and then start to look at where it's binding, but that's not a necessarily knowledge you need in order to first characterize and profile, the pharmacological and biophysical profile you want.

And then if I can ask just one follow-up. When you think about other indications, perhaps even those outside of epilepsy, would this also be a situation where you'd be looking at somewhat differentiated binding sites and sort of neuron firing than what you're seeing right now with -- in epilepsy models.

Nice try, Doug, we will be there soon.

Fair enough. I got to try.

Yes. So I can't -- I'm never going to punish you for try.

Maybe a very quick related question. So what you've patented is the clamping method, not the binding side or both?

No. The actual methods, you can go ahead and buy supercomputers and build the system and good luck to you. It's published. I'll actually say published a lot of that for all the years, not the final system. The ability to actually interpret this is so computation intensive that until recently was actually very difficult to completely integrate it. Now it's a little bit easier. There's much more computation. But knowing what to do with it for a while, this too shall pass as any understanding do, and we're going to figure out, others are going to figure out. But right now, the integration of how it works, how to derisk, how to derisk the tox later on, it's something that we have a significant amount of internal knowledge just by confidentiality and internal knowledge about how we design some of those drugs. And there's a lot of luck here as well. I mean if you know me, I'm not going to just say, just like a bunch of like people that can come up with anything, there is a lot of luck, right? The original idea for some of these molecules did not come through this, it came from like observing the environment, what happens. And then, oh, can we design something and we didn't know. And here we are.

What is the chance to rationally design state-specific activation inhibitors, inhibitors of hyperexcitable?

We would argue this is rational because we are precisely isolating that process and looking at its pharmacology. If you're saying rational from the perspective of from a protein perspective, with some of the new AI tools, I think that's in the future might hold that because we can look at -- we can model the binding of this drug docket into a model, do molecular dynamic simulations and ask it. I think that's in the future. Right now, I think that's things that we're looking at but not deploying it right now because we've got this ability to sort of not have to worry about the protein interactions because we can look at the functional interactions first.

Is there any structure of hyperexcitable sodium channel out there to be also docking...

There are structures of sodium channels on which you can impose a mutation and then say what would the confirmational change be and then use that. And we obviously -- this is a lot of way of our thinking. I mean we're obviously on top of all that literature on how to deploy. We will deploy at the right time.

We can make in silico predictions of the change, if that's the question. Yes.

You predict brain penetrants.

Through other methods, but yes. All right. I love the questions, but...

All right. So this is just a snapshot of where we're at with relutrigine. We talked about how well-positioned the in vitro, the biophysical, the in vivo properties are. And now we're going to start talking about what happens when you start to put this in people. And it's a bit of a highlight, and I'll talk in a little more detail on them. First thing is we've got robust seizure reduction in our first trial, the EMBOLD trial that read out last year and an incredible increase in seizure-free days, and that will be clear. Looking at the safety in our healthy volunteer studies, what was really remarkable with this -- and again -- another sign that this profile we've got is actually meaningful is that we couldn't find a maximum tolerated dose. And that's something that you can find with the sodium channel molecule quite easily. But we could dose these patients -- these healthy volunteers up, no MTD at fold levels higher than we would think is necessary for therapeutic coverage. DEEs were all mild to moderate. So it was a really good profile. Now if we look at the trial design, it was really -- I think these -- I think a lot of the trials that you look at, probably the best word to describe them without being too proud of what we've done is just elegant. They're very simple, but they're highly informative and they're very, very efficient in pulling the signal out from these patients. This was another design, obviously randomized to the placebo arm or the drug arm. The placebo arm was a very novel design where anywhere -- each of them is 4-week periods. Anywhere within one of those 4-week periods, you could have received a placebo period. And it was blinded to the certain investigators, was double complete, double blind on that when you got it, which we think contributed to the very favorable placebo rate that we saw here because no one had any idea. If you're on placebo for 3 or 4 months, you might guess it. But if you go, well, I could be on it, I may not be on it. You removed a big element of the human bias that influences, we think, some of these things. And then patients were able to go on to online open-label extension. Now this is perhaps the most remarkable result. You probably recall from last year where we talked about the top line data from the EMBOLD trial where -- and that -- these are the periods, baseline of 0 each month. At the end of the fourth period, the fourth month there, that was what we reported about a 47% reduction in seizures. This is now looking at the open-label extension. And there, we're assessing every 3 months. And what you can see is a continual reduction in seizures, absolutely remarkable. At 11 months now, we're seeing around a 90% reduction in seizures. Bearing in mind that these patients at baseline were already on the most number of drugs that the treating physician, they'd put them on. They've gone through all the trials and put them on this. And we put this drug, 70% of these patients were on sodium channel drugs. We put this on top of that and didn't see any new serious AE events emerge. So again, is that another sign that the profile of this drug only acting when you need it to. It's not inhibiting and contributing to the pathology that you'd expect if you put on a phenytoin or something.

Did you know that how many were on -- how many drugs they were on at baseline?

Yes. So there were an average of 3. There was one patient without any drug at baseline, incredibly unusual. Everyone else was in more. And that one -- we consider that one quite important as well. Why it's an [ N-of-1 ]? I think it's important because they had tried every drug available and they could not tolerate any [indiscernible] receiving a drug today and responding really well. So I think that it goes back to every piece of evidence helps us understanding the profile of this drug. It gave us enough confidence to design the next trial with even more liberties for these patients because one of the key, I would argue pushback we got from particularly families is the restriction of having one sodium channel blocker at baseline. So 70% of them were in one, as you can see here on the right-hand side, but we did not allow for stacking of multiples because I think it pays to be safe at the first time we are in pediatric patients, right? Most of these patients are not mobile. They have several seizures per day, as you might recall.

They were tried an average of 3, but they were only on...

They were on 3. The average is like for most patients, Brian, I was just reading some case actually this morning that came in on emergency requests and they tried, I think, 10 drugs. The majority of these patients to the point that they get to this trial, they tried everything. The other point that we got asked question before is not on the slide, but I think it's worth mentioning is, were they really at therapeutic levels? We measure therapeutic levels of every ASM during baseline. And no, they are not. They were at super therapeutic levels. They were actually higher at almost toxic therapeutic levels. So if something were to happen in this study, we can only talk about the study we are reporting. I think there was a high probability that if our hypothesis was wrong, we would have seen it. And quite the opposite, right? I think what you are seeing here is incredible seizure reduction that deepens the -- as you stay on the drug as it makes sense.

