Thank you and welcome to the call. Following prepared remarks from the Company, we will open the call for questions. [Operator instructions] This call is being recorded and will be available on the investors section of the ARCUS website. I will now turn the meeting over to Pia Eaves, Head of Investor Relations And Strategy

Hello everyone and thank you for joining us on today's conference call to discuss ARCUS's third quarter 2023 financial results and pipeline updates, including today's presentation of data from our EDGE Gastric trial at the ASCO monthly plenary. Today you will hear from Terry Rosen, Chief Executive Officer, Dimitry Nuyten, Chief Medical Officer, Jennifer Jarrett, Chief Operating Officer, and Bob Goeltz, Chief Financial Officer. For Q&A, we will also be joined by Juan Jaen, President and Head of Research. I'd like to remind you that on this call, management will be making forward-looking statements, including statements about our cash runway, our investigational products, our expected clinical development milestones and timelines, and the potential market opportunity and drug-treatable population of any indications being pursued by our program. All statements, other than historical facts, involve risks and uncertainties that may cause our actual results to differ. Those risks and uncertainties are described in our most recent quarterly report on Form 10-Q that has been filed with the SEC. We strongly encourage you to review the filing. For today's call, please refer to our latest corporate deck, which can be found on the investor section of our website. With that, I'll turn the call over to Terry.

Thank you very much, Pia, and thank you all for those who are on the line. This is the first quarterly earnings call we've had in some time. So in addition to the Domvanalimab data that was just presented at Ascol Plenary, we're also going to discuss our other upcoming data readouts, specifically for AB521, our potentially best-in-class HIF-2a inhibitor, and Quemliclustat, our first-in-class small molecule CD73 inhibitor, both of which we'll be sharing data with you just in the next few months. So just a few hours ago, many of you likely watched Dr. Juan Jaen, present initial data from our Phase 2 study, Edge Gastric, for Domvanalimab and first-line metastatic upper GI adenocarcinomas. As a reminder, Dom is the only Fc-silent anti-TIGIT antibody in late-stage development, and we and our partner, Gilead, are pursuing a very broad development program that includes four Phase 3 studies in non-small cell lung and upper GI cancers. Today's data are extremely encouraging on all fronts. ORR for the PD-L1 high population clearly exceeded historical benchmarks in this setting. Data in the overall patient population appears differentiated, particularly relative to the most contemporary Phase 3 studies in this patient population, or its keynote 859 in day genes rationale 305. Most importantly, six-month landmark PFS data for both the PD-L1 high and overall patient population appear to be quite differentiated from the benchmarks. These data are very supportive of our ongoing Phase 3 study, STAR-221, which is evaluating Dom in the same combination, same setting underlying today's dataset. STAR-221 is enrolling extraordinarily well, and we expect that today's data will further enhance interest and actually accelerate recruitment in this study. In addition to Dom, we continue to execute on the remainder of our pipeline, including our HIF-2-alpha and adenosine pathway programs. Regarding our HIF-2-alpha…

Thank you, Terry. I'll start by directing our audience to slide nine of our updated corporate deck. The data presented today are from cohort A1 from our Phase II EDGE-Gastric platform study, which is evaluating DOM-based combinations in both the first- and second-line setting. The A1 cohort is evaluating dom +/- zim +/- mFOLFOX chemotherapy in 1L Gastric, Gastroesophageal Junction (GEJ) and Esophageal Adenocarcinoma (EAC). As Terry mentioned earlier, this is the same setting and setting we are evaluating in our Phase III study, STAR-221. This cohort was specifically designed to generate safety and efficacy data to support regulatory requirements for the initiation of STAR-221 in certain countries outside of the U.S. Turning to slide 11, we described the baseline characteristics of this cohort. Forty-one patients were enrolled, and all patients are included in the efficacy analysis. The cohort enrolled a relatively diverse patient population from 20 different sites in the U.S., France, and South Korea. There was an even distribution of patients included in Asia versus the rest of the world, and 63 percent of patients had gastric cancer, 24 had junction tumors, and 12 percent Esophageal Adenocarcinoma. PD-L1 status was evaluated using the TAP scoring method, with 24 patients having TAP less than five, and 15 patients, or 37 percent, with a TAP greater than or equal to five. Two of the 41 patients included in the overall population did not have tumor samples available for PD-L1 testing. Note that the CPS and TAP are two methods for PD-L1 assessment, and both are used in gastric cancer. We'll show you later on this call some of these methods, sorry, that these methods have a high concordance. BMS and Merck have historically used CPS for their phase 3 studies of anti-PD-L1 plus chemotherapy, while Beijing is using TAP, including…

