Good morning, everyone. This is Daina Graybosch from Leerink Partners, and I'm glad to be able to host this morning Eran Ophir, the President and CEO of Compugen. I think you're hearing this on Monday. We are pre-recording this the Friday prior to our conference, because Eran and Compugen are in Israel, and he's unfortunately unable to travel to be with us in person. Eran, thank you for doing this ahead of time. I hope everything is as good as it can be. We'll spend some time on Compugen and the science and the clinical programs you guys have ongoing.
Thank you so much, Daina. Indeed, these are crazy days, but unfortunately in Israel, where I'm used to it and Compugen and myself, we keep on going. Yeah, let's have a great discussion. Thank you.
Good. We have 40 minutes, and in that I wanna cover two main topics and maybe time for a third. Wanna talk about the TIGIT bispecific that you have licensed to AstraZeneca, Rilvegogostimig, they have a large program. I wanna talk about your internally wholly owned PVRIG antagonist COM701. Finally, your pipeline and the third clinical program you have licensed to Gilead, your IL-18 binding protein, GS-0321. With that, let's jump in with Rilvegostimig. I think almost everybody listening to us probably knows that rovago could soon, not quite yet, be the last TIGIT antagonist standing in the clinic.
I wanna understand sort of the potential for this, why you still have confidence in it, and maybe even why investors should be excited about it. One, Astra has a large program of phase III trials that they're conducting with rilvegostomig. Many of these trials, as you have pointed out, do not necessarily depend on the contribution of TIGIT antagonism to be successful. I like to think about it as rilvegostomig as a durva replacement. Instead of using their anti-PD-L1, they can put an rilvegostomig in there because the bispecific is PD-1 combined with TIGIT. I wonder if you could talk through first those trials, which of them you're most confident that they'll succeed, and really how much of Astra's global program is developed in this approach, rilvegogostimig as a replacement durva.
Yes, sure. In general, AstraZeneca have 11 phase III trials with rilvegostomig. Yes, and by the way, it's important for me to mention, I'm not speaking on behalf of AstraZeneca by any means. This is my own personal and professional opinion, been studying this pathway for so long. Yes, looking at the program with our partner at Astra, rilvegostomig, some of the trials indeed, it seems that they are using rilvegostomig without really comparing it head-to-head versus pembro other PD-1. Kind of using it, as you mentioned, as a IO backbone as part of a novel combination. For example, TROPION-Lung10 in non-small cell lung cancer, they combine rova plus TROP-2 ADC and compare it to pembrolizumab. ARTEMIS-001, they combine rova plus chemo compared to chemo.
There are some other studies also, in combination with Enhertu, which is an very potent drug by itself, in which they compare it to different combinations. Yes, some of the trials exploit the fact it's a bispecific that doesn't really have nobody can ask to show contribution of components to it as a, as a very good IO backbone. In some of the trials, they are comparing directly versus a PD-1, and I think there is a reason to think while these trials could be successful as well, where others have failed.
Can you help us with like what proportion? Maybe actually go through one more slowly because I think it's a bit of a complex concept. Maybe TROPION-Lung10. Let's talk through it a little bit more slowly. Then of all the 11, what proportion are this type, where you're not teasing out the contribution of the TIGIT?
I think they have around four or so that are not really comparing head-to-head. Then I think they have additional two, and they're compared to a PD-L1, atezolizumab or durva, which is maybe a weaker competition compared to pembro. I would say almost about half are not comparing directly to PD-1 or to pembrolizumab. For TROPION-Lung10, yeah, those eventually the, it's two to two to one, so they have a small rilvegostomig arm there as well. Overall, the main readout is TROP-2 ADC plus rilvegostomig compared to pembrolizumab.
Got it. Maybe let's move to the trials that aren't like that, where TIGIT antagonism is important for success of the phase III trials. I wonder why should investors have any confidence that this approach is gonna work, given what we've seen from many competitors, Roche, Merck, now Gilead, Arcus, and their phase III trials with their TIGITs did not prove successful?
