All right, thanks very much. It's my pleasure to be moderating this fireside chat with Joe Lewcock, Chief Scientific Officer of Denali Therapeutics. I think, you know, instead of just having you give the canned overview of Denali, maybe we can just keep this conversational and talk more about the platform and some of the science and some of your programs. But maybe to start, right, we were just having a conversation about how, you know, talk three years ago, right, transport was intellectually interesting but not validated. Now we're at a point, right, where there's a number of companies trying to make sure they don't miss out on this new entry into the space. And it reminds me a lot of GalNAc, right, where you had like Alnylam as one of the first, but then, you know, many others coming.
You know, how, maybe give us sort of an overview of where you feel like Denali sits as differentiated within the transport space and your thought. It's kind of high barriers to entry, right? Like why can't others sort of copy what you're doing at the same level?
Yeah, yeah, great question. So first of all, thanks for having us. Great to be here and talk a little bit about what's going on at Denali. Yeah, and Denali as a company, you know, we've existed for a little bit under 10 years. And when we started the company, we really had three scientific tenets, you know, going after using human genetics to guide drug discovery, doing brain delivery, which was where our TV platform came from, and then doing biomarker-driven drug discovery. And you're right, at the time we started Denali, you know, I think a lot of, you know, what we heard was our, you know, our blood-brain barrier vehicle's necessary. Is this going to be needed for brain therapeutics? And we felt strongly that it was, right?
And so we've spent a number of years building up our expertise in the area as well as building a proprietary platform, which we call the transport vehicle or the TV. And you're right that over the past few years, I mean, the interest in the area has just exploded. And I think that's in part because of clinical validation from our DNL310 enzyme program, as well as to other TfR-enabled programs such as Tronte from Roche, really showing that these TfR-based vehicles can have a major impact. You know, I think we've spent a lot of time trying to understand the details of transferrin receptor biology. And I do think the details of how you engage TfR does matter in terms of developing a best-in-class therapeutic.
And so for us, you know, really the transport vehicle just by means of a quick review is engineering the Fc domain of an antibody to introduce low affinity binding to transferrin receptor. And what that does is it really enables, you know, optimal engagement of TfR to get the brain uptake while, you know, avoiding some of the non-desirable effects of engaging TfR and getting, you know, the optimal biodistribution once you get into the CNS. But then leveraging the Fc domain really allows us to be highly modular with our platform. And we're really able to use it to deliver a range of modalities, including enzymes, oligos, and antibodies, and really flexible. And one of the things that, you know, sets us apart is we really have strong IP protection around the engineering of the Fc domain.
So we're really the only player in the field that's not having to use, you know, added sequences, appended sequences in order to engage TfR. And that really, in addition to the way that we engage TfR, minimizes things like immunogenicity and keeps us with the, you know, the optimal drug-like properties of our molecule.
Yep. Okay. So, you know, across the space, right, there's different companies with different flavors of this, right, be it Fabs, full antibodies, engineering out effector function, maybe having some residual effector function, low affinity, high affinity. Like, is there, can we actually look across clinical datasets now and try to get a best sense of the best combination for a given cargo to optimize efficacy and safety?
Yeah, yeah, I think we can. And I think what we found is it's not exactly the same depending on what cargo you're looking at, but there is an optimized profile. You know, one of the things we've used and part of the benefit of using an Fc domain is in addition to engaging TfR, you are also engaging FcRn. So you're getting FcRn-mediated recycling, which is the mechanism by which antibodies have such a nice half-life and part of what makes them great as drugs. And so being able to leverage that gives us longer half-life in the periphery. And we found, I mean, speaking directly for our oligo transport vehicle, that that extended half-life in the periphery is super important for brain uptake because TfR-mediated uptake to, you know, through into the CNS is not instantaneous.
