Relay Therapeutics, Inc. (RLAY) Earnings Call Transcript & Summary

So good morning, and thank you again, everybody, for coming. I'm Pete Rahmer, the Chief Corporate Development Officer here at Relay Therapeutics. We're excited to unveil some new programs and give you a bit of a sense of how the platform continues to be extremely productive. Before we get going, just to let you know of some of the other relay team members that are here today. We have Dr. Sanjiv Patel, our CEO; Don Bergstrom, President of R&D; Jim Watters, our CSO of Late Research; Pascal Fortin, our CFO of Early Research; and Pat Walters, our Chief Data Officer. Additionally, we're fortunate to have with us today Dr. Adrienne Hammill from Cincinnati Children's Hospital, who will help kind of guide us all through the vascular malformations new program. And so we'll start by doing some prepared remarks on the programs. And then at the very end, we'll open it up to Q&A. The obligatory forward-looking statements -- in this presentation, we will be making various remarks that contain forward-looking statements that are made under the safe harbor provisions of the securities laws and which are subject to a number of risks and uncertainties including those described in the slide entitled Disclaimer and in the Risk Factors section of our most recent annual report and on Form 10-Q and K. Actual results or events could differ materially from the plans, intentions and expectations discussed here and those projected. The forward-looking statements in this presentation speak only as of the original date of this presentation. We undertake no obligation to update or revise any of these statements. Great. With that, now I'd like to turn it over to Sanjiv.

Thank you. Maybe -- right. Well, thank you. First of all, it's been a very busy week for all of you. I'm sure that many of you traveled from here to Chicago and then Chicago elsewhere and then finally, back here again. So we do appreciate it. We understood when we scheduled this, this is going to be challenging. And so I appreciate that you've all made the effort to attend. There's a lot to get through in the next 60 minutes. And so I'll try and just summarize what we hope that you'll take away, which is we were 1 of the first 8 years ago to use computational tools to try and make the discovery of medicines both more efficient and effective. And hopefully, as you'll see today, we have the 8 development candidates we have to date and 3 more that we're announcing that this is a very productive platform. And that over the medium term, we will have an extensive clinical program that is split across solid tumors, breast cancer and then genetic disease. And then over the medium term as well, we'll be able to show clinical proof-of-concept data that will derisk these programs. These programs are in large addressable markets. And they mean themselves to commercial models that are suitable to the companies of our stage and size. We have a team that's dedicated to research development and then bringing these medicines to patients. We have a capital balance sheet position of $750 million as of Q1. And so we have a wholly-owned portfolio outside of our ship too. So we have plenty of levers to bring all of this to fruition as we try and build a sustainable company. And so with that, let me give you an overview of the 4 updates that we'll make today. First update concerns PI3K-alpha. As you know, there's a huge amount of excitement around this target. Those of you who are at ASCO, those of you are at San Antonio Breast in December know that this target is now clearly therapeutically important in breast cancer. And we've always had the hypothesis that if you could build and create a mutant selective PI3K-alpha that dials out some of the toxicities associated with the current therapies and that greater tolerability profile should translate into greater efficacy for a broader range of patients. And over these last 5 years, our confidence in this hypothesis only grows. And so what you'll hear from us today is that we hope to plan to start a pivotal trial in 2025. We'll share the scope and scale of data that we'll share from our RLY-2608 trial at the end of this year with you. And that data will be focused on efficacy. We believe the data sets that we put out to date help derisk the fact that this is clearly differentiated on a safety and tolerability profile. And the question that I think we all have is how will that translate out into efficacy. And so we want as robust a dataset as we can with as much duration as we can and as like a pivotal population as we can. And so we'll be sharing that towards the end of the year. The other new news today is our press release that we put out last night, which is we signed a clinical trial collaboration with Pfizer to combine our selective PI3K-alpha mutant compound RLY-2608 with Pfizer's selective CDK4. We're thrilled with this because as you think about earlier and earlier lines of therapy in breast cancer, it's all about combinations. And so these patients are going to be on therapy for a considerable period of time. And so the tolerability profile of the stacking of combinations is important. So this is the first stacking of a selective CDK4 with a selective -- mutant selective PI3K-alpha. And so the goal is to be able to dial out as much of the tolerability challenges that traditional combinations have. And we believe that this selective, selective, novel, novel combination is very exciting. We hope to have this in the clinic before the end of the year. The second update we'll make today is an extension of our mutant selective PI3K-alpha portfolio into vascular malformations. There are over 170,000 patients in the U.S. that have a PI3K-alpha driven vascular malformations. Again, these patients are going to be on therapy chronically. So what's important to them is both efficacy, but also very important is the tolerability profile given the chronic nature of these treatments. And so again, we believe a mutant selective solution should offer a better therapeutic option for patients. And so again, we hope to have this in the clinic in Q1 of 2025. The next update we'll make is our second genetic disease program, Fabry disease. Fabry disease today inflicts about 8,000 patients in the U.S. It's a large commercial market with about $2 billion worth of sales of current therapies. One of those current therapies today is an inhibitory chaperone. And remember, the goal of these medicines is to try and activate alpha-gal. And so our team using our Dynamo platform has identified a novel allosteric pocket and created what we believe to be the first non-inhibitory chaperone. And the goal of that would be to increase the activation of alpha-gal and again, provide a better therapeutic solution for a broader range of patients. And so we hope to have this in the clinic in the second half of 2025. The final of our updates. And you can see now we've got a lot to say in 60 minutes, is that we're going to introduce you to the first selective NRAS inhibitor. This is a classic precision medicine play, which is there are about 28,000 patients with NRAS-driven solid tumors in the U.S. If we could produce as we have a deep and selective inhibitor of NRAS that hopefully avoids a lot of the off-target toxicities associated with the current therapies, again we believe this should be a better therapeutic solution for patients. And again, we hope to have this in the clinic in the second half of 2025. So that's a lot to update you on. And I think before we go into it, it's the first time we can see the totality of our mutant selective PI3K-alpha portfolio. If you take a few steps back, you start to see it. We've seen it for the last 6 years, which is why we built the franchise and molecules around it, which is you've heard us talk a lot about mutant selective PI3K molecules in hormone receptor positive HER2-negative breast cancer, 150,000 patients. What you're hearing from us today is we will hope to start a pivotal trial in 2025. We'll hope to share data with you that shows you a differentiated profile, hopefully, both in safety and efficacy at the end of the year. And we will, as we announced last night, start a novel, novel, selective, selective combination with RLY-2608 and CDK4. Our confidence in our hypothesis that the mutant selective inhibitor will reach a broader range of patients with greater efficacy is the same as it has been over the last 5 years. The second of our pillars that we've introduced today is a mutant selective PI3K-alpha inhibitors for vascular malformations. Over 170,000 patients. Here, we'll look to rapidly get proof of concept with RLY-2608 and then if over time, as we run the pivotal, we may use a distinct molecule for that. The final piece of the pillars is outside of breast cancer, solid tumors. And there, it's not something we're focused on today. We went straight into combination dosing, both in the doublet and the triplet. But we've gone back now and we've done some dosing as monotherapy. And we're seeing a range of partial responses across tumor types outside of breast cancer. So at the right time, we believe that this is a portfolio pillar that we can address. So if you look at the totality of it, the 3 pillars, the several hundred thousand patients, it's an opportunity that you really need a broad portfolio of assets to address. And we believe, as you know, we have a broad portfolio of mutant selective PI3K-alpha inhibitors with the first of those being RLY-2608. And so we're excited to what we believe to be addressing 1 of the largest precision medicine opportunities out there. So with that all said, let's get into where this all came from. The platform that we have is productive on any metrics that you want to look at, 8 development candidates over the last 8 years, 4 INDs, 2 clinical data sets that show proof of concept. We're announcing today 3 more INDs over the next few years, 3 more clinical starts. The platform is a combination of computational and experimental tools. And people talk a lot about this, as you know, there's a lot of hype, everywhere you look around AI and the impact on society. For us, it's a very granular thing using computational tools and the platform isn't a button. It's a series of tools, both on the experimental side and on the computational side used by bilingually qualified and versant scientists. And the tight integration of computational tools with experimental tools is what drives our platform. Eight years ago, we had computational tools and we had experimental tools. Today, we just have tools. And I think that's how biotech companies will look in the future. Again, people talk about AI as if it is a discrete thing, we're using AI. We make medicines through thousands of different steps all linked together. And we combine experimental and computational tools and with the scientists that we have to try and make each of the thousands of steps involved here slightly better, slightly more efficient, slightly more effective, and we stream it all together to make the entire process more efficient and effective. But what you'll hear from us today is not hype and hyperbole, but real tangible examples. And so for example, we'll go through each of the vignettes from the programs that we have just announced and show you how we've used these tools. A great example here is in our selective NRAS inhibitor. We found a valid hit experimentally for our NRAS inhibitor. Once we found it, we created using computational approaches a giant amount of computational evaluation of chemical space to see what kinds of molecules could potentially bind in that same binding site. And then using both computational algorithms and how medicinal chemists looking at these predictions we predicted about 30 compounds to actually make. And those compounds are made by a combination of humans with Pete in the wet lab and robots and the first 1 of those 29 compounds that we synthesized was a very potent inhibitor of NRAS. And obviously, we generated in vivo POC data very rapidly afterwards. That process took us about 120 days to do. And that was done through a combination of people, the wet lab, robots and computers. And I think what we've learned over the last 8 years is you need all 4 components. And the company of the future will be a company that consists of people, robots, computers and the wet lab, but it's the tight integration of how those 4 elements come together that really drives this. So what you're seeing from us is a very granular demonstration of how this actually comes to fruition. A lot of companies, I think, in this space talk about inputs. We've got a lot of inputs. That's great. The only reason for having inputs is that you then generate outputs. And so for us, what we're showing you today is 8 development candidates that we've generated with our inputs. And we're seeing 3 more that we've generated with our inputs. And I think we urge you to judge companies like us based on are they solving hard problems? And are they solving lots of hard problems? And are they bringing real medicines into the clinic? All this says that we have a very powerful platform, and we direct that platform against the lowest biological risk that we can. And so the targets that we choose are genetically defined patient populations clinically validated spaces, large addressable markets, which are commercially attractive. And that's why you see the portfolio that we have. And so we have a very deep portfolio now spread across breast cancer, genetic disease, and solid tumors. And so again, you're going to see lots of detail now, but I urge you to take away the key messages, which is we are one of the most experienced exponents of using computational tools in our industry to make the discovery of medicines both more efficient and effective. We've demonstrated that time and again, 8 development candidates, 3 more today. We have a broad clinical portfolio now heading towards generating meaningful data that will derisk it. These markets are large, they're commercially attractive. They are commercially amenable to a company like us. We have an extensive balance sheet of $750 million. We wholly own this portfolio outside of [indiscernible]. And so our goal is to generate a sustainable company in business. And I think we have all the tools to bring it all to life. And so with that, we'll get going into the update. So there's 4, and we'll walk you through them 1 at a time. The first 2 will be from our mutant selective PI3K franchise. And I think Don will also cover the third pillar here. So with that, I'll hand it over to our President of R&D, Don Bergstrom.

