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

Good day, and thank you for standing by. Welcome to the Relay Therapeutics Update Conference Call. [Operator Instructions] I would now like to hand the conference over to your speaker today, Mr. Peter Rahmer, SVP, Corporations Affairs. Please go ahead.

Thank you, operator, and good morning -- good afternoon, everyone. You can access the press release from this morning, the slides we are reviewing and a replay of this call by going to the Investor Relations section of our website. Before we get started, I'd like to remind you that during this conference call, Relay Therapeutics will make certain statements that are considered forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995, including our strategy, business plans and objectives. The expected therapeutic and clinical benefits of our product candidates, the potential of our platform and our product candidates and progress and timing and execution of our clinical trials. Such forward-looking statements are not guarantees of future performance, and therefore, you should not undue reliance upon them. These statements are subject to numerous risks and uncertainties that could cause natural -- actual results to differ materially from what we expect. I refer you to our SEC filings for a discussion of risk factors that could cause our actual results to differ materially from those discussed today. Today on the call, Sanjiv Patel, our President and CEO, will walk through a summary of the disclosures. Don Bergstrom, Head of R&D, will also take you through additional details. With that, I'll turn the call over to Sanjiv.

Thank you, Pete, for that introduction, and thank you to those of you on the line. Today, we want to share with you key updates on our progress to deliver new medicines to patients. This is the first clinical update since we were founded in 2016. And as you can imagine, this is a very exciting day for our company. Reflecting back over the last 5 years, we put a very high premium on setting tangible stretched goals for ourselves and executing against them. We believe we've achieved every goal we've laid out for ourselves at the time of our IPO in July of 2020. For RLY-4008, we've achieved our stated goals. As we are showing today, the interim clinical data suggests RLY-4008 is the first known highly selective inhibitor of FGFR2, not limited by any off-target toxicities, and it shows the potential to reduce tumors across a range of patients and alteration types. For RLY-2608, our PI3K-alpha mutant selective inhibitor, we've achieved our stated goals as we are rapidly advancing to initiate a clinical study in the first half of 2022 for the first of a franchise of therapies. The foundation in first of these programs has pan-mutant activity, and we believe has the potential to be among the largest precision oncology medicines opportunities ever. For RLY-1971, our SHP2 inhibitor, we've achieved our stated goals as we have determined a strategic path forward for the clinical and commercialization of this program, with the world's leading oncology development and commercialization organizations, Roche Genentech. With regards to our platform and capabilities, we've achieved our stated goals as we've continued to evolve our platform, build the team across all the functions required to make new medicines, seamlessly progress multiple clinical trials in the face of a global pandemic, underscoring our world-class development group, while all the time expanding the depth and breadth of our scale and research. To summarize, we have met or exceeded every single one of the goals we have set for ourselves. During a plenary presentation today given by Dr. Lipika Goyal from MGH at the Triple Meeting, interim clinical data from our first-in-human study for RLY-4008 was shared. To summarize the interim data, although early and from a heterogeneous set of patients treated across multiple therapeutic doses, it shows that we believe RLY-4008 has the potential to be highly selective. This shows our platform has the potential to achieve what we set out to achieve, i.e., creating a highly selective FGFR2 inhibitor. We did this using our Dynamo platform, which is a combination of leading-edge computational and experimental techniques. We created a meaningfully selective inhibitor that has been a challenge for conventional approaches to drug discovery. As of the data cutoff, our interim data shows that RLY significantly dials out primary off-target effects at all dose levels we tested, including FGFR1-related toxicities of hyperphosphatemia and FGFR4-related toxicities of diarrhea. At last, we believe our interim data shows that RLY-4008 is potentially not limited by these off-target side effects and can inhibit FGFR2 at levels that have not been seen with the pan-FGFR agents that have received accelerated approval. This is a clear validation of our Dynamo platform. We excited -- we are excited about what this means for our other programs in our deep portfolio of precision medicines. We believe RLY-4008 shows promising initial safety and tolerability data and expect to select a recommended Phase II dose optimized for safety without compromising efficacy, and we will commence opening expansion cohorts prior to the end of this year. Finally, from this interim data, we believe RLY-4008 showing promising early signs of tumor regressions. Note again that this stage is early into a first-in-human trial, where we have multiple doses deployed across heterogeneous patient populations. But with all that said, RLY-4008 has shown that inhibiting FGFR2 above previously known achievable levels can drive tumor regressions regardless of line of treatment, alteration or tumor type. 3 out of 6 of the pan-FGFR treatment-naive patients showed confirmed partial responses, all or on the once-daily dosing regimen at the time of response. Notably, a 35-year-old with ICC had a confirmed response with an 83% tumor reduction by the RECIST and went on to surgical resection with curative intent. This patient was initiated on a once-daily dose of 70 milligrams a day and remained on this dose for the duration of his treatment. This is a wonderful outcome so early in our trial and speaks to the potential of this medicine. The other 2 responders continue on a once-daily dosing regimen. RLY-4008 shows early activity against on-target resistance mutations, and we've also been able to see early activity of RLY-4008 against de novo mutations and amplifications in tumor types within and outside of cholangiocarcinoma. We are encouraged by this data, and we look forward to selecting a recommended Phase II dose and sharing more in 2022. The next target on which we deployed our Dynamo platform is PI3K-alpha. Our goal is to create a meaningfully selective molecule because this is one of the last significant precision medicine targets deemed intractable. With a pan-mutant PI3K-alpha inhibitor, the addressable patient population has the potential to be several fold greater than that of KRAS G12C, and this is over 100,000 patients annually in the U.S. alone. And we believe that this is among the largest ever for precision oncology medicine. The challenge is the industry has struggled to make such a medicine. Non-mutant selective inhibitors have shown efficacy but are limited by on-target toxicity. As you can see from this map of mutations on the protein, they are disparately located, requiring a pan-mutant approach to prevent on-target resistance mechanisms. We believe only a knowledge of allostery and the dynamic nature of this protein will unlock this challenge. As traditional approaches from our industry have tried and failed this task over several decades, we believe Relay Therapeutics is uniquely positioned to solve this challenge. In our recent abstract shown at the Triple Meeting this week, we highlight some interim data for the first known pan-mutant selective PI3K-alpha inhibitor created by Relay Therapeutics' Dynamo platform. We are very excited about this program entering the clinic in the first half of 2022. Key elements to note about this program are it has a novel mechanism of action. There is a mutant-selective binding to an allosteric binding site discovered by Relay Therapeutics' Dynamo platform. It's pan-mutant selective while sparing wild type. We believe the preclinical data show this is more tolerable than existing therapies, and it is fast approaching the clinic. As I said, we expect the first-in-human study to start in the first half of 2022. RLY-2608 represents the start of Relay Therapeutics emerging PI3K-alpha franchise. Today's updates allow us to take another step forward in validating our approach and platform. We have now shown multiple times that we can select a target, use our platform to identify a motion-based hypothesis to drug previously intractable precision medicine targets, use our team to create a molecule that generates promising preclinical data, select a development candidate, and today, we have shown for the first time that we can effectively progress a precision medicine clinical trial and generate promising early clinical data to support this. This gives us great confidence that we can continue to do this for programs in our pipeline. With each cycle and each additional program, we're able to improve our approach and strengthen our capabilities. We are very well set up to execute against our goals now and in the future. Over the last 5 years, we've grown our team to well over 200. We have a core research team with leadership that has 10-year now, well over 5 years as a group, and this experience as a group is clearly paying off. As we have made it clear, we believe there is a steep experience curve to what we do. In addition, we have shown today our development group is excellent in its ability to design and execute precision medicine trials. It is clear, the more we do, the better we get in all aspects of our work. We're excited about building on the momentum of learning and continue to bring forward our portfolio on all fronts. We believe we have a deep precision medicine pipeline, and we are motivated to bring multiple meaningful medicines to patients globally across both. Firstly, our wholly-owned innovative portfolio, which we believe will be Relay's core focus; and secondly, our challenger portfolio, which we will use to build our platform and derive value through development and commercialization partnerships. We've started work on the first of these partnerships with our partner, EQRx. I will now hand over to Don Bergstrom, Executive Vice President and Head of R&D, to talk in detail about today's updates. Thank you.

