Amlitelimab

Amlitelimab is Sanofi's fully human anti-OX40L antibody that completed a four-trial Phase 3 program (OCEANA) in moderate-to-severe atopic dermatitis. On 24 July 2026, Sanofi announced it would not file for regulatory approval in AD, concluding the asset "would not represent a meaningful improvement to the standard of care" despite COAST 1 hitting every primary and key secondary endpoint and SHORE and the long-term extension reading out favorably on durability [13]. COAST 2 landed mixed: amlitelimab hit the primary vIGA-AD 0/1 (clear or almost-clear skin on a validated physician global score) but missed the more stringent vIGA-AD 0/1 with barely-perceptible-erythema co-primary, and EASI-75 (75% reduction on the Eczema Area and Severity Index) crossed nominal significance only [14][15]. The dermatology story is effectively over. Development continues in Phase 2a/b nonresponsive celiac disease (NCT06557772) [8] and in a scleroderma platform trial [9]. For investors, amlitelimab is now a case study in positive-trial-but-commercially-unviable biologics in crowded second-line AD, not a Dupixent challenger.

Sanofi acquired KY1005 (now amlitelimab) via its $1.1B-upfront, up-to-$350M-in-milestones purchase of Kymab, announced 11 January 2021 and closed 8 April 2021 [11]. The Phase 3 OCEANA program ran four pivotal trials plus a long-term extension: COAST 1 (NCT06130566, n=601, monotherapy vs placebo in patients ≥12 y) [16], COAST 2 (NCT06181435, n=589, monotherapy vs placebo) [6], SHORE (background topical corticosteroids), ESTUARY (NCT06241118, n=464, biologic/JAK-inadequate patients) [5], and AQUA long-term safety extension (NCT05492578, n=1856) [7]. Readouts cadence: COAST 1 positive September 2025, additional data at AAD March 2026, mixed COAST 2 and positive SHORE between January and March 2026, and discontinuation of the AD program on 24 July 2026 [13]. No FDA Breakthrough Therapy, Fast Track, or Orphan designations were ever granted in AD. Live indications after the AD wind-down are a Phase 2a/b in nonresponsive celiac disease (NCT06557772) [8] and the Scleroderma Research Foundation platform trial arm (NCT06195072) [9]. The internal registry NCT (NCT05341336) was wrong - that identifier maps to an unrelated Assiut University surgical study.

OX40L is a protein on the surface of antigen-presenting cells (dendritic cells, B cells). When T cells engage it through their OX40 receptor, they survive longer, divide more, and generate memory cells, especially the Th2 subset that drives allergic inflammation [12]. In atopic dermatitis, this chronic T-cell co-stimulation is a core driver of the itch-scratch-inflammation cycle. Amlitelimab binds OX40L and blocks that signal. The pathway is mechanistically distinct from Dupixent (dupilumab), which blocks the IL-4Rα receptor downstream and specifically shuts off Th2 cytokine signaling: OX40L blockade is further upstream and broader, hitting Th1 and Th17 activation as well. Human genetics tie TNFSF4 (OX40L) variants to atopic eczema (Open Targets evidence 0.59), lupus (0.58), and asthma (0.56), supporting the target. The target-level validation remained unproven in commercial terms as of October 2026: no OX40 or OX40L antibody has reached approval, and the rival anti-OX40 receptor antibody rocatinlimab (Amgen/Kyowa Kirin) is still in Phase 3. The failure of amlitelimab to generate a differentiated clinical profile in the most heavily studied indication raises the question of whether the OX40 axis simply has a modest clinical ceiling in AD relative to IL-4/IL-13 blockade.

