Compass Therapeutics Inc. (CMPX) 2025 Q2 法說會逐字稿

Greetings, and welcome to Compass Therapeutics second quarter 2025 Earnings and Business Update Call. (Operator Instructions) As a reminder, this conference is being recorded. It's now my pleasure to introduce Anna Gifford, Chief of Staff. Thank you. You may begin.

Good morning, and thank you for joining us. My name is Anna Gifford. I'm Chief of Staff at Compass Therapeutics. With me today is Dr. Thomas Schuetz, CEO and Vice Chair of the Compass Board. Our CFO, Barry Shin, will also join us for a short Q&A following Tom's comments. Earlier this morning, we released our financial results and a business update for the second quarter. The slide presentation accompanying this webcast and a copy of the press release are available on our website. Please note, we'll be making forward-looking statements on today's webcast. These forward-looking statements are described in our press release issued today and the company's SEC filings. With that, I'd like to turn the call over to Thomas Schuetz. Tom?

Thank you. We are incredibly excited today to be hosting this call with you this morning. As Anna just mentioned, earlier today, we released our quarterly financials for Q2 2025. In addition to the financials, we also provided very important updates for 3 of our development programs. These updates are summarized on the next slide, and I'll then provide additional details for each of these program updates. Most importantly today, for our lead program, tovecimig, the DLL4 VEGF-A bispecific antibody in the ongoing randomized trial in patients with advanced biliary tract cancer, there are currently fewer total deaths in the study than we have projected. While this disclosure is a simple fact, it is very important to say this sentence differently. More patients are alive today than we have projected. I understand clearly that this is an investor call, but it's so important to reflect on what this could mean for the patients enrolled in this study. I'll update the timing of the survival analysis in 1 minute.

Next, for CTX-8371, our PD-1/PD-L1 bispecific antibody, very unexpectedly, we have 2 deep partial responses in the early dose escalation cohorts in the ongoing Phase I study. One partial response is in a patient with non-small cell lung cancer and one partial response is in a patient with triple-negative breast cancer. Later this year, we'll be initiating cohort expansions in patients with non-small cell lung cancer and triple-negative breast cancer. I will describe CT scans for each of these 2 patients in a few minutes. We hope to present these data at a scientific conference later this year. We are also disclosing results from preclinical head-to-head studies of CTX-10726, our proprietary PD-1 VEGF bispecific antibody compared to the leading drug candidate in the class ivonescimab. We will be presenting these data at a scientific conference later this year.

So let's begin with tovecimig. This slide summarizes the design of the ongoing randomized study in the United States in patients with advanced biliary tract cancer. This study is a 2:1 randomization of tovecimig plus paclitaxel versus paclitaxel alone. The primary endpoint of the study is overall response rate. We announced those results about 4 months ago. The secondary endpoints in this order are PFS/OS and duration of response. We're using, of course, the hierarchical testing methodology to control for alpha statistical spending in the study. The next slide summarizes the current status of the study that we call Companion-002. On the top right, as I mentioned, we achieved the primary endpoint. So the study is positive by definition. We had a 17.1% overall response rate, about tripling what was seen in the control arm with a p-value of 0.031. On the top left box, I'll come back to this point in 1 minute. The trial was fully enrolled in August of 2024, enrolled 168 patients with advanced biliary tract cancer treated in the second-line setting.

As of today, we are currently at greater than 17 months median follow-up in the study. So let's talk about the secondary endpoints of progression-free survival and overall survival. It's actually -- it's hard to say this, but the secondary endpoints are triggered by a total number of deaths in the study. So that's how these time-to-event analyses are commonly done. We need 80% OS events to trigger the analysis of progression-free survival and overall survival. And today, we have fewer total deaths in the study than we had originally projected when we projected that we would be presenting analyses of these endpoints in Q4. We made that projection in April of this year. And since that time, the number of deaths in the study has continued to decline. Clearly, the 80% OS event threshold has not been met. So the analyses of PFS and OS are now projected to occur in Q1 of 2026.

