Compass Therapeutics, Inc. (CMPX) Earnings Call Transcript & Summary

Greetings, and welcome to the Compass Therapeutics Data Call. [Operator Instructions] As a reminder, this conference is being recorded. At this time, it is now my pleasure to introduce Dr. Thomas Schuetz. Dr. Schuetz, you may now begin.

Thank you so much. Thanks, everyone, for joining us today. I'm Tom Schuetz, I'm the CEO and scientific founder of Compass. I'm joined here this morning by Barry Shin, our Chief Financial Officer. So it's 8:01, we're going to start in a couple of minutes. So I apologize, we've had a little bit of a technical mishap here. The slide deck is not available via the portal that was in our announcement last night, but the slides are available on our website. So I'm going to give everybody a couple of minutes to get the slide deck up. So briefly, so our website, compasstherapeutics.com. So in the upper right-hand corner, you'll see a navigation needle, a compass needle. If you click on the compass needle, you'll get a drop-down menu, go to Investors, second from the bottom. When Investors is pulled up, there's a blue bar in the middle called Investor Menu. Click on News & Events, and you'll get a drop-down menu, go to Presentations. And the first one that will come up is called COMPANION-002 Data Disclosure. Because the slides are not live, I will be calling out slide numbers as we move forward. So that's April 1, 2025, COMPANION-002 Data Disclosure. Again, I'll give everybody a minute or 2 to navigate over to that. Again, Investors section, News & Events, Presentations and then the very first one on the list is a PDF file labeled COMPANION-002 Data Disclosure. And apologies for the technical mishap this morning. So I'm going to wait a minute, and I'll start at 8:05 just to give everybody a chance to navigate to the slide deck. Okay. I'm going to get started. Again, apologies for the technical mishap. I'll be calling out slide numbers as we move forward here. So Slide #2, just a brief reminder. I will be making forward-looking statements today, and I refer you to our regulatory filings for risks associated with those. Slide #3. I'm just going to start out with 2 introductory slides just to remind everyone about tovecimig, formerly known as CTX-009 and talk a little bit about the study design. But on Slide 3, tovecimig, a bispecific antibody targeting both VEGF-A and DLL4. VEGF-A, of course, is the well-known soluble ligand for the VEGF family of receptors that is a proven anti-angiogenic mechanism. DLL4, delta-like ligand 4 is the cell surface ligand for Notch-1. We believe the bispecific is anchored in the tumor microenvironment via DLL4, where it can provide high local capture of VEGF-A in the tumor microenvironment. And tovecimig is the only bispecific to our knowledge to demonstrate monotherapy activity in patients with a set of GI solid tumors. So Slide #4. So Slide #4 is just a review of the design of the randomized study. And just a reminder, this study, of course, is ongoing. This study is a 2:1 randomization of tovecimig plus paclitaxel versus paclitaxel alone in patients with advanced biliary tract cancer who have received one prior line of therapy. The study enrolled 111 patients into the combination arm of tovecimig plus paclitaxel and 57 patients into the control arm who are being treated with paclitaxel alone. The primary endpoint of this study is overall response rate as assessed by blinded independent central review. So all of the data that I'm presenting today have come from the blinded independent central review, sometimes abbreviated as BICR, BICR, but all of these data are from the central review. The secondary endpoints on the bottom here, progression-free survival, overall survival and duration of response, we are using the hierarchical testing method to control for alpha spending in the study. That method, which has been shown to control alpha better than alpha splitting requires that endpoints are assessed in a hierarchical way. So on Slide #5, I'm happy to report that COMPANION-002 is positive based on a statistically significant effect of tovecimig on the primary endpoint of overall response rate. So let's take a minute to go through this slide as there are a lot of data on this slide. First of all, this is a full intent-to-treat analysis, 111 and 57 patients, respectively, in the 2 arms. The overall response rate in the combination arm is 17.1%. The response rate in the paclitaxel control arm is 5.3%. So in the combination arm, approximately triple the control arm. The p-value, which is a two-sided p-value for that comparison is significant at p equals 0.031. Moving to the best overall response box, some categories here. We had a confirmed complete response via the blinded independent central review. Of course, complete responses in this disease are quite uncommon with gem/cis and a checkpoint inhibitor in the frontline setting, the complete response rate is single-digit percentages. So I think seeing a complete response in this study is quite important. Importantly, also, I think, second from the bottom here, there's quite a difference in the number of patients that have progressive disease as their best overall response. 