My name is Maury Raycroft, and one of the biotech analysts at Jefferies. It's with great pleasure that I'd like to welcome Thomas Schuetz, the CEO of Compass Therapeutics. Thanks so much for joining us today, Thom. Thanks, Maury. Thanks for inviting us to the conference. Really appreciate it. We're going to do fireside chat format. Maybe for those who are new to the story, if you can give a brief intro to Compass. Sure. Compass Therapeutics, we're located in Boston. We're a monoclonal antibody discovery and development company in oncology. We currently have three drugs in the clinic. All of our drugs are either monoclonal antibodies or bispecific antibodies. The three drugs we have in the clinic, we have a DLL4 VEGF-A bispecific antibody. That drug is called tovecimig. That drug is currently in a randomized trial in patients with advanced biliary tract cancer. We have a next generation CD137 agonist antibody. That drug is called CTX-471. Our third drug that entered the clinic is a drug that we call CTX-8371. That is a first-in-class bispecific antibody that targets both PD-1 and PD-L1. We recently disclosed that we have responses in the post-checkpoint inhibitor patient population, and we can talk more about that. Finally, we have our fourth drug that's going to enter the clinic, which is a novel PD-1 VEGF-A bispecific antibody. That drug is called 10726. Happy to talk about any of those drugs during the conversation today. Got it. Yeah, it's a great intro and a lot of interest in tovecimig going into the updated data first quarter 2026. And you reported response rate data earlier this year. Maybe just talk about expectations for this upcoming readout and more specifically what OS and PFS results would you consider to be clinically meaningful? Sure. As Maury mentioned, we're currently conducting a randomized trial in patients with advanced biliary tract cancer in the U.S. Patients have received one prior line of therapy, and so it's a pure second line study. Earlier this calendar year, we announced that that trial hit the primary endpoint of overall response rate in the study, more than tripling the response rate, and that difference was statistically significant. The secondary endpoints in the study are PFS, OS, and duration of response. Those analyses are all triggered by 80% OS events. We're waiting for the OS events to accumulate. In August, we announced, and then we reaffirmed two weeks ago with our Q3 earnings, that we're seeing fewer deaths in the study than we had originally anticipated. Said a different way, more people are alive and they're living for a longer period of time than we had anticipated. Because of that, the analyses of the secondary endpoints will occur in late Q1. You asked a very specific question, you know, what would we consider to be clinically meaningful? I think in oncology, hazard ratios in PFS and OS less than 0.7 would be considered clinically meaningful. I often answer this question just by reiterating what our assumptions were in the power calculations. For PFS, we assumed 3.0 months in the paclitaxel control arm, and we assumed 5.4 months in the Tevesemig combination arm. That's a hazard ratio actually of around 0.6. If we see that, that would obviously be incredibly meaningful. For overall survival, we assumed 6.2 months in the control arm and 10.9 months in the combination arm with Tevesemig. That's a hazard ratio slightly less than 0.6. The study is 80% powered to detect that difference. Obviously, if we saw a hazard ratio in that range for OS, the study would just be nothing short of stunningly positive. All of those things would be considered very meaningful, I think. Got it. That is really helpful and sets the stage for this update. Based on the response rate data that you saw in the active and control arm, how would you anticipate that that could translate to durability? Sure. You know, it's a very important question because when we looked at the overall response rate data for the primary endpoint, we noticed something quite important in the data, which is there was a substantial difference between the two arms in the incidence of progressive disease as the patient's best overall response. All of the responses in the study were assessed by so-called BICR, B-I-C-R, blinded independent central review, obviously the most rigorous way you can analyze imaging results. We noticed that the incidence of patients with progressive disease as their best overall response was 42% in the control arm and only 16% in the combination arm. Now, this is maybe a bit of a nuance, but if progressive disease is your best overall response, by definition, that means it occurred at the first scan time point. Because if you had something other than PD at the first time point, such as stable disease, for example, you'd be in the stable disease category. That difference, 42% versus 16%, must have occurred at week eight, which was the first scan time point. We already have a suggestion that the progression-free survival curves are separating, possibly separating significantly. Yeah. Interesting insights there. Given that you allowed crossover for the study, how many patients crossed over from the control arm to the active arm? Sure. Another very important question. Approximately half the patients in the control arm crossed over to the active arm. In these studies in indications where there's no available therapy, these studies basically cannot be done without a crossover. One example of