Compass Therapeutics, Inc. ($CMPX)

Greetings. Welcome to Compass Therapeutics Top Line Secondary endpoints Call. [Operator Instructions] Please note that this conference is being recorded. At this time, I'll turn the conference over to Anna Gifford, Chief of Staff. Thank you, Anna. You may now 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. Earlier this morning, we released positive secondary endpoint data from the Phase II/III [indiscernible] study, which we look forward to discussing on this call. 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 Dr. Thomas Schuetz. Tom?

Thanks so much, Anna. And I just want to echo Anna's greeting. Happy to be with you all today, and we're very excited today to discuss with you the initial data from our COMPANION-002 randomized trial. Today, going to be discussing [indiscernible] of the secondary endpoint efficacy data, as well as demographic data and some top line safety data. The rest of the analysis from the study, we look forward to presenting at a major medical meeting later this year. I would also like to echo Anna's forward-looking statement comment. I will be making forward-looking statements today. Okay. So on Slide 3 is an executive summary of the data analysis so far in our COMPANION-002 study. So just to reiterate, we hit the primary endpoint in the study of overall response rate, more than tripling the response rate in the control arm, and we presented those data last year. The key secondary endpoint was progression-free survival. And today, we're reporting what can only be described as an absolutely stunning result. We have a 4.7 month versus 2.6 month median bicker assessed progression-free survival with a hazard ratio of 0.44, and a p-value less than [ 0.0001 ]. I'll talk more about the data in a minute. In terms of overall survival, the overall survival analyses were confounded by crossover from the paclitaxel arm to the combination arm, 54% of patients crossed over and 85% of patients in the study ultimately received tavecimig in combination with paclitaxel. I'll talk more about that, of course, in detail. Fascinatingly, those 54% of patients who crossed over from the control arm lived an incredible long time. The crossover patients had a mean overall survival of 12.8 months which is consistent with what you would see in the frontline setting with gemcitabine/cisplatin [indiscernible] patients who did not cross over had a median overall survival of 6.1 months. I'll show all of those data in this presentation. The safety profile was consistent with our previous studies. I'll talk more about that, of course, Importantly, we're now preparing for an FDA meeting in advance of a planned BLA submission. It is our intention to seek full approval. However, having hit on overall response rate and really dramatically hitting on progression-free survival, we feel at absolute minimum, this data set supports accelerated approval, and I'll talk more about that in a couple of minutes as well. So let's just quickly review the design of the study. COMPANION-002 was a randomized trial in patients with advanced [indiscernible] tract cancer who had received one prior line of therapy. So this is a pure second-line study. It was a 2:1 randomization of Tovecimig plus paclitaxel, versus paclitaxel alone. 111 patients were randomized to the combination arm, 57% were randomized to the control arm. In the control arm, after centrally confirmed progression, patients were allowed to cross over and receive Tovecimig plus paclitaxel. And as I mentioned, patients from the paclitaxel only arm crossed over to the control arm, to the combination [indiscernible]. So therefore, 31 plus 111, 142 patients in the study received Tovecimig at one point during the study. I'll talk more about that in a couple of minutes. On the bottom of this slide, is the statistical hierarchy. Overall response rate was to be analyzed first progression-free survival, second. Overall survival, third. In terms of duration of response, we don't have that analysis today, and we look forward to presenting that at the medical meeting presentation later this year. Okay. Let's dive into the data from the study. So let's start with the baseline demographics that are shown here. Men, women and [indiscernible] the bottom 3 rows are the most important. These were the prognostic variables that were the 3 stratification factors for the randomization. The first was anatomic location of the primary tumor, [indiscernible] [ cholangiocarcinoma ] versus other very well-balanced ECOG performance status 0 versus 1. You can see those are within single-digit percentages of each other. And then finally, disease status, disease outside the liver or locally advanced disease. The majority of the patients in this study had metastatic disease outside the liver. And all of these 3 important prognostic variables were well balanced between the 2 treatment arms. Just again, an important reminder here that we hit the primary endpoint in the study. All of these analyses were done via blinded independent central review, so-called [indiscernible] So the [indiscernible] assessed response rate was 17.1% in the combination on with Tovecimig, 5.3% in the control arm with paclitaxel, we had a centrally adjudicated complete response in the Tovecimig combination arm. Okay. In terms of [ bicker ] assessed progression-free survival here are the -- here are the PFS curves. The red line is the Tovecimig plus paclitaxel group, a median progression-free survival of 4.7 months. A median progression-free survival in the paclitaxel control arm of 2.6 months. That revealed a hazard ratio of 0.44 again with an extraordinarily significant p-value of less than [ 1x 10 to minus 4 ]. Just a reminder also that we powered this endpoint for a hazard ratio of approximately 0.6. So a hazard ratio of 0.6 represents a 40% reduction in the risk of progression. So 100 minus 60. And what we saw was a 56% reduction in the risk of progression, about 50% better than we powered the study for. Let's move to overall survival. This is the intent-to-treat analysis for overall survival. The overall survival analysis was confounded by crossover, and I'm going to talk a lot more about that in 1 minute, but this is the intent-to-treat analysis. Now what is most important to understand here is that the "control arm" includes 2 different sets of patients. The 54% of patients who crossed over and received Tovecimig in combination with paclitaxel, and the 26 patients who did not cross over and received paclitaxel alone. But this is the intent-to-treat analysis where the patients