Delocamten in obstructive HCM: BMS halts its own backup myosin inhibitor

What was tried

MERCUTIO (NCT05556343) was a Phase 2a, open-label pilot of delocamten (MYK-224, BMS-986435), an oral cardiac myosin inhibitor, in adults with symptomatic obstructive hypertrophic cardiomyopathy and left ventricular outflow tract obstruction. Bristol-Myers Squibb dosed delocamten in two non-randomized cohorts, as monotherapy or layered on standard-of-care beta-blockers, calcium channel blockers, or disopyramide. The study enrolled 18 participants, recorded a primary completion date of August 2024, and was then terminated. The registry reason is a single line: "Business objectives have changed."

Delocamten sits in the same pharmacologic class as mavacamten, which BMS already markets as Camzyos. Both reduce the fraction of myosin heads available to form force-generating cross-bridges. The MERCUTIO endpoints were almost entirely safety and pharmacodynamic: adverse events, arrhythmia counts, and the incidence of treatment-emergent left ventricular ejection fraction below 50 percent. This was a tolerability and dose-finding study, not a powered efficacy trial.

The biological hypothesis

Hypertrophic cardiomyopathy is the most clearly Mendelian of the common cardiomyopathies. Pathogenic variants in sarcomeric genes, led by MYH7 (beta-cardiac myosin heavy chain) and MYBPC3, drive hypercontractility, impaired relaxation, and dynamic outflow obstruction. Open Targets scores the MYH7 to hypertrophic cardiomyopathy association at 0.91, near the top of its scale, with strong genetic and clinical support. The SHaRe registry showed that a sarcomere-positive genotype tracks with earlier disease onset and a higher lifetime burden of events (Ho 2018).

Cardiac myosin inhibitors do not touch the gene. They target the downstream biophysics. Green and colleagues showed in 2016 that a small molecule reducing actin-myosin cross-bridge formation normalized contractility and blunted hypertrophy in mouse models of HCM (Green 2016). Mavacamten carried that mechanism into patients: EXPLORER-HCM lowered outflow gradients and improved symptoms in obstructive disease (Olivotto 2020). Aficamten, a structurally distinct inhibitor, reproduced the effect in SEQUOIA-HCM (Maron 2024, Coats 2024). The class is validated. Delocamten was positioned as a next-generation entrant with a claimed wider therapeutic window, the engineering goal being a flatter exposure-response relationship and less risk of overshoot into systolic dysfunction.

What actually happened

MERCUTIO did not read out an efficacy failure. It stopped because the sponsor changed direction. The trial ran open-label in 18 patients, the posted outcomes are tolerability measures, and no safety alarm was named as the cause of termination. The stated reason is explicitly commercial.

The competitive context explains the move. By the time MERCUTIO wound down, BMS held mavacamten as an approved, marketed myosin inhibitor and was watching aficamten advance toward approval from Cytokinetics. A second BMS myosin inhibitor in early Phase 2 would have competed against the company's own franchise for the same patients, the same prescribers, and overlapping trial sites. Delocamten arrived inside BMS through the 2020 MyoKardia acquisition, the same deal that brought mavacamten. It was the backup, not the lead.

Failure mechanism, best guess

This is a portfolio decision, not a biological refutation. The target is among the best-validated in cardiology, the mechanism is approved and selling, and the molecule reached patients without a disclosed safety stop. The most parsimonious read is that delocamten's differentiation, a wider dosing window, was not worth a multi-year, effectively head-to-head investment against an already-approved in-house drug and a fast-moving external competitor. The asset's value was cannibalistic rather than additive.

There is a quieter possibility worth holding open. A wider therapeutic window only matters if the lead drug has a window problem. Mavacamten does carry a REMS for the risk of LVEF reduction and heart failure. If the Phase 2a data had shown delocamten clearly separating efficacy from LVEF decline, the differentiation case would have strengthened, not collapsed. The decision to stop suggests the early data did not make that case compelling enough to fund.

