Welcome to the 46th Annual TD Cowen Healthcare Conference. I'm Marc Frahm from the Biotech Team here at TD Cowen. First up this morning, we are really pleased to have with us from Shattuck Labs, Taylor Schreiber, the CEO. He's gonna tell us. The way we're gonna run this is he's gonna run through a few slides and prepared remarks, talking about their program in IBD, targeting DR3. Then we'll do maybe the back half or so of the presentation slot as Q&A. I'll rejoin for that. With that, I'll turn it over to Taylor.
Great. Thank you, Mark. Sorry to keep everybody waiting. As Mark said, Shattuck is focused on a DR3 blocking antibody. DR3 is the receptor for TL1A. These are my forward-looking statements. Just to set the stage for anybody who's not already familiar, I think people are pretty familiar with the TL1A space today, the reason that folks have been so excited about this axis and the reason why there's been so much partnering activity in the last several years is really because of the three data points that have been generated with the leading TL1A blocking antibodies, tulisokibart, afimkibart, and duvakitug, which have all delivered consistent placebo-adjusted clinical remission rates in ulcerative colitis, demonstrating that this mechanism may have best-in-class or best-in-disease induction efficacy with a really almost pristine safety profile.
Some of the KOLs have described this axis as having, like, see without the toxicities that come with JAK inhibitors. Where we come in is that, like all biological pathways, TL1A binds to a receptor. The name of that receptor is DR3. This axis is very simple. TL1A is the only signaling ligand for DR3, and it does not cross-bind any other receptors. From a specificity standpoint, there's no reason a priori to say, "Well, I'm gonna focus on the ligand because that is going to give me a more specific effect." The ligand and the receptor are on equal footing from that perspective. The programs that we'll talk about today will probably mostly be what we call SL-325. This is a human DR3 blocking antibody. This is an Fc-silenced antibody. There is absolutely no Fc gamma receptor binding.
We may touch on SL-425. The only difference between SL-325 and SL-425 is that SL-425 contains a YTE mutation for half-life extension. Then there's a series of bispecifics, all of which target DR3 with 1 arm that we expect will be highly differentiated from the TL1A bispecifics. We're gonna get into this in the Q&A, but just to set the stage, the two main areas where blocking DR3 may provide superior efficacy in comparison to blocking TL1A, 1 has to do with the fact that DR3 is a more stable target than TL1A. Then two , there's an expectation that DR3 blocking antibodies will be far less immunogenic than TL1A blocking antibodies. Every single TL1A blocking antibodies causes ADA in most patients.
More than 64% of patients develop ADA to all of the TL1A blocking antibodies. The reason for this is because TL1A blocking antibodies create immune complexes with serum TL1A. This is an unavoidable aspect of targeting TL1A. You cannot engineer around it. You need to look no further than the TNF antibodies to understand this. TNF-α and TL1A are both soluble trimeric proteins. That means that every single trimer of TL1A has the potential to both be cross-linked by individual anti-TL1A antibodies or to be bound by multiple TL1A blocking antibodies. This is a protein aggregate-driven mechanism of immunogenicity. We're most of the way through our phase I clinical trial. We've finished enrollment in all cohorts of the single ascending dose part of this study. We finished enrollment in two of the three multiple ascending dose cohorts.
We will share data in Q2. We'll probably get into these key questions that are here. I'll leave this slide up once Marc and I start chatting. We will answer with a definitive result whether this is a best in mechanism way of blocking this pathway from an immunogenicity standpoint, and then we'll answer the other key questions that I think will draw a straight line to an expectation of higher efficacy in phase II. As of today, this is what the competitive landscape looks like. This is a non-exhaustive list of the companies that are developing TL1A blocking antibodies. This is an exhaustive list of all of the companies that are developing DR3 blocking antibodies today. With that, I think we can start.