Great. Thank you, Sanjiv. So I'll start with breast cancer. When you look at the breast cancer opportunity for PI3K alpha, the prevalent population of ER-positive HER2-negative PI3K mutant breast cancer in the U.S. is about 150,000 patients. So clearly, a very large opportunity. Now within breast cancer, we've also looked at building a franchise of wholly-owned assets. And as we anticipate where the treatment landscape is going in breast cancer, clearly, for PI3K, we feel it will move to meet the selective PI3K inhibitors. We have RLY-2608. As you look at targeting the estrogen receptor, we have at the IND ready stage are highly potent degrader of the ER receptor. As you look now at the CDK space, we have our own selective CDK2 inhibitor sitting at the IND stage. And with today's announcement, we've also accessed the selective CDK4 inhibitor. So we have 1 of the broadest portfolios of next-generation breast cancer therapies in the biotech space. And we're looking at really using our portfolio to evolve the treatment of breast cancer for these patients in need. Now the new news today is that we've entered into a clinical trial agreement with Pfizer to combine RLY-2608, our mutant selective PI3K inhibitor with atirmociclib, their selective CDK4 inhibitor. We feel that this combination of 2 selective agents will allow us to optimize the inhibition of these key driver targets in breast cancer, while avoiding the nontarget mediated toxicity associated with CDK6 inhibition and associated with wild-type PI3K-alpha inhibition. And again, it speaks to our commitment to making sure that RLY-2608 is a foundational therapy for the treatment of patients with PI3K alpha mutated breast cancer. In the RLY-2608 program, we have a very broad development effort that is actively ongoing. As a reminder, about 18 months ago, we started doublet development of our RLY-2608 with fulvestrant. We've now completed dose escalation. We have identified 600 milligrams as an optimal dose. We announced last year that we were opening cohort expansion at the 600-milligram dose. And by the end of 2023, we had enrolled about 40 patients with a combination of fulvestrant and RLY-2608 at the 600-milligram dose. As we look forward to the disclosure that we'll make later this year, we anticipate that at the time of that disclosure, we'll have over 60 patients who have been treated with 600 milligrams of 2608 in combination with fulvestrant. And of these patients, we anticipate at least 40 will have at least 6 months follow-up. Now that's critical because that will give us the ability to start estimating a clinical benefit rate and a landmark PFS analysis at 6 months. So really understanding what the durability of benefit is in these patients. In our safety database, we anticipate we'll have over 100 patients who have been exposed to RLY-2608 in combination with fulvestrant across the doses that we've tested including a 400-milligram dose cohort that we're enrolling for project optimist purposes. We also, this year, initiated triplet development of RLY-2608 with ribociclib, and today, we've announced that before the end of this year, we'll initiate another triplet with atirmociclib the selective CDK4 inhibitor. And we anticipate later this year to be able to provide early safety data on the ribociclib triplet. We haven't forgotten monotherapy in other solid tumors. That's the third pillar that Sanjiv mentioned. We continue to enroll patients in monotherapy and I'm pleased to announce today that we've seen partial responses across multiple solid tumor types. And importantly, given the pan-mutant selective mechanism of action of RLY-2608, we've seen responses across multiple types of alterations and types of mutations in PI3K-alpha. So again, in summary, very broad development program for 2608. We're enrolling patients across monotherapy, fulvestrant doublet, ribociclib, soon to be atirmociclib. And obviously, we'll look to continue to identify combinations with emerging standards of care across these disease areas, including breast cancer. The ongoing rediscover trial, currently enrolling to about 25 sites across 5 countries. That number will grow over the near future. So now as we look across the breast cancer space and look at the opportunity, if we look in the front line, we estimate there are about 18,000 patients annually in the U.S. with PI3K-mutated hormone receptor-positive breast cancer. These patients currently in the frontline are treated with anti-endocrine therapy and the CDK4/6 inhibitor, Current standard of care gives you about 20 to 24 months PFS. If we look in the second line, after patients progress on the CDK4/6 inhibitor, we estimate there are about 14,000 patients addressable annually in the U.S. currently treated in the PI3K space with alpelisib or capivasertib, current standards of care give you about 6 months progression-free survival. As we look at these opportunities, the data that we're generating over the course of 2024 will give us insights into how to position RLY-2608 for further development in the breast cancer space. And we anticipate with the data that we're generating this year, we'll be well positioned to understand what our options are for pivotal trial development. And we'll have a good data set to be able to take to the FDA and other regulators to have productive discussions around the design of our pivotal trials. And we anticipate starting our first pivotal trial in 2025. Now dialing in a little bit on the second line space, post CDK4/6 inhibitor therapy. As I mentioned, benchmark therapies in this space have about 6 months progression-free survival, whether you look at alpelisib in the real world, where it's been reported about 6 months PFS or in capivasertib in the context of the CAPItello clinical trial, where PFS in the post-CDK 4/6 setting was about 5.5 months. Now what we've seen with the recent launch of capivasertib is it appears to be taking significant market share. Why is that? While capivasertib is reported to have a lower rate of Grade 3 hyperglycemia than alpelisib, and this is a toxicity that physicians have told us repeatedly they don't like dealing with. Now even though it doesn't have better PFS, it appears that it's being used more broadly, despite the fact that capivasertib brings with it toxicities, including diarrhea and stomatitis that are challenging for patients. This population of post-CDK4/6 patients reflects the patient population we are currently enrolling in the rediscovery trial and the patient population that will report data out later this year. So what do we hope to be able to show when we do our data disclosure later this year. We hope to be able to show that we have lower rates of grade 3 hyperglycemia, than either alpelisib or inavolisib, lower rates of diarrhea and other toxicities than capivasertib and an overall favorable safety profile. We also hope to show, given what I've just mentioned is the benchmarks of 5.5 to 6 months that we have a PFS or CBR that would give us confidence that in a head-to-head trial against 1 of these agents, we would be superior in a pivotal trial. And that's why it's critical, as I mentioned to you earlier, that we'll have over 40 patients by the end of this year who have at least 6 months exposure to 2608. So again, just to summarize where we are with the breast cancer portfolio for 2608. We'll have a data update later this year in the fourth quarter. And again, we'll show doublet safety and efficacy data and initial triplet safety data with ribociclib. We'll initiate our CDK4 combination with atirmociclib. And then finally, we anticipate we'll have a potential pivotal trial start for RLY-2608 in breast cancer in 2025. So I'll now move into a new disease area that we're pursuing. And this is we're also pursuing initially with RLY-2608 given that it is the mutant selective PI3K-alpha inhibitor, where we have over 100 patients worth of experience to date. And this disease space that we're going to is vascular malformations. Now vascular malformations, are a set of diseases that affect about 170,000 patients in the U.S. These patients require chronic therapy. Currently, there's only 1 approved therapy in this space, VIJOICE, which is the VMs brand for alpelisib, but it's approved only in a very narrow population of patients with a disease called PROS, which we'll learn about in a few minutes. Now we feel that the lack of selectivity of alpelisib in this chronic disease provides us the opportunity for our mutant selective PI3K-alpha inhibitors to assess whether mutant selective potent inhibition of PI3K-alpha can lead to greater efficacy and greater chronic tolerability for patients with these diseases that require long-term therapy. So this is a disease area that for many of us is a newer disease area. We're very fortunate to have Dr. Adrienne Hammill from Cincinnati Children's Hospital here with us today. Dr. Hammill will now take a few minutes to talk to us about the biology of vascular malformations, the patient journey, current treatment options and some of the subtypes of the disease. So with that, I'm pleased to welcome Dr. Hammill to the stage to introduce us to vascular malformation. Thanks, Dr. Hammill.