Thank you, Sanjiv. I'll start today with an overview of our wholly-owned innovator programs. As a reminder, the interim clinical data we will review today shows that RLY-4008 is highly selective and validates our approach for drug discovery, shows robust inhibition of FGFR2 with promising initial tolerability and has early signs of activity across a range of alterations in tumor types. Fibroblast growth factor receptor 2, or FGFR2, has all of the hallmarks of a cancer driver oncogene, which makes it an attractive precision oncology target. FGFR2 gene alterations in tumors include gene fusions, local gene amplifications and gene mutations. And they're seen across a range of solid tumors, including breast cancer, endometrial cancer and cholangiocarcinoma. We estimate that a highly selective FGFR2 inhibitor could address between 8,000 and 20,000 advanced cancer patients with FGFR2 alterations annually in the United States. And despite recent progress with FDA-accelerated approvals of pan-FGFR inhibitors and FGFR2-altered cholangiocarcinoma and urothelial cancer, there remains a broad tumor-agnostic medical need for patients with FGFR2-altered tumors. So FGFR2 has been validated as a therapeutic target in FGFR2-altered tumors, but we believe the current generation of nonselective inhibitors are limited by off-target toxicity. These inhibitors -- these non-selective FGFR inhibitors are both not only in FGFR2, but also the highly similar family members, FGFR1 and FGFR3 and FGFR4 to a varying extent. So while inhibitors with accelerated approval for FGFR2-altered cholangiocarcinoma achieve 23% to 36% response rates in FGFR inhibitor-naive patients, we believe that the dose of these agents is actually limited by toxicity related to off-target inhibition, hyperphosphatemia related to FGFR1, diarrhea related to FGFR4. This ultimately limits how completely these agents can shut down signaling through FGFR2. Our therapeutic hypothesis with that a highly selective inhibitor of FGFR2 will be optimized for innovation of FGFR2, leading to higher response rates and better durability of response in patients with FGFR2-altered tumors. This is a hypothesis that has been proven at many times before us in precision oncology for other targets, including ALK, EGFR, HER2, RET, ROS, TRK and others. Designing a selective FGFR2 inhibitor is a major challenge for conventional drug discovery approaches. Here, you can see a representation of the static 3D structures of FGFR1 and FGFR2. And as you can appreciate, they are essentially identical. In contrast, our drug discovery teams use dynamic models of our protein targets to generate novel modulation hypotheses. In the case of FGFR2, as is shown in this movie reel to represent motion, we discovered that despite the similarities in static structure, there's a loop on the surface of both proteins that moves differently in FGFR1 compared to FGFR2. Our drug discovery scientists were able to use this insight into differential protein motion to design a compound able to irreversibly interact with the slower moving loop on FGFR2, potently shutting down the protein, while much less efficiently interacting with the analogous loop on FGFR1. We believe RLY-4008 is the first known highly selective and irreversible FGFR2 inhibitor designed to avoid engagement of other FGFR family members, which, as I mentioned, leads to clinical toxicities. RLY-4008 display selectivity not only within the FGFR family, but across the kinome generally, in contrast to the pan-FGFR inhibitors that are all equipotent against FGFR1, 2 and 3 and as well as many other off-target kinases, which further narrow their therapeutic window. Another unique feature of RLY-4008 is that it was designed to exhibit broad coverage against the spectrum of FGFR2-resistance mutations, including the gatekeeper position FD565 and the molecular brake position at N550. We have specifically designed RLY-4008 to potently target activity against these on-target resistance mutations. The grid here shows the potency of RLY-4008 versus these mutations. In a panel of tumor models with primary FGFR2 alterations and/or resistance mutations, RLY-4008 showed in vivo antitumor activity across tumor models, reflecting the spectrum of oncogenic FGFR2 alterations. These data support the broad tumor and alteration-agnostic therapeutic potential of RLY-4008. We report today on the interim data from the dose escalation portion of the first-in-human study of RLY-4008, which is enrolling patients with FGFR2-altered tumors as determined by local assessment, regardless of prior FGFR inhibitor treatment. Today's disclosure includes data on 49 patients enrolled in part 1 of the study between September 2020 and the data cutoff date of September 9, 2021. RLY-4008 was administered orally on once daily or Q day and twice daily or BID schedules. The primary objectives are to define the maximum tolerated dose, the recommended Phase II dose and safety and tolerability of RLY-4008. Secondary objectives are to define the pharmacokinetic profile, impact on pharmacodynamic biomarkers and preliminary antitumor activity. Given the resistance to FGFR inhibitors is often mediated by secondary mutations in the FGFR2 kinase domain, we check for these mutations in this trial in circulating tumor DNA at baseline and longitudinally on study. Once the recommended Phase II dose is defined in the study, the dose expansion will begin and enroll across the 5 groups being here on this slide. 40 of 49 patients enrolled had cholangiocarcinoma, as you can see on this slide. 80% of these patients had an FGFR2 fusion or rearrangements. This trial allows prior exposure to an FGFR inhibitor, and 76% of the patients with FGFR2 fusion positive cholangiocarcinoma had received at least 1 prior FGFR inhibitor therapy. You will also note that these heavily pretreated patients had large tumor burden, with the mean sum of longest diameters of target lesions nearly 10 centimeters across the study population. Pharmacokinetic studies for the once-daily and twice-daily schedules of RLY-4008 show that both schedules achieved predictable dose-dependent increases in exposure. They also show that the half-life of the drug is long at 15 to 30 hours, supporting a once-daily dosing schedule. Notably, PK/PD modeling based on the observed human PK and a receptor occupancy model validated in preclinical tumor efficacy models predicts greater than 85% median FGFR2 receptor occupancy across all dose levels tested in both schedules. All doses achieved exposures that were associated with antitumor activity in preclinical models of FGFR2-altered tumors. Now according -- acknowledging that our method of calculating receptor occupancy for RLY-4008 differs from the ex vivo blood phospo FGFR2 assay utilized to measure pharmacodynamics at pemigatinib, the target occupancy for all doses of RLY-4008 tested exceed that achieved by pemigatinib's label dose. To remind you, pemigatinib's label dose of 13.5 milligrams daily achieved 76% inhibition of FGFR2 at trough, assuming continuous dosing. However, the pemigatinib label dose calls for 2 weeks on, 1 week off dosing. And that's we believe is likely to be an overestimate true target inhibition over time. These data suggest that RLY-4008 can significantly inhibit FGFR2, with all tested doses achieving target coverage associated with activity in preclinical models and reported FGFR2 inhibition levels for pan-FGFR inhibitors. The first-in-human trial is a Bayesian dose escalation, which allows for flexibility to explore and enrich multiple dose levels in parallel. The trial began with a twice-daily dosing schedule. At the initial dose of 50 milligram twice daily, we were already achieving approximately 96% of the FGFR2 receptor continuously over the dosing interval. We believe this speaks to the broad therapeutic window of RLY-4008 preclinical species and the favorable drug-like properties of RLY-4008 in humans. In the early months of the trial, we focused on the twice-daily dosing schedule. And as we gathered more experience with the agent, we appreciated that the long half-life of RLY-4008 in humans was compatible with once-daily dosing. And the acute and chronic tolerability of the 50 and 100 milligram twice daily doses was challenging, as shown by the rate of DLTs and dose reductions at these doses. We initiated once-daily dosing at 50 milligrams and then 70 milligrams in early 2021. Once we gathered more PK information, we're able to model the once-daily dose range that we projected would cover the range of exposures associated with efficacy in preclinical models. The flexible design of the Bayesian protocol with enrichment allowed us to continue to enroll the 50 and 70 milligram daily cohorts, while in parallel, enrolling cohorts at 20, 30 and 40 milligram once daily. For protocol, we are continuing to enroll patients across all 5 once-daily dose levels. Based on acute and chronic tolerability in PK, the twice-daily schedule was deprioritized for further enrollment. The early data across the doses on the once-daily schedule show that only 1 DLT of 32 patients treated in the once-daily schedule has been identified as of the data cutoff date across all the doses. And this schedule appears to allow for maintenance of greater dose intensity