that, we have, I think, incredible regulatory and incredible regulatory precedent with the IDH1 inhibitor, Tibsovo. In second line BTC, the exact patient population we're looking at, Tibsovo did a study of Tibsovo versus placebo that allowed crossover. A little bit more than half the patients crossed over in that study, but they used a statistical technique, which just maybe to get into the weeds for 10 seconds here, it's called the RPSFT, which stands for rank preserving structural failure time. It's a statistical method that adjusts the Kaplan-Meier curve for crossover. Tibsovo used that, and it was incredibly positive with a hazard ratio of 0.49. Even though the intent to treat analysis was not positive, that single study got Tibsovo approved in second line BTC. We prospectively declared that the RPSFT would be the primary method of analysis for OS in this study. That statistical method should correct for the crossover that we observed in the study. Got it. Helpful. You mentioned the BICR analysis where you're blinded, but there's also the investigator assessed data too. Wondering if you're blinded to that data as well. I don't have that data. Okay. It's an interesting question because why don't I have that data? The statistical testing has been prospectively defined to occur in a specific order. That order is referred to as hierarchical testing. In hierarchical testing, you have to define the order ahead of time that you're doing the analyses. The primary endpoint was overall response, right? We go to PFS as measured by BICR, followed by OS. We cannot, we're not allowed to analyze the investigator assessed responses until after the secondary analyses are completed. Got it. We'll have that sometime next year. We'll probably save that data for presentation at a medical meeting. In late Q1, we're planning to report progression-free survival, overall survival, and top line safety. Got it. Okay. You'll have those in a press release, probably conference call. Yes. Got it. Then you'll have more data, more analyses afterwards. Correct. Okay. Given that FDA's view on statistical requirements remains TBD, what's your initial view on what needs to be demonstrated to receive favorable feedback? Considering that crossover is allowed, do you think a positive OS trend should be sufficient? I think you kind of already answered that based on. Yeah. So you know, just to be super clear here, I don't really know. I don't have a piece of paper that tells me what the answer to your question is. I am inferring a little bit, you know, from the conversations. I think the way you summarized it is probably correct. You know, that similar to Tibsovo, you know, we hit on PFS, have a favorable trend on OS. I think that puts us in a very good place to have a productive conversation because we would have hit on ORR, PFS, and a favorable trend on OS in an indication that literally has no labeled therapy in the U.S. Right. Yeah. A common question we get is just around what these patients typically get treated with. Maybe talk about that and also just the population size in the United States. Sure. Earlier this year, as we're preparing for potential commercialization of this drug in the U.S., we conducted some pretty comprehensive third-party market research, which gave us some data on these two questions. First of all, in terms of what these patients are usually treated with, it's a whole set of different chemotherapy regimens: FOLFOX, FOLFIRI, 5-FU leucovorin alone, the NIFTY regimen, 5-FU leucovorin, liposomal irinotecan, Abraxane. It's a whole set of different therapies that these patients are treated with, including restarting frontline therapy, which is somewhat crazy to think about. It's a whole different set of chemotherapy regimens. When we had conversations about the study design with FDA, FDA simply said to us they don't consider any of these to be standard of care in this patient population. What proportion get a Taxane based on your? I don't remember the exact fraction. I think more patients get a 5-FU based regimen than a Taxane in the second line setting, but I don't remember the exact number. The patient population, as you alluded to, you know, we did, we repeated a claims-based analysis as part of our market research. That claims-based analysis identified about 25,000 patients annually as newly diagnosed with biliary tract cancer in the U.S. It's an incredibly common cancer. Komodo Health, the claims aggregator, actually did an analysis that they presented at the Cholangiocarcinoma Foundation meeting in 2023. They did a claims-based analysis for the two-year period of 2021 and 2022, and they found just under 23,000 patients diagnosed annually. This is consistent with some published literature that suggests that the incidence of this disease is increasing. It's a very common patient population. In terms of the fraction of patients who make it to second line therapy, we believe it's something like two-thirds of patients make it to second line therapy. Our treatable patient population would be something like 15,000 patients annually. Again, that's in the U.S. alone. U.S. plus EU plus Japan is probably something like 60,000 plus patients annually in second line BTC only. Got it. A lot of patients for this tumor setting, how do you think about pricing and what are appropriate benchmarks there? Sure. You know, I think we also included some payer interviews as part of our market research. You know, there are some recent