45th Annual TD Cowen Healthcare Conference. I'm your Yaron Werber , along with my colleague Jane Hun from the TD Cowen Biotech team. It's a great pleasure to moderate the next fireside chat with Relay Therapeutics. The team here really needs no introduction. To my right is Sanjiv Patel, President and CEO. To his right is Peter Rahmer, Chief Corporate Development Officer. To his right is Don Bergstrom, President of R&D. Gentlemen, thanks for joining. We appreciate it.
Thank you. Thanks for the invite.
A lot to talk about. You just announced last week your phase 3 ReDiscover-2 trial design. Maybe before we get to that, can we talk about just the latest ReDiscover one, latest data from San Antonio, where you split up the data now in second-line metastatic breast cancer and also all comers, essentially, in that study, second-line, third-line onwards. You also broke it up by kinase mutations versus all mutations. There was obviously the reason you did that so we can compare apples to apples to the new Lilly/Scorpion data as well. Do you want to review the data, and then we'll have some follow-ups as well?
Yeah. I mean, as you know, the data has been a long time coming. We've been working on this program now for close to seven years now, trying to create a mutant-selective PI3Kα inhibitor. And with RLY-2608, we think that's exactly what we've done. The goal back when we started the research program was to dial out the wild-type toxicities of hyperglycemia, diarrhea, rash that hopefully would have led to greater dose intensity, and that would in turn lead to greater efficacy. It was a great pleasure in San Antonio to show that data. We have a big safety database now. The data that we showed was 64 patients in 600 mg b.i.d. with nine and a half months' worth of follow-up.
We showed what we believe to be a very differentiated safety profile, very low rates of diarrhea, rash, and most importantly, minuscule rates of grade 3 hyperglycemia. We hope that would translate into efficacy. As you know, the second-line, second-line plus population is really where the unmet need is. Early lines have great therapies, but five and a half months of PFS is where capivasertib is. Capivasertib is doing great in the market, as you know. It has had a great launch. It is on its way to being a billion-dollar drug very rapidly. Showing 9.2 months' worth of PFS across all patients, over 10 months' worth of PFS in second-line-only patients, is close to a doubling of the PFS. We were thrilled to be able to show that data, both translate for the safety being differentiated and then the efficacy.
In the kinase-only population, where we know Loxo, Lilly was showing data, and we saw some early data from Oncure, we showed a 67% response rate and again, over a 10-month PFS. That compares to, I think, the discontinued LOXO- 783, where it was kind of a single-digit response rate and obviously did not get to showing a PFS. I think we have a truly differentiated molecule, which makes us excited to go out now and start the ReDiscover-2 pivotal trial to go head-to-head with capivasertib and get this therapy to patients as rapidly as we can.
Yeah. The other thing, just as we launch into probably some follow-on questions about the phase III, the data, the patient population in which we showed our data, and we're about to run this phase III, just a reminder for folks, that's 13,000-plus patients in the U.S. alone that are CDK4/6 experienced that would have a PIK3CA mutation that we can address. It's clear that it's an unmet medical need, as Sanjiv alluded to, the launch of the capivasertib just four quarters into its launch on a trajectory of reaching $1 billion. Having an agent where we can now go into a phase III and potentially come close to doubling the PFS, we're very excited about the potential for that market for us.
Is the PFS that you showed at San Antonio the final PFS, or is it still maturing?
Yeah. It's still maturing. What we showed at San Antonio in our overall patient population, we had a median of 9.2 months PFS with nine and a half months of follow-up. We still had some patients who were before the median who were censored. Most of those patients who were censored were patients who were ongoing in response at the time of the data cut. Similarly, in the second-line patient population, our follow-up was a little bit shorter, is about seven and a half months. We're sitting at 11.2 months PFS, similarly with some early censored patients who are ongoing in response. We're continuing to follow with the data and anticipate that the data will be stable with regard to not eroding any from where those numbers are, given that the early censored patients will still contribute information.
Our patients are doing well with the drug, but we anticipate that we will continue following those patients and provide an update this year.