I think the bottom of this slide is very important. Recall the study that was titled ABC-06. That study was published in 2021 in Lancet Oncology by (inaudible). That was a randomized study of the 3-drug combination FOLFOX, 5-FU, leucovorin and oxaliplatin in patients with biliary tract cancer treated in the second-line setting. So the exact same population that we're treating in Companion-002. If you look at the Kaplan-Meier curves for that study at 18 months, the overall survival was less than 10% in that study with a median overall survival in the FOLFOX arm of 6.2 months. Where we are today, and again, it's important to point out, these are -- this is a pooled survival number.

We're greater than 20% overall survival with greater than 17 months median follow-up. And we probably -- although it's hard to predict, of course, we probably will not be at 80% mortality until we have something like greater than 20 months median follow-up. So very interesting data today, obviously, extremely important. And as we mentioned in our press release, obviously, we don't know this, but it appears that tovecimig could be affecting overall survival in this patient population.

Okay. Let's now move to CTX-8371, our PD-1/PD-L1 bispecific antibody. Just a reminder of the differentiated mechanism of action here. Recall that this drug emerged from a screen using a proprietary technique at Compass in which we can screen bispecific drug candidates for synergy. That screen identified PD-L1 as a synergistic partner for PD-1 blockade. As I've discussed many times before, that was a rather unexpected scientific discovery. And because of that, we spent a long time investigating the mechanism of action. We published all that data, the PubMed ID for that paper is at the bottom of this slide. We have always envisioned 8371 to be on the leading edge of defining next-generation checkpoint inhibition. This mechanism of action on the right-hand side of this slide, with the bispecific being unequivocally a cell engager and quite fascinatingly, the bispecific actually converts PD-1 positive T cells into PD-1 negative T cells by removing PD-1 from the surface of those cells.

So it is a very differentiated mechanism of action. And again, we believe that this drug could be on the cutting edge of defining next-generation checkpoint inhibition. We're currently running a Phase I study. This study is a standard 3 plus 3 dose escalation study. Importantly, we, of course, as all Phase I studies do, we started with a minimal dose of 0.1 milligrams per kilogram, and we have finished enrolling the first 4 dosing cohorts, 0.1, 0.3, 1 and 3 milligrams per kilogram. We have not seen any dose-limiting toxicities in those 12 patients.

So again, a 3 plus 3 design, 3 patients times 4 dose levels, 12 patients. We're currently enrolling the fifth dose level, which will be the 10-milligram per kilogram dose level. The patient population in this study, importantly, is all post checkpoint inhibitor. So patients with melanoma, non-small cell lung cancer, head and neck cancer, Hodgkin lymphoma and triple-negative breast cancer are being enrolled in this study.

As I mentioned earlier, we now have 2 deep partial responses in the first patients enrolled in this study. And again, I will emphasize to you that the first dosing cohort was really a de minimis dose. These are CT scans on Slide 10 from a patient in the study with non-small cell lung cancer. For reference, these scan images across the top, the patient is lying on their back, dark color is air, so that's the lungs, lighter color is tissue and the bright white color is bone. Inside of the blue circle at baseline, you can see a 45-millimeter metastatic tumor in this patient, which over time, completely disappears. In fact, this patient had 59 millimeters, 5.9 centimeters, more than 2 inches total of metastatic tumor, which actually all disappeared. Really interestingly in this patient, this patient actually had initial pseudo progression at a lymph node, which has been described with checkpoint inhibitors like Keytruda and Opdivo, and it's interesting to speculate on what that might mean.

Subsequently, all of these patients' target lesions disappeared. We also have 2 patients with non-small cell lung cancer among 5 patients treated so far with prolonged stable disease for a clinical benefit rate of approximately 60%. So on the next slide, I'm going to spend a little bit more time on this slide because this slide is incredibly important. So this is a patient with metastatic triple-negative breast cancer who had 3 metastatic target lesions at baseline. I'm showing 2 of these 3 lesions on this slide. The third lesion was a lymph node. Across the top, same thing as the previous slide, patient lying on her back. Black color is air. So inside the blue circle, you can see a metastatic tumor in the lung, which completely disappears by 8 weeks. On the bottom, this is a sagittal view. So this is a reconstruction where you're looking at the patient from the side.

