Arcus Biosciences, Inc. (RCUS) Earnings Call Transcript & Summary

Great. Welcome, everyone. I'm Yigal Nochomovitz, one of the biotech analysts here at Citi. Welcome to the Arcus Biosciences fireside chat, and it's my pleasure to have with me several distinguished guests. First, the CEO, Terry Rosen; second, the Chief Operating Officer, Jennifer Jarrett; and the President of Research, Juan Jaen. So welcome all of you. Thank you very much for your participation.

And Terry, maybe just to kick things off and level set for those a little bit less familiar with the Arcus story. Maybe you could just walk us through the broad strategy of the company at a high level and what's the near-term mission of the company and what you see as the more longer-term mission.

Sure. Thanks, Yigal. And thank you, everybody, for joining. We appreciate the interest. So I'll start with that high-level overview. The strategy of Arcus, which has been around since 2015, at the highest level, we look to create highly potent and specific small molecules for very well-characterized targets, but targets for which really they haven't been translated into good drugs. A second pillar of that strategy is then to access what we would call backbone antibodies, like anti-TIGIT, like anti-PD-1, where we can combine with those small molecules, which is really the secret sauce to create highly differentiated intra-portfolio combinations. I'd say the third component of our strategy is Arcus is looking to build a long-term independent biopharmaceutical company. So with that in mind, we look to have a broad portfolio that's targeting a fairly diverse set of mechanisms. We have 5 molecules in the clinic already despite a short time of existence and soon to be 6. If you look at our clinical programs, they target some of the most prevalent cancers, just a few examples, non-small cell lung cancer, colorectal cancer, prostate cancer, pancreatic cancer. And finally, given the breadth of what we're doing, mechanisms that we target tend to be very ubiquitous in nature. We sought to augment our independent activities with very strong, and I would say, strategic partnerships. So they also give us somewhat of a global footprint. We have a very major collaboration with Gilead. We have collaboration with Taiho. We have one with AstraZeneca. We have other relationships. And we view those as providing us the opportunity to accelerate our pathway towards becoming that fully integrated biopharmaceutical company, but doing it in a very capital-efficient way in a way that mitigates risk, particularly in our early days before we have revenues.

Okay. Perfect. Thanks for that overview. So the last time I checked, I believe you have 6 -- or about 6 randomized trials ongoing, including a pivotal Phase III called ARC-10 in first-line non-small cell lung cancer. So it would be super helpful, as a general starting point, if you could highlight which of these trials are mission-critical, and which are important but perhaps play more of a supportive role in the investment thesis.

