My name is Daina Graybosch. I'm a senior analyst here at Leerink Partners, and I'm excited to welcome Terry and Jen from Arcus to have a 30-minute conversation, which is nearly impossible for the three of us.
Nope.
We'll see what we can do. You guys have multiple clinical assets in phase three, and so we'll try to tick through a few of them, starting with casdatifan, your HIF-2α inhibitor. Maybe before we do that, I'll just turn it over to Terry just to introduce yourself and just a couple of introductory words about Arcus.
Yeah, Terry Rosen, CEO and co-founder of Arcus. I'll be brief since, as Dana said, it's hard for us to squeeze in 30 minutes and we'd like to maximize the questions. We have two what we consider very near-term value drivers: casdatifan, our HIF-2 inhibitor. We'll talk a lot about that today, which is getting close to phase three. We have an anti-TIGIT program that's quite evolved. We have three phase three studies, two of which will be reading out in the relative near term, one in the very near term, upper GI cancers that should be reading out sometime in 2026. We have a full portfolio of programs, including our CD73 inhibitor, which has just started a phase three study in pancreatic cancer. We hardly talk about it, but it's actually a very exciting program. Arcus is sort of a full company.
We have a very strong discovery group. What we do is sustainable, and we're well capitalized with over $1 billion. All of these programs that we're talking about, they're less theoretical in terms of what we're doing with them, but actually we're operationalizing all these programs and advancing them and pushing them towards commercialization.
Awesome. Maybe let's start with your HIF-2α inhibitor, casdatifan, because we're in the thick of early data, which is an exciting time. Recently, Gilead decided not to opt in to co-develop casdatifan. Can you just tell us why you're excited about keeping full ownership, even though they decided not to opt in?
Yeah, it's truly a transformational event for Arcus. There's a number of reasons to be very excited about having all of the rights to casdatifan. Probably the first is the simplest. It's the most valuable asset we have. The reason it's the most valuable isn't just the market, but it's because of the probability of technical success. Between an approved drug that hits this target, belzutifan, and our early data, which shows a clear differentiation from belzutifan, it's about as close to 100% probability of technical success that you could have at this stage. The second thing is it's going to be a two-horse race. This is a very difficult target to get a good small molecule. It's going to be us and Merck.
It's a very large market, and it's a market that's going to expand because it's going to be driven by duration of therapy. This is a great mechanism. The final piece is having this 100%, and it's actually two pieces. On one hand, it's going to enable us to be much more efficient in its development. We're about to start our first phase three study with this, but we have a very broad plan. The second component of that, it gives us a lot more strategic optionality where we own this 100% versus sharing 50/50 with Gilead, which has been one of the things about all of our programs. We like to say we'd rather have 50% of a lot than 100% of zero. This one we feel is 100% of a lot.
One thing that I would just add to that is we did the clinical collaboration with AstraZeneca. To the extent anybody had concerns when the opt-in did not happen, did Gilead see anything? AstraZeneca has seen all of our data. They are extremely excited about our clinical collaboration, combining casdatifan with volrustomig. They are putting a lot of resources against that program. For competitive reasons, we have not said a lot about what we are doing, but I think about mid-year we will be able to say a lot more. There will be just a lot more information out there about exactly what we are doing and what exactly AstraZeneca is bringing to the collaboration.
I think the biggest criticism is that on differentiation, you said it's clear differentiated, and I happen to agree, but that will this differentiation translate from ORR into a survival difference, or may that be sort of modest and indirect comparison? Can you win because Merck is already on the market, so they're several years ahead based on this indirect survival comparison?
Sure. Let me address two components. First, just the differentiation, and then the when will we win, not if we will win, component of your question. On the differentiation, I think a lot of times, and it's probably more investor than investigator driven, there's always, no matter what you have, there's more, I'll call it sometimes a narrative versus reality. The data, I think, are unequivocal. If you look and compare every endpoint that you can look at, ORR were about 50% better than what you see with belzutifan. That's in three different cohorts and with a more advanced patient population. The rate of primary progression, which is a big liability of belzutifan, it's about 34%. So 34% of the patients in that LITESPARK-005 setting, 34, almost a third, worse than ever alignments. The control don't make it to pass the first scan.