In fact, in many of our studies, pre-clinically, we see the Cmax at 24 hours. So if you have a compound that's cleared very quickly, then you're really not maximizing the level of brain uptake. So I think in terms of the way that we've engineered the TV, there's some advantages there. You also spoke about effector function. And effector function is critical for some mechanisms such as Aβ, but it's a challenge to pair with TfR because you can get safety-related issues by essentially binding TfR and then having effector function against those TfR-expressing cell types. We've developed a mechanism that's unique to the TV where you can essentially have effector function and have TfR binding, but not at the same time. And so that alleviates some of the safety issues.
We feel like that's a key differentiator for certain targets like an anti-Aβ where you need that effector function, but you want to be able to do that safely.
Yep. Okay. And then, you know, I mean, across companies, we've seen varying levels in some programs, not in some programs, a significant amount of, you know, hematological-related AEs, be it either anemia. There's been a couple stroke events across trials, which I think there's debate if they're drug-related or not, but there's at least some smoke. For you guys going forward, right, obviously maintaining this platform safety is really, really important. How have you figured out how to avoid those on-target issues?
Yeah, and I think we've been very careful about that. I think you're right that there's, you know, been some safety issues out there for TfR-targeting molecules that are, you know, likely to be molecule-specific, not TfR-related. But things like anemia and reticulocyte depletion are on mechanism for TfR. And so we've been, you know, very careful in terms of how we engineer our molecules to avoid that. So the vast majority of our TV molecules, they lack effector function. That's one of the major things that can drive that anemia. But they also engage TfR monovalent, low affinity. And it turns out this is one of the details of how the way you engage TfR matters. So higher affinity TfR engagement or bivalent TfR engagement can actually have a similar effect to effector function.
Is that because you induce internalization?
Yeah, you drive down receptor expression, right? And so you can end up having similar effects. And so, you know, when looking at our molecules compared to the space, you know, we have a number of details of those engineering, which end up actually that lower affinity TfR binding actually maximizes distribution to the CNS while at the same time avoiding some of those safety issues.
Yeah. Okay. Well, these are all interesting theoretical questions to kind of get, or not theoretical, but scientific questions to get into, right, as the space gets more and more crowded and as you guys advance, and so it's helpful to kind of talk through. Maybe being more tangible, right, on 310, walk us through what the next month or two looks like inside Denali as you get ready to submit a filing.
Yeah, so we've had very positive interactions over the last quarter with the FDA on DNL310. So that's our lead program for Hunter syndrome, which is an IDS enzyme conjugated to our TV platform. And what we were able to align with regulators on is that CSF heparan sulfate levels are going to be reasonably predictive of clinical benefit and that surrogate endpoint would be sufficient to file a BLA for accelerated approval. So we're super excited about that. I think the second thing we aligned on is that the totality of our development program with DNL310 would be sufficient to convert that from an accelerated to a full approval following the completion of the COMPASS study. So exciting times. And I think now it's really just about preparation for filing of that BLA, which we plan to do in early 2025.
Are you expecting an adcom? And if so, do you think that adcom can incorporate, you know, vetting of other products like from JCR or REGENXBIO?
Yeah. So I can't comment on the adcom at this time. You know, time will tell on that one. But in terms of the other products, I mean, I do think, you know, looking at paths for accelerated approval based on heparan sulfate does open the door for other companies in the space. And I think we have to acknowledge that. I do think we have a highly differentiated data package. And just to name a, you know, a couple of key points, I think we're the only company to show normalization of CSF heparan sulfate levels. So other companies will show a correction, but it's a partial correction. And we're able to normalize it back down to healthy levels and have that be sustained for over two years of dosing.
And then the second is when we recently released some of our neurofilament data where we were able to show, you know, prolonged dosing was able to get a significant reduction in neurofilament levels. And in some patients, again, back down to baseline. And, you know, notably, we've been the only company to show that neurofilament data. You know, and we believe that this is something anyone could collect. And so, you know, we think that if others had it, they would show it as well. And so we really feel pretty comfortable in terms of the differentiation of our biomarker data package.