No, that's an awesome question, Yigal. And so as you would imagine, none of these are throwing spaghetti up on the wall. So they're all important, but in a different way. And I would actually say that we're really in a nexus in our field for both our trials in patients and for the future. So I think one way to answer your question before we get granular is to think about what we're doing and perhaps put our programs into 3 different buckets. So the first one I would call the checkpoint inhibitor. So in addition to our anti-PD-1, which was just recently approved in China, we have the anti-TIGIT program. So we have 2 molecules there. And what I would call that component of our portfolio is the enablers. So they're essential, but if you're going to be a long-term player in oncology, you need those backbone of therapies. The second bucket I would talk about is our adenosine modulator. So things like AB928, our A2 receptor antagonist, and AB680, our CD73 inhibitor. And so those we really think of as the differentiator. So when I was talking about that notion of creating those intra-portfolio combinations, that's what really distinguishes us from the rest of the world. And then we have the earlier pipeline, which would be an example would be AB521, our HIF-2 inhibitor, which is going to go into clinic later this year. So when you think about that now and you move from the molecules to the trials, this will give you a sense of where these all fit in our ecosystem. So for anti-TIGIT, our lead molecule is AB154. It's an extremely important program. We recently presented interim analysis data from our non-small cell lung cancer trial with that. We showed interesting activity for both the doublet and then with that differentiated triplet that included AB928. Our other anti-TIGIT program called AB308, has also moved very well through its early clinical development. In fact, we got through our dose escalation in combination with our anti-PD-1 such that, that doublet is already in expansion trials and so interestingly, leapfrogged a number of other programs. What I would now talk about is like the adenosine program. So that field is getting more and more validated every day. Before I even talk about our programs, I think it's interesting to note, AstraZeneca is an awesome collaborator of ours, is also a competitor. And they'll be sharing data. They have a late breaker at ESMO on their CD73 antibody. It's a study called Coast. It's in Stage 3 lung. We think that might be a very interesting data set. In terms of our own program in that field, which we feel is like extraordinarily interesting. We presented data from what we call ARC-8. That's our study in frontline pancreatic cancer in January. Saw a very interesting response. We've been able to very aggressively enroll an expansion cohort, a lot of investigator enthusiasm. They have enrolled and gone well throughout the year. We'll be sharing data from that study later this fall. Not only are we interested in response rate there, but also it will be the first opportunity for us to start to talk about durability. I'd also say that's a very important program for the field for us. And so that study will certainly just be the tip of the iceberg. But I also think about adenosine, then I would say our AB928 program. I talked about ARC-7, where it's the key differentiator for that triplet arm where we have a few ongoing randomized studies. One that we've been talking about a lot, our prostate trial called ARC-6. We have a study ongoing in EGF receptor mutant lung cancer. That will be a data set that we disclose at the conference in the middle of next year. And finally, we have ongoing there a study in colorectal cancer called ARC-9. It's a platform study in later lines. So I would just finally say that one of the interesting things that if you think about, clearly, as this field is all coming together, and I was talking about that inflection, there's a nexus between the anti-TIGIT, the anti-PD-1 and these adenosine fields. And interestingly, whether large or small company, a lot of companies have multiple pieces. But we're -- I think we're the only ones we have all of the components of that puzzle.

Okay. Perfect. And then just one more high-level question before we dive into some details for each of your programs. You did mention the 3 partnerships with Gilead, Taiho and AstraZeneca. How are they different? How are they the same? And how does each advance your goals at Arcus in terms of your overall corporate objectives?

Yes. They really -- they were all done with a strong strategic intent. I would look at the Gilead clearly is the big anchor there. We did the Gilead collaboration to facilitate our growth as an independent company. It's a 10-year all-in collaboration. It really -- when you think about all the programs I just got done describing, these are all very ubiquitous mechanisms, and they all have competition that looks like a Merck, it looks like a Genentech, it looks like an AstraZeneca. So working together with Gilead will enable us to essentially compete, in parallel, in these very large fields and not have to do it as a -- just as a smaller company. Our collaboration with Taiho gives us access as you kind of start to think about a global presence. That gives us access to Japan. They're a big player in oncology in Japan. And interestingly, for example, they're starting to look at our anti-PD-1 in combination with their portfolio molecules. Now the AstraZeneca relationship, I think, is prototypical, something that you'll see a lot more from us. So when you think about it, as I was describing us having pieces of puzzle that could bring value to other portfolios, also bring value to us. So while we can focus on the core studies that we want to do, we would expect to do more collaborations like that where we can actually combine with others where they either have a standard of care that's quite relevant to us or they have an expertise at executing in a certain setting that's relevant to us. So we view those types of collaborations as being able to expedite and expand our opportunities. And then the final piece I'd point out is a company called Gloria in China. And while we don't have a formal collaboration with them, Gloria is the company that developed zimberelimab in China, just recently got approval for zimberelimab, and we're talking with them about ways that we may use some of their data sets to facilitate some of our regulatory pathways as well. So I would say the totality, collaborations and corporate development is a big part of what we've done, and it will continue to be going forward.

Okay. Perfect. Well, let's talk in some more specific terms, if we could, on ARC-7 and ARC-10 in lung cancer. So just in terms of ARC-7, obviously, you can't disclose the numbers quite yet. But in broad terms, what are you able to say as far as what Gilead is waiting to see from the ARC-7 data before they can make a call, the opt-in for the TIGIT and for etruma?