That's one of the biggest Achilles heels that investigators point out before we even had our data. We're about half that. We've demonstrated that in three separate cohorts. We're somewhere in the high teens, 15%-less than 20%. All those patients that make it past that first scan then contribute to median PFS. We should think of it the other way around. It's not only that they're contributing to meaningful PFS, but they're benefiting from the drug. Whether they're stable disease patients or responders, we're likely to end up with a PFS that's more than double that of belzutifan. I think unequivocally we're doing better than belzutifan. I think that's also demonstrated by certain pharmacokinetic data. Maybe we'll talk about that, or pharmacodynamic data. Maybe we'll talk about that.
The point about beating them also comes down to not only the molecule, but the combinations and the settings. Really, Merck is ahead. They're approved in late line, and it's a great proof of concept, but they're not in earlier lines. We're going into earlier lines, and we have a very differentiated development strategy. We can talk about that maybe as we go along, but I think that's the place where we'll also win is that we have a development strategy that leverages the advantages of our molecule.
Is that double PFS spitballing? Are you guys modeled it? That is.
No, so.
That's simulated. You could hit that.
No, so as we sit right now, we have one cohort that had a median PFS of 9.7 months versus belzutifan's 5.6 months. We have another cohort that isn't quite as mature where we've actually taken and combined those two cohorts together to get to 60 patients. If you look at that PFS now, it would come in over 13 months. If you look at our 100 milligram cohort, where we've had more early responders, you look at the spider plots, it's clear that it's going to end up in double-digit PFS. My guess is that the number actually comes up more than 2x that of belzutifan by the time it's done.
Yeah, and just I know you're going to get to KeyMaker U03B, but that was another very recent belzutifan data set. It was a phase one two study. To the extent people have any concerns about us comparing our data to LITESPARK-005, which was a phase three study. In that phase one two study, as you know, belzutifan showed almost exactly the same numbers, if not maybe a little bit worse than what they showed in LITESPARK-005. That was a 19% ORR and a PFS of 5.4 months. Even more contemporary data set, very similar patient population to what we're enrolling in ARC-20. Again, numbers that are well below what we've been showing in our.
The interesting thing, actually, speaking of the patient population, is that if you look at the demographics in all three of the cohorts that I was discussing when I was synthesizing those numbers, we have about 25% of our patients that would not have even been eligible for LITESPARK-005 in a more advanced patient population.
Got it. Maybe the most important question for anybody following you closely is we're going to have the first outcomes this year from your casdatifan plus cabozantinib combination from ARC-20 from proof of concept because that's the first combination that you're taking into phase three. You have said that you have submitted for an upcoming conference. I wonder if you could talk about, given the size and the maturity of the data, what bar in outcomes do you think supports your differentiation thesis?
Sure. I think the best way to think about it at this point, if you look at Cabo data, there are a number of studies. They are all somewhat different. The reason they are somewhat different tends to be the patient demographics. Interestingly enough, there is a Cabo Bells data set out there where you see an ORR of just over 30% and a median PFS, I think, on the order of 13 months or so. When you look at our data, I think the way you really want to think about it is our data are going to look better than Cabo alone, and they are going to look better than CAST alone. Keep in mind that this is a very early data set. This study was done primarily to support the safety of the combination.
We will have enough efficacy data that I think you're going to be able to see quite clearly that it's looking better than either of those individual components.
Why isn't the bar Merck's initial combo of belzutifan plus lenvatinib?
Yeah, just those are two different agents. The fact that they're combining with lenvatinib and what they did see was a bit of a higher ORR, but they also saw a lower PFS, which is pretty consistent with what we've seen in Lenva in different studies. Lenva is a more toxic agent than Cabo. We believe that you can probably juice up the ORR by giving the higher doses of Lenva. It's available in like six or seven different dosages, but then it comes back to bite you a little bit on PFS. If you look at Bel's Lenva versus Bel's Cabo, there's about a two and a half month difference in PFS that we're seeing in those different combinations. Bel's Cabo had a PFS of about 13 and a half.
Bell's Lenva in that KeyMaker study was, I think, a little under 11 and a half.