I think it's the format of the antibody, in this case, the bispecific antibody. It's the combination strategy and the overall clinical trial design. I will explain. For the format. The bispecific does give cooperative binding which have shown, at least in ex vivo patient material, to be more active than PD-1 TIGIT combination. None of the others have tested bispecific, and they also use Fc-reduced, which is, again, almost all the failures were the Fc-active TIGIT. Recently, Arcus reported one study which failed also with Fc-reduced. Eventually, the Fc-reduced does have, I think, more probability for better efficacy because it doesn't deplete effector cells, and for sure, much better safety that allows to combine very easily with nivolumab and other Fc-reduced TIGITs. This takes me to the second point, the combination.
Along the trials, AstraZeneca are really doing different and novel combinations, including mainly ADCs, including the recent phase III they added including 18.2 ADC combination. This was not tested by others. There is strong rationale for ADC IO combination in general and for nivolumab it could work for its behalf. Finally, the trial design. I think that AZ learned from the mistakes of others, they are doing the studies, really learning from what all of the others have said. They're really focusing on specific population. They are, for example, selecting for PD-1 in the non-small cell trial. They are separating squamous and non-squamous to have a homogeneous population. They also probably power the trial more correctly.
For example, I think that when Genentech planned the SKYSCRAPER-01 study, they relied on, and it was fair to do so, this is what typically is done, they rely on what they saw in the phase II in CITYSCAPE. I think that AstraZeneca now are more aware of the overall TIGIT magnitude of effect, and they're also powering the trial accordingly. Again, overall, it's the bispecific and reduced Fc format, it's the combination and the clinical trial design itself.
Okay, I wanna go through some of those. Why don't you talk more about the bispecific format, and the ex vivo/in vitro data, if you will, that gives you confidence that that's bringing something more than the targeting TIGIT and then combining it separately of a PD-1?
AstraZeneca have shown few things in their publication. One, they really showed that the bispecific and their playing with the affinities of the antibody, relying on the very high affinity COM902, our own TIGIT antibody, which is high affinity, and lower affinity PD-1 allows docking first on the TIGIT and then really co-blockade of the PD-1 TIGIT on the same effector cells. Eventually, what they've shown in ex vivo fragment system, which is really a very translational system, we take the tumor directly from patients, and this system was shown to correlate very nicely with actual PD-1 responses in the patients. They showed that the nivolumab is more active than PD-1, is more active than PD-1 plus TIGIT combination, as was tested by many others.
They also showed, by the way, that if you take nivolumab and make the FC active and not reduced, you also lose activity. In this relatively translational patient material system, they have shown both it's more active than the combination and the importance of the FC reduced.
The system is actually taking fragments from real patients. You do the in vitro experiment, then you can see how those patients responded to the therapy in the real clinic. It's validated based on PD-1, then they're taking that system. While we don't have that validating here, but they're applying that, those ex vivo samples to look at all these.
Exactly.
Got it. actually, I didn't know the affinity, docking point. The COM902, which is the TIGIT part of the bispecific, that's higher affinity than the PD-1 side? How have they balanced those two?
Yes. This exactly. I think this is one of the reason they initially took this antibody, which is a potential best in class, high affinity TIGIT antibody. They combined it with PD-1 to allow this docking approach. First docking on TIGIT with the higher affinity TIGIT antibody, and then the co-blockade of PD-1, allowing PD-1 and TIGIT combined effect on the same effector cells.
Right. You're getting TIGIT antagonism, but you're also getting TIGIT as a conditional targeting element. You're bringing two mechanisms essentially in one with rilvegostomig.
Yeah. Eventually, this results in kind of cooperative binding. I think most importantly that the bottom line is that is in this ex vivo system, it was more active than PD-1 TIGIT combination.
Got it.
Now we'll have to see if it will translate to patients. Again, I think this aspect of the bispecific is definitely have the potential to add, but it's not only that. I think it's definitely also the combination and clinical strategy that AstraZeneca are employing.
To your point that they have these novel combinations, I like that as well, and I think we already talked about TROPION-Lung10, where they're combining with the TROP-2 ADC. Then can you talk more about the recently started claudin 18.2 ADC trial? Does that one also rely more on the claudin 18.2 for effect or do you expect more synergy with this ADC? I think that one maybe they are teasing out the contribution of TIGIT as well.