Inside the blue circle on the bottom left is a metastasis to the pericardium, the lining of the heart. That metastasis is 52 millimeters in size, 5.2 centimeters, more than 2 inches. Both of these tumors completely disappeared. The other target lesion went from 15 millimeters to 7 millimeters. So the total tumor decline in this patient from 87 millimeters to 7 millimeters is greater than a 90% reduction in this patient treated in the fourth-line setting who had previously received pembrolizumab. This patient is 1 out of 3 patients treated in the study with triple-negative breast cancer. So moving to CTX-10726, our proprietary PD-1 VEGF bispecific antibody. So this is a drug candidate that we worked on internally at Compass for about 18 months. We nominated it as a development candidate earlier this year. We have disclosed previously that we have more potent PD-1 blockade in vitro than has been reported for other drugs in the class.

Over the past 6 months or so, since we disclosed this as a development candidate, we've locked down our CMC process, I think one of the things that we have not talked much about at Compass is we've developed a fair amount of proprietary know-how around bispecific manufacturing and our manufacturing process has commercial level yields for this drug already before we're in Phase I. We're on track to file our IND in the US in Q4 of this year. Today, we're announcing some really interesting preclinical head-to-head comparisons of CTX-10726 with ivonescimab. These next 3 slides are, of course, complicated preclinical experiments, and I'm going to go through these in some detail. In this study, we're using a transgenic mouse model that expresses PD-1 and PD-L1 as human. So the extracellular domains of PD-1 and PD-L1 are knocked in as human, so you can directly test human targeted checkpoint inhibitors in this experiment.

Importantly, though, there's no human VEGF in this experiment. I'll come back to that point in a minute. Here, we're directly comparing the PD-1 blocking arms of 10726 with ivonescimab. And you can see that in terms of tumor control in this mouse model in a head-to-head study, 10726 is superior to ivonescimab. For those of you who are looking at these graphs very carefully, you can see that the control and ivonescimab arms end at week 28 because those animals had to be sacrificed due to uncontrolled tumor growth. On the next slide, in the same model, we compare 10726 directly with pembrolizumab. So again, this experiment is simply testing the PD-1 blocking arm of 10726 and comparing that head-to-head with pembrolizumab and 10726 is equivalent to pembrolizumab in this study. The next slide is a little bit more complicated experiment. So this is a xenograft experiment in which a human tumor, a non-small cell lung cancer model called HCC822 is injected into mice. That tumor secretes human VEGF-A.

So this experiment tests both PD-1 blockade and VEGF-A targeting. These experiments, of course, are done in immunocompromised mice. So a human immune system is added back to the mice. PBMC is peripheral blood mononuclear cell. So on the bottom left, you can see the control in black. Bevacizumab and ivonescimab are about the same in this experiment and the best drug in this head-to-head experiment is CTX-10726. As I mentioned earlier, we will be presenting these data at a scientific meeting later this year and filing our IND, which is on track for Q4. So on my last slide, we have some updated milestones here. And I think over the next 6 quarters, we have an incredibly rich milestone list here.

So let's start with tovecimig. In Q1 of the coming year, we'll read out our important progression-free survival and overall survival from our randomized study. I would imagine that, that readout would be followed by a very robust interaction with the FDA, which would put us into a position to potentially file a license application in the middle of 2026. Of course, we have Fast Track status, Fast Track designation. So I would anticipate that we would get a priority review. We will be initiating our planned basket study for tovecimig following that analysis in patients with DLL4 positive tumors, including potentially gastric cancer, ovarian cancer, hepatocellular cancer, et cetera. Still working on that design, but that study should be ready to go in the coming months. For CTX-471, we're planning to initiate our NCAM positive Basket study later this year. That biomarker was discovered in the Phase I study of CTX-471, and we presented scientific data for that drug twice last year.