You know, one thing I think it's a good place to frame what we were talking about earlier in terms of the development strategy. I think it's interesting because we kind of get questions on both ends of this. I think if you put it together, it comes together well. Lenva, clearly, if you talk to the investigator community, it's probably the most disfavored TKI out there. In fact, Merck chose because they own the rights to Lenva. It's a self-inflicted decision. They ran their initial phase three study that's ongoing. It's Lenva, Bell's versus Cabo. In that investigator community, Cabo is just dramatically preferred over Lenva for the reasons that Jen was talking about. It's not as well understood, the best way to dose it. It's more toxic. There's more patients having to discontinue, et cetera, et cetera.
We often get asked the question, well, Merck is talking about doing studies with Zanza. Now, the two things tell you something, and people shouldn't have cognitive dissonance about this. The fact that they're starting something with Zanza sort of negates the thing, oh, you're so far behind Merck. If they're going to be doing something with Zanza, we may even be ahead of them because we're going to be starting our Cabo combination in the first half of this year. We're on full speed ahead to do that. There's strong tailwinds. The investigators love it, et cetera. Zanza is clearly, again, a molecule that is very little known about in the RCC context. Clearly, Merck, if they're moving away from Zanza, sees the same thing that everybody else sees.
If they're going to another TKI, it's telling you that Zanza is not the preferred combination.
You mean Lenva.
Lenvatinib is not the preferred combination partner.
Let's talk more about development strategy. First, we've been talking about the TKI, but there's a couple of other elements of your development strategy. There's where you're going to do single agent, and then there's your frontline combination and partnership of AstraZeneca. Let's talk about that one first because you're going to do a proof of concept study combining with their CTLA-4 bispecific, Volrustomig. And we have some experience of trying to combine with IPI CTLA-4 on the frontline. COSMIC-313, which was Cabo Nivo IPI versus Nivo IPI, was pretty disappointing. It had a PFS benefit, but nothing on OS. Why is this strategy to go with Volrustomig and CAS going to be different?
Yeah, I mean, you actually got to exactly why AstraZeneca was interested in that combination. The COSMIC study and the overall survival data was just presented at ASCOG. I know you saw that. It was interesting to discuss in presentation what they really honed in on was the toxicity of those two mechanisms in combination. That is what they were getting to as to why they did not hit on overall survival. AstraZeneca actually did think about it, and I think actually did combine Volrustomig with the TKI and also saw a lot of toxicity. That is exactly why they were very interested in combining with the HIF-2α inhibitor and specifically casdatifan because casdatifan does have that lower incidence of primary progressive disease. PD-1 CTLA-4, if it works, it works very, very well, and it can be a cure for patients.
For some reason, 25% of patients do not respond to Ipi Nivo and just progress right through therapy. One of the really appealing things about CASD for AstraZeneca was we do have this very low incidence of primary progressive disease relative to belzutifan. It made it, we think, the ideal HIF-2α inhibitor to be combining with. As I said, PD-1 CTLA-4 plus a TKI is just too toxic, especially in that first-line setting. That is exactly why PD-1 CTLA-4 plus HIF-2α inhibitor should make a lot of sense because the toxicity profile is so much more benign for HIF-2α versus TKI.
You know, Dana, I think the thing that this also illustrates is what probably will end up being a big shift for those who aren't overly familiar with the clear cell RCC space. The thing about HIF-2α inhibition is it's an extraordinarily safe mechanism for an anti-cancer agent. The only AEs really are on mechanism in their anemia, which is very well managed, as well as occasionally hypoxia, which is also well managed. What you're going to see is a shift. We think every single patient that gets clear cell RCC in the life cycle of their treatment is going to get a HIF-2α inhibitor. This is the part where the Merck molecule demonstrated great proof of concept in late line and is doing quite well there.
We think that what's going to happen is HIF-2α inhibition is going to move up in the line of treatment. What you're going to see is that physicians would like to hold off as long as possible before they have to give the TKI for just what Jen was saying because of the toxicity. The HIF-2α inhibition really looks good for moving into the earlier line of therapy. What I'll contrast belzutifan and casdatifan is that in going into that early line of therapy, that limitation of that high rate of primary progression is something that physicians hate to be giving early line patients. They really don't want to do that. You have a third of the patients that don't even make it to a scan. That's where we see another advantage as you move this into earlier lines of therapy.