Yeah. Okay. And so you recently described top-line data for the Sanfilippo program. Maybe walk us through that and then I'll have a few follow-ups.
Yeah, sure. So yeah, in the same quarterly release when we talked about our progress on DNL310, we also announced that we were able to get a robust biomarker response for our DNL126 program. So this is a program against a related enzyme, SGSH. And so it's in the same pathway, but it's a different indication, Sanfilippo syndrome. And we were able to show, we were able to share in our phase I data that we were able to get robust biomarker response, including normalization and early data from those patients.
Yeah. Okay. And how is this disease similar and different to Hunter?
Similar disease in that the two enzymes are actually in the same pathway in terms of breaking down heparan sulfate. And so that's where there's biological similarities. And what it means is you can actually use essentially the same biomarker of that CSF heparan sulfate for both indications. So we're really able to leverage some of the expertise we've built in biomarker discovery, but we're also hopeful then, you know, to leverage some of the regulatory interactions we've had on DNL310 and hopefully that there's read-through to the DNL126 program as well.
Actually, the other thing we announced as part of talking about this top-line data is that we're expanding our phase I study in Sanfilippo syndrome to include a couple of additional or few additional cohorts, three actually, to really make sure we have a robust biomarker data package to enable accelerated approval for that program as well.
The thought is you could file on this individual study.
Yeah. I mean, we haven't come to a conclusive agreement on that with the FDA yet. We do have START designation. So we're able to have frequent back and forth with the FDA on the 126 program, which is fantastic. But we do feel like there, you know, there's again, that strong read-through from a biology perspective. So, you know, we're very optimistic that we can follow a similar path for DNL126.
You know, we've all learned a lot about neurofilament over the past five years as it relates to not just, you know, what diseases it's relevant in and what it might be telling us, but also the kinetics of neurofilament across indications. You know, based on your experience in Hunter, do you think that you can show a quick signal in the Sanfilippo population that you've selected for this phase I too? Or is it going to be one of those things where we got to kind of wait a year and a half to show some clear-cut decrease?
Yeah, and great question, and we just don't know the answer to that quite yet. You know, to talk about, you know, neurofilament, a great biomarker used across indications, of course, great data in ALS, you know, with tofersen and in MS and other indications. You know, we were the ones to really blaze the trail for neurofilament in the MPS disorders. So it hadn't really been looked at before we looked at it in MPS2. And so there's a little bit about understanding how the biomarker behaved in addition to looking at it in terms of a treatment readout. The same is somewhat true with Sanfilippo, where there's actually no published reports of neurofilament levels in these patients. And so it's a little bit, you know, challenging to say like the timing of response or how it's going to respond to treatment.
I think, you know, based on the neuronopathic component of disease, I think it's reasonable to predict that neurofilament levels will be elevated in these patients. But time's going to, you know, time will tell in terms of the, you know, details of the treatment response.
Yep. Okay. And as it relates to confirmatory studies for 310 and then in the future for the Sanfilippo program, I mean, I think, and I'm saying this transparently, right? Like I believe 310 firmly works. Like I think there's no doubt about it. That said, in some sense, the phase three study is essentially a clinical proof of concept study, right? So from your perspective, right, how did you guys think about designing a trial without really knowing how a placebo might behave on a neurodevelopmental endpoint? Like obviously there's a chance that that trial fails the drug. And I'm curious how you mitigated those risks.
Yeah. Great question again, and I think, you know, of course we thought carefully about that, and remember that when we, so we have a, I should take a step back and say we have a currently ongoing phase three study for DNL310, the COMPASS study. When we designed the COMPASS study, actually the regulatory environment was very different, right? There wasn't necessarily that path for accelerated approval, so this study was designed to enable approval of a molecule, right, so just to give a quick rundown of the details of that study, there's two cohorts. There's a neuronopathic cohort and a non-neuronopathic cohort, and that's really to, you know, make sure that we feel like DNL310 can address both the central and peripheral aspects of disease, and we wanted to be able to show that directly as part of this trial.