Thanks, Dimitry. I'd like to start by talking about the TIGID field overall. Our most recent feedback from KOLs indicates a clear and growing data-driven enthusiasm for anti-TIGID as an innovative therapy. Following the release of data from the second interim analysis of Roche SKYSCRAPER-01, we engaged a third party to conduct a survey of 730 oncology KOLs regarding their sentiment on anti-TIGID. You can see these results on slide 35 of our investor deck. Approximately 80% of respondents believe it is likely that Turago plus Atezo, Roche's anti-TIGID and PD-L1 combination will receive FDA approval. Additionally, based on data they've seen to date, 83% expect anti-TIGID to play a moderate or larger role in the treatment of first-line PD-L1 high non-small cell lung cancer. These survey results are telling. As the evidence on anti-TIGID accumulates, physicians are increasingly convinced of the potential for this mechanism to change the treatment paradigm in lung cancer. Since we are combining DOM with the current standard of care, FEMO plus anti-PD-L1 and adding almost no toxicity, this is a very easy value proposition for physicians. Today's results continue to demonstrate the potential for anti-TIGID to improve outcomes and indications where anti-PD-L1 is successful. As Terry mentioned, gastric GE junction and esophageal adenocarcinomas represent a $3 billion plus market opportunity with approximately 25,000 patients in the U.S. alone and over 100,000 patients across the G7 countries. Note that we are pursuing adenocarcinomas, a different histology than that which is being pursued by Rosh and Beijing in their respective phase 3 and 2 studies. In fact, we in Gilead have the only anti-TIGID in phase 3 development for first-line upper GI adenocarcinomas, which is one of the fastest-growing cancer types in the U.S. and Western Europe. Many of you are aware that there is another class of agents being developed in gastric cancer directed against CLADIN 18.2s, which were also discussed in today's plenary session. While the efficacy data are encouraging, published literature estimates that only anywhere from 20 percent to 40 percent of patients are high expressors of CLADIN 18.2, and these molecules are associated with significant GI toxicities. We also expect it will take several years for CLADIN 18.2 testing to be successfully rolled out. With anti-TIGID, we believe we have the potential to achieve similar efficacy without the toxicity associated with the anti-CLADINs and potentially avoid the need for any testing, allowing patients to go directly to treatment. Today's data set will support both study recruitment and regulatory filings for a combination, and we are moving aggressively to secure our position as first-to-market and best-in-class in this indication. I'll now turn it back to Terry to discuss our HIF-2-alpha and CD73 programs.

Thanks very much, Jen. So, first let me start by level-setting a bit on HIF-2-alpha. It's a transcription factor, has no ligand-binding domain, and as a result, it's proven very difficult to create molecules that are high-quality inhibitors but also that have ideal drug properties against the target. That's why there are very few other HIF-2-alpha blockers in clinical development today. Belzutepin is the first and only HIF-2-alpha inhibitor to be approved to date, and it's only approved for BHL-associated cancers, although Merck recently filed for Belzutepin's first approval in Clear Cell RCC. Slide 47 of our corporate deck describes the value proposition for AB521. It's really pretty simple. The primary limitation of Belzutepin is substantial. Its oral absorption saturates at a dose of 120 milligrams. So what does that mean? Even when higher doses of Belzutepin are administered, systemic drug exposure does not meaningfully increase. In fact, it's very clear from the published data for Belzutepin that the approved dose of 120 milligrams was chosen because doses beyond 120 milligrams did not result in meaningfully higher plasma levels. This had nothing to do with dose-limiting toxicity. Merck, in fact, just released results for LightSpark 13 at ESMO, which evaluated doses of 120 milligram versus 200 milligrams of Belzutepin. No surprises. They concluded that there was no difference in ORR between these two doses and that the results support their dose selection of 120 milligrams for their trials. However, it's super important to note that they made no mention of whether they were achieving higher levels of drug exposure with the 200 milligram dose, and they did not share any PK or PD data for this study. So consistent with the previous literature, we just view this as more evidence of Belzutepin's PK and PD limitations and the inability, in…

