All right. We're going to go ahead and get started. I'm Stephen Willey, one of the Senior Biotech Analysts here at Stifel, and glad to have with us to wrap up the morning session, Eran Ophir, who is the CEO of Compugen. Eran, thanks for joining us. I actually ran into Anat at ESMO.
Oh, really?
I hadn't seen her in a while.
Okay. She's still involved?
Yeah. She's still Chair of the Board, right?
Absolutely.
Okay. Any opening statements you want to make before we jump into Q&A?
Yeah. Maybe for those who are new to the story, you can say a few sentences about Compugen. Compugen is relying on our computational AI-based engine that we built for years to identify new drug targets in the field of immuno-oncology, and we're pursuing these targets clinically through internal pipeline and also through collaborations. The engine is, if you may, validated because we have shown we can identify computationally drug targets, take them to the clinic, collaborate with pharma. Internally, we have the COM701 molecule, which is an anti-PVRIG antibody, which is now in a randomized study, the MAIA-ovarian trial, in a maintenance settings for platinum-sensitive ovarian cancer. We have two ongoing collaborations, one with AstraZeneca on the molecule rilvegostomig. The TIGIT arm of that bispecific PD1-TIGIT antibody is actually our own COM902, derived from our own COM902.
This interesting molecule is now in 12 phase III studies. We have another collaboration ongoing with Gilead on another first-in-class program that we identified computationally. Altogether, we have more than $1 billion in milestones and royalties from these collaborations. Behind all of that, we continue to use Unigen, recently announced the expansion into UnigenX for identifying targets also for T-cell engagers. This goes into the early pipeline. We have a solid cash balance with cash runway into 2029, allowing us to progress all our programs, so to get into catalysts in 2027, 2028.
Okay, perfect. Maybe we can start with MAIA-ovarian. I think you announced yesterday you completed enrollment into the trial. This is a kind of adaptive platform-type study that is evaluating single-agent COM701, this is your anti-PVRIG, as maintenance therapy. These are in platinum-sensitive ovarian cancer patients. What specifically about the biology of COM701 kind of gets you excited about the program?
PVRIG is a checkpoint, but it is a very, very different checkpoint, different from PD1, different from TIGIT, and all the others. I will not go through all the details, but what we have shown extensively that it controls a different stage in the cancer immunity cycle, stem-like memory T cells, and so it really showed activity in our earlier clinical trials in places that typically checkpoints are not working. For example, in the platinum-resistant settings, in last-line patients who failed everything in ovarian cancer, we had responses in COM701 monotherapy in combination. For example, in PD-L1 in ovarian cancer, also in PD-L1-negative ovarian cancer, which historically even PD1 plus TIGIT in combination shows zero response rate. It is different biology that leads to different outcomes, and specifically ovarian cancer was focus indication for us from the beginning.
The pathway is extremely dominant there, and we are glad to see entering to the clinic that we saw signals, again, monotherapy in combination in the last line ovarian cancer patients. The rationale of the current studies is to try to focus on the patients which are most probable to respond to the biology of COM701. When we initially started in the platinum-resistant, we are now moving earlier into the platinum-sensitive settings. Patient is second, third -line that responded to their last line platinum, so they have very low tumor burden. Some of them do not have tumor at all.
We exclude patient with liver metastasis. They have less exhausted immune system, low tumor burden, and by that, we hope that the signal we have seen in fraction of the patients in the platinum-resistant setting will translate into improving median progression-free survival in this important maintenance settings that have a huge unmet need.
Okay. Maybe you could just kind of remind us a little bit about kind of the adaptive trial design that you're looking to leverage here, and how these answers will either inform or not inform what you choose to do next with this MAIA platform trial.
The study is focused on second, third-line platinum-sensitive patients. These are patients who receive platinum, they respond to the platinum. We take only those who responded to the last line platinum. We exclude patient with liver mets, and then the patients are randomized. 40 patients receive COM701 monotherapy, and 20 patients receive placebo. Because of the no standard of care, we could actually have the luxury of using here a placebo control and compare it to COM701. The goal of this arm of the study, and it is an adaptive trial design.
Meaning we have now the first arm, which checks COM701 monotherapy, but then we can open more arms very easily for different combination, if you may. But the goal of this arm is to show clearly COM701 monotherapy activity versus the internal placebo control. Now, if we see significant monotherapy activity, and if they were improving median PFS for more than six months, we'll discuss maybe a bit later about what we look for, what is clinically meaningful. Then, of course, we could start planning for phase III, but also in parallel, we can now open more arms for combinations, like ADCs or whatever, bevacizumab.
