Deramiocel Heart-Derived Cellular Therapy in Advanced Duchenne Muscular Dystrophy (HOPE-3): A Phase 3, Randomised, Double-Blind, Placebo-Controlled Trial

Major Points and Findings:

The study reported here is the largest and most rigorous cell therapy trial completed so far in Duchenne muscular dystrophy (DMD). It enrolled 106 boys and young men at 20 US hospitals and was randomised, double-blind and placebo-controlled. The product, deramiocel (previously CAP-1002), consists of cardiosphere-derived cells grown from donated human hearts. Deramiocel is not a mesenchymal stem cell product and is not used at our clinic, so its results cannot be carried over to other cell types.

The trial met its primary endpoint. Arm function declined more slowly on deramiocel than on placebo over 12 months. The key heart endpoint did not reach statistical significance. The treatment does not restore lost function or replace dystrophin. The manufacturer, Capricor Therapeutics, funded the study and took part in the analysis and writing.

Aim:

To test whether intravenous deramiocel every 3 months for a year slows the loss of arm and heart function in advanced DMD, and to confirm the much smaller HOPE-2 trial.

Methods:

The trial enrolled males aged 10 or older with genetically confirmed DMD of any mutation type. They were either non-ambulatory or “late-ambulatory” (more than 10 seconds to walk or run 10 metres), were on stable corticosteroids, and had a Performance of the Upper Limb (PUL 2.0) entry score of 2 to 6 and a total score of 40 or lower (normal is 42). Exclusion criteria were a left ventricular ejection fraction (LVEF) of 35% or lower, forced vital capacity below 30% of predicted, or severe elbow contractures. Of 139 screened, 106 were randomised, 54 to deramiocel and 52 to placebo. Median age was 15 (range 10 to 22). Only 16 (15%) could still walk. About a quarter were also taking exon-skipping drugs.

Donor hearts that were eligible for transplant but not used came from organ procurement organisations (donors aged 22 to 50, both sexes). Tissue pieces were cultured into cell clusters called cardiospheres, then expanded for five passages. Each lot was tested for viability, identity, purity, sterility and potency.

Participants received 150 million cells, or an identical-looking placebo, by outpatient intravenous infusion on day 1 and every 3 months for 12 months. Because earlier studies produced allergic-type reactions, everyone was pre-treated with high-dose oral corticosteroids plus H1 and H2 antihistamines.

The primary endpoint was the percentage change in total PUL 2.0 score at 12 months. The key secondary endpoint was the change in LVEF on cardiac MRI, read centrally by blinded assessors. Other endpoints included the mid-level (elbow) part of the PUL, heart scarring on MRI, and a video-scored test of eating ten bites of food.

Change to the primary analysis: The primary analysis was changed late. In July 2025 the FDA declined to approve deramiocel on HOPE-2 data alone. After a meeting with the FDA, protocol version 9.0 (September 2025) confirmed total PUL 2.0 as the primary endpoint and changed the analysis from absolute change to percentage change from baseline. The last infusion had been given in June 2025. The database was unblinded on 25 November 2025, after the statistical plan was final. The change was therefore made before anyone saw unblinded results, although dosing had already finished.

Results:

Arm function (primary endpoint, met): The difference in percentage change in total PUL 2.0 was 4.55 percentage points in favour of deramiocel (95% CI 0.47 to 8.63; p = 0.029). In absolute terms this corresponds to a 1.2-point advantage on a 42-point scale. The placebo group lost 2.72 points over the year, in line with published natural history. The authors describe the result as about a 54% slowing of decline. The prespecified sensitivity analysis using absolute change, the original method, gave p = 0.0503, just short of the usual threshold.

For elbow-level function, the mid-level PUL difference was 8.12 percentage points (p = 0.008), equal to 1.0 point. The video-scored eating test also favoured deramiocel (nominal p = 0.0176).

Heart function (key secondary endpoint, not met): Only 83 of the 106 had usable MRI scans at both time points. The ranked difference in LVEF change was 9.51 (95% CI -1.64 to 20.65; p = 0.0935), an absolute difference of 1.92 percentage points. Because this endpoint failed, the authors state that every p-value after it in the testing sequence is nominal and descriptive only. The following results fall under that restriction:

• Boys with existing cardiomyopathy (64 of 83): LVEF change +0.45 on deramiocel against -2.33 on placebo (nominal p = 0.0165).
• Heart scarring was measured in only 22 participants with paired scans. The difference was -3.01 affected heart segments (p = 0.022). Scar as a percentage of heart muscle did not reach significance (p = 0.0645).
• No difference in heart volumes, the cardiac composite test, troponin or CK-MB.

Baseline LVEF was unequal, at 54.4% on deramiocel against 59.6% on placebo.

Safety: 105 participants were dosed and 594 adverse events were recorded, 98% of them mild or moderate. No deaths occurred.

• Any adverse event: 50 of 53 (94.3%) on deramiocel, 43 of 52 (82.7%) on placebo.
• Hypersensitivity reactions: 22 of 53 (41.5%, 55 events) on deramiocel against 8 of 52 (15.4%, 18 events) on placebo, despite the pre-treatment.
• Events judged related to the product or the infusion: 44 (83.0%) against 19 (36.5%).
• Most common: headache 35.2%, cough 15.2%, fever 14.3%, nausea 12.4%, tachycardia 12.4%, mostly gone within 24 to 48 hours.
• Serious adverse events: 1 on deramiocel, 5 on placebo. Grade 3 or higher: 1 against 3 participants.
• Two serious events were treatment-related: a moderate infusion reaction on deramiocel (the boy continued without recurrence), and a grade 4 anaphylactic reaction in a placebo recipient needing adrenaline, presumably due to an ingredient present in both preparations.

Conclusions:

The authors conclude that deramiocel slows progression of DMD with a favourable safety profile. They note that the mild to moderate reactions mean it must be given under medical supervision in an infusion centre.

The limitations they list are a modest sample, 12 months of follow-up, wide variation between patients, more slow progressors on placebo (5 against 1), unequal baseline heart function, and MRI scans lost to poor image quality. Patient-reported outcomes are not yet published, and the smallest clinically important percentage change in PUL 2.0 has not been established. Longer follow-up is needed for durability, long-term safety and any effect on survival.

Background Information:

Cardiosphere-derived cells survive only 1 to 2 weeks after infusion. They do not become muscle or supply dystrophin. The proposed mechanism is the release of extracellular vesicles carrying RNA signals that shift macrophages and fibroblasts away from inflammation and scarring.

A 1.2-point difference on a 42-point scale is small in absolute terms. The authors argue that it is clinically relevant because one point at elbow level can decide whether a boy can still feed himself, the ability parents rank highest. The heart findings are encouraging, but under the trial’s own statistical rules they remain unconfirmed.