Thank you. All right. So I'm really excited to be here this morning to talk about vascular anomalies. This is something that's really important to me, and I spend the majority of my time taking care of these patients. So I'd like to say thank you to listening to this for my patients who live with this and their families who deal with this and all of my colleagues who are trying to do better for our patients. So first, I'll say that like a lot of the medicines that we use in this field, I actually started out with a focus on oncology. And along the way, I actually chose pediatrics, realized that I preferred clinical research to bench research and was eventually recruited into the vascular anomalies field in 2009 in time to join the first-ever prospective clinical trial for vascular anomalies that was done in Cincinnati by Dr. Denise Adams, my mentor. I now serve a variety of roles at Cincinnati Children's in vascular anomalies and my interdisciplinary group takes care of a lot of patients with these. We see about 900 new patients a year. We have about 3,000 patient visits, including about 2,200 unique patients, and we follow a cohort of over 4,000 patients who are active, have been seen in the last 3 years. So vascular anomalies is a big umbrella term. We use this ISSVA classification. ISSVA is the International Society for the Study of vascular anomalies that was founded in 1992 and it's multidisciplinary. And they came up with this classification so that we would all use the same words and talk about the same things because the literature in vascular anomalies is very messy with a lot of different terms. So this is the definitive classification. It was approved in 2018 with some updates, there's another 1 coming, but it's not too different. And the big first division that we see in vascular anomalies is between tumors, which are things that grow and malformations, which are things that are the result of abnormal vascular development, they're present at birth. They stay with that person as they grow and develop, they do not go away on their own. This overall group of vascular anomalies is not rare because infantile hemangiomas happen at such a high frequency, but many of the individual diseases qualify as rare disorders and are very uncommon. Moving down to the vascular malformations. These are generally divided first by the type of vessel that's abnormal within that vascular malformation. And this can be what we say, simple with a single type of abnormal vessel, capillary malformation, lymphatic malformation, venous malformation, 1 of the arterial malformations or you can put them together in any combination you can imagine. Big alphabet soup, the most common we see is capillary venous, lymphatic venous or capillary lymphatic venous. And they can also be associated with other anomalies and syndromes that I'll talk about in a minute. The other big way that we talk about vascular malformations is by where they are and how many they are. So they can be localized with a single lesion on a single body part. There on the left, you see 1 little venous malformation on a tongue. But the one in the middle is also a localized isolated lesion, but obviously much more significant taking up half of a face on this child. And they can also be found in multifocal with multiple distinct lesions or diffuse where they involve large body segments, cross joints, cross tissues. And as I mentioned, can be part of a syndrome like you see there with a lot of overgrowth. So you see the capillary malformation, the lipomatous overgrowth. There are other anomalies in that child. And it turns out that it is all about cancer. So we realized that the inherited vascular anomaly syndromes that we see are actually due to germline mutations that are inactivating in well-known tumor suppressor genes, The tumor suppressor genes are shown there in red as well as stop signs. So just like in cancer, loss of negative regulators result in over activation of these pathways with consequences. It took us another 15 years to realize that the rest of the vascular malformation that we hadn't been able to figure out by looking at blood are actually due to somatic mutations that are activating in well-known oncogenes. And those are shown here in green. And it turns out that the phenotype is really largely related to the degree and the distribution of the mosaicism for those abnormal tissues. And when you put these all together, you get a really nice picture first of most of our vascular malformations shown all over here in blue, but also of these 2 pathways that are well known to people who work in cancer. This PI3 kinase/AKT/mTOR pathway, and the parallel RAS/RAF/MAP kinase pathway. We know that they have a lot of cross talk, but there's a lot of back and forth between them. And we know because they cause vascular malformation that these 2 pathways are critical for normal development of the vascular system in utero. And in fact, 1 of these spots, the PIK3CA gene, which encodes for PI3K-alpha is so important that it can generate a number of vascular malformations and syndromes. Anything from isolated venous or lymphatic malformations all the way to really extensive syndromes. There are so many things that result from pathogenic variants in PIK3CA that we've called it a whole spectrum. The PIK3CA-related overgrowth spectrum or PROS, which was already mentioned. And these can include vascular malformations or not. Some of them are simple overgrowth. They can include overgrowth in many different tissues, bone, fat, muscle, and they can be just like vascular malformations, they can be isolated, for example, focal cortical dysplasia in the brain or macrodactyly, over growth of a single digit or they can be really complicated diffuse extensive and dangerous like something called CLOVES, which is an acronym there, you can see in the third line for congenital lipomatous overgrowth, vascular malformations, epidermal nevi, scoliosis, skeletal and spinal anomalies. So some of these are really dangerous and difficult -- and altogether, all of those gene changes from -- in PIK3CA that result in vascular malformations, make up the majority of the phenotypes that we see in our vascular anomalies clinics. On the right, you can actually see 2 of our patients who are affected with significant overgrowth. So those are patients with PROS. And these 2 particular little girls, I would actually say they have CLOVES due to the lipomatous overgrowth on their trunk. And so you can imagine how this can interfere with function, with life and cause pain and various other issues. So how are patients diagnosed, referred and treated. Unfortunately, it could take a really long time for this diagnosis to be made partly because even with the same genetic change, even with the same syndrome, they can look all sorts of ways depending where the disease is. And they also present differently depending on which symptom bothers them the most. So you can imagine that if it's capillary malformation that they're worried about, they'll go to the dermatologist for laser. If they're worried about this big mass on their back, they might present to a surgeon to take it off or if they have a growth on their tongue, they might present to an ENT. And because these are so rare, chances are that the person that they show up to see has never seen it before and might not recognize it as a syndrome or part of that spectrum and may even offer the treatment with the tools that they have. Ultimately, diagnosis is usually made when the tissue comes back with a PIK3CA pathogenic variant or potentially when they make it into the office of someone who's really familiar with vascular anomalies and recognizes this pattern and calls it what it is. And then once that diagnosis is made, you can develop a multidisciplinary, comprehensive plan for treatment for that patient that's individualized to their needs. And those treatments can include anything from expectant management, watchful waiting if they're not having symptoms and it's not progressing. We can also offer supportive care things like compression therapy for patients who have venous or lymphatic malformations. If their symptoms are just local, we might offer a local therapy such as laser, sclerotherapy or surgery for resection or debulking of a problematic area. And then systemic therapies have usually been reserved for patients who failed local therapy, either because the response was inadequate or they had multiple recurrences or those diseases are too multifocal or diffuse to be addressed by other methods. And our systemic therapies have changed the landscape for our patients. It's completely different than when I started in 2009, when the conversation was more like parents asking us, will my child grow up? Will my child ever walk? Can my child ever go to school? So we're doing a lot better, but it's not good enough. The quality of life and other issues for these patients is still a challenge, partly because of the very real adverse effects that can happen with any of these systemic therapies that we have available. So that limits the doses that we can use and even limits our patient selection. Currently, our systemic therapies are still used off label in many cases. Sirolimus, which we've been using for over 15 years for a variety of lymphatic anomalies was never approved for any vascular anomalies indication. And alpelisib has been approved for PROS, for PIK3CA-related overgrowth spectrum, those overgrowth phenotypes in patients over the age of 2. But you can see that there are an awful lot of patients who are not covered by that indication. And there are a lot of people out there who could really benefit from PI3K-alpha inhibition, particularly if we can make it more efficient or less toxic or both. And with that, I'll hand it back to Don.

