Morning, everyone. Thank you for joining us at the first day of the Needham & Company Healthcare Conference. It is my pleasure to have with me this morning Eran Ophir, the CEO of Compugen. Eran, please go ahead and start your presentation.
Thank you, Gil. I'm delighted to share some of the advances Compugen's making, utilizing our AI-based computational engine to identify new drug targets to activate the immune system against cancer. Maybe I will start with a few words about the history of Compugen. Actually, we were established in the 1990s, initially as a hardware company, really trying to deal with the vast amount of biological data that started to accumulate these days. Later on, we listed in Nasdaq in 2000, and we moved to deploy more software and doing collaboration with pharma companies using our computational capabilities.
But the Compugen is a Compugen where we realized that to make the most of the value out of our computational engine, the best thing would be to take the computational discoveries, the new drug targets that we identify, but to develop the drugs themselves and take it to the clinic ourself. Around a decade ago, we started to employ this strategy, really not only doing the discoveries, but also developing the drugs and taking to the clinic.
After around a decade of this strategy, I think this is a very exciting time to talk about Compugen because we both have a whole pipeline derived from this computational engine, and hopefully in the near future, we're going to have our first drug derived from our computational discovery to be approved, as we have one of our important drugs now in 11 phase III trials ongoing by AstraZeneca. We're active in the field of immune oncology, meaning to activate the immune system against cancer. The poster child of this approach is Keytruda, the best-selling drug in 2024 and 2025, and in general. Keytruda doesn't, and other PD-1 blockers and other immune oncology drugs, doesn't work for all patients. Certain patients and patient populations really do not respond to Keytruda and similar drugs.
Immune oncology and oncology, in general, are trying now to employ different combinations, new mode of actions, and really to eventually bring immune oncology to a larger fraction of the patient. This is exactly where Compugen fits in. We're using our validated computational engine, Unigen, to identify new mode of actions, novel biology, to really expand and improve treatment option for cancer patients. I'm saying the engine is validated because using this engine, we've shown again and again, we can bring new drug targets to clinical testing with clinical signals, and most importantly, the platform is validated by the partnerships on the drug targets and drugs we developed with AstraZeneca and Gilead. We have two fully owned program in our pipeline, clinical programs in our pipeline.
COM701, which is a potential first-in-class antibody checkpoint blocker against PVRIG, a target that we identified, and COM902, potential best-in-class anti-TIGIT antibody. We also licensed to AstraZeneca the right to use COM902 as part of their PD-1/TIGIT bispecific antibody, rilvegostomig. This antibody is now moving in 11 phase III trials, hopefully toward registration from 2028 and on. Now, the field of TIGIT, and we identified the target in parallel to Genentech in the past, and the field has it definitely ups and downs for those who follow. We truly believe that rilvegostomig can be successful in places where other TIGIT have failed, and I will explain in the following slide why do we believe so. In addition, we licensed to Gilead more recently, a potential first-in-class anti-IL-18 binding protein antibody.
The target was licensed to Gilead, and we are now moving this target into phase I. We also have additional early-stage pipeline, multiple undisclosed assets, all derived from our validated computational discovery engine, Unigen. Compugen in 2026 is uniquely positioned with rich clinical pipeline and with validated partnerships with potential more than $5 billion in milestones plus royalties. We also ended 2025 with a very solid cash balance of more than $145 million, and this should allow us to continue to run all our operations, clinical, pre-clinical, without any further potential milestones that could arrive into 2029. Let's start with our pipeline. First is our COM701 antibody, anti-PVRIG antibody. This target is now moving in the MAIA-OV trial.
It's a randomized adaptive trial study, and the subtrial one is a randomized study of COM701 monotherapy versus placebo, and we expect results in Q1 2027 for this study. Rilvegostomig is driven by AstraZeneca now, PD-1 TIGIT bispecific across 11 phase III trials and 14 phase I or II trials across multiple indications with multiple combinations. The GS-0321 partner program is run by us in the phase I study that was initiated at the beginning. The first patient was dosed at the beginning of 2025, and again, the early stage pipeline of undisclosed assets all derived from Unigen computational engine. Let's talk about COM701. COM701 is targeting PVRIG, a checkpoint that was not known to the scientific community when we identified computationally. Moving forward with the first-mover advantage, the first to enter the clinic to show clinical results.
