Everyone, we're on day two in the afternoon session of Citi's Biopharma Back to School Summit here in New York City. I'm Yigal Nochomovitz, senior biotech analyst at Citi. We have our next company, which is IDEAYA. Great platform, great pipeline, and we've been covering them for a number of years, and it's really been a great journey, and a lot more to come. I'm pleased to introduce Yujiro Hata, who's the Chief Executive Officer, founder of the company, and then also we have Josh Bleharski, who's the Chief Financial Officer, and Daniel Simon, the Chief Business Officer. So welcome, all of you. So Yujiro, a lot's going to happen in this next several months with regard to readouts. But before we get to all that, for those a little bit less familiar, just identify the key programs that you're focused on, and the key opportunities in the different market segments that you guys are pursuing. Yeah. First, thank you, Yigal, for the introduction, and thank you to Citi for the invitation to participate again this year. So IDEAYA Biosciences, we're a leading precision medicine oncology company. I think where we're unique is that we have a very late-stage asset in darovasertib. The lead indication there is metastatic uveal melanoma. We are in the process of submitting the NDA under RTOR. The first three modules have been submitted, and we have the final module to submit here very soon. Most of our analysts are projecting a first half of next year commercial launch. So again, a very late-stage asset. In addition to that, we have two additional indications we're pursuing, phase III studies, both in the neoadjuvant setting, as well as in the adjuvant setting. So a very broad set of clinical activities to also expand the opportunity into the earlier line settings. In addition, we have IDE849, a DLL3 Topo ADC. We announced recently that we had a successful Type C meeting with the FDA to advance into a phase III registrational study for monotherapy approval in extensive stage small cell lung cancer. Here, the primary endpoint for accelerated approval will be overall response rate. For full approval, median overall survival. Beyond that, Yigal, as you know, we have a very broad set of assets in the area of MTAP deletion and CDKN2A. We've announced two recent collaborations with Roche, with their both our pan-RAS inhibitor and their KRAS G12D inhibitor. And that will be exclusively focused on the area of pancreatic cancer. So for those that may or may not be familiar, the overlap between all of the KRAS mutations with MTAP deletion, we believe is about 40% of that population. So a very significant population. We announced those relationships in the last few months. We have more in the pipeline, but I think if we covered that, we would have done very well. Okay. All right. Let's go to the first thing. You said you did three out of the four modules, and potential launch in the first half of next year. What is left to do with this? What is the last module? What is left to do? It seems like you would need a PDUFA date assigned pretty soon in order to do the one half 2027 launch. You have the RTOR and you have the rolling submission, so you have some leverage there. Yeah, that is correct. We have fast track designation, so we will have expedited review through that, and as you noted, RTOR in terms of the rolling review. There is no specific gating item at this point. It is largely QC that has been ongoing, related to summaries of both clinical safety and clinical efficacy, as well as the clinical overview section. So that is the good news. We are getting very close. There is no specific gating item that is right now in front of us. It is mainly a QC process that is ongoing. Okay. But the stated guidance to submission is this year? Correct. Oh, okay. All right. Related to that, obviously, there is an important data set in the HLA A2 positive group. Can you speak to that? I believe you are going to be showing some of that data relatively soon. There is a potential for getting that into the marketplace through the compendia potentially as a strategy. Can you elaborate on that? Yeah. We have two fairly sizable updates that is going to be upcoming at ESMO. First, Yigal, as you noted, we have about 100 patients worth of data in metastatic uveal melanoma in HLA A2 positive, so this is where KIMMTRAK is approved. We will share response rate PFS and median overall survival data. We will break that out into first line, and then all patients, which is consistent with how we have presented in the past. Our sense is that the biggest investor focus is, even though it is going to be in the subset of patients, which is what does the frontline OS data look like? As you can appreciate, that will be the second time we will share OS data in the first-line setting. Just as a reminder, at ESMO last fall, we did share OS, which was just over 21 months. Here, we think the control arm will likely be in that 12- 13 month range. For neoadjuvant, we will also have about 100 patient data set that is