Presenter at the B of A Annual Healthcare Conference in Las Vegas. My name is Jason Gerberry. I'm one of the Biopharma Analysts. Pleased to be introducing Relay Therapeutics. We're joined by Pete Rahmer, Chief Corporate Development Officer, and Don Bergstrom, President, of R&D. Gentlemen, first, thanks, for joining us. Yeah. Thanks to Jason, and thanks to the entire BofA team for having us. Yeah, I think figure, two relevant discussion points would be your PI3K and, you know, your FGFR2 programs, which are your most advanced programs. You know, for those of you less familiar with Relay works on validated targets in the precision oncology realm, looking to make better drugs than those that are, I guess, available to patients, oftentimes improving the therapeutic window for those medications. I thought it would be good to just start with the PI3K program, just because of your more recent data update that you had at the AACR medical meeting. You know, maybe just if you can just talk a little bit about the data that you presented in breast cancer specifically with the combination arm, you know, the good, the bad, and I guess ultimately the TBD? Yeah. Which is always the case with, you know, the first initial, phase I updates. Yeah, definitely. Maybe I'll start and then pass it over to Don. I think the one thing, just to be clear on, our goal of this program is to create a unequivocally differentiated medicine for patients with PI3Kα mutations, as measured by the ultimate regulatory endpoint here, which is median PFS. We believe that the initial data that we showed for RLY-2608 at AACR are the first step in being able to demonstrate the fact that we have a molecule that could potentially end up doing that. I think, to remind folks, you know, as we had been talking about for almost a year going into this initial disclosure, similar to our initial disclosure we made on RLY-4008, our FGFR2 inhibitor, the goal here is to demonstrate clinical proof of mechanism. We had been defining that as being able to demonstrate that we could reach our target exposures. In this case, the target exposure that we defined was consistent coverage above the IC80 of the mutant form of the protein. Being able to do that at tolerable doses, showing being able to demonstrate in a robust number of patients a safety profile that is in these initial data set, differentiated from the non-selective inhibitors. Then being able to do that in the context of supportive pharmacodynamic data, showing also showing that we are hitting the PI3Kα mutant form of the protein. I think the initial data that we showed checked all those boxes for us. We showed that we had multiple doses in both monotherapy and combination that achieved those target exposures above the IC80. We were able to do that at a very favorable safety profile in this initial data set. You know, the key toxicities that plague the non-selective inhibitors are hyperglycemia, rash, diarrhea, all with Grade 3 rates that are certainly limiting the dose intensity of those agents. We saw no Grade 3 events for at those doses above the target exposure. That was extremely encouraging to us. Lastly, we were able to demonstrate that we were hitting the target through a number of measurements. Our ex vivo PD assay that we created, showing that we can inhibit phospho-AKT at or above the 80% threshold. We showed ctDNA decreasing across a number of different doses. Lastly, we showed antitumor activity. I think where there was probably a bit of a focus that resulted in kind of the stock price reaction was probably on the expectation of seeing a definitive response rate in the kind of breast cancer combination patients. Factually, we did not see a partial response in those combination patients. Mm-hmm. I think we, you know, we gain confidence in the totality of the data. You know, the key metric that we're focused on here again is can we get confidence in being able to drive to differentiated median progression-free survival to the existing agents. I think that all the data that we've seen to date in the AACR presentation start to give us confidence that we have the profile of the medicine that can accomplish that. Yep. Yeah, you mentioned sort of the market's reaction to it. You know, clearly from our perspective, you established safety and good target coverage out of the box. With the combination arm, were you at all surprised by the lack of responses, or would you just chalk that up to small sample size and you need a little bit longer duration follow-up ultimately? I mean, the latter, but, you know, I don't want to sound dismissive of the fact that we didn't see a partial response. Like, you have to remember that if you were to go back and look at even the early combination phase I-B studies with inavolisib and alpelisib, you know, you're talking about these early response rates are somewhere in the high single digits to low teens. Mm-hmm. When you look at the number of patients we had above the target exposures in combination, there are seven of them in this initial data disclosure with measurable disease. You know, that's a difference of one of those patients, you know, tripping over that 30% line. Yeah. When you break that down to the actual amount of tumor shrinkage you're talking about, you're talking about the difference in millimeters of tumor shrinkage. Yes, I think we would have all felt more comfortable with this initial data disclosure if there was a couple of those patients that tripped over that 30% line. I