are allocated to these 2 curves based on the treatment to which they were randomized. But again, it's very important to keep in mind here that the "control arm" is really 2 different sets of patients, and we're going to talk a lot more about that in a minute. One of the things that we have discussed in the past with all of you is the crossover adjustment statistical methodology. That methodology is called the [ RPSFT ], or the rank-preserving structural failure time. So the right preserving structural failure time statical assumptions were not met. So you can see that the RPSFT, and the intent to treat analysis are virtually identical. So one of the prespecified secondary endpoints is an endpoint that we called [ PFS 2 ]. And when we saw the attempt to treat OS curve, we wanted to look specifically at the PFS 2 analysis. So PFS 2 is an analysis of progression-free survival in the crossover patients. So you can see on the bottom left, where you see number at risk 31 and 31, those are the same individual patients. So pre-crossover in red, post crossover in blue. And you can see that there is a significant difference in progression-free survival post crossover. The patients who crossed over actually had a very rapid initial progression with a median PFS of 1.9 months. Then after crossing over to receive Tovecimig, their PFS improved to 3.5 months. Now times 0 on these two curves is different. Time 0 for the red line is randomization. Time 0 for the blue line is time of crossover. So once these patients who have progressed very rapidly on paclitaxel crossed over to receive Tovecimig, their progression-free survival improved significantly and those two curves are different with a p-value of [ 0.0016 ] and a hazard ratio of 0.36. Another kind of subtle aspect of this curve, perhaps is you can see that there is a rather long tail on this progression-free survival curve. So there are many patients here. And by the way, this blue curve, of course, is now patients treated in the third-line setting. So many patients here who have not progressed 6, 9 and 12 months out after crossing over to receive team. That observation led us to do this analysis. So this is an analysis of overall survival in the control arm. What we did is we looked back and we split the patients into the crossover patients and the patients who did not cross over. The patients who crossed over, [ N equals 31 ], had a remarkable median overall survival of 12.8 months, compared to 6.1 months median overall survival with paclitaxel alone in the patients who did not cross over. Now obviously, you all know this, and we know full well that when you do an analysis like this, you might have simply selected the patients who were going to do better anyways. And maybe the patients who happened to cross over to Tovecimig were the patients who might have done better anyways. So what we did is we did this analysis for progression-free survival. So as I go to the next slide, just keep an eye on the red and blue curves. For progression-free survival, they are reversed. So the patients who actually crossed over in red actually did worse on paclitaxel than the patients who did not cross over, despite progressing much more rapidly, actually significantly rapid -- more rapidly with a p-value of 0.007. Despite those patients progressing more rapidly. I'm just going to go back 1 slide, they ended up living significantly longer once they cross over and receive Tovecimig. So we believe this analysis suggests strongly that the addition of Tovecimig to paclitaxel is associated with an improvement in overall survival. These slides, by the way, are available on our website this morning. This is top line safety data. Lots of numbers on this slide. Just want to highlight one number here. The main difference between the 2 treatment arms is hypertension. Of course, hypertension is a mechanism [ or VEGF ] blockade. There are now published algorithms for managing the hypertension associated with angiogenesis inhibitors. So here's a summary and a brief discussion of our next steps. So the study, obviously, is positive by definition with a significant improvement in overall [ response ] rate more than tripling that what you saw in what we saw in the control arm. For progression-free survival before crossover, there was an incredibly significant treatment effect with a 56% reduction in the risk of progression, outside of targeted therapies to my knowledge. This is the lowest hazard ratio on progression-free survival that's ever been reported in this disease. The OS analyses were confounded by crossover. And the patients who crossed over, despite initial very rapid progression, lived an incredibly long time at 12.8 months median. As I said earlier, 12.8 months is the exact median overall survival in the experimental arm of the study [ TOPAZ-1 ]. So in the front-line setting, gemcitabine/cisplatin plus [ Imfinzi ] has a median overall survival of 12.8 months and those crossover patients now treated in the second and third-line setting, had a median overall survival of 12.8 months. Our safety profile was consistent with our prior studies, and the drug was well tolerated. And as I mentioned earlier, our next steps will be to meet with the FDA to discuss these data in advance of a BLA submission. As I mentioned earlier, it is our intention to seek full approval based on these data. But after a discussion with FDA, we certainly believe that the combination of ORR and PFS supports accelerated approval as it has for many, many dozens of other oncology drugs. We feel quite strongly about our data because when our data are put in context, compared to several different chemotherapy regimens that are used in the second-line setting, such as FOLFOX or FOLFIRI, where one randomized trial was done, you can see that FOLFOX had a 2.8-month median progression-free survival compared to 2.1 months with FOLFIRI, median overall survivals 5.7 and 6.2 months with our study, the pooled median overall survival is approximately 9 months, 50% longer than you see with any chemotherapy regimen in the setting, significant improvement in overall response rate, very significant improvement in progression-free survival. Overall survival ITT analysis confounded by crossover and the crossover patients lived an unbelievably long time and just a very, very satisfying observation for us to see how long those patients lived. So next steps for us put all the data together, meet with FDA and discuss the pathway to a biologics license application submission. Thanks for your time this morning. Happy to take questions.