How to prevent this next time

Two quantitative gates would have surfaced this outcome before the trial opened, not after 18 patients.

First, a decision-analytic portfolio model. The relevant quantity is not delocamten's standalone probability of approval but its incremental expected value given that BMS already owns the class. A second-in-class asset against your own marketed drug should clear an explicit hurdle: incremental risk-adjusted net present value above zero after subtracting franchise cannibalization. That number was almost certainly negative once mavacamten was approved, and it is computable at the go decision.

Second, a Bayesian success criterion tied to the actual differentiation claim. The differentiator was a wider window, so the trial should have been powered to detect a difference in the rate of symptomatic LVEF below 50 percent between delocamten and a mavacamten benchmark, not merely to describe delocamten alone. To detect a halving of that event rate from 10 percent to 5 percent at 80 percent power and a two-sided alpha of 0.05 requires roughly 435 patients per arm. An 18-patient open-label cohort cannot move that posterior. Formalizing the readout as

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

would have shown that the predictive probability of ever winning a differentiation claim was low given the planned sample, independent of the biology.

The single highest leverage change would have been to define delocamten's go criterion as incremental value against mavacamten, quantified before first dose, rather than as standalone class activity.

What this means for similar programs

Second-in-class assets inside the company that already owns the class face a different bar than first entrants. The biology is not the question. The question is whether a measurable, clinically meaningful separation from the in-house standard exists, and that separation has to be specified as a powered endpoint from the first study. Aficamten succeeded as an external competitor precisely because it could be framed against an incumbent it did not own. A backup myosin inhibitor at BMS had no such framing. Programs acquired as part of a platform deal deserve the same scrutiny: a pipeline-in-a-product acquisition often carries redundant assets whose value evaporates once the lead is approved.

Open questions

Did the 18-patient cohort show any separation between symptom or gradient response and LVEF decline that would have supported the wider-window thesis? Were the posted arrhythmia and LVEF data ever analyzed against mavacamten's Phase 2 profile? Does delocamten have a future in non-obstructive HCM or in genotype-defined subsets where mavacamten was less studied, or has BMS retired the molecule entirely? And does the broader field need more than two marketed myosin inhibitors, given that aficamten and mavacamten already crowd the obstructive indication?

Sources

1. ClinicalTrials.gov. A Phase 2a Open-label Pilot Study of MYK-224 in Symptomatic Hypertrophic Cardiomyopathy and LVOT Obstruction (MERCUTIO), NCT05556343. https://clinicaltrials.gov/study/NCT05556343.

2. Olivotto I, Oreziak A, Barriales-Villa R, et al. Mavacamten for treatment of symptomatic obstructive hypertrophic cardiomyopathy (EXPLORER-HCM), Lancet, 2020.

3. Maron MS, Masri A, Nassif ME, et al. Aficamten in symptomatic obstructive hypertrophic cardiomyopathy, results from SEQUOIA-HCM, J Am Coll Cardiol, 2024.

4. Coats CJ, Maron MS, Abraham TP, et al. Dosing and safety profile of aficamten in symptomatic obstructive hypertrophic cardiomyopathy, SEQUOIA-HCM, J Am Heart Assoc, 2024.

5. Green EM, Wakimoto H, Anderson RL, et al. A small-molecule inhibitor of sarcomere contractility suppresses hypertrophic cardiomyopathy in mice, Science, 2016.

6. Ho CY, Day SM, Ashley EA, et al. Genotype and lifetime burden of disease in hypertrophic cardiomyopathy, insights from the SHaRe registry, Circulation, 2018.

7. ChEMBL. Delocamten (CHEMBL6068426). https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL6068426/.

8. Open Targets Platform. MYH7 to hypertrophic cardiomyopathy association (EFO_0000538). https://platform.opentargets.org/evidence/ENSG00000092054/EFO_0000538.