Relay Therapeutics

1.1 History & Prior Art

Scientists could not selectively inhibit FGFR2 for years. A peer-reviewed abstract conceded that “selective targeting of FGFR2 has not been achieved.” [T2: PMID 37270847] Pan-FGFR inhibitors proved that FGFR2 drives fusion-positive intrahepatic cholangiocarcinoma, but FGFR1-mediated hyperphosphatemia, FGFR4-mediated diarrhea, and FGFR2 resistance mutations restricted their benefit. [T2: PMID 37270847] PI3Kα faces the same constraint: orthosteric inhibitors such as alpelisib block wild-type and mutant enzyme equally, producing severe hyperglycemia and rash that limit clinical use. [T2: PMID 37623743] A selective mutant PI3Kα inhibitor could separate antitumor efficacy from metabolic toxicity. [T2: PMID 37623743] Relay designed its candidates to close both gaps.

1.2 Novelty Assessment

RLY-4008 is the first selective FGFR2 inhibitor active across FGFR2 alterations and resistance mutations. [T2: PMID 37270847] In vitro it achieves >250-fold selectivity over FGFR1 and >5,000-fold over FGFR4, sparing patients the class toxicities of pan-FGFR agents. [T2: PMID 37270847] In xenograft models harboring FGFR2 resistance mutations—the same mutations that drive progression on pan-FGFR inhibitors—RLY-4008 induced regression while sparing FGFR1 and FGFR4. [T2: PMID 37270847] Early clinical testing confirmed the therapeutic potential of selective FGFR2 targeting: RLY-4008 produced responses without clinically significant off-isoform FGFR toxicities. [T2: PMID 37270847]

Relay discovered RLY-4008 through structure-based design that exploited ATP-pocket residue differences between FGFR1 and FGFR2. [T2: PMID 38300868] This validated the platform’s ability to drug a kinase-family selectivity gap that prior scientists considered undruggable. [T2: PMID 38300868] The discovery was reported in PNAS. [T2: PMID 38300868; T2: C1221]

The authors of PMID 37916956 described RLY-2608 as the first allosteric mutant-selective PI3Kα inhibitor. Scorpion Therapeutics’ STX-478 has since entered clinical trials with an analogous allosteric mechanism, so the claim of being the first allosteric mutant-selective PI3Kα inhibitor is contested. Relay discovered RLY-2608 via a DNA-encoded library screen, cryo-EM-optimized constructs, differential enrichment, and an orthosteric-blocking compound. [T2: PMID 37916956] In PIK3CA-mutant xenografts, the molecule inhibited tumor growth with minimal impact on insulin, a surrogate for glucose homeostasis. [T2: PMID 37916956] Early clinical testing produced objective responses in two patients with advanced HR+ breast cancer, showing no wild-type PI3Kα-related toxicities. [T2: PMID 37916956] The authors titled their report: RLY-2608 “Decouples Antitumor Activity from Hyperinsulinemia.” [T2: PMID 37916956]

Migoprotafib (RLY-1971/GDC-1971) emerged when researchers applied structural biology and motion-based design to a medicinal-chemistry campaign. [T2: PMID 37774359] The discovery paper identified GDC-1971 and emphasized that this approach produced a clinical candidate now in Roche’s pipeline. [T2: PMID 37774359]

All three programs close a selectivity gap that prior agents could not.

1.3 Differentiation

Relay’s isoform-selective small molecules use structure-based design against ATP-pocket residue differences. This selectivity removes class toxicities while preserving on-target efficacy—something pan-inhibitors cannot achieve.

2.1 Target Biology

SHP2 drives RAS-MAPK signaling in solid tumors. [T2: PMID 39632729] Migoprotafib inhibits SHP2 potently and selectively for RAS-MAPK-driven cancers, especially in combination with other targeted agents. [T2: PMID 39632729]

PIK3CA ranks among the most frequently mutated cancer genes, with hotspot mutations in the kinase and helical domains. [T2: PMID 37916956] These mutations occur in approximately 8% of all cancers and roughly 40% of hormone receptor-positive breast cancers. [T2: PMID 37916958] Alpelisib and other orthosteric PI3Kα inhibitors block wild-type and mutant enzyme equally, causing hyperglycemia. [T2: PMID 37916956] Molecular dynamics simulations and cryo-EM revealed an allosteric network that explains how mutations favor PI3Kα activation, informing the design of mutant-selective molecules. [T2: PMID 37916956]

