Voyager Therapeutics, Inc. ($VYGR)

Great. Thanks very much. It's my pleasure to be moderating this chat with the CEO of Voyager Therapeutics, Al Sandrock, who sure many folks know well from his current role and famous roles as prior. Al, maybe we can get into it and talk more about Voyager and some of the key events this year for you guys in the broader neuro space. If you want to just give a quick snapshot of 2025 as a year for Voyager and the key things that people should be focused on, and then we'll do more specifics?

Yes. So first of all, we have -- we've been calling it the year of tau. We have 2 assets directed against tau. We have -- 1 is an antibody, a C-terminal antibody that we expect to get tau PET imaging data by the end of the year in a multiple ascending dose study. And then the second is the gene therapy tau knockdown asset, very much akin to BIIB080 which, by the way, has a pretty important readout coming up midyear time frame. Second sort of pillar of value would be that we get -- this is the first year that we put our capsids, our newly discovered BBB penetrant capsids into the clinic. And we have 2 assets doing that. One is the tau knock down gene therapy, I just mentioned, but the other 1 is a neurocrine partnered program for Friedreich's ataxia. So we should be getting set up, if you will, this year for getting a proof of concept that our capsids can work and produce gene expression broadly in the brain. And the third is that we plan to show more data on our shuttle platform that we're developing which is kind of an outgrowth of our capsid discovery platform. So 3 important pillars, if you will, of potential inflection points this year.

Yes. Okay. Great. All right. Well, let's start off with the year of tau. When you think about the scientific evidence behind tau versus a beta, maybe 2 questions. One, can you sort of help contrast for people where each plays a role in the driver of disease? And then second, like how clear is it that tau is truly a disease driver and not an innocent bystander?

Yes. So that's a very important question. So first of all, if you look at human genetics, it turns out that it's really hard to implicate tau based on human genetics for Alzheimer's. In fact, we do know, of course, that mutations in tau can cause neurodegeneration, but at least to diseases like frontal temporal dementia, for example. So the genetics would not necessarily imply that tau is important. But I would look at a lot of other data, including human data, in particular in, for example, in the South American cohort, the PSEN1 mutants. So over 1,000 of them in Colombia. And if you look at those patients, they all get demented roughly in their early 40s, and they develop a lot of amyloid accumulation and then they get tau progression. But there were a few outliers, people who did not get demented in the early 40s, even though they carried this highly penetrant mutation. And when you look at those patients, it looks -- they don't get demented until they're 70s. It turns out that they have a brain full of amyloid, but they didn't get the normal progression of tau pathology, which strongly implicates that the real cause of dementia is actually not amyloid, but it's tau. So a lot of people say, tau is better correlated with cognitive decline. And I think those are clear cut examples of that. And so I believe that -- so when you look at natural history studies, whether it's in sporadic Alzheimer's disease or familial Alzheimer's disease, the first abnormalities are in the A beta pathways. You get abnormalities in A beta, in the brain, include biomarkers. And then later on, you get tau accumulating and spreading in the brain. And the staging of Alzheimer's is actually based on where tau has spread to. So the back-end brock staging is based on spreading a pathological tau. And it's a very characteristic spreading pattern, follows a very similar course from patient to patient. And my belief is that amyloid is necessary to trigger that spread of tau progression, the spread of pathological tau in the brain. And that -- in fact, getting a little bit of misfolded tau in a particular part of the temporal lobe called the rhinal cortex is actually part of normal aging. So the initial misfolding and hyperphosphorylation if you will, of tau is actually not pathologic. It's part of normal aging, but the spreading outside of that region into the other parts of the brain is what's actually abnormal and what causes the progression of disability -- of dementia in Alzheimer's disease.

Yes. Okay. That's a great overview, Al. So I guess in the context of that, right, the BIIB08 data this year is going to be a big catalyst for the space, especially for your gene therapy program. Do we have anything to look to that can kind of give us a guide on how much we might need to lower tau for clinical benefit?