I think as you look at the data, as we look at the data over time, even the response rates are deepening. I think this has been one of the unexpected findings with this drug, right? The cardinal mantra usually in oncology is your response is early or they're improved slowly. In this case, the response is declared late. I think investors, because of that, were potentially taken aback or a little skeptical or confused, but we've seen that over and over again now and consistently with the data. Do you have a sense what's driving that?
Yeah. I think what we've noticed here is at our lower doses, where we're not getting as much target coverage, we can achieve responses, but they happen later, right? We made our initial data disclosure. We didn't have that much follow-up. We had a median of about four months of follow-up. We had a lot of patients at lower doses. We weren't yet seeing responses. Now that we've gotten to our optimal dose, so 600 mg fasted, which is the data we've reported at San Antonio, we're moving forward with 400 mg administered with food, which gives us the same blood concentration of 2608. You get better absorption with food. What we see at that dose is that actually patients who respond do tend to respond quite quickly, right? Median time to response is about one cycle. It's typically first scan where we see a response.
I think this is now a property just of when we've disclosed data. The initial data disclosure, again, lower doses, short follow-up. We hadn't seen the responses yet. Now that we've got patients with longer follow-up at our optimal dose, we do see responses happen quickly.
When are we going to see the next data cut? Is this sort of an ASCO event, an ESMO, San Antonio?
I think you're hitting on the key potential venues. I think we'll fall short of giving a specific venue as of yet. The SCBCS data cut was from November. We do want to let reasonable maturity delta in between those two time points. I think getting out past a median follow-up of 12 months would be pretty informative. The other key objective of an updated data disclosure is really to continue to drive awareness for the global phase III study that we're running. Therefore, it makes sense to choose one of these major international medical meetings. I think you hit on kind of the three major ones that are coming up in kind of between the middle of the year and the end of the year.
Right. If you're at seven and a half and you want 12 months data, we're probably looking at kind of later on ESMO, San Antonio.
Yeah. I think ASCO and San Antonio are the bookends. Yeah.
OK. Would we start having survival data at that point as well, or is that still going to be too early?
Yeah. I mean, I think we can potentially characterize survival. We're following it in the trial. We do expect it to lag, given that we had a large proportion of our patients still being actively treated with 2608 at the time of the November data cut.
Are you still enrolling patients into the study or no?
No.
Sixty-four patients was on the 600, so that's the upper end, right?
Yeah. Fifty-two of those were for efficacy variable, no AKT, no PTEN mutation.
Let's maybe talk a little bit. I want to talk about the phase two trial design. Maybe before we go to that, let's talk about the 400 milligrams fed versus the fasted. That's based on your modeling. Maybe give us a little bit of experience. How much data do you have with 400 to get yourself and to get FDA comfortable? Is this based on simulations? How did you get there?
No. Let's start from the beginning. Right back at the beginning, we dosed fasted right from the beginning of this when we put the 2608 into the clinic. All the data that you saw up until recently was fasted data. We were lucky enough to start this in the era of understanding the rules around Project Optimus. We did a pretty extensive dose exploration. Obviously, we chose expansion cohorts to move forward with. The clear 600 mg b.i.d. was the one that we initially started with. That has ended up being the one that we've shown the efficacy data with. As we got into 2024, we started to think about dosing with terbicyclib, which is dosed with food.
We started to think about the vascular malformations trial that we'd run with our RLY-2608, which is going to be done in adolescents and then even younger children over time. Dosing with food became a question for us. We did the food effect study in 2024. That, obviously, as we started to do that, showed that there was a positive food effect. Eating increased the exposure slightly of 2608. In the back half of last year, encompassing all the work we did for our own Project Optimus, we started to dose a reasonable number of patients in the teens at 400 milligrams b.i.d. fed.