FGFR2 drives fusion-positive intrahepatic cholangiocarcinoma clinically, as pan-FGFR inhibitor efficacy validates. [T2: PMID 37270847] These agents suffer a limited therapeutic window: FGFR1-mediated hyperphosphatemia, FGFR4-mediated diarrhea, and FGFR2 resistance mutations limit their benefit. [T2: PMID 37270847]

2.2 Validation Layers

RLY-4008 induced regression in multiple xenograft models containing FGFR2 resistance mutations while sparing FGFR1 and FGFR4 in vivo. [T2: PMID 37270847] In early clinical testing, it produced responses without clinically significant off-isoform FGFR toxicities, confirming proof of concept for selective FGFR2 inhibition. [T2: PMID 37270847]

RLY-2608 inhibited tumor growth in PIK3CA-mutant xenografts with minimal impact on insulin. [T2: PMID 37916956] It produced objective tumor responses in two patients with advanced HR+ breast cancer without wild-type PI3Kα-related toxicities. [T2: PMID 37916956]

Migoprotafib demonstrated pharmacodynamic pathway modulation (peripheral blood phospho-ERK) in a Phase Ia trial and produced stable disease in 10 of 56 patients (18%). [T2: PMID 39632729] The drug showed predictable, dose-dependent PK and an effective half-life supporting once-daily dosing. [T2: PMID 39632729]

Early clinical results across programs validate the mechanism and selectivity hypothesis. Completed trials (NCT04252339, NCT04526106, NCT05487235) have posted no results. Small cohorts preclude confirming breadth.

2.3 Target Validation Score

The evidence validates the targets. PIK3CA mutations occur in ~40% of HR+ breast cancers, and FGFR2 fusions drive intrahepatic cholangiocarcinoma; approved inhibitors have validated both. Preclinical studies demonstrate molecular and cellular selectivity, and early clinical signals from RLY-4008 and RLY-2608 show activity with clean safety profiles. High druggability, top-tier publications, high disease prevalence, emerging safety data, and early clinical activity all support strong target validation. Pivotal data have not arrived, preventing a ceiling score. The directional signal aligns with strong target validation.

3.1 Data Summary

RLY-4008 (lirafugratinib): Elevar Therapeutics led REFOCUS (NCT04526106), a first-in-human Phase I/II study of 490 patients last updated 2026-02-27. [T1: NCT04526106] Investigators have not posted results from the expansion cohorts. [T1: NCT04526106] Published early-phase results show that RLY-4008 produced responses without clinically significant off-isoform FGFR toxicities, confirming selective FGFR2 targeting. [T2: PMID 37270847] As of May 2026, investigators are recruiting 30 patients for a Phase 2 single-arm study in non-CCA solid tumors with FGFR2 fusions (NCT07359820). [T1: NCT07359820]

RLY-1971 (migoprotafib, GDC-1971): Migoprotafib completed a Phase I dose-escalation/expansion study (NCT04252339) in 56 patients in November 2022. [T1: NCT04252339] Investigators have not posted results. [T1: NCT04252339] A published Phase Ia analysis set the maximum tolerated dose at 100 mg and the recommended Phase II dose at 60 mg daily. The study reported rapid absorption (~0.5–2 hours), dose-dependent exposure, pathway modulation, and predictable PK supporting once-daily dosing. [T2: PMID 39632729] Migoprotafib produced stable disease in 10 patients (18%); frequent adverse events included diarrhea, peripheral edema, dyspnea, anemia, constipation, fatigue, AST increase, and thrombocytopenia. [T2: PMID 39632729] Investigators completed combination trials with atezolizumab, osimertinib, or cetuximab (NCT05487235, 57 patients; NCT05954871, 10 patients) but have posted no results. [T1: NCT05487235; T1: NCT05954871]

RLY-5836: Investigators enrolled 41 patients in a first-in-human Phase I study (NCT05759949) and completed it in April 2025. [T1: NCT05759949] Investigators have not posted results. [T1: NCT05759949]