The point that I would also make that I think is really critical is we spend a lot of time, and I think it's going to start to shift as the year goes along and we put out more and more data, that it's not only that casdatifan looks better than belzutifan. Casdatifan looks better than TKI monotherapy. The idea of moving casdatifan earlier in the treatment paradigm is well supported by its profile to date.
Can you talk about your strategy to have casdatifan approved as a single agent and how that's different?
Yeah, so we haven't said anything about a registrational strategy for casdatifan mono, but we have some different ideas. We also just added some new arms to ARC-20 that's looking at casdatifan mono. There is one arm that's evaluating casdatifan mono, or sorry, casdatifan plus PD-1, so not casdatifan mono, but casdatifan plus PD-1 first line all comers. There is a casdatifan mono arm that's evaluating favorable risk patients. That is about a third of the first line patient population. There is another casdatifan mono arm that's looking at patients that have had one prior line of therapy. To Terry's point, that is the arm that's really looking at, can we displace TKIs because we're evaluating specifically in the setting where you'd expect to be using a TKI monotherapy.
We'll look at that data and continue to think about what kind of CAST monotherapy strategy might make sense. At a minimum, we think these cohorts are going to continue to expand upon them, could support an NCCN guideline inclusion. We're already hearing anecdotes about belzutifan moving up, interestingly, which speaks a little bit to Terry's point. If you talk to MD Anderson, kind of very consistent with what Terry was saying, they start basically every patient that doesn't get PD-1 CTLA-4 on a TKI. Instead of going to another TKI after, they really want to give patients a TKI break. They are using belzutifan in second line, even though I think most places are probably preserving belzutifan for third line. That's the type of setting that we would love to be able to evaluate CAST mono in.
That's exactly what we're looking at in ARC-20 with that new cohort.
Did you ever consider adding a CAST arm to PEAK-1?
It's funny that someone else brought that up, and we have thought about it a little bit. I think it's probably too late at this point. Really, the primary objective of PEAK-1 is to get it enrolled as quickly as possible. If you add another arm, it's just more patients and more time. We think right now we have a really well-designed study that's teed up to enroll very, very quickly. We just want to get it going and get it enrolled. We are thinking a lot about what a PEAK-1, I'm sorry, what a casdatifan mono strategy might look like.
I want to spend a few minutes on your other programs in phase three. You have domvanalimab, Dom, say Dom, your TIGIT antagonist. I guess the pressing question here is, why should any investor believe in TIGIT after Merck and Roche's failures? I know there's an Fc difference, but a lot of investors are like, the simplest explanation is TIGIT just doesn't work that well.
Yeah, I think that investor thesis and not trying to kidnap anybody into believing something they don't want to believe. But if you're a data-driven person, what you have to recognize is that Fc difference in this particular case is not some subtle difference. In one case, you get dramatic, and it is, it's probably above 80% depletion of peripheral Tregs with an Fc-enabled anti-TIGIT like Merck had with Vibo, Velro, Vibo. And what that leads to is immune AEs. If you deplete 80% of your Tregs, you're going to see immune AEs. In Merck's case, they were being cute. They did a co-formulation to extend the life of Keytruda, and they put it in the same bag. Patients have to come off of both drugs.
They end up seeing a lot of immune AEs in the study arm, and that leads to withdrawals and discontinuations and loss of efficacy. On the other hand, and by the way, just yesterday, if you want to look at proof of concept between us and AstraZeneca, AstraZeneca has a bispecific anti-TIGIT, anti-PD-1, where the anti-TIGIT is also Fc silent. These Fc silent anti-TIGITs are extraordinarily clean. Everything that's going on is going on in the tumor. They just started their eighth registrational study yesterday, and they started one in high PD-L1 non-small cell lung. Essentially what we were doing in ARC-10, which we kind of gave them the proof of concept data there. You have between us and AstraZeneca, we've run five studies. They've all been positive. Our ARC-10 data is a randomized data set. It's in high PD-L1 non-small cell lung.