Why? Why should it deepen over such a long time period? Why shouldn't you see the maximum benefit within a month?

From a neuronal perspective, you've all heard about plasticity, homeostatic plasticity. The systems change and there's the immediate effect, but then having a lower firing rate has a longer-term effect and as the system completely resets its function. And you see that in many other states. So there is a strong acute effect, but continual exposure starts to actually really to a new set point that doesn't happen because you first put a drug on, the neuron wants to go back to where it was because that's what homeostatic plasticity does. I'm firing at this high rate, even though you pulled me down, I'm going to change other things. But eventually, everything comes down to a new set point. And we're seeing, I think, a really good manifestation of that in this data.

Yes. And this might help as well, I mean, on the next slide here. We showed another version of this before where we like show the baseline in the controlled periods. But when you look by months of exposure, what we are seeing is not only the seizures are reducing, but there is spacing, right? So you might still have one seizure, but now instead of every day or every 3 days maximum, it takes 2 months for one session to reemerge or 3 months or 4 months or 10 months, whatever it is. And it might be -- now we don't fully understand as humans, scientists why those seizures show up. So I think we all should be incredibly humble as well that we don't know the key for any of this, right? We can observe a phenomenon as we're observing. This phenomenon is quite positive, but we don't completely understand why it resolved. There is an increase in concentration as well of the drug. And there are for many patients, a reduction on their background therapy. And it might be -- it's not an unreasonable hypothesis, not one that we aim to study here, so I can tell you definitively that what was causing some of this was actually those nonspecific drugs as well. They might reduce to a level, but they might create other issues in the brain. So we just don't know, but it might be a synergistic combination here about reducing a little bit of toxic drug and increasing the concentration of one that is not.

So when you say increasing exposure, do you mean like the CSF in the brain or serum or both?

Yes, both. But the -- all the patients started at 0.5 milligram per kilogram per day in this study. There was an opportunity just once during this study to go up to 1 and in the open label, so after month 4, there was an opportunity to go up to 1 again if you hadn't, right? And what we can tell you is that at this data, the majority of them got to those exposures or patients have different clearance, right? We all have different clearance to different drugs. So some of them can get to 1 milligram exposures by staying at half. That's why we always like to talk about exposure versus dose because exposures have a lot of variability around them. And we do see a very strong correlation between exposures around 1 milligram or target 1 milligram and resolution of seizures and reduction of seizures. So it's a multisystem like multifactor system. So isolating one parameter is not going to be the solution here. So all of this might be contributed.

So we've noted out on technical reasons. But if you look at the clinical phenomenology, one of the earliest things that people learn in clinical epileptology is seizures beget seizures. It's a really old saying in the field that you start having seizures, you start having more seizures. And you look at the wording for the definition of a DEE, it's described where the seizures themselves contribute to the pathology. Now that doesn't happen overnight. That's a slow feedback that makes the seizures worse and worse. Now when you're unwinding that, that's going to be unwound on a similar time frame. So that's the phenomenology. I think Marcio has very elegantly told us about some of the underlying reasons why that might happen. But that's, I think, part of the response to your question.

One other aspect just to add to about relutrigine profile is there's no titration required. It's an oral pill or it can also be given to the G2, J2 once daily. So very, very easy relatively to some other drugs out there to give and the dose. I think the second part, too, and it will lead into the next slide also, it's not just the patients who see the benefit of these extended period without seizures. There's an entire caregiver ecosystem as well, who is supporting these patients and the children who also see the impact of fewer seizures, less severe seizures also.

Yes. Doug, I think you have a question.

Yes. Thank you, Marcio. So Steve, I think you sort of described the sort of this reset of sort of what is needed for sort of like the firing for an epilepsy. And I think in the literature, they sort of referred to like this kindling effect. I'm curious, so it sounds like you would characterize this as maybe like a dekindling effect that you see with relutrigine. And I'm curious, is this manifested -- does it change in terms of sort of the confirmation of the neurons?

When you dekindle, reset whatever the terminology ends up being for this phenomenon, you do change things and you change what things express in a neuron. You change the amount of potassium channels and other things that express that homeostatic response that keeps neurons firing at one, right, is because of that. And so now if you change it, you cannot to reduce. The things that express, that's how you change the set point, Doug. It's not necessarily -- it could be structural parameters, the amount of dendrites and the size of the field, other things can change. And this is a whole area of investigation, and we won't nerd out on the science, but it is -- there are -- we do know how this can happen.

And I guess as a quick follow-up, though, that as you change the set point, would that imply to get further gains that perhaps you might need a different drug with a different binding site and availability?

As most disease, there is no silver bullets. I think it would serve us really well when you are like at about 90% or whatever, 90% reduction should be very arrogant right now and to say that's it, that's the be all for these patients, right? But I think one must remain fairly humble about this. So there might be -- and we know it, they are on standard of care. How much is contributing each of those drugs? I think the important point here is those patients are completely uncontrolled. Let me remind you, this was the most severe ever DEE study conducted. There was 50 seizures per month at baseline. Most Dravet -- and I'm not -- it's not a ding on Dravet. It's a very important condition. It's 5x less than that, 3x less than that. So we already started an incredibly high bar and seeing this is incredibly encouraging. Is that going to be like that for every condition? I think, to your question before, likely not, right? We're likely going to see variability throughout. But I think at this point, it should be encouraging. Maybe we can move a little bit? I know Ken has a question because we're trying to get to the end. We're probably not going to be able to stop at 11:30. I love the dynamic. We're going to stick around.

FDA commissioner has supported expediting drugs for rare diseases publicly -- that's granting traditional approval based on an RCT a little bit vague, but post marketing as well. Do you have any plans to explore conditional approval for relutrigine SCN2A and 8A based on the cohort 1 data, which appeared quite strong?

The pace of dosing for cohort 2 is fairly strong, and we believe there might be a point in time in the near future that makes sense for us to ask the question more formally to the agency and try to understand the overall available data at that point. I think right now, head down, executing gets the 80 patients enrolled on that study and submit as quickly as we possibly can. I'm sure it did not get lost because I repeat that all the time. We're talking about no later than the first half of next year, which means we are really working hard for that not to be as late as the first half. And if there is an opportunity for a subpart H submission under the current federal codes, we're absolutely going to explore here. But I think right now, it's a little bit premature with 15 patients or 16 patients of data.