RLY-2608: Investigators are evaluating RLY-2608 in the Phase I/Ib study NCT05216432 across single-agent and combination arms (fulvestrant, CDK4/6 inhibitors). [T1: NCT05216432] Public data from NCT05216432 remain unavailable. Published proof-of-concept results report objective tumor responses in two patients with advanced HR+ breast cancer harboring kinase or helical domain PIK3CA mutations, without wild-type PI3Kα-related toxicities. [T2: PMID 37916956]

Investigators are recruiting 277 participants for a Phase 2 study in PIK3CA-Related Overgrowth Spectrum and vascular malformations (NCT06789913); the study includes a randomized, double-blind, placebo-controlled portion plus a basket expansion. [T1: NCT06789913]

The registrational Phase 3 ReDiscover-2 trial (NCT06982521) compares RLY-2608 plus fulvestrant against capivasertib (an AKT inhibitor) plus fulvestrant in PIK3CA-mutant HR+/HER2− metastatic breast cancer after CDK4/6 inhibitor progression. [T1: NCT06982521] The trial began in August 2025, plans to enroll 540 patients, and uses progression-free survival by blinded independent central review as its primary endpoint. [T1: NCT06982521] Investigators are actively recruiting. [T1: NCT06982521]

3.2 Source vs. Company Claims

Relay has not posted results from any completed efficacy study on ClinicalTrials.gov. The published early-phase results represent small cohorts: two responders for RLY-2608, an unspecified number for RLY-4008. Early results show clean selectivity-driven safety profiles in these subsets. Larger, powered studies have not verified response magnitude and durability.

Data from the ReFocus trial reveal on-target FGFR2 kinase domain mutations and RTK-MAPK bypass alterations as principal resistance pathways. [T2: PMID 41571046] M538 and/or L618 kinase domain mutations appeared in 11 of 16 evaluable patients with acquired resistance (69%). [T2: PMID 41632446] Lirafugratinib and futibatinib show complementary activity against these mutations, supporting sequential treatment but highlighting finite monotherapy durability. [T2: PMID 41632446]

Secondary resistance mutations emerged in orthosteric PI3Kα inhibitor-treated patients and altered the inhibitor binding pocket. RLY-2608 preclinically overcame all such mutations. [T2: PMID 37916958] Clinical data have not yet proved whether this preclinical advantage holds or whether new resistance pathways will surface.

Management has sold stock systematically while pivotal data remain unpublished (see verified claims C001–C014). This pattern could weigh on investor confidence ahead of key catalysts.

Single-author concentration risk. For RLY-2608, three or more mechanism and efficacy claims come from a single publication [T2: PMID 37916956] (first author Varkaris A); similarly, three or more for RLY-4008 come from [T2: PMID 37270847] (first author Subbiah V). Relying on individual lab publications for core claims creates reproducibility risk: if those results do not replicate, the primary scientific rationale narrows materially. Independent replication has not been verified for the core efficacy claims.

4.1 FAERS Disproportionality

PI3Kα inhibitors carry a dominant hyperglycemia signal. Alpelisib generated 109 death, 151 diarrhea, 192 rash, 74 diabetes mellitus, 325 hyperglycemia, and 231 malignant neoplasm progression reports, with combined glycemic events exceeding 450. [T1: FAERS alpelisib] Rash ranks third; pneumonitis appears in 19 FAERS reports but lacks independent confirmation as a formal class warning. [T1: FAERS alpelisib; T2: C1413 — unverified as class warning] RLY-2608 must beat these benchmarks to win share in PIK3CA-mutant HR+/HER2− breast cancer. [T1: FAERS alpelisib]

Inavolisib, approved in October 2024, already shows hyperglycemia as the top signal (35 reports), with neutropenia (19), diarrhea (15), thrombocytopenia (14), and rash (5). [T1: FAERS inavolisib] Four reports of Fournier’s gangrene flag a serious post-marketing concern. [T1: FAERS inavolisib] The inavolisib FAERS hyperglycemia signal raises the hypothesis that even a mutant-degrader may not fully spare wild-type PI3Kα tone; if that interpretation holds, RLY-2608 must demonstrate a meaningfully cleaner glycemic profile to differentiate. [T1: FAERS inavolisib; T2: C1480 — mechanistic inference, not verified]

