Good afternoon and welcome to the I-Mab KOL webinar. At this time, all attendees are in a listen-only mode. A question and answer session will follow the formal presentations. As a reminder, this call is being recorded and a replay will be made available on the I-Mab website following the conclusion of the event. Before we begin, I would like to remind everyone that statements made during this conference call will include forward-looking statements under the safe harbor provisions of the Private Securities Litigation Reform Act of 1995, which involves risks and uncertainties that can cause actual results to differ materially. Forward-looking statements speak only as of the date they are made, as the underlying facts and circumstances may change. Except as required by law, I-Mab undertakes no obligation to update these forward-looking statements to reflect future information, events, or circumstances.
This presentation includes statistical and other industry and market data that was obtained from industry publications and research surveys and studies conducted by third parties, as well as our own estimates of potential market opportunities. All of the market data used in this presentation involves a number of assumptions and limitations, and you are cautioned not to give undue weight to such data. Industry publications and third-party research surveys and studies generally indicate that their information has been obtained from sources believed to be reliable. Although we do not guarantee the accuracy or completeness of such information, our estimates of the potential market opportunities for our product candidates include several key assumptions based on our industry knowledge, industry publications, third-party research, and other surveys, which may be based on small sample size and may fail to accurately reflect market opportunities.
While we believe that our internal assumptions are reasonable, no independent source has verified such assumptions. For more information on forward-looking statements, please review the disclaimers in today's press release and the risk factors in the company's SEC fillings. With that, I'll now turn the call over to Dr. Sean Fu, I-Mab's Chief Executive Officer. Please go ahead, Sean.
Thank you very much. Welcome, everyone, to the company webcast. My name is Sean, CEO of the company. Today, I'm joined by Dr. Phillip Dennis, CMO of I-Mab, and Dr. Sam Kleppner, Associate Professor of Massachusetts General, one of the PIs for Givastomig . Next slide. We have arranged today's agenda around the latest data release from ESMO GI for Givastomig , our 18.2/4-1BB bispecific, and with a balanced summary of overview from monotherapy and forward-looking next-step clinical plans. Hopefully, you'll find it helpful. Next. Givastomig , this 18.2/4-1BB bispecific, is an asset that we are actively developing as a bolt-on to the current standard of care for the frontline gastric cancer. The current standard of care is IO chemo. In our study, we're adding Giva without changing the current standard of care. Giva, as an asset, is clearly active in monotherapy studies.
Now we're presenting you with details of the data when we add it to the IO chemo standard of care. Gastric cancer, we don't talk about it as often as, say, lung cancer, but it is still a significant unmet medical need. When you look at the incidence rate, when you look at the death caused by gastric cancers, it ranked at number five in both categories worldwide. With that unmet medical need, the commercial opportunities or the addressable market is estimated to be $12 billion by 2030. We look at this data from ESMO GI with 17 patients over three doses. We see a few clear differentiating factors for Giva Stomach. The first one I will point out is that we found that the Giva can be safely added to the standard of care.
When added, Giva showed that the combination with Giva showed markedly improved efficacy for patients in frontline metastatic gastric cancer. The second thing is also remarkable, which is in our clinical study, we have the broadest claudin 18.2 coverage. We require patients to have 1% of the tumor cells expressing 1+ staining intensity, which translates to one of the largest, if not the largest, patient coverage in the ongoing active clinical studies for 18.2 assets. This is a very exciting profile, and we believe this is a differentiated asset. As far as we know, Giva is the front runner in the U.S., EU, and Japan market for frontline gastric cancers. Finally, I would say that I-Mab is in a strong position when it comes to finance.
We have $168 million on our balance sheet, which means we are in a position to see through some of the important clinical readouts in the next period of time. With that, let's turn it to Phillip and Sam to give you details of our phase I-B combination study, Giva plus IO and the chemo. Phillip.
Thanks, Sean. Again, it's a pleasure to have the opportunity to present to you and to have our colleague Sam Kleppner really present the data in a very eloquent way. I'm going to provide some background. In terms of gastric cancer itself, Sean hit on the unmet medical need. It's one of, I'd say, a handful of solid tumors that unfortunately are diagnosed late at an advanced stage. When they're treated, even then, the five-year survival rate is dismal, less than 10% of patients who survive five years. Nonetheless, there's a lot of opportunity for clinical improvement and a lot of opportunity for market opportunity as well, with a $12 billion market and for sales expected by 2030. The right side of the slide shows the current standards of care, the newly developed standards of care for frontline gastric cancer that address two targeted populations.