Great. Taylor. Thanks for that overview. It's good. Just a second to get set up here. I think you kind of walked through it nicely some of the rationale of why you'd wanna target DR3, but I guess is there any theoretical risk of it, of inhibiting DR3, particularly in, some of the tissues where maybe TL1A hasn't even been turned on yet, in an inflammatory process and isn't present? Is there a risk of kinda over-inhibiting the pathway?
Yeah. Because TL1A is the only thing that turns on DR3, if the only thing that happens to DR3 when your antibody binds it, then you wouldn't expect any other adverse events, because the cell that's doing that is expressing DR3 is waiting for TL1A to be present. Otherwise, it's a null receptor on the cell. We engineered the antibody so that it doesn't bind Fc gamma receptors, so that reduces any risk of other off-target effects happening when the antibody binds. We also engineered the antibody so it does not cause receptor-mediated endocytosis of DR3, so there is no what's known as residual agonism. That is the principal concern that people have and probably what kept most folks from going after DR3 earlier.
That's something you have to engineer out very carefully. It depends on the epitope, it depends on the affinity of the antibody, it depends on how you engineer the Fc part of the antibody. I think you wouldn't expect any new toxicities is where we've always been thinking. You might ask, okay, if this really is a more durable way of blocking the axis, might you see either higher grade or more frequent AEs than the AEs that are observed with the TL1A antibodies? That is a fair question. We always believed that there would not be, again, because it's a monogamous receptor ligand pair. And certainly, the non-human primate data gave us, I think, a concrete result there. Tulisokibart caused lethal AEs in monkeys because of secondary immunogenicity. We had pristine acute and chronic tox. That's the expectation for humans.
Okay. When we do get the immunogenicity data, I mean, I think most antibodies have at least some level of ADA detection. What is a acceptable level of ADAs that's, in your view, clearly different than what's been shown with TL1As and maybe more of a kind of quote, unquote, "standard" antibody ADA response?
I mean, no TL1A blocking antibody causes ADA in less than 64% of patients, period. In both our acute and chronic non-human primate tox studies, we had about 9% ADA, and this antibody is much more human than it is monkey, so we expect lower ADA rates than that in humans.
Okay. Maybe kind of tying that, one is to show the ADA levels are low, but that hope that that is ultimately gonna translate to an improved clinical benefit. Are there downstream biomarkers that we should also be looking at that are really well correlated to TL1A activity to show that you're really inhibiting the pathway to the same extent?
In a healthy volunteer trial, there aren't disease-modifying biomarkers you can look at. The receptor occupancy assay that we developed is a TL1A binding assay. We measure whether 325 inhibits any TL1A binding to DR3 positive cells, that's as good as you can get. There are concrete data points you can look at in the peer data sets to quantify how much efficacy is being left on the table, both at induction and maintenance with the TL1As.
Is the goal b lock 100% of TL1A blocking or again is, do you wanna allow some level of, signaling to happen?
A 100% blockade at all times.
Yeah. Okay.
It's the goal.
You mentioned some of those looking to the TL1A program, some of those proxies of correlating the clinical efficacy. I guess what are those assays as you maybe get more into patients?
Yeah. I mean, the proxy data sets come from all the leading TL1A players, afimkibart, duvakitug, and tulisokibart. For afimkibart, the Roche antibody, there was a publication in The Lancet Gastroenterology last year where they went into detail on their phase II data, and in the data supplement... What everybody, what every TL1A developer is saying today is, "Yes, we have high rates of ADA, don't worry, not many of those ADA are neutralizing." They're asserting that maybe efficacy doesn't suffer. If you look at the afimkibart paper in The Lancet and go to the data supplement, what you'll see is that they broke down exposure, drug levels of afimkibart by ADA titer. Also, they broke down efficacy by ADA titer.