Pan-FGFR inhibitors carry a different liability: hyperphosphatemia, retinal toxicity, and nail disorders. Erdafitinib’s FAERS dataset contains 97 death, 39 diarrhea, 19 hyperphosphatemia, 10 retinal detachment, and 10 blood phosphorus increased reports. [T1: FAERS erdafitinib] Dry eye (17), blindness (5), onycholysis (26), and nail disorder (12) reflect FGFR1- and FGFR4-mediated toxicity. [T1: FAERS erdafitinib] The FAERS signal associates FGFR1 inhibition with retinal toxicity, though the underlying mechanism (C1440) is UNSUPPORTED by independently verified data; the FAERS observation is consistent with FGFR1 biology but not confirmed. [T1: FAERS erdafitinib; T2: C1440 — unverified] RLY-4008’s >250-fold FGFR1 selectivity should reduce this risk. [T1: FAERS erdafitinib]

Futibatinib’s smaller dataset mirrors the class signal: hyperphosphatemia (5) and blood phosphorus increased (1). [T1: FAERS futibatinib] Drug-ineffective (7) and disease-progression (6) reports point to durability limits in the refractory setting. [T1: FAERS futibatinib]

Infigratinib generated too few FAERS reports for meaningful signal characterization (maximum 3 per term), consistent with a narrow second-line FGFR2-fusion cholangiocarcinoma indication that generated insufficient real-world exposure. [T1: FAERS infigratinib] The FDA approved infigratinib in May 2021; distributors stopped distribution in late 2022; the FDA withdrew approval in May 2024. [T1: FAERS infigratinib]

Pemigatinib’s pan-FGFR class toxicities dominate: hyperphosphatemia, nail disorders (discolouration, onychomadesis), dry eye, and stomatitis. [T2: C1502–C1529 — pemigatinib FAERS counts not independently verified; report as qualitative class signal only] FGFR1 and FGFR4 mediate these off-target effects. RLY-4008’s >250-fold FGFR2 selectivity should spare patients from them.

4.2 Class-Wide Signals

Hyperglycemia and rash limit dosing for PI3Kα inhibitors. Hyperphosphatemia (FGFR1) and nail, retinal, and mucosal effects (FGFR1 and FGFR4) define the FGFR class. Relay designed its molecules to avoid these through mutant-selective (RLY-2608) or isoform-selective (RLY-4008) inhibition. Early clinical results show clean off-isoform FGFR safety profiles for RLY-4008 [T2: PMID 37270847; C1112] and clean wild-type PI3Kα safety profiles for RLY-2608 [T2: PMID 37916956; C1173]. The observations derive from small, early-phase datasets. Larger completed trials have not published safety data, limiting the ability to fully certify class-signal avoidance.

4.3 Disqualifying Findings

FAERS contains no entries for Relay’s own candidates. No post-marketing signal warns of unrecognized toxicities. Investors must rely on competitor benchmarks and preclinical predictions until mature clinical data arrive. On-target resistance mutations have emerged for lirafugratinib (FGFR2 M538/L618) and for orthosteric PI3Kα inhibitors; although RLY-2608 overcame these preclinically, resistance will still develop. Large trials must still verify long-term clinical safety. Published patient-level data support RLY-2608’s glycemic safety only through the two-patient proof-of-concept. The Phase 3 trial will serve as the definitive arbiter.

5.1 Chain of Title

This category contains no verified claims.

5.2 Maintenance & Lapse Risk

This category contains no verified claims.

5.3 Freedom to Operate

This category contains no verified claims.

6.1 Verified Credentials

This category contains no verified claims.

6.2 Publication Record

This category contains no verified claims.

6.3 Flags

This category contains no verified claims.

7.1 Capital Raised vs. Claimed

The current evidence set verifies no capital-raise or financing-round claims.

7.2 Cash Runway

The current evidence set verifies no cash runway figure from financial filings.

7.3 Insider Transactions

Senior officers sold shares systematically across November 2025 – April 2026 while pivotal Phase 3 data remain unpublished. All transactions are VERIFIED from SEC Form 4 filings.

