ALX Oncology Holdings Inc. (ALXO) Earnings Call Transcript & Summary

Thank you for standing by. My name is Luella, and I will be your conference operator today. At this time, I would like to welcome everyone to ALX Oncology, ASPEN-06, Phase II Clinical Trial Data Results Conference Call. [Operator Instructions] I would now like to turn the conference over to Jason Lettmann, CEO of ALX Oncology. You may begin.

Hi. Thank you all for joining. I'm Jason Lettmann, CEO of ALX Oncology, joined here with Dr. Sophia Randolph as well, who is our CMO. We are very excited today to be presenting data from our randomized ASPEN-06 trial and HER2-Positive Gastric Cancer. And as many of you know and have been tracking closely, ALX is leading the field in CD47. We're focused on developing a Evorpacept or Evo, which is a CD47 blocker, that has the potential to both be a first-in-class and best-in-class IO agent. This data today is particularly exciting for us. I know for many of you, exciting for the clinical community; and most importantly, for cancer patients as this represents a couple big firsts. It's the first randomized data to read out in a prospective randomized clinical study in solid tumors in the CD47 space. And as we're going to walk through here shortly, ASPEN-06 is the first randomized data ever to demonstrate both the tolerable safety profile and the first randomized trial to report a clear and durable improvement over a standard of care. So what's really exciting, of course, about today's new news here is not only the compelling data in Gastric and GEJ cancers, but the fact that, for us, it just truly validates a truly unique approach to blocking CD47 and validates Evo's broad potential when combined with any Fc active antibody across multiple tumor types in addition to Gastric. So the plan today is for me to first walk you through the program and a reminder of our mechanism. I'll then hand it over to Sophia to walk through specifically the data that we're seeing with the ASPEN-06 top line results, and then wrap it up with where we're going from here. So, first on Slide 3 is a quick reminder of the biology and the mechanism of a Evorpacept. And what you can see here on the left is how CD47, or the "Don't Eat Me" signal, works. It is one of the primary markers of self, and it's upregulated by tumors to evade the immune system. And what you can see here is how CD47 signals the Don't Eat Me, signals Don't Eat Me to both cancer as well as healthy. And this is what's been really challenging for the field to solve. And on the right is our molecule, again, Evorpacept, which has a very high affinity CD47 binding domain, as well as an inactive or dead Fc, which is a truly unique attribute of a Evorpacept. So, on Slide 4, this next slide shows how Evorpacept works and how we selectively target cancer while sparing the on-target toxicity that has led to other failures in the space. On the left, Evo drives a potent blockade of CD47 signaling. And because there is no positive, there's no Eat Me signal, if you will, the macrophage will not attack either cancer or healthy cells. But in this middle panel here, you can see that when combined with an anticancer antibody, which is seen in green here, this together will cause a macrophage to attack cancer. And this attack is targeted specifically to cancer, as the targeted antibody partner provides that specificity. And that's what then drives what you see on the right, which is a very targeted, selected killing of tumors. And importantly, in a particularly fundamental concept to recall throughout today, is that a macrophage will eat both. It needs to have the Don't Eat Me signal block as well as a positive Eat Me signal. And that together is what makes up our mechanism. So when you turn to Slide 5, you can see how this is different than conventional CD47 targeting. It is, as you can see on the left, the conventional CD47 blocker tries to provide both of these signals on the same molecule and effectively tries to use CD47 as a tumor targeting antigen. But unfortunately, because CD47 is not specific to tumor cells, this also results in ADCP or phagocytosis activity against normal cells, like red blood cells or platelets, resulting in CD47-targeted destruction of both normal and cancer cells, and then results in associated toxicity. Additionally, and