Marcio, can you help us understand the patient disposition once you finished the 16 week, how many opted to be participating in open-label expansion?

Yes. So 13 of those entered the long-term. One patient was -- I'm going to call -- it's not really a lifestyle, but I can't find a better word right now. They decided not to transition for other reasons. One patient, unfortunately, had a significant infection and could not, and he's actually no longer with us. And we have, in these data sets, 12 patients are being represented.

From month 5 onward, the number of patients are consistent.

12.

Yet the seizure rates reduced. So it's not just because some severe patients dropped out that you got better. It's...

No, no, no. This is not a survival effect, if that's the question, which is incredibly common. And you might be thinking about another AES poster. That was -- look at how everyone knows what I'm talking about. Sorry, it's supposed to keep this precaution, right? I'm going to go that. You all know us. It's fine, yes. It's a good point. No, no, this is not driven by a survival effect, statistically speaking.

And this is just a sort of the final -- if you're looking at other functional domain, this is really -- I found this really exciting because if you are going to do disease modification, because we know stopping some of the seizures in these DEE patients is very important. You're going to see improvements in other functional areas. And these are some of the ones that are very important to the parents and caregivers obviously, disruptive behavior, communication, alertness, getting the kids gaze is really important. You can see 60%, 70% great concordance between the clinician and the caregiver here, which gives us a little more credibility, I think, in believing this data. And I think just one final thing here, the seizure severity. So even if you're going down 90%, each of the individual seizures that are now remaining are much less severe than they were before. So it's not -- that looks binary when you just look at the counts, but when you start to look at the gray levels and then the sort of analog side of this, the seizures are going down, all of -- which you probably would expect...

Did you do CGIS then?

So we did as well, but the comparison here was on the improvements that was based on both of them. What we can tell you about it's not on the data set, but it's a very, very high concordance on severity as well.

So now let's just turn to cohort 2. So we had a really good baseline start with EMBOLD. What's the next step for this project? It is to do a pivotal study, obviously. And this is the design. Very, very similar design. What we have done is we can start at a higher dose. And it was important in the other trial, the reason we're confident is, not one dose reduced in the previous trial. And that's one thing that you might do here. You can start at 1 milligram, same innovative design on the placebo. We think that will be great for controlling placebo rate, a good balance. We know the drug is effective. It sort of -- it allows us to have some good ethical perspectives as well. We want to make sure that these kids are treated. That is important. We do care a lot of our patients at Praxis. And this trial is currently underway. It's going to be measured in the same way as we did the other one. And we think it's going to make an obvious way that we can get to move registration as efficiently as possible. The duration is good, and we know that we can see effects in this time period. So we're poised now for registration.

Do you need to have a certain mix of 2A versus 8A in the study?

We do not. Yes, we do not. We're attempting to have as many patients with each one, but I think there is a very good understanding from the scientific community and the regulators that, that is not necessarily a requirement.

When thinking about like, if, for example, the placebo is the second 4 weeks, how many weeks of those is the drug still on board? Or should we just assume it's going to be the first 4 weeks, so they're actually placebo?

You're free to assume whatever. We're going to stick to the design and the disclosure...

What you said about the offering.

Yes. So maybe I'll answer your question about the carryover effects, right? Of course, with a 5 days half-life. And I would be speaking both sides of my mouth, if I say there's no carryover effect, it must be, right? What would that do is maybe the question that we have to -- and what that would do is just reduce the ability to compare drug and placebo negatively to Praxis. So if you believe that, that's what happened, should be even more impressed about the results that we had. So that is an obvious placement for this. I think we talked about this before. There's an obvious way to look into this data. I encourage you to think about what will be the most obvious here. But because you worked incredibly well, we thought it would be ill-advised to actually change the design. I think we hear again and again from all of you and from our shareholders, try to keep things as stable as possible, especially when you are winning, when things are going well, and that's what we attempted to do at this study. But the second part, and I can't reinforce that enough. We all worked with a lot of incredibly severe conditions before. These patients are not in great shape. Keeping them on several months of placebo was not an option. It's not that it's like oh, we thought about this because like we want a clever design or something like this. Yes, that was part of it. But the most important is they decompensate. They die. They have like a series of infections that leads to aspiration pneumonia, sudden death. And that was just not okay, right? So there is an ethical component of this. When you consider all of that, what is the most efficient way? And if you think a little bit, you're going to figure out which is the period, and the beauty is it didn't. When you actually ask, can you guess which period they were in? People cannot guess which period they were in because there is no expectation that you could always be in. We know very, very little about the psychology of placebo effects, number one. And two, I think it's oxymoronic to think that a kid that is seizing several times a day, a parent would be just like, "I'm not going to count." Compliance to diary was through the roof. They feel every single day. These people are incredibly compliant. Human recall is -- but we don't remember anything, right? That's what the literature on recall shows. So...

Variance on severity of these seizures, such that if there is a placebo, could it be exerted on a parent's judgment of what is a seizure and what is not?

Yes. We are only showing countable motor seizures, right? So the patient described the seizure. All epilepsy studies do the same way. You describe the seizures, then there's a shorthand definition, right, meaning a simpler way. They put that on the diary. What we see in the diary is your description of the seizure. But what's on the back end is the classification of the seizure by the PI and in most studies, by the epilepsy study consortium, including this study. So the patients never know if they are counting a motor or non-motor. Maybe that helps with the question. They're just counting what they are seeing. So you might see a lot of absence seizure. For example, most absence are non-motor, as you all know. And then they wouldn't be counted because like you want to see reduction in general seizures. You always want to do that. Now 2A and 8A patients, virtually all seizures are motor, they are very severe. But in general, in epilepsy, you have different types of seizures that count. So I have a hard time with placebo rates, as you can see in epilepsy because seizures are real manifestations of like -- and they're very severe, and they lead to a lot of other consequences.

And then the final, really comes back to everything we started saying at the beginning, how do you -- because of the properties of this drug, it's incredibly well positioned to serve the broader DEE market. And that's what EMERALD is all about. And this is -- we've got a great biological rationale that we went over. We have clinical proof of concept in the 2A and 8A population of phenomenal reductions in seizures, phenomenal increases in seizure-free days, improvements on a whole bunch of other functional domains. So everything is really well positioned. And from a patient and a commercial perspective, a huge large unmet need, where this drug could be used and deployed for a really broad range of these DEEs. And there's a market of 200,000, no matter how we count it, as Marcio said, in the U.S. alone.