Great. Thank you, Dr. Hammill, and thank you for joining us here today. So as we heard Dr. Hammill mention, there are currently only 2 systemic therapies that are routinely used in the treatment of patients with vascular malformations. Alpelisib currently has a label in the PROS subtype of the disease. But of course, we know from the experience in the cancer population that alpelisib has challenges achieving full inhibition of PI3K-alpha while being able to remain tolerable because of inhibition of the wild-type PI3K alpha. And then sirolimus used exclusively off label in these diseases was designed originally as immunosuppressant for solid organ transplant, clearly immunosuppressive by design. And we feel, as we look at these therapeutic options and as we talk to physicians like Dr. Hammill, who indicate that there's clear need in this patient population, that the ability to be able to bring forward a mutant selective PI3K-alpha inhibitor to be able to hit the driver in these lesions in these patients while maintaining systemic tolerability chronically could give us the opportunity to both help patients who are currently treated with systemic therapy as well as potentially expand the pool of patients for whom systemic therapy might be considered. So again, we look at RLY-2608 has the profile of what we think we need for these patients. We're bringing it forward as a -- to establish rapid proof of concept in the vascular malformation space. And then because we have such a deep understanding in the PI3K office space, we've been able to build a portfolio of pan-mutant selective PI3K-alpha inhibitors and feel we could have the option and opportunity to then use a distinct molecule for pivotal studies in this space. Just as a reminder of where our portfolio of PI3K-alpha inhibitors came from. This is really the application of our Dynamo platform. starts in the lab with solving the first full-length structures of mutant and wild-type PI3K alpha. Then goes in silico, being able to do long time scale molecular dynamic simulations using that experimental data to identify for the first time a novel allosteric pocket that would allow for mutant selective inhibition of PI3K-alpha. And then finally, using our platform for drug discovery and discovering molecules like RLY-2608 that we've now shown are mutant selective PI3K-alpha inhibitors where we can hit the target while sparing many of the toxicities associated with wild type PI3K-alpha inhibition. So as a recap, completely novel approach for the treatment of this disease to our knowledge, the first mutant selective PI3K-alpha inhibitor that will go into the clinic for the treatment of PI3K alpha-driven vascular malformations. Large population of a potential 170,000 addressable patients in the U.S. with an opportunity to grow the use of systemic therapy in this patient population, should we be able to show that mutant selective PI3K-alpha inhibition is both more effective and more tolerated over chronic use. We feel, again, that the ability and the opportunity here to establish rapid proof of concept with RLY-2608 will establish the second pillar of our PI3K-alpha franchise portfolio that Sanjiv just mentioned earlier. So with that, I'm going to hand over to our CSO of late research, Jim Watters, to talk about our next new genetic disease program that we'll be advancing into development.

