All right, good morning, everyone. My name is Michael Schmidt. I'm one of the senior biotech analysts at Guggenheim, and it is my great pleasure to start this fireside chat this morning with Compass Therapeutics. With us today, we have Thomas Schuetz, CEO of Compass. Welcome, and thanks for joining us.
Thanks, Michael. Thanks for inviting us to the conference. Really appreciate it.
So why don't we just start out with a high-level question before we talk about your pipeline in more detail? And this has to do with your platform. So can you just walk us through your discovery capabilities and your StitchMabs platform? And perhaps also talk about your general approach to bispecific antibody discovery.
Sure. So we have the sort of internal capacity to do naive discovery campaigns for monoclonal antibodies against antigens. We've got a couple of different discovery platforms there where we generate our CDRs. I think your question is very interesting and very important to us, of course. So at the company, we developed a technology that we call StitchMabs. It's a screening tool that allows us, in a 15-minute room temperature incubation, to covalently link any two monoclonal antibody fragments together. And we use that to screen for synergy in bispecific drug candidates. So if you think about a bispecific, I think there needs to be some advantage to the bispecific over the simple combination of two monoclonals. Because if there's no advantage over two monoclonals, just give two monoclonals together. So we developed a hypothesis that we could screen for combinatorial synergy.
We use that tool to screen for synergy in bispecific drug candidates. Synergy can mean a couple of different things. Synergy could simply be co-localization. I think the best example of that would be a BiTE. The antigen and CD3 separately would have no effect whatsoever. The co-localization from the BiTE is very important. Also, drugging two signalling pathways simultaneously. I think all of those things, I think, are part of our approach to bispecific drug discovery.
Great. And so yeah, then why don't we talk about some of your key value drivers, starting with CTX-009, which is your DLL4 VEGF bispecific antibody. And so again, remind us, perhaps, first of the mechanism of action behind this product candidate.
Sure. So DLL4, delta-like ligand 4, that is the cell surface ligand for Notch 1. VEGF-A, of course, is the well-known soluble ligand for the VEGF family of receptors. So those two signalling pathways, DLL4 Notch 1, VEGF-A, VEGF receptor, have different phenotypic effects on angiogenesis in the tumor microenvironment. So importantly, I think dual blockade has been shown to be synergistic in preclinical models. And I think what ultimately could turn out to be really interesting is that DLL4 upregulation in the tumor microenvironment has been shown to mediate resistance to VEGF-targeted therapies, such as Avastin. I think that's going to turn out to be really important. So its 009 is a next-gen angiogenesis disruptor, if you will.
Right, and then yeah, I know you're focusing on BTC here in your lead indication. Perhaps just remind us of the commercial potential of this opportunity.
Sure. So we're currently running a randomized trial in patients with biliary tract cancer who've received one prior line of therapy, so that study is a randomized study of CTX-009 plus paclitaxel versus paclitaxel alone in patients who have received one prior line of therapy, and I think, excitingly, we're going to get a readout from the primary endpoint of that study in the next couple of months, so it's a very obviously exciting time at the company, and because we're sort of so close to potentially commercializing this drug, we've done a lot of work to address your question. How many patients are out there, so we were really helped, I think, by an analysis that Komodo Health did. They looked at U.S. claims data in 2021 and 2022, and they discovered that there were about 23,000 new patients annually, and the annualized prevalence was actually 30,000.
So It's a really, really solid patient population. Just unfortunately, in BTC, only about 15% of patients have an actionable mutation. Believe it or not, the other 85% of patients in the United States have no labelled second-line therapy. There's no second-line standard of care in a disease that is so common. Our estimate is that about 15,000 patients annually will progress to receive second-line therapy. It's a very substantial patient population, something like three times larger than the platinum-resistant ovarian cancer market.
OK, great. And then, as you mentioned, you are in an ongoing or nearly complete phase III study right now. Just remind us of the data that supported the design of that study.
Sure. In a phase II study where the phase just backup one step back in an all-comer phase I-B, where we tested 009 in combination with either irinotecan, paclitaxel, we had two deep and very durable responses in a couple of patients with cholangiocarcinoma. And both of those patients received 009 in combination with paclitaxel. So it was a purely empirical observation at the very beginning. But it was an important enough observation that we had to confirm it in a phase II study. So we treated 24 patients with advanced BTC who had received either one or two prior lines of therapy, so a mixed second- and third-line study. And we had, surprisingly, a 37.5% overall response rate in that study. And those were confirmed responses. And what would you expect in the second-line setting?