So Jen, do you want to talk a little bit about both of those and the Gilead relationship and how that will all play out?

Sure. So as they have said and we said, contractually, they have the ability to see more patients of data, more mature data. So that's what they're waiting for. We will do another data cut just a little bit later in the year. And so the purpose of that data cut is to help inform their opt-in decision. So it's really just more patients, more mature data, the way the study has been enrolling since enrollment has accelerated over time. It's like literally every week that passes. In a way, there's almost like exponentially more amount of data, particularly if you think about the maturity of the data, so it's not just linear. So we hope to be able to present exactly the package that they're looking for.

And at what point would you be able to tell us when to expect the numerical ORR data for each arm of ARC-7? When might we know when that will be presented?

Yes. So right now, our plan is to present that at a medical conference. There's a medical conference that's late this year that we're looking at and then there's 2 next year. We obviously haven't submitted the abstract yet, so we're still thinking through which conference we want to target, and there's present cons of each. But it will be late this year or sometime in the first half of next year.

And just so I understand, I want to make sure I understand the Gilead agreement properly. Is it separate opt-ins for TIGIT and etruma, or are they linked?

Those are separate opt-ins. So the dom and AB308 are linked. So if they opt in to dom, they would get both TIGIT antibodies, but etruma is separate. So yes, that is one of the things that we're definitely very focused on for this next look at the data, is understanding the triplet arm better. I think we've been very transparent that what we saw at least in the first data cut on the triplet was super interesting. And this next data cut, we're definitely going to be interested how we're feeling better, how much of that effect was just dom versus how much was etruma contributing. And as you might imagine, I mean, the more that it seems to be etruma, maybe there's a potential that they would opt in to etruma earlier than we would expect to be able to start with the triplet right away.

And then I just want to make sure I understand correctly. There's no downside for Gilead to trigger the opt-in review period, except for the fact that once they do that, then the clock can't stop. Is that correct?

That's correct. The clock can't stop, so they can't stop the clock and then restart it at a future point in time. They get one bite big at the apple. And so we've always said that we feel like the decision to start the opt-in process is an important event, almost like a critical event because it's unlikely that they would start that process, unless they thought it was likely that they would exercise the opt-in. And it's probably a few months, they have a few months of time between the time that they start the clock and they can make the opt-in decision. And going back to your point on the data. One thing that we do think is going to be important, obviously, is Gilead's decision when they see the data. So we get a lot of questions about are we going to hear about the Gilead opt-in first, are we going to see the data first. At this point, it's not totally clear because it will depend on when we present the data, but there is a chance that we would hear something from Gilead on the opt-in before people see the actual data. And that would obviously give people a signal on the data set that we saw.

But then, okay, if we were to go that route, you would still preserve your strategy of presenting at a medical meeting or once -- if Gilead were in...

[indiscernible] our plate right now. We want to be able to use a medical meeting to really put the data in the spotlight, I think get the appropriate attention to it.

Yigal, I would say we really look at those as independent variables. We think this is going to be a very meaningful big deal data set. So we definitely want to do it at the right medical conference in the right way and have it be a very meaningful and important data set for the field.

And just -- I think I know the answer to this, but just to confirm. I mean, if Gilead were to opt in, they don't have the ability to disclose the numbers themselves. Is that correct?

We don't think they would. I mean everything should be held in confidence. I mean the idea is to try to save information for the conference presentation. So we don't have total control over what they do. But then I would think that our communications would be relatively consistent.

We're pretty close interacting as a team, I mean, that they wouldn't -- we -- I mean, to be honest, even everything we disclose all the time we share with each other. I would have no expectation they would share something that we wouldn't share.