launches of drugs, you know, that I think are relevant here in second-line BTC. The Jazz Zymworks drug, Zanhera, priced at about $36,000 a month. I mean, that's, you know, that's obviously that's a high number, but, you know, both that's about what these, you know, what these drugs are priced at now. Some very recent launches in oncology are actually higher than that, you know, in the $40,000 a month range. Making some very conservative assumptions, you know, that maybe a patient is on drug, say, an average of six months, just as a ballpark number, you know, six times 35 is about $200,000 per patient annually. Fifteen thousand patients times $200,000, second line BTC in the U.S. alone is approaching a $3 billion annual market. Interesting. Yeah. Okay. Maybe talk about the competitive landscape. We do not see a lot of activity. We see some activity in the mutation agnostic approaches for biliary tract cancer. What is your perspective on the landscape and how do you view Akeso's first line BTC program? Sure. You're correct. You know, the development of mutation agnostic approaches in this disease have been few and far between. We would be probably the first. In second line BTC, all in the U.S., all of the recent drug approvals have been mutation-specific drugs, FGFR2, IDH1, or HER2. Akeso with ivonescimab, I believe they're going to test that in a frontline study. I'm not certain what the control arm there is. I believe it might be pembrolizumab. You know, I hope that study's positive because it'd just be good for everybody. Obviously that's going to be a big study, probably just, I don't know when that study's going to read out, probably early 2030s, I would think, something like that. That's a fairly long time away. Yeah. Okay. You have got an investigator-led study with MD Anderson testing tovecimig in frontline BTC. What are expectations for this study and how many patients will it enroll and when can we see initial data? Sure. You know, that was a very interesting development. You know, our second line study is open label at the sites. You know, we're doing our best to stay blinded internally, you know, just to preserve the integrity of the study, of course. MD Anderson in Houston was the second largest enroller in our second line study. After they had enrolled something like seven patients or so, they sent us a proposal to add tovecimig to the frontline regimen of gemcitabine, cisplatin, and durvalumab. That study got going last year. It's enrolling patients now. I'm going to get an update on that study from the investigators early next year. It's planning to enroll 50 patients. I know it's actively enrolling now. I don't have a specific update on that today, but I'll get an update from that study. In terms of what we could get out of that study, you know, I think we're certainly going to get safety and tolerability data, you know, with another chemotherapy regimen, including a checkpoint inhibitor. You know, that'll be important because I think drugs like bevacizumab, you know, have been very successful because they can be combined with basically almost anything, you know, any chemotherapy regimen, checkpoint inhibitors, et cetera. We'll get safety and tolerability data. I think, you know, ideally we could get an early read on efficacy, which depending on what that is, you know, potentially that could lead to, you know, maybe an NCCN citation, maybe, or I think at minimum it would help us power a frontline study. Got it. Okay. Interesting. You also plan to start a basket study in DLL4-positive patients. Which tumor type should we expect to see most frequently and what would constitute success that could allow you to pursue an indication agnostic? Sure. One half of tovecimig targets DLL4, delta-like ligand 4, which is the cell surface ligand for NOTCH1. It turns out there are many, many tumors that are enriched for DLL4 expression. Most of these tumors are intra-abdominal malignancies. Tumors that are known to be DLL4 positive in addition to biliary tract cancer are colorectal cancer, gastric cancer, hepatocellular cancer, renal cancer, ovarian cancer. A whole set of tumors are really possible for expanding the use of tovecimig into other indications. This study that, you know, we're still finalizing the design of it, but this study would be sort of a classic signal finding study that we would use the data to inform the design of post-approval label expansion studies following what we hope will be the approval of tovecimig in the U.S. Got it. Okay. What's your strategy for ex-U.S.? You mentioned EU and Japan earlier. Would the current study be sufficient for those regions or? I don't know that yet. We have not had an interaction with the EMA in Europe. Just, you know, just to be fair about that, you know, that would be something that we would want to, you know, work on in the coming year. We would need to do the same thing in Japan. You know, I think there's, you know, the possibility that, you know, this study could, you know, support an application in those jurisdictions, but I don't know that yet. We have, you know, disclosed publicly that, you know, we're having some, you know, conversations with, you know, strategic partners about ex-U.S. geographies. Got it. Okay. Makes sense. Let's shift gears to 8371. Maybe talk about just the rationale with this bispecific. You mentioned you've seen some responses already. Maybe talk about just more about what you're seeing there. Sure. I think 8371 has been