Voyager Therapeutics, Inc.

Happy to be here and moderating this chat with Al Sandrock, CEO of Voyager. I'm sure everyone knows Al who's here listening in and got to know Al at his days at Biogen. So Al, maybe just give a quick overview of Voyager and then we can do a fireside chat and dive deeper into different programs.

Yeah, great. So we are a multi-modality neurotherapeutics company where we're trying to optimize delivery. We have two platforms. We have a gene therapy platform platform where we're discovering capsids that cross the blood brain barrier after IV delivery. And that platform identifies not only the capsids, but then the receptors that the capsids leverage to get into the brain, we're now gonna be looking to see if we can use them as shuttles. And the first one of these is called ALPL, so you see that's already appearing on our pipeline chart there. So the idea is that these are validated receptors in the sense that we know they can carry large viral particle across the BBB. And so we're gonna make ligands against these receptors and conjugate oligonucleotides, put them on various protein therapeutics, and optimize delivery. So multi-modality focused on optimizing delivery. We have a heavy emphasis on Alzheimer's disease, as you can see. And we have multiple partner programs with some great partners, Neurocrine, Novartis, and AstraZeneca. And so we're heavily into gene therapy. And I guess I should end by saying we have one program that is in phase one. It's in a multiple ascending dose study, the anti-Tau antibody study. which we expect to read out next year. Maybe of note is that there's a lot going on in TAO, as you know, Paul, and we think that not only our program, which we'll read out next year, could be an inflection point, there's J&J, that has an antibody that we expect to read out early next year. And then there's BIB80, the biogen antisense oligonucleotide that we expect to read out in mid-year. I think both of those could have read-through to our programs because we also have an antibody and an siRNA tau silencing gene therapy as well.

Yep. Okay. Great. Well, on the antibody side, I think some people, myself, someone included, have interpreted the failures of other antibodies as maybe concerning that maybe the antibody strategy for tau can't really access the majority of the target. What would you say to that? And how do you think the shot on goal here is fundamentally different from what hasn't worked?

Yeah, so it's funny because when I first started as CEO of Orger, I was asked. Actually, I was still on the board. I wasn't even CEO in my first meeting. I was asked what my opinion was of the tau antibody. Todd won't remember that meeting well. And remember, I had just come off a big failure with the biogen and terminal antibody. And so I was all set to kill it until I heard a couple of things. First of all, that the animal model that Steve Paul brought in from Cornell was used to pick this antibody. It's a pretty intriguing model. It's a mouse model that expresses human tau. It's a P301S human transgenic mouse model of tauopathies, essentially. And what you do is you take Alzheimer's brain derived perihelical filaments and you inject it into the brain in one region and you look at the spread of tau. Because after all, what's pathologic in Alzheimer's is the spread of tau out of the temporal lobe. In fact, we all get a little bit of pathological tau, if you will. I use that in quotes because if we all get it, it can't be pathological. But in normal people, it stays in the renal cortex. It only starts to spread in the presence of amyloid. So it's the spread of tau that's abnormal. This model, we hope, will predict whether or not antibodies work. The two N-terminal antibodies that failed in the clinic failed to block the spread of tau. In that animal model. In that animal model. In fact, in a side-by-side study, we had essentially a zero effect of one of the N-terminal antibodies, and we had a 70% effect blocking the spread with our antibody. I thought that was pretty interesting.

Yeah, that is interesting.

Because, you know, the issue always is, well, we had... So the other thing we did was we chose an antibody that was specific for pathological forms of tau. In the case of amyloid, that was important, that we had antibodies that were specific for pathological forms of amyloid. But even that, we had half a dozen or so antibodies specific for pathological forms of tau scattered across the tau molecule, and we had to pick one of those. And so I used this, we used this animal model empirically, thinking that whatever mechanisms mediate the spread in humans might be replicated in that mouse model expressing human tau.

When exactly does that spread start? And what does that tell us about the window for intervention for the tau antibody strategy?