think we really have to keep in mind the disease context here. You know, this is a disease setting where, one, response rate doesn't correlate that well with PFS. It's not an ultimate regulatory endpoint. When you really look at some of the metrics that actually matter to being able to push towards or getting to a meaningfully differentiated PFS, you know, one of the key things you keep hearing us talk about is dose intensity. The reason why we do that is one of the most more robust assessments of why that matters in this patient population is when you look at the SOLAR-1 study, which was the approval study for alpelisib. Mm. They did a subset analysis by dose intensity in that patient population. The label dose for alpelisib is 300 mg once daily. The median dose intensity able to be maintained in the SOLAR-1 study is 248 mg once daily. When they bifurcated that patient population and looked at the Kaplan-Meier PFS curves by those that were able to maintain dose intensity above 248 and those below, you saw the median PFS in the above patient population at 12.5 months. In the patient population below that 248, it was only 9.5 months, 9.6 months. Again, we keep coming back to this concept of dose intensity. What makes people uncomfortable about that is that it takes time to define. Sure. We think we have a lot of the initial signs to point towards where we should be able to get there. The more acute measurement that you can get in oncology studies i s response rate. you know, the fact remains that we didn't see a partial response in the combination patient. As you mentioned, ORR and PFS haven't always correlated in this area with other medications that we've seen. Correct. Yeah. Again, you gotta keep in mind, this is the patient population where only 5% of the patients are diagnosed in the metastatic setting. Mm. The bulk of these patients are found very early on in their disease course. They're not in the HR+/ HER2- patient population, these aren't patients that you're typically finding with extremely bulky disease where the benefit of seeing response rate early on, it gives you the comfort that you're debulking very symptomatic disease. That's not the case here. They're being monitored largely through CT scans and not because they are generally feeling very symptomatic progression of their disease. Therefore, that's why this concept, when you speak to the investigator community, which we've spent a lot of time with post these data, there's a very different perspective than what Wall Street is seeing, how Wall Street is assessing these data. They're very encouraged by these early data. It is giving them confidence for the first time that they have the option of a truly mutant-selective PI3Kα inhibitor at their disposal. Yeah. I guess we jumped right into a very micro-level topic. Sure. This is metastatic breast cancer, really where the opportunity is and why that's relevant, I think, for you guys is, you know, FGFR2 was very good data, but I guess the pushback is small, maybe smaller tumor opportunity, whereas the breast cancer opportunity for PI3Kα is viewed to be kind of one of the bigger, more attractive opportunities, which is why it gets a lot of investor focus. Look, totally makes sense we've got to kinda wait and see how the data mature ultimately and how the PFS profile looks. One thing that comes up a little bit is just the reliability of the ex vivo target coverage data that you've generated. This came up with the discussant in the, in the Q&A at the AACR medical meeting. With FGFR2 inhibitor, you were able to corroborate really good target coverage with clinical efficacy. Do you use the same assays with 2608? Just kind of if you can speak to the reliability of these assays in predicting target coverage. Yeah. Yeah. I think one of the challenges of developing a mutant-selective inhibitor is the only tissue that the true target of the drug is expressed in is tumor. Frequently, you will see people use a surrogate tissue. They'll look at, you know, peripheral blood mononuclear cells or hair follicles or some other tissue that gives you confidence that in tissue your drug is doing what you would hope it to do. Ultimately, people will look at tumor biopsies. Tumor biopsies, you tend to not get very many paired tumor biopsies. The assays that are used to measure biomarkers in tumor biopsies frequently are not quantitative, so it's hard to build a quantitative PK/PD model. As we approach this, what we did is we said we're obviously gonna look at PK and relate that back to all of our preclinical PK data, and PK/PD efficacy in preclinical models. We will look at an ex vivo assay that in and of itself as a single data point is not necessarily definitive for saying what's happening in the tumor. In the context of all of the other data we're collecting, again, it's another corroborative data point where we can look for consistency, of effect across the various measures we're looking at to get confidence in the dose. We look at ctDNA, which is now a more direct measure of the pharmacodynamic effect of RLY-2608 in the tumor because what you can see is that you actually are having activity against the specific mutant clone that is expressing your PI3K mutation, and you can look for clearance of that ctDNA data. Here is now where I think you see all the data come together, where you say we would project that at 400 mg monotherapy or 600-800 mg combo, we should be at our efficacious threshold. That is corroborated b y the ex vivo