And the first question is from the line of Jonathan Chang with Leerink Partners.

First, how do you view the efficacy and safety profile of Tovecimig plus paclitaxel, compared to chemotherapies used in this treatment setting? And just as a follow-up, can you provide more color on the crossover adjusted OS analysis based on RPSFT? How should we interpret those data given that the statistical [indiscernible] that analysis weren't [indiscernible]?

Okay. Thanks, Jonathan. So in terms of the first question, of course, we didn't -- when you look at paclitaxel in comparison to FOLFOX or FOLFIRI, fairly similar safety profiles there. The real difference here is, again, hypertension which again is on mechanism for angiogenesis disruption. There are now published algorithms for managing the hypertension associated with drugs like Avastin. And in the Avastin label, there's language about treating the hypertension rather than discontinuing Avastin, and that's the approach that we took in this study. Your second question is paradoxically very complex and very simple. So the statistical assumptions for the RPSFT analysis were not meant. So the RPSFT numbers that we're reporting here, we're reporting for full disclosure -- just for full disclosure. But the RPSFT analysis is meaningless, which is why we presented the RPSFT -- why we presented the [ ITT ] analysis on Slide 8.

Our next question is from the line of Maury Raycroft with Jefferies.

This is [indiscernible] on for Maury. A couple of questions from us. First, on just to make sure we're interpreting the data correctly and using the right comparisons. If we look at the [indiscernible] patients versus patients who received paclitaxel without crossover, it looks like the OS figures are like 8.9 months versus 6.1 months. Is that the right way to think about the underlying treatment effect here? And I have a follow-up on the FDA interaction.

Yes. That is completely correct. So the combination arm 8.9 months, the paclitaxel only arm 6.1 months. correct.

Okay. That's clear. And have you had any regulatory discussions with the FDA in the past few months to see how they view the bar for successful study? And if not, when do you expect to have your next discussion with the FDA?

Okay. Two-part question. The answer to the first part of your question is no. We have not had any discussions with the FDA in the past few months. The second part of your question, we expect to complete the analysis of all of the other endpoints in the study within the coming approximately 1 month. And then we would put everything together into a meeting request, and package and send all of that into the FDA, which should put us in position to have a meaningful conversation with the FDA in approximately the middle of the summer. If that conversation goes well, then that would put us in a position to submit a BLA application before the end of this year. And because we have a fast track designation, of course, that could potentially put our PDUFA date, sort of inside the second half of next year.