So if you look at the natural history of Alzheimer's disease, the first thing that happens is abnormalities with amyloid. And then secondarily, you see an effect on tau. If you look at normal human aging, as I said, we all accumulate a little bit of misfolded tau in the part of the temporal lobe called the renal cortex. And it's only when you get amyloid that you spread beyond that region. For example, the Colombian cohort of PSEN1 patients. It's about 2,000 or so of those patients. There are people who have a brain full of amyloid who actually, a few people, who don't get demented. In fact, in that cohort, the natural history is that you get demented sort of in your mid-40s, plus or minus just a year or two. So it's very, very regular. It's a single gene mutation. A few patients actually don't get demented until their 70s or 80s. And when you look at those patients, They have a brain full of amyloid, but they actually don't spread tau normally. So in that situation, it looks like it's the spread of tau that's more critical for causing dementia. It's data like that that makes me think that the spread of tau is actually pretty critical. And now we're starting to understand how amyloid can trigger the spread of tau, but that's a whole other story.

Okay, so what's the development path for actually showing in people pharmacodynamically that this antibody is different than what's been tested before?

Well, so there's two types of measurements for tau. There's fluid-based biomarkers and tau PET imaging. We're going to rely more heavily on tau PET imaging. And the main reason for that is that Biogen presented some interesting data at the ADPD meeting earlier this year, where you remember they had that N-terminal antibody that failed. what they showed at ADPD this year is that there was a 40% effect on MTBR tau, a fluid-based biomarker. And MTBR tau was the one that people were most excited about because it correlated the best with tau PET imaging. But in that trial, it was a 40% effect on MTBR tau, no effect on tau PET imaging and no effect on clinical measurements. The lesson that I learned from that is you've got to be a little cautious with fluid-based tau biomarkers. I don't think we know enough about them. I don't know that we know enough to rely on any one of them for sure. But tau PET imaging does look like the real thing.

And you'd be focused on specific brain regions. The idea wouldn't be you're removing the tau from where it already exists.

Bib80 actually removes tau from where it already exists. Which is pretty darn remarkable. But here what we're looking at is We're going to have people at various stages, Brock stages, stage 2, 3, 4, etc. What we're going to have to do is look where is it predicted to go next. If you start at stage 2, you're predicted to go to stage 3. What we're looking for is whether it can impede developing into that next stage, if you will.

Yep. Yep. Okay. And so is that the kind of question you can actually answer in a phase one trial? Or is that?

Yeah, to work? We believe our trials well powered to see an effect on on tile spread based on top head imaging using that kind of?

Okay, super interesting. And if that's positive, what would you do with that asset? Would you actually try to take that fully forward through a to be your partner at

Well, you know, we're too small, I think, to even think about commercializing it for sure. And if you're not going to commercialize it, it's better if you have a partner do phase three. And so the question is whether or not we would have all the questions answered to go into phase three. Right. Some people would say we would, potentially, because we're doing more than one dose. So you have to know the dose. And yeah, so I think we would be looking for a partner.

Yep. OK. OK. And then on the tau silencing approach, so I, like you, share a lot of enthusiasm for the Biogen Ionis anti-tau program. So let's say those data are positive. How much read-through and de-risking would that have on something like 1706?

Well, I think it reads through quite a lot because what we have is a vectorized siRNA. Very similar concept. You're just decreasing the expression of tau. It should affect all forms of tau, intracellular, extracellular. Epitope doesn't matter here, obviously. So I think it reads through quite well. And what we're looking for is do we see again whether or not there's a decrease in tau pet signal. That's the remarkable thing is that, you know, I used to think neurofibrillary tangles were pretty irreversible. and these tau-pet ligands are specific for pathological forms of tau, to see a decrease relative to baseline must mean that there's some equilibrium, if you will.

There's some sort of endogenous clearance mechanism. Yeah, because if you just block the synthesis of new tau... No, that could have easily just looked like a base inhibitor for tau, right?

Yeah, exactly, but that's not what we saw. And then compounded with that, the effect size, if we can believe... comparisons to natural history and to the placebo groups of other trials. It looks very interesting because the effect size is pretty large. Right. So both of those, the imaging and the clinical outcome measures, portend well for pretty interesting effect, I think.