So FOLFOX, the three-drug chemotherapy combination, has about a 5% response rate in that patient population. This was just an unequivocal efficacy signal. That phase II study informed the design of our current randomized trial.
Right. And then yeah, can you talk about the randomized study, some of the primary endpoints, and what your powering assumptions were?
Sure. So the primary endpoint of the study is overall response rate. The study is 90% powered to detect a difference of 33% versus 10% in overall response rate. So 33% in the combination arm, 10% in the chemotherapy-only arm, so a delta of 23%. So the study fully enrolled last August. And the primary endpoint is triggered by time. The analysis is scheduled for 28 weeks after the last patient was randomized. And we picked that because the latest first time that a patient responded in our phase II study was at 24 weeks, six months. So an additional four weeks for a confirmatory scan gets you to 28 weeks. So 28 weeks from mid-August is mid-March of this year. So we're coming up on a readout from the response rate endpoint. Key secondary endpoints, we're using the hierarchical testing method to control for alpha spend. PFS and OS are the key secondary endpoints.
Right, and then can you discuss a little bit your regulatory interactions with the agency around filing requirements? Is there the need to demonstrate a significant effect on PFS and OS to support potential filing or approval?
So I don't know the answer to that. In our exchanges with FDA around the design of this randomized study, they never told us what would specifically be required for filing. So as they commonly do, it depends on the data.
Is this an accelerated approval opportunity or full approval? Do you have any visibility on that?
Sure. So that question is interesting. And we had a specific interaction with FDA on that specific question. So we asked if overall response rate could potentially support full approval, not just accelerated approval. And I just want to be fair with you. FDA did not say yes, and they did not say no either. So thank you. So they said it depends on the magnitude of difference. And they didn't tell us what that magnitude needed to be.
Right. And then just on the secondary endpoints, I know it's a fairly small study. Technically speaking, can you actually demonstrate a statistically significant difference in PFS? And if so, what magnitude of benefit on some of those outcomes could you technically demonstrate?
Sure. Yeah. So I think technically, to use your word, the answer is yes. So PFS is 80% powered to detect a hazard ratio of approximately 0.6. And in order to calculate that, we used the lower bound of the 95% confidence interval on PFS from our phase II study. So the assumption for PFS is 5.4 months median in the combination arm, 3.0 months in the control arm. That gets you to a hazard ratio of approximately 0.6.
OK, gotcha. And then yeah, how much information will you be able to include in your top-line announcement if possible?
Right. So let me just go back briefly to the power calculations for PFS. So we used 5.4 months as the assumption of median PFS in the combination arm. In the phase II study, the median PFS was 9.4 months. So again, we used the lower bound of the 95% confidence interval. So when we have our top-line data in a couple of months, that's only going to be a response rate. Because we don't have enough events yet occurring in order to be able to have a readout on the time to event analyses. And the best that we can do right now, as we've said publicly, is that it's going to be in the second half of the year. I realize that's a broad range. Maybe in the coming months, we'll be able to get a little bit more accurate on the timing of those analyses. But it's going to be in the second half of the year.
That makes sense. Will you be able to do an interim analysis on PFS and OS?
Definitely no. The reason for that is the alpha spend. With the hierarchical testing, you're not allowed to do an interim analysis.
Gotcha. All right, cool. Then maybe just going back to the market opportunity. So this is presumably for second-line BTC, correct? How widely is paclitaxel used, which is your combination chemotherapy choice? And just talk a bit about the bigger context, other drugs that are perhaps used for BTC.
Sure. So let's start with the 15% of patients with an actionable mutation. So an IDH1 inhibitor, an FGFR2 inhibitor, or a HER2-targeted drug, which is something like 3% to 5% of the patient population. So excluding that 15%, the other 85% of patients get a wide variety of different chemotherapy regimens. Some are getting FOLFOX based on the randomized trial against best supportive care. Docs don't really like that regimen because they don't like to give two platinum agents back to back, because cisplatin is part of front-line standard of care. So patients are getting 5-FU leucovorin. Some patients are getting liposomal irinotecan based on the NIFTY study. Some patients are getting Abraxane based on SWOG 1815. So it's a complete mix of second-line therapies that is, frankly, largely investigator-specific.