What is unique about PVRIG is that the biology is very, very different from TIGIT. It is very different from most of the other checkpoints in general. This unique biology, and we'll not go through all the details today, has the potential to really drive T cells into less inflamed indications, which are typically not responsive to checkpoints. Specifically, the pathway is very high in ovarian cancer, which again, is an indication which typically is responsive less well to checkpoints. Ovarian cancer was a target indication for us before we even started. We started the clinical trials, and this is definitely an indication where we treated last-line platinum-resistant ovarian cancer patients who failed every treatment option. We saw clinical results with COM701 in monotherapy and in combination. The patients who did respond have very durable responses with excellent safety.
What we decided, taking these signals in the last-line patients and to take them where we thought, along with the KOLs we discussed with, would fit the most to this type of drug. This is the platinum-sensitive settings. We talk about patients who are the second or third line. They received four to six cycles of chemotherapy. This early line to what we've seen before, meaning they have a more potent immune system, less compromised by repeated cycles of treatment. We take only patients who responded to the chemotherapy, so they have low tumor burden. This, we believe, makes it a better situation for the COM701 to really induce its effects.
The goal of this study, and that's why we call it the maintenance settings, is to prolong the PFS, to really prolong the time the patients are following responses to chemotherapy, to prolong the time before they relapse. The subtrial one is randomizing 60 patients after responding to chemotherapy. They're randomized to COM701 versus placebo. There is no standard of care for this patient population, and the readout is expected in Q1 2027. Following successful readout, clearly showing that COM701 has monotherapy activity in this patient population, then we can definitely think about path to registration as monotherapy. We can think about a backbone for combinations, for example, with ADCs. Obviously, following positive results with clear monotherapy signal, we can think about going earlier in ovarian cancer, later in ovarian cancer, and also to other indications in which we saw signals for COM701. Rilvegostomig.
Why do we think rilvegostomig can succeed where other TIGIT have failed? First is the antibody format. What AstraZeneca did is they took COM902, our antibody, and put it as part of a PD-1/TIGIT bispecific antibody. One arm binds PD-1, like Pembrolizumablizumablizumab , and the second arm binds TIGIT. The antibody binds two targets. What AstraZeneca have shown in a very translational ex vivo system, meaning taking tumors directly from patients and testing them with the different drugs, and this system was shown to correlate very nicely with actual clinical activity. They showed that this bispecific format of rilvegostomig, through its cooperative binding, is not only more active than PD-1 in this system, but it's even more active than PD-1/TIGIT two-drug combination that was tested thus far by others. The bispecific could be more effective than PD-1/TIGIT combination. This is one.
Looking in the whole clinical strategy of AstraZeneca, and I'm not speaking on their behalf, but really looking at what seems to be a very comprehensive plan, they use rilvegostomig in a way trying to replace PD-1, PD-L1 IO backbone across different indications, across different combinations. Notably, I would say, the ADC combinations, Enhertu, Dato-DXd, that was not tested by others. Also the clinical trial design that they employ, in some cases, it seems, again, looking from the side, they learn from the mistakes of others and are basing the clinical design based on the mistakes of others. We believe rilvegostomig could succeed where other TIGIT players failed due to the antibody formats, the combination strategy, and the clinical trial strategy. From this collaboration, we received to date $95 million. We're eligible for additional $195 million in milestones.
Most importantly, with this extensive clinical strategy, AstraZeneca previously predicted more than $5 billion on our peak sales for rilvegostomig. Being eligible for up to mid-single digits to royalties from this potential blockbuster drug could really drive future financial stability for Compugen in a very meaningful way and will allow us to continue and use our engine to bring more assets using this financing. While the phase III studies are ongoing, and the results will be after 2027, meaning 2028 and on, there are multiple other trials ongoing, each of them showing different results. Next week in AACR, actually, there are five presentations, both research and clinical, that I believe could shed additional light on why us too are so excited for rilvegostomig. COM503, called today GS-0321, another demonstration of our computation capability.