So before I move there, I think Yas has a question as well.

I guess given that you saw in cohort 1 percentage of already are titrating to their lower background therapy, did you pick that up at the 16 weeks? I'm just trying to think about if you could, in cohort 2, potentially also show that because that's going to be really important for physicians as well, not only seizure reduction, but sort of down titration. So if you could just remind us of that and how you're thinking about this?

Yes, we do. It's fairly individual. So after a number of months on drug, the physicians can assess if they have a tolerability issue only, right? So they cannot simply reduce the dose arbitrarily because that could create other issues. So there's a trigger. So let's say, a few of these patients are a little bit more somnolent, for example, and then they reduce the background meds, and it tends to resolve. It tends to be a one-to-one relationship. You reduce -- seizures don't increase, and the side effects decrease. So we are capturing all of this. It's going to be all in the [ safe stables ] or reporting to the FDA, all of this, as you can imagine. But because it does not happen for all patients every single day, it is slightly harder to make in trials this small as an end point. But I think as we dose more and more patients, that's going to become a quite important feature, right? At the end of the day, and we say this with all cautions, this drug should become monotherapy, right? The overall promise of this drug, vormatrigine, I know it's not vormatrigine day today, but is that we would be able to get as close to monotherapy as possible in some of those patients. So not the time to explore. There is no regulatory framework in the United States to do a study like that. You probably know, you cannot run a monotherapy study in the United States in epilepsy patients. Rightly so, by the way, but it is certainly something we're going to continue to explore in the future.

And then again, this is just another example of a very simple, elegant and efficient trial to how to answer this question. It's straightforward. It's 1:1 randomization. Either go to the drug arm for 16 weeks or you go to the -- this is a full placebo arm for 16 weeks, around 160 patients. And really, really important, these patients will be phenotypically defined, the DEE patients. That's how DEEs were done until 10 years ago when we didn't know anything about the genetics. Everything was phenotypically defined. And that's how LGS trials are still run to this day because we have got no idea of the number of etiologies that give rise to LGS. A couple of them are genetic, but the most -- majority of them, we don't really know. And that's very important because this is how most physicians come and see their patients that have got DEEs. And that -- because of the way we set the drug up, that ability for it to be downstream of the etiology, it's really a natural consequence of it. And we think this trial -- the drug, as Tim said, there's no titration once-a-day dosing. Pretty much very unheard of for a lot of these types of medicines. I'm sure you know about a lot of the drugs in development that are 2 or 3 times a day dosing requirement. All of these things matter for the conduct of the trial, for the attractiveness of the trial and then for the long-term benefit of the patients and the families in dealing with this. So we think this is really well set up to answer this question in this broad population.

Do you see your ability to enroll a broad DEE population when -- despite all the great data you've shown so far, [indiscernible] the perception, don't give sodium channel blockers to certain types of DEEs.

Yes. I think it's particular, Joe. The 1A, right, not all Dravet patients have mutations on 1A gene, as you all know, but let's call Dravet 1A something similar there. It's probably going to be less rightly so. There's a lot of stuff that works, right? So there's a lot of patients on Dravet that have actually very good control. But if you go back to the pie, right, that we showed before, we talked a lot about Dravet throughout the years because there are drugs approved. They are a very small proportion of DEE patients. So we might not end up having a lot of that in the study or maybe any, we just don't know. We need to start. But the premise of 200, let's call just to round it up, patients with DEEs that would be available for this trial. If the early indication on the -- all the conversations we've been having with sites is -- can be used, there are going to be no problem whatsoever, actually getting this off the ground very quickly because the majority of the patients are not controlled on that. There was a lot of conversations with the agency, as you can imagine, about like, let's make sure we don't put like all 2As and 8As in this study. And maybe that is more of the concern, and I wouldn't necessarily call concern, it's more for us to be careful on allowing for a lot of 2A and 8A would do the overall approval before randomization anyway, it's medical monitors and company because in the words of even people that were in the meeting with us, the kind of you know it works, right? And if you know it works, we shouldn't be adding a lot of those, but we are not seeing that as a problem like whatsoever here, but something to watch. The perception change a lot when data is available. And we've been doing, I believe, a quite a good job and Steve, particularly in our medical team, meeting with all the key opinion leaders in the U.S., have like this once or twice per week, evening meetings where there is a discussion. And I think it's quite -- they had the same initial bias, prejudice and very quickly question that when you know anatomical is not on the same place, why are we using the same things. These are very smart people, very well intentioned. They want to do good for the patients as well.

Is the FDA going to require you to have a certain representation from each DEE type? Or are you going to try to make sure you have a certain representation?

No. The premise is really if you see -- if those seizures are motor, right, quite important that if those seizures are not in a cluster, then go back to what Steve showed on, as Tim calls it, tree of life, right? All the -- there should be no reason not to try this drug. And at the end, there is a disproportionate amount, of course, we're going to need to run subgroup analysis, make sure they are consistent and things like that. But our priority, there is no restriction on the mix.

There were over 1,000 DEE genes now.

There's no way to do that then.

It would be futile. But LGS is the ultimate proof that, that is not necessary, right? And maybe that's the discussion we don't have publicly. LGS is not a condition. LGS is a basket, if you want to call basket because people call things baskets of things that manifest similarly. A patient with a focal epilepsy does not have a single etiology. It's just manifest in locus, in a focus in the brain, and sometimes they stay there, sometimes, generalize. It's the same general principle, right? Like what are you treating the manifestation or the overall etiology?

The target label here would be ages 2 and older DEE, no other virus?

Yes. With presence of seizures, right, because there are some of them that don't have seizure as the primary driver.

Why the incorporation of the full placebo arm in this study versus the prior...

Yes. Levels of comfort or I think from -- particularly from us and the agency, right, in terms of, okay, this is a little bit more -- that should be a little bit less severe population based on what we discussed before. Keeping this longer is going to allow for like better understanding on this diverse population. Some of them might take a little longer to act. Some of them might take a little bit like shorter to act. So it's just -- I'm going to call this the bread and butter. It's normally 12 weeks, as you know. So there's a question here as well, why 16, right? So we know 2 things so far about this drug. Many things, but 2 of them are important. One, the more time, the more effects. So far, that's why we've seen this. But the other is the ability to go to 1.5 milligram is relatively late is at day 35. And with about 20 days to get to steady state, you want to make sure that if those patients went to 1.5, they reach steady state. So it just required a slightly longer steady here.