as we've observed now in the clinic, these toxicities can prevent the exposures needed for full CD47 blockade and limit the effective dose. And because conventional CD47 blockers may compete for the Fc receptor, it presents a significant challenge. And the ultimate result is what you see on the right, which is broad and indiscriminate targeting of both cancer and healthy. So that brings us to where we are today on Slide 6, and what we're really excited about sharing with you all. Again, ASPEN-06 represents a few key firsts for the field. It's the first prospective randomized study in solid tumors in the space, and the first randomized study to read out both a strong ORR gain as well as a durable improvement versus standard-of-care. In terms of the full population, and as Sophia will dive into, we are seeing a 40.3% ORR on treatment, compared to 26.6% on the control arm, which is nearly a 14% delta versus control. Importantly, and really exciting for us, is what we're seeing in terms of durability, with a median duration of response of 15.7 months, which is more than double the 7.6 months that we've seen on the TRP control arm. What we're also going to dive in today, and what we're really excited about is what we're seeing as it relates to HER2 expression. Again, very important for us, mechanistically, to have both. And in patients with fresh HER2-Positive biopsies, which is indicative of patients with the strongest HER2 expression, we're seeing a 54.8% ORR, compared to 23.1% for the control arm, which is over a 30% delta. Last, and certainly not least, given some of the challenges in the field, we continue to see a safety profile which is consistent with the over 500 patients that we treated to-date, as Evo was well tolerated versus control, and of course, that's an important component when we think about taking this drug forward. So turning to the next slide on Slide 7, this is just a reminder of our registration strategy and how we got here. We first tested a Evorpacept in Gastric Cancer in our ASPEN-01 study. This provided a very strong signal, and in this same combination with the tras/ram/pac backbone, provided proof of principle for us and a strong signal of activity. Back then formed the basis for what you see in the middle, which is what we're going to be reporting on today, which is our ASPEN-06 study. This is a key proof-of-concept study for us. Again, testing the exact same combination, but this time versus the control of TRP, to really isolate what is the contribution of Evorpacept. And then last is, this is again a Phase II/III design where we have agreement with FDA as to what it will take to drive registration. So on the right, you can see the Phase III design, which will then compare Evo-TRP versus the global standard of care of ram/pac or RP. So with that turning to Slide 8, this just touches quickly on the unmet need in Gastric Cancer. It is truly a global disease, a challenging disease to treat with a global unmet need. And it's really an area where we need more novel and better tolerable treatment options. While the incidence is highest in Asia, it has a significant presence in Europe, the Americas and other global regions. And it is really characterized by a relatively poor 5-year overall survival, which can be as low as 7%. On the right, you can see the standard-of-care by line of therapy. You have the Tras+Chemo doublet in the front line, as well as Pembro now, which is approved in front line patients. Second line is either ram/pac or in HER2. And again, we are targeting the second and third line here as part of this study. So turning to the next slide on Slide 9. This is just a reminder as to what are the benchmarks in the field. The first is the Rainbow study. Rainbow really established ram/pac as a global standard of care in the second line, and that study had showed about a 28% ORR versus a 16% in terms of paclitaxel alone. Then following that, the DESTINY-Gastric-01 study in the third line reported a 40.5% ORR within HER2 versus physician's choice chemo. Both of these studies demonstrated a survival benefit of 1 year or less and really is highlighted where we see the unmet need going forward. So with that, I'll turn it over to Sophia here to walk through the mechanism briefly that we're testing within gastric and then share our ASPEN-06 data.