The take-home from each of these targeted populations is that there's a lot of room for improvement. CheckMate 649 was a pivotal, large, well-performed phase III study comparing Nivolumab plus chemotherapy, either FOLFOX or another regimen versus chemotherapy alone. In that study, the ORR was improved from 37% to 47% in the evaluable population in that study. This led to an improvement in progression-free survival. Note that at the bottom, those two gray bars for CheckMate 649, the improvement is less than two months. Similarly, Spotlight was another pivotal phase III study. This led to the approval of the first 18.2 asset in the class to be approved in frontline gastric cancer, zolbituximab. In the Spotlight study, zolbituximab plus FOLFOX was compared to FOLFOX alone. In this study, you can see that the objective response rate was 40% for each arm.
Nonetheless, there was an improvement in both progression-free and overall survival. Again, the improvement in progression-free survival was less than two months. There is market room for improvement in terms of objective responses and prolonged disease control. Next slide, please. This slide shows that for zolbituximab, the first 18.2 asset in the class, its activity is limited and its indication is limited to patients whose tumors express 75% or more of tumor cells expressing a very high level of protein expression of 18.2, namely 2+ or greater intensity. Astellas, the developer of zolbituximab, has done a very nice job in publishing the data set that we're looking at here, which is a distribution of over 4,000 tumors for claudin 18.2 expression. If you look at the data in the light blue to the right, that is the area.
Those are the patients who would fall under the indication for zolbituximab in combination with chemotherapy. What's clear is that if you look at that, that's a very small percent of the population. If you think about our eligibility criteria that we use in all of our clinical development programs for Giva Stomach, our cutoff is a minimal threshold of 1% of cells, 1 plus or greater. We do this to provide, have the opportunity to provide Giva to as many patients as possible. If you quantitatively compare the area in the light blue to the 1%- 74% shown again from basically on the left up to that light blue area, we basically double the patients that would potentially benefit from Givastomig compared to those that would be eligible for zolbituximab. We think this is a very important differentiating feature of Givastomig. Next slide, please.
I'm going to talk a little bit about the background on Giva that led us to the phase I-B combination study presented by Sam just six days ago. Next slide, please. Giva is, I think, a potential best-in-class 18.2 therapeutic for gastric cancer. The presentation that was done six days ago was the first to present an 18.2 asset in an early-stage study in combination with immunochemotherapy in frontline gastric cancer. There are some important features about Givastomig that I think underlie its unique profile. I'll get into some of the detail in the next slide. What's important is that in our monotherapy studies in advanced gastric cancer patients that had received a median prior number of therapies of three, we had clinical activity as monotherapy across a range of claudin 18.2 expression, including a patient whose expression was as low as 11%.
We think this is due to very high affinity binding of the IgG1 portion that binds to 18.2. That's, again, I'll discuss that in the next slide. The important thing about this conditional activation, that it's only in the presence of the tumor microenvironment where claudin 18.2 is expressed, is the SCFE fragments that interact with 4-1BB. Are they able to activate T cells? What we hypothesize, and I think what's been borne out in our monotherapy study as well as the combination study, is that we have a minimal of liver toxicity associated with many 4-1BB assets. As well, we have very limited systemic immune toxicities. Next slide, please. The right side of this slide shows the features of Giva that, again, I think are highly differentiated. In terms of the binding of the IgG1 moiety of Givastomig to 18.2, the affinity, again, is very strong.
In fact, the KD is a log lower for Givastomig than it is for zolbituximab. We have a very avid binding to 18.2. Yet, if we look at the FC effector function, we've intentionally mutated it to eliminate ADCC and CDC. This was intended to minimize unintended systemic immune activation driven by FC receptor-mediated 4-1BB clustering. What this leads to is the SCFE fragments at the bottom of that diagram. Those only activate T cells in the tumor microenvironment.
This is shown schematically on the left, where if we envision Giva, and this was shown preclinically as well as in our clinical studies, again, with the tox profile, that if Givastomig encounters a T cell in circulation because it has its FC effector function mutated and because there's no claudin 18.2 in circulation, if Giva interacts with a T cell, there is no clustering of 4-1BB, and there is no co-stimulatory signal stimulated by 4-1BB. In contrast, in the right side of the cartoon on the left, when a tumor cell expresses Claudin 18.2, several molecules of Givastomig combine to 18.2. This is where the clustering of a neighboring T cell, the clustering of 4-1BB occurs, and the co-stimulatory signal is delivered to the internal apparatus of the T cell, leading to T cell activation and elimination of the tumor. Next slide, please.
In the interest of time and in the interest of getting to the combination, I want to summarize the monotherapy data that led us really to spur on development. That is in our monotherapy data that's been presented in both poster form and recently published in Clinical Cancer Research literally a couple of weeks ago. Our monotherapy study in heavily pretreated patients showed that Givastomig had efficacy. The objective response rate as monotherapy in heavily pretreated gastric cancer patients was 18%. The toxicity profile was very favorable. We did not encounter a dose-limiting toxicity. We did not hit a maximum tolerated dose. The incidence of important clinically relevant treatment-related adverse events, grade three and above, was in the single digits. This was a very favorable monotherapy profile.
