Compugen Ltd (CGEN) October 2026 Conference Transcript

Good morning, and thanks for joining us to have a conversation with Iran Ofer, President and CEO of Compugen. Compugen uses computational discovery to identify cancer immunotherapy targets and advance them into both own and partner drug programs. It's proprietary anti-PBRIG program, which is COM701, is being evaluated as maintenance monotherapy in randomized MAIA ovarian study with interim median progression free survival data expected in the first quarter of 2027. So partnered assets include AstraZeneca's PD-1-digit bispecific Rilvegostomac, now in 12 phase 3 trials, and Gilead's GS-0321, an anti-IL-18 binding protein antibody in phase 1. And at its October 5th event, the company also outlined an expansion of its Unigen discovery engine into tumor-associated antigens. So to discuss what the next data can establish and how discovery translates into economic value, let's get started with Iran. Iran, I'm glad to see you and appreciate you accepting our invitation to talk to our audience this morning.

Thank you for having me, Artie. So Iran, across COM701, the AstraZeneca program and also the 0321 with Gilead, what's the clearest way for investors to separate Compugent's scientific differentiation from what Compugent actually owns? so um so out of these assets com71 is wholly owned by us while revagostamig and gs321 are partner programs that provides potential milestones and obviously royalty economics um the co-incentive part is that all three are coming from the same ai power computational platform unigene that this is the tool we use to identify new drug targets in the field of immuno-oncology.

Perfect. So at your recent October 5th event, you outlined how unigens expansion into tumor-associated antigens, especially with T-cell engagers and targeted immune activation is being the first focus. So with only two T-cell engagers approved so far in solid tumors, you know, why is the target quality the bottleneck? And also what What positions Compugen, you know, to solve this?

So, if we talk specifically about this in engagement, but this could be true also for other immune activators or immune engagers, that these are very powerful molecules and they can identify very low density of antigen. So, for example, in hematological cancer in which TCs are approved for a while and it's broadly used, the markers for the cancer, for example, CD19 on oncogenic B cells is a very clean marker to use because it depletes the cancer. In this case, you also deplete the normal B cells, but one can survive without normal B cells. For solid tumors, this is very, very challenging to find a marker that will be expressed only in the tumor and nowhere else. And typically, these kind of organs, if you do have toxicity, it's an important and critical organs. So DLA3, for example, which is a very clean target, the target of tarlatamab, when you have a good target, then T-cellin just can be approved, and indeed tarlatamab is approved, and it revolutionized the treatment of small saline cancer, for example. But the challenge of really finding such rare, extra clean, and specific tumor antigens is a huge challenge in the field and we believe that CompuGen with our database and expertise for finding first-in-class assets for immuno-oncology in general but we build along years and all this know-how of how to deal and how to validate and novel targets expanding now into UnigenX that explores and uses multiple data layers to identify such new TIAs we believe it's a this is a place where we can really bring more value in an area of huge unmet needs.

Okay. You know, you said a few targets have passed the computational stage and are currently moving through validation and that you will be disclosing timelines once an asset is de-risked enough.

So what is that bar that you're talking about and would you develop the first TAA program internally or would you be seeking a partner for that so in general when you're talking that such a new target we say that the first-in-class approaches the initial stage is yes identified computation and this is where we start but for us to really commit and discuss timelines with the investment community we want to be really sure that the target is passing all the stages of validation and this includes not only computational this is just the initial bar you then need to take and validate experimentally and to bring all the therapeutic package and we put very high bar for ourselves before we take a target to the clinic we really want to make sure in in ex vivo systems human systems that the target is meeting the bar to take it into patients and once all of that is clear and we have very high probability that this target is indeed something that is going into patients this is the the stage in which we disclose it and and give guidance on specific milestones on for the specific targets about partnering or moving internally this absolutely depends on on the target on this on the company needs and for some programs it might be the right way to go to partner and to let big pharma push it aggressively and for us to get milestones and royalties as we did with gilead and asperzenica and some programs like com701 we take ourselves to the clinic if we think that we have the capabilities to do the right experiment and to bring value for the target ourselves.