presented at ESMO. This will include both the iodine plaque therapy patients, consistent with prior disclosures related to eye preservation rate, predicted visual acuity. The actual visual acuity will be still too early. My understanding is we will have some commentary related to relapse. I think that should be helpful as well. We will utilize this data with a strategy towards a NCCN guideline usage in the neoadjuvant setting with the anticipation of commercial launch in the metastatic setting. Okay. For the HLA positives, that would be also with this compendia strategy, or that could potentially work its way into the label if you fold that into the rolling submission? Is that a plan of action, or you may not necessarily need to do that? Sure. Josh? Okay. Yeah. We've had a lot of good dialogue with the FDA and, through the RTOR designation, have had the chance to share some of the data that we'll be having at ESMO, and they've seen that data now. I think our sense from the interaction is that they're encouraging us to try and seek a broader label. I think they understand the unmet need there. I think the data sets are remarkably consistent and I think our intention is to have that dialogue with them and see if we can get an HLA agnostic label at approval. That's a little bit of an evolution from where we started, but I think it's really a testament to the drug's benefit and the unmet need, and I think the FDA has been very aligned with us in trying to move this forward as quickly as possible so we could get it- That sounds like a relatively new development. Yeah, it is. I hadn't heard it quite like that before, so that sounds good. Yeah. Yeah. Okay, all right. Yeah, and maybe part of that, Yigal, is we did have their pre-NDA meeting. Yeah. That was largely the focus of the pre-NDA meeting topic. That is why now we had that meeting, we got that feedback, and so you are seeing our communication evolve from that feedback. Which was recent, I gather. Correct. Okay. Yeah, it was recent. Okay. So then for neoadj, as you pointed out, compendia listing, but does this mean you may take another think as far as what to do with OptimUM-10? Yeah. I think in any scenario, we'll submit for compendia for guideline inclusion based on the data we've generated already. What we're thinking about now is just whether that's enough, and whether in light of some of the enrollment delays we've had with the 10 study that we've talked about, would it be better for us to sort of pivot, rely exclusively on the compendia strategy for the neoadjuvant subset, and reallocate some of the funds we'd set aside to run that registrational study towards other high priority programs in the pipeline? We have not yet made a determination there, but we are actively kind of thinking through the options and trying to do the analysis to support a decision, which we'd expect would be before the end of the year. Okay. The gating factors there are just showing the data at ESMO, getting more KOL buy-in on that data, I gather, or? Yeah. We've had the combination of our own market research and some of the work we've done with KOLs and people who have sat on the NCCN panel. Our take of that is that there's a reasonable likelihood that we would get included in the compendia. If that were the case, it's a question of what would that mean commercially. We're doing some of that work to try and understand that a little bit better. But suffice to say, we think that may be a viable route in light of some of the other challenges I highlighted, and could allow us to invest in other things in the pipeline that we'll get to. Okay. All right. Yeah. Related to all that, can we talk a bit about just the commercial setup, the build? You're getting very close, obviously, to a launch, which would be great, the first launch. So, where are you with the commercial build? Who have you hired? All the market intelligence work to set yourself up to launch? Yeah. Josh and I could tag team on this. I think it's a fairly straightforward build-out. We're just focused on the U.S. launch. As you know, Servier is handling ex-U.S. for us. The sales force size, we believe approximately 25 individuals, very similar to what Immunocore's done with KIMMTRAK. The base commercial leadership team has been hired, so that's all set. Everything on the supply chain side is also very much on track. We do have an early scenario commercial readiness and preparation that we'll be prepared for. But everything is all systems go right now. And anything early that you can communicate in terms of how you see the early launch progressing, relative to the obvious competitor, KIMMTRAK? Do you mean specifically in the A2 population? Or sorry, the positives? Yeah. The positives. Yeah. Yeah. Look, we have seen and heard a lot of enthusiasm from the people that we have talked to about the profile of the drug. Obviously, when we were at ESMO presenting this, the discussant after our presentation noted that this is a big advance for the field where there is a