And then finally, for 8371, very important update on that program today. Next step is initiating the cohort expansions in patients with non-small cell lung cancer and triple-negative breast cancer. Those cohort expansions will begin later this year with clinical data from that -- those cohort expansions next year. Hopefully, presenting the dose escalation data at a scientific meeting later this year. Finally, for 10726, the preclinical update that we've provided today, we're going to present that data at a scientific conference later this year, IND filing in Q4, which should put us in a position to read out clinical data next year. And lastly, we also, as part of our disclosure today, we ended Q2 with $101 million in cash, which is cash runway here at Compass into 2027, executing on all these programs and delivering the milestones that you see here. So with that, thank you again for joining the call today. I'm happy to take questions.

(Operator Instructions)

Andrew Berens with Leerink Partners.

Two questions from me. You're allowing crossover on the PAC arm to tovecimig. So is there any chance that the decreased you're seeing reflects performance of [tovecimig] in -- from the drug crossover? Can you give us any idea how many patients are crossing over on (inaudible) from the control arm? And then you mentioned interacting with the FDA. Any comments on potential breakthrough designation? And then I have one on the DLL4 biomarker testing after that.

Okay. Two important questions. Thanks, Andrew. So on your first question, I just went back to the study schema. So yes, we allow progression -- I mean, we allow crossover in the control arm following centrally confirmed progression. I'm going to answer your second question next. Ballpark, about half the patients crossed over in the control arm. So the statistical methodology that we're using, it's called the rank-preserving structural failure time, was the same methodology used in the IDH1 inhibitor analysis in biliary tract cancer. That statistical analysis adjusts the overall survival analysis for crossover. That is the defined primary method of analysis of the overall survival endpoint.

In terms of the general first question, I think potentially, the answer to your question is yes, that potentially even in patients treated in the third-line setting, tovecimig could be extending overall survival, and wouldn't that be fantastic, because if you look at the presentation of the Clarity data, the rank-preserving structural failure time, that analysis was a little bit better than the intent-to-treat analysis, which indicates that the drug was active even after crossover.

Now in our presentation of the overall response data, and that data are currently in our corporate deck on our website, the rate of progressive disease at week 8 was substantially different in the control arm than in the combination arm, 42.1% progression at week 8 versus 16.2% progression. So it doesn't seem like paclitaxel alone is particularly effective. So happy to take your additional question, Andrew.

Yes. And then just on BTD, any thoughts about doing that? Or would it not be -- if you're waiting for the OS analysis, would there not be any real benefit to applying for that at this point?

Probably. Yes. No, I think we'll -- that's sort of a work in progress, but it's obviously something we would -- obviously, we're thinking very, very seriously about.

Okay. And then just wondering, the DLL4 biomarker testing, how has that changed from that which was employed in some of the Korean trials?

It's the same. We tech transferred that analysis into the US, and we're using -- we used that and we're continuing to use that in the evaluation of biopsy specimens from some of the studies that we've done.

Okay.

Thank you.

Thanks.

Maury Raycroft with Jefferies

This is [Amin] on for Maury. A couple of questions from us. You mentioned the PFS OS analysis happening in 1Q '26. Can you clarify whether enough events might come in during Q4 to allow for analysis in early Q1? Or are you expecting that 80% of event threshold to actually be reached in Q1 itself? And I have a follow-up.

Sure. Thanks for the question. That's a very hard question. What we had originally projected is that we would have 80% -- we would hit the 80% event threshold by the end of Q3. What we know today is that's not going to happen. So beyond that, it's almost impossible to project accurately. And what we have simply said is we believe that the analysis will read out in Q1.

All right. Sounds good. And when it comes to the readout, should we be thinking about a full data set release? Or will some of the data be held back for medical meeting presentation?

Yes. I think we're planning to present a priority data set at that time, which would be PFS, OS, demographic data and top line safety data with the rest of the data to be presented at a medical meeting.

Okay, very helpful.

Michael Schmidt with Guggenheim

Thanks for taking my questions. I had just a logistical question. So if the Companion-002 study, in fact, succeeds on PFS and OS, I guess what else needs to be done to support a possible BLA submission around CMC, for example, how far along are you with some of the other sections that are required for BLA submission? And then I had a separate question on 8371.

Okay. Thanks, Michael. Yes, great question. Our so-called PPQ batches; process, performance, qualification, the batches that are required for a BLA submission are all underway. So our CMC process should be very much locked down. So that should not be limiting.