Sanjiv Patel (President and CEO) sold 62,073 shares at $7.00 on 2025-11-03, retaining 703,215 shares [T1: C014], and 43,168 shares at $7.82 on 2026-01-06, retaining 661,041 shares. [T1: C013] The CEO sold near multi-month lows before the stock recovered to the $13–15 range in early 2026.

Thomas Catinazzo (CFO) sold: 13,820 shares at $7.62 on 2026-01-27, retaining 250,996 [T1: C011]; 1,695 shares at $8.45 on 2026-01-28, retaining 249,301 [T1: C012]; 17,717 shares at $10.06 on 2026-03-09, retaining 231,584 [T1: C006]; 17,717 shares at $13.01 on 2026-04-07, retaining 213,867 [T1: C004]; 1,800 shares at $15.00 on 2026-04-09, retaining 213,867 [T1: C005]; and 972 shares at $14.79 on 2026-04-28, retaining 212,895. [T1: C001] The CFO sold across five price points from $7.62 to $15.00.

Donald A Bergstrom (President, R&D) sold: 18,895 shares at $7.62 on 2026-01-27, retaining 422,733 [T1: C007]; 2,686 shares at $8.45 on 2026-01-28, retaining 420,047 [T1: C008]; and 1,490 shares at $14.79 on 2026-04-28, retaining 418,557. [T1: C003]

Peter Rahmer (Chief Corporate Development Officer) sold: 11,684 shares at $7.62 on 2026-01-27, retaining 277,961 [T1: C009]; 1,354 shares at $8.45 on 2026-01-28, retaining 276,610 [T1: C010]; and 753 shares at $14.79 on 2026-04-28, retaining 275,857. [T1: C002]

The pattern is consistent with a scheduled diversification plan but concentrated in the pre-catalyst window. Whether these represent pre-scheduled plan sales or discretionary selling remains unspecified in the verified Form 4 filings. The pattern warrants disclosure in any investor communication.

7.4 Implied Valuation

No verified claims support an independent implied valuation calculation. Market cap and enterprise value figures would require current trading data not in the verified set.

8.1 FDA Precedent

Alpelisib plus fulvestrant won FDA approval for PIK3CA-mutant HR+/HER2− breast cancer, establishing a regulatory precedent for PI3Kα inhibitors. [T2: PMID 37916958] Inavolisib, a mutant-degrader, gained approval in October 2024, further validating the pathway. [T1: FAERS inavolisib] Futibatinib secured approval for FGFR2-fusion cholangiocarcinoma. The FDA withdrew infigratinib in May 2024 after limited commercial uptake—a caution that narrow genetic subsets carry execution risk. [T1: FAERS futibatinib; T1: FAERS infigratinib]

8.2 Designation Status

The verified set contains no claims regarding breakthrough therapy, fast track, orphan drug, or other special designations.

8.3 Approval Path

ReDiscover-2 (NCT06982521) uses progression-free survival by blinded independent central review as its registration endpoint, with capivasertib plus fulvestrant as the active comparator. [T1: C831, C832] Investigators are recruiting 540 patients, consistent with a registrational design. [T1: C801] The Phase 2 PROS study (NCT06789913) includes a randomized, double-blind, placebo-controlled arm that could support a future rare-disease or tissue-agnostic submission.

The verified claims omit FDA feedback on these programs. The pathway architecture aligns with prior successful submissions.

9.1 Verified TAM

The verified evidence set quantifies no total addressable market size.

9.2 Prevalence Data

PIK3CA mutations occur in approximately 8% of all cancers and roughly 40% of hormone receptor-positive breast cancers. [T2: PMID 37916958; T2: PMID 37916956] FGFR2 fusions drive intrahepatic cholangiocarcinoma, but the verified claims do not specify population incidence. RIT1 mutations occur in 2.4% and RIT1 amplifications hit up to 14% of lung adenocarcinomas; patients with RIT1 alterations currently fall into the “oncogene-negative” category and lack targeted therapy options. [T2: PMID 40608319]

9.3 Standard of Care & Pricing

The verified set contains no claims regarding standard-of-care treatment patterns or pricing data.