Yep. So, I mean, I think I can understand that the gene therapy approach would have tremendous advantages just from like a patient convenience and access perspective. On the other side, like across the gene therapy space, right, it feels like, you know, every other program there's some sort of inflammatory sae and for certain rare diseases that kind of risk is acceptable not saying alzheimer's is not a terrible disease but are we ready for a gene therapy that could target this big of a population like if you run a program that has you know 500 600 person safety database like are you worried that you're bound to see something that could kind of shift the risk benefit against you yeah

Well, first of all, if you look at the data we showed in non-human primates, we use a relatively low dose. Yeah. Where are you relative to others? We're at 1.3 E13 VGs per kg, which is essentially an order of magnitude lower than the E14 VGs per kg doses that are typically used for systemic ABV. So we're an order of magnitude lower.

That's because of the BBB penetration.

Yeah, it's a very potent and so we get 50 to 80 percent knockdown across the brain with that one injection of that one pretty low dose. The second thing is that we use a capsid that's detargeting the liver by 30-fold. So not only is it a lower dose, we detarget the liver. So as you know, other companies have shown liver toxicity, including that it could cause death. So look, I agree that safety should be very good. But on the other hand, Alzheimer's is a pretty bad disease. I mean, I think we often think of it as, huh, but it's fatal. And it's pretty bad before you die. So it's always a benefit-risk.

Do you think we are out of the woods on any on-target risk with knocking down all forms of tau in all areas of the brain, including quote-unquote healthy tau? I don't even know if we know the function of tau in adult brains, but how do you make sense of that?

Well, there are some conditional knockout experiments that have shown some subtle differences. So first of all, even just regular knockouts, the animals are surprisingly viable. You would have thought that... The full knockout? Yeah, the full knockout. I mean, knockout from embryogenesis. the animals are actually viable. They're a little smaller. They're even fertile. And so that right off the bat tells you there's probably some redundancy. But if you look carefully, there are some problems in the brain. So now we have to turn to conditional knockout, which is more similar to the situation that we have here. There are some subtle differences in very specific parts of the brain. You know, that's something we're going to also learn a lot from BIB 80, because not only have they been following those patients from the phase one trial for many years now, but we have hundreds of patients now exposed to BIB 80. And so the safety piece of that trial readout is just as important, I think, as the efficacy. That's where the antibody may have, I mean, look, Bipranumab was incredibly safe. It did block the spread of tau. And look, if you have an antibody that's specific for pathological forms of tau, you would predict that there's going to be less safety liability. Of course. But of course, it has to work.

Right. Yep. Yep. OK. OK. So what's rate limiting for 1706 for getting that into the clinic? go do the usual GLP talks and get the manufacturing ready to do our first in human studies but we we are on track and we expect to do that next year yeah well maybe let's like use that program as just sort of an opportunity for you to go a little bit deeper into the mechanism by which this crosses the BBB I know you kind of you know first screen the capsids and then work backwards on mechanism but what have you learned on how this works and I think you and I have had these conversations around the analog to transferrin. With transferrin, there's this concern around interfering with the receptor, and how do you kind of toy with that? Is that at all a concern here with ALPL?

Yeah. So we have disclosed that the capsid we're going to be using in this program is an ALPL utilizing capsid. So with transferrin receptor, we have hematologic adverse events. We don't see that with ALPL. We don't see it with our shuttles. And we don't think we're going to see it with our gene therapies as well. Gene therapies is once and done, essentially. So even if you have an effect, it's going to be very temporary. Now, if we make a shuttle, though, we're going to have repeated dosing. And that's where we have to be cautious about potential safety issues. And that's something we're going to learn about now. you know, the advantage of gene therapy is that capsid is employed, gets across the BBB and the capsid itself has gone pretty rapidly. Right. So, so I, you know, I think that the liability here is, is, is low relative to sort of the endogenous function of ALPL. Yeah. That's my guess.

Yep. Yep. Okay. Okay. Great. What do you want to talk about next?

Whatever you want to, man. I mean, look, we have a bunch of partner programs. I know.

You guys are more limited in what you can say in those, but I'm still going to try. For taxon gene therapy, I think it's tremendously interesting. I agree. Now, can you talk about the capsid? And do you think that this is the kind of thing where you can treat both neurological and cardiac in one? And can you do that safely? Yeah.