with fewer dose reductions compared to the twice-daily dosing schedule. All of the once-daily doses remain in consideration for the recommended Phase II dose, which will be identified based on safety and tolerability, PK and early efficacy signals. We feel the comprehensive database we will have on safety, PK and early efficacy across multiple biologically-active doses is promising and will enable us to determine the appropriate recommended Phase II dose. We have focused on hyperphosphatemia and diarrhea as hallmark toxicities of the pan-FGFR inhibitors associated with inhibition of FGFR1 and FGFR4. RLY-4008 shows low rates of hyperphosphatemia and diarrhea across all doses in both schedules. The observed hyperphosphatemia was transient and did not require medical management. Our PK and receptor occupancy data indicate we are exceeding exposure thresholds required for efficacy in preclinical models of FGFR2-driven tumors, and the reported target coverage for pan-FGFR inhibitors. These interim data suggest we've created a truly selective FGFR2 inhibitor. This speaks to the power of the Dynamo platform as this drug discovery challenge has alluded traditional discovery approaches for several decades now. We believe this high selectivity for FGFR2 now allows us to be the first to be able to optimize dosing schedule for inhibition of FGFR2 and not to be limited by toxicities due to off-target effects. This slide shows the stability of serum phosphate levels over time for all doses in both schedules tested, further indicating the selectivity of RLY-4008 for FGFR2 over FGFR1. Given our ability to inhibit FGFR2 at levels not previously seen in the clinic, we can now understand for the first time on-target toxicity for such an agent. The adverse event profiles of RLY-4008 on the once-daily and twice-daily schedules are shown here. Overall, most treatment-emergent adverse events were low grade. No grade 4 or 5 AEs were seen in either dose schedule. The AE profile was consistent with sparing FGFR1 and FGFR4. These data provide insight into which AEs that are thought to be class effects of FGFR inhibitors may specifically be related to highly selective FGFR2 targeting. The most common treatment-emergent adverse events were palmar-plantar erythrodysesthesia, stomatitis and dry mouth. Both schedules exhibited a manageable safety profile, but the AE profile favored continuing with once-daily doses. This slide summarizes the overall tolerability profile for the small molecule pan-FGFR inhibitors as established in single-arm Phase II data sets in patients with FGFR2 fusion cholangiocarcinoma. The AEs listed here are the most frequent AEs seen across the class. This slide shows the rate of AEs with once-daily dosing of RLY-4008 across the spectrum of the most frequent pan-FGFR inhibitor-associated AEs. While the data are early and still immature, the promising AE profile has been achieved at exposures exceeding our preclinical efficacy thresholds and at all once-daily doses, inhibiting FGFR2 more potently in humans compared to the pan-FGFR inhibitors. We will select an optimal dose for further characterization of safety, tolerability and efficacy in the part 2 expansion cohorts in the ongoing study. We look forward to learning more about the safety and tolerability profile of RLY-4008 over time as the data mature. In patients with fusion-positive cholangiocarcinoma who are FGFR inhibitor-naive, 3 out of 6 patients exhibited a confirmed PR, all on the once-daily dosing schedule, all at doses which are under consideration for the recommended Phase II dose. Despite our plot shows that 3 out of 6 patients remain on treatment, and a fourth had greater than 80% response and recently underwent surgical resection with curative intent. The pan-FGFR inhibitors available under accelerated approval have shown 23% to 36% response rates in this patient population at an optimized dosing schedule in their registrational Phase II trials. All of the patients ultimately ended up being treated on the once-daily schedule. You can see the 1 responder who initiated treatment at 100 milligrams twice a day did not achieve response until treatment on a once-daily dosing schedule. And this patient remains on treatment with a deepening response at the 30-milligram once-daily dose. These initial early data highlight the encouraging activity of RLY-4008 in FGFR inhibitor-naive fusion-positive cholangiocarcinoma patients. I will highlight the patient who underwent resection. He's a 35-year-old patient with multifocal left-sided liver lesions and an enlarged portal lymph node to a greater than 80% response after 6 cycles of RLY-4008 treatment. He remained on treatment at 70 milligrams once-daily throughout the duration of therapy. Given the marked tumor reduction, the patient underwent partial hepatectomy with curative intent. This is an exceptional landmark in the current management of patients with advanced cholangiocarcinoma, and this then yet highlights the potential of RLY-4008 in the FGFR inhibitor-naive setting. The second subpopulation is patients with fusion-positive cholangiocarcinoma who were previously treated with 1 or more pan-FGFR inhibitors. These are the same data as presented earlier at the Triple Meeting, but broken into patients with detectable acquired resistance mutations versus patients without a detectable acquired resistance mutation or for whom mutation status is unknown. As designed, RLY-4008 is active against acquired resistance mutations in FGFR2 that arrived in response to treatment with pan-FGFR inhibitors. In 10 patients with detectable acquired resistance mutations and circulating tumor DNA, the mutant clones were cleared to below the level of detection by cycle 2 day 1 for all mutant clones in 7 out of 10 patients. This included clearance of both N550 molecular brake and V565 gatekeeper mutations. And there was tumor regression in 9 out of 10 patients with detectable acquired resistance mutations at the time of enrollment. These data are consistent with RLY-4008's mechanism of action and preclinical activity against FGFR2 resistance mutations. This is not a population we are primarily focused on for further developments of RLY-4008. We will continue to focus on the FGFR inhibitor-naive patients, but consistent with the experience for other precision oncology targets, we feel the demonstrated activity of RLY-4008 against established on-target acquired resistance mutations suggest that RLY-4008 may have greater durability of response in naive patients through suppression of common mechanisms of acquired resistance. This patient's vignette is a 50-year-old -- 51-year-old woman with recurrent multifocal cholangiocarcinoma in her pelvis, who is treated with futibatinib for nearly 1 year and then was found to have an FGFR2 kinase domain mutation at baseline prior to starting on RLY-4008. The patient has achieved greater than 20% tumor shrinkage on 4008 and her FGFR2-E566V mutation was rendered undetectable at cycle 2. The patient is being treated with 50 milligrams once-daily and has had no dose interruption or dose holiday and treatment is still ongoing in cycle 7. We're also sharing a vignette today for a patient who achieved a response subsequent to the database block for the Triple Meeting presentation. This is a 65-year-old male who received prior infigratinib. Circulating tumor DNA at baseline detected polyclonal on-target acquired resistance, with N550K, N550D and V565I resistance clones. The patient commenced treatment with RLY-4008 at 30 milligram once-daily with a 72% reduction in target lesions at the end of cycle 2 scan with complete regression of a manubrium bone lesion associated with symptomatic improvement of chest pain. The partial response is pending confirmation and the patient is tolerating 30 milligram once-daily well and remains on treatment. Additional subpopulations that showed evidence of tumor regression on RLY-4008 were patients with FGFR2 oncogenic mutations or FGFR2 amplification. We notably saw a confirmed partial response in a breast cancer patient with an FGFR2 N550K mutation, a difficult-to-treat mutation in the molecular brake of the hinge region of FGFR2, in which most pan-FGFR inhibitors lose significant potency. Two additional breast cancer patients with activating mutations in FGFR2 remain on treatment and cycle 3 at the time of data cutoff. Additionally, a cholangiocarcinoma patient with a C3 [indiscernible] activating mutation treated at 40 milligram once-daily showed a 72% reduction in target lesions at the first scan and remains on treatment pending confirmation of response. A cholangiocarcinoma patient with low level amplification of FGFR2 and prior treatment with derazantinib started therapy at 100 milligrams twice daily. This patient remains on study with stable disease at cycle 12 and is currently being treated with RLY-4008 50 milligrams once daily. This vignette is in a 60-year-old woman who had 12 prior lines of systemic therapy for advanced breast cancer. The patient's tumor has an oncogenic FGFR2 N550K mutation. She received prior treatment with palbociclib, an emerging data in the literature suggests that FGFR2 alterations are potential resistance mechanisms to endocrine therapy and/or to CDK inhibitors. This patient has been treated