Yes. In that study, they are really. Again, this is the way AstraZeneca, their clinical strategy, oncology strategy in general, from what I see, they're really stratifying the patients and give to each patient population what they need. In this relatively big phase III trial, it's more than 2,000 patients, they have the PD-1 negatives, claudin 18.2 positive, which are not getting bevacizumab. In the PD-L1 positive claudin 18.2 positive, then they treat with relevant combination with the ADC. In this case, they do compare to the standard of care of nivo plus FOLFOX or nivo plus the claudin 18.2 ADC.
Got it. They're going to tease out. Not a direct comparison of Rilvigo ADC to nivo ADC, right? They're both probably comparing to the nivo.
Oh, yeah.
... Chemo arm.
No, in this trial, also, as far as I recall, they compare also directly Rilvigo ADC-
Ah.
... To nivo ADC. Yeah.
No Rilvigo ADC to nivo ADC plus nivo is compared to nivo plus chemo, right?
Yes.
Got it. you know, investors tend to give outsize focus to lung cancer, there's several Rilvigo or non-Rilvigo trials Astra has ongoing in frontline lung cancer. I wonder if you could just give an overview of holistically what's going on there with Rilvigo and Astra? Also, how much is the success of Rilvigo tied to Astra's TROP-2 QCS biomarker and their TROP-2 ADC Dato-DXd?
Talking about phase III trials, they have two trials into PD-L1-positive in combination with chemo, one for squamous, one for non-squamous, again, highlighting how they focus on specific population and work on homogeneous populations. They have a study of a monotherapy versus pembro in PD-L1-high, above 50. The TROPION-Lung10 that we discussed in PD-L1-high, combining TROP-2 ADC versus with rilvegostomig versus pembro. In this regard, the TROP-2 ADC biomarker is really interesting. I think it's one of the first to really employ AI-based digital pathology, and they're not only focusing as other biomarkers on cells which express TROP-2 or not. They look at the distribution in the cell membrane versus cytoplasm, kind of identifying the patients which have the capacity or tendency to internalize.
This is kind of a interesting biomarker, which is almost a functional assay to identify the patient who could efficiently internalize the TROP-2 ADC. It makes a lot of sense, and it's really something novel. I think most importantly, at least by now, at least until now, retrospectively, in the TROPION-Lung01 study, when they looked retrospectively on the biomarker-positive patient, the clinical efficacy of the TROP-2 ADC was much better compared to the all-comers. I think eventually there is good reason to think, and again, until you test prospectively, you don't know, that the combination with rilva plus the TROP-2 biomarker is definitely increasing yield, probability of success.
You ran through the trials really fast. Which of the trials is going to be looking specifically in the QCS biomarker population?
Only the TROPION-Lung10.
Yeah.
Which is the TROP-2 ADC combination. This is the only one who combines the TROP-2 in combination with the rilva. The other trials are either monotherapy or combination with chemo in the.
Got it.
... In the non-small cell.
Got it. Got it. You get that one overlap, and that's the QCS, and it's PD-L1 positive, right? It's the double-
High.
Positive. Yeah, high. Thank you.
They're selecting for PD-L1 high. They are stratifying by TROP-2, and eventually the primary readout is in the TROP-2 positive first and then also in intent to treat.
Oh, you could get both?
I believe so.
They added the QCS later, right? That's why you have both and then changed the stat plan.
Yeah. I'm not sure about that. I think it makes sense, but I'm not sure exactly about the chain of events.
Got it. From Astra's perspective, they have their TROP-2 strategy, which is tied to QCS, but they also have the rilvegostomig strategy not tied to the ADC. In case QCS or their ADC falls short, you still have that second and third trial going after the broader population. I heard that correctly?
Exactly. In the TROP-2, the ADC combination is in PD-L1 high. You have also the monotherapy in the same population, PD-L1 high, but also the PD-L1 positive in combination with chemo. Again, this broad clinical strategy, really with multiple shots on goal, some more relying on TROP, some maybe a bit less, all relying on the bispecific format. I think a really promising strategy.
There's all these trials going on. I wonder if you could tell us when your best estimate of when we'll start to see some of these randomized control readouts?
Again, since AstraZeneca are the ones running the studies and they fully control it, I can comment only on what they say. Per their records, all the phase III studies gonna read out after 27. Potentially, if you look at the analysis when the trial started and the actual benchmark PFS or OS, one can estimate that it could be a readout in 27. Per AstraZeneca, and we absolutely need to relate respectfully for their guidelines, all the phase III gonna read out after 27. They will have non-phase III readouts along 26 and 27 that I believe when accumulated will show maybe part of the reasons why these phase III were initiated to begin with.