Well, that's 1 of the things we need to learn more. In the case of the amyloid treatments, we had -- there was a lot of sort of trial and error, if you will. The initial anti-amyloid antibodies did not show a very significant effect on the cognitive endpoints. And that's -- and so we had to learn that epitope mattered, that the stage of disease mattered, that it was important to screen to make sure patients actually had Alzheimer's and actually had amyloid, et cetera, et cetera. But it took a lot of early failures and eventual successes to learn how much amyloid reduction you need to get a clinically significant effect and it turned out to -- there had to be a substantial amount of lowering. And so we're in the early stages of tau now. We're having failures. And basically, the antibodies or the treatments that actually didn't lower the -- or it didn't affect the spreading of tau. That's not very helpful, right? So we know the epitopes that led to those failures. So we're learning about which epitopes matter and which don't matter. But what we don't have enough of is examples of drugs that help -- that actually have effects on tau spreading and to know what the relationship needs to be between how much tau lowering or how much blockade of tau spread, how much is necessary to produce that cognitive outcome that we desire.

Yes. Okay. Very interesting. For your siRNA gene therapy and then we'll talk about your antibody, but maybe just talk about your confidence that with this gene therapy, you can get robust knockdown that's in the 60%, 70%, 80% range in the right areas to bring that matter to Alzheimer's?

Yes. So we use these novel blood-brain barrier penetrant capsids, precisely because we get broad brain distribution of our gene therapy. And I think that's what you need. The tau is spreading from 1 cortical region to the other. So the key thing to hone in on is what are all the cortical regions that the gene therapy can get to. And how much does it lower the expression of tau in those regions. And we've looked at hippocampus, central rhinal cortex, temporal cortex, frontal cortex. Essentially, all the cortical regions get affected roughly similarly in the sort of 50% to 75% range. And that's very similar to what you get with BIIB08. So we're kind of trying to replicate what BIIB08 does except for the onetime IV gene therapy where the siRNA is actually produced by the cells in the brain.

And from a development perspective, so let's say the BIIB08 data are positive. How do you think about the development path here? And we've obviously -- and we might be in a more fluid situation with the FDA and leadership over the next few years, so we'll see how it goes. But for right now, right, it feels like we're in a -- maybe a more stringent FDA environment with gene therapy, it's hard to say. So what's your thought on the development path? And kind of the safety database or safety evidence you might need to get a product like this for market -- to market for a non-orphan disease?

Yes. So we've actually had couple of touch points with FDA on this tau knockdown gene therapy program. We had a pre-IND meeting about a year ago. And then we asked for and received a Type C meeting more recently. Because we wanted to get their latest thinking on how to do the studies in the safest possible way. Any other learnings that -- because FDA, they know everything that's going on in the U.S. They can't necessarily tell everybody everything that they've learned. But I think when you go with the program, they share a very important information and so we actually didn't have to, but we decided we would have another Type C meeting. And I can say that we're having -- we're not having any difficulty interacting with the FDA, and we're getting a lot of really good interactions -- helpful interactions, I would say, with FDA. So with that as a backdrop, I'd say, look, our first step is to see if we can lower tau in the brain. And again, we're going to rely on tau PET imaging primarily. And here, as opposed to blocking the spread of pathological tau, we're going to be asking to see whether or not we can reduce from baseline of pathological to, which is what BIIB080 showed in their Phase I trial. That's a pretty striking result that you get a reduction from baseline just by blocking the synthesis of new tau, which suggests that the parent helical filaments is or pathological tau, it's not like -- it's in some sort of dynamic equilibrium because if you block the production of tau, you actually remove it seems the pathological tau, which is pretty striking.

Yes, absolutely. Good. Well, let's talk about the antibody now. So we -- I think the oversimplified negative case on the antibody approach is that all these antibodies try to stop the spread of exosome or tau, but the vast majority of it is inside the cell. But you were alluding to kind of different binding domains, different properties, these antibodies. So what's the case for why you think your antibody could actually show a change on tau PET where others have failed?