with 70 milligram once-daily and remains on this dose in cycle 5. The patient had a 41% confirmed PR with response deepening between initial response and confirmatory scan. We believe this is the first response reported for an FGFR inhibitor in a breast cancer patient with an oncogenic FGFR2 mutation and suggests the potential for RLY-4008 in the patient population outside of cholangiocarcinoma. This is the swimmer's plot showing the time on treatment according to FGFR2 alteration. The data reflect the early stage of the trial with limited follow-up. However, 54% of patients remain on treatment with duration of treatment ranging from 4 to 45 weeks. The majority of patients who discontinued, did so due to progressive disease. A number of patients with progressive disease were treated initially on the twice-daily dosing schedule and did require prolonged dose interruptions, which significantly compromised the ability to maintain RLY-4008 dose intensity in that schedule. In summary, we believe this investigational medicine is the first known highly-selective FGFR2 inhibitor in the clinic, the target's driver alterations and the broad spectrum of FGFR inhibitor required resistance mutations and oncogenic mutations in FGFR2. We observed robust FGFR2 inhibition with greater than 85% receptor occupancy and minimal hyperphosphatemia or diarrhea across a range of doses. The once-daily schedule was favored based on both the PK and safety profile, and we now have initial data showing the AE profile of a highly selective small molecule inhibitor of FGFR2. We believe the initial overall tolerability profile is comparable to or favorable to that of the pan-FGFR inhibitors across all doses tested in the once-daily dosing schedule. The drug also showed encouraging early antitumor activity. An FGFR inhibitor-naive FGFR2 fusion-positive cholangiocarcinoma, 3 out of 6 patients with confirmed partial responses in the FGFR2-positive cholangiocarcinoma population resistant to prior FGFR inhibitor treatment, clearance of circulating tumor DNA associated with acquired resistance mutations and tumor shrinkage in the majority of patients, and early signs of activity in FGFR2 mutants and amplified tumors beyond cholangiocarcinoma. Overall, these results validate selective targeting of FGFR2 and suggest RLY-4008 has potential to overcome the most common mechanisms of on-target FGFR inhibitor resistance. We will select a recommended Phase II dose and open expansion cohorts before the end of 2021 and look forward to updating you on the next set of data for this agent in 2022. As we mentioned earlier, we also shared at the Triple Meeting, an abstract from our first pan-mutant PI3K-alpha inhibitor, which I'll walk you through. Key elements to highlight about this program: it's a novel mechanism; it's pan-mutant selective; in preclinical models, it's more tolerable than existing therapies; and it's approaching the clinic. RLY-2608 represents the start of Relay Therapeutics' emerging PI3K-alpha franchise. We can see from Novartis alpelisib, marketed as PIQRAY, that it is effective when combined with fulvestrant in ER-positive PI3K-alpha mutant breast cancer. But its tolerability is challenging given the adverse events of hyperglycemia, which limits the dose of alpelisib that can be administered acutely and leads to dose reductions and discontinuations with chronic dosing. Ultimately, the inability of agents like alpelisib to maintain continuous high level inhibition of PI3K-alpha is efficacy-limiting. The hyperglycemia AE is due to inhibition of wild-type PI3K-alpha in normal tissue. And there are also AEs related to inhibiting other isoforms of PI3K as well. The objective of this field has become to create mutant-selective PI3K-alpha inhibitors, meaningfully selective for mutant PI3K-alpha relative to the other PI3K family members and the kinome generally. We've learned from the experience with mutation-specific inhibitors of oncogenes like the KRAS G12C inhibitors that just targeting a single mutation in the context of the RAS-driven tumor leads to relatively short duration of response and multiple on-target mechanisms of acquired resistance. Analogously, we know that PI3K-alpha mutations are also clear oncogenic drivers, a cluster in a number of oncogenic hotspots, the most common of which are H1047, E542 and E545. Based on the experience of KRAS G12C mutant-specific inhibitors, we would anticipate that mutant-specific inhibitors of PI3K-alpha, for example, inhibiting only the H1047R mutation, would likewise be subject to the emergence of multiple on-target mechanisms of acquired resistance. We used our unparalleled knowledge of PI3K-alpha to design the first known pan-mutant selective PI3K-alpha. All of the current drugs targeting PI3K-alpha, target the active or orthosteric site, whereas mutations are distal to it, indicating the need for an allosteric inhibitor to achieve mutant selective inhibition of PI3K-alpha. The Dynamo platform gave us novel understanding of the allosteric networks in PI3K-alpha that allowed us to design what we believe is the first pan-mutant-selective PI3K-alpha inhibitor. We are building a franchise of PI3K-alpha mutant selective inhibitors where the foundation is exemplified by RLY-4008, which is potent and mutant-selective for all of the PI3K-alpha mutant hotspots. We call this a pan-mutant selective inhibitor. We will then augment this foundation with mutant-specific inhibitors, which are only active against a specific mutation, such as H1047R, to allow us to potentially double-drug specific mutations to maximize mutant PI3K inhibition. We believe we know more about allosteric targeting of PI3K-alpha than anybody else in the industry, and we'll leverage our insights to build a portfolio of mutant selective and mutant-specific PI3K-alpha inhibitors that will allow us to inhibit all PI3K-alpha hotspot mutations more completely and more durably than any mutant-specific approach can achieve on its own. We started by solving the first full-length structure of PI3K-alpha in complex with its regulatory subunit, both for wild-type and mutated forms of the protein. We then used these data to perform long-time scale [indiscernible] molecular dynamic simulation of wild-type and mutant PI3K-alpha to identify potentially druggable differences in the mutant and wild-type proteins. We find a novel druggable allosteric pocket, the mutant protein that we thought could provide us an opportunity, to develop a pan-mutant selective allosteric inhibitor of PI3K-alpha. We believe these insights are unparalleled in the field and culminated in the discovery of RLY-2608. We investigated the activity of RLY-2608 in biochemical enzymatic assays. RLY-2608 has potent activity against the 3 most common PI3K-alpha mutations and is 8- to 12-fold selective compared to wild-type PI3K-alpha in these assays. This is a profile that we have -- has not been shared by any other PI3K-alpha inhibitors. In addition to wild-type selectivity, RLY-2608 is over 500-fold selective against all of the other PI3K isoforms. RLY-2608 is also meaningfully selective against the rest of the kinome. We believe this is the profile of a true precision oncology therapeutic candidate. The selectivity that we observed in biochemical assays translates into signaling assays in cells. We generated cell lines that express the mutant protein in the absence of wild-type to allow us to tease apart mutant from wild-type signaling. While orthosteric inhibitors, as expected, inhibit wild-type and mutants with equivalent potency, RLY-2608 is significantly more potent at blocking downstream signaling in the cell lines that express mutant PI3K-alpha. When cancer cell lines harboring endogenous hotspot mutations and the kinase or helical domains are incubated with RLY-2608, we observed potent inhibition of downstream signaling and potent anti-proliferate effects on PI3K-alpha mutant cell lines. In contrast, RLY-2608 is markedly less potent at inhibiting proliferation in PI3K-alpha wild-type cells. Maximum efficacy is observed across different hotspot mutant models, efficacy comparable to or greater than orthosteric benchmarks dosed at doses above the exposures achievable in the clinic for these agents due to on-target toxicity. In contrast to the orthosteric inhibitors, which significantly induce serum insulin levels due to inhibition of wild-type PI3K-alpha, RLY-2608 shows lower impact on serum insulin levels compared to the benchmark compounds. We believe that this is in vivo proof-of-concept that RLY-2608 is selective for inhibition of mutant PI3K-alpha. In higher species at exposures that are matched for the efficacy exposures in the most models, there is no glucose increase nor do we observe any histopathological or ophthalmic findings associated with hyperglycemia in the reported preclinical studies for the benchmark PI3K inhibitors. RLY-2608 has properties consistent with moving into clinical development. The projected human bioavailability is approximately 60% and the projected human half-life is approximately 16 hours, compatible with once-daily or twice-daily dosing in the clinic. We anticipate initiating clinical studies for RLY-2608 in the first half of 2022. I'll now pass it over to Pete before opening it up to questions.