Great. Thanks, Jason, and for all. I'm very happy to take you through our final analysis of the ASPEN-06 randomized Phase II study. And first up on Slide 10, you can see as Jason mentioned earlier, Evorpacept binds to CD47 on the cancer cell and this allows it to inhibit the SIRP alpha CD47 myeloid checkpoint. And so, in this mechanistic figure, you can see that in combination with an anticancer monoclonal antibody here, Trastuzumab, which binds to the Fc gamma receptor on the macrophage, you can see that the 2 drugs are able to fully activate the macrophage, directing the ADCP activity against the HER2 expressing gastric cancer cell. So on Slide #11, here we can see the randomized Phase II design of the ASPEN-06 study of Evorpacept in combination with TRP or tras/ram/pac. In this study, patients were enrolled with second or third line, HER2-positive Gastric Cancer that had progressed upon an anti-HER2 agent and they were randomized to receive either treatment arm, the Evo-TRP or the TRP. Eligible patients must have had no prior treatment with an anti-CD47 or SIRP alpha agent nor Ramucirumab. All patients, as I mentioned, must have been treated with prior anti-HER2 therapy. And as such, prior treatment within HER2 was allowed as was prior checkpoint inhibitor therapy as well. So 127 patients were randomized in a one-to-one fashion to receive either of the 2 study regimens. The primary study endpoints where investigator assessed overall response rate with key secondary endpoints of durability of response survival, progression-free survival as well as safety. So we've previously presented data from a prespecified interim analysis of 54 randomized patients at the end of last year and today we're now presenting the top line data from the final analysis of the 127 randomized patients. So this study had 2 primary objectives: the first was to look for a 50% improvement over an assumed historical ram/pac objective response rate of 30% in this HER2-positive population; the second was to establish the contribution of Evorpacept to the TRP backbone. Here we were looking for a clinically meaningful magnitude difference between the 2 arms of greater than 10%. So these 2 objectives were evaluated in the full intent-to-treat randomized population. These were patients who were identified as HER2-positive based upon a tissue biopsy that could be obtained at any time during the course of their previous disease, as well as in a subset of patients whose HER2-positive gastric cancer was recently identified based upon a fresh biopsy obtained after prior anti-HER2 therapy. And the second population was of particular interest to us, because HER2 expression we know there is greatly in gastric cancer. And we wanted to prospectively evaluate a population of patients who are more likely to be enriched for HER2-positive disease at the time of the study start. On Slide 12 on the next slide, the study was conducted in 13 countries and patients were randomized across 90 sites. The dose of Evorpacept in this study was 30 milligrams per kilogram administered once every 2 weeks. Trastuzumab was administered with a 6 mg per kg loading dose followed by a 4 mg per kg dose once every 2 weeks. Ramucirumab and Paclitaxel were dosed according to their label in a 28-day cycle. All patients enrolled had received a prior anti-HER2 therapy and there were several stratification factors that were used to balance the patient characteristics across the 2 treatment arms. So these included cancer type, fresh versus archival biopsies, Asia versus ex-Asia region, the line of treatment, HER2 IHC score and then use of prior in HER2. On Slide 13, the demographics of the patients randomized are shown here, so the majority of patients were Male and Asian or White. Ethnicity was missing in a proportion of patients as some countries do not collect this particular demographic. The patient characteristics were generally well balanced across treatment arms. There were, however, some features which did differ between the population that contributed to the interim analysis and those who were enrolled in the latter half of the study after the interim analysis. So post interim analysis, there were fewer patients enrolled using a fresh biopsy. So 46% of patients were randomized based off of a fresh biopsy in the interim population versus 32% in the more recently enrolled post interim patients. Also, patients on the Evo-TRP arm who enrolled into the study after the interim analysis did have characteristics suggestive of more aggressive disease such as higher ECOG score, faster time to their first progression, and overall just a shorter prior disease course. And then lastly, it's noted there that patients who did enroll based upon fresh biopsies had those biopsies on average within 1 month of coming on study and that's within the screening window, whereas those using archival samples came from biopsies obtained really over 1 year before enrolling on this study. So Evo's safety profile is shown on the next slide, 14, and it was consistent with the prior experience of Evorpacept in over 500 patients that we've treated to-date, so Evo plus TRP was generally well tolerated. The incidence of adverse events by grade due to any cause was comparable by arm and there were no on study treatment related deaths on either arm. On Slide 15 you can see a summary of all causality Grade 3, 4, 5 