And then patients would be allowed to be on their background therapy, but then only one sodium.

No, not anymore. And on the previous study, we were missed on EMBOLD cohort 2. There is no restriction for 1. They can be on 2 and they are on 2 as of today.

They should have been moved from 1 to 2 in...

Yes.

And it's global. I mean, that's going to be a critical part of getting to the patients. And we have developed a portal that you're going to see for a lot of these trials. It's important for patient engagement, alignment. So it's going to -- this will really unify the way we look at these trials and the way that we understand the cohorts that we're dealing with.

Those are the committed geographies. So geographies where we either have applications or we have commitment from physicians and working on applications. So this is not theoretical. This is the practical footprint at this point.

Just a clarification for EMERALD, will you be excluding SCN2 and 8?

We're not going to be excluding, but we do not expect that to be a significant number of patients to be included. That was not a requirement to exclude just to watch that they don't become 80%.

Both studies will have SCN2 and 8? What if you see some different efficacy, what does that mean? It's hypothetical...

Yes. That size -- I think it's incredibly unlikely that we would see no efficacy. Yes. And we know there is variability every time we repeat a study anyway. And what you're looking here for is a large significant effect, and we don't believe that, that is a large risk.

How do you arrive at 1.5 milligram?

At 1, we are already seeing the results we are seeing. But the trigger is seizure reduction. We're not going to tell you how much, but the trigger to go to 1.5 is a level of seizure reduction. So we're gauging in terms of like, let's say, one etiology requires more than what is one. It's almost an insurance policy, right? And then we have that ability to go there. Are we going to require to go there? No, at 1.5, it is very high the exposure. You can try to simulate that with a drug that has the properties of relutrigine. So I think that is proper that we stay at the maximum.

Did FDA say that there was a ceiling of any particular...

We did agree on a ceiling of 1.5 for this indication.

So ceiling of 1.5, but of any proportion of the subtype?

No.

Okay.

There was a very interesting debate that I would call. Yes, and whether or not proportions would be necessary and whether or not like -- you've seen being avoiding to use the term baskets, right? And there's a reason why we're avoiding because a basket is something that you know what you put it in and you try to keep proportional. This is not a basket study. This is a phenotypically defined study. A basket would require proportions, but...

Well, somebody who moved departments recently, I fundamentally disagree with you what's in a basket. God only knows what's in a basket after you moved. Are you saying that it is really random?

It is not random if you...

But you are very deliberately putting in...

No, this is not a basket, meaning patients come in, they manifest a given way, and they get randomized. A basket would be...

That you underlying...

Correct. Correct. Inclusion criteria met, as Steve just described, that can be randomized. A basket would be one patient with mutation A, matching mutation A on placebo. That would be infeasible, unreasonable, scientifically idiotic trial to run with 1,000 potential. You can even -- from a combinatory perspective, just that trial couldn't be run.

So how do you ensure a minimum representation of [ non-DS LGS ] SCN2A and 8A?

We don't. Like if you have a drug that can work on -- across the board on -- because mechanistically, as Steve showed, it's like this is a fundamental mechanism to reduce seizure. Why would you want that?

I'm not saying not have those patients. I'm just saying limit them because you want enough representation from the others.

Obviously, we're going to monitor this. I mean, so we're not going to let any phenotype be overwhelming on this study, but a priority, there is no reason to believe that there would be an issue.

Did you do ever consider taking an approach like Longboard had done where they had carved out certain groups such as LGS separately from a broader DEE population, just sort of as maybe a little bit of a hedge as to getting to a larger segment should the overall trial have any issues?

Yes. No, absolutely. I think it's incredibly important to consider what everyone else has been done and then you go back to the science as well, right? Serotonergic mechanisms are not downstream, as Steve showed. They've been shown -- they've been in the market for a long time, cannot ignore the evidence in the market. Fenfluramine does not work in most of those indications. We love narratives, but I love data more, and the data shows that it doesn't work. So other serotonergic mechanisms are unlikely. If I believe that was unlikely, if we all believe it wasn't likely, we would have designed that study because that is a hedging. But if you don't believe, then you have to follow the science, and the science is incredibly clear that it's not a requirement.

We say this at the moment, sodium channels, you block, there is no -- they're all going to be get to a max. There is almost no other mechanism that can guarantee that you'll stop a seizure. You will stop a seizure if you block enough sodium channels guaranteed because it is the only way to generate an action potential is with the sodium channel. And whether you're modifying the output of an intraneuron, which is what a serotonergic system does, there's a limit to the effect you can do it through that particular mix. So it limits. We don't have that with the sodium channel. So that brings a lot of conviction that it's going to work broadly.

You should never disclose the genotypes.

We? That's a great advice. Thank you.

But we also happen to like Acadia...

Not always wise. That's what you're saying. All right.