Thank you, Don. I'm happy to talk about our second genetic disease program, which is the first non-inhibitory chaperone of the protein alpha-galactosidase for the treatment of Fabry disease. Fabry is a debilitating lysosomal storage disorder that's caused by mutations in the gene that encodes alpha-galactosidase or alpha-gal for short. There are about 8,000 patients in the U.S. suffering from Fabry with existing therapies, all having limitations. So this opportunity really fits our strategy of solving technical problems others haven't been able to solve. And keeping translational risk low because here, we have genetically defined patients, a clinically validated target, large unmet medical need and compelling commercial opportunity. As I mentioned, Fabry is caused by mutations in alpha-gal. There are over 1,000 different mutations that had been reported in alpha-gal that could cause Fabry. And what these mutations do is they remove or reduce the function of alpha-gal. Alpha-gal is normally degrading its substrate, which is called Gb3 in the lysosome of cells. So in the presence of these disease-causing mutations, the level of Gb3 accumulates them cells to toxic levels. That results in broad clinical manifestations across multiple organ systems with patients frequently experiencing life-threatening cardiac or renal dysfunction amongst other symptoms. This is a large and established market with multiple approved therapies. The first mechanism to be approved in Fabry was enzyme replacement therapy, or ERT. This is when patients are administered the normal or wild-type enzyme through biweekly intravenous infusion. And ERT is an active therapy, but it has multiple limitations. For example, it requires chronic biweekly IV infusion. The half life of their protein is short, so the active protein is not present for the duration of this 2-week interval. Patients can experience injection site reactions. They can develop antidrug antibodies. So limitations here for ERT despite that projected peak sales of ERT at around $1.6 billion annually. There's another approved mechanism in Fabry, which is a small molecule chaperone, known as migalastat, marketed under the trade name Galafold. And by chaperone, we mean a small molecule that can bind to alpha-gal mutant protein and stabilize it in an attempt to recapture some of the normal wild-type activity. And migalastat can bind to a subset of alpha-gal mutations and stabilize them, but it has a critical liability, which is it is also a potent inhibitor of the activity of alpha-gal and it's this inhibitory function that results in multiple clinically important liabilities. For example, we get limited activation of alpha-gal mutants, there's limited mutational coverage. Only about 40% of Fabry patients have mutations that are responsive to migalastat, otherwise known as amenable to migalastat. And this categorization of amenable nonamenable is on the drug label. Finally, migalastat is not combined with enzyme replacement therapy. So there's no clear pathway for migalastat to access more than this, about 40% of patients. Despite those limitations, the peak sales of migalastat are projected to be around $780 million annually. That really shows you the extent of the unmet medical need here and the size of the potential commercial opportunity. And, in our view, creates a clear need for the development of the non-inhibitory chaperone of alpha-gal, which we believe will overcome these limitations resulting potentially in a best-in-class therapy. Let me talk a little bit more about alpha-gal, mutations and chaperones. Here on the left-hand part of this slide, you can see in pink a representation of normal alpha-gal protein with the lightning indicating its activity. It's normally degrading its substrate Gb3 in the lysosome of cells. Here, we show a disease-causing mutation that destabilizes alpha-gal. The protein becomes unstable, it unfolds and then degrades. That's what causes reduced activity that leads to Fabry. I mentioned chaperones and migalastat is an inhibitory chaperone. What you can see here is that migalastat can bind to alpha-gal and stabilize it, but it binds in the active site. So it's also a potent inhibitor. So the protein is actually not active in the presence of the drug. Only when the molecule dissociates do you get a transient window of activity before the protein then degrades again. So you have this transient window of activity in the way that migalastat deals with that is by dosing every other day in the clinic, if you dose it more frequently, you paradoxically get less activation. What we've discovered is the first non-inhibitory chaperone that we show here in blue. It binds to a novel allosteric pocket, stabilizes the protein. But now because it's not inhibitory, the protein is active in the presence of the molecule. So our molecule can be given continuously resulting in continuous activation. And this is continuous as opposed to transient activation that we believe will result in a better therapy. So we use multiple components of our platform to enable this discovery. We integrated computational and experimental approaches here. First, we discovered a new allosteric pocket, utilizing a combination of structural biology and computational approaches. We then found chemical starting points in that pocket using multiple experimental and computational hit-finding approaches, Then we progress these chemical starting points into lead series toward development candidate, using other tools like high-throughput, automated chemistry, ADME, PK modeling and others. And it is through this broad integration of computation and experimentation that we identified molecules that met our target profile of being non-inhibitory alpha-gal chaperones. Here, we show biochemical inhibition of alpha-gal, by migalastat in red and 1 of our lead molecules called RTX1 in blue. I think you can see that migalastat is a potent and strong inhibitor of the biochemical activity of alpha-gal, whereas RTX1 in blue shows no inhibition whatsoever. It is this lack of inhibition that we believe will overcome the main liabilities of migalastat, namely allow us to show superior activation of alpha-gal mutants compared to migalastat, give us a broader mutational coverage and should be readily combinable with ERT since it's non-inhibitory. And over the next few slides, I'll show you data supporting these 3 points of differentiation. First, we show head-to-head activity of RTX1 and migalastat for a known disease-causing mutation, which is called R301Q. On the left-hand part of this slide, we show an in vitro cellular assay, looking at alpha-gal activation in cells by RTX1 in blue and migalastat in red. First, you can see that the blue curve is less shifted significantly compared to the red curve showing that we have more potent activation of alpha-gal in cells. Also, you can see the blue curve goes up higher showing we have a higher overall maximum effect and a larger degree of alpha-gal activation. The middle panel shows that this improved page activation translates in vivo. The middle panel shows a mouse model, where we've replaced the mouse alpha-gal gene with the human alpha-gal gene that contains this R301Q mutation. And here, we dosed this model with vehicle the clinically relevant dose of Migalastat in red and 2 doses of RTX1 in light blue and dark blue. I think one, you can see we have a dose-dependent activation of alpha-gal in vivo by RTX1. And two, you can see that we have more activation of alpha-gal by RTX1 compared to the clinically relevant dose of migalastat. And on the right, we show that this increased activation translates into better substrate reduction, so better Gb3 reduction as expected because this molecule degrades -- this protein degrades Gb3. So as we get more activity, we should see less substrate. That's what we see, dose-dependent decrease in the substrate by RTX1, and we get more degradation of the substrate compared to the clinically relevant dose of migalastat. So in vitro, in vivo, head-to-head, superior activation and substrate degradation. There were no adverse findings in an exploratory rat toxicology study of RTX1 at exposures that are equivalent to the high dose, the dark blue of RTX 1 in this mouse experiment suggesting that this mechanism should be well tolerated. Next, we show that this superior activation is not limited only to 1 mutation. In fact, here, we see it broadly across a panel of 38 different mutations that we've tested in cells. And this plot shows increase in alpha-gal activity by RTX1 relative to migalastat, which is set at 100% for each mutation. And a couple of observations here. We see RTX1 has a better alpha-gal activation across the board on every single mutation. And second, that is true whether or not the mutation is categorized as amenable or not amenable to migalastat according to the drug prescribing information. This suggests that a non-inhibitory chaperone will have a broader mutational coverage in migalastat, also suggests that a non-inhibitory chaperone can be a treatment option for patients that cannot be treated by migalastat because they have non-amenable mutations. And finally, given the non-inhibitory nature of our molecule, we believe it should be readily combinable with ERT, and that's what we show on this slide, in vivo combination of RTX1 plus ERT compared to ERT alone. And this is a mouse model with the alpha-gal gene knocked out. On the left, we show alpha-gal activity, the combination of RTX1 plus ERT results in about eightfold increase in alpha-gal activity compared to ERT alone. And on the right, we show substrate reduction where the combination of RTX1 plus ERT results in a 45% more degradation of the substrate compared to ERT alone. As these data suggest that a non-inhibitory chaperone is combinable with ERT and in fact, can increase the activity of ERT, providing it a pathway for a non-inhibitory chaperone to potentially access all Fabry patients. So just to summarize and wrap it up, we've used our platform to discover the first non-inhibitory chaperone of alpha-gal. We believe that this will allow our molecules to overcome the key liabilities of migalastat potentially resulting in a best-in-class therapy in a disease that presents an attractive commercial model. And we look forward to a clinical start in the second half of 2025 for this program. With that, I'll hand it off to Pascal Fortin. Thank you.