Those patients are dosed both in combination and in monotherapy across a range of different tumor types to try and get us a pretty robust data set to show PK equivalence between the 400 milligrams fed and the 600 milligrams fasted. We took the whole data package, extensive dose exploration, the expansion cohorts that we ran all fasted, this PK study that we did in patients, oncology patients with the fed dosing. The FDA was in full agreement with us to move forwards with the 400 milligrams fed in our pivotal trials. Obviously, we'll use fed dosing now as we go forwards, both in the triplet trials that we're running as well as our planned oral SERT trials. In terms of just the equivalence, maybe Don, you can just comment on the actual kind of clinical experience.
Yeah. I mean, I think what we see when we look at any PK parameter, the 600 milligram fasted dose and the 400 milligram fed dose look essentially identical across CMAX, AUC, CTROF, any PK parameter that we've looked at.
Maybe on the safety profile?
Yeah. The safety profile is generally consistent. We see the same rates of any of the common AEs that we see, with the exception of some of the upper GI AEs, so nausea, loss of appetite. Both of those benefit from dosing with food. We see both rates of both of those go down at the 400 milligram fed dose compared to 600 milligram fasted.
Would you release this data at an upcoming conference?
The PK data, the majority of the data we have, we've shown a sample of it in the corporate deck and the press release we made last week. In terms of efficacy data, the study was never run for efficacy. These were monotherapy combo patients across multiple tumor types. The goal for us was just to show equivalence, which we've done.
The PK we have in the deck is trough data. We could show these other PK measurements because it was really just optimized to really show equivalency on the PK parameters.
Is there accumulation? It's fairly predictable.
The behavior across any parameter we look at is exactly the same as 600 fasted.
The type of food that you have is also not.
Doesn't matter.
Not specific. Yeah. It's just with food. Right. So they need to eat with a, what, a half hour before they take the drug or concomitantly?
Yeah. I mean, compared to what patients have to do fasted, which is not eat for two hours before taking the drug and not eat for an hour after, now it's pretty much we can remove those restrictions and just instruct patients to eat a meal proximal to when they take 2608.
OK. By the way, if anybody has any questions, feel free to raise your hands. Happy to call on you. Let's talk about the phase III trial design. The primary endpoint is PFS. As you said, it's head-to-head against TrueCap. What's interesting is you have a hierarchical PFS, right? First in patients with kinase mutations and then in all mutations. Number one, would you enrich for one or the other? Do you need to have certain numbers of one or the other? Secondly, why do kinase mutations first and not all mutations and then kinase mutations? Which sample is bigger?
You want to take that one, Don? Maybe you take the second one. Yeah. The breakdown of these mutations in the population in the positive HER2 negative breast cancer population is roughly 50/50. That is what we aim for in the trial. That is what we have had in ReDiscover 1 so far, is we have been about 50/50 kinase, non-kinase. As we enroll the ReDiscover-2 trial, we will be ensuring that we stay at about 50/50 so that you have the representational distribution of the two types of mutations. In terms of the second question, typically in these types of hierarchical designs, you start with the more specific population first, right? Because ultimately, you are testing that statistical hypothesis, and then you are broadening out from that.
I think the decision to go with the hierarchical design was partially driven by the fact that we have seen very, very encouraging kinase domain activity in ReDiscover 1. As Sanjiv mentioned earlier, in ReDiscover 1, we're sitting at a 67% response rate, 11.4 month PFS across all lines of therapy. We want to be able to ensure that if that holds up in the phase III trial, that we are able to make label claims about that population so that we do have a formal statistical analysis. This is the most statistically efficient way to make sure that we're able to perform a formal analysis in the kinase patient population, as well as being able to perform the analysis across all patients with mutations.
Let me ask another one. I mean, capivasertib is on its way to become a big product. That's a PI3K selective, right? Sort of AKT focus kind of pathway. The other opportunity is inavolisib, which is a PI3K alpha, but not selective like yours. That's even moved up. Why go against capivasertib if not inavolisib?
You kind of regulate your standard of care in the space that we're in, which is traditionally second line or in our trial post-CDK4/6. There's only really two approved agents in there. One is capivasertib, as we've seen, is now essentially being used pretty broadly. Sales last quarter, $160 million for the quarter, and that was kind of four quarters in. And then alpelisib from Novartis. Those sales have kind of stalled over the last two to three years. Doesn't seem like it has much of a following now amongst doctors, just given the toxicity profile. So those are the two choices for us. It would be very hard for us to kind of enroll the trial using alpelisib. Just people don't want to use it. So having to randomize against it would be a real headwind for us. I think the opposite is true with capivasertib.
