RO7204239

RO7204239 (emugrobart, GYM329) is Roche's anti-latent myostatin monoclonal antibody, originated at Japanese subsidiary Chugai. As of March 23, 2026, Roche and Chugai discontinued development in spinal muscular atrophy (MANATEE) and facioscapulohumeral muscular dystrophy (MANOEUVRE) after both Phase 2 trials failed to show consistent improvement over standard of care; safety was clean, the issue was efficacy [11]. Two programs remain active: the GYMINDA obesity combination trial with tirzepatide (NCT06965413, n=285, primary completion August 2026) [2] and a small Phase 1 in type 2 diabetes (NCT07137585) [5]. The SMA failure was not a surprise to anyone tracking the class - bimagrumab, domagrozumab, landogrozumab, and trevogrumab all hit muscle volume targets without delivering functional benefit. Scholar Rock's apitegromab, a same-mechanism latent-myostatin antibody, broke that pattern in the Phase 3 SAPPHIRE SMA trial (HFMSE +1.8 points vs placebo, p=0.0192) and is the only program in this space with a confirmed Phase 3 win [3]. The story for emugrobart is now obesity, where Roche is going after the muscle-preservation problem that dogs GLP-1 weight loss alongside Lilly's bimagrumab (acquired via Versanis in 2023) and Regeneron's trevogrumab.

RO7204239 has no approvals anywhere and no FDA designations disclosed. As of June 2026, the program is materially reduced from its early-2026 footprint. MANATEE (NCT05115110, Phase 2, n=259, SMA add-on to risdiplam) was discontinued March 23, 2026 after Part 1 failed to show consistent muscle growth or functional improvement over risdiplam alone [1][11]. MANOEUVRE (NCT05548556, Phase 2 in FSHD, n=51) was discontinued the same day for similar reasons - failure to demonstrate the hoped-for muscle growth and function gains [4][11]. Two programs remain active: GYMINDA (NCT06965413, Phase 2 obesity combo with tirzepatide, n=285), which measures percent body weight change at 52 weeks and has stated primary completion in August 2026 [2]; and a Phase 1 in type 2 diabetes (NCT07137585, n=30) using the hyperinsulinemic-euglycemic clamp - a procedure that holds blood glucose at a normal level via continuous insulin and glucose infusion, then measures glucose disposal as a gold-standard read on insulin sensitivity [5]. The GYMINDA obesity readout is the next and likely final binary catalyst for this molecule; expected late 2026 to mid-2027 based on the stated primary completion date.

Myostatin is the body's brake on muscle growth. A protein secreted by muscle cells, it tells those cells to stop dividing and stop bulking up. The biology is locked in by genetics - mice without the myostatin gene look bodybuilder-jacked, Belgian Blue cattle with a natural myostatin mutation carry roughly double the normal muscle mass, and rare humans with myostatin mutations have been described with similar phenotypes [6]. Block myostatin and muscle grows. That part of the story is not controversial. The hard part is doing it cleanly. Myostatin belongs to the TGF-beta family, a group of structurally related signaling proteins, and earlier antibodies in this class cross-reacted with cousins like activin A and GDF11. Bimagrumab (Novartis, now Lilly/Versanis) blocks the activin type II receptor; in the Heymsfield 2021 obesity/T2D Phase 2 trial, mild diarrhea and muscle spasms were the most common adverse events, with acne reported as a recurring class-associated effect [7]. Trevogrumab (Regeneron), domagrozumab (Pfizer), and landogrozumab/LY2495655 (Lilly) all grew muscle volume on MRI but did not produce functional improvement in their target indications; landogrozumab's pancreatic cancer Phase 2 in particular failed to confer clinical benefit and raised survival concerns specific to that population [8][12]. GYM329 is engineered to bind only the latent (inactive precursor) form of myostatin, which sits in an inhibitory complex with its own propeptide, and uses a recycling/sweeping antibody design that further reduces circulating active myostatin. The same selectivity logic underpins Scholar Rock's apitegromab. Latent-myostatin binding therefore looked like the cleanest mechanistic bet in this class - but the March 2026 MANATEE/MANOEUVRE failures show that mechanism selectivity alone does not guarantee a functional clinical effect, and apitegromab remains the only confirmed positive readout in the entire myostatin-blocking field [3][11].