If we take the monotherapy profile summarized in this slide in the column on the far left and compare it against other assets in the 18.2 class, namely zolbituximab, the column to the right of Giva, and arguably the lead antibody drug conjugate, AZD0901, as well as the leading T cell engager from Innovan that activates T cells through a CD3 mechanism, the first thing I want to point out to you is that in comparison of Giva against these other lead assets, our 18.2 threshold is the lowest in the industry. We really can't get any lower without eliminating the requirement for any 18.2 expression. It is as low as it can go, hopefully providing enough clinical opportunity for as many patients as possible. The objective response rate is higher for Givastomig than zolbituximab.
In fact, if you look at the cutoff for claudin 18.2, the ORR for zolbituximab early in its development as monotherapy with a cutoff of 50% of cells being positive was only 9%. We come in at 18%. It is well noted that for both ADCs as well as T cell engagers, the objective response rates are higher, here shown as 28% and 26%. These assets are important. They are being developed in later lines of therapy. Another important comparison is in the safety table shown below. What I highlight are two things. First is in the red box for treatment-related adverse events. If you look at grade three or above treatment-related adverse events for Givastomig as monotherapy, what you can see is that we come in at 33% incidence.
If we go to the right and look at the ADC and T cell engager that have the higher objective response rate, those grade three and above events come in at almost double the incidence at almost 60%. I think this is important for a couple of reasons. First is that to use these agents, you expect more efficacy and you expect more toxicity. Knowing that Giva had this differentiated profile of having clinical activity, an ORR of 18%, and a well-tolerated tox profile, if we look at the grade three for the Giva column, look at neutropenia, nausea, vomiting, AST, ALT, GGT, which is another liver enzyme, you can see we're in the single digit for grade three and above treatment-related adverse events.
That says to me that if you know that you want to provide the best benefit for the most patients with really any solid tumor, you want to move new therapies into the frontline setting. Giva, given this profile, we thought was perfectly situated to move into the frontline setting. In contrast, if we look at the ADC and T cell engager, while it's clear they have efficacy to move into the frontline, it's our opinion that, in fact, it's going to be a bit more difficult to move something into the frontline, for example, with Nivolumab and chemotherapy, with monotherapy coming in at 60% treatment-related adverse events. I think it's going to be more difficult for those assets to move into the frontline space without modifying the standard of care.
The advantage we have is that we have shown now, as of six days ago, that, in fact, Giva can be added safely to an existing standard of care, Nivolumab plus chemotherapy. The one question we had for the phase I-B combination study is how do we choose the doses? Next slide, please. I won't belabor the next slide, but the next slide shows the pharmacokinetic and pharmacodynamic analyses that we did in our monotherapy study. The first thing I'll point out to you on the left is the pharmacodynamic effect. When one uses 4-1BB as a mechanism to activate T cells, T cells respond by secreting 4-1BB, and you can measure 4-1BB in the circulation. It's called soluble 4-1BB. That's how we measure the on-target effect of adding Giva when patients receive Giva.
What you can see in the top left is that if we look at over the course of time in the monotherapy study, the doses of 5, 8, and 12 milligrams per kilogram were doses in the monotherapy where we saw activity, where we saw our responses. You can see from the different shades of green and the blue at the top, those are the best responders in terms of soluble 4-1BB, but there is still a range between 5 mg, 8 mg, and 12 mg. That's shown in a box and whisker plot on the lower left, where again, if we look at the three boxes in the middle, we see kind of a plateau, but we can see also that the range of the box and whisker plot is a little bit different. We chose 5 mg, 8 mg, and 12 mg per kilogram because we saw responders in the monotherapy study.
We saw brisk induction of soluble 4-1BB, but we did seem to see differences in that secretion of soluble 4-1BB. Importantly, on the right, we can see that there appears to be linear kinetics for the exposure to Givastomig at really all doses. We can see that 5, 8, and 12 are clustered in the middle, but there are some minor differences between them. Based on this, we said, let's look at the profile we had for efficacy, toxicity. Let's look at the PK and PD data that we had for monotherapy, and let's design a phase I-B study in collaboration with our investigators and see whether or not Giva can be added safely and effectively to the frontline standard of care. This is where I'll turn it over to Sam to talk about the phase I-B combination data. Next slide, please.
Sure. Thank you, Phillip. It's happy to be back for an encore presentation of this data. I'm not limited by five minutes for ESMO GI, so I'm going to take a little bit more time going through a couple of these slides and maybe editorialize and contextualize data a little bit if it's helpful. As you can see here, this is the design of the phase I-B escalation and expansion study. A few elements to call out. One, as Phillip mentioned, the dose levels were selected based on monotherapy totality of data. In immunomodulatory agents where DLTs are often not reached, I believe this is a very common way to select doses and optimal biologic dosing. This is a U.S. study. Only U.S. patients. Historically in gastric and esophageal cancers, Western patients do worse than Asian patients. Most of the ADCs that were shown before are done entirely in Asia.