Okay. Great. And then yes, on 8371, interesting new data here today. Yes, could you just provide some additional commentary on the dose level where these responses were seen? And yes, have you seen any other patients with tumor size reduction? Or in fact, I think you mentioned two other patients with lung cancer with stable disease. Did you see tumor shrinkage in those as well?

So a couple of questions there. The non-small cell lung cancer response was at the 0.3 milligram per kilogram dose level and the triple-negative breast cancer response was at the 3.0 milligram per kilogram dose level. We're not releasing any other clinical data at this time. Again, hoping to present all of that data at a scientific meeting later this year, including the patients from the 10 mg per kg cohort.

Okay. Anything else you can share on the treatment history of the two case studies? Obviously, they did have a PD-1 inhibitor before, but anything else you could share there?

Yes. Most of these patients, sort of a typical Phase I population, many lines of therapy. This patient with triple-negative breast cancer had three lines of therapy in the metastatic setting, including previously some neoadjuvant and adjuvant therapy. So typical Phase I population, heavily pretreated, all the patients in the study having received prior therapy with a checkpoint inhibitor.

Great. Thank you.

Biren Amin with Piper Sandler

Hey, yeah, hi guys. Thanks for taking my questions. Maybe just to start on tovecimig. Your projections for the OS analysis and time lines for Q1, is that based on the current event rate that you're seeing? You mentioned that the event rate had slowed since your projection earlier this year? Or is the projection based on a lower event rate from what one would anticipate? I guess I'm trying to assess confidence on the Q1 analysis.

Yes. Great question. So I think the short answer to your question, Biren, is yes. Our projection is based on the current mortality rate that we're seeing. And I think over the last approximately four or five months, we've clearly seen a decrease. So we're basing the Q1 projection on the current rate.

Got it. Okay. And then as far as the Phase I IST, any update on that in terms of when we can expect first data from the quadruplet combination?

No, I don't have an update on that study at this time. That study is enrolling patients at MD Anderson, but I don't have an update today.

Okay. And then I do have several questions on 8371. Congrats on the data in the dose escalation cohort. So I just wanted to maybe ask for the non-small cell lung cancer patient, the patient had 0 centimeters of tumor. why is that patient not considered a complete response?

Yes. That is a great question. And actually, that question applies to the triple-negative breast cancer patient as well because -- so with this patient, this patient had some non-target lesions that did not qualify the patient as a CR. For the triple-negative breast cancer patient, the target lesion 3 is actually a lymph node and a lymph node less than 10 millimeters in the short access, and this is 7 by RECIST 1.1 is not pathological. So this patient's target lesion response was actually read as a CR. This patient also had nontarget lesions that did not disappear. So because of that, this patient is a RECIST PR.

Got it. And then maybe just a few more on 8371. You mentioned that the triple-negative patient had prior pembro. What about the non-small cell lung cancer patient? Did they have prior PD-1? And what was the therapy? And then I guess -- and then also, can you talk about what their PD-L1 status was when they received 8371?

What a great question. So the answer to that question is, I don't know. We did not read biopsy patients before the Phase I study. I do know that the non-small cell lung cancer patient at their diagnosis, so take that for what it's worth, had a very low PD-L1 expression. But we did not re-biopsy patients for the Phase I study.

Stephen Willey with Stifel

This is (inaudible) on for Steve. Congrats on the progress. I just have two

questions, one on tovecimig, one on CTX-10726. Just starting with the tovecimig, Tom, like I know that you will present patient demographics later when both PFS and OS analysis are ready. But like the fact that you haven't accrued like enough -- like any death events, how confident are you that these patients are actually reflective of like historical clinical trials and just like a real-world data. So that's the first question. And second question on the CTX-10726. By the way, very nice, impressive preclinical data in terms of tumor shrinkage. What can you -- what can you say about the safety data? And are you guys planning to present any preclinical data like prior to or soon after IND submission?

Thank you.