one of the most exciting developments at the company, really, in the past several months. Several years ago, we began a scientific discovery program where we set out to discover a next generation checkpoint inhibitor. We, in order to do so, needed to develop a technology, and we did so. That technology is called StitchMabs. That technology allows us to build small libraries of biologics to screen for synergy. We asked a question, what's the best combination partner for PD-1 blockade in a bispecific antibody? Built a small library, did a screen, and fascinatingly, turns out that PD-L1 blockade is the best combination partner for PD-1 blockade. In a bispecific antibody, the combination of dual ligand and receptor blockade is between 100 and 1,000-fold more potent than either antibody alone. It turns out to have an incredibly unique mechanism of action because it's not just dual ligand receptor blockade. The bispecific is unequivocally a T cell engager. Really interestingly, the bispecific actually strips PD-1 off the surface of effector T cells. It converts PD-1 positive T cells that, of course, can be suppressed. It converts those into PD-1 negative T cells, which cannot be suppressed. We brought that into a phase one study last year. It's a three plus three dose escalation where we were testing five different doses between 0.1 and 10 milligrams per kilogram. We've completed the dose escalation, and there have been no dose limiting toxicities at any of the dosing cohorts, suggesting that we may have a differentiated safety profile, which I think is plausible mechanistically because we believe that this drug is anchored in the tumor microenvironment via PD-L1. Fascinatingly, because we did not see any dose limiting toxicities, I said it was a three plus three design, but we never needed the plus three. We only have three patients at each of the five dose levels, so 15 total, and we already have three responses, all in the post-checkpoint inhibitor setting. We have a confirmed PR by RECIST in a patient with non-small cell lung cancer. That patient's target lesions completely disappeared from nearly 6 centimeters to zero. We have a confirmed RECIST PR in a patient with metastatic triple negative breast cancer. I think this has been the single most important development at the company in the past six months. This patient relapsed while getting adjuvant Keytruda. Technically, the patient is actually refractory, then received Trodelvy, and then two additional chemotherapy regimens. Treated with 8371 in the fourth line setting, this patient started out with nearly 9 centimeters of tumor burden, and it is currently down to approximately 4 millimeters. This patient has had more than 90% decline in their linear tumor burden. This patient could convert to a CR conceivably, very, very nearly a CR. This patient is now durable, continuing to decline, and continuing on drug. Just an incredible observation. Based on those two observations, we've announced that we're initiating cohort expansions this quarter in patients with non-small cell lung cancer and triple negative breast cancer. That study's ready to go and about to be up and running. We recently announced that we have a third response in an indication that we have not yet disclosed at the highest dose level. We now have three responses out of the first 15 patients treated. I think we could be off to the races with that drug. Interesting. For the three responses, were they at higher doses? I guess, are you seeing a dose response, sir? Interestingly, you know, with checkpoint inhibitors, you don't commonly see a dose response. Two of the three responses were seen at the two highest dose levels, 3 mg per kg and 10 mg per kg. Those are the two dose levels that we're taking forward into the cohort expansions. Got it. You'll have another data update first half of next year. Correct. We hope to present the dose escalation data and some early cohort expansion data at a medical meeting in the first half of next year. Got it. A lot of interest in the PD-1 VEGF program as well. Maybe provide a status update there. Sure. Just two weeks ago at SITSI, we presented the full preclinical characterization of that drug, 10726, a novel PD-1 VEGF bispecific antibody. We've completed all of our preclinical work. We've got manufacturing all locked down. We have commercial scale yields in our manufacturing process for that drug, so ready to go there. Finishing up the last dotting of the i's and crossing of the t's on the IND submission, phase one study to begin early next year. We're going to go into patients with gastric cancer, hepatocellular cancer, renal cell, and endometrial cancers. Indications where both PD-1 blockade and VEGF targeting are effective. Those are the indications we've selected for the phase one study of that drug. Got it. Thom, I think we're out of time, but maybe just to close out, just highlight key catalysts ahead. Sure. 2026, very important year for the company for sure. Going in reverse order, phase one initiation for 10726, cohort expansion in 8371, basket study for 471 that we didn't have time to discuss, and then the PFS and OS readout in late Q1 from our randomized trial in patients with tovecimig. That'll be followed by an FDA interaction if that goes well, a license application, setting up commercial launch of the drug in the first half of 2027. Big year next year. Big year next year. Thanks so much for joining us today. Thanks, Maury. Thanks, everyone.