Just a point to keep in mind. The eligibility here is pretty standard for a frontline population, HER2 negative, and very broad claudin expression, and no cutoff for PD-L1 expression required for entry. All comers were allowed, which is notable. As you know, a couple of months ago, the U.S. FDA relabeled Nevo and Pembrol and Tizolizumab to CPS PD-L1 positive by any assay. Dose escalation is complete. That's the data set that is presented. Dose expansion is very close to completion, completed at 8 mg and near completed at 12 milligram per kilogram cohorts in combination. Standard primary endpoint here of safety with secondary endpoints, including the efficacy measures, which I will walk you through. Next slide. This is the table one, typical patient demographics across the cohort and in total on the far right. I think what you see here is a very typical Western gastroesophageal population.
Three rows, I think, to call your attention to are towards the bottom of this table. One, there are no MSI high patients in small cohorts. Those patients are particularly immunoresponsive and can drive outsized effects. That should not be contaminating these efficacy results with IO agents. The PD-L1 CPS distribution here in the overall population is pretty aligned with what we've seen from large phase III data sets. If you look at 649 and 859 and RATIONALE- 305, etc., you'll see that roughly about 70%- 80% of patients are PD-L1 positive, which leaves about 20% PD-L1 negative. Largely aligned. Similar to, as Phillip mentioned, with that histogram of claudin distribution, you can see here that the greater than 75% moderate intensity is the zolbituximab cutoff. There is about 30% of patients here who would not qualify for zolbituximab.
I'll also just remind people who look at that histogram, as a lot of people reference it, that histogram is built on moderate to strong, so 2+ or 3+ staining intensity. That's the distribution. It's probably even a larger patient population if you were to do 1+ by percentage of tumor cells and recapitulate that. That is unknown in the field. We don't know the exact proportion of patients who would be captured with 1+ at 1%. In our own internal data at Massachusetts General Hospital, this is probably somewhere upwards of 70%- 80% of all gastric and esophageal patients. We're still working through that data and have submitted it to a future conference. Next slide. This is the punchline. This is the waterfall plot for the combination. In the evaluable patients, you can see a response rate of 71%.
The absolute values of the PD-L1 expression and the claudin testing are shown in the bottom below the bars. Highlighted in red are the patients with low CPS and low claudin. Again, the patients who would not qualify for zolbituximab in particular. On the far right is really a more granular breakdown. If you look at particularly the cohorts that are being expanded upon, so the 8 mg and 12 mg, and yes, this is small numbers and only 17 patients overall and 12 in the 8 mg and 12 mg dose levels, you can see that response rate in those 12 patients was 83%. Those are the dose levels that are being expanded upon. If you look at the PD-L1 strata here, I think the absence of a pattern is probably the finding. There's not a clear relationship between PD-L1 expression and response.
Similarly, in claudin expression, again, you see activity across the broad range. That is summarized really in the bottom two rows. Next slide. Like we like to see with immunotherapy and immunomodulatory approaches, initial responses in this combination seem to be persistent and deepening and rapid, as you can see in this spider plot. You can see a few patients having achieved 100% RECIST responses. Everybody had a decrease in some level of target lesions. There's no initial bumps upwards, so no primary progression. As you can see, a large portion of these patients at the 12 mg have a subsequent deepening of response on follow-up scans. That is always an encouraging feature to see. Next slide. This is the swim lane plot and only a few things to comment on. This is as of a data cut in May of 2025.
In the far left are the columns with the absolute values of CPS and claudin scoring. You can, for yourself, at an individual patient level, relate back to the waterfall plot. You can see a median follow-up here is relatively short of nine months. With that amount of follow-up, you can estimate a six-month PFS rate, which in frontline trials has been used as a surrogate for median PFS and overall survival in several randomized phase II trials that led to subsequent phase III studies like KEYNOTE- 811, originally the Mega trial, which was 5-FU oxaliplatin, TRAS, and pembrolizumab used this that launched the phase III KEYNOTE- 811. There is some precedent to six-month PFS rates here. The six-month PFS rate is 82% for the cohorts being expanded upon and 73% when you include the 5 mg per kilogram cohort and the total patient population.
As marked by the greater than sign, there are multiple patients here still with ongoing treatment, several well beyond a year at this point. Next slide. This is really an important slide. I think as we continue to learn about claudin therapies in patients, greater experience with zolbituximab, many of the other claudin-directed strategies, we are becoming more and more accustomed to the class-related toxicity profile. We have a little bit more data to sort of understand and predict about what we're going to experience in the clinic with these drugs based on some of the early phase trial. As was shown in the monotherapy data, the overall toxicity profile here is relatively consistent with what we see from 5-FU oxaliplatin and immunotherapy with PD-1. I'd say the notable features to pay attention to is there is a slight increase in the infusion-related reaction.