Thanks, Jim. Moving us to our last new program update for today and the newest entry to our solid tumors portfolio NRAS. So NRAS is a known frequently mutated oncogene that is known to drive the growth of multiple tumor types. It represents a large [indiscernible] market with a significant unmet need. In fact, in the U.S. alone, more than 28,000 patients every year get diagnosed with NRAS mutated cancers. Curing therapies have been used against this target of suboptimal due to the fact that the therapies themselves are not selective for the target. So what we'll show you today is how we've been able to deploy the Dynamo platform and for the first time, uncover this cryptic allosteric pocket, that allows us to actually inhibit this target exquisitely with a highly selective mechanism. So the RAS MAP kinase pathway is the most frequently mutated pathway in cancer. And as such, has been targeted through multiple of its nodes using a diversity of [indiscernible] agents that many of which are in current development today. However, until now, nobody has been able to generate an NRAS selective agent to really inhibit this target towards the top of the pathway with high selectivity. In cancer, NRAS mutations are observed broadly, notably in non-small cell lung cancer and melanoma and also in colorectal cancer. So these patients represents a high medical need and could really benefit from a much more efficacious, much better tolerated drug. So what you see on the top here is the RAS MAP kinase pathway again. Multiple agents have been developed against all of the nodes of this pathway, namely Pan-RAS,Pan-RAF, [indiscernible] inhibitor. However, none of these agents can selectively target NRAS. So it leaves a lot of efficacy on the table because it brings a lot of toxicities to the equation and to the treatment paradigms of these patients. Namely, we see toxicity such as rash and liver toxicities. This results in very limited target inhibition and engagement that you see and many patients have to go through dose reductions, but also those discontinuation. This ultimately lead to limited efficacy in the clinic where we see poor response rates, but also very limited progression-free survivals. So what we've done at Relay is deploy the Dynamo platform and really bring protein motion and protein dynamics to the center of drug discovery. And what we've been able to see is what you see on the left side of this slide here is different RAS isoforms, NRAS, HRAS and KRAS, again, the lightning signifying the activity of these proteins. What we've been able to observe is the fact that NRAS moves very differently from these other off-target isoforms, HRAS and KRAS. This allowed us to actually identify a new cryptic pocket that exists in NRAS that is rarely seen in the other RAS isoforms. So what you see in green is a molecule that we could design to potently bind the NRAS isoform while leaving its off-target isoform NRAS and HRAS unengaged. So -- if you want a little more of a sneak peek to exactly how we've done this, we've been able to integrate to the platform experimental and computational tools to further validate the existence of the critic pockets here on the left, where we have applied structural biology techniques, fragment screening and virtual screening methods to really not only validate the expense of that pocket but find early chemical matter that will be able to bind that pocket. Through these concepts, we've actually deployed some of these principles and designed multiple hit-finding techniques to identify early chemical starting points that would engage this pocket very selectively while it wouldn't have no ability to bind a pocket in HRAS and KRAS that didn't exist in these isoforms. So this led to the discovery of these early chemical starting points then we could validate through biophysical techniques such as NMR and 2D and also use computational methods to merge these early chemical starting points together and generating much more potent, much more selective leads. Ultimately, we were able to use AI-enabled free energy calculation methods to further our designs of these molecules and rapidly make these molecules using our high-throughput automated chemistry platform that ultimately led to the generation of development candidates. So again, what we've been able to do is to uncover new opportunities in the protein, find a cryptic pocket to target NRAS specifically and made molecules against it. So the end result here was NRAS inhibitors were very potent, very selective and very active against NRAS mutations. On the left, you see the biophysical binding of these molecules, here's 1 example, RTX2, that can specifically bind to NRAS mutations and wild-type in the on in off state. Now this is an important point because the on state of this target is the 1 that's triggering MAP kinase signaling that leads to cancer cell growth. And while it could bind these NRAS states very potently, it was unable to engage HRAS and KRAS off-target isoforms, again, highlighting the high degree of selectivity of these molecules. This also translated well in cells. And so what you see in the middle panel and the blue lines is how these NRAS mutated cell lines are potently inhibited at the phospho-ERK level, which means we can inhibit the MAP kinase signaling in these cell lines that are NRAS mutated, while if you take an NRAS wild-type cell lines with normal NRAS, but is otherwise mutated in another oncogene KRAS, there was no inhibition observed. This also translated in proliferation assay. So if you measure how these cells are growing, we can actually inhibit the cell growth of NRAS mutant cell lines shown here in pink on the right side. But when we tested cell lines that were otherwise NRAS wild-type, but mutated at other oncogene levels, so KRAS, HRAS and BRAF, for example, there was no inhibition observed, again showing a high degree of selectivity for our molecules against NRAS as a target. This also translated in vivo. So we've ran in vivo [indiscernible] assays. And on the left side here, we show 1 example of 1 of these experiments where we've taken an NRAS Q61R mutant melanoma cell line xenograft, that we've treated the benchmark molecule, so binimetinib and exarafenib-binimetinib combination. And we've used those combinations at doses that were relevant to clinical exposures that we see in patients. And we also treated these models with RTX2 and RTX4, our 2 NRAS inhibitors, in this case, with RTX4 being a more advanced molecule that selectively inhibit NRAS. Important to note here that RTX2 and 4 were dosed overly on a daily -- once-daily schedule. So we can observe here in this graph is the red line, which is the exarafenib-binimetinib combination benchmark, where we see very limited tumor responses. In contrast to this, if you look at the blue lines on the left panel, you see a deep and complete regression of the -- in this model for RTX2 in the light blue and RTX4 in the dark blue. For RTX4, we could see complete regression for only 14 days at a dose of only 10 mg per kg dose daily. On the right side, you can see that treating these models with these molecules, these NRAS substrate molecules were very well tolerated. There was no change in the body weight and we could also see no adverse events when we run an exploratory rat toxicology study using doses of RTX2 equivalent to 100 mg per kg dose daily. So here you have it, we've been able to show you that we've been able to generate a first-in-class NRAS elective inhibitor through the power of the Dynamo platform, and we very much look forward to a clinical start for this program in the second half of 2025. With that, I'd pass it over to Pete to wrap up this presentation.

Thanks, Pascal. And thanks, everyone, for tuning in for the entirety of today. We're very excited to have been able to show you the 3 new programs that were announced today, new innovative programs coming directly from the Dynamo platform. Obviously, they continue to address very large market opportunities. They're unique in each of their own way, but have very common themes that you've seen throughout the entirety of the Relay pipeline, which is they address clinically or genetically validated targets, address very large markets. And most importantly, we're solving problems that can't be solved otherwise. These are very known problems that needed to be addressed for patients, and we were able to put to force the power of the platform and the scientists at Relay to come up with novel ways to address patients with high unmet medical need. These new programs continue to be firmly underpinned and anchored by our PI3K-alpha franchise. We've talked for a number of years about the breast cancer opportunity. Today, we got to go in depth on the vascular malformations opportunity in front of us and look forward to continuing to expand upon the solid tumor opportunity that we intend to address in the future. As we continue to evolve and learn about the most critical tools that we use daily within the platform, we've also been able to evolve and start to integrate some of these capabilities more in-house where we've traditionally relied on external vendors or partners to help us with some of these techniques and tools. And some examples of those you can see here, namely long-time scale molecular dynamics. It was a -- when we originally started the company, we had a foundational partnership with D.E. Shaw research, but over time, have been able to now fully integrate molecular dynamic capabilities in-house that serve all our needs for our discovery stage programs. In the middle, you see something called Rel-Del. That is a -- over the years, we've been able to create bespoke DNA-encoded libraries that are specifically tailored for our machine learning needs in-house, and we have that deployed throughout a lot of the early discovery research that we do. Lastly, high throughput automated chemistry, which allows us to really increase the cycle times for chemical synthesis, which as you all know, is 1 of the most time-consuming portions of the drug discovery process, and it's allowing us to more rapidly get to development candidates. And again, as Sanjiv mentioned, the platform is a living thing. It continues to evolve. We continue to bring more tools online, both experimentally and computationally and the team is getting more even tightly integrated, where these capabilities really transcend either side of this circle and the circle becomes more tightly integrated. As we mentioned at the beginning, this all only really matters based off the outputs that you're able to produce. What are the number and how distinct are the molecules you're producing, the difference you're making for patients. Again, we hope you continue to measure us by that metric and measure others in the field by that metric. So in closing, again, thank you for tuning in, both online and in person. I hope you take away a greater understanding of the power of the platform, how it continues to produce amazingly novel programs that are serving high unmet needs for patients, how we're focusing our research and where we spend our time there. It's not something we've talked a lot about over the past few years. So there's an immense sense of pride both here in the room with the 6 folks that have been able to present it to you. But most importantly, the way we're able to represent this research from all our folks back in Cambridge. And there's a lot of updates coming over the next couple of years. On the data side, obviously, the highly anticipated data coming from our breast cancer program, RLY-2608 in doublet therapies. This, we think, at the end of the year, will really help establish why we believe we're definitively first-in-class in this field. And then we have that backed up by moving into standard of care combinations with CDK4/6. And most importantly, we're moving to where the future of standard of care is going in this field in combinations with CDK4. And so the robustness of this breast cancer franchise has never been stronger. Our confidence in our ability to win here has never been stronger. And obviously, as we talked about today, a lot of clinical starts over the next 18 months. As we continue to progress forward, the fundamentals that brought us here will continue to be what drives us forward into the future. That's namely great science, amazing people, and a very strong balance sheet that will help us prosecute across everything we have in front of us and continue to build this into 1 of the greatest biotechs we've seen yet. So with that, I'd like to ask the speakers to please come up and join me on stage, and we'll do -- move into the Q&A portion of the event. Please wait for the microphone to come to you, so everyone on the webcast can hear you speak.