When you do that math, sort of which trial started first? Which of the phase IIIs based on when they started and the control arm survival should we see first?
Yeah, I think it's a complex analysis because not necessarily the one that started first is the one that will read out first. It depends really on the control arm, OS, PFS. I believe that maybe some of the non-small cell trials have probability to read earlier, but again, Astra didn't give any guidelines, and I would be cautious about interpreting exactly which one will read first, and it obviously will depend on the actual events on the study.
Since I have you and you've worked on TIGIT for so long, I felt like we should talk about why vanilla TIGIT antagonism has sort of failed to show the large clinical benefit that we had all hoped. There was a recent, I thought, nice article in Nature Reviews Drug Discovery, and others just thinking back, like, there was a lot of money invested in this across the industry and what happened. I think I've heard and actually in the article, two hypotheses put forward. The one is something we've always known, that TIGIT is somewhat redundant with PD-1, and that they both modulate co-stimulation with DNAM-1, otherwise called CD226. That maybe for most patients and most T-cells, PD-1 is sufficient.
That redundancy is there for a reason, and if you could-- most PD-1s just fully give you everything you need on DNAM-1 co-stimulation. The second I think is a little bit more of a complex rationale, but maybe provides a more interesting path forward, and that's the TIGIT's ligand, CD155, or otherwise called PVR. When expressed at really high levels in the tumor, and you do see differential expression, may actually be driving down DNAM-1 expression on the surface of T cells. Basically, DNAM-1 becomes less relevant in those T cells. When you block TIGIT, you just don't have an impact because there's no DNAM-1 on the cell surface to provide the co-stimulatory signal. I think the complex part here is people have to realize that DNAM-1 and TIGIT compete for this ligand.
When this ligand's really high, if you have this compensatory effect, even when you block TIGIT, that might not have any impact because the co-stimulatory ligand is not on the cell surface. I wonder if you favor either of these hypotheses or one completely different, and if the bispecific Rilvigo overcomes one or both of these challenges on the that we could be having with the core TIGIT antagonism, failing to provide benefit.
I have to say, I am not sure about any of these two hypotheses. I will explain. Yes, the TIGIT and PD-1 are intersecting in the DNAM-1, also PD-1 has other activities, TIGIT signaling signals also through its tail. I will not go through into all the biology. Eventually, if you put T-cells in a tissue culture plate and you block TIGIT and PD-1, if you co-block them, you do see the activity. I don't think there's a redundancy in the signaling. Also, for the PVR high, if anything, Arcus did show not many patients, and yet that in ARC-7, that in the PVR high subsets of patients that respond poorly to PD-1, they did show very nice activity of TIGIT plus PD-1. Again, there are some other reasons to think why there's a complex biology. I agree.
I'm not sure this is the reason for the TIGIT failure, that's this downregulation of DNAM-1 by PVR, which is true. I think it's more about the fact that the TME needs certain composition eventually to respond to checkpoints in general or to PD-1. For example, sufficient numbers of exhausted T cells, this is a typical one, and probably other attributes of the TME. TIGIT, like almost all the other checkpoints, is expressed on exhausted cells dominantly. This is the dominant and high expression, like PD-1. PVRIG, the target of COM701, that we'll describe soon, is very different in this regard.
TIGIT is expressed dominantly on exhausted T cells, and I think that eventually the type of TME that has accumulation of exhausted T cells, and again, maybe some other immune factors that would respond to PD-1, has the tendency now to respond better to TIGIT. Eventually, TIGIT is active clinically, and it was shown in quite a few studies. To get a phase III success, I think you really need to identify the right patients who will benefit. For example, selecting for PD-L1 is a good start. Not all of the trials did that. Some did and failed. This is true. Again, obviously, yes, you need to have the right drug format, as mentioned before, the bispecific, Fc-reduced, really to have the to get the maximum you can get from TIGIT.