Thank you, Don. Beyond this, our partner programs, what we're calling Challengers, also represent a significant opportunity. For RLY-1971, we have determined a strategic path forward for clinical development and commercialization with one of world's leading oncology development and commercialization organizations, Roche Genentech. In August 2021, we announced a worldwide strategic collaboration with EQRx, through which we intend to discover, develop and commercialize a novel medicine against a validated oncology target, which will help us bolster our AI/ML capabilities. Over the coming year, we have additional meaningful catalysts, including RLY-4008, where we will select a randomized Phase II dose and move to opening expansion cohorts by year-end. We expect to give an update -- data update next year in 2022 on this program. On RLY-2608, the clinical start is expected in the first half of 2022, and we will make an additional preclinical data presentation at the San Antonio Breast Cancer Symposium in December of this year. Our next precision medicine target will be disclosed in the first half of 2022. For RLY-1971, as I mentioned, has begun its combination trial with GDC-6036, which is being run by Roche Genentech. Over the medium term, we will also build value through advancing our 5 additional innovative programs, pursuing challenger targets through partnerships, continuing to evolve our Dynamo platform, continuing to expand our pipeline scope and scale. We have $671 million in cash on our balance sheet as of the end of Q2 2021. And in the most aggressive success scenarios, this will take us into 2024. We look forward to executing on our plans. I hope that you've seen the execution is something we've been able to do to date and would like to thank you for your support of our efforts over the years. Thank you for joining us today and hearing our progress to change the way precision medicines are created. With that, I'll turn the call back to the operator to open it up for Q&A.