Thanks, Terry. Arcus continues to be in a very strong financial position. Our cash at the end of September 30, 2023 was $950 million and our net cash utilization through three quarters of 2023 has been $188 million. Importantly, we share cost 50-50 with our partner Gilead on a large portion of our portfolio, including all of our ongoing phase three studies. We now expect cash utilization for full year 2023 to be between 2$65 million and $290 million, down from our prior guidance of $295 million to $325 million. And we continue to expect our cash balance to fund operations into 2026. Our cash runway guidance excludes potential opt-in payments and approval milestones from our partners. Turning to our P&L, we recognize gap revenue for this quarter of $32 million, which compares to $29 million for the second quarter of this year. Our revenue is primarily driven by our collaboration with Gilead and we expect to maintain similar levels of revenue in the near term with small fluctuations due to updates to our clinical development plans and timelines. In addition to our partnership with Gilead, Taiho is our partner for DOM in Japan. In the fourth quarter, we received a milestone payment of $14 million from Taiho related to their participation in our STAR 221 pivotal study, and will receive another $14 million in the first quarter of 2024. We are also eligible for additional milestone payments from Taiho related to STAR 121. Our R&D expenses for the third quarter are stated net of reimbursements from Gilead and were $82 million as compared to $84 million in the second quarter of this year. In the current quarter, non-cash stock-based compensation represented $8 million of our R&D expenses. We expect modest increases in our R&D expenses as our phase three studies mature. However, spend may fluctuate based on the timing of clinical manufacturing activities and the purchase of standard of care therapeutics for our clinical trials. For example, in the third quarter, spend in these areas was lower than in the preceding quarters of this year. G&A expenses were $30 million for the third quarter of 2023 compared to $28 million in the second quarter of this year. Non-cash stock compensation represented $10 million of our G&A expenses for this quarter, and we expect G&A to remain stable over the remainder of 2023 and through 2024. For more details regarding our financial results, please refer to our earnings press release from earlier today and our 10Q. I'll now turn the call back over to Terry for some closing remarks.

Thanks very much, Bob. So to wrap up, first off, we're really excited about today's data set. It provides evidence that supports DOM as potential best-in-class anti-TGIT and first-line upper GI adenocarcinomas, a setting we're well on track to be first to market, and first to market by a distance with an anti-TGIT antibody. I think it's also important that if you elevate a bit, it really continues to support DOM-ZIM as a best-of-class anti-TGIT, anti-PD-1 doublet, the combined holistic profile of efficacy and safety profile. As we execute on our four phase three trials for DOM, we continue to advance the remainder of our pipeline, and we're really looking forward to sharing more data in the very near term on QMLE and pancreatic cancer in our HIF-2-alpha program. Both of these programs have been ongoing for some time. I think we're going to have some data that you can really look at as inflections for both. So with that, I'd like to thank you for your continued support as we work to develop, we believe, very innovative combination cancer therapies for patients that are in need. With that, I'll open up the floor for questions.

Thank you. [Operator instructions] Our first question comes from Yigal Nochomovitz from Citigroup. Please go ahead.

Yeah, hi, thanks very much for taking the question. I just wanted to follow up on the comment Dimitry made around the design for star 221. Dimitry, you mentioned that you, in the plan for the stats, it's a pre-specified percentage of TAP high and TAP low. Can you comment on what that is, and is it similar to what you've seen in the EDGE-Gastric?

Thanks for the question. So as my standard answer around statistical analysis plans is we don't comment on the exact details. I made the comment specifically because I want people to be sure that the trial is designed to look at both the overall population and the high population. In order to do that, we have specific criteria about the number of patients, number of events, and thereby the power for the survival analysis in those groups. What I can say in general is what I've said before. We are expecting a slightly lower number of PD-L1 high patients. As you can see in our current data set, we are at about 40% PD-L1 high. That's something that we were expecting to be lower than checkmate, given that checkmate changed the standard of care more for high expressor than low expressing tumors, and thereby we anticipate that patients with PD-L1 low disease disproportionately would want to be on a clinical trial. That's why we've done comprehensive planning to make sure that both of the primary populations have a good shot on goal for success.

And the primary endpoint is on the overall, or do you have a, how does it work? Do you have a primary on the PD-L1 high?