adverse events by arm. These events were again generally well balanced across both treatment arms. There was some increase in neutropenia on the Evo-TRP arm. However, as we'll show you in the subsequent slide, this increase may be explained by the fact that as many patients stay on the Evo containing treatment arm for longer periods of time, they have more opportunity to experience cytopenias, certainly in a regimen that contains chemotherapy as part of its backbone. There were 3 unstudy deaths due to sepsis across both study arms. However, these were not considered related to either treatment regimen. So on Slide 16, here we see in the full intent-to-treat population, the addition of Evorpacept to TRP demonstrated an ORR of 40.3%, which compared favorably to the assumed historic ram/pac's overall response rate of 30% with a p Value of 0.095. Evo contributed a clinically meaningful benefit, this is our second objective of more than the prespecified 10% delta in ORR over the TRP arm. When an additional analysis was performed comparing Evo-TRP to the observed in-trial TRP overall response rate and that within the study was 26.6%, there was a p Value observed of 0.027. Responses were durable with a median duration of response in the Evorpacept combination arm of 15.7 months compared to the TRP arm control of 7.6 months. The activity of Evo-TRP compared favorably as well, the published data of available therapies in the second and third line gastric populations. So in Slide 17, as shown in the waterfall plots, you can see there's substantial tumor shrinkage seen in patients who received Evo-TRP compared to TRP. And on Slide 18, when we look at the overlay of those waterfalls, it becomes really important, but you can see the relative greater depth of tumor shrinkage seen in the Evo-TRP arm, again supporting the contribution of Evorpacept to that TRP backbone and the subsequent clinical benefit of the regimen seen in this trial. When combining with Trastuzumab on Slide 19, Evorpacept's mechanism of action depends upon the expression of the HER2 receptor in order to drive that maximum phagocytosis activity against the gastric tumor cells. And in fact, HER2 expression is an important biomarker for Evo-TRP response. When we look at the activity on Slide 20 of Evorpacept plus TRP in the prespecified population of patients who were enrolled into the study based upon fresh HER2-positive biopsies, we see that Evorpacept more than doubled the tumor response in patients when compared to the TRP control. So this improved benefit in this prespecified population of patients who are enriched for recent HER2 overexpression underscores the importance of this biomarker in Evo-TRP response and it continues to validate Evorpacept's mechanism of action. So as is shown, you can see patients with recent HER2 positivity demonstrated an ORR of 54.8% compared to 23.1% for the TRP control and that had an observed p Value of 0.0038. Now, as shown on Slide 21, this was no surprise, as in gastric cancer, HER2 expression is highly variable. So shown in this Slide from a recent paper of Shitara et al, based off the DESTINY-01 experience HER2 expression in gastric cancer can vary over time due to downregulation of the receptor after treatment with an anti-HER2 therapy, and can even vary from tissue sample to sample within the original tumor and its metastases. This does seem to be more specific to gastric cancer than to other HER2 overexpressing tumors such as breast cancer, where there's more homogeneous expression of HER2 throughout the tumor. And this has been well documented by others. So in gastric cancer, having a fresh HER2-positive biopsy can allow for the identification of a population enriched for HER2 expression, and this is certainly consistent with the data that we've seen in this study. So shown a different way on Slide 22. And again, consistent with Evo's mechanism, patients had a greater response if they were identified using a fresh biopsy. So in contrast, you can see that patients receiving TRP and who are being now retreated with Trastuzumab did not see much benefit beyond the backbone of ram/pac, regardless of HER2 expression or whether a fresh biopsy was used to characterize their current HER2-positive status. So again, this would be consistent with earlier reports, notably from the [ TAG ] study, that HER2-positive patients who have recently progressed upon Trastuzumab are likely resistant to Trastuzumab and do not benefit from Trastuzumab retreatment in the context of systemic therapy. On Slide 23, the mechanism of how Evo works with Trastuzumab in combination is fundamentally different from how Trastuzumab works by itself. So without blockade of the [ CD40 ] SIRP alpha checkpoint, minimal phagocytosis of cancer cells using macrophages activated by Trastuzumab will occur. And again, this is consistent with the ASPEN-06 TRP arm data. However, when combined with Evorpacept's CD47 blockade and Fc Active antibodies such as Trastuzumab can drive phagocytosis of the cancer cell, translating into real clinical antitumor response. And I think that's seen on our E-TRP arms within this study. So there continues to be much to learn from this clinical trial, and we thank the patients, the investigators and the clinic staff for their contributions and support of this study. And now I will turn it back to Jason for final remarks. Jason?