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

Can you maybe talk about the influence of drug discontinuations, or dose reductions with the [indiscernible] arm versus [indiscernible]? And then maybe another question. I know Tom, you're not talking about duration of response today and that's going to be at a medical presentation. But what were the data there and read through on PFS? If you could just maybe tell us, I guess, from like a high-level perspective.

Okay. A 2-part question. In terms of all of the data for dose reductions and treatment discontinuations. I don't have the totality of that data set today. I did because hypertension was most common. I looked at that specifically, and there were 4 patients, 3.7% who had drug discontinuation from hypertension. The rest of that data is in process for analyzing. In terms of your second part of your question, in terms of DOR, we did not prioritize that analysis for this presentation. The only thing I'll maybe add to that is one of the things that was very, very interesting to us, is in the PFS2 analysis. So again, patients who very rapidly progressed on paclitaxel only, then crossed over, there are additional responses in those patients after crossover. So that's also a very interesting piece of data that we're scoring through now.

And maybe if I could have one follow-up. On the patients that crossed over, were they more [indiscernible]? Or were there any baseline characteristics? Because if I look at [indiscernible] on PFS 2, I think you reported 3.5 months on the patients that crossed over on PFS 2. And if I calculate and add the 1.9 months that they had on PFS 1, that gets you to about 5.4 months. And so where I'm going with this is your OS was significantly higher at 12.8 months, compared to the Tovecimig arm at 8.9. And I think the difference is on those patients that cross over? I mean, they seem to have been slightly better. So was there anything from like a baseline characteristics of the crossover patients?

So I don't know the specific answer to the question as you phrased it. We are examining all of the information for those 31 patients incredibly carefully. And we're going to be in a position to talk more about that. I very much, Biren, I think that the caveat that you made here is quite important, right? We need to make sure that these patients who crossed over that we're not biased here by selecting a patient population that was going to do better anyway. And that's why we did the PFS analysis on paclitaxel that we show on Slide 12. And as if PFS is a marker here, not only did the crossover patients not do better, they did much, much worse on paclitaxel only. So as measured by PFS, the crossover patients were actually a worse subset. And despite that fact, they lived significantly longer once they crossed over to receive Tovecimig. [Technical Difficulty]

[indiscernible] your question.

Tom, can you hear me?

Yes. Yes. Sorry.

Sorry, everything went silent there for about a minute or so.

Yes. Sorry about that.

Maybe to follow up on Biren's question. Do you have much, if any, patient biopsy data? And are you going to be looking at retrospect, DLL4 expression in these patients, I guess, specifically in those there on patients who crossed over and did better on Tovecimig, with the intention of trying to see again, if you can [indiscernible] out anything that's specific about this patient subset?

Yes. So we collected archived biopsy specimens in the study, and we're now -- we're going to embark on sort of a comprehensive biomarker analysis with -- the real -- I think the real onus on us is to do as much work as we can to try to [ tease ] out why those patients who crossed over did so amazingly well. So yes, to your question, we collected biopsy specimens and we're going to embark on a comprehensive biomarker analysis of those specimens and in addition, blood samples, of course.

Okay. And then to the extent that this potentially becomes a conversation with the FDA about an accelerated approval pathway. Any additional, or just initial thoughts right now as to where you would look to conduct a confirmatory Phase III trial, whether that would be in the second line? Or would you look to do this in the front line? And just how quickly you think you might be able to get this up and running?

Sure. So again, just to repeat, it is our intention to seek full approval based on this study. These data are only single-digit days old here for us. But just some high-level thoughts on your question. Adding Tovecimig to the frontline regimen of gemcitabine/cisplatin and [ devolumab ] is something that we're doing right now. I don't want to go too far into a rabbit hole here with you, Steve. But something else that we could do is we could swap our own PD-L1 blocking antibody [indiscernible] for durvalumab in that study, which would be something that we'd be incredibly excited to do. Our frontline study adding Tovecimig to gemcitabine/cisplatin [indiscernible] ongoing at MD Anderson, and we're continually reviewing data from that study with the investigators, again, to help us think about future studies. Another study that we think about would be Tovecimig/paclitaxel versus FOLFOX in the second-line study. Maybe that's a study that we do in Europe, where FOLFOX is probably used a little bit more commonly in the second-line setting than it is in the U.S.