You can see that there's only one patient who had a grade 3 reaction. That's over on the right side, right below the green box. The class-related effects center around nausea and vomiting. You saw the table early on for monotherapy and SPOTLIGHT data, but this is all low-grade nausea and vomiting. There is no grade 3 or grade 4 nausea or vomiting. The liver function increases relate back to some of the learnings from naked 4-1BB agonists like Urelumab or however you say it. Those drugs were plagued by some hepatotoxicity. Here, there's only a single patient, and this patient was in the 5 milligram per kilogram cohort and had some underlying comorbidities that may confound this a little bit further.
You can see only one patient had ALT and AST increases out of the 17 patients. It is encouraging to support the idea that this is a local effect and not a lot of systemic 4-1BB toxicity related. Next slide. This is just a summary slide comparing this data to other first-line combinations. You see on the left is the Givastomig dose escalation experience. On the right are the chemo IO and chemo plus claudin benchmarks, in this case, 649 and SPOTLIGHT. The obvious discrepancy is in the number of patients. I think a few things to call out. The AST/ALT are, again, not a worrying signal in this Givastomig data set. The neutropenia rate, and you can see that in the all grade and grade three for the Givastomig combinations, is a little bit higher early on.
We learned this, and now with standard growth factor modifications and clinical discretion, this has not been a concern, at least from my own experience on the trial. You can see it's numerically a little higher than the CheckMate 649. This was something we saw with the zolbituximab and chemotherapy. It is certainly not a toxicity that oncologists have any real clinical concerns about with the use of growth factors. You can see the nausea and vomiting comparison really stands out when you compare against the SPOTLIGHT phase III data. Next slide. I will hand it back to Phillip here.
Thanks, Sam. Just to summarize, we're very excited by the data. Giva is the first 18.2 asset tested in the U.S. in an unmodified way with the standard of care of immunochemotherapy in frontline gastric cancer patients. We have an 83% ORR in the doses selected for the expansion study.
I'll refer you back to the U.S. product inserts for both zolbituximab and Nivolumab in the frontline gastric cancer setting where the ORR is 47% and 40%. Very early data, very small numbers. This is, I think, a signal that we're all really intrigued by. The other is the toxicity. We have a low level of gastric side effects, either nausea and vomiting associated with zolbituximab or with hepatic toxicity associated with 4-1BB assets. You saw the spider plot. Responses were rapid, durable, and they deepened over time. Importantly, we're seeing signals that we may be providing benefit in patients who would not be expected to respond well to either of the standards of care, namely those with low PD-L1 and/or low claudin 18.2 expression. What we haven't talked about is the potential in gastric cancer.
Sam, just a few weeks before ESMO GI, had the pleasurable role of presenting the Matterhorn data in locally advanced, where now it appears as though the standard of care will change from FLOT, which is a different chemotherapy regimen, but now in combination with another IO agent, durvalumab from AstraZeneca. One of the obvious questions is, is there an upside for Givastomig ? We've shown that continuing to play this data out, showing that Giva can be added safely and effectively to IO chemo in metastatic disease. What about locally advanced disease? Moreover, gastric cancer is one of three GI malignancies that have a very high prevalence of 18.2 expression. The other two are pancreatic cancer and biliary tract cancer. I think there's a tremendous upside to Giva that certainly we're not acting on yet, but it's in the back of our minds.
We want to deliver on frontline metastatic gastric, but we know there is a tremendous upside. Next slide, please. Just to summarize, we just presented, obviously, our dose escalation data. We have two other important readouts with Givastomig coming up. We are presenting updated monotherapy data in Q4 of this year just to, again, provide additional biomarker data, association with claudin 18.2, association with PD-L1, and updated clinical outcomes. Importantly, the dose expansion data that we hope will validate and verify what we've seen in escalation will be presented in Q1 of 2026. We're very excited about that. With that, I think we'll close the formal presentation and we'll start the question and answer period.
Great. Thank you, Phillip. Yes, as mentioned, we'll begin conducting a question and answer session with our speakers. Please hold for a brief moment while we pull for questions. Our first question comes from Anpui Yu at Jefferies. Please go ahead.
Thank you for taking my question. This is Anpui on behalf of Kalishi. Our question is, could you discuss the clinical bar for the safety profile to really differentiate Givastomig /nivolumab combo or chemo combo from approved therapies in the frontline setting? Thank you.
Sam, did you want to take that?
Sure, sure. Happy to. Yeah, very good question. I don't know that there's an absolute bar that would say we were not going to use this or we're definitely going to use this for any specific claudin-directed therapy. There are definitely things that impact patients for sure, nausea, vomiting being a very bothersome symptom for patients as opposed to lab abnormalities which most patients don't have any feeling of. It's a difficult question to answer in absolute terms, but I think the features that we look at and we know is that if patients can stay on these drugs longer, they have a greater chance of benefit. There's a clear relationship in zolbituximab between exposure and efficacy. That's not a surprise, but it's harder to stay on drugs where there's a lot of side effects.