Got it. Got it. That makes total sense. And then, Terry, I think on the call back in the summer when you disclosed the interim results for ARC-7, you made the point that ARC-7 could behave similarly to Roche's CITYSCAPE in terms of evolution of response rates. So I was wondering if you could just kind of unpack that statement again and help us understand what you were trying to get at there.

Sure. So the point there was just that mechanistically, we're talking about essentially the same biology. And in fact, everything we'd seen in the early data set looked similar to what you would expect when you're looking at those immunotherapy-type responses. And so to give you an example, 25% of the patients that achieved the response did so after a -- not until at least the third scan. So when you look at the shape of the curves, what you essentially see is that very typical, and we had certainly a large enough data set to see that responses coming in sometimes later, deepening with time and being quite prolonged. And in fact, when we looked at the interim analysis, the vast majority of the patients were still on study. So it looked very similar to what you might have anticipated. And if you went back, for example, and you looked at the Genentech sort of evolution in their data, for example, when they put out the press release about the CITYSCAPE data, their response rate at the time was 55%. By the time the medical conference came around, it was 66%. And if you just looked at the trajectory of even our curves, they look very similar to that. You have patients that were still deepening and prolonged. And so I think it's very likely, not just for us, but that's how the field looks with these. It's a very typical immunotherapy. Even when you move outside of the anti-TIGIT space, we do see that type of behavior. But we think it will end up behaving very similar to what Genentech saw.

Is that -- are you making that statement in applying that to the zim plus dom arm and also the triplet arm? Or are you mainly referring to the zim plus dom arm in that statement?

I would say it's -- with time, we may see some divergence, but it's definitely applicable to both the doublet and the triplet.

Okay. Got it. Got it. And then I had an ARC-10 question, which I've gotten from a bunch of investors over the summer. Just wondering why, for ARC-10, you decided not to do the triplet of zim plus dom plus etruma. And I'm just wondering whether you have second thoughts about not including the triplet, given your comments over the summer suggesting that the triplet in ARC-7 was looking quite good.

Yes. So really, we don't have second thoughts about that, and I'll explain how we think about all that. So the reason we went with the design that we did, first of all, that study has 2 purposes. It's also to get approval for the monotherapy, zimberelimab and then as well as the doublet with dom. So it's a hierarchical analysis that we'll do that. But insofar as the design itself, given that we had both the biology and then what we know from Phase I and what we knew about the behavior of our anti-TIGIT and what we knew about the field, so that, in fact, the CITYSCAPE data, in addition to our own, said this is a foot race, there's no reason we shouldn't get this going sooner rather than later. We had all the hypotheses about what the tripling might do, but we didn't have an empirical data set. So we certainly weren't just going to jump in and design a study around that. In fact, felt that what would come out of ARC-7 would inform the smartest possible design for potentially a triplet. Now interestingly enough is we were speculating, we thought we might not have a good idea as to whether the tripling was bringing something to the table until later on. We didn't know if it would just -- won't be related to response rate, to durability. We clearly saw something that looked interesting. As Jen said, we'll see if that continues to go with the same trajectory. But our view would be that we would need and want to design study and studies around the triplet prospectively, and those would be independent trial. So the choice for ARC-10 is that it was driven by data, both our own and the rest of the world. We didn't have an empirical data set on the triplet at that point. And so we didn't even think about including it. And in fact, now we would design a new study or probably even studies based upon what we're learning, and we'll continue to learn around the triplet itself.

And when is -- have you said when ARC-10 is going to read out?

No, we haven't. It's -- obviously, we started the trial early this year, and we haven't commented yet on when we expect that to read out.

Okay. Let's move on because you've got a lot of clinical trials to cover here in the next 20 minutes. So ARC-6 in prostate cancer. Love to get your thoughts as to how do you view the single-arm response rate data that you've got so far. I think 35% PSA response, 27% radiographic response. How competitive do you think that is? Or what do you think the benchmark is for being a -- having a competitive profile there?

Jen, why don't you go ahead on that, talk about ARC-6?