People have comfort with it. The interesting thing about the two things is on the label, obviously, alpelisib's kind of nominal PFS is greater than capivasertib. Capivasertib is obviously being used because it just has a kind of easier to take than alpelisib. 2608's profile, I think, has kind of the benefits of both, has what we believe to be a good tolerability profile and then a better efficacy profile.
You asked about inavolisib. The current label inavolisib has is in CDK4/6 naive patients, endocrine resistant, and pretty narrow experience in the traditional Western patient population. The trial they ran was in very metabolically fit patients, younger, predominantly Asian. You are seeing some of that reflected in its first quarter of sales. The first full quarter, essentially first full quarter of sales for inavolisib was $8 million. In contrast, CAPI's first quarter of sales was about $43 million. We had surmised they would likely run into some challenges commercially, given the way they ran the study. I think we need to get some more experience for a couple more quarters of sales, probably. In the early days, it is playing out like we had expected.
It also speaks to this question around people, I think, have a view that early lines have this bigger commercial opportunity. In these later lines, it gets less and less. This is a real example to show CAPI is really a big product in the later lines. And inavolisib here is not doing too great in its first quarter in an earlier line.
I think that I'm going to pass it to Jane in a second to talk about your triples and some of the other doublets that you're doing and the pipeline. You've said at this point you have cash to get through the phase III study. This is going to be a global study. Does it make sense to partner at some point? Would you do it sooner than later?
Yeah. Look, I think everyone in this room understands the capital markets are about as bleak as we've ever seen them. The position that we're in, having $780 million of cash at the end of 2024, is one that we don't take lightly. You'll see us preserve that cash to make sure we can get well beyond the top line for our ReDiscover-2 trial. We'll also keep the cash to make sure we can get proof of concept data with our vascular malformations program with 2608, as well as reach an IND with our NRAS selected program and get a development candidate for our Fabrics program. Everything else we have, we have to make sure that is prioritized to do those things.
In terms of just when and how to partner, I mean, I think the breast cancer market is complex and has lots of combinations that we will have to deploy. The second line or post-CDK4/6 trial that we're going to run, I think, is manageable for us to run. Both execution lead team has run a pivotal trial before. We can manage to do that, especially with the capital that we have. As we said, given capivasertib sales, this is a significant commercial opportunity. That is kind of the baseline for us. In terms of going forwards and kind of going into earlier, larger potential indications, we'll do the groundwork this year. We'll continue to push forward with triplets with 2608, fulvestrant, and ribociclib, as well as 2608, fulvestrant, and Pfizer's abemaciclib.
We also will start dosing with some of the new next generation oral SERDs. All of that work in terms of finding the dose, understanding combinability, getting a robust pack together will allow us to be in a position, I think, in 2026 to understand what that next pivotal could look like. That is a development program that we will need to do in partnership with another player. As we bring 2608 into the post-CDK4/6 commercial world, it is probably another time for us to consider a strategic partner. I think we are in one of those very unusual positions in biotech at the moment as a pre-revenue company that we have the capital to execute a meaningful trial in a meaningful commercial space. We will look to get the strategic partners we need at the time that we need them.
Great. I want to dig a little bit more deeply into your triplet combinations. You mentioned the first one is a triplet with fulvestrant and ribociclib. That started in fourth quarter of 2023. What was the rationale for adding ribociclib in particular? I seem to recall the data was a little bit delayed from the year end of last year, just to get a more fulsome data set. Can you just expound on your thinking behind that?
Yeah. I mean, in terms of ribociclib, obviously, in the kind of evolving CDK4/6 landscape, we think that that one, given the data that we've seen, kind of moving towards being the market leader. That is why we wanted to combine with it. In terms of atermociclib, I'm sure we'll get back to that. Obviously, that could become the evolving standard of care as we move into the next decade. In terms of just the speed and what we said earlier in 2024, we obviously noticed a DDI with ribociclib that increased the exposure of 2608. Obviously, that took us a little bit of time to work through as we started to try and find the right dose of 2608 to combine with ribociclib.