[Operator Instructions] Your first question comes from the line of Yaron Werber from Cowen.

Congrats on the data. I have a couple of questions. Number one, the profile of 4008 looks very selective on-target for FGFR2 and obviously sparing 1 and 4. In the QD dosing, as you go from 20 to 70 mg, can you give us a little bit of a sense of how does some of the tolerability relate to dose? Do you see more AEs as you go higher because you're still not really seeing DLTs? Or do you not see necessarily that trend? And then secondly, in terms of efficacy, how quickly do you see onset of activity instead of RECIST criteria responses or does it really deepen with time?

Yes. So thanks for the question, Yaron. I'll start with the same caveat for both answers, which is it's still early, and we're still collecting data. I think on the once-daily dosing schedules, the rate of AEs has still been relatively low. So it's been challenging at this point to necessarily assign a dose relationship based on the current data. Certainly, we have seen a low rate of acute toxicity as measured by DLT as we go through that dose level. And the dose reduction with chronic tolerability is still emerging. We anticipate as we continue to enroll out the 20 to 70-milligram once-daily cohorts and gather more acute and chronic tolerability data, we'll be able to provide a more complete picture of the dose relatedness to the AE profile as the data mature. With regard to the timing of the onset of response, again, I think the data are early, and I'm going to be extrapolating from a small number of responses here. But we certainly have seen responses that have occurred early in treatment. I think 4 of the responses that I highlighted today were all the responses that were seen at the first scan. But we've also seen patients where the response has come markedly later, where we haven't seen response until potentially the third scan. And what we've seen across the patients with responses, in many cases, is further deepening of response with continued treatment. So I think, obviously, as we continue to collect more data and as the data mature, we'll be able to share with you a more precise estimate of the onset of response. But we certainly have seen both fast responses and responses that have seen more time to build to.