Okay. You talked about kind of the 2:1 randomization that this study leverages. I know you've changed, or I guess you revised your assumption around what control arm PFS should look like in this study, and that happened, I guess, after the initiation of enrollment, after the trial had started. Can you just kind of provide some color as to why or how those assumptions changed over time? I guess, what informed that change in viewpoint of what a control arm's going to do?
Yeah. When we started the study, the data available to inform on second, third-line platinum-sensitive ovarian cancer was from old trials when the treatment algorithm was different. For example, PARP inhibitors were not approved in first-line maintenance in these days. The usage of bevacizumab might have been a bit different. More recently, there were two studies in platinum-sensitive ovarian cancer, not exactly the same patient population as ours, the TIDAL and DUO-O, actually a bit more heavily pretreated. Maybe they enrolled also patients in later lines, but overall maintenance studies for platinum-sensitive. For them, the placebo arm had a median PFS of 2.8 months. Because these are a bit heavier pretreated patients, we did our interpretation, and our current assumption is that our control arm will be around four months.
This is really reflecting in the changes in treatment algorithms and having more contemporary studies to inform us of what the placebo control arm should look like. Anyway, that's why we have an internal control arm that we can compare not to historical data, but actually compare COM701 to the internal control arm, and having the historical data to give context is always important.
Okay. You've guided to this interim update happening in the first quarter of next year. Can you maybe just talk about either, A, what triggers the disclosure of that interim? Is it predicated upon some pre-specified information fraction? Is it representative of a given number of progression events that have accrued? What is the trigger for this disclosure? Any guess as to how many patients that might represent and how you as a company plan on communicating that?
Our assumption that by the time we report the results, by the end of the first quarter of 2027, we're going to have sufficient patients of study or sufficient patients off study to represent the median PFS results, and maybe even a bit more than that to really see the separation of the curves.
We estimate that by then, and still dependent on the way the study will develop, because obviously, we are still enrolling patients until recently, but we estimate that by then, we are going to have sufficient patients of study to really see the separation of the curves and to understand clearly, do we see here COM701 monotherapy activity, which is the goal of that study.
Okay. COM701 has a very clean safety profile, right? Obviously, the control arm is placebo. I would imagine if and when we see those KM curves of progression, there shouldn't be a lot of contamination from censoring due to dropouts or
Yeah, the dropout, the discontinuation rate was very limited. It was very small numbers in the earlier trials.
Okay.
I don't expect anything different here.
You are now thinking about a placebo control arm that looks like four months, give or take.
Probably, yeah.
What is your bar for success relative to that four-month number with COM701?
There are two aspects for what success looks like. The study is a 60-patient study. It is not planned, not intended, and not powered for statistical significance. This is not the study. The next study, the phase III, so hopefully the registration trial, the phase III study, obviously going to be powered for significance. What we do want to see here is, first of all, to be convinced that there is separation of the curves and that 701 is showing activity in monotherapy. Then when you ask our PIs, "What is clinically meaningful for you?" It's not only to see monotherapy, right? If you want to go to registration, it has to be also a clinically meaningful effect.
For them, to have the patients remaining platinum-sensitive, meaning if the PFS is at least median of six months for the COM701 arm, this is going to be clinically meaningful. Because in the ovarian cancer settings, patients who are relapsing after less than six months following the platinum are considered platinum-resistant. They are being moved into the different treatment settings that have worse prognosis. For physicians to keep patients in a platinum-sensitive, meaning if most of the patients will end the study platinum-sensitive, meaning that the tumor will relapse eventually, but after more than six months, this is clinically meaningful for them.
Yeah.
I think the totality of the data plus the six-month bar is going to be meaningful and probably could be sufficient to now start thinking about registration.
Yeah. No, it is kind of interesting, right? I guess that six-month kind of treatment-free or platinum-free interval is kind of this binary definition of what is refractory versus sensitive. I am sure it is probably much more continuous, right?
Of course. This is almost a technical bar.
Yeah
But it does maintain the platinum-sensitive. Eventually, we need to see how they respond to the next platinum, and we need to see the overall survival, of course. But for this type of study, we think that the six-month bar is a good bar for what is clinically meaningful for patient.
Okay.
Physicians treat ovarian cancer.