You know, but the COMPASS study is, you know, designed to really, you know, have two primary endpoints. The first primary endpoint is heparan sulfate at 24 weeks. That's an endpoint that we feel very confident that we hit based on our phase I data showing such robust biomarker readout. The second primary endpoint is the Vineland scale, the third edition at 96 weeks. And suffice to say, we were very careful to use the data we had in hand to power that study to make sure we gave ourselves the best chance of seeing a readout in that study.
I think, you know, our hope is that when we look at that in a placebo-controlled fashion, the totality of the data set that we're going to be able to bring to bear, including robust biomarkers, impact on neurofilament, impact on other functional outcomes such as hearing, which we've also seen in our phase 1 study, that the totality of that data package is going to be supportive of approval.
Yeah. And are you trying to work through what that study could look like in Sanfilippo too? Or is it too early?
It's a bit early for that. Of course we're thinking about it. And, you know, we are doing long-term planning there. You know, I think our initial focus will be on generating a data set to enable accelerated approval of that molecule. But of course, you know, again, following a similar path to enable, you know, full, you know, a fuller data set for 126 as well.
Okay. And so as you think about the Transport program broadly and, you know, whether or not this is not just modular, but sort of cargo agnostic, you know, what are some of the challenges in translating what you've accomplished with enzymes to, you know, siRNAs or ASOs?
Yeah. So I think many of the rules for enzymes are similar to ASOs, and, you know, but some are a bit different. I think that we are super excited about the data that we have thus far on our OTV platform, our oligonucleotide transport vehicle platform, of the ability to deliver some of these either antisense oligos or potentially other oligonucleotide platforms to the CNS. What we found is I think, you know, there's details around how you engage TfR that, you know, I think there's subtleties to it in terms of enzymes versus oligos. But really I think for the oligos, it's about maximizing those drug-like properties and making sure you have reasonable half-life in the periphery to enable brain uptake.
The exciting thing about some of the oligonucleotides that we found is, you know, one of the benefits of intrathecal delivery of oligonucleotides in the CNS is a very prolonged gene knockdown, prolonged pharmacodynamic effect. And we've optimized our molecules such that even though they're delivered through IV and delivered through transferrin receptor, that you still retain that prolonged knockdown and that similar benefit. So we've just recently shared some data on our lead program targeting MAPT that we're able to sustain knockdown for, you know, up to 15 weeks. And that's as long as we looked. It could be longer.
That was an NHP.
After the final dose. What's that?
That was an NHP.
That was actually, that data is in mice, but we're doing.
And sometimes that might be better, right? 15 in mice might predict longer and longer in mice.
Exactly. Mice tend to metabolize things pretty quickly. Yeah.
Right. Okay. Okay. Interesting. I mean, I think, you know, and this came up on an earlier panel too. I mean, one of the challenges with oligo-based drugs in the CNS is the long half-life is great for a commercial product or a clinical product, but it's a challenge for actually getting an IND time because the toxins just take a really long time. Where are you with all of that?
That program, our lead program targeting tau is currently in IND enabling studies and we're planning to file in 2025.
Okay. Okay. Got it. And so the initial study for something like that probably would be in patients, is that right? Or would you think you'd go into healthy volunteers?
Yeah. We haven't provided guidance on that quite yet. What I do think is encouraging is you can get in what I think the path has been somewhat paved by Biogen and their intrathecal delivery of a Tau ASO. We do feel like we're differentiated there. But what they were able to show is you can use CSF levels of Tau as a biomarker readout to essentially get a relatively quick readout of target engagement in early stages.
Their tau PET data looks really interesting.
The tau PET data is very encouraging, right? And so, I mean, I do think combining a CSF biomarker readout with an imaging readout of tau pathology, I think you can very quickly, you know, validate that that molecule is working the way you would hope it would.