Your next question comes from the line of Salveen Richter from Goldman Sachs.

[indiscernible] and yet, what's the mechanism for disease progression after the drug was administered here 4008 was for patients?

Could you repeat, Salveen?

Sorry, Salveen. I think we missed the first part of the question. Sorry, could you repeat that, please?

Trying to understand what you think the mechanism for disease progression was here, whether you think it's because of acquired resistance or what's playing out?

Yes. So thank you for the question, Salveen. I think the first thing I'll point out is that amongst our responding patients with confirmed responses, they all remain in response and have not progressed. So we have not yet obviously been able to assess what the mechanism of potential progression might be in those patients. We are in this trial doing -- circulating tumor DNA profiling both the baseline and longitudinally on study. We've shown you data so far showing the early profiling and showing you our ability to clear mutant clones. But obviously, one of the questions that we'll be asking is what is the profile in ctDNA with continued therapy with RLY-4008. We anticipate that will be the subject of a later disclosure as our data mature.

Your next question comes from the line of Eric Joseph from JPMorgan.

I was just wondering if you could speak to the level of prior systemic treatment experience in the TKI-naive fusion-positive cholangiocarcinoma patients specifically? And how those compare to the sort of the prior treatment exposure in the Phase II studies with the pan-FGFR inhibitors, firstly. And then secondly, as it relates to tolerability among the patients who underwent those reductions with the QD schedule, I think, on Slide 23. Can you talk about the specific AEs that led to that positive dose modification, how and -- how quickly and whether they resolved and whether there were any dose interruptions in between?

Yes. Thank you for the question, Eric. So with regard to the first question, about the profile of the patients who are in our FGFR inhibitor-naive population, these are patients who, in our study, have all received prior chemotherapy for advanced disease. And we anticipate that it's a patient population that likely is reflective in terms of intensity of prior therapy to the data sets that have been reported for the pan-FGFR inhibitors. With regard to your second question about the AEs that are associated with dose modification, what we've seen so far in terms of the on-target toxicity that we think is attributable to FGFR2 based on the experience so far has been stomatitis, DPE, retinal AEs. And those have been the AEs that have most commonly required dose modification, especially in the BID dosing schedule.

Okay. That's helpful. And just I also wanted to revisit the activity in the prior pan-FGFR-treated cholangio patient population. This is still a focal population in part 2 of the study, although it sounds like you're going to focus on those where there is a resistance mutation identified at baseline. I guess is the breakdown here that your -- I guess, the split that you're reporting here in a study reflective of the kind of the prevalence of baseline resistance mutations out there in the broader population? Just trying to get a sense of sort of the FGFR refractory population that are being scoped as you look to develop further in part 2?

Yes. Yes. So our population has been very reflective of what's been reported in literature. At the Triple Meeting last year, Dr. Goyal, who is our presenter this year, presented her work on characterizing mechanisms of acquired resistance to pan-FGFR inhibitors and presented a series of 46 patients treated with pan-FGFR inhibitors, of which 23 showed evidence of on-target resistance mutations, and many of those patients had multiple resistant clones of polyclonal resistance. And consistent with that, in our study, roughly half the patients have detectable on-target resistance mutations at baseline. And as we showed today, many of these patients have multiple on-target resistance clones with the 2 most common sites of acquired resistance being the molecular brake at N550 and the gatekeeper at V565. Again, very consistent with the data that Dr. Goyal presented at the Triple Meeting last year.

Your next question comes from the line of Michael Schmidt from Guggenheim Partners.

I just had one. Don, could you just comment on the DLTs that were observed in the study, were they expected to be on-target FGFR2-related DLTs? And then a question on the efficacy. I mean, it looks like the majority of the responses were seen at patients who received doses of 50, 70, perhaps higher doses per day. I was just wondering if that's just a function of still no follow-up at the lower QD dose cohorts because based on your PD data, you would -- I think you would expect to see responses at those doses as well. Or if there -- or whether there is indeed unexpected dose response with respect to efficacy in the study?