Finally, the trial design and the, and the power of the trial that really take into account the contribution of TIGIT. I think eventually the bispecific for sure is a strong word, but at least from the preclinical data could add quite a lot. Eventually the whole trial design and the, and the, and finally the combination partner, as mentioned before, could be the reason why Rilvegostimig eventually will be approved, not only in the trials that compare it, that use it as an IO backbone, but only also in the trials that compared rilve to a PD-1, PD-L1.
A lot of the trials, I think you're saying, because they combined with chemo, went broad. Did we see any subgroup analysis in PD-L1 high from any of those failed trials that supports that hypothesis? As you said, we did see failures in trials that were PD-L1 high.
For example, the, not the post-hoc analysis, but even, even the SKYSCRAPER-01, which used the Fc-enabled TIGIT that I believe is not the right format with the discontinuation and all of these, side effects of the Fc-enabled. Even in that study, eventually the overall survival, and it was in PD-L1-high, was improved by seven months. It was not enough. The hazard ratio was only 0.87. It was not enough for success, but I think there was activity there. Again, using the right format, Fc reduced, getting more activity with the bispecific and focusing on the right patient population and powering the study better, I think eventually could lead for a win in phase III.
Do you think the win is gonna be clinically meaningful? I mean, I keep thinking about that when you talk about the power that, okay, you can overpower a study, but is it clinically meaningful?
Yeah. I mean, I don't think that AstraZeneca overpowering the study. I think that maybe Genentech underpowered that specific study. I'm not saying that all the phase III studies were underpowered. Eventually, you know, if you are improving by median significantly, then you are prolonging patient's life. I mean, also, LAG-3 in melanoma, people are saying that maybe the effect was not dramatic, but eventually this is a standard of care, and this drug does prolong more the life of patients.
I hear you. Okay. Let's move to PVRIG in COM701. You have a trial ongoing, a very important trial in platinum-sensitive setting. I wonder if you could just talk about your strategy with PVRIG and this trial called MAIA ovarian?
Absolutely. What we're doing in the MAIA ovarian, we're taking platinum-sensitive ovarian cancer patients in the second or third line, meaning typically they are still responding to the platinum chemotherapy. That's why they're considered platinum sensitive. They've probably already received bev on pembro, which is a standard of care for the earlier lines or they're not eligible, there's no standard of care for these patients. What typically happens in the patient journey, they receive the four to six cycles of platinum. If they're not responding, they are platinum resistant. If they are, they are platinum sensitive, there's no maintenance treatment to really maintain this therapy. Eventually, typically after around four to five months, they are relapsing and eventually becoming platinum resistant. The unmet need is clearly there.
What we're doing in our study, we're taking these patients in the second, third line, platinum-sensitive. We're excluding, and I will explain a bit later why, we're excluding patients with liver metastasis, and we're randomizing them 40 to receive COM701 in monotherapy. Because there is no standard of care, 20 will receive placebo. It's an adaptive trial design. This is the first arm of the study. We want first to see clearly that there is monotherapy activity of COM701, and then we can consider either adding more arms or maybe consider adding more patients. This is the MAIA study.
Why this setting is the right setting to understand the monotherapy activity of COM701?
There are a few reasons why we think this is the right trial for COM701. First, like everything we do in Compugen, it's the biology. The PVRIG pathway in general is very high in ovarian cancer, not specifically in platinum-sensitive, but in ovarian cancer in general, and it was a target indication for us to begin with. Also the unique biology of PVRIG. As I mentioned before, TIGIT and other checkpoints, also PD-L1, were expressed mostly on exhausted T cells. In PVRIG, the dominant expression, and we published a nice paper in Cancer Immunology Research two years ago, the dominant expression is on stem-like memory T cells, which have very strong proliferative capacity.
This unique biology have led to when you block it with COM701, we really have the potential to drive proliferation of cells and to see a burst of T cells, a wave of T cells into tumors, also in tumors which are less responsive to checkpoints which have less T cells there to begin with. This is exactly what we've seen in patient samples taken before and after treatment. We saw that COM701 can drive T cells into these less inflamed tumors. We saw clinical responses in these less inflamed indications. For example, ovarian cancer. Ovarian cancer, if it's PD-L1 negative, the probability of responding, or at least in platinum-resistant settings, last line patients, to respond to PD-L1 or even PD-L1 plus TIGIT is very low. Historically, PD-L1 plus TIGIT showed 0 response rate in PD-L1 negative platinum-resistant ovarian cancer.