16.2% in the combination arm versus 42.1% in the paclitaxel arm. So just to be clear, the first scan that was obtained in this study was at week 8. So by week 8, 42.1% of patients in the paclitaxel control arm already had radiographic progressive disease. I'll come back to that point in a minute when we review the waterfall data. So on the bottom of this slide, because the study is ongoing, we don't have a full safety analysis yet, but the safety profile has been consistent with our prior studies. Importantly, we had an independent data safety monitoring committee that reviewed safety data at 4 separate times during the study. And at each occasion, they recommended that the study continue. Moving on now to the next slide, Slide 6. Slide 6 shows the waterfall plots from the study. So you can see, I think, quite clearly that these waterfall plots are quite different. On the top here, in the paclitaxel monotherapy arm, the majority of patients are -- have had increases in their linear tumor burden. You can see the inflection point in the waterfall plot is to the right. On the bottom, with tovecimig, the inflection point has been shifted quite a bit to the left and the vast majority of patients have actually had a decline in their linear tumor burden. I just want to highlight on the bottom right-hand side of this slide, there are actually 4 patients in the study who have had 100% declines in the combination arm in their linear tumor burden. One of these patients was officially categorized as a complete response. The other 3 patients are partial responses based on some subtle differences in their non-target lesions, something that is commonly seen in these studies. So given the fact that we have such a difference in patients who have progressive disease combined with this waterfall, gives us confidence in the progression-free survival analysis. And on Slide 7, we give a summary of the treatment landscape in this disease, and let's talk a little bit about the timing of the secondary analysis. So because we're doing the hierarchical testing method, it's extremely important that the study was positive on the primary endpoint. That now allows us to allocate a full alpha of 0.05 to the next analysis, which is progression-free survival. The time-to-event analysis, of course, are triggered by total number of events. At 80% total events, these analysis will be triggered. And these events have been tracking less than we had originally modeled. So currently, we believe that the time-to-event analysis will be triggered in late Q3 of this year, which would put us in a position to report PFS, OS and duration of response in Q4 of this year. And I think you can see also on this slide, one regimen that has been used in patients with this disease in the second-line setting is the 3-drug combination FOLFOX, tovecimig plus paclitaxel, again, is more than triple the response rate seen with FOLFOX. Moving to Slide 8, just a summary of the current treatment landscape and brief discussion about the patient population. So believe it or not, there are only about 15% of patients with this disease that have a genetic alteration that makes them eligible for a targeted therapy, be that an FGFR2 inhibitor, an IDH1 inhibitor or a HER2-targeted drug. The other 85% of patients have no labeled treatment option in the United States. So we believe that tovecimig plus paclitaxel could be positioned to become second-line standard of care in this disease. Next slide, Slide 9, how many patients are out there. This slide is updated with the 2025 SEER numbers that were published in January, about 23,000 patients annually in the United States based on the SEER data, the claims aggregator, Komodo Health, presented an analysis about 2 years ago. Based on claims data, they determined that there were about 22,800 patients annually in the United States. And there was also a recent publication in the bottom box here, an epidemiologic analysis indicating that by 2040, primary hepatocellular and intrahepatic bile duct tumors were actually going to become the third most common cause of cancer-related death in the United States. So this incidence will continue to increase over the coming decade. So the last slide is Slide 10, just a summary of our anticipated milestones for the rest of this calendar year. With tovecimig, we now have a positive randomized study driving toward an analysis of the time-to-event endpoints in the second half of the year, probably at the end of Q3 or at the beginning of Q4. Other important milestones for us this year include analysis of our Phase I data for 8371, our PD-1, PD-L1 bispecific antibody, which should occur in the second half of the year. A reminder that, that's a Phase I study. But of course, if we see any activity in that study, that could be very meaningful. And as we announced earlier this year, we also have a PD-1 VEGF-A bispecific antibody that we're driving toward IND filing by the end of this calendar year. So thank you very much, all of you for joining today. Again, my apologies for the technical mishap this morning, but I'm happy to take questions now.