Okay. One additional question in terms of Unigine and, you know, general purpose AI tools that seem to be progressing really at a great speed. You know, as you said, Unigine actually has been, you know, the tool that identified all three of them, COM701902 and also the GS0321. on. You know, as I said, you know, there are various general purpose tools that are being developed, you know, by different parties now. So what is it that actually makes it unique? You know, is it proprietary data or the experimental feedback that you get that makes your system unique?

And it's difficult, you know, for general purpose tools to actually identify a new target. it's an excellent question and it's actually both i mean the the biology i mean you can use the eye for many things in in the r d process of of a drug you can maybe enhance the clinical trial design you can potentially and maybe do a faster antibody discovery stage but the first stage of really identifying the target itself which is the target that we think that if you target you can and make meaningful impact to cancer patients, the AI is very, very far from just pushing a button and getting a target. And our know-how that we developed along the years, it includes the proprietary databases, the know-how to do experimental validation for novel targets, and eventually how to take such targets into preclinical and clinical packages.

This is very far from just being achieved by by pushing a button in the general purpose tools so definitely the know-how includes both the algorithms the data set billing along wheels and experimental expertise okay so moving on into your programs now you know especially the maia ovarian study you know this study is designed to randomize 40 patients to com 701 and 20 to placebo as maintenance therapy in relapsed platinum-sensitive ovarian cancer. So in plain terms, why should blocking PVRIG as maintenance matter in this population, and what does this study test, you know, that your earlier platinum-resistant cohorts could not test?

So PVRIG is a checkpoint, but the biology is very different from PD-1 in digits. That unique biology probably is the reason why we see in general, PVRG blockade activity, COM701, in places where typically checkpoints are not working, like ovarian cancer. Specifically, the PVRG pathway is very high in ovarian cancer. And actually, it was a target indication for us from the beginning. So one reason is the biology. And then the actual clinical signals. When we treated last line, heavily pretreated, platelet-resistant ovarian cancer patients, clearly a subset of patients had clear benefit. and the activity was durable. These patients had durable, very favorable outcome and clinical benefits. And the rationale of the Maya study and the reason why we think that in the earlier line, second, third line, platinum sensitive maintenance settings, we could now enhance this response or in simple words, getting more patients, ideally more than 50%, because this is what you need to do to improve median, to get more patients responding compared to what we saw in the last line and this is the exact rationale for the Maya study. We are going earlier into patients with second, third line platinum sensitive. The immune system is not compromised by repeated rounds of chemotherapy like the patients we treated in the platinum resistance settings. We are taking only patients with low tumor burden. We treated patients with extremely high tumor burden. Now we take only patients to respond to the platinum. So there are patients, some of them don't have tumor at all and they have low tumor burden. And then we're also excluding patients with liver metastasis, because even in the last line patients, when we excluded retrospectively patients with liver meds, we already have had almost 40% clinical benefit, almost reached the 50% already in the last line patients, but doing only exclusion of liver meds. So overall, really, the design here is to maximize the probability by selecting the patients that are really most fitted to respond to COM-7-1 mode of action. And this is the goal of the study. It's a randomized study that should be informative on COM-7-1 monotherapy activity in these settings.

Perfect. Right. So at your CAROL event, you know, Dr. Yeku said he wants a comfortable buffer between the treated and the placebo arms rather than a specific PFS number. And that time to next therapy on the PFS2 on the next platinum really matters in maintenance. So will those measures be part of the interim disclosure? And how do they factor, you know, in your own decisions?