Okay. So maybe you could activate the mics again. Thank you . There was a lot there. And maybe we appreciate 1 last time you coming and attending during the very busy week. So Megan, do you want to -- Andrew?

Thanks, and thanks for doing this. Very, very helpful to see the drug discovery engine. For the NRAS program, I think I saw an inverse response in the in vitro data. Just wondering, are you planning to combine that with a MEK inhibitor? Are there any concerns about paradoxical activation?

Let me hand it over to our CSO, Jim Watters.

Yes, sure. We've not observed any in paradoxical activation of pathway signaling after NRAS inhibition. As you saw from the data, with just our inhibitor, we have complete suppression of signaling tumors going to 0 in vivo. So far, we don't see any evidence of pathway reactivation, there's strong monotherapy activity preclinically.

Salveen?

Could you just speak to how IRA is being taken into consideration as you think about trial design and regulatory outlook on the Ford? And so with PI3K-alpha, will you use different assets within breast cancer but also in the vascular indications and how you bring CDK4 into your registrational pathway here?

Maybe, Pete, you can just take that one.

Yes. So we've thought about the implications of the IRA throughout the recent evolution of the pipeline. Certainly, as you mentioned, Salveen, the ability to have multiple PI3K-alpha inhibitors as we think about the breadth of indications we can pursue is extremely helpful, very unique to us. And as we -- the other consideration is even within the RLY-2608 application in breast cancer, you see us moving quickly to establish safety, dose and eventual activity in what we believe to be the future frontline standards of care here, and we'll be moving very rapidly to find a way to initiate those registrational trials in addition to the second line 1 that we have discussed earlier.

Yaron?

Yaron from TD Cowen. Maybe the first question, Don, why go into venous malformation specifically and not go more broadly with that program? Secondly, for NRAS, are you going to target both on an off mutations? And then maybe thirdly, for 2608, when we see the updated data on efficacy in Q4, is it -- are you really only focusing on PFS? Or do you need to see differentiation on response rates still?

Thanks, Yaron, for your 3 questions. We will start maybe by taking the first one. I think that's just more of a clarification around vascular malformation as an umbrella term. So maybe, Don, you can start and then maybe Dr. Hammill, you can comment on like what does the term vascular malformations mean.

So I'll just start by saying we intend our initial development to focus on a broad range of malformations, including pros and other severe PI3K-driven vascular malformation. So it's not just venous malformations. I think as we get closer to a clinical start, we'll come back and be able to provide more information for what that trial looks like specifically and what the time lines would be reporting our data for that trial. I'll it over to you, Dr. Hammill, just to clarify that.

I agree. I think that some of that's the term. Unfortunately, we abbreviate vascular malformations VM and venous malformation VM. So it's a mess. But yes, I do think that this is venous malformations might not be the ones who need it the most, although there are certainly some severe venous malformations that could benefit.

The second question was NRAS on and off. Maybe, Jim, you could take that one.

Sure. First, just to get everyone on the same page, on means GTP bound so active and off GDP-bound inactive. So the mutations tend to lock protein in GTP on state. As we showed in the slides, molecule clearly binds in the on state as well as off. It's the on state that you want to inhibit and we are inhibiting the on state completely and against every NRAS mutation that's been tested. So [indiscernible] NRAS mutant inhibitor that's inhibiting the on state.

And then the final question around the data set on RLY-2608 and what kinds of data we would show. Maybe Pete, you can take that one.

Yes. So the reason why you hear us focus on progression-free survival in that update is obviously because that's the regulatory endpoint. Response rate has not been a primary endpoint in HR-positive HER2-negative breast cancer for over 2 decades now. And in fact, across many different data sets in this space, especially in the pathway inhibitors, you see no correlation, sometimes inverse correlations between response rate, progression-free survival. That said, I think when we disclose our data set at the back half of the year, as we kind of hinted to today with some of the monotherapy PRs that we're seeing, we're very confident that not only the progression-free survival data that we will show but also response rate in tumor control will match up very nicely with what we all hope to see. And we'll show all of it.

Brad?

Brad Canino from Stifel. So Dr. Hammill really nicely outlined the broad spectrum of the vascular malformations. So 2-parter for me on that aspect. I'm wondering, first, beyond just the alpelisib label, how much of the opportunity across all the subsets has been derisked by off-label clinical use and literature reports of such? And then second, how heterogeneous or homogeneous are the regulatory paths and end points to obtain potential registration across this broad spectrum of disease?

So maybe Dr. Hammill, you take the first part of that around the use of off-label therapies.

So certainly, we've used a lot of these, a lot of different vascular malformations have been treated with off-label alpelisib. And obviously, it's showing up as case reports, but there's a lot more experience out there that's even -- than is even reported. So yes, we have seen evidence of improvement in venous malformation, sometimes regardless of the underlying mutation, so it's been seen for the TIE2 TEK mutation as well as for PIK3CA. And it's been used in a number of other things. Again, all very off-label and some very rare diseases like GLA, [indiscernible] and other things. We do have to be careful because you all know from cancer that sometimes inhibiting 1 pathway upregulates the other. And so we do have some NRAS-mutated diseases that may respond the opposite of what you'd like to PI3K inhibition.

If only we had a selective NRAS inhibitor. And the second of the questions, maybe, Don, you could take, which is around the regulatory pathways.

Yes. So we've only seen 1 FDA approval in this space. And it was based off of retrospective chart review of a compassionate use program for alpelisib in about 37 patients. So there's not a well-defined regulatory landscape in these diseases. And it speaks to the fact that this is a collection of diseases. I think the molecular understanding of these diseases, as Dr. Hammill pointed out, is really quite recent. But I think this is a place where given the severity of the nature of the diseases gives an opportunity to work both with patients and with regulators to define a rational path forward.

Excellent, Jason?

Jason Gerberry from Bank of America. On Fabry, just curious what you hope to test in early clinical testing in Phase I beyond safety. Can you get to an answer regarding that more patients are amenable to your therapy? And then we also look at both monotherapy and potentially combination with ERT to test the hypothesis that you might be able to get added benefit that way.

Don?

Yes. So I think we anticipate that the initial clinical evaluation of this asset will likely be in healthy volunteers, but there, you can look for upregulation of the wild type. So at least gives you the pharmacodynamic end point, along with your safety and tolerability in PK that you're hitting the target. But then as you point out, I think you can move then into mutated patients. And you can start asking the question about whether there's activity of our molecule in the so-called nonamenable patients, primarily that would be looking at biomarker endpoints in early phase clinical trials. Lysis of Gb3, which we showed you here in the slide is a clinically translatable endpoint into a plasma measurement of the substrate. So I think we'll be able fairly early on to start getting a read for breadth of activation across mutations as well as how much activation we're able to see with our mechanism. And then obviously, over time, in the development, we could look at the ability to combine with ERT as well. But that likely would come later.

Thank you. Peter?

Peter Lawson from Barclays. Just on the CDK4 collaboration, I'm kind of -- how should we think about how you select and move forward with Ribo versus CDK4 and are you seeing any kind of DDI issues with Ribo or any other CDK4/6 that we should be thinking about?