This program, which targets IL-18 binding protein, which is an inhibitor of a cytokine, really a first-in-class program, but also first-in-class approach. Nobody ever tested such an approach that has the potential to really overcome a lot of the challenges that others had, trying for years to engineer cytokines to make these potent immune stimulants into drugs. While there are some approved cytokines in the clinic, most of the attempts to reduce toxicity while maintaining a good therapeutic window have failed to date, and we believe this program has the potential to really make a difference. Gilead thought so as well, and we licensed the program to Gilead, we received to date $90 million for this program that was licensed at the preclinical stage. We're eligible for additional $758 million in milestones and up to low double digits to royalties.
Compugen is responsible for the ongoing phase I trial that started. The first patient was dosed in the beginning of 2025, and Gilead will take it then to the later-stage trials and hopefully to commercialization. To summarize, I think that Compugen is really well-positioned today with a solid financial stability and then also the computational engine, a validated computational engine that could use this financial stability to bring more assets to continue and build our own pipeline, while in parallel, the validating partnerships are ongoing, hopefully with an approval on the near horizon. I would like to thank you for your attention, and if you have any questions, please don't hesitate to send us.
Thank you, Eran. At this stage, we will open the floor up for questions, but while we poll for additional questions, maybe a couple from us. Maybe I'll start with something a bit fundamental. Investor disappointment in this space has been driven by the fact that all these new IO agents, and I'm speaking broadly in the last decade, haven't really reached this sort of breadth of activity that we've seen in PD-1s in general. Is there an underlying biological reason why the PD-1 axis seems more amenable?
Yeah. It's a great question, and I think that overall, there are specific types of patients who respond to PD-1. These are specific indications, specific patients who have probably normally some baseline inflammation in tumor microenvironment, some baseline T cells. This patient profile, either they respond to PD-1 or to date, when added other IO agents, if they didn't respond, meaning the tumor microenvironment was not prone to respond to IO in general, and they didn't respond also to the combination. This is, I think, one of the reasons that PD-1 obviously is very dominant. All the IO agents that arrived afterwards had, for example, TIGIT. TIGIT did show activity in randomized studies. It just was not sufficient to achieve a phase III success.
LAG-3, by the way, I've no choice, much more active, and it did, with a properly designed trial, eventually, it was approved in melanoma. If we talk about and take this into the two assets, I would say that the two checkpoints that we work on, PVRIG and TIGIT. On the one hand, PVRIG, first of all, did show some monotherapy activity. That is something that was missing for many, from some of the other IO agents who failed. We didn't test many patients, but we did see monotherapy activity. In general, the biology and the signals that we're seeing in various other indications, less inflamed indications, is very unique, and we'll have to see how it plays now in a randomized study, in the MAIA-OV study. For rilvegostomig, yes, TIGIT, I think it's quite clear, is active.
Definitely not as active as PD-1 monotherapy compared to chemotherapies in these inflamed indications. TIGIT is active in places where PD-1 is active. Now I think it's really the game of having the right format, having the right combination, and also designing the trial properly, understanding the magnitude of effect that TIGIT can bring.
Maybe as a lead on there, do you think that other programs just didn't design the correct study?
I think it's, first of all, most of the other programs who failed have an Fc-active TIGIT, which I really think that even though the initial result of the Fc-active looked good, eventually they had worse safety, that high discontinuation rate. They took the activity of TIGIT, which is not huge, and in a way, eroded it with the safety and maybe a bit less of efficacy. This is one. I think having the right drug format is one thing which is important. Yes, also having the right combination and the right trial design, definitely is critically important. I think along the 11 phase III trials of AstraZeneca, you can really see, and some of them are comparing head-to-head versus Keytruda, for example. They, I think, were rightly powered with selection of PD-L1, which is important, that not everybody did.
Also some of the trials are really doing this combination, also with the rilvegostomig plus ADC, and they just use it as an IO backbone. They don't even really compare. This is a bit of a different story. Some of these trials don't really measure TIGIT activity directly. Because it's a bispecific, it's a bit of a different story also for AstraZeneca in this case.
Maybe switching gears here a little bit, a bit on the rationale of moving forward in platinum-sensitive versus platinum-resistant, not the most common path forward.
The rationale was that after we saw quite clearly, again, not many patients, but we saw the signals in the last line, especially, for example, PD-1 or PD-1 even plus TIGIT in the past showed zero response rates in PD-L1 negative. We saw responses in PD-L1 negative ovarian cancer, again, last line, platinum resistant with COM701 in one patient and two stabilization of disease and in combination. We're quite sure that we have a signal there. The patient who did respond, the response was very durable. In the last line patients, when you have such highly progressing tumors, some of them never even had a chance to respond to IO, few, and this is what we tried, IO-only treatment. We had, I think, two options.