Yes. So we were quite excited about the data, our investigators were excited about the data. And I think even one of the analysts hosted a call with another investigator that we don't work with, but he was also very positive about what we presented. And if you benchmark that data that you just mentioned to what we would think are the appropriate comparators, if you look at docetaxel alone, typically, you see about a mid-20s PSA response and sort of a mid- to upper teens ORR response. So we certainly look a lot better on the docetaxel on its own. Another interesting data point to look at is the COSMIC-021 data. So this is Exelixis' cabozantinib plus atezo and basically the same patient population that we're looking at in ARC-6, which is patients with castrate-resistant prostate cancer that had failed at least 1 novel hormone therapy. They presented, I think, almost exact same resist ORR that we showed based on investigator assessment. They showed then a lower ORR if they look to the independent assessment, but their data set actually look very close to ours. And we actually think that our regimen should be better tolerated than theirs. So it was good to see that data set, and we feel very comparable to theirs. So right now, we're looking for this data set to hold up. As a reminder, the data that we presented was really the gate to the randomized portion. So when we do present the next data, it will be the randomized portion of the study. So we'll just be single arm, and we're trying to make comparisons to other data sets. We will actually have docetaxel as the comparator in that randomized data. So it will obviously tell us a lot. One of the things I do want to call out on this study, make sure that we point it out is there's a few different arms that we were looking at. One arm that hasn't gotten a lot of attention externally that we're excited about is looking at etruma plus AB680, which is our CD73 inhibitor, plus PD-1 and in late...

You read my mind. That's my next question.

So this is we're calling a chemo-free regimen in late-line patients. We've actually seen some really interesting early results there. So we just made the decision to expand that cohort to the second stage. So it will now enroll about 25 patients. So we actually could have that data set before we have the data from the randomized cohort that I was just mentioning earlier. But this is something we're excited about. We literally have a waiting list of patients that want to get on that study. Investigators have also been very excited about that regimen. So that's something you'll probably hear us talking more about in the future. And that is also our first data set looking at the etruma plus AB680 combinations.

That's really interesting. And I was very curious about that combination because you've combined etruma with AB680, as you said, in one of the cohorts of ARC-6 as well as one of the cohorts of the ARC-9 trial in metastatic colorectal cancer, as I understand it. So my question is what's stopping you from combining etruma and AB680 for other solid tumors? Is there some belief that prostate and mCRC is more susceptible to blockade of the adenosine access? Or is it just you haven't gotten there yet?

It's a combination of both. So we clearly -- when we started out, we recognized there's certain studies for -- prostate is a very unusual biology. So prostate has -- there's another enzyme called PAP that's very prevalent. In fact, it used to be used as a marker for prostate cancer just like PSAs. And it's a phosphatase and it's capable of hydrolyzing adenosine monophosphate to produce adenosine. So there's actually a CD73-independent mechanism in prostate to produce adenosine, hence the receptor blocker becomes very critical. With that said, and this is one of the reasons we're so excited about where this field is right now, once you've shown that adenosine is not only important biologically, but muting its effect has clinical manifestations, which clearly we're starting to see that, AstraZeneca is starting to see that, it definitely forces you to want to go in and at least look in settings where you know that adenosine is important to make sure that you squeeze all the -- I like to use the term you're squeezing all the water out of the stone. So one thing to think about in terms of adenosine is that flux of adenosine. So is that adenosine being produced in a setting or in an individual patient as if it's coming out of a fire hose or is it being produced as if it's coming out of a fire hydrant? And so when we think about this, a priority, you would just want to block. If you could block with perfect fidelity, it's formation, which, again, CD73 would be your first choice. But if anything was slipping through, you'd want to block at the receptor. So there's a case for being made for each. De novo, you probably want to just block CD73. But in situations where you may have other mechanisms [ and so you'd ] want to block the receptor. And then the final piece to your question is recognizing we started the company in 2015. We started from a blank piece of paper. And the practical reality is that we started working on both of those programs at the same time, and AB928 simply came first. So that's why you're starting to see us. Now the real -- the first study that we did with CD73 post the dose escalation study was in this pancreatic population. But we like to say that's the tip of the iceberg. And so now we're starting to explore it on its own, but also you can expect us to be combining it more. The colorectal cancer is a perfect example where we've already seen a signal from that ARC-3 in the late-line population. So we want to make sure we're using the platform to see if we might get any leverage out of attacking both of the targets.