So in this study, in the interim analysis, obviously what we will report is the activity of COM701 in this line of treatment. So the question we are asking is in the second, third line, after platinum is inducing responses, can COM701 prolong the time by which these patients are surviving without the cancer relapsing? Obviously, eventually for approval, like for every drug, you need to see the long-term follow-up. You need to see how the patient is also responding to the next line of treatment, which is always important, and this will eventually translate to over-survival. So this is true for every drug, but from what our understanding of the mechanism of action of COM701, if we will see a good enhancement of PFS in this trial, there is no reason to think that this will not be translated to long-term benefit, it, but this will obviously have to be verified in longer-term follow-up and eventually in a phase three study.

Okay. So talking a little bit about, you know, what to expect on the commercial plans, you know, Dr. Yekou estimated 40 to 60% of these patients could be potential candidates for maintenance option other than bevacizumab, right?

How does that shape the commercial opportunity and how do you plan to position 701, especially when we are seeing a disease you know moving into maintenance there yes so um as dr yeko mentioned indeed in the first line maintenance patients are receiving power or vivacizumab or sometimes both and when they reach the second or third line actually there is nothing really approved that's why we can do a trial with the control arm is placebo and not vivacizumab but yes in many centers and i think that we mentioned that as well it really depends on on the center and the in the philosophy of the center sometimes they give the system but overall many of the patients already received and progress on the system many of the patients cannot tolerate the system because of medical conditions so overall we estimate that the initial population for com 701 if it's positive in this trial is around 10k per year, of course, upon approval and upon the pay-through success. About the ADCs, and this was a major topic that Dr. Yaku discussed, the ADCs have definitely shown activity in the last line of platinum-resistant ovarian cancer settings. ADCs eventually are chemotherapies, and there are two main problems with ADCs. One, that you You cannot sequence, the current thinking is, because most of them carry the same payload, so you cannot sequence them. So if you use it in the maintenance settings, you now don't have it as a tool for the platinum-resistant settings. At least some physicians think that they prefer to keep ADCs for the later stage settings because of that kind of chemotherapies. And for me, what is even more convincing, that there is a reason that patients cannot take more than four to six cycles of platinum. It's too toxic. That's why you need the maintenance drug. Patients receive four to six acres of platinum, they respond to the platinum, but they are not taking more platinum because they cannot tolerate it. So many of these patients will not want to receive another type of chemotherapy, which ADC is, or they cannot tolerate ADCs. So even if ADCs will be approved, and as Yaku mentioned in the recent ASCO, there was a failed trial for ADCs in a treatment plus maintenance settings, O'Brien Platinum said to ovaric cancer. So even if they will be approved, there are going to be a large fraction of patients that will prefer for maintenance to maintain the quality of life and to get a break from the chemotherapy and get COM701. So from what I know, COM701 is very differentiated from all of these phase three trials. All of them use ADCs and all of them, the safety profile, some more and some less, but eventually all of them are very different for COM701. So, if COM701 is able to meaningful use, prolongation of PFS, I think the commercial opportunity is quite clear, even if ADCs are eventually approved.

So, moving on into the partner programs and starting off with AstraZeneca, you know, as I said, AstraZeneca is running 12 phase three studies. Can you tell us, like, what's the, you know, what should we expect in terms of these programs, especially what, you know, which will be the first one that we'll start to hear?

And also in terms of milestones, you know, what should investors really track and, you know, what steps, which steps actually translate that into economics makes for confugent rather than just seeing another clinical activity so basically what all of us are waiting for is to see the phase three studies there are 12 phase three studies ongoing across multiple indications in different combinations and and the readout for those according to astrazeneca starting after 27 probably there could be some trials to be read already in 28 if you're doing some high level calculations in a very good scenario that is currently not the guidance of AstraZeneca. There could be interim and early results also along 27. But let's discuss the guidance, the ones that AstraZeneca are guiding for. So we should expect to start to see phase three results from 28 and on. Now, remember that we are eligible for additional, we received to date $105 million from AstraZeneca. We're eligible for additional $155 million, $195 million, sorry, in regulatory and commercial milestones, plus mid-single-digit tiered royalties from their sales. And they're predicting a non-risk-adjusted revenue of above $5 billion. So what you also said, that when we negotiated the monetization of a small portion of the royalties last year for which we received $60 million, we also said that we negotiated additional $25 million increase in the BLA acceptance milestone. So to wrap it up, we expect to see some phase three results probably from 28th and on. And the next milestone is 25 million plus, whatever was there before, for BLA acceptance. Total milestones, 195. And then, and I think more importantly, seeing all this broad development plan mid-single-digit third-worldies.