So we'll switch to the Solidus platform now, which is our antisense oligo platform. It's been in the company for some time. There are 4 molecules, 1 late stage, 3 preclinical. And it's a very well-defined platform here. We've got a computational start to help us select ASOs. And then we've got a very well-trodden way we think about development as well. And I think it's important. The platform encompasses that. For elsunersen, the ASO is a gapmer. The gapmers are very specific instantiation of antisense oligos. They've got a chemical modification in the middle that's different to the chemical modification on the wings. That sort of ASO when it binds an RNA, it targets it for destruction. It's like a tag for destruction and it tracks the attention to this molecule you call RNase H and that destroys the RNA. So you can -- with a [ disease ] like SCN2A gain of function, the therapeutic hypothesis is, let's just reduce the number of channels, and that's going to be beneficial because you've got too many overactive channels. And we've got a whole -- I'll show you a whole slew of preclinical work that validated that concept. And this is -- you see here, this is a mouse harboring a human mutation, and these mice are really sick. They're dead by day 30, if you do nothing, on the left. And what you can see, when we inject at birth with 2 different concentrations of an SCN2A gapmer ASO to reduce the amount of channel, either by 50% in blue or by 80% in orange, you get an extension of survival. That's remarkable. Now if you give a second injection to compensate for development changes and size of the animals, that said you can see an even more pronounced increase in survival. These animals, the second dose that at a 50% reduction extended out to 120 days before you start to see any -- and that's because of the pharmacokinetics of the ASO wearing off and then the disease remanifesting. The final test was what happens if you intervene a little bit later in the natural history of the disease because in a human, you can't always do it at birth. What we did it as late as we possibly could before the animals start to die. And you can see, at the point where we inject it with the control ASO, they all just died within a few days after that. But then you can see in the 50% and 80% for that later-stage injection, incredible. And really, that delay was just the time it took for the ASO to kick in and start to work. The time lines for the human are so different that this is not issued for people because the disease doesn't develop as quickly. So the ASO kinetics is much better suited to the disease kinetics, if you want, in people. Right. And this is the overall program to give you like a road map, but we've got -- it can be a little bit confusing because it's EMBRAVE Part 1 that we -- I think it was the fall of 2023, we first released that 4 patients, open label, and saw the first signs of remarkable seizure reduction. It was consistent. I mean this is really great proof of concept that this ASO is doing what we wanted to do, and it's having the clinical effect. EMBRAVE Part A is an ongoing dose escalation study. It can be between 8 and 16 participants, and this is an option, 3:1 randomization. So this is a proper control study, and we've got 3:1 for the treated arm and the sham arm, and so we're going to get a very good look to see what happens. And then EMBRAVE3 is really, again, I think, a simple elegant design that lets us get a registrational cohort in a controlled fashion with a bunch of patients. And then there's 2 additional cohorts that let us wind back the intervention time to birth. And that's a really important thing. So the label, we're shooting for a label that's going to say from birth onwards, you now can take this ASO for this disorder. And we've -- and that we've got a very powerful way of understanding the genotype and how to include patients in this study. So we know that they're validated as gain of function. And in the other study, this just gives you a taste of what that EMBRAVE1 from 2023 readout look like. You can see 30% reductions, et cetera. In Part A, this is the design of the trial that's ongoing right now with the randomization, and we expected to finish this by the midyear. So again, these programs are moving ahead at light speed. And then this is going to be very important for supporting registration with the EMBRAVE3, which is the registrational study.

Finishing enrollment or finishing...

Finishing enrollment, yes.

And this is a registrational trial. Again, very simple design, 1:1 randomization between sham elsunersen for 24 weeks and set to start shortly, very shortly, where we're in really good shape with this. And this is the cohort 1 and you see it's a control part. If you look at the next part.

Yes. Just maybe -- sorry for interrupting you. One of the important things here as well is this is the shortest disease-modifying study being conducted with an ASO for a CNS condition, was incredibly productive discussion with the FDA on what is expected based on we know on this drug, what you've seen the overall biology of the channel and the number of dose we've been given to show a disease-modifying effects. And we're pretty happy that we landed at 24 weeks because I'm going to bring back to -- these patients are incredibly severe, and keeping them on sham for 1 year, 18 months, it is -- when necessary, it's necessary, but in this case, it was not necessary. So we're incredibly happy with the fact that we're limiting the overall exposure here as well. I would say even happier because that is even more unique on what Steve is going to show now.

Yes. The way to reach back to 0 is actually, really, I don't think you can get more streamlined than this, 2 cohorts, uncontrolled. So there's no placebo arm, [ N-of-5 ] in each group, and you can start with first between the 1- to 2-year group and have a look at some of those and then start to introduce it to the younger group. So we're going to very rapidly move in. And there can be a bit of overlap potentially here between these cohorts as well. So we can very quickly get to the point of 0, which we think this is a disease that manifests early. So the earlier options are absolutely very critical for these patients. And then we're doing what we can to be able to include them and give them that option.

Can you remind us about the 4 patients where you gave us the data initially? Are they still on treatment? Where are they now?

Yes, it's a good story and maybe one that sometimes you got a little bit sad to report. So at that point in time, we could only dose those 4 patients and then you wait for a review by the FDA on the next step. Remember, it was a seamless design. So we had 4 patients review.

How many doses did they get?

They got 4 dose, yes. So there was a natural stop, a planned stop on the dose at the end for a decision on the next cohort. Unfortunately, we saw the natural history playing full blow. After we stopped, 2 patients died after stopping the drug several months after through -- they end up getting more seizures, as we would expect, that they had regained, and they got into the overall complication.

Were they the most severe or the oldest?

They were similarly severe. It's very idiosyncratic sometimes happens, right, infections, ICU, status. So I don't have complete confirmation, but I think they helped a lot with these conversations with the agency. It's unfortunate what happened with them. But because they're not on drug and because it was several months later, that could not be any association whatsoever, other than that the drug was keeping then alive. The other patients are on drug and doing well, the other 2 patients.

The other 2, they were allowed to stay on drug?

They were quite quickly when certain things are -- certain decisions are made for administrative purposes, and they end up being -- resulting to patients die. Those bodies making the administrative decisions tends to reverse -- tends to revert their decision quite quickly. So they are on drug to today.

And they've been treated for roughly...

18 months.

18 months.

The highest mortality in most well, the DEEs are the 2A patients. So the natural history is terrible.

6x higher chance of die by childhood than Dravet. 2A patients has.

If those patients, like, let's say, you have 2 commercial programs here, like if they have inadequate control on one of these genetic, could they still go on?

We think there's natural complementarity there. The ASO tackles the genetic root cause. And this is now a brain is different because that's developmentally, everything about it, there's going to be residual excitability due to other things that have happened. And the relutrigine is very well positioned to address that. And it's not going to be only for this case. Relutrigine could be a drug you might add on for a whole host of other genetically defined therapies. And we think that, that combination of root cause, residual hyperexcitability is a really important match. And so we are definitely thinking deeply about what this synergy could look like.

Can you give us some sort of sense of -- I think you guys reported the percent reduction in seizures, but the residual seizures, how many are they? And put it in context with the SCN2A4...

Yes. So they're definitely way less severe. And I think that, that's something we see as well here, right? I can imagine there is slightly different motivations to participate on intrathecal trials than on oral drugs, right? And these patients tends to be fairly uncontrolled as well. But relatively anatomically stable. There's a lot of anatomical change that happens because like wheelchairs and other things here. So what we see in the short run, we didn't talk about what's happening now with 18 months and so on. It is -- I wouldn't call modest, but about 50% reduction is not modest on that. But there is somewhat of that plateau there for a while. The reason why we have this slide here as well. I think the combination because they are on other drugs that are not necessarily adequate to control. The combination of both of them might be and is likely to be the way to really round it up and get the best treatment for those patients. I believe Yas has another question as well. I don't know if she's still on the line.