phospho-AKT assay that's showing us about 80% inhibition of PI3K in that assay at those doses. When you go to the patients who are treated at those doses who have detectable circulating tumor DNA, we're seeing significant reductions or complete clearance of circulating tumor DNA actually at a rate higher than what you're seeing in patients at the lower threshold. Now I think, you know, we're getting more experience with these doses for efficacy readouts. Here it's looking not only at tumor shrinkage and the potential for response, we've talked about response quite a bit, but you're also going to be looking at how long patients stay on therapy, how long patients stay in disease control with stable disease or better as their response. That will translate into clinical benefit rates of stable disease or better at 24 weeks. All of these data, I think, will be taken together in aggregate as we select doses to take forward. Just to follow up on dose, I think you framed 600 mg and 800 mg as therapeutically active doses. Curious, you know, that's 80% target inhibition with these doses and in the preclinical setting, have you looked at sort of antitumor activity when you push beyond IC80 to maybe IC90? Maybe it's a question, are you leaving, you know, efficacy on the table by not pushing to greater target inhibition? In preclinical models, no. We could push to IC90, but we had already maxed out efficacy at IC80. Got it. Okay. Maybe how comfortable are you guys with the 600 mg hitting, IC80 across patients, given there was some data variability just judging by the error bars, and the drug exposure in the PK data? Yeah. I mean, there's, you know, clearly, I think, fairly standard interpatient variability for an oral molecule that we're seeing with 2608. When we're showing those PK curves there in average with error bars, you know, there could be patients who are just at the threshold or slightly below it. You know, I think the threshold is a useful tool, but it's not necessarily a, you know, a line etched in concrete. You know, if you look at our patient where we did see a monotherapy confirmed partial response, that individual patient's PK exposure was actually slightly below the 80% exposure threshold line. It is, it is a continuum here. Right. I think with the 600 mg dose, we're confident that the bulk of patients, you know, will be at or above the line. Got it. Maybe thinking a little bit about sort of next steps here as you advance 2608 to the dose expansion here in the second half. Maybe just the timeline between now and AACR, and going forward and what you aim to achieve in the dose escalation portion? Any metrics? I mean, I would assume that the focus will be probably on still response rates, but also PFS with the next data update. Yeah. Maybe two parts. Don can start with some of the characteristics we'll take into consideration as we choose the dose or doses to take into expansion, and then I'll come back and talk about the 2024 update. Okay. Right now we are continuing to enroll the 600 mg and 800 mg doses in combination in dose escalation. The protocol is flexible and allows us to enroll up to 12 or potentially even more patients and still be in dose escalation so that we can get more robust data there. We are also opening a 1,000 mg BID cohort. You know, we think that is probably excessive compared to where we want to be in terms of exposure, but we're doing it to be able to define the full dose range and to be able to set up productive end of phase I interactions with FDA. The parameters we'll be looking at as we choose a dose or doses to take forward into expansion cohorts is we'll be looking at our target coverage, we'll be looking at the PD endpoints, we'll be looking at the safety profile, and I think importantly, we'll be looking at the ability to maintain dose intensity. We'll be looking at the efficacy readouts that come in conjunction with it, and we'll look at the totality of the data to be able to support a dose to take forward into expansion cohorts, which is very similar to what we did at about this time last year in our RLY- 4008 FGFR2 inhibitor program, where we had a very productive end of phase I interaction with FDA based on that robust phase I data set. Yeah. The nature of the guidance in 2024 is so we can then take some of those data that Don just mentioned and let that mature and be able to have what we would anticipate to be an interpretable, robust disclosure that starts to give us a sense of this concept of dose intensity at a larger end at those doses above the target exposures and start to get an idea of this, of what CBR could start to look like, albeit early and likely somewhat immature, to start to really inform that ultimate endpoint of median PFS. Yep. Got it. Maybe just talk about past discussions with FDA around Project Optimus and sort of given the agency's push, you know, regarding dose exploration, how this sort of strategy maybe melds into this dynamic. You know, we hear a lot about this from your peers that it's sort of impacting how they're going about sort of clinical exploration. Yeah. I mean, I think there's certainly more dose exploration in the Project Optimus era than there was before a few years ago when this, when this regime came into place. You know, I think we have learned from our FGFR2 RLY-4008 experience with FDA, where we went into that end of phase one meeting, we had 115 patients worth of data across 14 different cohorts. We had robust profiling across multiple different