I think as we talked earlier, as we went through last year, we obviously also saw a food effect for 2608. What has really slowed us down in that is obviously stopping dosing as fasted and starting to dose now with food. All dosing that we do going forward has food. You have the food effect, as well as this DDI, put it together, slowed us down. I think we have every confidence that we'll continue to progress. Obviously, as we got to the end of the year, we started now dosing with atermociclib, fulvestrant, and RLY-2608.
Drug development, sometimes you do need to be deliberate and thoughtful, exactly as we've dealt with 2608 as we went through the dose escalation and then the fed versus fasted to get what we believe to be a great result in our interaction with the FDA. We want to do exactly the same here with ribociclib, atermociclib, and then see all the oral SERT combinations that we'll start with as well.
Great. Does that mean that you've landed on a dose for the ribociclib triplet?
No. We continue to kind of move through our dose escalation and expansion. We will update you when we are ready to do that.
OK. Also, of course, you mentioned the triplet with Pfizer's abemaciclib. Given that this may be the emerging standard of care, are you perhaps more excited about this triplet versus the ribociclib, which is just the current standard of care?
Sorry, children. We love them all.
All right.
I do think you can look at the current profile of atermociclib in the data that's been reported. We all have a goal to put together regimens that are the best for patients. It is clear that atermociclib thus far is showing to have been able to mitigate some of the toxicities that you've seen with the traditional CDK4/6s. If that continues to play out, it is hard not to be excited about that prospect for patients.
Right. Do we have any timing for data for the atermociclib triplet?
No.
In the future, when we do perhaps see data, what kind of efficacy and safety do you want to see to move it forward in the first line setting?
Look, I think the first thing for us to establish is just dose and combinability. Obviously, we want to see a safety profile that is manageable. I think just given the kind of stacking of these agents, you would imagine that the efficacy would play out. I don't think we have any specific numbers against these.
You have post-Monarch, you have the MAINTAIN study. Those are two decent benchmarks to look at in the CDK4/6 retreatment setting with Palbo and Ribo respectively. I think we would, as we get interpretable efficacy data in the CDK4/6 experienced patient population, those are two probable data sets to be looking at to see how things stack up.
All right. You mentioned that you're moving forward with perhaps combining with next generation oral agents, oral endocrine agents. Where are you in terms of your progress here? And kind of what's the strategy?
Yeah. I think we're in the, as you know, there's only one approved of these agents with elacestrant from Menarini Stemline. That's one we could just choose to do whenever we want to because that's commercially available. Any of the other agents would require some sort of arrangement with the other sponsors of those molecules, which we're in the process of doing. It's tough to guide exactly which ones and at what time. We clearly have an alpha protect in our own pipeline. We have a bit of an affinity for that mechanism. Fortunately, I think we'll see the VeroTac 2 data imminently. We're watching the data sets as they unfold. The way we think about it is to generate data with different distinct mechanisms within the verticals, if you will, of these next gen oral endocrine therapies.
Kind of more broadly, for breast cancer patients who have already failed Piqray, how often is this because they have a PI3K pathway mutation? Would these kind of potentially be addressable by RLY-2608?
Yeah. I think the first thing to point out is many of the patients who stop taking Piqray actually stop because of toxicity as opposed to disease progression. We saw what was stated as a 25% rate of discontinuation in the registrational SOLAR-1 trial. In a patient population with 11-month median PFS, median duration of treatment with alpelisib was only 5 and 1/2 months. Patients cannot stay on the drug chronically. As such, I think there has been limited ability to really define the biological mechanisms of resistance to the drug. It is a question of whether there is actually resistance in the tumor as opposed to just insufficient coverage of the target and the inability to maintain dose intensity.