Yes. Yes. I hope you can hear me. I guess, team, as you're thinking about sort of the complementarity of both a small molecule as well as an ASO that you're developing, is there an opportunity, I guess, if patients, obviously, with genetic mutations are going to be identified earlier? Like as we think about modeling both of these drugs that have 4 billion total opportunity, how do we think like which population is easier to recruit, get to pivotal data, think about it, obviously, relutrigine has a broader population. So maybe help us understand sort of the developmental time lines and first-to-market and uptake between these 2.

Yes. So we've been working on the recruitment for EMBOLD 2 for about 7 months or so now. I'm not going to give you exact numbers, but I'll give you order of magnitudes. There are more patients interested in slots on the trial right now. So the question is more eligible, it's not necessarily whether or not we're going to finish the study. It's going again, as we said, incredibly well. We haven't -- we don't have a single site open for EMBRAVE3 yet. They are in the process. Every single patient needed for that study is already identified. Not making any promise here that all of them are going to randomize.

The site -- you know what the sites are. They're just not open, but the site directors have their records?

Patients, their willingness to participate their seizure counts, their concomitant map. It's just unheard of, that are not even open. It's a small study as well. But together, there are 120 patients. It's not a small study. When you look into other GE studies taking 2 to 3 years to recruit similar size, right? And again, this is not a promise, not a guidance that we're giving. It's just the expression of interest is clear here. I can't like help myself but to think, and Joe is going to remember this really well because we worked together back then on different situations. On similar questions are being asked about Type 1 SMA back in the day and a lot of you do parts there as well. And what -- why would you need 3 different drugs, 3 different mechanisms for if you have like a gene therapy, an ASO, oral splicing modulator for that. And fast forward today, and you all know the markets better than I do at this point in time. And I think what we learned is that biology is tricky. The majority of them probably needs a little bit of and/or there are specific situations and each one of them are multibillion-dollar drugs. So not a straight line to get to the same results, but that is one good comparator because pediatric severe, high mortality, similar mechanisms as well. The only difference is like a gene therapy will never, and I don't say never often, going to be developed for this because precise control is important for these channels, and we cannot deliver at this day and age like vector-based gene therapies with precise control.

Again, I think it's just another sign that we're going global. We've got these sites lined up. Portal is the same, so we're taking advantage of that interface with the participants. And I think, again, just poised to really move as fast as we possibly can on this. What I wanted to just in closing for the -- for my part is just to give you really a brief introduction to the pipeline for the next ASOs coming forward in the company. There are 3 of them: PRAX-080, PRAX-090, PRAX-100. One of them for PCDH19, where we've -- Praxis, again, I think, is sort of leading the way in a novel way of thinking about delivering a therapy. And it's a disease of mosaic expression. Our idea is to completely ablate because we know that males that have none are clinically normal. So that's -- it's much worse to have a mutant and wild-type expressed. PRAX-090, SYNGAP1, haploinsufficient, you need more of it. What I'll focus a little more time on is PRAX-100. That is a haploinsufficient autism, one of the most common monogenetic causes of autism, not enough 2A. So the channel that when it's gain of function gives epilepsy, when it's truly haploinsufficient, 50% less than normal, autism manifests. And there's about 20,000 of these patients. And it's the analog to Dravet. Dravet is the same thing in 1A. This is 2A, the excitatory gene and the excitatory neuron. When you don't have enough of that, you don't have enough activity now, and that gives rise to autism. So the therapeutic hypothesis, can you double the amount that you've got. And this is just giving you a taste for what we've seen with 4 of Praxis development candidates measured in a humanized animal that contains mouse, that contains a full human gene. We're giving it the human ASO that is designed to increase the amount of protein. You see all of them are pushing it more or roughly 200%, which is a doubling, which is a restoration of the normal amount of 2A. And we think we couldn't be more excited by the fact that -- and this is hard, as Marcio said, being able to increase the amount of a protein is difficult. And we're not disclosing the full mechanism here that we've deployed. We obviously know now about retained introns, we know about everything. But we think we're just very excited by these first signs of direct increase in protein. It's the first time we've described this publicly. And just giving you a feel of where this platform could go. And I think, Marcio, I was a bit floored by the size of the potential market here as well. Very excited by what we can do for these kids with 2A haploinsufficient autism. And we'll hear more about this soon.

So this is not getting rid of the Poisson Axon strategy? It is not.

It's undisclosed. We looked at everything, and we'll reveal more later, but there's other ways to increase the amount of protein that don't involve that. There are other way -- other things that you can do, bind RNAs and different RNAs and in different spots that make you end up with more protein.

Another complementary way to answer your question. We are certain that retained intron strategies are important in certain situations. They are highly dependent on the rates of the events in the population. And when you understand the population, the rate of events, people should do a little bit more advanced molecular biology and not the 101. If the event is not there, there is no effect, right? The event is not 100%, no event is 100% expressed, and retained internal events sometimes naturally disappear throughout life. And we like when they stay, so you could assume maybe one here is, maybe one that is not here is, we had hundreds of those. We like when they stay. When they don't stay, they are a big problem because it means they're going to work for some patients, but not for others. We like the idea to start with working for a lot, not a few of the patients because we can always walk back from there, but not walk forward from now working into people. And maybe I'll leave there until next time when we show the many more things we have in this program that get us so excited. All right. I know Doug has a question, and then we can get more questions, I'll wrap it up. Doug?

Just going back to EMBRAVE3 with the additional cohorts to get the age lower. I'm just curious, how quickly do you think those would be completed? Would those be sort of -- would you seek labeling to like an sNDA process? Or is there any way that you could perhaps even get those completed and included at the time of launch?

It is fairly fast, and we're incredibly grateful to the division of neurology to work with us on this from the FDA. There's a requirement for a few doses for the first patients on the first cohort, but not for the complete time, not for all patients, 3 patients, 4 months, and then we can move to the second cohort, and then we can move to the third cohorts as one would expect, right, to be. So in a sense, we are the rate limit. You dose 4 patients today, it means that before the end of the year, we're already dosing the next cohorts. There are integration of the different studies here. As you can imagine, that might allow for that to go faster. So stay tuned. I would say right now, without guiding too strongly, probably the cohorts 1 and 2 would be the initial application, and it might take a little bit longer for cohort 1 because it's birth, right? So which means that we need to know the patient was just born, available, willing, able to consent, stable enough. So those studies tend to take a little longer. Those cohorts in general. So it might be the case. It might be that we get lucky, and we are accelerating that. So I think all the way to 1, we should be pretty certain, all the way to 0, supplemental NDA is probably the way to go.