Both, so we have dual primary endpoints for overall survival in both the high expressors and in the full population, the intention to tree population. So both can be tested independently, and the study will be declared positive if either one of those is positive.

Okay, and then the discussion today mentioned the FC silent might potentially be the reason why you saw the higher efficacy in the TAP high population that were more immune sensitive. Just, can you comment on that? Do you agree with that hypothesis?

I think, so Yigal, I’ll take that. I think it's an overstatement on the number of patients, et cetera, to get into any differences. The one thing that we would comment on the FC silent is clearly that safety profile, you know, is something that enables patients to stay on better, et cetera. So we think that manifests itself in a better overall profile. I think it's a little too nuanced to get into differences in with respect to the PD-L1 status of those patients.

Okay, got it. And then one on 521, Terry. I'm just interested that, you know, you've made the argument that you have better PK than the Merck compound for several quarters. Can you comment a little bit on what's driving that? I mean, some of the hypotheses I was thinking about was maybe less protein binding or engineered to avoid certain metabolic pathways or better gastrointestinal absorption. Can you comment at all as to what design features you've introduced that are giving you that edge you refer to? Thanks.

Yeah, so what we know in a black box sense is simply their absorption, you know, is limited. And if they go, actually, if you look at the data closely, there's very little difference between 80 milligrams, 120 milligrams, 160 milligrams, and 240 milligrams. You have an issue with absorption. In their case, it probably relates more likely than not simply to solubility. Our molecule doesn't have that liability and our molecules simply will behave like most small molecule, organic molecules, and we don't have that problem. So it's probably something as simple as that, physical chemical properties.

Our next question comes from Jonathan Miller at Evercore ISI. Please go ahead.

Hi, hey guys, thanks for taking the question. I'll follow up on HIF-2. I guess it makes sense that you've got better exposure than Merck does, let's take that as read. But I guess what we don't know yet is if hitting the target harder really does result in better outcomes that Terry mentioned on the call. So I guess my question is, what data gives us that clear signal of additional efficacy in addition to better PD? Is it likely that we could get something in escalation that does that, or do we have to wait for expansion data later in the year?

Thanks, John. And that is a great question, and of course that's the key question. I think the arguments that we make are a combination of factors from our own data as well as the data with the Merck molecule. So I'll start with the latter. What's clear from their data is that if you look, the vast majority of the patients, for example, in their earlier study in India, you see this 20 some odd percent response rate. You see tumor reduction, but you see slow kinetics. You don't see the sort of response rate that you might envision, particularly in a setting that is known to be driven very strongly by HIF-2-alpha. So you might expect that you would see stronger results. So for example, if you had an ORR that was 90%, and you could hit the target much harder, that probably wouldn't buy you much. Now if you come to what we know about our molecule, and first of all, let me finish on their molecule, we know that they can't hit the target harder. So as you know, usually in oncology, you've either modeled something or you've hit an MTD. In their case, they simply can't get to a higher exposure, and in fact, that's been demonstrated in multiple ways by them. What we have shown, in fact, is that basically, if you look at the pharmacodynamic endpoint that's easily detectable, we can get the same effect that Merck gets with their clinically used dose at about one third to one quarter the dose. And the dose that we're using now go forward is four-fold that dose. So clearly, we should be hitting the target harder, and then you said the obvious question will be now going to randomized study in a large trial, do we see that manifest itself? Certainly from the escalation data, you'll get a good feel that the molecule is behaving well, that roughly half or a little more than half of that 12 patients that has RCC. Keep in mind, that's a very late line group of patients. It's heterogeneous, but you'll be able to pick those patients out and see that it looks like the drug is working. So I think that what we would say is that the phase two data in the expansion cohort will be one that we will get the real true sense of how much better we might be doing by hitting the target harder.

Great, great, that makes sense. And then I guess on Tidget, it seems like the plenary discussion this afternoon wasn't very excited about the EDGE results, I think because of how early the PFS curve is at this point. So your PR suggests that mature PFS is coming second half, but I know that the virtual plenary series usually comes with ASCO presentation, is it fair to expect we'll get updated PFS at ASCO next year as well?