[Operator Instructions] The first question today is from the line of Maury Raycroft with Jefferies.

Congrats on the update. Maybe first question, if you can just talk more about upcoming next steps for BTC and whether you would meet with FDA prior to getting durability data? And if you do that meeting, what would be the objectives of that meeting?

Sure. I think we would like to get these data in front of FDA as soon as we can. I think ideally, we'd like to have an interaction with FDA this quarter, Q2. The objectives of that meeting would be to get their perspectives on the data and have an initial conversation on the data set that might support a license application in the U.S. I think -- thanks for the question, Maury. I think we have -- our base case assumption has kind of always been that the full trial results would be required for a license application. And I think -- today, I think we would continue to maintain that base case assumption. So just having a conversation about the time-to-event analysis, the timing of those and follow-up from those analysis with FDA and thinking about the regulatory significance of this study, I think, would be the main goal of that interaction.

Got it. Okay. That's helpful. And then based on this data cut that you have, you're seeing a big difference in progressive disease for the paclitaxel monotherapy arm. At this point, do you have any perspective into how this could influence durability? And since the response rate data are different from the Phase II data that you have, do you think OS and PFS from the Phase II are the proper benchmark to consider?

It's an interesting question. I think I have a couple of thoughts on that perhaps. Yes, I think the overall response rate is lower than we saw in Phase II. I think that's not some surprising result when you go to a large randomized trial. I think one of the things that we did when we modeled the time-to-event analysis, we used the lower bar, the lower boundary of the 95% confidence interval in order to model the time-to-event analysis. And I think just simply based on the fact that we're not there today, suggests that the PFS and OS data from that study are probably relevant to our modeling for this study. And I would say that it appears just given where we are in time that we're probably actually higher than the lower bound of the 95% confidence interval.

Our next question is from the line of Biren Amin with Piper Sandler.

Congrats on the data. Maybe to start on baseline patient characteristics. Can you talk a little bit about the distribution of patients across intrahepatic, extrahepatic and gall bladder?

Sure. So we don't have that data today, Biren. But I will add that the tumor subtype was one of the stratification factors for the study. So the stratification factor was intrahepatic cholangiocarcinoma or not. So intrahepatic cholangiocarcinoma has to be perfectly balanced between the 2 arms. That was a specific FDA request by the way. I suppose it's possible that gallbladder and extrahepatic could be slightly different. That's possible. Ampullary possible as well. But we don't have that information yet. Again, an ongoing study. The baseline characteristics is probably something that we could -- we'll be able to get probably earlier than our time-to-event analysis, but we don't have that today.

And then maybe a second question on the PFS and OS analysis that's expected in late 3Q. Given that the last patient enrolled was in August, and so it's -- august of last year. So it's about a full 12 to 13 months after that. What do you think is driving the effect? Because clearly, this is a 2:1 design. You're going to need the active arm to progress. And as you mentioned, that you're running longer than your Phase II analysis. So what exactly would be the drivers? And I guess, is there a risk around censoring of events due to safety issues, for example?