The pivotal OCEANA program used vIGA-AD 0/1 (clear or almost-clear skin on a validated physician global scale) and EASI-75 (≥75% reduction from baseline on the Eczema Area and Severity Index, a standard composite of lesion area and severity) as co-primary endpoints at Week 24. Dosing was a loading phase followed by Q4W (every 4 weeks) or Q12W (every 12 weeks) maintenance, supported by population PK/PD modeling [4]. COAST 1 (NCT06130566, n=601) read first in September 2025 and met every primary and key secondary endpoint across both dosing regimens in patients ≥12 y [16]. COAST 2 (NCT06181435, n=589) hit vIGA-AD 0/1 (25.3% Q4W and 25.7% Q12W vs 14.8% placebo) but failed the more stringent vIGA-AD 0/1 with barely-perceptible-erythema co-primary, and EASI-75 reached only nominal significance in some analyses (41.8-48.1% treated vs 32.3% placebo) [14][15]. SHORE (background topical corticosteroid arm) met its primary endpoint. ESTUARY (NCT06241118, n=464) was the biologic/JAK-inadequate differentiation study with primary at Week 36 [5]; AQUA (NCT05492578, n=1856) was the long-term safety and durability extension [7]. There was no head-to-head against Dupixent; cross-trial comparison to Dupixent's roughly 45-50% EASI-75 at Week 16 was always going to be unforgiving, and the mixed COAST 2 EASI-75 numbers against that benchmark are a core reason the program could not differentiate commercially.

Our model estimates a 42% chance this drug is eventually approved. It starts from the historical base rate for Phase 3 drugs in this area (about 61%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by a non-randomized design and its light or open-label blinding; it is held back by smaller-than-typical enrollment for this phase and weak or limited earlier-phase results. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Primary risk is already realized in AD: Sanofi walked away from a positive Phase 3 program because the magnitude of effect did not clear the Dupixent + Ebglyss + Adbry commercial bar. The mixed COAST 2 EASI-75 (41.8-48.1% treated vs placebo 32.3%) sits below Dupixent's ~45-50% EASI-75 at Week 16 across its registrational trials, with the placebo arm running unusually high - a known difficulty in modern AD trials but one that compressed amlitelimab's absolute benefit. Residual risks on the surviving program: celiac is a difficult indication with no approved disease-modifying biologic, no validated biomarker for Phase 2/3 transition, and a long history of failed candidates (larazotide, IMGX003 etc). Scleroderma platform arm has even weaker read-across. Safety signal from AQUA (n=1856) remained acceptable through the AD wind-down, so safety is not the open question. The commercial read-across for the OX40 pathway more broadly is now negative: rocatinlimab (Amgen/Kyowa Kirin, anti-OX40 receptor) is running Phase 3 against the same AD benchmark that just defeated amlitelimab. If rocatinlimab delivers a cleaner EASI-75 profile, the mechanistic story reopens; if not, the OX40 axis in AD looks structurally limited versus IL-4/IL-13 blockade.

Watch value is now low and historical, not forward-looking. The amlitelimab story is a clean worked example of how clinical success and commercial viability can decouple in a saturated biologic market: COAST 1 hit every endpoint, SHORE hit, long-term extension showed durability, and Sanofi still shelved it. The signal that matters for the BioCosm pipeline model is the gap between "met primary endpoint" and "filed for approval" in indications with multiple established biologics - that gap is a real failure mode PoS models should capture via competitive_density, not treat as sponsor behavior noise. Forward-looking, the pathway question re-anchors on rocatinlimab (Amgen/Kyowa Kirin) Phase 3 readouts in 2026-2027: a positive rocatinlimab result argues the ligand-versus-receptor distinction matters and amlitelimab's shelving was execution-specific; a negative result argues the entire OX40 axis has a modest ceiling in AD. For amlitelimab specifically, check back on the nonresponsive celiac disease Phase 2a/b (NCT06557772) [8] - that is a greenfield indication where modest clinical effect sizes retain commercial value and the drug could still find a label. IP position: Kymab transferred patent families covering KY1005/amlitelimab composition-of-matter to Sanofi in 2021 [11]; projected composition-of-matter exclusivity runs into the mid-to-late 2030s on standard biologic patent timelines, so the shelved AD program does not create an immediate loss-of-exclusivity concern for the residual indications.