Got it. Thank you, Terry. And I also wanted to ask about some of these additional cohorts in ARC-6 that do not include chemo, etruma plus zim, and then etruma plus zim plus AB680 and then etruma plus AB680. Is the thesis there that you believe you'd be able to eliminate the standard of care? Or could you just help better understand why you're running those cohorts?

Jen, do you want to comment on that? [ Or talk about the ARC-6? ]

[indiscernible] just to look at all the different lines of therapy that we can go into and just to go on top of the standard of care in that line of therapy, so that's really what the design of the studies and how we chose these different cohorts. For the XTANDI combination, we will probably deprioritize that for a few different reasons. One, it was the one that made the least biological sense. Second, there was just a lot less interest in that cohort than the other cohorts. There's some reimbursement decisions associated with it. Physicians like weren't totally excited about putting patients that had just failed abi on XTANDI, just given there isn't a lot of great data showing the patients that failed on abi responding to XTANDI. So based on a few different things, we'll probably deprioritize that one. And then like I said, we will prioritize the docetaxel cohort where we see really interesting data as well as the AB680 plus etruma plus zim cohort where we've also seen very interesting data.

So are we no longer going to see the etruma plus zim plus XTANDI data and that randomized part of that study? Or is it...

Yes. But we probably -- we haven't decided yet, to be honest. But most likely, we will just focus on the randomization of the docetaxel containing cohort, given just that being a bigger market opportunity, like a more exciting setting for us and generate a lot more investigator interest.

Okay. All right. Let's continue and move on marching through the pipeline. So on to ARC-8 in pancreatic, where you showed the data at the beginning of the year. I mean, from my perspective, you could have picked etruma as well. I'm just wondering why you decided to pick AB680 and blocking CD73 in the case of pancreatic. What's the biological rationale for that?

Sure. So pancreatic was, again, the first setting. So we picked what we thought might be a poster child. So if you look at the sort of history, we did a study in healthy volunteers, showed we covered the target. And then instead of going into heterogeneous population, we went into pancreatic because pancreatic is a super high CD72 tumor. It's super high KRAS. KRAS, by the way, is -- it can almost be viewed as a marker for CD23. So if you're [indiscernible], you're going to see high CD73. So it was a perfect sentinel indication. We obviously got a very exciting data set. It's one that we may ultimately look at the combination of the both, but it was a perfect choice as an initial setting. And as I said, that will be the tip of the iceberg as to how we use 680.

And I was also wondering, I believe that for AB680, you're developing both an oral and an IV. Whereas for the adenosine AB928, I believe you're only developing an oral. Is that correct? And if so, could you just help us understand why the difference there?

Sure. The difference, very typically for the small molecules, you would generally be going after an oral, particularly in settings like this where you want to dose on a few weekly basis. The more unique situation was when we were developing the series of molecules that led to AB680, we actually recognized they had an extraordinarily long half lives and then they were also extraordinarily potent. We haven't really talked about the molecule properties, but that molecule is a reversible inhibitor of CD73 that inhibits the enzyme with an IC50 of 5 picomolar. So between the potency and the half-life, we recognized there was an opportunity to develop that as an IV formulation, which from a convenience standpoint, was perfect that you could basically come and get it together with most of the therapies we thought would be in conjunction with such a therapy, like, for example, an anti-PD therapy, things like gem Abraxane. Now what we recognized along the way is it might be useful to have an oral formulation in the future, depending if we went into certain other combinations where -- or where you might want to give that oral as a maintenance therapy. For example, give a loading dose with IV and then let a patient take home the pills. But from now, we're very focused on the IV formulation for 680 and basically you have the oral formulation on the shelf if and when we need it. In the case of AB928, it's more like just a typical small molecule. Either you really wouldn't even be able to come up with something that would be able to give you the opportunity to dose it on a biweekly or triweekly basis.