Gotcha. OK, understood. So looking forward to the top-line announcement then in a few weeks. And then I know you've also been doing work with CTX-009 in colorectal cancer. I know there has been work done with other DLL4/ VEGF bispecific, more or less successful. So how do you think about the potential of the antibody in colorectal?
Sure. So in a phase II study in the U.S., we had monotherapy responses in patients treated in the fourth-line setting. So we had two responses out of 41 patients that were evaluable in that study. And as a largely fourth-line study where current standard of care is either fruquintinib, regorafenib, and those two drugs have a 1%-1.5% response rate, believe it or not. And those are approved drugs in the fourth-line setting. So we have a signal there. What we're doing right now is we're analyzing biopsy specimens that we collected from that study for DLL4 expression. And we're trying to figure out if there's a correlation between clinical activity and DLL4 expression. We had a hint of that from our phase I study, but we're trying to repeat that with a larger sample set.
I think in the coming months, we'll probably finish that up and make a decision about whether or not to sort of go down the pathway of developing a companion diagnostic. I'd like to have that decision made before the end of the quarter.
OK, great. And then perhaps switching gears and talk about CTX-471, which is your 4-1BB antibody. Again, it's been a category that has been challenging for others. And so perhaps just talk about the general concept behind your development strategy.
Sure. So you're correct. Targeting 4-1BB has been a very challenging field. And there are many, many drugs in development now targeting 4-1BB because it's such an important receptor in the immune system. So we developed an IgG4 monoclonal antibody that is both epitope and affinity optimized for agonizing 4-1BB. So we did a standard dose escalation phase I, which was followed by cohort expansions in the post-PD-1 patient population. So everybody that got 471 had to have progressed on a labelled PD-1-containing regimen. Of course, PD-1 blockers now are labelled in so many different diseases. So we enrolled 60 patients that had 17 different tumor types. And we had five responses as a monotherapy in that population. I mean, look, I mean responses of any kind to any agent in the post-PD-1 patient population are very uncommon.
So we had three responses out of 11 patients treated with melanoma for a 28% response rate. We had a very durable PR in a patient with mesothelioma, one out of four patients treated. And then we had, believe it or not, a complete response in a patient with small cell lung cancer, one out of three patients treated. And in that study, we collected biopsies from patients. And we analyzed those biopsies last year. And we discovered that the small cell lung cancer patient who had the CR had a tumor that was highly positive for NCAM, neural cell adhesion molecule, also known as CD56. That made us look at all the other biopsies for CD56 expression on tumor. And we found that all the patients with clinical benefit, response, or stable disease, their tumors were NCAM positive. All the patients whose tumors were NCAM negative had progressive disease.
A really interesting observation. We presented all that data at SITC last fall. It's really interesting because we then did some research in the lab on, does this make any mechanistic sense? NCAM and CD56, it interacts with itself. It's a homotypic interaction. CD56-positive NK cells are a major subset of effector NK cells. We believe that we may have discovered sort of a novel immunophenotype. NCAM negative tumors cannot attract NK cells. NCAM positive tumors attract NK cells. What do you know? That interaction upregulates CD137 on the NK cell, whereas 471 can activate it. We're now going to do a basket study. We're well on the way to planning that now. That'll start probably middle of this year in the NCAM positive patient population.
That's a substantial patient population, probably 60,000 patients annually in the U.S., because most solid tumors have subsets of patients whose tumors undergo so-called neuroendocrine differentiation. They express NCAM. And so we're going to study that in a phase II basket study.
OK, and that'll be presumably post-PD-1 and then prospectively selected for.
So that's a fascinating question. So not all of the NCAM-positive indications have labelled PD-1. So that's one of the reasons we're kind of excited about this, because we're going to be able to treat some IO-naive patients. So neuroendocrine tumors in general are not amenable to PD-1 blockade. That'll be the first time that we treat some IO-naive patients.
Gotcha. OK, so looking forward to that and then maybe spending the last five minutes discussing your PD-1 portfolio now at this point, so CTX-8371 is a PD-1, PD-L1 bispecific, a very unique concept, perhaps. Just walk us again through the rationale for targeting both in one molecule.