I think as someone who's treated a bunch of patients on this trial, it has been something that is a notable difference, the ability to remain on drug and exposed to drug, which in theory should improve the chances of a benefit. To put percentage numbers to it, you know if nausea and vomiting are more than 60% and 80% in the zolbituximab data sets, we look at numbers that are anything substantially less than that as encouraging, but it's hard to put an absolute term to that if that answers your question.
Great. Thank you.
Thanks for the questions. Our next question comes from Christopher Liu at Lucid Capital Markets. Please go ahead, Christopher.
Thanks for the questions. Once again, congrats on the data. Maybe one for the first question for Phil. In terms of the data that we saw at ESMO GI, going beyond May 15, is there any sort of color you can give us on the durability you've been seeing in patients? Maybe one more for Dr. Kleppner. How do you think about the place of claudin 18.2 assets while we also have other assets like FGFR2b inhibitors?
All right. I will start. Since the data cutoff of May 15, we obviously continue to follow these patients. I think we've seen no, and in addition to following these escalation patients, again, we now have more than 30, almost 40 patients now enrolled into the expansion cohorts. We are looking at now a large number of patients, right, close to 50 or 60 patients. We don't see anything new. We continue to follow patients. We know that in phase I-B studies, the longer you follow patients at doses that are not effective or less effective, patients can come off, but at the same time, patients are remaining on. The color is we pretty much are seeing what we saw in escalation, and the expansion data continues to unfold. I think with longer follow-up, we're going to learn a lot more.
I will also say that even though the median follow-up was nine months, we had the objective response rate data much earlier, and we chose to wait to present the data to ensure we didn't, knowing that we're combining a 4-1BB asset with a checkpoint inhibitor, having a lot of experience with development of checkpoint inhibitors at Big Pharma before, that I wanted to make sure we weren't seeing late-onset immune-related adverse events, and we're not. I think that that continues to be true. We're not seeing late-onset irAEs, and we continue to see the efficacy play out in safety in a pretty consistent way at this point. Sam, I don't know if you want to address the FGFR2b story.
Yeah, I'm happy to. This is like my whole career. We're great for patients that we're having more and more biomarkers. The standard ones currently are MMR or MSI, PD-L1, HER2, and claudin 18.2. Those are the tests that are linked to approved agents. We saw the press release from Amgen like 10 days ago with positive top-line results for the FORTITUDE-101 of the bemarituzumab in combination with chemo. There's no immunotherapy in that trial. We're waiting for the 102 readout. Presumably, FGFR2b will enter the biomarker world in the near future. Then we'll be obligated to test our patients for all of these things. Your question of how do you decide if someone has this and where does claudin 18.2 fit? I think MSI stands out as its own thing that sort of trumps all the other biomarkers, but it's only 3%- 5% of our patients.
Tiny, tiny slice of the pie. You've got HER2, which is firmly entrenched at about 15%- 20% of patients. Let's say that leaves 80% of the patient population out there for claudin and immune therapies. As clinicians, you know what we want to do and our gut instinct is find ways to expose our patients to as many active approaches as possible. We know immunotherapy is active, and we know claudin 18.2 approaches are active, but we don't have an approved combination for all of them together, which is what we would ideally want to think towards. I think claudin is clearly here to stay. How to extend claudin-directed therapies into later line therapies is probably going to be the world of ADCs.
How to improve upon frontline claudin strategies is really where I see agents like Givastomig fitting in and ultimately maybe even into the non-metastatic populations, as Phillip alluded to. I think the broad threshold here is an attractive clinical feature because it will probably capture the vast majority of patients, period, is my guess. If we have a toxicity profile that's perhaps advantageous over zolbituximab, that remains to be seen in a larger data set, then you would essentially take those patients away from zolbituximab potentially if the efficacy is equal. We look at biomarker, we look at efficacy, and if all things are equal on the efficacy side, we look towards toxicity to make clinical decisions. I think if you have a good balance of comparable efficacy with favorable toxicity profile, you win.
If I could add to that a little bit, one of the things that we think about, and it is great to have more biomarkers to select patients to identify patients most likely to respond to all therapies. From Amgen's press release and from our own internal research, FGFR2B overexpression at the threshold they use, they said it was 16% of patients of the HER2 negative population. Our own internal data calculated is 11%- 19%. I think it is a minority of patients who would benefit from this therapy, but it's great for them if they have this expression. The other thing I think we're all curious about is the discontinuation rate in the preceding randomized study was 40% of patients. Of those 40%, 70% had eye toxicity. Having led an ADC program where there was eye toxicity, it becomes oncologists get very timid about managing eye toxicities.