How do you think about the commercial opportunity for COM701 here? Maybe you can speak to that specifically in the context of some of the eligibility criteria that you're requiring, right? So I believe you require patients to have at least a PR or better to prior platinum, right? So no stable disease patients. I know you're excluding patients with liver mets because we know the totality of IO data would suggest those are hard patients to treat. I guess when you make those cuts, what are you kind of left with in terms of a patient number that is addressable for you, should you choose to move into registrational phase III?
Overall, the second, third-line are roughly in the 30K numbers. If we focus specifically on MAIA population.
Yeah.
Which is PR, CR, exclusion liver mets, patients who already received Bev or PARP or not eligible for Bev or PARP, our estimation that you're somewhere around 10K.
Okay.
Having positive monotherapy data, you can easily, even just by adding Bev, by the way, could already increase the numbers. After showing monotherapy activity, we can go earlier or later in ovarian cancer.
Not to mention that PVRIG pathway is relevant also in other indication. We had clinical signals in other indications. So the potential obviously is much broader than just ovarian, but the exact same patient population, roughly 10K annually.
How do you just think about ovarian from kind of just a broader competitive perspective, right? I know the platinum resistance setting has been pretty competitively crowded. There's kind of a gazillion topo-conjugated ADCs that are somewhere in development. I think we've now seen some of the PD1 VEGF bispecifics trying to go into.
Starting, yeah.
Earlier lines of platinum-sensitive disease. Some of the ADCs that may have a therapeutic index, which is not many of them, but that suggests that they can maybe go into a more proximal line of therapy or trying to go there. It's a busy space, right? As you think about maybe having to try to design and execute on a registrational study for COM701, given that there's stuff happening downstream and upstream, how do you factor that into your decision-making process?
This is a very important question, and for now, most of the competition in the maintenance settings, in that patient population, in the platinum-sensitive maintenance settings, is mostly ADCs. There are already a few phase III trials ongoing for ADCs. COM701 is very, very differentiated. There are huge discussion in the community about, in general, the use of ADCs for these settings, because there is a reason why patients take four to six cycles of platinum and they stop. In this, I mean, it's second, third-line, but this is an aggressive cancer, and some of them just stay for observation, take nothing, because the platinum is very aggressive chemotherapy, and they cannot tolerate more than that.
So, to finish six cycles of platinum and then move into a different type of chemotherapy, even though it's better tolerated, it's not an easy decision, especially not for long-term maintenance treatment. Some patients will not be able to take it, some will not want to take it, and COM701 is significantly differentiated. Not to mention that some of the physicians, and all of them have the same payload, so the one that will win will be the one, right? Some physicians think that actually they prefer to keep the ADC for the platinum-resistant settings and not to use it for maintenance in the earlier line. I think overall, if the ADCs will be approved, and actually in ASCO, there was the first results of treatment plus maintenance for MIRV that failed.
But if the ADCs will be eventually successful, especially as you mentioned, the next generation, the safer ADCs, COM701, I think, will have its place in that competition. Not to mention that if we show significant monotherapy activity and all these same payload ADCs are competing with each other, to add an IO arm to that combination to differentiate them and give them the durability is another opportunity that you can test fast in our adaptive trial design.
Okay. Maybe we can shift gears to GS-0321.
Yep.
This is an IL-18 binding protein targeting antibody that is partnered with Gilead. Kind of interesting to me that if you, I guess, look at the Compugen discovery engine, right, this is kind of like the third novel target that we've seen, I guess, manifest off the tech platform itself. But maybe you can just kind of talk about how you think this asset is differentiated relative to some of the traditional cytokine approaches and maybe what that allows you to do from a clinical development perspective, where some of those other modalities are a bit limited.
We used Unigen platform to scan for molecules which could mediate resistance in the tumor microenvironment. The observation we made using the platform is that in the tumor microenvironment, there is naturally a cytokine called IL-18, which is a very potent cytokine, which is naturally there in high levels induced by inflammasome, high in the tumor, low in normal tissues and blood. But it's completely not active because there's another molecule called IL-18 binding protein, and by its name, it's exactly what it does. It binds IL-18 with high affinity and blocks its interaction. So what we identified computationally that while for most cytokines, if you want to use them as therapeutics, you need to put them into the patient's veins.
Normally, you first induce side effects, sometimes you get something into the tumor, you get some effects. People are trying to engineer cytokines for years. Maybe there are now some emerging interesting clinical data, and still the safety profile is problematic. We've just seen the data from Innovent molecule, right?