lot of unmet needs still. We feel pretty good about the uptake, particularly in the A2 negatives where there are no approved therapies, but certainly in A2 positives, where there still remains a lot of unmet needs. Time will tell, but we continue to believe that the drug delivers a lot of benefit and, for patients where there is quick progression or even in the post-KIMMTRAK setting, we think that this is a nice viable alternative should it be approved for those patients, and we would expect to see reasonably quick uptake. Okay. That sounds good. Let's move over to talking about DLL3. First of all, in terms of the Type C meeting and the alignment on the phase III design, did that sort of square with your goals for what you wanted to get agreement on there in terms of the endpoints you mentioned, obviously ORR and OS? Yeah, it was, Yigal. I think the biggest question for us when we had the Type C meeting was around how do we think about pre-IMDELLTRA versus post-IMDELLTRA patients? That was really the key question for us. The feedback was very clear that if we did do a pre and post-IMDELLTRA study, that it would be viewed as more of a heterogeneous population, at least for the U.S. approval portion. We just felt it was a cleaner study to do it as a purely a post-IMDELLTRA study population. In addition, we also did it this way with the mindset that you always want to do a study that's going to be somewhat relevant at the time of hopefully potential approval. We did anticipate that IMDELLTRA would hit their endpoint for frontline approval, and as you probably know, earlier this week, they announced that they did hit their OS endpoint for frontline maintenance. That exactly, I think, further just reinforces the trial design that we're pursuing. I think the main questions we've been getting is, are we seeing responses post-IMDELLTRA? Is there any concern that you might have DLL3 downregulation post-IMDELLTRA? I think there we can confidently say at least what we know of, that that's not a concern. Second, we are clearly seeing responses in the post-IMDELLTRA setting, which gives us confidence in the study design. The last two parts I'll mention is we are interested in neuroendocrine carcinoma. We believe we're clearly seeing monotherapy activity there. We do have another meeting with the FDA set up in the fall to get more clarity on what a registrational path will look like there for monotherapy, both accelerated and full approval. Here, it's a lot less crowded of an area, a lot less activity with antibody-drug conjugates. We don't have really IMDELLTRA here as well. There's a lot more white space. Lastly is around frontline small cell lung cancer. There, I think we can maybe talk about it more, but we think we have more work to do, thinking through what the right strategy is for frontline. What are the factors that need to be weighed on the frontline that you are weighing? Yeah. First is the landscape is evolving rapidly, right? As we just saw, IMDELLTRA hit their endpoint in frontline maintenance. They have another randomized phase III study that has not read out yet in frontline induction. Those type of things, it would be nice to get some sense of where that is all coming out, just so we know we have the right things in the comparator arm. That is one. I think second, when you think about what you combine it with, at least our view is PD-L1 is fairly table stakes. I think there has been some questions around plus or minus chemo. Obviously our antibody-drug conjugate. Then, of course, as you know, Yigal, we are quite interested in the PARP combination that we are doing, which is a novel combination. If that does enhance durability, that would clearly provide us a differentiated path forward than others. If we did do a PD-L1 plus ADC versus PD-L1 plus chemo, and now with possibly IMDELLTRA in the control arm, that may be a high bar study, right? I think that is what we want to think through before we just sort of launch into the trial, that we have just a better sense of all these various parameters. Have the KOLs suggested even combining IDE849 with IMDELLTRA, or has that come up or not really? Yes, it has. Yeah. That would be another- Yeah Piece of the story, right? Once you do that, you have to think about a lot of other considerations from cost, logistics. I think there's a lot of things that you would have to think through there. That's why I think as we mentioned, versus sort of rushing into a frontline study, just making sure we have clarity on what is the highest POS study. In the interim, making sure that we get these monotherapy later line accelerated approval studies off the ground first in small cell. Then next we'll evaluate if there's a viable path forward for neuroendocrine carcinoma as we collect more information on this question with frontline. Because, yes, you could consider it. Yeah. Well, because some other companies are combining, as you know, right? That work has just begun. Yeah. It doesn't look like it's going to be in the