Patients don't like it as well. I think we're all very encouraged. I can't wait to see the data. I worked in FGFR2 as a PhD student. I'm very interested in the data, but I think that, again, in terms of from a commercial standpoint, I don't view it as a significant commercial threat to any future Giva potential.
Got it. Thank you for your answers, Phil and Sam.
Thanks for the questions, Christopher. Our next question comes from Andres Maldonado at H.C. Wainwright. Please go ahead, Andres.
Hi, everyone. Thank you for taking my questions. Maybe I'll start off with one for Dr. Kleppner here. Could you talk about your experience with zolbituximab infusion times? What percentage of patients that you've treated with zolbituximab or that your colleagues have treated as well, that you've had to extend the infusion times to really give the patient a better tolerability profile? In that same context, could you help us shape how you believe Giva may aid in attenuating these infusion times, especially in the context of maybe future potential stronger chemo combos?
Yeah, good question. I've been involved in the zolbituximab development since quite near the beginning in the Alastro trial where we explored multiple cohorts and gained some experience and then on into the phase three trials. This drug was approved in October and really started being deployed in the U.S. probably closer to December. I know we first dosed standard of care zolbituximab in December here in Boston. The infusion time, the first infusion with the loading dose is four hours and sometimes longer, depending on toxicity. It's given prior to the FOLFOX chemotherapy. Four hours plus four hours in optimal setting is a long day and a lot of infusion chair time. At academic centers where we have infusion centers that operate 24/7, we can accommodate this. In community and many practice settings, it is a potential barrier.
I would say there's a learning curve to the drug management and the toxicity to the point that I think we've only had to decouple zolbituximab from FOLFOX over separate days once or twice. I know of colleagues around the country where this has been a more common phenomenon. One of the most effective management strategies is interrupting the infusion, giving patients 60 minutes to let the nausea and vomiting potentially subside and then resuming. I think there will continue to be a learning, but it is a potential clinical implementation barrier for zolbituximab that, in my experience with Givastomig , has not really applied. As you saw in the toxicity, there have been some infusion-related reactions, and that is a real phenomenon. We have been able to mitigate that with increased premedication strategies, just standard things we do for anyone with infusion reactions to cetuximab or imivantamab, for example.
With that, I think we have, at least in my own experience, and I don't know how many people I've treated, more than 20 in the Givastomig program, we have not had recurrent infusion reactions. I don't think I've had a single patient where we had to do the decoupling of the Givastomig and the FOLFOX and the nivolumab. I hope that answers your question.
Yes, very much so. Thank you for that. One last question. Given that there's not a clear pattern in teasing out a correlation of expression levels between PD-1 and 18.2 thus far, can you help us get some historical context? How common is it that there's a correlative expression correlation with PD-1 across cancers in your experience? How important is this? How important should this variable be? Does this hold true also in maybe a HER2 positive context? Thank you very much.
Yeah, I think as we see larger and larger data sets, and I think transparency in big pharma trials has improved where all of the companies now are largely publishing the biomarker data. Phillip referenced the recent JCO Precision Oncology paper from Amgen investigators looking at the FGFR2 landscape and correlating with some other standard of care biomarkers. Claudin 18.2 was not on that, but I do have an understanding that they may go back and do claudin 18 to understand claudin 18 FGFR2 overlap. Honestly, I think the overlap idea is a little bit blown out of proportion. I think PD-L1 and claudin 18.2 run through all of the other biomarkers at essentially independent, not really biological relationships. In HER2 positive disease, about 85% of patients are PD-L1 positive in the KEYNOTE- 811.
There may be some pre-selection in any clinical trial where investigators already know the results of local testing. In large retrospective studies, it's a little bit more like the standard distribution of PD-L1, like 80% positive, 20% negative within HER2. It's pretty much the same within claudin. I suspect it's the same within FGFR2. The only one where there's probably some mutual exclusivity is really the MSI high population. Those are almost universally PD-L1 positive and exceedingly rare for them to be HER2 amplified. We looked at 1,800 patients, and there's only two patients that were overlapping between MMR deficient or MSI high and HER2 positive. That's probably a true biologic phenomenon. Outside of the MSI, I think they essentially run through all of the biomarker strata at the same prevalence that's been reported.
We actually wrote an editorial on this for ESMO Open, like, I don't know, a year ago, if you want to reference something.
Thank you very much.
Thanks for the questions, Andres. Our next question comes from Kumar Raja at Brookline. Please go ahead, Kumar.
Thanks for taking my questions and congratulations on the data. First, with regard to the tumor location as well as the ECOG performance status of the patients in the trial, how does that compare with the real-life setting? Once we have the dose escalation data, how soon can we get to a randomized trial? How do you think this randomized trial is going to look? Thank you.