Great. Thanks so much, Sophia. I think as Sophia mentioned, what we've demonstrated here is really how Evorpacept can harness and engage the innate response, and it's really a first-in-class molecule here that, as Sophia said, we continue to learn about, but continue to also be incredibly excited about. And I think what we feel we know now is that we have an active agent. We have a very strong signal, a response rate that is robust and that is clearly durable, and one that is also clearly different than a HER2-targeted agent like Herceptin. On the durability front, to see a DOR of more than double and knowing that there's typically a tight correlation between DOR and PFS, we're encouraged by what we're seeing there as well. To the point that Sophia just made. What we're seeing is in line with our mechanism. As we've talked about here, have talked about at length over the last decade or so of this company, is just the importance of both the blockade of CD40 -- CD47 and the positive signal. And there's been questions as to what Tras would do in this setting. And we feel what we're seeing is that Tras by itself is doing very little and that Evorpacept is bringing something new to the table that's fundamentally different and driving benefit. And in the population where we have patients that had the most recent biopsy, which again, should be indicative of HER2 expression, to see a 54.8% delta versus 21.1% on the controller over a 30% gain truly helps validate the mechanism for us. From a safety perspective. Again, as Sophia walked through, we continue to see safety in line with our prior experience, continues to be a very tolerable agent that is well behaved in the clinic. And then last is just to highlight again the novelty of what we're doing. This is truly a novel IO agent that has the potential to be first and best-in-class. And to see this data come through in a randomized study for the first time is very compelling. So, on the next slide, again, this just pieces together the Why. On Slide 25, it speaks to how Evorpacept was designed. We have a very potent CD47 blockade molecule, which is resulting in less sink, allowing us to drive much higher dose. And again, this is why we think we are seeing what is truly the first CD47 to C-cell tumor efficacy, and now to do so in a randomized setting. All of these things are what's working together to create this effect, and I think really helps validate where we're going from here. So on the next slide, it touches on that. Again, there's 2 mechanisms at play, 2 mechanisms that we are now testing in the clinic. The first on the top left is what we focused on today, which is how does Evorpacept help harness the innate immune system. And we're also testing how CD47 signals to the adaptive immune system. And that's the second mechanism here, where we're going to have over 300 patients randomized reading out also Q4 or Q1, which will be another big event for the company. But with this data, with ASPEN-06, this now adds yet another positive clinical readout. We now have 6 positive clinical studies now with randomized data in support to single arm studies that we delivered in the past. And for us, this really does derisk and drive our development going forward as we think about combining a Evorpacept with other Fc Active antibodies across multiple tumor types. So certainly excited about next steps with Gastric, but also excited about what this means for other malignancies where we can combine with an antibody across both [ heme ] and solid. So again, excited about the data. Thank you all for the time and frankly, support of us and of this company and this molecule. As Sophia mentioned, a big thank you to our clinicians and an even bigger thank you to the patients who participated here. So with that, we'll open it up to questions.