There's been some suggestion, and there is a publication that came out of Mass General last year that was looking at a rapid autopsy series and looking at metastatic lesions in patients who had died of their breast cancer and characterizing PIK3CA acquired resistance mutations for the orthosteric PI3K inhibitors. It did suggest there were some mutations that arose to alpelisib or inavolisib or other active site PI3K inhibitors that could be addressable by 2608. That could be a mechanism where we could work. Any other mechanisms that would be common with 2608 would be, for example, loss of PTEN or selection for AKT mutations. Again, it's basically based, everything I'm saying is largely based on one rapid autopsy paper right now.
Got it. Also for your PI3K program, you're going to expand into vascular malformations, phase one starting very soon, in Q1 of 2025.
Yes, not long now.
I think this phase one is going to be specifically in PROS, that subset of VM. Can you give us a sense of the opportunity here? What % of VM is due to PROS? What's the standard of care here? What's the bar in terms of efficacy that you're looking to clear?
Yeah, a couple of things there. Correct, we're imminently going to start this study. It will be biased towards PIK3CA-related overgrowth spectrum patients, but it is not limited to that. We can enroll other vascular malformation subtypes. We will be looking to include lymphatic malformations, venous malformations into this study. That will be important to get some of those patients in. The reason why we are biasing it towards PROS is because that is the one subtype. It's about 10% of the 170,000 patients in vascular malformations with PIK3CA mutations. Alpelisib is from Novartis, marketed as VIJOICE, has an accelerated approval there. That's part of the reason for biasing the study a little bit into that patient population so that when we generate data, we do have a relevant benchmark to be able to do the cross-trial comparison to.
It is important to note that when you talk about the hurdle, once we do get robust and interpretable efficacy data, alpelisib's confirmatory study, EPIC-P2 in PROS, recently failed. It failed in a way, though, that was quite validating for the mechanism and the approach. We believe they used a subpar dose in that study. Not very surprising that they were not able to achieve the response rate hurdle that was necessary for the primary endpoint. They did see effect in both adults and pediatric patients. Where in the control arm where placebo was given, there was a 0% response rate. There was enough of an effect to continue to validate the mechanism and the approach. It leaves open the accelerated approval regulatory pathway to us because that confirmatory trial failed.
Right. Interestingly, if you establish proof of concept here in VM, you're actually not going to take RLY-2608. I think you're planning to take another asset forward, another PI3Kα. How is that going to work?
Not necessarily. We are availing ourselves of the option to do both, either take 2608 forward. As you referenced, we do have an additional molecule that's kind of close in to be able to give us the option to move into a distinct molecule. That really comes back to some of the strategic outcomes that happen over the next few years. It's good to have that optionality should we want a distinct brand and molecule in breast cancer and the ability to have a distinct brand and molecule in vascular malformations.
Right. I am going to squeeze in one last one. Obviously, in order to prioritize phase three development of RLY-2608, you're maybe going to try to phase the phase one, the upcoming phase ones for your other pipeline programs, NRAS, and also your Fabry disease program. How is that going to work in practice? Are you going to pause them altogether? Are you going to try to partner them? What's going on?
I think for this year, we'll continue to move them as fast as we can towards IND. Once we get to the IND, I think it will be a question around what's our clinical team doing. At the moment, it's fully focused on getting the pivotal up and running, running these triplets, running these doublets. I think it's a bit like what we won't do is hire a bunch of extra people for a little phase. We'll just see what they are focused on. Once the kind of bolus of work kind of settles down, we may bring either one of them into the clinic. It is really just around managing our resources in a conservative way. Obviously, we will look at options such as partnering.
Look, in the ultimate kind of situation with capital markets the way they are as we get into 2026, if we have to, yes, we will pause to make sure that we always have the capital to bring 2608 Rediscover to fruition, get the vascular malformations proof of concept data because we think those are the nearest term, highest value catalyst for the company. It's a position that if you biotechs are in with this kind of capital, our future is in our hands. We're not dependent on having to raise this money.
Team, terrific. Thank you so much for joining us. We really appreciate it. It's great to see you. We will follow closely.
Thank you.
Thank you.