I can comment on the patient demographics in the dose escalation. If I understood the question, it was about the anatomic tumor location breakdown in the Givastomig dose escalation versus typical U.S. or Western patient population. What I would say is quite common, if not perhaps a little bit of a trend towards more gastric cancers in this data set. With 17 patients, I think it's really hard to say much about that. In most Western or U.S. trials, looking at these populations, it's often somewhere in the 60/40 gastric GE junction. That's sort of just back of the envelope calculations from Western-oriented phase III studies. This small data set is enriched slightly towards gastric. Gastric and GE junction, you know, of course, therapeutically, we lump them together. Biologically, proximal gastric cancers are the same as GE junction cancers.
If you really want to get into the weeds, you'd have to look at the location within the stomach. Distal gastric cancers are a little different than proximal, but I'm not aware of that granular level of data right now.
We have a cash runway of $169 million that takes us into 2027. We recognize that randomized data is an important data set to generate. We are equipped to execute a randomized phase II study in our defined runway. We have not announced anything regarding such a trial, nor have we finalized the design of such a trial. We recognize the importance of randomized data, and we're well equipped to generate that data within our runway.
Thanks so much.
Great. Thanks for the questions, Kumar. Our next question comes from Ethan Markowski at Needham. Please go ahead, Ethan.
Hi, everyone. This is Ethan off of Gail Bloom, and thank you for taking our questions and congrats on the data. I know we still need to see the dose expansion results, but based on the data thus far, how are you thinking about potential dose selection between the 8 mg and 12 mg per kilogram levels, given similar efficacy and safety profiles demonstrated thus far? Maybe just a clarifying question. You mentioned that several patients achieved 100% reduction in lesion size. Were they not considered complete responders because there were still non-target lesions present, or was there some other reason? Thank you.
Yeah, I can take that. It is true, you nailed it, that the patients who had 100% reduction in target lesions did not have resolution of their non-target lesion. That was the reason they were not complete responses. The other part of your question about dose selection, there are important, right? You saw from the efficacy table, there are subsets of patients, claudin high, PD-L1 high, claudin low, PD-L1 low. With only 17 patients, it's really hard to discern differences between 8 mg and 12 mg. We had five out of six responders in each group. The toxicity profile is well tolerated in each group. We saw no evidence of dose-dependent toxicities. The expansion data is going to be very important for us to be able to determine the best dose moving forward.
I will say that I think it's likely, based on informal consultations we've had with people with FDA experience, that this approach that we have with sequential enrollment of our dose expansion cohorts will satisfy Project Optimist. We anticipate having a much more robust data set to make a determination of dose, but it's premature at this point because, again, we may end up seeing subtle differences between the two groups in terms of toxicity, right, as we play out with 25 patients- 30 patients in each group as opposed to six patients in each group, as well as efficacy in subgroups that I mentioned, the low PD-L1 and low claudin. It's a bit early, but we anticipate we'll be able to make a decision with a more robust data set.
Thank you for taking our questions.
Thanks for the question, Ethan. Our final question comes from Justin Zelen at BTIG. Please go ahead, Justin.
Thanks for taking our questions and congrats again on the data. I want to ask about expectations on durability. The estimation on six months PFS rate is helpful. I want to ask if you expect median PFS to exceed eight months in CheckMate 649, if that's the appropriate benchmark, and any commentary on what a clinically significant delta on PFS could be. My second question is, any commentary on how implementation of growth factor and clinical discretion could impact the neutropenia rates that you've seen thus far? Thank you.
I'll take a first swing. Obviously, durability of response is important. That is unfolding as we speak. To get to a median, we obviously need events, and we simply don't have the requisite number of events. We just need to follow the data for that to happen. I would also say that with IO, which often does not give a significant improvement in objective response rates, and again, given my prior experience developing checkpoint inhibitors, I think that if our ORR is maintained to the level that it is now, I would be surprised we don't see an improvement in PFS. We'll just have to see in terms of the median once the events happen. I will mention that in relation to the neutropenia, we saw a higher incidence of grade 3 neutropenia at the lower dose levels.
Once we recognized this and conferred with the investigators, after the implementation, rather uniform implementation of growth factor support, what we found is that the incidence of grade 3 neutropenia was actually lower at the 12 milligram per kilogram dose than it was at the 8 mg and the 5 mg. We think that it was highly effective in doing that. In the early data from our dose expansion cohorts, again, neutropenia has not been an issue. I think we nipped that in the bud.
Thank you for taking the question.
Thank you. With that, I think we'll turn it over to Sean for some concluding statements.
Great. Thanks, everyone, for your participation today. I hope you enjoyed the conversations. We're very excited to share the data from Givastomig , which we believe is a potential best-in-class asset, differentiated obviously by its ability to be added to the undiluted standard of care in frontline metastatic gastric cancer and also by its broad claudin 18.2 coverage or wide patient coverage. We are encouraged by the clean safety profile and the strong efficacy signal from the escalation study. We are moving quickly in the expansion cohort as we speak. Looking forward, we want to continue to share additional clinical data with you guys in the near future. Thanks again for your support and for your participation today.