Yeah. Okay. So we have 10 minutes. What do you think we should talk about next?
I mean, I think that we are super excited about our TV platform broadly. So we talked a little bit about OTV and the MAPT Tau molecule. We could talk a little bit about our ATV:Aβ molecule.
Yeah.
This is a molecule we're super excited about.
So why didn't Biogen take that forward?
So, you know, we signed the agreement with Biogen, which the Aβ molecule was a part of back in 2020. So this is pre-ADU readout clinically even. I think it was a much different space for the field as a whole and a much different space for Biogen. What's important to note is Biogen did opt into that program, but they opted into a specific Aβ molecule that was of interest to us.
When did they opt into it?
They opted into it, I believe, in 2022.
The inference is that they may have opted into an Aducanumab.
Yeah. Yeah. And we're not, I'm not at liberty to say what antibody that they did opt into, you know, but suffice to say that there was, you know, I think that there were limitations to the ability to move forward, you know, or their interest in moving forward with the molecule. And so they did make the decision to give it back to us. We feel like that, you know, we're, it's something we're very excited about. And so what we've done is we actually have transitioned away from that antibody that they licensed and moved on to an anti-Aβ sequence that we feel is differentiated from that. And we're pretty excited about it.
I mean, is it a novel sequence?
It's a novel sequence. It's a Denali-owned sequence, and so we're excited to move that forward. We're moving it forward both on our TfR platform as well as our CD98 platform.
Right.
TfR, the TfR molecule is a bit more advanced than CD98, but, you know, we feel like there's, you know, a lot of rationale to move forward TfR based on the data that's out there, you know, really exciting landscape. CD98 could just be, you know, completely differentiated from anything else out there. So we're quite excited about that as well.
What's your view on the Aβ secret sauce in terms of subtype affinity? Like do you want to just be a plaque buster, a soluble Aβ ?
My take is I think if you, in order to be efficacious, if you look across the clinical molecules in Alzheimer's disease, in order to be efficacious, you probably need to bind plaque and have effector function. And you have to have both of those things in order to really have the most robust activity. And so in terms, you know, whether you need to be plaque specific versus just binding plaque, I think that that's, you know, I just say less clear from the publicly available data. But I do think effector function is important. And, you know, we are excited about, and I mentioned this a bit earlier, about the ways that we feel like we can retain effector function safely on our TfR molecule.
And we feel like that's going to differentiate us from, you know, some of the other competitors in this second generation TfR Aβ space.
Do you believe this idea that with the TfR-based approach, you can decouple amyloid-loading and ARIA?
So we do. And I mean, we have shared some preclinical data that suggest that. And I think the data with the Tronte molecule from Roche also, you know, suggests that that's possible in the clinic. What we were able to show in a mouse model of ARIA was that even when you dose the molecules to, you know, achieve like similar doses, so much higher brain concentration of a TfR-enabled Aβ than non, that you end up getting significantly reduced ARIA. And we think that that's due to a couple of things, the most important of which is biodistribution. So of course these Aβs , they do get into brain, the first generation molecules, they are able to engage plaque.
But in order to engage plaque, you get high concentrations of antibody in perivascular regions, which is the exact regions where you might see ARIA-like signals, right? So you're loading up that response to antibody in those regions. Whereas TfR is delivered to the brain via capillary mediated delivery. And what we see preclinically is a very different distribution. So you don't get that perivascular distribution. And we think that, you know, that is one of the major drivers behind the ARIA signals that are significantly reduced with our molecule compared to the first generation of Aβs.
So there was a. I think this was in a BioCentury article years ago, but there was an interview with George Scangos about Aducanumab after the phase I-B data. And he talked about the investment they made in Aducanumab just kind of going right into phase III and all the manufacturing and everything. And I think you're talking about an investment that was over $1 billion. So do you feel like this is something that Denali can do as a standalone company? And at the very least, how far would you take this independently?