So first of all, we did see a hint of 1 of the antibodies actually working. So the terminal antibodies both failed, the Biogen one and the Lilly one, but they didn't do anything on tau PET imaging. But the Bepranemab, the UCB antibody did affect the spread of pathological tau. So for the first time, an antibody was able to do it, that was a mid-domain not a terminal. Now, so we had hundreds of antibodies to choose from. And at Voyager, we decided to focus on 2 things. One, we wanted to get antibodies that were specific for pathological forms of tau, because the Priam-like hypothesis, if you will, of how tau spreads requires that there's a pathological form of tau that gets secreted that it's in -- and hopefully, it's in the extracellular space accessible to the antibody and that, that pathological form of tau gets into the neighboring cell and in a template-driven manner, usurps the normal tau essentially informs pathological tau that neighboring cell. So we wanted to catch it while it's moving from 1 cell to the other. And so we said a, we want something that's specific for pathological forms of tau. By the way, in the case of anti-amyloid, that turned out to be really important, that only the antibodies that bind the pathological forms of amyloid actually work. The ones that don't do that don't work. The second was that we relied on this animal model, where it's a mouse that expresses human tau, and we inject Alzheimer's brain material into the mouse. And we look for the spread in the brain of the mouse. And we look for the antibodies the most robustly, that the antibody, the most robustly blocks that spread and most reproducibly. Now we could have chosen a mid-domain antibody ourselves. We had 1 in the mid-domain that was pathologic tow specific, but we didn't choose that. We could have chosen 1 in MTBR. But -- and by the way, Interestingly, none of the end terminal specific antibodies that we had blocked the spread. So that animal model correctly predicted the failure of the anterminal antibodies, which I thought was pretty interesting. It actually also correctly predicted that Bepranemab would block the spread somewhat. But ours was better. And so we relied empirically on this animal model because there's really no other way that we thought where you could choose among the handful or so of pathologic specific anti-tau antibodies.

Right. Okay. So you have your MAD data coming up later this year, correct?

That's right. .

So what do you think is a realistic goal for what you could see on tau PET?

So what we're hoping to see is an effect that's better than Bepranemab because Bepranemab, by the way, not only did a block to spread of tau by tau PET imaging, it actually had an effect on ADAS COG. It was a prespecified secondary endpoint, and it was not a subgroup analysis. So it actually did show some evidence of clinical efficacy. The problem is it failed on CDR sum of boxes, which was the primary endpoint. So I had mixed results on cognitive or cognitive/functional endpoints clinical outcome measures. And we know that CDR is what FDA is going to want to see in a pivotal trial, right? So what we hope to see is that we have a bigger effect on the spread of pathological tau by tau PET imaging than Bepranemab had and therefore, have a more clear-cut effect on clinical outcomes. Of course, we want our studies too small to really evaluate clinical outcome measures. But we're hoping that if we have a bigger effect on impeding the spread of pathological tau that we will eventually have a bigger effect on the clinical outcome measures that are required for approval.

When in the disease, if you're trying to stop the spread, when in the disease do you have to intervene? Do you have to intervene even earlier than you might within ASO, which seems to be blocking the production and clearing intracellular or is that the wrong way to think about it?

No, I don't think that's the wrong way necessarily. So -- but as I mentioned, by the time you have even mild cognitive impairment, you've actually reached the maximum of amyloid burden in the brain, ironically. Even though it's very -- still very early, amyloid is maxed out by the time you're even at MCI. Tau is different. Tau is still progressing and hasn't quite reached the maximum amount of pathological accumulation or spread in the brain, even at the MCI stage. So I do think that the -- and we've done trials now in MCI and early mild patients. I think that's a pretty good population to target for tau. In other words, you don't have to go as early as you might have to with anti-amyloid. Now we'll learn later this -- maybe in a year or so from other studies, Lilly and Biogen, whether or not earlier anti-amyloid treatment, even before you have mild cognitive impairment, could actually be more effective. That's a pretty important study. My prediction is that it will be more effective because there it hasn't quite maxed out yet. But the BIIB080 or the knockdown treatments, to your point, may not need to go as early as MCI. So even if you progressed into the mild to moderate dementia state, maybe even moderate, I don't know, but if you can remove the pathological tau, it has a chance of helping those patients.

Yes. Okay. Great. So maybe take a step back, Al, and just talk a little bit about like the novel capsid platform, right? Because we can talk about this through the lenses of tau, different ones but platform more broadly. But when do we get proof of concept, right, that this thesis is correct around you can give these capsids IV, you get greater, greater brain penetrants, wider therapeutic index, like when will we get the answer there?