In terms of Gilead, you know, they have paid Compugent $90 million and Compugent remains eligible for up to $758 million more. So the O321 targets IL-18 binding protein rather than administering, you know, cytokines directly. So what pharmacodynamic and safety findings would actually establish a usable therapeutic window, you know, based on whatever information that both you and the partner have provided.

Yeah, so this program is a first-in-class approach and another demonstration of our ability using AI-powered computational platform to identify new biologies and new targets. And the program here is blocking IL-1 binding protein, therefore releasing other cytokine called IL-18 to perform its anti-tumor activity by enhancing anti-tumor immunity so what we would like to see pharmacodynamically is basically that we see active IL-18 that we are able to release IL-18 in the downstream effects of immune activation this is obviously the goal and this should lead also eventually to anti-tumor effects and like for any drug we want to see that we can do that in tolerable doses but we're able to have a therapeutic window in which we are not inducing significant antitumor immunity without inducing in parallel significant toxicity. Most cytokines suffer from a very problematic therapeutic window because they are giving systemically to the blood, and there you are, first of all, inducing immunity in the blood and in non-tumor tissues before you are inducing activity in the tumor. And the basic of this program, to begin with, and this is why Gilead is paying $90 million on a preclinical asset is that according to the biology IL-18 and IL-18 binding protein are very high in the tumor and lower in the blood so we expect to see in contrast to all the other cytokines that most of the activity will happen in the tumor and for that you need to have pre and on treatment biopsy to really see that you're modulating immunologically the tumor and the effect is not translating to significant side effects in the blood then this is our expectation for the mode of action okay so for for both for not only maia but also other you know discovery programs you know where is this uh the the computational

platform simply just a discovery tool and where if if anywhere at all um does the model output both um i mean output uh influence a clinical or a regulated decision you know that that you would require a separate validation package if you really need it.

So the main goal of Unigen, and this is where really our expertise is, is to identify first in class targets. So identify the targets and then we take it. But the initial computational package includes also, I mean, we always start with them in When we start a program, we already know where do we think we'll take it in patients. So I mentioned before that ovarian cancer was targeted indication for PBRG, for the COM701 program, from day one. So part of the output is also which patients we think are going to be benefiting and which indications will benefit. So definitely patient selection and biomarker is part of the computational platform. We also use it to analyze along the clinical trial, to analyze the data and try to further identify the patient was probably to benefit. But eventually, eventually, the clinical and regulatory decisions are based on validated assays and standardized packages like also like non-AI companies are doing. Eventually, the clinical development is doing in a conventional manner while relying on the computational foundations.

So, Yiran, just to close this conversation, which single achievement over the next, say, 12 months would change investors, how they think about Configen? And what are the two or three milestones that we should be looking out for over the next, say, six to 12 months?

So definitely the first one is the Maya readout by the end of Q1 27, which is a randomized study with a clear readout that should be very informative and a very important milestone. If we look at 12 months and maybe a bit more, talking before about the potentially phase three readouts for AstraZeneca from after 27 to 28 and on. So this is definitely a place where the first phase three success is going to be very meaningful. And investors that will see it, hopefully will see a phase three success from Zeneca, will start to understand that Rilvogostamik is different from the other digital programs. And then we don't have guidance yet, but if you look at the coming 12 months, there could be also some developments in other early pipeline assets in COM503. But for now, we are not giving specific guidance for that.

Thank you very much, Iran. Thanks for taking time to be on this call today. And good luck. Appreciate your time.