Sorry, you cannot start cohort 2 until you dose -- at least give one dose to all of the 40 in cohort 2?

No, no, 3 patients.

Only 3?

Correct.

I have a question for Tim. With so many things going on, can you remind us what your cash position is and your runway and what is baked into that?

So we disclosed this morning that we had $472 million at the end of Q1. I'd continue to confirm that our runway is into 2028 and it funds all of the trials through the readouts that we talked about today. So we've got the cash to execute all of this.

Right. We get a chance to read the press release...

No, it's totally cool.

Marcio, just to clarify, for the EMERALD study, are you doing site activation in the middle of 2025? Is that when it's going to start?

Yes. But our goal is to actually get patients enrolled at that point in time. It's happening basically right now. No pressure on Brian to activate. Yas has a question for Tim.

Yes. I guess I'll let Tim take a start and then Marcio, you can add on. I mean, obviously, with this really rich pipeline, how do you think about which assets could warrant to be partnered at an earlier stage of development versus taking them all the way to the end? How do you -- or maybe help us understand how do you think about factors that determines partnering versus bringing to the finish line and potentially even commercializing on your own? That's one. And then the second one is, you guys have always done a great job using great resources for getting interested patients. You did that for your essential tremor study. You're currently doing that for your focal epilepsy studies. Or I'm assuming that same resource is being utilized for the entire pipeline. So if you could just talk about that aspect because I'm sure as enrollment continues, we're going to learn more, but maybe both would be helpful for us.

Sure. So I'll kick off and the idea about partnering these assets. We're very excited about them. I would expect there's other people who would be excited about them as well. But the best thing that we can be doing for patients, I think for shareholders as well, is focusing on the execution of advancing these trials to further derisk them and to validate the mechanism and what the applicability is. I think when you were -- when we look at which ones we might partner in different ways, there's different things to think about there. For example, on vormatrigine with more of a focal epilepsy, a much broader market, maybe a partner who has broader commercial reach might make sense, although neurologists are a fairly targeted market and something that we feel that we could do. But again, if there's a partner that comes with a good offer, and I think that part is critical. We don't want to be wasting our time on deals that don't make any sense. But that might make, from a commercial use perspective or use of resources and realizing the full potential of the molecule might make sense for some of the smaller market areas like what we talked about today, rare diseases can be done with a fairly focused field force and commercial effort. They also have commensurately higher pricing as well that we talked about. So even globally, I think there's a opportunity for us to look at commercializing that as well. But again, there are companies set up with infrastructure outside the U.S. that might make sense to partner with also from a bringing to market perspective and expanding -- being sure we realize the full value of what these molecules can do. All of those conversations, we are best prepared for if we continue to advance these trials. And so we're not waiting for somebody to come as you'll get from looking at our pipeline side, all 4 of these assets that are in late stage right now, if we were to include vormatrigine and essential tremor as well are all in parallel enrolling quite quickly. And so we're not taking our foot off the gas with any of those, which then I think leads to your second question, and let me just address that quickly, and I'll leave it to Marcio to round out as well. We have done some very thoughtful, I think, innovative ways to recruit and identify patients for essential tremor. I think what we disclosed today is now, it's over 200,000 people who have been interested in the Essential3 study. For vormatrigine. We're utilizing the EMPOWER registrational study to find patients. We will apply a similar type tools as well for both the DEE programs that we talked about today. But I think the other aspect that we will lean into more on the development of epilepsy side is the patient organizations. They're very well organized. They have a good sense of who in their networks are good candidates for these trials as well. And so that gives us, if you will, a third leg of the stool, on ways to identify and find these patients sooner. So there's a whole breadth of tools that we'll be using. And we very much feel that it's important to invest on the recruiting side of the trial because the faster we can get the right patients into the study, the less resources we need in the long run, but also more importantly, the faster we bring a new therapy to these patients as well. So I'll pause there and open it up for Marcio to...

Yes. And maybe on the recruitment, we get that question a lot, right? Like we get patients faster, the quality of the patients is great, like the quality of the study is great. And like how do you guys do this? And maybe I'll take a little bit of a detour for a second, right, back into when every auto manufacturing in the world suck at producing cars. People would ask how does Toyota do this, right? And you might recall back then, you'd say, come and look into the plants. I will show you everything. We don't have any problem if you actually take a look under the hoods. And we take the same general approach with competitors, with colleagues. You cannot imitate and you cannot copy passion, love for delivering this competency and lack of silos. Go ahead and try. And that's the way, I would say, right? I would be -- I would leave people behind, but it looked like Taylor, who leads our recruitment, and Megan and Kelly Dolby and many more people here, they are obsessed on getting this done. And it's impossible to copy obsession like no one stop until they understand what is necessary. So I could give you financial reasons, system reasons, the fact that we actually look for the best possible technology that we lever technology. Yes, that is all important. But at the very end, at the very core that's something no one else in Cambridge, in Boston or anywhere else can copy. People love it and they want to deliver to these patients and I'm incredibly grateful to everyone here at Praxis that does this. That's the long answer. All right. I think that was it. I do have a closing slide, but I'm not sure we even need to go for this. We went through absolutely most of it. Maybe one aspect that we haven't touched before, I believe is quite important here, is all these molecules are pretty new, right? They are all discovered very recently. And one of the questions that is quite important in drug development. As you know, currently, the past was important. I think currently is even more important is, how long is the intellectual property runway? And for how long can we actually see the benefits of this, while we are working on the next thing. And while we are not going to give you specific times for each one of them, but they always start with a [ 40 ] or they will end with a [ 40 ] here. I think most of us, because we're all very young in this room are still going to be doing this and still going to be loving this. But a lot of people are not even going to be with us, that develop drugs right now when those patents expire. So we have a lot of grounds to bring to life to patients the promise that we have here today. I'm insanely grateful for all of you, not only attending, asking great questions, engaging but staying 40 minutes longer than -- it shows the importance of these events and shows the passion that we all share about helping these patients as well. So thank you very much. We shall meet again, and have a wonderful weekend, everyone.

Thank you.