first wave of those frontline studies. Right Just because that dosing just began. But yeah, those are the questions. But we know in the past, Amgen had done B7H3 ADC combinations. Those were halted. So why were they halted? So I think those are the type of questions I think are going to be important to have answers to. But for neuroendocrine, it's more white space. So there you could do frontline more easily or not necessarily, or? There's not a real, depending on which subtype of neuroendocrine carcinoma, but some of the subtypes we're considering, there's not really a clear standard of care. Yeah. So yes, absolutely, there's more white space. You see less activity from Amgen there, as well with IMDELLTRA in the NEC space. As you know, there's not as much activity ongoing with other ADCs. So there could be a unique opportunity for a DLL3 triple ADC in the NEC area. As I mentioned, we are clearly seeing activity there as a single agent. Okay. The question of the dose. So 2.4, 3.5 are the finalists, I guess. So what's left to determine there? I mean, 2.4 looked pretty good. What else do you want to know about 3.5? Yeah, we know that our closest peer company has picked an expansion dose of 1.6 mg per kg. At 2.4, we're already 50% higher. At 3.5, we would be over 100% more. Which means it appears that we could deliver more payload. Then now the question will be, how will that translate into a response rate PFS and ultimately OS? But I think the good news is, we did share all of the data we had with the FDA as part of the Type C meeting. They were in agreement us that those should be the two doses we should pick from. It appears that we can go higher the more hungry decided to expand at 2.4. So if we can deliver that payload and we could do it safely, then I think 3.5 would be a great dose. But 2.4, as you noted, Yigal, is also very viable dose. Is this sort of under the Project Optimus umbrella, or is it a little bit separate? Actually, because sometimes the FDA just asks for this because check boxes, but you guys already know what you are going to do for phase III. In this case, it sounds like there is more of an actual debate as to which one to take forward. I would say the majority or the vast majority of interactions we have had on dose optimization with the FDA, as you enter into a registrational study, this is a regular topic now at this point. Okay. There was discussion around, not a blinded, but a randomized dose optimization that we do in this process as we initiate the phase III. We are going through that right now. So taking the existing data we have, but also doing these two cohorts randomized. You are going to tell the markets the answer by the end of the year? Are you just going to announce the design and say that is the dose? Are we going to see the comparative data or not really? Yeah, we haven't said exactly when that would be, but we're defining that right now with the data we're developing. But yes, we anticipate we will disclose ultimately what that dose is. Okay. Another important question, just as you pointed out, they're competitors, so it would be helpful if you could sort of delineate what you view is differentiated from your drug, both on the efficacy and safety side, both in terms of response rates, CNS activity, AE profile, et cetera. How do you see it as a unique asset? In terms of within the DLL3 area? Yes. No, within the DLL3. Yeah. Look, I think at the end, our view is directionally. Obviously, we need to prove that with more data. That is one that is going to get presented. I should have mentioned, we will have about 100 patients with the data presented at ESMO. About two-thirds of that will be in small cell, the remainder neuroendocrine carcinoma. There will be a robust data set from response rate PFS and survival data. We will also share full AE information. We will also provide an update by the end of the year in the U.S., Europe population as well. Big picture, our view is that we believe directionally, we believe we are seeing higher response rate. As you may know, in the World Conference on Lung Cancer, we also reported roughly a 6.5-month PFS. IMDELLTRA was roughly 4.5 months or so in their approval readout. We will share for the first time 12-month landmark OS data. You may know that IMDELLTRA's OS in that later line setting was 12.5 months. We think anything above 50%, in terms of that landmark OS, would be a win there. I think from a safety perspective, we will share more information here, but I think big picture, we feel very good about what we are seeing. Okay. That sounds good. All right, let us talk a little bit about some of the other assets then, because you have a lot in the mTOR pathway. Tell us a little bit more about the scientific rationale for the combo work with both the Roche drugs, the pan-RAS and their KRAS, their two, the pan-RAS and the G12D. I will let Daniel Simon answer. Okay. Yep. Great. Thanks, Yujiro. Great questions. We have these two collaborations with Roche that we executed early this