Makes sense. And then I understand that there will be an update for ARC-8 later this half at some point. So could you just quickly review what we should expect data-wise for that? And then secondly, when would we get the randomized data for 680 plus zim plus gem Abraxane versus the backbone of Zim plus gem Abraxane?

Jen, do you want to answer those two?

Yes. So we are very excited about this update that will be in the fall. The update will include probably about 30 patients. So we'll be focusing on the patients of the [ 100 mg ] as well as the 125 mg dose. So those are our going-forward doses, as people probably know. And that data set will include ORR as well as probably early PFS data. So we should have 6 months of follow-up on everybody by then. So it's something we're super excited about. On your question on the randomized portion of the study. We started enrolling that a few months ago. We're actually way ahead of schedule just because of the high interest in that study. So we now expect the randomized portion to enroll by year-end. So we'll be presenting that data, I think, much sooner than we probably expected. Our guidance right now is that we would present that data around midyear next year. And just a quick correction on the randomized portion. What we're looking at is basically the quad versus the triplet. So we're looking at AB680 plus gem Abraxane, plus PD-1 versus AB680, plus gem Abraxane alone. And so really the intent of this randomized portion is to tease out whether or not PD-1 has any activity and whether or not we need to include the PD-1 in our Phase III, which we're planning for right now.

One other thing, Yigal, that I'd just note about that study because I think it's an important aspect. Given how things have been going, we've also opened up a cohort in second line where those patients really have nothing. The big part of that also has to do not only with what we've been seeing in the frontline population per se, but also the safety profile. So it's a really -- we're seeing that same pattern that we've seen in other studies where you're getting that AB680 biology without bringing any additional toxicity.

Okay. I just want to make sure I got this correct. So Jen, it's AB680 plus zim plus gem Abraxane versus AB680 plus gem Abraxane?

Exactly. So kind of like I said, just really determine whether or not PD-1 is adding anything. So as a reminder, prior studies looking at gem Abraxane plus PD-1, PD-1 has added nothing. So we'll see if maybe, for some reason, went to add AB680, the PD-1 does provide some activity or maybe it doesn't. And just AB680 gives you everything you need, so that's [indiscernible].

Okay. Perfect. Well, let's see if I can squeeze in a few more questions here before we run out of time. ARC-12 for the FcR-enabled version -- Fc-enabled version of your TIGIT AB308. So now I think you make the argument, Terry and team, that it's -- this is best suited for hematologic malignancies because you can deplete the TIGIT bearing cancer cells such as in myeloma and non-Hodgkin's lymphoma. So I was curious as to why ARC-12 is also enrolling solid tumors in the dose escalation and dose expansion cohorts, given the statement around being suited for hematologic malignancy.

Sure, that's a great question. And actually there's multiple reasons. The first one is it simply allows you to enroll more and quicker, and so we'll continue to build up the safety database for that. Second piece is that those expansion cohorts actually will also serve as signal-seeking. It may encourage us to go into a direction with 154 that we might not have gone otherwise from something that we might see in that segment of the study. And the final piece, which I think is something that is less recognized, but we've certainly realized, is that's a valuable asset. And from a corporate development standpoint, it may be even something that, at some point, we all licensed. We think of anti-TIGIT as something that, in the end, as important as it may be that anyone who wants one is going to be able to get one. We'll have one that's advanced. We'll have one with a great database. And when you think about competition, so things like AB680 and AB928, remember, I call them differentiators, it's going to be very difficult for someone to get a molecule, if even possible, that is exactly as those properties. Anybody who wants an anti-TIGIT, just like anybody who wants an anti-PD-1, is ultimately going to be able to get one. So if the time value of having a well-tested molecule is -- if there are others out there who might want one, we would certainly want to be -- not just throw that on the shelf, but we think it may have value in the future. So building up a data set for it will be value from that perspective as well.