Okay. That's helpful. And then just lastly, I understand you're going to have kind of a more comprehensive update of safety data at a future medical meeting. But is there just anything that you can say about [indiscernible] toxicity? And I guess I'm specifically referring to something like pulmonary hypertension?

Sure. I never -- you and I have talked about this before. As a [ bispecific ] antibody, it's hard to ascribe anything to either end of the molecule, right, either safety or efficacy. But since you asked specifically about pulmonary hypertension, you -- as you all review this slide deck, you'll see that on Slide 13, pulmonary hypertension is not on that slide. So in the study, there were 3 [indiscernible] as of pulmonary hypertension 2.8%. So something substantially lower than we saw in our Phase I and Phase II studies in Korea. I don't have an explanation for that. I do know that we looked much more thoroughly for that in this study with serial BNP measurements, and serial echoes, and we saw it at an incidence of 2.8%.

Our next question comes from the line of Michael Schmidt with Guggenheim Securities.

I'm still trying to wrap my head around the OS data, I suppose, and I guess, Tom, any additional hypotheses why the crossover patients did so much better than patients who received Tovecimig plus paclitaxel in the treatment arm. And I'm just curious if there is anything, maybe you can comment on post-progression treatment in the Tovecimig treatment arm, where patients perhaps had less opportunity to receive third-line therapies, then across to patients, for example. And also [indiscernible], if you saw any imbalances in prior PD-1 use [indiscernible] treatment arms?

Okay. multiple different questions there. I'm going to take those in reverse order. More than 90% of patients in the study had a prior PD-1 blocking regimen. So the vast majority of patients had either [ durvalumab or pembrolizumab ]. We don't have all of the post-study therapy data analyzed so far. But the one thing I did look at the end of last week only because it was so topical, I think there was only 1 patient in the study who got a RAF inhibitor. So I don't think post progression therapies are an explanation for that. But again, that's very preliminary. Michael, the first part of your question is, of course, fascinating, which is the first part of your question is why do we see this? So one obvious hypothesis is maybe 2 months, 1.9 months, 2 cycles of paclitaxel. Does that condition the tumor microenvironment to be more acceptive to angiogenesis inhibition with Tovecimig? That's sort of a fairly straightforward hypothesis. I think we're thinking about that. We're going to do some preclinical work on that. That might be an interesting thing to test and maybe other indications as we move forward in the future, now that we have this unequivocal demonstration of efficacy here. As we think about moving into other indications, maybe that's a question we think about in [ gastric ] cancer where paclitaxel is part of the standard of care maybe a couple of cycles of pack and then add Tovecimig into that, obviously, incredibly early days in our thinking about that observation.

Okay. And maybe just a follow-up. I'm just curious if you've had any additional regulatory interactions with the FDA since the study started originally. And I'm just wondering how you feel about [ approvability ] with the 1.5 [indiscernible]. I mean, generally, I believe FDA is looking for an HR that less than 1 in OS for these types of studies. But obviously, in BTC [indiscernible] there's a lot of history of accelerated approval. The drug clearly works on PFS and ORR. And I'm just curious if you've had any other interactions with the agency to look the trial.

So no is the specific answer to your question. But I think, again, it's very, very important, I think, to think about this hazardous show, right? When you see a hazard ratio of 1.05, right, the knee-jerk reaction is that somehow there is toxicity in the experimental arm. But that's not the case here. The reason that the hazard ratio was 1.05 is because the control arm patients who crossed over did so amazingly well. In terms of the very broad general question you asked, Michael, there are a number of examples of approvals, with either ORR or PFS where the hazard ratio on OS is above 1. So I'm not worried about that, especially because it's so clear that the patients who crossed over really drove the hazard ratio. So I think because of that, this analysis of the crossover patients, I think, is a demonstration of efficacy of the drug.

Our next question is from the line of Li Watsek with Cantor Fitzgerald.