10.1 Competitor Stage Map

The PI3Kα field is crowded with approved agents. Alpelisib (Piqray) holds FDA approval with fulvestrant but carries hyperglycemia (325 FAERS reports) and rash liabilities. [T1: FAERS alpelisib; C1398] Inavolisib (Itovebi), a mutant-degrader approved in October 2024, shows hyperglycemia as its top FAERS signal, indicating residual wild-type PI3Kα inhibition. [T1: FAERS inavolisib] Capivasertib (Truqap), an FDA-approved AKT inhibitor for the same population, serves as the active comparator in RLY-2608’s Phase 3 ReDiscover-2 trial. STX-478 (Scorpion Therapeutics), an allosteric PI3Kα inhibitor, avoided metabolic dysfunction in xenografts and combined well with fulvestrant and CDK4/6 inhibitors; Phase 1 data remain unpublished, but a 2023 Cancer Discovery paper established its preclinical profile. [T2: PMID 37623743]

The FGFR2 field is populated by pan-FGFR agents with class-wide toxicities. Futibatinib (Lytgobi), an FDA-approved irreversible pan-FGFR for FGFR2-fusion CCA, shows class-consistent hyperphosphatemia and durability limitations in FAERS. [T1: FAERS futibatinib] Pemigatinib (Pemazyre), an FDA-approved pan-FGFR, carries class toxicities including hyperphosphatemia, nail disorders, and stomatitis. [T1: FAERS pemigatinib] Erdafitinib (Balversa), an FDA-approved FGFR1-3 inhibitor, generates a retinal-toxicity FAERS signal; boxed-warning status is not independently verified from the FDA label. [T1: FAERS erdafitinib; T2: C1439 — unverified] Infigratinib (Truseltiq) won FDA approval and later had it withdrawn because of commercial fragility and limited exposure—a cautionary example of niche vulnerability. [T1: FAERS infigratinib]

Lirafugratinib (RLY-4008) delivers >250-fold FGFR2 selectivity and should avoid the class toxicities that handicap pan-FGFR agents.

10.2 Differentiator Durability

RLY-4008’s selectivity profile could eliminate hyperphosphatemia and retinal toxicity; early clinical reports support this, though investigators have not yet posted REFOCUS results. Acquired resistance mutations (M538, L618) have emerged, but lirafugratinib and futibatinib show complementary mutation coverage, suggesting sequential therapy may cap monotherapy durability while opening a pairing strategy. [T2: PMID 41632446]

RLY-2608’s allosteric mechanism preclinically overcomes the secondary resistance mutations that limit orthosteric PI3Kα inhibitors. The Phase 3 head-to-head trial against capivasertib will test whether this translates into a clinically meaningful advantage. [T2: PMID 37916958] The PROS indication, if successful, provides a non-oncology niche that no competitor in the verified dataset addresses.

Pivotal data will determine whether RLY-2608 achieves durable differentiation. A Cancer Discovery commentary already characterizes mutant-selective allosteric PI3Kα inhibition as a “class breakthrough that decouples antitumor activity from hyperglycemic toxicity.” [T2: PMID 38327193]

11.1 Contradicted Findings

The verified claims show no source disagreement.

11.2 Unsupported Claims

Several claims in the verified set lack support. Among them, C949–C994 (site locations for NCT06982521) and material claims such as C1247, C1267, C1268, C1269 could not be confirmed.

Two unsupported claims address material safety topics and require explicit disclosure:

• C1413 (pneumonitis class warning): The report notes 19 pneumonitis FAERS reports for alpelisib. No independent regulatory source confirmed that pneumonitis constitutes a formal class warning. Investors should not rely on this characterization for safety benchmarking.
• C1480 (glycemic mutant-degrader inference): The claim that a mutant-degrader mechanism incompletely spares wild-type PI3Kα tone—implying glycemic toxicity persists even for RLY-2608—is a mechanistic inference from the inavolisib FAERS signal, not a verified regulatory or clinical finding. The inference may prove correct, but it is not established.

11.3 Existential vs. Manageable

Relay faces several existential risks. The company still lacks mature clinical data despite completing Phase 1 studies for its three lead programs; sponsors have not posted results. The Phase 3 ReDiscover-2 study for RLY-2608 continues recruiting. Valuation would suffer materially from failure to demonstrate superiority over capivasertib or from an unacceptable safety signal.