Got it. Got it. And then I had a question on ARC-4. So this is your trial in EGFR mutant non-small cell lung cancer, and you made the decision to do the combo of etruma plus zim plus chemo. Curious on the thought process there, why etruma and not the CD73 block or AB680 in that one?

Yes. Honestly, there would be -- that's purely a timing. It's not that necessarily is a better rationale for one or the other, it's just what we're talking about before. That trial was started before we could have even gone forward with 680. We do think it's an exciting trial. We think that there's interesting data that AstraZeneca generated with their CD73 antibody, together with OC. But we think that tumor setting, it's a high CD73. There's increased levels of CD73 associated with that EGF receptor mutation. So we think it's a very smart trial to run. It's one that, well, as Jen said, we'll be talking about a randomized data set from that. We'll have PFS by the middle of next year. But it's one that we would also consider looking at with 680 and/or a combination of both.

Okay. And just in the final few minutes here. I think just I want touch on some of the other programs that are earlier stage. You have a HIF-2 alpha inhibitor, an AXL inhibitor program and a PAK4 inhibitor program. To the extent that you can comment, when would these be entering the clinic? And is there a biologic rationale for combining any of these with the adenosine access blockers?

So Juan, if you're there and you can say a few words about HIF, that would be awesome, if I don't see you go off from mute, I'll speak to that. Okay. I'll speak to that. The HIF inhibitor will enter the clinic later this year, first in healthy volunteers. We definitely think there's a very strong biological rationale for combining with one of the adenosine modulators. And in fact, that was the reason we went into this space in the first place. So one of the biggest drivers of CD73 expression is hypoxia. So you'll be hitting that from 2 directions by bringing in a HIF-2 inhibitor together with either a CD73 inhibitor or an adenosine receptor modulator. And I'll just make a note that the molecule we developed in addition to the combination opportunities, which is really just -- really the strategic focus of the program. We also feel that given our projected PK profile for the molecule that it may give us advantages relative to the Merck molecule that would enable us to hit the target harder. So it's clear that the Merck molecule seems to be limited in its PK profile by absorption. They're clearly looking to go to higher doses and higher exposures. And given what we know about our molecule today, it's still preclinical. We think we will be able to hit the target harder. We don't know if that will translate to anything better, but it's certainly something that we will explore. But our main focus there was with respect to the combinations with the adenosine modulators.

Then maybe just one final housekeeping question, Jen. If you could just go over the cash runway for the company. And again, to the extent that you can comment, how much nondilutive capital would potentially be coming in from the Gilead collaboration over the near term?

Yes. So we ended last quarter with over $800 million of cash and stated that even without an opt-in, we have funding for at least 2023. That's also assuming that we would be going pretty full steam through 2023. So operating almost as if an opt-in had happened. So it's looking at a little bit of a worst-case scenario for us, which we don't think is likely. And sorry, what was the second part of the question, Yigal?

Nondilutive capital from Gilead -- from the Gilead opt-in.

Yes. So the opt-in would be $275 million upfront when they exercise and then they would start paying for 50% of everything related to TIGIT. So that's both dom as well as AB308. The way we actually sized those opt-in payments was the idea was it would cover our 50% share of those programs. And so to some extent, like once they opt in, those programs start to become a little bit cash flow neutral for us. So that's an important aspect of how we structured the deal with them.

Got it. All right. Well, thank you all very, very much. Really interesting discussion and best of luck as you progress with the pipeline.

Thanks, Yigal. We appreciate it. And thanks, everybody, who joined. We appreciate your time and engagement. Thank you.

Thanks.