Thanks, Complex first question about demographics in this study. I'll simply say it's hard to know without the final data set and cross-trial comparisons to demographics. It's -- I think I probably can't really answer that. I think I'll simply say that the randomization in this study was stratified by three important prognostic variables, performance status, 0 versus 1. Metastatic disease outside the liver, yes or no. The vast majority of these patients had metastatic disease outside the liver, something like 80%.

And it was finally, it was stratified by anatomic subtype, intrahepatic cholangiocarcinoma or other. So because the randomization is stratified, I think the demographics in the two treatment arms will be very well balanced. So I think that's very important. Yes, thanks for your comment on 10726. Yes, we will be presenting scientific data for 10726 at a scientific conference later this year.

Aydin Huseynov with Ladenburg Thalmann.

Hi, good morning. Congrats on the progress this quarter. A couple of questions from us. So first, on tovecimig, the question is about the duration of response in the tovecimig arm of the trial. Could you provide any comments on the duration of response? And are these patients who initially responded still on the trial?

Thanks, Aydin. So I went back to the study schema here. So we don't have any analysis of duration of response at this time because in the statistical methodology, the first two secondary endpoints to be analyzed are PFS and then OS. So duration will be the last secondary endpoint analyzed. So I don't have any information on that. I don't -- most of the patients are in survival follow-up. I don't have a number today on how many patients are still on the study. I'm sorry about that Aydin.

Okay. That, that's.

Okay.

Okay. That's okay. Another question is on VEGF PD-1 bispecific. So I'm just trying to understand how the future is going to look like for this asset. So in the future, how do you see the regulatory path for your VEGF PD-1? And is it going to be as a single-arm path or like a head-to-head to pembro or nivo, just curious about the -- your thoughts on this.

Sure. I think great question, obviously. So I think the way we've been thinking about this is I think the regulatory path depends on very thoughtful indication selection. And where our thinking is with this drug is where we want to explore indications where both VEGF targeting and PD-1 targeting as monotherapies have been demonstrated to be effective.

So what are some of those? So renal cell, nivolumab and VEGF kinase inhibitors, gastric cancer where the VEGF receptor blocking antibody, (inaudible) is approved as well as PD-1 targeted agents. Obviously, hepatocellular cancer, where bevacizumab and atezolizumab are frontline standard of care and maybe something like endometrial cancer, where VEGF kinase inhibitors have also been shown to be effective.

And I think for some of these indications, say, the post-PD-1 VEGF patient with renal cell cancer, I think those could be single-arm pathways to approval. I think the non-small cell lung cancer indication is going to be, obviously, as you know, incredibly competitive, and we would probably not go there first.

Very helpful.

Robert Driscoll with Wedbush Securities.

Thanks. Good morning, Tom. Maybe just a follow-up on Biren's question. For the 8371 expansion cohorts, do you expect to select patients based on PD-L1 expression at this stage?

So we don't. We expect to leave the selection criteria the same as for the dose escalation portion of the study.

Okay. And then anything you can say with regards to immune-related adverse events here kind of acknowledging that we're still in dose escalation.

Sure. Yes. It's interesting. We believe that the way this drug might work, although we have some data to support this preclinically. Ultimately, we're going to need a lot more data to support what I'm about to say. So just that caution. We believe that this drug could be anchored in the tumor microenvironment by PD-L1, where it can provide very high concentration PD-1 blockade in the tumor microenvironment. So I'll simply say that in the first 4 dosing cohorts, we've not seen any dose-limiting toxicities. And look, as a next-generation checkpoint inhibitor, there's no reason to believe that the safety profile might not be better.

Got it. I'm looking forward to the update here. Thanks, Tom.

Thanks, Robert.

Sean McCutcheon with Raymond James

Hey guys, thanks for the question. Just a couple for me. How are you thinking about the necessary magnitude of benefit over paclitaxel on PFS given that paclitaxel not a commonly used chemo in second-line biliary tract cancer. And what gives you confidence that you've cleared the PFS bar for clinical adoption over FOLFOX? And then separately, could you give some context on kind of how you're thinking around OS and how FDA may be approaching OS and what you need to see as a trend on an ITT basis versus a [process adjusted] basis?