Yeah. So great question. Maybe a couple different answers to that. I think first in terms of what it takes to really understand whether you have a molecule that works. The benefit of Aβ is what you have is a target that has essentially been somewhat validated in Alzheimer's disease along with a clinical development path that helps you get there. That's coupled with a first generation of anti-Aβ molecules that are effective, but not as effective as people would like. So it creates this unique opportunity. And so in terms of our development of Aβ, I think we can leverage some of the tools that have been built through ADU and some of these other molecules in terms of understanding things like rate of plaque clearance, as well as any impact on ARIA or vascular abnormalities.
Those are things that can be assessed in pretty early stage clinical trials prior to making a big investment.
Yep.
That being said, I think we're, you know, we are always open to potential partnerships. I think that's particularly true for Alzheimer's and other larger indications where, you know, potentially tapping into the resources of a large company with an existing infrastructure can be a benefit. So, you know, I think that there's potential to, you know, continue to generate and value ourselves and really bring that molecule into a position where it's clear the impact it's going to have for patients on our own, but also something that we're open to potentially in the future in terms of partnerships more broadly.
Okay. Great. In the last few minutes, should we talk about the eIF2B program?
Sure. Let's do it.
Okay. Great.
Okay. I mean, eIF2B, so.
B, not beta. My bad.
Yeah. It's fine.
Due to the, you know, the Latin letter. Yeah. Go ahead.
Yeah. It's a very, very, very exciting program for, you know, it's currently part of the Healey study, which is a platform trial for ALS, and the way that study works is it has shared placebo arm among a number of different therapeutics, and so that's being currently run by MGH.
Can I ask you about that study?
Sure.
Just like looking at some of the products that have been evaluated, looking at like, you know, there's some of the data that's been public, there's been some pretty significant imbalances between the control arms, the drug arms. I think one of the drugs that showed a signal in that study has a very atypical mechanism of action, right? Like how, I mean, I know, look, they're your partners, right? It's a very legitimate academic center, but as it just relates to the quality of the data you're going to get from that trial, like are you comfortable this is going to be clear-cut proof of concept or not?
Yeah. So I, you know, I think we do have a lot of confidence in the Healey team. If you look at who's participating in the trial, it's some of the top ALS experts in the field, those that you may also tap into if you're running your own study. I think one of the limitations that we've been open about with the Healey study is it is a six-month study. We would ideally like for the study to be a bit longer in order to, you know, really bring out some of those clinical outcomes we are interested in. One of the challenges there is, you know, Healey has been so successful with the ALS patient community, it becomes pretty challenging to compete with Healey and run your own study. And so we, you know, we made the strategic decision.
We felt like this was a, you know, time-efficient and resource-efficient way to generate, you know, a data set to understand the proof of concept of this molecule. And what we were able to do is really tap into the infrastructure that Healy already built, right? So as a company that, you know, hadn't run a trial of this kind before, that enabled us to move much more rapidly and enroll the trial very quickly, which were advantages, right? So I think that there's pros and cons there, but, you know, we feel like we're pretty comfortable with our decision to do that and looking forward to that data readout.
So when's that readout?
So MGH is running the study. I mean, it should come soon. We're not guiding on exact timelines. And that's because it's really been driven by MGH. We are, of course, in close contact with them, but, you know, they're really driving that data analysis. And so we'll look to them for the exact timing there.
Okay, so in the last few seconds here, Joe, anything else you'd like to add?
No. I mean, I think we're, you know, renewed focus, you know, on our TV platform. I think, you know, Denali, we're, you know, excited about the future of our portfolio. We're well capitalized to be able to execute on our three TV franchises. So our enzymes are ETV franchise, our oligos OTV, and our antibodies ATV. We think that there's potential across the board there. And that along with, you know, the filing for our DNL310 program early next year is definitely an exciting time at Denali right now.
Great. Well, thanks for your time. Interesting conversation.