Maybe just a follow-up for crossover versus non crossover patients. Just curious, any difference in the subsequent treatments, or other factors could have also contributed to the OS difference that you observe? And what other crossover analysis that you guys maybe would you [ conduct ] to show there is OS benefit to [indiscernible]

Sure. That's a fantastic question. So in terms of the first half of your question, Li, I don't have all of that analysis done yet. But our very early review of that data suggests that it's not post-study therapy. The second part of your question is absolutely fascinating. Because as many of you know, in the CLARITY study, which was [ ivosidenib ] versus placebo the crossover adjustment, RPSFT analysis was done post hoc in that study. So while the RPSFT was prespecified here, the statistical assumptions were not met, but there are multiple other statistical methodologies that can be used to look at adjusting OS curves for crossover, and we're exploring those other methodologies now.

The next question is from the line of Charles Zhu with LifeSci Capital.

I believe these may have been somewhat addressed from different angles. But how do you think the FDA might looking at some of your OS data here in light of hitting ORR as well as PFS, particularly since the FDA seems to place a bit of an emphasis on no detriment on overall survival relative to hazard issue that is at least numerically above 1. And related to that somewhat, maybe just sort of going back to the [indiscernible] what exactly were those certain statistical assumptions that were not met in this particular study? And could you provide some color as to perhaps why it looks like when you ran that analysis, the OS hazard ratio seems to have gone up to 1.13?

Yes. Okay. Very, very important questions. So let me take the second part first. which is a 2-part question. So when the ITT hazard ratio is above 1, the RPSFT cannot work. So it's that simple. We're providing the [indiscernible] numbers here today solely for full disclosure. Those numbers are completely irrelevant because the assumptions for the statistical method were not met. Now the first part of your question, I want to just take head on, right, which is -- and specifically, I'm going to use your word [indiscernible], right? So again, the assumption is with a hazard ratio of 1.05, that that's evidence of [ harm ]. That is fundamentally not the case here. And that is the single most important take-home message of these data. Because the hazard ratio of 1.05, I would strongly argue is [indiscernible] evidence of benefit, because the crossover patients that did so well are the patients who drove the hazard ratio on the ITT analysis above 1. Those patients that live 12.8 months, booked the ITT hazard ratio above 1, because those patients were included in control arm, because of where they were randomized. So the ITT analysis is not evidence of harm. It's evidence that the crossover patients did incredibly well. This is a super, super important point.

Our next question is from the line of Matt Phipps with William Blair.

Can you remind us for patients who did cross over, what including price you had needed to be met for them to be allowed to cross over? And I guess were there patients didn't cross over who wanted to but didn't meet any of that criteria? And then I guess, just confirming you have not yet done something like a 2-stage adjustment versus probability of censoring as other sensitivity analysis around this crossover [ OS ].

Okay. Second part of your question, very straightforward. No, we have [indiscernible] 2-day analysis on the crossover. And there are a couple of other statistical analyses that we're doing on the crossover as well that are in progress. In terms of the first part of your question, I went back to the schema. The main criterion for a cross[indiscernible] was a centrally confirmed progression event so the patient was required to have a [ bicker ] confirmed progression event and then meet some of the selection criteria for the study. Matt, I'm sorry, this very specific question you asked, I don't know the answer to. How many patients wanted to cross over, but didn't for whatever reason. I'm sorry, I don't know the exact answer to that. I can look that up as we continue our evaluation of the data.

Our next question is from the line of Sean McCutchen with Raymond James.

Just one from us. Can you speak to the PFS results in the context of what would be expected for FOLFOX and second-line BTC. Obviously, we see this question a lot. But given the median PFS below 5, I think some context necessary for how you're thinking about the competitive profile relative to standard chemo in this setting?

Sure. Sure. So just put up the PFS slide again for 0.7 months in the combination arm versus 2.6 months in the paclitaxel only arm. Now I think the CHOICE study, which was a randomized trial of FOLFOX versus FOLFIRI is the best benchmark here for what to expect. 2.1-month median PFS for FOLFIRI, 2.8 months median PFS for FOLFOX. That's what people see with these chemotherapy, with each of these 3 drug chemotherapy regimens out there in ABC-06, which was FOLFOX versus best supportive care, there was a very long scan frequency there. And more importantly, in ABC-06, the control arm was not scanned. So we don't know whether or not FOLFOX in ABC-06 was better than supportive care. Also very importantly, our study is blinded, independent central review. So all of our PFS data are [indiscernible] assessed progression, ABC-06 all investigator assessed. So I don't think there's any KOL who believes that the PFS with FOLFOX is greater than 3 months.