MANATEE (NCT05115110) was a two-part Phase 2 study of emugrobart on top of risdiplam in SMA patients aged 2 to 10 years. Part 1 was the safety/dose run-in; Part 2 was the placebo-controlled efficacy portion, with motor function measures including the Hammersmith Functional Motor Scale Expanded (HFMSE) and the Revised Upper Limb Module (RULM) - standardized scored checklists that grade what an SMA patient can physically do, like sitting unsupported, lifting an arm, or moving from lying to sitting - as key outcomes [1]. The add-on design was scientifically sound: SMN-restoring therapies (risdiplam, nusinersen, onasemnogene) stop motor neuron loss but do not restore muscle that was already lost, and apitegromab's positive SAPPHIRE readout (HFMSE +1.8 points vs placebo at 52 weeks, p=0.0192; 30% of treated patients had >3-point HFMSE improvement vs 12.5% on placebo) [3] suggested the class could fill that residual gap. Emugrobart did not. The March 2026 announcement stated that participants on emugrobart plus risdiplam failed to show consistent improvement over risdiplam monotherapy in Part 1 [11]. MANOEUVRE (NCT05548556, n=51) used contractile muscle volume by MRI as its primary endpoint, which is a structural surrogate, not a function measure [4]; this is the same general approach that failed losmapimod (Fulcrum) in the Phase 3 REACH trial in FSHD after a partial Phase 2 ReDUX4 signal on MRI fat infiltration did not replicate functionally [13]. The MRI-anchored design has now misfired twice in FSHD. The still-active GYMINDA obesity trial (NCT06965413, n=285) layers emugrobart onto tirzepatide for 52 weeks with percent body weight change as the primary endpoint; the more diagnostic secondary will be lean mass preservation [2]. The T2DM Phase 1 (NCT07137585, n=30) uses the glucose clamp to test insulin sensitivity [5].

Our model gives this drug a 17% chance of eventually being approved. That figure starts from a baseline of about 34% - the historical approval rate for Phase 2 drugs in this area - then adjusts based on ten facts about the trial and sponsor. The estimate is pushed up by more secondary endpoints than usual, larger-than-typical enrollment for this phase, and the sponsor's strong track record of getting drugs approved; it is pulled down by weak or limited earlier-phase results. The remaining facts fall close to average for this stage, so they leave the number roughly where the baseline set it.

The headline SMA competitive risk has materialized. Roche conceded the SMA niche to Scholar Rock in March 2026 [11]. Apitegromab is the only confirmed winner in the class and per Scholar Rock's October 2024 topline announcement was planning Q1 2025 regulatory submission to the FDA [3]; investors tracking emugrobart should now treat apitegromab as the reference asset for any anti-myostatin readout in SMA. The class efficacy risk also materialized in both SMA and FSHD. Domagrozumab (Pfizer) failed its Phase 2 DMD trial in 2018 despite hitting target and growing muscle volume [8]. Landogrozumab/LY2495655 (Lilly) failed in pancreatic cancer with safety concerns specific to that population and did not advance further in oncology cachexia [12]. Bimagrumab grew muscle in inclusion body myositis without meaningful function gain, and a sarcopenia program was halted. Emugrobart now joins that pattern in SMA and FSHD. The remaining obesity program faces a different risk profile. Safety risk is moderate; the latent-myostatin binding approach should reduce cross-reactivity with activin and GDF11, but on-target effects on cardiac or smooth muscle remain theoretically possible. The bigger risk for GYMINDA is competitive: Lilly already has Phase 2 data for bimagrumab plus semaglutide showing fat-mass loss with lean-mass preservation, and Regeneron's trevogrumab is in late-stage development. Roche is not first to this combination concept. Commercial risk for the obesity indication is mainly market-structural - pricing, payer dynamics, and durability of muscle preservation past one year. There is no longer a meaningful SMA or FSHD commercial path for emugrobart.

The investment case for emugrobart narrowed sharply in March 2026. Two of four programs are dead; the remaining two are obesity (GYMINDA, primary completion August 2026, the real catalyst) and a T2DM Phase 1 (mechanism-of-action study). For BioCosm-tracked competitive context, the most important durable takeaway is that latent-myostatin binding does not by itself rescue a Phase 2 functional readout - apitegromab succeeded with mechanism plus dose plus indication match, and Roche did not replicate that. Watch GYMINDA for two things: (1) is the body composition signal real and consistent (lean mass preserved while fat mass falls), and (2) is the magnitude commercially meaningful versus tirzepatide monotherapy, which already produces substantial weight loss. A noisy or marginal result here likely ends the molecule. For SMA market sizing: roughly 9,000-10,000 prevalent SMA patients in the US across Types I-III based on indirect prevalence estimates; risdiplam (Evrysdi) generated approximately $1.8 billion in 2024 global sales with 16,000+ patients treated worldwide [14]. A second anti-myostatin entrant in even a fraction of that on-treatment population would have been a viable commercial asset - that opportunity now belongs to Scholar Rock alone. Roche reported CHF 60.5 billion (~$67 billion USD at prevailing 2024 exchange rates) in 2024 group revenue and is one of the few sponsors that can absorb a two-indication Phase 2 failure without dropping the molecule entirely [10]. Check Roche's 2026 H2 Pharma Day or earnings call for any GYMINDA interim color, and watch Scholar Rock's apitegromab BLA progression as the live competitive benchmark.