year. In the first, they are providing their pan-RAS inhibitor, which is a molecular glue, to us to combine with our PRMT5 inhibitor. In the second, we had additional conversations. We are providing our PRMT5 to Genentech to combine with their G12D. As everyone is probably well aware, I think 92% of pancreatic, of PDAC patients have KRAS mutations. Co-occurring in roughly 40% of those patients are MTAP deletions. There is strong scientific rationale to hit both of those drivers simultaneously. Clearly, the data that has already been put out there from the combination of divarasib and Tango's PRMT5 is hugely compelling. These were a 94% response rate with divarasib. In the G12D population with zoldonrasib, I think it was a 62% response rate. We are excited to have those two combinations going within the RAS mutant population within PDAC. G12D mutations are roughly 40% of those patients. 40% times by 40% is roughly 15%, 16%. That is still a significantly sizable population within PDAC. Okay. There are some other combinations too beyond that, though, right? There is potential for even, if I am not mistaken, a triple combination with the CDKN2A asset. Is that right? Or how are you Yeah. There are several other combinations. One is in the Pan RAS agreement with Roche. It is contemplated that we also have the ability to enable MAT2A, PRMT5, and Pan RAS. That is one. Yigal, as you mentioned, which will be part of the R&D Day in London that we'll showcase in the fall, which is CDKN2A deficiency, which we know is prevalent in roughly 70% of all pancreatic cancer. There you could enable a doublet with pan-RAS. You could also do a doublet between pan-RAS and PRMT5, and then you could also consider a triplet with all of those three for pancreatic cancer. Yeah. Right now, we're probably most excited about the CDKN2A pan-RAS combination. Some of that data that we'll have, we'll plan to share that at that R&D day. Perhaps there you may not even need to do patient selection. Because, as you point out, there's a lot of combinatorics here as far as how you could do this. Maybe just to reiterate for everyone, the one that you sort of see, is the one you just mentioned, you see as the front runner? Or how do you prioritize or what's the hierarchy in terms of your views on putting all these different assets together? Yeah. With PRMT5 and CDKN2A, because they are both on the 9p21 chromosome, and they are about 2,000 base pairs apart, when MTAP is deleted, CDKN2A is almost always co-deleted. Because of that, PRMT5 and this CDKN2A asset, we can pursue as a combo in any MTAP solid tumor indication. That will be a plug and play all across- Okay All of those indications. And we obviously wholly own that, right? In pancreatic cancer, because the CDKN2A deficiency is even higher now than MTAP, and we do not believe others have this molecule of this type, except for us, if we get this into the clinic, that uniquely positions us to do a doublet with pan-RAS in pancreatic cancer to capture even larger population. And yes, based on our preclinical data, we do believe that data actually looks the most exciting. We are building a larger and larger data set, and hopefully by now until when we hope this molecule will be in the clinic, which is first half of next year. But we are already getting interest to do this doublet, with RAS, in pancreatic cancer. Now, we maybe have time to go into some of the other assets like the KAT6 program. Can we talk about that one briefly? Yeah. KAT6/7 and the lysine acetyltransferase area were also relevant to the biology of chromatin remodeling. Yigal, as you know, there has been a lot of activity with KAT6 in breast cancer, in particular in combination with fulvestrant, in the ESR1 mutant setting. Pfizer has presented multiple times and has now initiated a phase III study, so clearly in our eyes, a validated target. Where we think this could hopefully be brought to a step change, advancement in this area is by hitting what we will describe for now as a paralog with KAT6 and 7. As you know, we presented that data in peer-reviewed settings like AACR multiple times, where we clearly see more activity preclinically with a 6/7 dual inhibitor versus 6, both in breast, CRC models, and lung models. That is the plan. We are in dose escalation now. The priorities are breast, CRC, and lung cancer. Pfizer also has a 6/7. They are side by side with us. They are also evaluating CRC. It looks like they also have an arm of just monotherapy in CRC, which is exciting. We have cleared multiple dose cohorts. We believe we are now entering into the clinical efficacious range. Hopefully by the end of the year, we will have a lot more information. The last part I will mention, two combinations we are focused on. One is with SERD in breast cancer. The second is with pan-RAS and KRAS CRC, and we did publish that data as well, at this last AACR. Okay. The significance, because that is an important point, that you hit the seven isoform too, which gives you what exactly? What