Thanks, Sean. So for the first question, I have sort of a two-part answer perhaps. So the first part, I'll simply summarize what the statistical power calculations are for the PFS analysis. So the PFS analysis is 80% powered for a hazard ratio of approximately 0.6. So a hazard ratio of 0.6 or lower would, I think, obviously be spectacular. I take your point about paclitaxel, and I think our overall response rate data where we had 42.1% radiographic progression in the paclitaxel arm at week 8 suggests that the median PFS in the paclitaxel arm is going to be in the 2 to 2.5-month range, about the same as has been seen with FOLFOX, for example, in the FOLFOX versus FOLFIRI randomized trial.

We recently completed some market research that was done by sort of a very well-known leading market research firm. And I think we were very pleasantly surprised at the KOL feedback on PFS from that market research, which frankly suggests that a hazard ratio of 0.6 would be off the charts. So KOLs would be happier just given what they know about FOLFOX with even less benefit. But a hazard ratio of 0.6 would be incredible. The second question, I don't have any information on your second question, never got any specific feedback from FDA about the difference between the ITT analysis and the RPSFT analysis of overall survival.

Understood, thanks.

(Operator Instructions)

Joe Pantginis with H.C. Wainwright.

So first, on tovecimig, first, a logistical question. Can you just remind us the level of meetings that you've already had with the FDA and what is planned and what role the potential for accelerated approval based on response rates have had in those discussions, number 1.

And then number two, obviously, there's a lot of talk here about the control arm. There's a lot of variables that are in this study. You've obviously talked about -- it doesn't seem like paclitaxel in the control arm is contributing to anything with regard to the data you're seeing. So with that, I'm going to ask about what are your thoughts on the perceived risk of any better-than-expected impact from the control arm, specifically from the fact that, obviously, these patients are receiving excellent clinical care. I mean it sounds like a rhetorical question, but I just wanted to get your views on the perceived risk

Yes, sure. Thanks, Joe. Maybe I'll take the second one first. So again, in terms of the control arm, again, I'm going to go back to the rate of progressive disease in the control arm at week 8, 42.1% radiographic progression at week 8. So we already have substantial progressive disease at week 8. And PFS, of course, is defined in the standard way, either radiographic progression or death.

We, of course, have not done any analysis of PFS in the study because we're trying to be very rigorous statistically. But just based on the rate of progression at week 8, it doesn't seem like we're seeing something outlandish in the control arm. In terms of a formal interaction with the FDA, we had a formal interaction with the FDA regarding the design of this study. So that was, of course, some time ago. So we would plan to have a much more robust and formal interaction with the FDA after we get the PFS and OS readouts from the study in the first half of '26.

I appreciate that. And then just quickly on 8371, can you take any broad strokes right now about the design and/or numbers expected within the dose expansion cohorts for the 2 indications stated?

Sure. Yes. So what we are currently planning in patients with triple-negative breast cancer, non-small cell lung cancer, a randomized study to two doses, a small randomized study, something like 50 patients ballpark in order to start to explore a couple of doses, have not picked the doses yet because we want to get all the data from the 10 mg per kg dose level. We should have most of that data by the end of this quarter, which would put us in a position to initiate those cohort expansions in Q4.

There are no further questions at this time. I would like to turn the floor back over to Tom for closing remarks.

Great. Thank you so much. And I'm just going to go to the last slide here again. Thank everyone for joining today and just highlight what the next 12 to 18 months could look like for us, readout from our randomized trial in patients with biliary tract cancer, followed by a robust interaction with FDA and then potentially a license application, which would obviously be incredibly exciting.

A couple more clinical trials with tovecimig and 471. And I think really exciting today our cohort expansion for 8371. And as I mentioned earlier, we have always positioned 8371 to be on the cutting edge of defining next-generation checkpoint inhibition and the data that we've reported today really puts us on track to achieving that goal.

And then lastly, our PD-1 VEGF-A bispecific antibody 10726. We have preclinical differentiation from ivonescimab, really looking forward to getting that drug into patients in 2026 and reporting Phase I clinical data. Thanks again, everybody. Happy to follow up with any questions that any of you folks might have.

Thank you. This will conclude today's conference. You may disconnect your lines at this time and thank you for your participation.