Resistance biology threatens durability. The majority of patients with acquired resistance develop polyclonal FGFR2 kinase-domain mutations and RTK-MAPK bypass pathways as principal resistance mechanisms against RLY-4008. Short response durability would limit the commercial opportunity to sequential use dependent on diagnostic logistics.

Insider selling creates perception risk. Management has sold stock systematically while pivotal data remain unpublished (see verified claims C001–C014). This pattern could weigh on investor sentiment ahead of data readouts, regardless of whether sales reflect scheduled diversification or skepticism about near-term catalysts.

Reproducibility risk stems from single-author thesis anchors. Three or more efficacy and mechanism claims for each of RLY-2608 [T2: PMID 37916956] and RLY-4008 [T2: PMID 37270847] derive from single-PI publications with no verified independent replication. If these results do not hold in broader data, the core scientific rationale narrows materially.

The company also faces manageable risks. Relay designed its molecules to eliminate known class toxicities, and early clinical signals align with that design. Clean safety data from Phase 3 would differentiate the therapeutic window.

Regulatory precedents reduce risk along the path. Alpelisib, inavolisib, and futibatinib earned approvals that indicate a well-trodden regulatory route, as do standard PFS endpoints.

Programs beyond oncology diversify value. The PROS study and lirafugratinib’s non-CCA expansion diversify revenue potential if an oncology indication underperforms.

11.4 Risk Severity

Existential risks: clinical failure in Phase 3 ReDiscover-2 or an unacceptable safety signal. Manageable risks: class toxicity design, regulatory precedent, and pipeline diversification each reduce but do not eliminate execution risk.

12.1 Platform Beyond Lead

Relay’s discovery engine relies on motion-based structural biology and computational methods. A 2024 Nature paper described AF-Cluster, a technique that clusters multiple-sequence alignments to enable AlphaFold2 to predict alternative protein conformations. The authors identified a fold-switched state of the circadian protein KaiB and designed mutations that flip the protein between states. [T2: PMID 37956700] Dorothee Kern, a Relay scientific founder, co-authored the study. [T2: PMID 37956700, metadata]

Lirafugratinib validated the platform’s ability to drug a kinase-family selectivity gap that prior scientists considered undruggable. [T2: PMID 38300868] Using DNA-encoded library screens, cryo-EM, and motion-based design, Relay produced RLY-2608, an allosteric mutant-selective PI3Kα inhibitor. [T2: PMID 37916956] An independent Cancer Discovery editorial called the approach a “class breakthrough,” suggesting the platform can extend to other targets with selectivity windows. [T2: PMID 38327193]

12.2 Repurposing Potential

Relay is expanding its programs beyond their initial indications. Investigators are evaluating RLY-2608 in a Phase 2 trial (NCT06789913) for PIK3CA-Related Overgrowth Spectrum and vascular malformations, including patients aged 2 and older, with a randomized, placebo-controlled portion positioning the program to enter a genetically defined non-oncology indication that lacks approved therapies. [T1: NCT06789913]

Investigators are testing lirafugratinib in non-CCA solid tumors with FGFR2 fusions (NCT07359820), broadening beyond intrahepatic cholangiocarcinoma. [T1: NCT07359820]

Investigators are testing migoprotafib in combination trials with osimertinib (EGFR-mutant NSCLC) and cetuximab (CRC), positioning it as a backbone agent across RAS-MAPK-driven cancers. [T1: NCT05954871]

12.3 Acquirer Profile

Roche partnered with Relay on migoprotafib during development. The discovery paper listed both Relay and Genentech authors and placed the candidate in Roche’s pipeline. [T2: PMID 37774359] Roche sponsors the ongoing combination trials (NCT05487235, NCT05954871). [T1: NCT05487235, lead sponsor Genentech; T1: NCT05954871, lead sponsor Genentech]

Relay retains full ownership of its PI3Kα and FGFR2 programs in validated oncology targets. Large pharma oncology divisions regularly acquire late-stage targeted assets with novel selectivity profiles. Relay’s platform for tackling undruggable kinases adds value for buyers seeking next-generation inhibitors.