Not a safe usage molecule.
The PD1 IL-2 was a little disappointing.
I agree. Exactly. Especially the safety in general. We do not need to give IL-18. It is there. It is naturally in the tumor. We developed an antibody that inhibits the inhibitor, displaces IL-18 bp away from IL-18, and by that, at least pre-clinically until now, or reported pre-clinical data, is that we are having the potential potency of a cytokine, but without the side effects of a cytokine. The activity was really confined to the tumor microenvironment, and almost nothing happened peripherally, and this was promising. This is why Gilead and some other pharma companies chased it, and we had this nice deal in the pre-clinical stage. We are now doing a study in monotherapy and combination, that started at beginning of last year, and it is ongoing, and it is exciting.
Have you preselected tumor types for that initial study? Are there certain tumors where IL-18 binding protein might maybe have more of a biological role in tumor progression, or is this kind of your classical phase I all-comers?
The dose escalation is composed of. The study is, the phase I, composed of dose escalation, some backfill cohorts, and expansion cohorts. The dose escalation is typically all-comers.
Because you want to escalate and test your drug. We did not give specific guidance on the tumor types for expansion and backfill, but what we did say that pre-clinically we saw that you need some level of T-cell or NK infiltration in the tumor to respond to the IL-18 being released. The indication we look at are probably, the pathway is expressed in many indications, but maybe not the coldest ones, not the immune desert ones, but more that have some basic level of inflammation. This is what we saw pre-clinically, and this is, I think, what we can say at this stage.
Okay. Who discloses this data? Will it be you or will it be Gilead?
We are running the trial, so we control the data, but Gilead, which owns the molecule, controls the data disclosure. Eventually it will be a discussion between us and them about the right timing for release. Obviously, there are competitive reasons. This is a first in class.
The moment we release data, if it's positive, the competition will arrive very fast, and I understand Gilead, and also ourselves, that want to keep the cards close to us, and we'll see when would be the right time to release the data. It will be a discussion with them.
Okay. I know Gilead has talked about, or maybe they've even articulated this dose expansion portion where they're going to be evaluating this drug in combination with their PD1 inhibitor. I know you just talked about kind of hot and cold tumors. Presumably, combo with zimberelimab would allow you maybe to expand that reach of that biological effect to a broader number of tumor types, or how do you think about that combo specifically and the synergy that you might be able to extract from it?
I think that eventually the tumor types will be dictated by the ones that we think that COM503 or GS-0321 will be active in. Based on the pre-clinical data and the biological rationale, there is a big I think that the combination with PD1 definitely should amplify the effect that we see.
Okay. What is Compugen's economic leverage to this program? Can you just remind us of what that is?
We received to-date $90 million on the program.
Nine zero. We are eligible for additional $758 million, and up to low -double-digit tier royalties.
Okay. In terms of pursuing additional development opportunities beyond phase I, is that just a Gilead-driven decision? Do you have any say? Do you have a seat on a steering committee that allows you to give an opinion on how this drug should be developed?
Absolutely a seat to give our advice on. First of all, we are running the phase I trial.
Yeah
The later trial will be run with Gilead. Obviously, we will have a seat. We know the drug. We developed it. We understand the biology. This will be their decision, but we will be consulted.
Okay. Maybe we can switch gears to rilvegostomig, which is a drug name that I incorrectly pronounce 57% of the time. This is a TIGIT PD1 partner with AstraZeneca. I think they have 12 phase III studies that are now ongoing across multiple indications. I think there is even more in phase I and II. Broadly speaking, I mean, the TIGIT narrative is what it is. It has not really played out. How does rilve change the narrative in your mind around that legacy TIGIT narrative and the ability to win in all these different tumor types that AstraZeneca has pursued?
Yeah. First of all, rilve, I think, still didn't change the narrative. People think negatively about rilve because I think they are missing why, and they are right. After all the TIGIT failures, why should this molecule be successful? I think there are few reasons for the, I think, good potential of rilvegostomig. One, it's a bispecific molecule. All the others tested PD1 and TIGIT in combination, and we know that for bispecific molecules, in many cases, one plus one equals more than two.