Sure. That's the first time I've heard the phrase PD-1 portfolio. I'm going to steal that from you. So we used our StitchMabs technology. And we asked an incredibly simple question. What is the best combination partner for PD-1 blockade in a bispecific antibody? So we made a small library, tested those constructs. And much to our surprise, the best partner for PD-1 blockade turned out to be PD-L1 blockade. Now, we did this experiment. I want to be fair about one thing. We did this experiment about four years ago. So we did not include VEGF in that original screen, which I'll come back to in a minute. And what we discovered is that dual ligand receptor blockade in a bispecific is between 100 and 1,000-fold more potent than either antibody alone or the combination of the two. It was a very surprising outcome.
We spent about a year investigating the mechanism of action. It turned out to be not just dual ligand receptor blockade. The bi-specific is unequivocally a cell engager, no question about that. We published on that. Fascinatingly, the bi-specific exposes a metalloproteinase cleavage site on PD-1. It leads to the cleavage of PD-1 off the surface of effector T cells. We turn PD-1 positive exhausted T cells into PD-1 negative T cells. It's an incredibly unique mechanism of action. We confirmed that result, actually, in monkeys. We showed that we cleaved PD-1 from T cells in monkeys. That was fascinating. The bi-specific is better than known PD-1 or PD-L1 blockers in preclinical studies. We're now in a phase I study. We just fully enrolled the third dosing cohort, which is one mg per kg.
We're going to probably move to the fourth dosing cohort, which is three mg per kg, in the coming month or so. And we've had no dose-limiting toxicity so far, which is in and of itself quite fascinating, actually. It's possible that the bi-specific actually could have maybe a better safety profile. We don't know yet, early days. But I think in the second half of the year, we'll be able to share some data from that study.
Just remind us through what types of patients are enrolling in the phase I?
Sure. So we actually limited the basket a little bit based on our 471 experience. So these are post-PD-1 patients with melanoma, non-small cell lung, head and neck, triple-negative breast, and Hodgkin's lymphoma. That's our first hematologic malignancy, because as you know, Hodgkin's lymphoma is a PD-L1-driven disease.
OK. So looking forward to that in the second half. Then lastly, you recently announced a new product candidate, 10726, I believe, your PD-1 VEGF antibody. And obviously, there's been a lot of excitement in the industry around that mechanism of action. And so perhaps just talk about the design of your molecule and perhaps any sort of differentiation or not, or similarities to even Ivonescimab, some other product.
Sure. So from my point of view, when did this field get started? So Roche did a study where they combined Avastin with Tecentriq in patients with liver cancer. And that combination was superior to sorafenib. And that really started the whole field. And there are lots of studies now on the immunological effects of blocking VEGF and combining that with PD-1 blockade. So a few years ago, we started thinking about bi-specifics that could do this. And it's not so straightforward, right? Because you've got four targets here, all of which have approved drugs: PD-1, PD-L1, VEGFA, VEGF receptor. So how do you think about this? So we spent a fair amount of time investigating the various combinations. And we did, to be fair about it, we landed on a very similar type of construct to Ivonescimab, two-by-two valency, N-terminal VEGF capture, C-terminal PD-1 blockade.
We have, we believe, some differentiation. Our PD-1 blocking component of 726 is better than IVO in vitro. It's about five times better PD-1 blockade. We'll see what that means. I don't know. And both components of that bi-specific actually came from 009 and 8371. So the CDRs have already been in the clinic. And we've also, I think importantly, a very important point, we've developed some bi-specific manufacturing expertise at Compass that we've used in order to develop 10726. And I think that'll ultimately, hopefully, turn out to be very important as well.
Right. And then I think you said IND by the end of this year.
Yes.
Just philosophically, how do you think about longer-term clinical development, perhaps relative to Ivonescimab and the BioNTech product and others?
Sure. I think we should start with indications where both PD-1 axis blockade and angiogenesis inhibition are approved therapies. So what's an example of that? Maybe something like renal cell, where nivo is approved, as well as VEGF kinase inhibitors are a big part of the therapeutic regimen in renal cell. So thinking about indications where both of the components have been shown to be effective, I think, is where we're going to start. So indications like renal cell.
Right. OK, great. Well, I think it's time to wrap up.
Great.
I really appreciate the time and exciting things happening at the Compass.