We have shown in monotherapy and in combination response in multiple patients with PD-L1 negative exactly along the lines of that unique biology. First is the biology, the dominant expression, and the biology that could enable activity in this indication, ovarian cancer. The actual clinical signals. When we published in ESMO last year, they pulled analysis from the phase I we did in last line patients, the platinum-resistant patients, heavily pretreated, and the signals of clinical activity were there.
Even in that last line, patients platinum resistant, if we just excluded retrospectively the patient with liver metastasis, representing patients with maybe a bit less immune suppressed immune system, a bit lower tumor burden, we already saw almost 40% of clinical benefit. What we're doing in the MAIA, we're going earlier. We're going to patients in the second, third line platinum sensitive that have less compromised immune system. We also exclude there patients with liver metastasis. These patients have low tumor burden.
I think this is exactly the place where a drug like COM701 when used in maintenance to show its durability, combining with its safety, this is exactly the right place to test COM701 to translate the activity in the last line patients to the earlier lines.
Do you have sufficient power with 20 patients in each arm? I guess not power isn't the right word. This still is a maintenance study, so you're looking at survival. How big of effect do you expect between the two, if COM701 is working? Will we be able to essentially really see the signal based on the 40 patients?
Yes. First of all, we have 60 patients. It's 40 in the COM701 arm and 20 in the placebo.
Thank you.
We also have a very, quite solid historical control. We are combining also this in the analysis. Eventually, I'm not saying we're gonna have here a p-value that will drive immediate approval of the drug, but I think this data should allow us to clearly see if there is monotherapy activity. We're also looking at PFS, not in OS. It will really want to see. Ideally, I think, if indeed the placebo will be around the four or five months, if we get a three-months improvement in the progression-free survival, it's gonna be very clinically meaningful. If it will be a bit shorter than that, it really depends on the distribution of the hazard, the confidence intervals, et cetera.
Overall, I think that if we are expecting a strong effect, this trial should allow us to see that signal. Then we can make the next steps and decisions once we are sure about the monotherapy signal.
You're essentially setting just a really high bar. Like, you need to really see clinical benefit here, and, we'll see that in this small trial if that's the case.
Yeah. I think again, the combination of a randomized blinded study, 60 patients with solid historical control, with setting a bar, a high bar for activity, which is justified based on what we have seen in the platinum-resistant settings, is the right way to go. After we really prove that COM701 is active, then we have a few options. We can think about accelerated approval because of the unmet need. We might need to add more patients, obviously, but yet. Then, of course, phase III trial or other combinations, depending also how the landscape will evolve while we're doing the trial.
Yeah. Can you talk to me about the you quickly went through, but the steps forward if you do see a strong signal here? You mentioned accelerated approval. There's certainly many combinations you could consider. We have a pembro approval in a refractory setting now based on the data we saw last year. There's a lot of promising ADCs in proc. I mean, there's a lot of options certainly into just how you're thinking about that.
If we talk about the maintenance settings, second, third line, once we show clearly the monotherapy activity due to the unmet needs, and again, this will be have to go through a thorough discussion with the FDA. This is something that was not done yet. It will be with the data. After showing a very clear activity in the monotherapy, definitely thinking about moving to registration with COM701 alone is an option, either accelerated or/and eventually a phase III trial. Another option obviously will be to combine after we are sure we have monotherapy. It could be combination with Bev. This is open also maybe to population, to a bit wider population. PD-1, an option.
If ADCs by then will show activity in the maintenance settings, if they are safe enough for maintenance, then it's a very different mechanism. The COM701 MOA could give the durability to the ADCs. I think once showing the signal of the mono, we have quite a few options, not to mention to go also broader. I mean, I think initially we'll focus on the exact observation we will see and try to go to registration in the specific patient population. Eventually, after showing the signal, we can go earlier, we can go later in ovarian, and we can go to many other indications that we did saw signals also in other indications. We start by focusing on the monotherapy signal, which is required, definitely by us.
Also these days, everyone wants to see monotherapy before moving forward, and it was justified.
I hear you. Okay. Maybe now let's talk about your third clinical program, the IL-18 binding program, that's partnered with Gilead. Maybe just give us an overview of where that is in the clinic, when we might see data, and then we could talk more about the mechanism?