As you go from the first ADA quartile to the fourth ADA quartile, there is a relative 50% drop in efficacy at the time of induction. afimkibart didn't underperform the other TL1A blocking antibodies. Unless you think afimkibart is a fundamentally better antibody than duvakitug or tulisokibart, which no preclinical data suggests that it is, this is a class effect. The other reason that you can be confident this is a class effect is that this is also what is seen with TNF inhibitors, the other types of antibodies that bind soluble trimeric proteins. For TNF inhibitors, you can also find publications showing you that if you measure the relative reduction in efficacy to a TNF inhibitor based on whether that patient's ADA are neutralizing or not, non-neutralizing antibodies reduce efficacy to the same extent as neutralizing antibodies.
The reason for that is that both neutralizing and non-neutralizing antibodies can still accelerate clearance of your drug. That gives you a sense of that there's maybe a 50% relative efficacy to gain at the time of induction. When you think about what's left to gain between induction and maintenance, you can look at the Merck data. They had a presentation at UEGW in 2024 from the ARTEMIS-UC study where they showed you that there was a percentage-based improvement in the number of patients in clinical remission and response from week 12 induction to week 50 maintenance, but there was a numerical loss in the actual number of patients in response from induction to maintenance.
They went from 45 patients in response at induction to only 30 patients in response in maintenance. That's now the same thing you can see with the Teva data. The point is that the patient in front of you is not getting better from induction to maintenance. When you ask yourself, "Okay, is this just an aspect of this mechanism? Why should we believe that more longer term blockade of TL1A should lead to increasing responses the way you see with IL-23 inhibitors?" The ARTEMIS-UC study also had a reinduction arm. The patients that didn't achieve clinical response with the first three months high dose tulisokibart, instead of being removed from the study, they were given another three months high dose tulisokibart.
They were able to convert the same proportion of patients into response in the reinduction arm as achieved response the first time around. You go high dose tulisokibart to high dose tulisokibart, you double response rates. You go high dose tulisokibart to low dose maintenance, you lose them. ADA is the only logical explanation for why that result happens.
One of the other things you mentioned earlier was , how difficult it was to design the antibody that is SL-325 to get around the Fc, but also the internalization and the potential for agonism. If it turns out that the assays aren't perfect and there is a little bit of agonism happening in patients or in healthy volunteers, how would that show up and what are kind of the assays people should be looking to confirm that agonism's not happening?
Yeah. Pretty quickly. Lymphocytes don't respond to TL1A binding to DR3 unless those cells are also receiving stimulation through their T cell receptors. The most sensitive in vitro assays that you use to test for residual agonism are ones where you take primary human cells, you tickle them with a CD3, CD28 agonist, and then you bring in your antibody with or without TL1A. Many of the antibodies that we made in those assays show you that you get signaling happening through the DR3 cytoplasmic domain. It's a rare clone that you find one that doesn't do that. In vivo, in non-human primates and in humans, the most sensitive readout for whether there's any DR3 agonism happening is looking for proliferation of different T cell subsets between five and eight days after you give your antibody.
That's what we looked at in non-human primates in both acute and chronic tox. There was zero evidence of T cell proliferation of any subset in any animal at any dose at any time. And so we were confident that that would also be the case in humans, but I understand people wanna see the data in humans, and that's coming soon. For belt and suspenders, we also look for any changes in serum cytokines. We look for any changes in cell surface markers of activation. The expectation is that you see zero changes in any of those parameters.
Okay. We're starting to run a little close on time. Maybe we'll turn to what are the likely next steps after? W e're gonna read out the healthy volunteer data, in the next few months, but what are the next steps planned after that?
Yeah. We will wrap up the study, the phase I study very shortly. We will start a proper phase II-b randomized controlled study in Crohn's disease in Q3. This will be a high dose versus low dose versus placebo phase II study. Reasons to start in Crohn's before UC are, number one, we believe there's similar amount of mechanistic validation of the pathway in Crohn's as there is in UC. The translational link between TL1A polymorphisms in Crohn's disease has always been stronger than it is with ulcerative colitis.