So first, let me just come back to the actual data. We think the data actually look quite encouraging. As you know, the six month PFS were mature, and they're quite meaningful, both in the high PD-L1 as well as the totality of the patient population. So if you think about it, clearly that 93% landmark PFS for the high PD-L1 is quite meaningful, and even the 70 some odd percent is quite meaningful. If you go back and for example, look at the Checkmate study, what you've got is a median PF on the order of six, I'm sorry, seven to seven and a half months, and that's in the PD-L1 high. So in our total population, we're still at 73% landmark at six months, which bodes pretty well if you even start to compare it to like what would be 100% high PD-L1 and we're diluted. With that said, insofar as expectations, we do not know, our guidance has been that we expect median PFS to be mature in the second half of next year. So we'll just have to see, obviously that's event driven. We had a large, as you saw, a large number of patients remain on the study, so we'll see how that plays out as the year goes along.

But do you have a presentation at ASCO in the summer as well?

Oh yes, absolutely.

Okay, thanks. Hi everyone, this is Pia. If you could please keep it to one question as much as possible, we'd like to get to everybody.

Pia, I was liking the multiple questions. It's a bonding experience.

So you can call the next person.

Thank you, our next question is from Peter Lawson at Barclays, please go ahead.

Great, thank you. Thanks for taking my question. I'll try and make it a multi-part of you. Just on your HIF-2 alpha, the data we're going to see early 2024, do you think that differentiation starts emerging, whether it's in ORR, or do you think it kind of has to, we have to wait to see that emerging in PFS?

Yeah, I don't, go ahead, Jim.

Yeah, I think, so first of all, Terry started to talk about this, but maybe just again to tell you exactly what that data set is going to be. So the dose escalation is going to be 12 patients, approximately half of those patients, so approximately six of those patients do have RCC. All of those RCC patients were treated at a pharmacologically active dose. As far as will the ORR be differentiated relative to the 20 to 24% you see in cells, I think the problem is the sub-dominators are still going to be really small. So whether it's really high ORR, really low ORR, I think the denominator's too small to really tell you much, which is why Terry was saying we wouldn't rely on the ORR for the dose escalation. So I think for the dose escalation, what's going to be most interesting to see is, one, just what does the waterfall clot look like, and then RCC patients, are you seeing activity? Two, are there any signs that maybe the responses are happening a little bit faster, because as we know it's called Zutophan, for a lot of patients it can take six months or more to get a response. And then once we get to the expansion cohort, so that'll be sometime in 2024, that will be a higher N in the denominator, and so the ORR will start to be more meaningful.

Our next question is from Robin Karnauskas from Truist. Please go ahead.

Great, I have 10 questions. I hope you can do them all in a very short period of time.

Robin, you can have 100 questions.

All right, I just was noticing, given the curves, like Bill and I were talking about, like how patients who experienced tumor reductions were PD-L1 negative, and that's so unusual in the waterfall clot, so how do we think about that? How do we put that in perspective for PD-L1 high, and can you help us with the median 21 months of follow-up in pancreatic cancer, and put that into context? Like those are three, I'm sorry I got three, but like I'm just putting them out there. You can cut one off and give it to another person. But I thought the waterfall clots were very interesting.

So one thing to keep in mind with PD-L1 levels is that once you introduce chemo into the mix, you sort of have a dynamic situation, and that's an important aspect of this therapy. So it's not surprising if you heard one of the discussants talk about immunogenic chemotherapy and what that does. And so if you get an immune response, now having immunotherapy, you can generate what might be surprisingly good response. And I think it's something we talk about, since we work on a number of immunotherapies, and we talk about this a lot in the context of QMLE, you can have a patient who might not even have that much of a tumor reduction, but from a survival standpoint can really benefit, because if they can amount to T cell response, you can enhance it. So we just feel that that's certainly amongst the rationales for when you start bringing in chemo together with immunotherapy, that the PD-L1 levels don't tell 100% of the story. What was your second question, Robin?

So if you, for the pancreatic trial, like with the median follow-up of 21 months, how do we put that into context? Is that a long time? That seems like a long time for pancreatic cancer?

Yeah, yeah, it's a long time. And to be clear, when we think about what the median OS readout is, that tends to range somewhere between like nine to 11 months on the high end. If you talk to clinicians, given this is so devastating, even something that was two to three months longer than that would be meaningful. So we think we're going to have something that really looks like it's moving the needle.

All right, I'll put my another eight questions back in the queue, enjoy.

We can always talk later.

Our next question is from Kaveri Polman at BTIG. Please go ahead.