Thanks, Michael. I'll start with the second question first. So you are right that we have not, at this point, compiled as many scans on patients who are coming on at the lower doses. We have a lower number of patients who are eligible for response evaluation. We expect those data to mature. With that being said, we have now seen first scan partial responses in patients at 30 milligram, 40 milligram, 50 milligram and 70 milligram once daily. And the patient who was an FGFR inhibitor-naive patients who started treatment at 100 milligrams twice a day, actually did not go into response until they started on a 50-milligram once-daily dose. And they maintain -- they remain on therapy with deepening response at a 30 milligram once-daily dose. So we feel that any of these -- and we also shared the vignette today for the patient whose response came in following the database lock who also was able to achieve response at first scan at 30 milligram once daily. So we feel these data are consistent with all of these doses being biologically active. What we're doing right now with the continued enrollment across all 5 of these dose levels is precisely to answer the question you asked and to identify the dose that gives us the best target coverage, the best evidence of early efficacy. And the best acute tolerability profile and chronic tolerability profile that would be consistent with being able to maintain dose intensity over the course of chronic therapy. So I would say, to really answer the dose response question, stay tuned as we continue to collect data and it matures. With regard to your first question in terms of our understanding of what the AE profile might be going into our trial, this is the first trial that's been run with a selective small molecule inhibitor of FGFR2. We're able to triangulate in from the pan-FGFR inhibitors and then from bemarituzumab, the FGFR2 antibody on what AEs you might expect to be related to FGFR2 inhibition. But of course, we had no way of knowing from clinical experience, given that those are imperfect surrogates for small molecule FGFR inhibitor. But I think with the data that we've generated here, it definitely suggests that PPE, stomatitis, dry mouth and retinal AEs may all be on target FGFR2 AEs but we think with the once-daily dosing schedule, these are well-managed and will not impact our ability to identify a dose that will allow us to maintain chronic dose intensity.

Your next question comes from the line of Jason Gerberry from Bank of America.

Just a follow-up on the recommended Phase II dose between -- somewhere between 20 and 70 mg QD. So just sort of curious, when you look at the rates of dose reductions in DLT for either the 30 mg BID or 50 mg BID, thinking about like the total daily dosage, I'm wondering if that was cautionary at all as you guys think about going above like 50 mg QD or 70 mg QD ultimately? Or do you look at the BID and QD dosing a little bit differently with BID? You have 2 Cmax events. So maybe that's driving a differential sort of tox profile. And then if you can just comment on the baseline ctDNA levels for those 7 patients where you got those levels down to -- I think they were down to nil at the second scan. Just sort of curious if you can comment on that.

Yes. So I think with regard to the first question of BID versus QD, you're right, I mean the dose may be the same, but what we've seen is with the long half-life of 4008, the BID dosing, even at 20 or 30 BID, is giving us a high prolonged exposure at roughly 95% receptor occupancy or greater that we think is associated really with exposure-mediated AEs that seem to be ameliorated with going to the once-daily dosing schedule. And so far, the data would suggest to us with actually more data maturity than we've seen for the BID dosing that the 50 Q day and 70 Q day do appear to be better tolerated to date than we saw with the BID dosing schedule. So I think we're comfortable continuing to enroll in the 50 to 70 BID -- [ 50 to 70 ] Q day doses. And as I alluded to, we'll continue to enroll these 5 cohorts in parallel. We'll have up to 12 patients per cohort per protocol where we'll be able to look at, again, the acute and chronic tolerability PK and early efficacy to really be able to select a dose to move forward. In terms of the level of the ctDNA alterations, the assay we're using, I believe, has a lower level of detection of a variant allele frequency of 0.1%. The allele frequencies we were seeing varied from being in the 1% or less range to being much higher allele frequencies, but the common thread was that we were able to suppress them all to below 0.1%, which is the lower limit of detection on the assay we're using.

Your next question comes from the line of Silvan Tuerkcan from JMP Securities.

Congrats on the data. I dropped off there for a second, so this might have been asked. But could you comment on -- or do you know what the average dose per patient is that the patients received in the naive population and the overall population in the QD schedule? And then I have a follow-up.

Yes. So for the patients who have come on, on the Q day schedule and the treatment-naive patients, all of them remain on treatment in the Q day schedule on the dose they came in on that they were admitted to the trial when it started dosing on exception, the patient with the dose modification was the patient who started at 100 BID that has had to be dose reduced to 30 Q day.

All right. Great. And could you comment on the severity of the mouth sores compared to the pan-FGFR inhibitors? And also on the severity of retinal disorders, sorry, as I understand, look for them in the trial and could you please comment on what that -- how that impact will result?

Yes. Yes. So for us, stomatitis has predominantly been grade 1 or grade 2. Grade 3 has been rare. In terms of the frequency of what we're seeing, this is one of those AEs that is comparable to or favorable compared to what's been reported for the pan-FGFR inhibitor. So we're seeing in once-daily dosing, 31% stomatitis, again, mostly grade 1 or grade 2, whereas for the pan-FGFR inhibitors, that's gone up to as high as 56% in the case of infigratinib. In terms of the retinal AEs in the once-daily dosing, overall, we've seen 9% all grade 1 or grade 2, all reversible. The one caveat that I'll point out is the experience with the pan-FGFR inhibitors that first identified this AE as a potential class effect came in the trial of those agents. The trials of those agents did not require screening ophthalmologic exams where patients were on trial. Patients only went to a comprehensive ophthalmology exam if they were symptomatic, if they had visual changes. Our protocol does have scheduled ophthalmology, including OCT to look for retinal detachments, whether patients are symptomatic or not. So the rate we're seeing of 9% includes a small number of patients who actually had symptomatic retinal toxicity, but the majority of patients were grade 1 are asymptomatic retinal toxicity picked up on the scheduled exam. And as such, largely -- or potentially overestimates the rate compared to what's been reported earlier for the pan-FGFR inhibitors.

Thank you, operator, and thanks to all of you for joining us today. We're excited about the potential of our programs and appreciate your support of Relay Therapeutics in our mission to transform the drug discovery process with the goal of bringing life-changing therapies to patients. Thank you.

This concludes today conference call. Thank you for participating. You may now disconnect.