Yeah, so this is Neurocrine's program, so I have to be a little cautious. But yeah, here we need a capsid that crosses into the heart and into the brain as well. And so it's likely not to be the same capsid that you would use for purely brain disorder, right? I think that... And by the way, AAV9 or AAVRH10 or RH74, they actually generally work in the heart unmodified. So AAV, sort of the natural strains, if you will, transduces the heart pretty well. What we've done is then discovered caps that also gets into the brain. Because I think you have to do both. And I'm excited about what Lexio has told us about the path to approval. Yeah. I mean, you know, right ventricular biopsy and left ventricular mass, right? One, a single arm trial, natural history comparison. I'm pretty excited. I don't think they're gonna get very much brain transduction though. And it's called Friedreich's ataxia for a reason. The disability, especially in children and teenagers, is the ataxia. Now the cardiomyopathy is what kills the patients, but usually later in life. So both are important. And so ours would be differentiating, I think, in the sense that we would not only have the cardiac effects, hopefully, but we would also potentially have the brain effects as well.

You know, maybe it's premature to ask this question, but just when you think about your TPP, AgCover Lexio, I think their data is really promising. And I actually think there's a lot of scientific support in this disease that you may only need a little bit of frataxin.

In fact, you don't want too much either.

Right. Right, exactly. Definitely not in the liver, too. Now, they're making, I think, in the kind of low to mid single digit frataxin amounts. Like, do you have anything from your animal work that... predicts how much you might be able to make in the heart and the brain with kind of the human dose range you predict?

Yeah, we do. But you can't say? Yeah, I think now I'm getting into neurocriminology territory, but I think there are ways of predicting where you're going to land in humans based on animal work.

Yeah. Okay. Okay. And GBA one gene therapy, Gaucher's Parkinson's makes a lot of sense. If you're successful in Gaucher's with something like this, how de risking is that actually a Parkinson's?

Well, that's an interesting question. You know, 10% of patients with Parkinson's have mutations in GBA1. And there is data, even in sporadic PD, that there are lysosomal problems in perhaps not every patient with Parkinson's. But I think there's pretty good data that suggests that G case, is involved in Parkinson's. I think one of the main reasons for that is that we know that alpha-synuclein metabolism is partly controlled by enzymes in the lysosome like GKs. And so, since I believe that alpha-synuclein is central to Parkinson's disease, and since lysosomal enzymes like GKs can affect the metabolism of alpha-synuclein, I think there's a pretty good chance that it could be effective in Parkinson's. I'm looking at my CSO, Todd, to see if he's agreeing with me or not.

Okay. Now, from a delivery perspective, the deep brain or the structures involved in motor function have been harder. You know, I mean, I think, like, for the ASO LARC II program, like, Can you deliver there with your approach?

So we've already shown this, I believe. We look at pretty much every brain region. We look at 14 regions, I think, when we do. My brain only has six. Well, we look at 14, including the spinal cord, of course. So we leverage the vasculature. you know, with IV delivery, and we get transduction in the deep gray structures just as well as we get with the... In fact, in the nigrostriatal neurons, have we shown this data, that we have 98% of neurons transduced? Of the dopaminergic neurons transduced. That's amazing. Yeah. And in the caudate putamen, we get very high levels of transduction. So leveraging the vasculature is actually a pretty good way to get down to those deep gray structures.

Right. Right. Right. OK. OK. Interesting. Maybe taking a step back, a hot topic that is not directly impacting you right now, but maybe will someday. you know, how are you kind of making sense of maybe evidence on both sides of the argument around CBER and how flexible they'll be in gene therapy?

Well, I mean, we all saw what happened with Unicure, right? And I mean, I wasn't privy to all the conversations they had with FDA and everything. You know, listen, I do think that for these rare diseases in particular, It's kind of hard to do well controlled trials typically. So I would hope that they stick to their previous, what they said previously, that relatively small single arm study with natural history comparisons are gonna do the trick. I'm always of the opinion that if the effect sizes are large on hard endpoints, it's kind of hard to argue that the drug works, right? But there is this idea that with conditional approval, you're willing to accept that some of the time you're gonna approve drugs that could, could be ineffective. But you have to do the confirmatory study, right? And so I think our, I'm very proud of the fact that our industry has shown that, you know, for example, the ALS drug. I mean, those guys from Amalix were... Amalix, I mean, look, they showed what can be done. They got the drug approved based on accelerated approval. The confirmatory trial did not confirm. They took the drug off the market. The more we do that kind of thing, do the right thing in our industry, the more likely we are to get things like accelerator approval. And I think these rare diseases, particularly these horrible childhood diseases, I mean, I would hope FDA bends over backwards.