endpoints. you know, I think we had a database that really allowed us to defend the 70 mg once daily dose that ultimately we took forward. FDA ultimately agreed for that to be the dose that we would use in our pivotal design. I think it's important to point out that it's not any one single data point that FDA is looking for. They're looking for data, you know, across multiple different parameters. When we went and had the discussion around 4008 with FDA, response rate wasn't a driver of that dose defense. Mm. Right? Because at that time, we actually had, you know, very few patients who were FGFR inhibitor naive who had been exposed to 4008. We didn't even talk about response rate, right? It was all focused on PK/PD safety endpoints. I think consistent with that, with 2608, we're trying to build a similar type of robust data set where we're profiling multiple biologically and potentially clinically active doses across a number of different parameters to be able to hopefully have another productive interaction with FDA. We've not had it yet. Mm-hmm. I can't comment on what will happen, but I think based on the successful interaction we had last year in FGFR2, I think we're well-positioned as we go forward with 2608. I know it's a little early, but just maybe talk about, do you feel like you've, in some respects, optimized the profile, right? You've said that getting greater target inhibition, you're not seeing much bang for your buck there. The reductions in off target AEs looks pretty good. There are some other pipeline programs, competitors, right, that are touted as more selective. You know, as you kind of just look at yourself versus the field and what we understand about the different profiles of the different molecules, Eli Lilly, Scorpion, maybe if you can just sort of offer a little bit of your perspective. Yeah. Make the obvious statement that, you know, we only have clinical data on our molecules. I think as we gain more experience in the field and if these other peers end up showing data, it'll help inform this conversation more. You know, the factual differences between the molecules are, you know, we are running our clinical study agnostic to the PI3Kα mutation. As you can see, we had, you know, all different potential mutations represented in our initial data set. Both Loxo and Scorpion are limiting their initial clinical exploration to H1047R or X. There's just a much more limited patient population either of those molecules appear to be serving today. You know, the question to selectivity, it was a big question going into our initial data disclosure. You know, all we can say is that we are able to achieve the target exposures we were looking for. We are able to do that with no apparent impact or on the closest off target here, which is PI3Kα wild-type, nominal elevations in glucose across all the different doses we've been exploring. It's driving to meaningful pharmacodynamic markers that's showing we're doing that selectively. Mm. As we deduced pre-clinically, there's no benefit in going higher in selectivity pre-clinically, and that seems to be playing out clinically thus far in our experience. Yep. Okay. As we think about how the data can ultimately, when it matures, de-risk the ultimate regulatory endpoint here, you guys as a company have talked about being in the business of creating step change in efficacy. I assume that we, you know, we'll wanna see, from a PFS perspective, something that can beat alpelisib on a PFS superiority endpoint. Is that sort of the right way and the fair way to be thinking about this? Or, you know, if you were comparable to, but have a much different tox profile, is there ultimately an alternative pathway? Just trying to think through the different scenarios in the profile that you're seeing. The latter that you mentioned is not the base case. Our base case goal here is to create a meaningfully differentiated molecule for patients in the context of efficacy. As it stands today, the most relevant study from alpelisib with fulvestrant is the BYLieve study, in which the results there was a median PFS of 7.3 months. The CBR in that study was 46%. Those are some of the relevant efficacy benchmarks that we are focused on as we look and continue to follow the evolution of our own data set. Got it. Okay. Maybe just quickly, 5836. You as you advance that into the clinic, what you're hoping to see and how we should think about that relative to 2608. The original goal of RLY-5836 was to be an insurance policy, if you will, against potential idiosyncratic tox with RLY-2608. Obviously, with this initial data set, that's largely been dispelled. This is the way we view it, RLY-2608's game to lose at this point. It checks all the boxes that we wanted in an early clinical data disclosure. Now RLY-5836, we have it. It is at the stage of being able to go into the clinic. Given the size of the patient population and the opportunity for us, we will explore both of them clinically for a period of time. It's likely a decision we would make in 2024. Again, you know, there... There's no liability that we see in RLY-2608 that we're hoping RLY-5836 solves for. Yep. It's a different chemical scaffold, but otherwise very similar profile. Okay. Anything you think we've missed on PI3Kα before we jump to 4008? No, I think we covered all the points. Actually, can I make one point? Go. Sure. Yeah. You know, I think you were, you were asking the question about, you know, how we think about ultimately developing