Yeah, good evening and congrats on the EdgeCastric trial results. For PD-L1 expression, can you provide any color on your rationale for using TAP assessment and how different it is from the CPS score that was used for the 649 trial? Also, if you could tell us about your rationale for using cutoff of five versus one or can like some other trials have used?

Dimitry, you'll take that.

Yeah, so rationale to use TAP, there's a couple of different things, right? So there's the scoring method, which is TAP or CPS, and then there's the different essays. And we need a combination of those for regulatory purposes. And we need to have the work done as companion diagnostic to use it as stratification factor in a trial. That work cannot be done on every single essay. That's why we work with SB263. When it comes to TAP and CPS, my simplest way of answering the question is, as the discussant said, the TAP is kind of a visual scoring method. She didn't use those words, but it's easier for pathologists to do. CPS has a strong scientific rationale, but if you look at BMS's initiated trial that was published, looking at 13 pathologists who have been in practice for 20 plus years and got training, the concordance of the inter-observer availability for the CPS essays is not great, wit in the 50 to 60% range. So I think CPS is somewhat challenging in implementation in a larger setting, and TAP has the benefit that it's easier to implement. The 5% threshold is a reasonable threshold extrapolated from the data for CPS in the public domain. We think it's a good threshold. And then lastly, there was one more sub-part of the question. No, that was it? Okay.

Can we get everything to burn?

No, sorry, no, and I wanted to make the point, we showed the concordance, right? So TAP is easier. I think that that makes it easier to implement it on a larger scale, also beyond clinical trials. But we have provided, let's say, the CPS data on our own data set in the corporate deck as well, to provide confidence that the number of patients captured by both essays is not widely different. There's a high concordance between these two methods.

Got it. And if I can just squeeze one more. For ARC8 trial, the comment you made about sharing data from externally generated synthetic control with GenAbraxane, can you give us some color around that?

I'll give you minimal. I mean, what I'll say is it was done with metadata. It was done in an extremely rigorous way, very much to get matched patients. And quite frankly, when you start to think about, particularly in trials that aren't like 1,000 patients or 800 patients, the reality is you actually, from a control standpoint, get a better patient population that's more matched to your data set, assuming that they have a large enough, well-annotated data set, which they do. And that was even, that took a couple months to just even do that part of the analysis to make sure that they could do it. So I think everyone's going to be very pleasantly surprised by the quality of that data set. I think it's going to bring a lot of confidence. It's the type of data set that you can certainly even take to regulatory agencies in discussing your plans going forward. So we think that's going to be a big deal in part of our total package for that program.

Our next question comes from Mara Goldstein from Mizzou. Please go ahead.

Great, thanks so much. During the plenary today, there was a comment along the lines of the presentation of this data. There's a hope that this would also help facilitate STAR-221 enrollment. So can you talk a little bit about where that trial is from an enrollment characteristics perspective and what the timing looks like from here?

Yeah, so I love the question. Even though we can't give you a bottom line answer, the trial is enrolling gangbusters. So when we talk about help, it isn't so much like, oh, it's in need of help. It's that it's enrolling incredibly well. In fact, I would go and say that more likely than not, if you ask us to bet, I think that's probably going to be our first trial to read out. Our guidance has been that our trials will be, let's say 2025 plus. We're not changing our guidance, but it's enrolling well ahead of schedule. A couple comments on the reasons. I think there's been a lot of enthusiasm that's data-driven what's going on just in the TIGIT field. In this particular setting, as you can tell from the discussions, there's also a lot of clinical trial need. There's simply not a whole lot happening in an innovative sense. And we believe this data set unquestionably, so all of the investigators we've spoken to are really excited about the data and we think it's just going to accelerate things further. So I keep an eye on this. I think what's interesting about this, when you think holistically about the field, for us, this trial is going to help us close the distance on who and what ends up being the first doublet that comes to market in the field.

The next question comes from Donna Grabas from Lerink Partners. Please go ahead.

Jeff, hi, this is Dana. Do you think you could share your thoughts on the potential difference in characteristics of DOM relative to the various TIGIT antibodies beyond Fc function to these points? I mean, what else do you think could explain the arguably disappointing phase two data that we saw from osaprilumab, the competing function antibody that was recently presented to ESMO? Thank you.