One was what Merck did recently, to combine with chemotherapy to provide some kind of debulking to add it to the IO. We decided not only to use debulking, but also to go earlier, to go into patients in the second, third line, platinum-sensitive. They had less rounds of treatment, so their immune system is less compromised. We took only patients who are responding, so they have debulked tumors to begin with because we take patients to respond. We really give the opportunity to COM701, with its excellent safety and durability of response, to prolong the response with the chemotherapy.
If I'm to rephrase this, you're basically looking at patients with a higher T cell quality.
Absolutely
As it relates to MoA, basically like consolidation.
Yeah
Debulk, there's less target cells to go after, basically.
Yeah, I think this concept was shown also by others, for example, by AstraZeneca in their strategy in IO, going earlier into maintenance, adjuvant, neoadjuvant. We didn't invent the concept of going earlier, and I think really immuno-oncology has better potential to work in earlier settings with less compromised immune system, again, and also with low tumor burden. I think this is what exactly we did. Yeah, it's very interesting to see the results.
Okay, that all makes sense. Why bispecific, instead of just using both agents at the same time for rilvegostomig?
Yeah, it's an excellent question. First of all, as I showed in the presentation, when you use a bispecific, you have this kind of cooperative binding. What AstraZeneca have shown that you have induced the gains of affinities, you are first docking with the high-affinity antibody that is derived from COM902, our antibody, and then co-locate of PD-1 and TIGIT on the same cells. This cooperative binding, eventually, and this is what they showed in this ex vivo system, is more active in PD-1/TIGIT combination. I think this is one aspect that the bispecific seems to have, and we need eventually to see it in the clinical setting, so obviously, but seems to have the potential with its unique cooperative binding to deliver more activity than just PD-1 plus TIGIT combination.
Going into the clinical combination strategy, when you have a bispecific, in some cases, you don't really have to prove contribution of components because nobody can ask you to take out TIGIT and show it's active. Since the drug is using also the Fc-reduced approach that we utilize in contrast to most of the competitors, so the safety is excellent. Now you have a drug which is safety that until now seems quite similar to PD-1 blockers alone, which enhance activity. Now testing it in the trials, again, in some of the trials, for example, in TROPION-Lung10, they are comparing with rilvegostomig plus ADC versus Keytruda. They don't really have to prove. I mean, they use rilvegostomig as a IO backbone, if you may, a better type of PD-1, and combine it with ADCs and compare it to Keytruda.
There are other trials with Enhertu, for example, plus rilvegostomig versus Pembrolizumablizumab plus IL-18. Overall, I think that the strategy gives potentially more active drug, but also much easier combination strategy and clinical path forward in clinical trials.
Great. Maybe as a last question on your Gilead program, can you walk us again through the differences on your IL-18 strategy? Why do you think it would be more effective than just providing IL-18 or is this all a safety profile question?
It's a matter of therapeutic window. I mean, eventually cytokines, if you put T cells in a tissue culture plate, you put IL-18 or any other cytokines on them, they are exploding with activity. Typically what people are doing, and actually IL-2 is an approved drug, right? You put it into veins of the patients and then you have systemic side effects because the first thing that the T cells in the blood will see is the IL-18 or IL-2 or any other cytokine. Sometimes some of it can reach sufficient level to the tumor to induce the actual activity you want to see. What we are doing, we're utilizing the fact that in a tumor microenvironment, naturally, there is high level of IL-18 blocked by IL-18 binding protein, and it's very low in the blood because it's induced by inflammation.
We use a blocker antibody. We're not using a cytokine. We're using a blocker antibody. Again, this is a first-in-class approach. The concept itself was never shown by others. We are inhibiting the inhibitor. What we have shown pre-clinically, that we're modulating only the tumor and not the periphery. This is a conceptual difference in the approach. We are utilizing the potency of the cytokine, which is naturally a tumor microenvironment. By that, the therapeutic window, again, at least pre-clinically, is completely different from other attempts of the cytokines, using recombinant cytokines, others have tried.
Right. I don't currently see any further questions from the audience. Eran, I would like to thank you for attending today.
Thank you.