Yes. So we have 2 shots on goal this year. So we start with -- so we have our tau knockdown gene therapy. We're planning to file an IND in Q2 of this year. And we hope to be in the clinic in the second half of this year. So that's our own wholly owned program. And then Neurocrine has said that they are planning to file an IND and be in the clinic this year as well with 1 of our novel capsids for Friedreich's ataxia. So that's 2 sort of shots, if you will, on human proof of concept. Now in the case of gene therapy trials, you have to do the studies in patients with disease. And you have to use doses that have a chance of helping the patient. It's considered unethical to do anything else. And so we should be able, at some point next year, I would say, to be able to start to get evidence that we can dose safely and get gene expression in the brain. And I think that's going to rely primarily on fluid-based biomarkers initially. Ultimately, as I said, tau PET imaging, but that's going to require some more time.

Yes. Okay. Great. And you're also working on a brain shuttle as well using ALPL?

Yes. So we have a brain so these capsids. It turns out that these capsids are a great way to probe for receptors that can be leveraged to get things into the brain. And so we've leveraged those capsids as we've now discovered 5 to 8 receptors, many of them were actually surprising. I mean you would never have guessed that they would be receptors for BBB penetrant. But it turns out -- and then the first 1 of these is ALPL. ALPL is kind of surprising. It's a GPI linked protein, you wouldn't necessarily think that would be the greatest way to get across the BBB, but it turns out it's a very good way to get across the BBB. We've now made ligands against ALPL. We're going to use them as shuttles much the way people do with TFR transparent receptor.

Right. Okay. And what do you see as the most interesting potential applications of that technology?

So far, our ALPL shuttles are quite differentiated from the TFR-based shuttles. So we have very different PK. So we have a much longer half-life. So if you want to block something, for example, 24/7, most of the time when you block, you don't -- you want a 24/7 coverage, right? TFR has a short half-life. By a week, usually, it's gone unless it's bound to something in the brain. Ours will be -- even 3 weeks later, we still have very nice exposure. So we haven't even calculated the half-life yet, it's so long. So that's a nice characteristic for many applications. It also has no effect on any of the hematologic adverse events. We would like actually looked at reticulocyte count. We have no effect. Of course, we may have our own safety issues. So we're going to be looking carefully at that this year. We're doing nonhuman primate studies. And so we -- and so -- and also the distribution is different. TFR distributes broadly into many other tissues in the body, not just the brain. ALPL has a different distribution pattern. So -- and we have, like I said, 5 to 8 different receptors. Each 1 is going to have its own safety, distribution and pharmacokinetic profile. And I think that's going to give us a lot of flexibility in choosing the right shuttle for the right application.

Yes, makes sense. What would be the safety risk with ALPL, if anything? Like do we -- is there anything from genetics or loss of function or anything like that?

So in humans, if you have severe loss of function, so greater than 70%, you can get -- there's a disease called hypophosphatasia, which is bone and teeth mineralization issues. So -- but the human genetics would indicate that it has to be a severe loss of function. And so hopefully, we'll thread the needle so that we don't have that much loss of function.

Right. Okay. And from a company building perspective, Al, like how much do you want to take all this stuff forwardly yourself versus -- because I'm sure you could do BD around, a novel shuttle or different products like that. You've obviously done some with Neurocrine, but what's sort of the 2- to 3-year vision from here?

Yes. No, you can expect us to -- we're always talking to potential partners. If you look at what we did with capsids, we did a number of partnerships. We have a total of 5 programs right now with Neurocrine. We have 3 programs with Novartis, 1 with AstraZeneca because they inherited the Pfizer rare disease portfolio. And so I like these -- so look, if our shuttles work, there's no way a little Voyager can do all the potential targets, my goodness. It would be hard even for a large company to address all the targets, right? So the way to do that is to leverage partners. For example, in our -- with our Neurocrine programs for FA and GBA, we actually not only -- they do all the heavy lifting essentially to get to an IND and even to Phase 1. And then we have the ability to opt in, 40% of U.S. rights for FA and 50% for GBA1. Those are really good not only for patients because it gets these programs into the clinic, but it's good for our shareholders, too, because not only do we get the milestone payments and royalties potentially, but we may be able to opt in for a bigger chunk. So those are the kinds of things that I'll do all day long with the right partners. And I believe Neurocrine, Novartis and AstraZeneca, they're very good partners. They know what they're doing, and we learn from each other.

Yes. Okay. Great. Well, thank you, Al. It's great conversation, as always. And yes, best of luck this year. We're all rooting for [ you, Al ].

Thank you very much, Paul. Thanks for having me.