There's slightly less competition right now in Crohn's than in UC. Talk to most GI docs and they'll tell you, "Look, if you guys have a signal in Crohn's, you're gonna get credit for UC, but it doesn't necessarily work the other way around." The final reason is a mechanistic one that we didn't touch on this, but DR3 is constitutively expressed. TL1A is expressed in short bursts just at sites of inflammation. There's an expectation that blockade of DR3 will lead to more durable inhibition of TL1A signaling, and we think that our ability to show that will be greater in diseases that have more foci of inflammation. That is the case with Crohn's, a bit more than UC.
Okay. So you mentioned two doses versus placebo, r ough size of the trial. How big of a trial should we be thinking about?
Yeah. This is gonna be between 170 and 180 patients. We're making conservative assumptions with regard to endoscopic response rates in the placebo arm. We're powering for a 20% improvement even though 20% wouldn't be that interesting. We need 25% or greater. That gives us wiggle room for some of the things that have happened in recent trials with our peers.
The expectation is that trial will allow you to then select the dose that's phase III, or potentially need to do some more dose ranging before you would be able to move into phase III?
It should give us the dose.
Okay. In terms of the length of that trial, in terms of follow-up, how long? Is it just the induction phase? Do you have the tox coverage that you can move people into maintenance and keep them on drug long term?
We do. Yeah. We just finished the chronic tox study. Both 325 and 425 were in that chronic tox study, by the way. This will be a induction through to maintenance trial.
Okay. Maybe just with SL-425, as you mentioned, it's through the chronic tox study. I guess, what is gating to putting that into the clinic? Is it, just the healthy volunteer data? Do you wanna see the efficacy data to see that these hypotheses about better efficacy is playing out with SL-325 before you would advance SL-425? Just what's the plan there?
I mean, the plan, it's IND ready now. I don't think we're gonna need it. I think three two five will achieve the dose profile that we want. The reason for that is that we expect the duration of receptor occupancy to be very long. W e live in reality, and if we don't have a YTE extended version of this antibody, someone else will. We also don't wanna be wrong about that. We will move four two five into a phase I trial while the phase II is going for three two five, and if there's a good reason to bridge that antibody, we could always do that. I don't think there will be. I think it's more likely we use it for either indication splitting or life cycle management.
I guess the question is where is that cut point in your mind of longer duration between doses just doesn't really matter anymore? Is it three month dosing? Six. Yeah. Is it monthly dosing, three month dosing, six-month dosing, annual like where is that point where you're like, "Yeah, that's just really not that much of a benefit to a patient anymore?
Efficacy drives it all, right? I think we saw this a little bit in the reaction to the Upstream data. We expect to have higher efficacy, period. We also want to have a drug that can be given subcutaneously no more often than Q4 weekly. You talk to most docs and they say, "Q4 weekly is fine." Is that where the world is three years, five years from now? I don't know. I think with this drug we'll be looking at probably Q8 if I had to guess today, subcutaneous dosing and maintenance. We'll see. We'll get the rest of that PK data in the next couple of months.
Okay. You mentioned on the pipeline slide you are working on bispecifics that contain DR3 as one of the arms. Just when will we start to see what the combo on the other arm is? When are some of those agents gonna be ready to go into the clinic?
Soon. Later this year, we'll share that, and we're not going novel here. We're looking to pair DR3 with another validated mechanism. There's a small list of targets that make sense in doing that. TL1A is not an approachable target with a bispecific. We know that from the AMG966 data. Now we know it from the Roche TL1A by IL-23p40 data. Just ask yourselves, why isn't there a TNF-α directed bispecific? It's not for lack of trying. There have been years of attempts at building TNF-α by IL-17s, IL-23s, any number of other things. Look at those early data sets, and they look exactly like the emerging phase I data sets for the two TL1A-directed bispecifics.
Unfortunately, that's the time we have. We're gonna have to cut off there. Thanks a lot, Taylor. We're looking forward to seeing the data in the next couple months.
Thank you, Marc.