is the difference between that and ones that are just hitting the KAT6? Yeah. We have preclinical data which we have published on publicly, both in vitro data as well as in vivo data. We believe there are specific bypass type mechanisms with just KAT6 alone, that you cannot fully suppress that pathway. By doing that, our hope is, can we then deliver monotherapy activity? Because as you may know with KAT6, it has largely been dependent on this fulvestrant combination. Right? If you show real monotherapy activity right out of the gates, it's clearly delineated. We also know with KAT6 alone, there has never been a pursuit of CRC, but now that's being pursued by Pfizer with the 6/7. Again, I think clearly shows there's fundamentally important biology by hitting both 6 and 7. This is similar to how you think about CDK2 and CDK4 or MEK1 and 2. No different. Okay. Versus Pfizer, you're basically doing something relatively similar in terms of the molecule. Have you done the comping? Has yours got more potency on one or both of the isoforms, or is it more going to differentiate on the clinical development strategy? Because you pointed out that they're already in late-stage trial. Yeah, we haven't seen much information published from them, so it's hard for us to say at this point. All we know is that they started dosing the clinic this year, and as did we. I think we're basically side by side right now, and hopefully there'll be more information and data that's presented out there. But clearly an exciting area, an area that there's a lot less competition right now. There's only two companies with a six-seven inhibitor in the clinic. Right. Okay, and then briefly, we want to ask all the companies about AI really fast. If you could, anyone want to comment on how you're using AI tools, either with your rolling submission, with clinical work, with data analysis, with internal efficiencies, other specific modules or platforms you're using more than the others, like Gemini versus Claude versus Copilot? Yeah. I know it's a little bit- I think Josh and I could tag team on this one. But yeah, Yigal, the answer is it is a priority for us. We're integrating across drug discovery, clinical site selection from a regulatory perspective. I know we're even having discussions around further downstream, such as reimbursement processes as well. I don't know, Daniel, do you want to? Sure, I think in answer to your question, I could just say yes and put the mic down. We're looking at this across the entire pipeline. We've been most public historically about the work that we're doing within discovery using AI. For example, if you know the protein structure and you've got a specific binding pocket you're trying to generate a hit for based on the distribution of large and small amino acid side chains, positive and negative, hydrophobic, hydrophilic, there is a series of physical constraints within that pocket that we use AI to then try and generate a series of hits that may optimally fit within that space. We use additional software to then pass those molecules and say, "Actually, of these 100 possible hits, focus on making these 20." Clearly, that saves huge amounts of time and cost. We also analyze molecules from both sides, and yes, the ones that the software recommends to actually make do seem to generally do better, although there's still some less good fits within those. There've been a number of discussions this year about using AI to improve the process around the incredibly heavy lift, to get all these documents together, generate all the summaries within the regulatory documents. That is ongoing work, and I hope there'll be many more filings and regulatory documents to come. Using AI to help identify where we'll find patients when it comes to commercialization, to say nothing of all of the process efficiencies in gathering competitive intelligence, running analyses, helping with finance. It's been a big strategic priority this year, and there remains a lot to do. To your other question, we use ChatGPT, some folks use Claude. We've just started rolling out Copilot, so we're exploring multiple platforms to support the company. Any closing remarks, or just maybe recap the key catalysts for the next through the balance of the year? Yeah. Sure. Josh? Yeah. Lots coming. Exciting 12 months ahead. I think to recap, ESMO updates on the DLL3 asset, the neoadjuvant study, NDA submission on track for the second half. Into next year, preparing for commercial launch, hopefully in the first half. And additional progress that you should see throughout 2027 with regards to the MTAP portfolio, both the Roche collaborations, the IDE892 program, hopefully getting additional registrational studies started there. So lots more to look forward to. When is the London day you mentioned? It's in November, Josh. Yeah, no, it's November. It's during the Jefferies conference in London. Okay. So maybe I need to get a plane ticket. All right. All right. Thank you very much. Great discussion. Thank you.