We know that AstraZeneca designed this bispecific very smartly, and there is supportive data that shows that it's more effective than just combination. Then you look at the trial design, and I think they learned from the mistakes of others. They're powering the studies differently. The SKYSCRAPER-01 had a hazard ratio of 0.87, but it had only 500 patients, and the alpha allocation was problematic. This study could have been positive even for that problematic TIGIT molecule from my perspective. So they are designing the trial differently. They're powering it differently, selecting for PD-L1. For example, Arcus didn't select for PD-L1. They are selecting for PD-L1. Then comes maybe the most interesting observation that I'm not speaking on behalf of AstraZeneca. The way I look at the trial design, we don't have a consulting seat, by the way, for rilvegostomig.
They are running the trials. We are definitely an audience in this case, enjoying for the potential milestones and royalties that we will get. Almost half of the studies are not asking the question of TIGIT. Because it's a bispecific molecule.
Yeah.
Nobody can ask them, "Okay, take out TIGIT and now see if it works." For example, they just announced that in the second half of 2027, there will be a first phase III readout. This is a study in BTC, in biliary tract cancer adjuvant settings. It's studying rilvegostomig plus chemotherapy versus chemotherapy. There is no TIGIT question here.
If the molecule because the molecule is safe like KEYTRUDA, in terms of efficacy, it at least covering PD-1. So it at least like KEYTRUDA, TIGIT is effective and gives something. So in the worst case, it's a very, very good PD-1 molecule tested in an IO-sensitive indication in adjuvant settings, an indication in which IO is approved in metastatic settings, an indication in which rilve have shown in a small study to be more effective than KEYTRUDA.
Due by historical data. Now in adjuvant settings, let's see if rilve is working.
Yeah.
There are other trials in combination with HER2, TROP2 ADCs that are asking, "Is the combination effective?" Not asking if TIGIT is active or not. So I think at least half of the studies, even if you completely don't believe in TIGIT biology, could be positive. I think also some of the others because of the bispecific design and the potency.
Yeah. No, I think that's actually a really important point in terms of just trying to bypass the question of what is TIGIT actually doing, right? I guess the analogy here would be Genmab developing petosemtamab in frontline head and neck cancer, right? Like, do I have any idea what LGR5 is doing? No. Like, does it matter? No.
Yeah. The drug is active and safe enough. That is it. Exactly.
It is a good point. Maybe just talk a little bit about the economic leverage you have here. I know you have monetized a portion of those economics that you do. Maybe you can just kind of remind us of how important this is to Compugen and maybe also kind of walk us through the decision to monetize that portion of the royalty stream as well.
We monetized a very small portion of the royalties. Because we strongly believe that rilvegostomig is going to be a huge commercial success, given that it is being used as the IO backbone for AstraZeneca in many indications. So, in this case, we are eligible for additional $195 million in milestones, and most importantly, mid-single -digit royalties. If you think about the commercial opportunity, I mean, AstraZeneca have recently guided that they estimate more than $5 billion of sales. I think the potential even is broader. Thinking about it as in like their internal KEYTRUDA molecule for combinations. So the royalties are very important. We monetized the small portion. It was, we cannot say the exact numbers again because of the pharma biotech relationship, but it was mid-single -digit royalties before, which remained mid-single -digit royalties after.
We use that small portion we monetized, we received $60 million upfront, an additional $25 million for the next milestone, which is BLA acceptance, hopefully from this BTC trial in 2027. By that, we extended the runway. We can continue and progress our internal pipeline, the MAIA study, the early pipeline that all comes from Unigen, and reach the milestones both internally and externally, while maintaining the long-term potential from benefiting from rilvegostomig as a potential blockbuster.
Okay. Any questions? Anything we didn't talk about that you would like to quickly cover before we wrap things up here?
I think that another thing that we announced recently that we're extending Unigen, our discovery engine into UnigenX, which is aimed at identifying tumor-associated antigens, which are targets, very clean targets, which are tumor-specific for T-cell engagers, for example.
We also in ESMO how a good T-cell engager looks like when the target is clean and good, and how it can turn a so difficult indication like small cell lung cancer with a good clean target like DLL3. We really try now to find additional such targets, and hopefully, we'll make it. We have already some initial hits, and we'll be able to commit for some more concrete guidance, hopefully, in the near future. We'll see how this will progress, but this is very exciting new expansion of the computational platform.
Well, the world needs more DLL3-like targets.
Absolutely. There are a handful of such targets, and when they are active, we saw also the Xencor results.
Yeah.
When you have clean, good targets, then the clinical results are stunning. But there are very few such targets, and we are employing everything we build along the years to try to find these kind of targets.
All right. Eran, thank you very much. Really appreciate it.
Thank you.
Thanks, Eran.