The next question from...

Sorry, that's why we use 3.0 months as the powering assumption for median PFS in the control arm in our power calculations. Sorry, John, go ahead.

Just kind of going back to a couple of questions. I think as a follow-up on Biren's original question. Do you have any sense as to the discontinuation rate for the combination arm versus the patients who crossed over from the control? I'm just curious if you've been able to look at that to tell whether or not, perhaps for whatever reason, that discontinuation rate was different, either one way or the other?

We have not looked at that. I'm sorry, something that we'll be looking at in the coming weeks.

Next question is from the line of Ren Benjamin Citizens.

What was the -- can you just remind us what was the study power to detect in terms of OS delta between the 2 arms? And if you look at the OS of Tovecimig-treated patients, crossover and the original treated patients versus those not what do those 2 numbers look like? And just -- have you done any sensitivity analysis regarding prior therapies and how that might have impacted the FSR OS?

Okay. So a multipart question there. Let's start with the power calculations. The original power calculations were for OS were 6.2 months in the control arm, and that was based on multiple different FOLFOX studies, both [indiscernible] and ABC-06 had a median OS of 6.2 months in the [indiscernible] arm. The assumption on left in the combination arm was 10.9 months. The pooled median there was right around 9 months based on a 2:1 randomization. So those were the power calculations. Okay. The second part of your question is partly here. The median OS in the paclitaxel only arm was 6.1 months, very consistent with what you see with FOLFOX and the median OS in the original combination on patients was 8.9 [indiscernible] So that's probably the correct [indiscernible], 8.9 versus 6.1. Ballpark, 40-ish percent improvement.

Got it. And then any other just subset analyses that you're -- that you looked at regarding maybe prior therapies or anything else we should be looking at that might further support the OS benefit?

So no, we have not done that specific analysis, Ren. Just as I mentioned earlier, so many patients greater than 90% had [indiscernible] plus a checkpoint inhibitor. Perhaps we could check [indiscernible] versus pembrolizumab, but we have not done that.

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

My follow-up was asked and answered.

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

This is [ Sara ] on for Joe. I just wanted to kind of hone a little deeper on the PFS numbers as you discussed potentially going or seeking accelerated approval. Looking at PFS, are you thinking the numbers you got -- that you presented today represent an approvable, now metric? And just kind of curious how you're weighing that against your data for OS and even the ORR as well, maybe as you look to move ahead with your FDA meeting?

Sure. I think the most -- in addition, obviously, to the extraordinarily low p-value. The hazard ratio here of 0.44 is without any question an approvable number. That is a 56% reduction in the risk of progression.

Our final question is from the line of Albert Lowe with Craig Hallum.

The portion of patients that cross over differ from your expectations in the design of the trial? And did this aspect affect the feasibility of the RPSFT analysis?

So that's an interesting question. We didn't -- it was very, very hard to estimate ahead of time the fraction of patients [ who ] would cross over. I think 54% is maybe a little higher than we had initially thought of. And in addition, I think there is something that was also affected the RPSFT a little bit was the timing of those crossovers. It's a very good question. It's a very, very complicated question also. But I think it's maybe a little higher than we had originally thought when we were designing the study because, again, crossover patients would be treated in the third line setting, where the number of patients that make it out of a second-line therapy into third-line setting is that's not a huge number. So it's a little higher, I think. But good question, tough to answer.

At this time, this will conclude our question-and-answer session. I'd like to turn the floor back over to Dr. Thomas Schuetz for closing comments.

Great. Thanks so much. Just wanted to thank everyone for joining the call today. I really appreciate your time this morning. I think I'll probably just close by sort of going back to the next steps slide. We're super excited about the data we have. We have an unequivocal demonstration of efficacy, multiple different ways here. So we're going to put everything together, meet with the FDA to discuss. These data in advance of a planned BLA submission. If that discussion goes well, we should be in a position to initiate the BLA submission towards the end of this calendar year. Look forward then to BLA review and potential U.S. approval of Tovecimig in 2027. Thanks again.

Thank you. This will conclude today's teleconference. Ladies and gentlemen, thank you for your participation. You may now disconnect your lines at this time, and have a wonderful day.