Yeah, so I mean, I think primarily, we're going to come to that FC as being a key differentiator. So, you know, it's absolutely clear that FC enabled anti-TIGITs, and you wouldn't know this until you get the empirical data, but they do deplete Tregs. And so by definition, that's not going to be a good thing. If you deplete Tregs in the periphery, you're going to see enhanced immune AEs. But secondarily, dose is different. And in fact, I think it's telling. Bay gene is actually on the higher end of the spectrum for the FC enabled anti-TIGITs. And what you have to realize, it's sort of surprising when you think about, because one of the fundamental, it's not a shock that anti-TIGIT is turning out to have the safe mechanism that we see because it's a tumor specific mechanism where you have that high CD155 is just being way over expressed as a survival advantage by the tumor. So people keep asking questions about TIGIT expression. The key thing is CD155 expression. If you want to draw an analogy, you know how everybody focuses on PD-L1? PD-L1 is the correlate of CD155. So it's not the PD-1 that's the driver, it's not TIGIT that's the driver. But that component of then the higher dose, particularly with an FC enabled anti-TIGIT, you're simply getting, don't forget that that depletion of T cells, that's an on target mechanism. So by going a higher dose, you would expect to see more. And in fact, Beijing is on the higher end of the spectrum. What's telling? Okay, so this is where we sort of look at these data and point people to the forest in addition to the trees. And so far it's not just this setting, but as you know, Merck is using co-formulation. And not surprisingly, that co-formulation is a lot about Pembrol. It's protecting Pembrol. And Merck explored 200 milligram dose of VBO. They explored a 700 milligram dose of VBO. And in the co-form, they're using 200 milligrams. And that just speaks to this point we were talking about that going up in dose is more likely to give you more on target AEs and so we think that's, it's as simple as that.

Thank you. I think what I'll point out is, you know, OC Prelimab, particularly the data that was just printed at CENTIS, it as though was in very different settings than anything that we're doing. You know, and they tended to be going after the second line setting. So their data was presented second line cervical, second line esophageal squamous cell carcinoma. So a different histology that we're pursuing with gastric esophageal. But like I said, going after the second line setting without chemo and tumors that are progressing very rapidly where you have a PFS of sort of two to four months. So not only is their antibody different when you look at a superior amount, but their development plan and the tumor types that they've been pursuing and settings have been very different than ours.

Our next question comes from Lee Watzak from Cantor Fitzgerald. Please go ahead.

Hi, this is Rosemary on for Lee. Thanks for taking our question. So for STAR-221, hypothetically, if you were to hit the survival endpoint for the PD-L1 high patients, but not for all comers, do you think you still have a chance to get approved in that patient subset given that the PD-L1 status is a stratification factor?

We don't just think that, we know that. The study is set up for exactly that it was designed to detect both. And that one, Demetri was talking about the co-primary endpoints. Both of those are opportunities for the way the trial is designed and both opportunities for approval. And we think it's a very meaningful opportunity.

Thank you. Our next question is from Salveen Richter from Goldman Sachs. Please go ahead.

Hey, thanks, this is Matt on for Salveen. On the gastric TIGITdata, could you share anything on how median PFS is currently tracking? I know you're going to share more next year. And what do you view as a meaningful improvement over the seven to eight months of standard of care? And then could you share anything on how current standard of care breaks down between Keytruda and Nevo combos? And are most patients actually treated with an IO combo? Thanks.

That's a lot of questions. So when it comes to are most patients treated with an IO combo, it depends on where you are, what the reimbursement structure is, and what the experience of investigators is. There are plenty of investigators, especially in the US, that give Nevo-based regimens to everyone because you don't have to wait for the testing and there is a modest benefit for low expressing and there is, of course, significant benefit for high expressers and the toxicity profile is very manageable. There's definitely markets where reimbursement would come to be an issue. And in that case, typically reimbursement focuses on 5% and higher. When your question about the median PFS, we can't really say anything about it. The median hasn't been reached. We need more follow-up to reach a median that requires a number of events for Kepler-Meier to calculate that so that there's nothing what we can say except that we emphasize the numbers we have right now look very, very promising. And obviously, they are significantly away from the medians that we think we should be reaching. When it comes to a meaningful improvement, that to me is actually a question about survival benefit in a phase three. If the PFS benefit would be in the same range, then let's say in hazard ratio of 0.75 starts to get interesting. For survival, probably a little bit higher. That's as much as I can say. When it comes to absolute differences in months, I think that's very hard to state on a single arm trial. I would be looking at the hazard ratios in the phase three setting. I think you had another question in there.