Yes. I think what we're trying to show there is certainly the current standard of care in frontline patients is CDK4/6 plus aromatase inhibitor. Ribociclib has been gaining the dominant market share. So we want to establish a clear understanding of the combinability with the current standard of care. Ribociclib is 1 of the tougher ones to combine with. None of the other nonselective inhibitors have been able to establish a dose with full dose ribociclib. We're happy with the progression that we're seeing in our early clinical exploration of that triplet and are confident we will establish a dose of 2608 with full dose ribociclib. We'll give you a clear update on that towards the end of the year. In terms of the combinations with CDK4, again very excited to be skating where the puck is going here. We firmly believe that the future regimen in earlier lines of HR-positive HER2-negative breast cancer are going to include an all-oral regimen, inclusive of more selective CDKs, CDK4 and Pfizer's atirmociclib is the first-in-class there quite clearly. And we'll look to establish additional combination regimens over time in parallel with our initial Phase III registrational study that will start next year.

[indiscernible] with Guggenheim. Congrats on the progress. Two quick questions from us on PI3K-alpha VM. I'm wondering what level of PI3K inhibition do you need to achieve to get a clinical -- meaningful clinical benefit compared to the oncology indications? And clinically, do you plan to go with sort of like lower hanging fruit PROS, for example? Or do you want to pursue other indications that alpelisib cannot pursue. And for Fabry disease, just curious about the 1,000 mutations you pointed out like what's the distribution in the clinical setting? Like are they more evenly distributed? Or are there any other certain dominant ones?

All right. So maybe we'll take those back to front. So the Fabry one maybe, Jim, you could take around just the spread of distribution and mutation.

It's a very wide distribution. It's a disease where you really don't see hotspots. A couple of them, you can see in a small handful of patients, but they're widely distributed throughout the gene. So it's not like a cancer oncogene story. They're kind of spread out, they're spread out everywhere.

And then the second of those questions was around the unmet need and our ability to go after it. So maybe Dr. Hammill, you could just speak about the unmet need across those various subtypes of vascular malformations and then Don, you could speak as to how we want to address them.

So I think there are -- obviously, patients with PROS are the most visible changes that you can see when you put up pictures of CLOVES or [indiscernible] it's pretty impressive. But they're not the only ones that need medication. And so I think, thinking about it more broadly, we often think about them as severe because it's affecting quality of life because it's interfering with function, they think that they could qualify across diagnoses if they had that mutation.

Yes. And again, I think as we design the first-in-human clinical trial for VMs, we'll be looking to not just focus on PROS, but on severe other vascular malformations to the PI3K mutations. With regard to the target coverage question, that's a question where there's not at this point a data-driven answer. In the case of alpelisib, the doses that were used were not rationally selected, the pediatric dose was actually very low exposure. And this is in lesions that only have a single driver, which is PI3K. So again, I think with the profile that we've generated so far for 2608 in cancer patients, we'll look to maximize the target engagement that we can get. And I think we're confident with the experience that we've had so far in humans that we'll be able to achieve meaningful inhibition of PI3K for these patients.

Can I add one tiny point?

Yes, absolutely.

I mean the other thing that I would say is that in a non-mutant selective drug in children, we don't want to turn off all of the signaling, they need it to grow this as a normal pathway. And so we're certainly limited with our current less selective drugs.

And I think buried down deep in there was how did we find the allosteric pocket in Fabry. So I'll ask Pascal and Pat maybe to comment on that.

I mean, old feedback. We started really trying to understand the protein confirmations that are involved in alpha-galactosidase, right? And then we're able to uncover this [indiscernible] would form in a way that was almost impossible to visualize just using normal [indiscernible] technique. So it's really when we think about the ensemble of confirmation that this proteins form in a cell that we can actually start on covering these cryptic pockets. And then with this, we integrated computation that we'll have Pat comment on.

Yes. We used our molecular dynamics platform to then explore the confirmations of the protein. We were then able to run a fragment screen, identify fragments that bound and then use our artificial intelligence machine learning platform to search through databases of billions of molecules, use that to find starting points and then continue to use our physics and machine learning engines to do the optimization.

Excellent. All right. I think Sean.

Sean McCutcheon, Raymond James. Just a quick one. You guys mentioned combining with ERT in Fabry. Just what commentary do you have on the necessity of combining with ERT and maybe the feasibility on a toxicity basis as well as financial toxicity basis?

Maybe Pete, and then Don, you can take the financial question. Don, maybe you can just take the feasibility question.

Sure. So the financial question is a good one. I think a bit premature not having even started a clinical study to comment about pricing and combinability. I would note that there is a number of biosimilars in development for enzyme replacement therapy. I do think there's going to be opportunity to optimize the cost for these combination regimens that best address a clear unmet need in patients, which Jim talked a little bit about in his presentation, but can talk you through a little bit more, too.

Yes. And I think in terms of the need for combination with ERT across the majority of Fabry patients is probably not going to be the need. There is going to be a subset of patients across those 1,000 mutations that have essentially complete elimination of the gene with essentially 0 expressions, nothing to really chaperone. Those are patients that will continue to be dependent upon ERT, but where the addition of a small molecule noninhibitors -- non-inhibitory small molecule chaperone could give you, as we've shown in preclinical models, significantly more enzyme activation and more clearance of substrate. But we expect that to be a minority of cases.

This is Manoj [indiscernible] from Jefferies. Just 1 quick question. So Lilly recently mentioned about like optimizing the PK and drug-like qualities of their next-gen PIK3-alpha. So how confident are you in your PK/PD of the drug?.

Sorry, could you just repeat who did that.

Lilly.

Lilly. I mean I think we can be confident in the data that we've shown to date that we feel very confident. I think what you're hearing from us is we're going to see a robust data set at the end of the year and that we feel confident to engage in a discussion with the regulators to start a pivotal trial in 2025. And I think what you're hearing from us secondly is we're expanding our mutant selective PI3K-alpha franchise into vascular malformations, which you've heard all the details around. And I think the third thing you're hearing from us is that in monotherapy dosing, we're seeing a range of partial responses outside of breast cancer. So I think we can be as clear as that around how confident we are in our PK/PD relationships. Billy?

Billy Smith, JPMorgan. You mentioned you'd be using 2608 for POC in VM and then possibly move on to another compound. What would you look for in that next-gen compound that kind of suits it better to VM, for example?

Don, do you want to take that one?

I think as Dr. Hammill described to us, a number of the patients with VMs who will need treatment will be pediatric patients. So down to 2 years or potentially even younger. So 1 of the questions will be a pure development question, just identifying a molecule that would be compatible with pediatric formulation and pediatric dosing.

Yes. And just to add to that, to be clear, it's not a matter of a molecule that's better than 2608. It's just one that from a development standpoint is, most importantly, a distinct molecule from a branding pricing perspective would be helpful to have 2 different molecules, 1 for oncology, 1 for VM and then 1 where we can do some purpose-built late-stage development and formulation for the use case here.

Excellent. So again, thank you for being patient with us after a very busy week. Hopefully, you take away exactly what we laid out, which is we have a very productive platform. lots talked about around computational approaches to drug discovery. Hopefully, you can see it's coming to fruition. We have a broad pipeline now in the near term across genetic disease, solid tumors, breast cancer. You've seen for the first time the full unveiling of an auto mutant selective PI3K-alpha could do. We believe it's 1 of the biggest opportunities in precision medicine. These are all large markets. We'll be able to derisk them with clinical data relatively fast. And we have a team now and a capital that is able to try and make sure that we can bring all of this to fruition. So with that, I know there's a lot to take in over the coming weeks. We'll be very happy to engage with you as you get more educated through this. I'd like to thank Dr. Hammill for educating all of us this morning. So thank you.