This program was licensed to Gilead, and we dosed the first patient in a dose escalation study at the beginning of 25. The study is doing monotherapy and combination with Gilead PD-1 zimberelimab, and it's split it into two sub-studies. The first one, which is dose escalation, mono combo, plus some backfill cohorts, and then the expansion. Due to our obligation to Gilead, we cannot say too much. But in general, it's moving really nicely and the enrollment is working for us. I mean, there is it's a new mechanism and there is excitement and yes, we report data. For now we cannot give guidelines, but in line with Gilead, when the data matures, I'm sure that we'll report data as well.
There's been some recent successes of IL-18 binding protein in immunology. Is there any read-through to this program? What gives you confidence in the relevance for oncology?
Yes, we know that IL-18 is an important cytokine, also from immunology. When IL-18 is high, there is activation of the immune system, and this is reassuring to see that IL-18 is an important cytokine. This combined with our observation using Unigen, our AI-based engine, we identified that in oncology, in patient samples, there is accumulation of IL-18, but in an inactive form, so it doesn't activate the T cells and NK cells, even though it could, because of IL-18 binding protein, which is expressed there in high levels and blocks its activity. What we did here, exploited this computational observation and developed an antibody. Most people are developing cytokines for oncology and trying to engineer them, and it's very challenging, the therapeutic window. For years, though, people are trying and it's still challenging. We're not using a cytokine.
We use an antibody that blocks IL-18BP, inhibit the inhibitor, if you may, and by that enables IL-18 that we know that is an active cytokine, you just mentioned observations in immunology, that really drives an activity in the TME and not in the periphery because both of them are high naturally in the tumor microenvironment. This is kind of approach that should overcome the limitation of most other cytokines, giving activity in the TME without all the peripheral side effects.
That's great. I know you said already that, you know, Gilead, your partner, doesn't want you to talk too much about this. Should we expect any milestones this year, or just sort of generally when we might expect to learn?
Yeah. Like in any deal, I can say something maybe a bit generic, and yeah, this is the best I can say. Normally you get milestones when moving between stage to stage. This stage 1 is divided into phase I of the trial, or this phase I is divided into stage 1 and stage 2, and then the phase II and phase III. Typically, one could expect milestone between the different stages, but we cannot commit specifically about if and when, unfortunately.
Okay. Maybe the last question then is you mentioned your computational platform, and you've been doing computation at Compugen for a long time, before the AI era, if you will, and I wonder what it looks like now in the AI era, how that platform has been evolving and anything we should expect coming from your computational discovery this year?
Unigen, this is really the platform we built for years, and I think in the core of this platform is the data that we're accumulating along years. This data is really looking at the tumor microenvironment from every different angle you can imagine with every technology, proteomics, spatial, transcriptomics. Eventually, we have the patient's data mapping the tumor microenvironment very deeply. The way we utilize it today using AI, and actually we were doing AI before it was actually called AI, was, you know, at least before the trend of AI. We are using AI to actually feed the database itself in automatic manner into QC, use AI to query the data itself. We use AI tools to ask the system or ask database how to find specific targets.
Not less importantly, sometimes non-artificial intelligence, some might call it human intelligence, still have some role to play. I think having this experience was knowing how to translate computational output to having the end in mind. What do you want to target in the clinic? Which patients? How to validate? What experiments to do in the lab? I think this all combination is what's unique about Compugen, and this is what makes our platform maybe different of some of the others, and eventually it is a validated one. This platform brought PVRIG, TIGIT, ILDR2 and others.
Whether we should have any expectations for new programs from it in the near?-
Yeah. The biggest group in Compugen is the one working on the early pipeline. We started with the DNAM axis, PVRIG and TIGIT. We moved then to very different biology, cytokines, COM503. What I can say for now that, yes, the we are working on new biologies, new formats, new MOAs, focused on oncology. We are still focused on oncology. For now, we don't have specific guidelines about when, but definitely once we progress more and the risk and sometimes new biologies takes time to de-risk. Of course, we will report externally the MOAs and the new targets that we're moving forward with. For now, we don't have specific guidelines for that.
Okay. Well, thank you very much. I appreciate this. I hope for lots of good things for Compugen in the year forward and for you and your colleagues, in Israel, as well.
Thank you so much, Daina.
Thank you.