Okay. A very complex question, of course. Let me try to take those in order. So your math, of course, is correct that by -- the study was fully enrolled in August of 2024. So late September, so late Q3 would be September, of course, of 2025, that would be 13 months from the last patient enrolled and the study enrolled from approximately July, August of '23 to August of '24. So the median follow-up would be much longer than 13 months. So what's driving that? So it's always hard to know when you're monitoring total events. We're not monitoring the events in the 2 treatment arms. We're just monitoring total events. So I think there are always a couple of possibilities. Of course, the control arm could be performing better than modeled. But I think that seems a little unlikely because on Slide #5 I would refer you to the fraction of patients with progressive disease. And I think if anything, paclitaxel could be performing slightly worse maybe than we had modeled, but we don't know. Of course, we could be seeing a drug effect in the fewer total number of events, but this is something that we just can't know yet. Your question about censoring is it very, very important. So we're defining progression in the standard way. So progression is either radiographic progressive disease or death. That is the common definition -- commonly accepted definition of progression for progression-free survival analysis. And so I think that unfortunately, in second-line biliary tract cancer, I think a patient comes off this study for a reason, whatever that reason might be other than radiographic progression. Unfortunately, they're probably relatively close to death. And that would then be the progression event. So I think we're not expecting a particularly high number of censoring. And I would also refer you to the very bottom row of Slide #5. So our number of patients who are not evaluable is actually low, 14%, 15%. So not evaluable means they -- the patient did not make it to week 8 for whatever reason. And in the ABC-06 study, I'll point out, and you can go back and look at the survival curves, but I'll point out that the 2-month mortality, mortality in ABC-06 was almost 20%. So with this disease, believe it or not, the 2-month mortality is about 20%. So we had 14%, 15% of patients that didn't make it to week 8. Not all of those are for mortality. So I think our censoring is probably going to be less than we had originally anticipated.

Our next question is from the line of Michael Schmidt with Guggenheim Partners.

Tom, this is Rosy on for Michael. Congrats on the data. Just a couple of questions here from me. I guess just thinking about the time on therapy, do any patients still have like ongoing responses or still on treatment at this point in time? If you can provide any color there, that would be great. And then I guess my second question is if tovecimig is approved, I guess, how do you anticipate the drug being used over something like the [ NIFD ] regimen, which also showed a comparable mid-teens response rate?

Yes. I think -- so we currently have just under -- to the first part of your question, we currently have just under 20% of patients remaining on study. So which is interesting. Here we are 7, 8 months from the last patient randomized. Yes, I think the second part of your question is, I think, interesting. I think we would position this drug to be sort of second-line standard of care in this disease. And I think ultimately, that treatment decision, I think, will require some information about progression-free survival and overall survival.

Our next question is from the line of Stephen Willey with Stifel.

Congratulations on the results. Just a quick clerical question here. So I know the responses are all blinded independently, right? But can you just -- can you just confirm that they are indeed confirmed responses?

So Steve, you mean by RECIST 1.1?

Correct.

So confirmation of responses in randomized trials is not required in RECIST 1.1.

Okay.

But we did collect that data, and we'll be reporting that as a secondary endpoint.

Okay. And then I know you -- I know the DMC has looked at safety. Just curious if you have seen safety data, whether it be on a blinded or a unblinded basis? And if so, is there anything that you can say about some of the DLL4 specific adverse events of special interest, whether it be pulmonary hypertension or more cardiac related?

Sure. So we have been monitoring safety data, of course, and working with the data monitoring committee. The last data monitoring committee meeting, the study had just been fully enrolled. So I'm somewhat hesitant to make a general comment about safety based on that information. But to your specific question, I think as of the interim looks in this study, the interim looks at safety that is, there appear to be less of the DLL4 type toxicities than we had seen previously. Again, take that with a grain of salt because that's an interim look, not the full data -- not the full safety package. So we really, today, don't have a full answer to your question.

Okay. And then just lastly, I know you talked about wanting to perhaps engage with FDA at some point here soon, maybe second quarter. Are you hoping to have at least some level of clarity around what a path to registration might look like as a function of that conversation? And I guess if you don't are you thinking about potential confirmatory trial designs that would perhaps be needed if the FDA determines that the approval here is warranting an accelerated one? And I guess, is that something that you would be willing to start at risk? And is that something that you would move to the second line? Or would you try to move this into a frontline regimen?