[Operator Instructions] Your first question comes from the line of Michael Yee with Jefferies.

I think there are 2 areas I would like to ask important questions on. It seems that perhaps the first half of the study did better, the interim. If you go back to the data and everyone can do math around the second half of the study in terms of the incremental patients, and you can see that the response rates are more evenly balanced in the second half, i.e., the incremental patients. They're very similar, 32% and 30%. So without just looking at response rates, can you tell us, is the duration of response in the second half of the study much better for the drug arm versus the control arm, appreciating overall it was doubled. But in the second half of the study, the duration of response also better, much better, the drug arm versus the control arm. And then on PFS, which would also be important, can you please clarify why it's not mature? Because we would -- math would imply that -- certainly a lot of events have happened and should be mature. So appreciating that it could be all driven by the first half of the study. Tell us, is the second half of the study PFS also very different and there's a big separation?

Yes. Great. Thanks, Mike. Appreciate the question, and apologies to everybody on the technical difficulties here, but we'll take those in reverse order. Both good questions. I think fundamentally, we do see a difference in the population between interim and the post-interim population. We can talk about that. I think first, just to address PFS, obviously, that's important and a metric we're very focused on. I think, as we touched on -- what we're excited about, I think is a couple things. One is that PFS is tightly correlated to DOR. And we're seeing that play out in our earlier patients. And so when you look at our DOR versus control, and it's more than double and it's better than the benchmarks, we think it's really compelling. We have a DOR here of over 15 months, as a reminder. And as you know, Mike, the Rainbow study showed 4.4 months on ram/pac versus 2.8 months. The DESTINY study within HER2 showed about 11 months versus 4 months on the Chemo arm. So again, 15 months is significant here. And when you look at what we reported on the interim to now, it has remained that way. So we think that's really important. And then the second point I would make is just around enrollment. And again, this won't be lost on you, but we did see enrollment accelerate. We overenrolled this study to 127 patients. So it tells you something about the bolus of patients that came in relatively late here. So we just have a number of patients that are still early here and haven't progressed. But again, I think DOR should correlate tightly, and that's what we're really encouraged by. Sophia, do you have anything to add to that on the clinical front?

No. It's exactly what you said, actually, with the sort of exponential accrual on the back end of the study, the follow-up on that population is actually fairly short. It's almost the same as just having had 2 rounds of imaging. So the first half of the study is definitely more mature. But again, we'll continue to follow this out, but the follow-up on the second half is much less.

Okay, so let me clarify those [indiscernible]

Go ahead. I'm sorry.

So the duration was to wrap up on duration of response. The duration of response is very clearly different in the first half, but in the second half, those responders 12 for the drug, 11 for the control, immature, so the [ swimmer plots ] are still going on for those in the second half. That's what you're saying on duration response?

Yes, absolutely. And I mean, our hope and our expectation is that it will follow the first half, but it's just too early to tell from right now.

And then, therefore, how does that relate to PFS? I guess was the second question.

So I think.

I think, it should be tightly correlated, but go ahead, Sophia.

Yes, yes. No, I was going to say the same thing. So it's just the follow-up is short. In a way, it's good. It means that patients are benefiting from the combination, but more details on that will be presented. But for right now, the follow-up time for those time-to-event type metrics, it's just very short in the second half of the study.

So then, just to your first question, Mike, I think you talked about just the interim versus post-interim population, and I think we highlighted 2 really important factors there. One, just the rate of fresh biopsy, which was roughly half of the population at interim, and that number dropped significantly in the second half. And then as Sophia touched on the second really important point, which isn't a huge surprise when you put up data like we did at the interim, is just what happened in terms of overall disease status, if you will. And when you look at the Evo arm in particular, post-interim, it's pretty clear that we have just a more aggressive patient population, and we do think that's a significant factor. So those 2 things working together are what's different. But again, overall, we've seen a 14% delta here, which was more than our bar, certainly of 10%. And then we look at this fresh population with an over 30% delta. That is just a very, very strong signal. So I think that's what's driving a lot of our enthusiasm here.

Your next question comes from the line of Chris Raymond with Piper Sandler.

Yes, 2 from me, and maybe the first one to follow up on the subject matter from the first question, just on the characteristics of patients in the second part -- second half of the study. So can you maybe help quantify the variability, I guess, in the characteristics between the interim and the post-interim patients? And I understand the dynamic there of enrollment accelerating and getting sicker patients. But any sort of color you can give in terms of the difference between control arm and an active drug arm that sort of drove that ORR, it'll be pretty much the same in that second half. Anything that you can do to sort of characterize that? And how -- maybe, I guess a question on that is how did that escape your sort of balancing for enrollment? And then just also on the stats, just -- can you just confirm -- I'm just looking at Slide 16, the p Value was 0.027. Can you just confirm the bar for [ stat.sig. ] was 0.05, correct?