It has a hazard ratio of 0.64 for OS. That's with two years of follow-up with anti-PD-1 having an OS of over two years. Our upper GI gastric study, we showed a median PFS in the phase two correlate that's called EDGE Gastric of over 12 months, where the standard of care Nivo chemo or also Keytruda chemo or Tisle chemo show about seven months. We'll be sharing OS from that same study later this year. We have publicly said that at 18 months of follow-up, over 50% of the patients were still on study. Keep in mind the standard of care there has an OS of about 12-13 months. Our PFS was in that range. In that study, the phase three correlate, which is called STAR-221, was fully enrolled as of June of last year with 1,047 patients.
You can do the math, the standard of care at 12 months. We will be out at 18 months since the last patient in by the end of this year. There is a ton of data in the anti-TIGIT field. What you should be thinking, if you want to get convinced, and you want to get convinced in time, is that in this particular field, Fc enabled versus Fc silent, you should almost think of them as a different mechanism. In one case, you have got a blocking antibody where the biology totally supports the role of anti-TIGIT as a checkpoint, enhancing the activity of anti-PD-1. In the case of the Fc enabled anti-TIGITs, you have something that is going around and basically chewing up cells that have TIGIT on them. That is a dramatically different mechanism that leads to dramatically different outcomes.
In the last three minutes, let's talk about CD73 and adenosine perhaps. Just a quick question. You said you started to enroll your phase three in pancreatic with quemliclustat, your CD73. I know that those could enroll and read out pretty rapidly. Sort of what's the timing on that? Also, just that's another one that you sort of wholly own, although Gilead used to be partnered, and then you're taking that phase three yourself. I just wonder if you could help us with why you're so confident there.
Yeah, so our confidence stems from the study we ran called ARC-8, where we saw an OS of on the order of 16 months, pretty unprecedented in pancreatic cancer. We also had a synthetic control arm that was well matched that showed that advantage not only in it showed in basically every demographic we looked at in subpopulation. So it's a great data set, well matched synthetic control. And to your point, while we haven't given any guidance yet, you are absolutely right in terms of the enrollment. Even though it's only been running for a month, we feel like this is one that could probably even be fully enrolled by the end of this year. It's obviously a big extrapolation. I wouldn't give that as guidance. But given how fast it's already enrolling and we hit the ground running, investigator enthusiasm is enormous.
There's been an IST that's sort of looking in even earlier line that has very positive data. The surgeon's been very positive. It is looking extraordinarily promising. I don't know, would you add anything to that, Jen?
Yeah, no, just the study will enroll very, very quickly. I know RevMed was making comments yesterday or two days ago about their study and how quickly it's going to roll. I think ours, based on what we're seeing so far, will be very, very similar. We'll probably enroll well ahead of schedule. That is data that will probably now come into next year. Just to your point, unfortunately, OS is not long in that patient population. Given how fast we expect the study to now enroll and just the fact that OS is not particularly long, that would be a readout that would happen next year.
To Terry's point, and I think you were the one that probably first saw the presentation at ACRIO, we are doing a very interesting IST with Zev Waenberg at UCLA, where he is looking at quemliclustat, plus sim, plus chemo in the neoadjuvant setting. The data has not been publicly presented, but when he compared that data to what he has seen historically in that patient population, it was very, very interesting. He's now going to expand that study with a control arm. You saw some of the biomarker data that was generated from that study. We just think of that as more evidence that quemliclustat is definitely doing something pretty profound in this disease.
I'll say in our remaining minus 14 seconds that the one other piece that enhances confidence is we do have biomarker data from ARC8, which was the progenitor study to the registrational study. In fact, what's nice about the biomarker data is you have a couple of markers that are actually negative prognostics. Those were the patient that did better in the study. The patients that would have been expected to perform more poorly actually did better on treatment. Both of those markers that we are talking about are correlative with enhanced adenosine activity.
It's on target.
Correct.
Negative prognostic, and you flip it.
Right.
Which is interesting.
Yeah, that's pretty positive when you have those sorts of data.
Awesome. Thank you. We got that.
Thank you.
I appreciate it.
Thanks.