One other question you had was I think the mix of NEVO versus PEMBRO maybe in gastric. And today, NEVO is the only PD-1 that's approved. Mark has filed for approval with Ketruda for their study. Keynote 859 was conducted later. And I think their PDUFA dates in December. But our assumption is that NEVO will continue to have the vast majority of share in the PD-1 market for gastric. And that is the comparator for our phase three.

One other thing I just think is probably worth highlighting because it's something that like questions that somehow sometimes evaporate like FC versus non-FC, which for a year or two was the continuous question and overnight, somewhat evaporated. I think another one that's evaporating, but I'd like to address it is the ZIM-Ketruda comparison and I think this data set is another one that just the anti-PD-1 performance clearly looks good. It also was got another approval in China recently, but I think that's another example of ZIM behaving as you would expect an anti-PD-1 to behave.

Our last question for today comes from Jason Zemanski from Bank of America. Please go ahead.

Thanks. Hey guys, congrats on the progress this quarter and appreciate taking the question. So I have to ask during asthma, one of the big debates was obviously over how the treatment paradigm would evolve in gastric. And I think one of the big kind of takeaways here was that in any cloud and positive patient who was also CPS greater than or equal to one, the algorithm would likely move towards combining both agents to get that potential synergistic benefit. Now, obviously it's still very early, but I'm curious, how are you working to position TIGIT as the conversation starts to evolve? I mean, if KOLs are already thinking about this type of combination, what do you need to do to enter the discussion in favor of a TIGIT PD-1 combo, particularly as cloud and is biomarker driven?

Yeah, and I think, and we've spent a lot of time with KOLs talking about anti-clotting and we're working with the same KOLs as the anti-clotting companies are. So the first thing that I'd say about anti-clotting, if you look at the prevalence of high clotting 18.2 in patients, I know there was a number cited today, 40%, which was the percentage that was seen in spotlight, but there's literature out there that puts that number well below 20%. And my hunch is that if you go outside of Asia, the prevalence of high clotting 18.2 is lower. The other thing that's interesting, if you look at literature, is there seems to be more of a prevalence of high clotting 18.2 in patients that are PD-L1 low for some reason. And we've heard statistics like only 10% of patients that are CPS high are also high clotting 18.2 expressors. So even just looking at that piece, they may not be viewed as directly competitive. I'd say on top of that, when people look at anti- TIGITversus anti-clotting, we're obviously much less toxic. So if you look particularly at the Estella's antibody, which is the most advanced antibody in clinical development, there's a lot of GI toxicity in the plenary presentation today. They obviously talked about some of the other toxicities that they're seeing with the ADC, so that's something else to keep in mind. And then the third piece is clotting 18.2 testing does not happen today. And as we've talked to the KOLs, they think it's going to take up to five years or more for that to roll out. There's a lot of sites that don't even test for PD-L1 today, and because PD-L1s are relatively safe agents, they just feel like, you know, I'm not going to test, I want to put my patient right on treatment, and so I'm just going to give them PD-L1 antibody anyway. So we do think the need to roll out testing is going to be another impediment to the uptake of anti-clotting. So we're definitely keeping a very close eye on the class, but at least today, I think we feel very, very good about our chances to compete with anti-clotting, and maybe, you know, way down the road, there's a potential to combine PD-L1, TIGIT, and anti-clotting as well, anti-clotting ADC.

Yeah, I do want to add to that, right? So people typically talk about toxicity in grades, but I would encourage you to look up what it actually means. So the clotting antibodies, they have at least 10 or 15% grade three nausea and vomiting. If you hit that as a patient, that means you're hospitalized, you're on tube feeding, you get parental feeding. It's a really, really big deal. We talk about grade one and two toxicities that are obviously a burden on patients, but grade three nausea and vomiting is a really, really big deal. So I think that's another uphill battle for the clotting field. Obviously, the efficacy data is absolutely promising, but to the point that Jen made, I think it's going to be a small subset of patients that's funneled into the clotting high, PD-L1 high population and then honestly ask yourself the question, like, could you deal with grade three nausea and vomiting? I think that is incredibly, incredibly bad toxicity for patients that are in grade three.

That concludes today's conference call. Thank you everyone for joining, You may now disconnect.