Yes, sure. So, thanks, Chris. Appreciate it. Both good questions, I think, on the differences we highlighted, again, just the importance of fresh here. And I do think that that [ ITT, ] if you will, will have a important factor on what we're seeing post interim. And I think, as Sophia mentioned, on disease severity, it's always tricky to quantify that with one metric. But when we see ECOG and time to progression on a prior line, et cetera, all going the same direction, which is really the direction of suggesting it's a very severe population, I think that tells us something. And to your last question there, again, this was a well-randomized global study. We had 6 stratification factors across line, Asia and HER2 use, et cetera. So very well randomized. However, we didn't randomize or stratify across these 2 populations. So I think, again, it was randomized. It was well randomized and well conducted. But as can happen, patient characteristics can change over time. So I think that was your first question. And on your second question, can you just remind me of what that one was again?

Yes. Yes, sure. So I just want to clarify the stats. Just the E-TRP. The p Value versus TRP was 0.027. Can you just clarify? Was 0.05 the bar for stat.sig., just clarify that.

Yes. So we had 2 objectives here. One was to show benefit versus historical control, and that was ram/pac. Right. And that's where we used the 30% number. And we did not hit statistical significance on that. Again, we'd argue that is not as important as the second objective, which was to compare versus our control. And that is the 0.027 number that you highlighted. And again, our bar was to show a 10% delta there, which we achieved and getting close to 14%. So happy about that. And again, we think that's the most important way to look at a randomized study.

But what was the stat.sig. bar? Was it some number other than 0.05? That's my question.

Well, so 0.025 is what we use for the bar, right. Yes.

Your next question comes from the line of Li Watsek from Cantor.

Also a couple here. Maybe a follow-up to the question about the patients enrolled in the second half of the study seems to be a little bit secure. So I wanted to understand was that driven by perhaps the interim data that you've put out in physicians may be more comfortable putting sicker patients in the trial? Or was there any other factors that may be contributing to this? And also, could that potentially introduce any bias into the study?

That was a prespecified futility analysis and to see the data that we did, we certainly thought it was worthy of sharing with the Street and communicating externally. I think with that decision, of course, you're actively enrolling a study and have months to go to complete enrollment. That is going to drive physician interest. We certainly saw it in our enrollment and now we're seeing it in our data. And so, I think that is a valid question. Again, I would just say we're early at looking at these things. This data is relatively fresh for us as well, but we do think that was a factor. Sophia, anything you want to add on them -- on that front?

Yes. Yes, I mean, I would just add that. I would say it's unclear. I think that for sure, when we think about the stratification factors that we identified upfront, those are balanced, whether it's from the beginning of the study or the end of the study, those are balanced across the treatment arms in the full subset -- excuse me, in the full population. When it comes to these other things. So things like time to progression, time to overall disease course, these are not things that you can obviously stratify. Whether it's because of the interim being out there, I don't think we can say that conclusively. I mean, sometimes it's just as you get to the end of the study and there's a rush for folks to get patients on study somehow. Sometimes the type of patient changes or what's out there in terms of other trials that are available to accrue. So there's a lot of different factors that go into it, not necessarily seeing the data at the interim. There are other -- obviously, it's a complex data set where just plowing through it how these things impact response versus how they impact time to event endpoints is also not completely known. So I think it's a little bit early, but they're definitely ideas that we're all contemplating. But, unclear if that is truly the driving factor or not.

Okay. Then I have another question on the HER2 expression. So is there a plan to maybe go to the FDA and then try to implement fresh HER2 biopsy into the Phase III study? And can you also comment on the feasibility of using that?