our PI3K franchise. I think one thing that I, you know, think these data did teach us that's important is ultimately what our goal is in breast cancer is to be able to move into higher order combinations earlier in the treatment of the disease to really maximize benefit for patients early on. That could be in the early metastatic setting or even in the neoadjuvant, adjuvant setting. Those are places where in our target population, the CDK4/6 inhibitors, you know, have really established themselves. What we've seen in the PI3K class is challenges combining with the CDK4/6 inhibitor class. Alpelisib was not able to be successfully combined with ribociclib. You see inavolisib from Roche's Genentech only combined with palbo, not combined with ribo or with abemaciclib potentially because of the risk of overlapping GI tox. I think with the profile we've seen for 2608, it looks to be very combinable, and we think it, you know, could be combinable with any of the CDK4/6 inhibitors that really will give us maximal flexibility as we're continuing to develop this asset. Maybe actually you've sparked a thought. Outside of breast cancer, right? PI3K is expressed and, you know, but it's also a drug that has some synergism with fulvestrant. Yeah. As you think about other tumor settings and what you've learned single agent activity-wise, just maybe curious sort of combinations that you're thinking about 'cause it would seem logical that you'll probably need to combine this in other settings as well. For single agent going in, the four populations we have prioritized for exploration are head and neck, clear cell ovarian, cervical cancer, and patients across tumor histologies with double mutations. Amongst our 19 monotherapy patients, we only enrolled four that were in those populations, and only one who actually was at our target threshold in those populations. That was our responder who had a double mutant. I should mention one of the other patients in that population was at a lower dose, but it's a head and neck cancer patient, non-measurable disease, but clinically measurable disease with significant clinical improvement, and the patient stays on trial now over 10 months. I think we are seeing early data suggestive that those populations could be populations to target for monotherapy. Mm-hmm. For combination therapy, again, I think we have a profile that suggests you could combine with MAP Kinase pathway inhibitors, especially when you start looking in colorectal cancer and other diseases where you have co-mutation of MAP kinase pathway and PI3K pathway. Okay. Maybe transitioning to RLY-4008, your lead drug for cholangiocarcinoma. Just, you know, in the ASCO, you'll have a phase I dose escalation update. Just anything incremental you'd wanna flag into the update versus the prior update at the triple meeting, which I think was the last update. Yeah. No. This disclosure at ASCO is only from the dose escalation part of the study. It's an exercise to close out that study, give those investigators their due recognition for running probably one of the most robust dose escalation studies in modern targeted oncology, clinical development. Given that we presented twice from that data set already, and actually ESMO was more expansive of a presentation 'cause it included some of the expansion cohort of patients. W e don't anticipate that the investor community is gonna learn anything meaningfully new from this data disclosure. It is certainly one that's more of a closing out the study, continuing to build awareness in the investigator community, but there's not gonna be any meaningful new learnings within the cholangiocarcinoma FGFR and IE patient population or, you know, we've gotten the question of could we learn more about the non-CCA patient population? The bulk of the 115 patients in the dose escalation sit within cholangiocarcinoma, and actually a large portion of them are FGFRi experienced patients. Obviously not one of the key target patient populations of go forward. The more important data disclosure that we'll be making for 4008 this year will be the non-CCA expansion cohorts, which we're on track to disclose in the second half of this year. Those are three non-CCA, tumor-agnostic cohorts in fusions, amplifications, and mutations. And we do think that the data that we disclose in the second half of the year will be informative against potential additional indications to explore developmentally for 4008. Yep. Okay. I think you guided to full enrollment of your pivotal cohort second half completion. Just thinking about timeline for top line results, that'd be 2024 presumably? That's it's a reasonable estimate at this time. You know, let us get that cohort fully enrolled, and we'll come back with probably more precision around when we can see that. Would you expect to present any interim data on that or just go? We would not. Okay. It is most likely the next time you'll see data from the cholangiocarcinoma patient population at the registrational dose is when we show what is, you know, nearly gonna be the NDA data set. Got it. Okay. We have about 10 seconds left, but anything else you wanna flag on RLY-4008? No. I think the main thing to, the last point to hit on is we have $938 million in cash, takes us well into 2025, and we are confident that we will drive to meaningful value creation in the window of that cash runway. Great. Well, gentlemen, thank you so much for joining us at the conference. Thanks for having us. All right, everybody.