Sure. Very complicated question, of course. I don't know the answer to the first part of your question. It would certainly be our hope that we could sort of get to some conversation with FDA about path to approval, but I don't know the answer to that today. If this study supports accelerated or full approval, I think that will be a function of the effect of the drug on PFS and OS. In terms of the concept of a confirmatory study, I think that when we look at the top line data from this study, I think it would be quite difficult to repeat this study in this patient population. I think we have enough evidence here of a clear treatment effect that I think it would be quite difficult to just do a second study. So that would get us to a frontline study, which would be adding tovecimig to something like gem/cis/durva. The study -- that's part of the point of the study that we're doing with MD Anderson. And ideally, by, say, around the same time that we had PFS and OS from this study, we would have some preliminary data from the frontline study at MD Anderson that could help guide us one way or another. And I think to the specific question that you asked, I think if we got a hint from the FDA that we were going one way or another, I think, yes, we would start a confirmatory study at risk if they told us that this study would support accelerated approval. I think we would do that.

Our next question is from the line of Laura Prendergast with Raymond James.

Congratulations on the data. Can you provide any clarity on what the median time -- what the median follow-up time at the cut for ORR analysis was? Also any clarity on median time to response. And if you saw any deepening of responses between scans?

Sure. Thanks, Laura. So yes, the first question you asked, obviously, super important. So the study -- 95% of patients in the study were enrolled between August of '23 and August of '24. So the median time that -- the median patient that was enrolled was approximately the end of February of '24. So today, we've got about 13 months of median follow-up. So to your second 2 questions, time to response and deepening of response, I don't have that data today.

Our next question is from the line of Joe Pantginis with H.C. Wainwright.

Congratulations as well. So 2 questions. I'm looking forward here into the future. Number one, how would you define your manufacturing needs right now for tovecimig. And also based on the data you're seeing today and the mechanism of action, any potential read-through for the CRC study?

Thanks, Joe. We've been doing our standard license application-ready manufacturing. We've got a contract manufacturing organization that we use, which is based in the U.S. So I think that's important. So I feel good about where we are on the CMC side. Your second question, I think, is fascinating, of course. We clearly have a drug effect here. And I think one of the things that we're going to be doing is thinking about the best next indications to begin to think about because as you know, there are sort of an enormous number of -- an enormous set of DLL4 positive malignancies, gastric, colorectal, renal, ovarian, liver, BTC, of course. So it's quite substantial. And I think we're going to -- once we take a little bit more thorough dig into the data, we'll begin to think about that. And I think maybe sort of in the next -- maybe in the next quarter, we'll have sort of a better idea about our thoughts on other indications.

Congrats on this great data.

The next question comes from the line of Aydin Huseynov with Ladenburg Thalmann.

Congratulations with the nice ORR order of data. I've got a couple of questions. So out of 24 patients on the paclitaxel arm who progressed, how many patients do you expect to cross over to the tovecimig arm?

Thanks, Aydin. Great question. I don't have the exact number today, but something that's quite important. And I think your question and Biren's question also are related because I just want to remind everyone that in order to cross over from the control arm, you have to have centrally confirmed progressive disease. In addition, you needed -- patients needed to have fulfilled selection criteria again. So I don't have the exact number of patients that crossed over. So I'm just going to give you a ballpark number here. I think about half of those patients ultimately crossed over after progressing in the control arm.

That's helpful. And do you have any thoughts on -- or maybe data on the patients who progressed on the tovecimig arm in terms of their prior therapies? Did they have durvalumab, pembrolizumab? And also on that one particular CR patient, what was the prior therapy for that patient?

Yes, great question. I don't have that data today. That's something we're going to look at specifically. Patients were required to have a gem/cis containing regimen, but where that's something we're going to look at carefully as we begin to dig in this data over the coming months.

Okay, that's fine. And the last question on the confidence interval for ORR for both arms, if you could share those?

Sure. I don't have that, but I'll look that up for you.

At this time, there are no further questions. I would like to hand the floor back to Dr. Schuetz for closing remarks.

Okay. Great. Again, I just want to thank everyone today for joining. Again, we're super excited to report that we have a positive study here. And I think we're really focused on driving to the time-to-event analysis. And we're just -- we're happy to follow up with any of you all if you have follow-up questions, we're available to meet with people. But I just want to thank everyone again and really want to again thank all the patients and their caregivers and the investigators in the study and the clinical team at Compass for delivering these data.

Thank you. This does conclude today's teleconference. We thank you for your participation. You may now disconnect your lines at this time.