Yep. Yep. So for Voyager in 2026, how many more clinical stage programs do you expect to have?

Well, we're certainly expecting the tau silencing gene therapy in the clinic by then. We hope Neurocrine is in the clinic. We do have the ability to opt-in after phase one. I don't think we'll be ready to opt-in next year because we'll just be getting started, but at some point we may be able to opt-in to those. I don't know, the tau antibody, if that's positive, that'll still be in the clinic. We'll be looking for a partner. Yeah. Is that all, Trista? Did I get them all right?

And then one last thing. You have APOE gene therapy. How similar or different is this to what Lectio was trying to do there?

Well, here, they had three different programs. I'm always confused as to which one we're talking about.

Right, so we never saw the program with the Christchurch mutation, I believe. Right?

I never saw it. I knew that was one of the three. That was the third one. That was the next gen. I think, yeah, so what we're trying to do, I don't know what they're doing exactly, but... What we're trying to do is to decrease the expression of E4 and increase the expression of E2. E2 is considered the protective allele, while maintaining overall levels of ApoE constant. And so we're trying to switch from the harmful allele to the protective allele. And the reason why we're excited about this.

Are you silencing E4?

Are you trying to help compete E4? We're trying to silence E4. And we're going to increase APOE expression, essentially. OK. So if you're a homozygous for E4, you have a 90% likelihood of getting Alzheimer's disease. It's almost like a single gene mutation. And these patients are typically demented in their 50s. So they get it at a younger age. And it's a more rapid course. And we had a patient come in who showed a picture of her family. So she got diagnosed in her 50s. She had a very good job, and she was having memory issues at work. She had to stop working. She showed a picture of her family, and more than half of her family had been affected. And so these are people who've seen their relatives suffer, and they're worried about their own children, too. And so if we can prevent Alzheimer's in those patients by converting from E4 to E2 essentially, that would be a huge positive.

You have a couple minutes left. Anything else you'd be interested in talking about that you're passionate about?

Yeah, well, you know, TDP43 is something we did. And we just talked about that. You know, ALS, you know, we had a program in SOD1 ALS gene therapy that we had to terminate because the payload It was toxic, essentially. TDP43, we think, is central to ALS. Probably 90% of cases, you see certainly inclusions of TDP43 in the cytoplasm.

How much you actually address that with a small molecule? Try to move it back to the nucleus or change it at the RNA level?

Well, so the issue here is that TDP-43 has normal functions. If we get rid of TDP-43, we're in trouble, right? And so T-Bio is making small molecules that affect the condensates. So TDP-43 seems to be sequestered into these condensates in the cytoplasm of cells. And they have a way to screen for drugs, high-throughput screen, to get small molecules that remove the TDP-43 from the condensates so they can get back into the nucleus and affect the splicing. So that, in other words, restore the normal function of TDP-43. I can't think of too many other ways to target TDP-43. uh... you know i mean because you have to be delicate also you don't affect stress granule formation because tdp forty three is often found in stress granules as well and stress granules are important for cells to deal with stress so so we we had very high hurdle for this program And they took all the risk and they met the hurdle. Honestly, I didn't think, I thought it'd be hard for them to achieve what they did. We got very excited when they showed us the data and we paid them a single digit million dollar milestone. And so we're pretty excited about that as well. So we're, look, we are a multi-modality neurotherapeutics company. I wanted to underscore that. We have gene therapy. We have the emerging shuttle platform. We even have a small molecule. We are agnostic as to the modality. We want to go after some of the worst diseases affecting humans, and we want to apply the best modality that fits the target and the disease. And what we're trying to do is optimize delivery. So many examples now, trontinamab, gantanarumab. Avidity and Dyn are dealing with ASOs. Solving delivery for these newer modalities is, we think, is going to be very helpful.

Yep. Yep. Okay, great. Thank you, Al. Appreciate it.

Always a pleasure. Thank you.

Thanks. That was great.