Three patients, six months, and an average gain of 370 milliliters of lean muscle volume, roughly 0.8 pounds of tissue that, in every prior study of their disease, would have been disappearing.
Epicrispr Biotechnologies reported interim data on June 26, 2026 from its Phase 1/2 trial of EPI-321 in facioscapulohumeral muscular dystrophy (FSHD), a progressive genetic disease that slowly strips skeletal muscle from roughly 870,000 people worldwide, beginning with the face and shoulders and working downward until some patients cannot walk, and for which there are no approved disease-modifying therapies. Every clinical program that has reached a late-stage trial has measured the rate at which patients lose function and asked whether a drug can slow that decline.
EPI-321 did not slow the decline; it reversed the direction entirely.
What the Numbers Actually Show
As of a May 12, 2026 data cutoff, nine patients had received a single intravenous infusion of EPI-321 across two dose cohorts: six at the target dose of 2×1013 vector genomes per kilogram and three at double that level. The first three evaluable patients from Cohort 1, those with six months of follow-up, all demonstrated gains in lean muscle volume compared to their own baselines, as measured by whole-body MRI quantified by Springbok Analytics' AI platform, which segments and measures up to 140 individual muscles per scan. Individual gains ranged from approximately 0.5 to 1.3 pounds, with some individual muscles gaining 15% in lean volume. No serious adverse events were reported across all nine treated patients.
MRI findings were corroborated by reductions in a circulating cell-free DNA biomarker that tracks DUX4 pathway activity, the molecular cascade that drives FSHD's muscle destruction. This biomarker convergence matters: imaging data alone in a three-patient open-label trial would be interesting but fragile, whereas alignment between structural imaging and a mechanistic blood marker begins to suggest the therapy is acting on the disease's underlying biology, not just noise.
Context: The Fulcrum Phase 3 Failure
To grasp why gaining 0.8 pounds of muscle matters, you have to understand what happened twenty-one months earlier. In September 2024, Fulcrum Therapeutics reported that losmapimod, the most advanced FSHD therapy at the time, had failed its Phase 3 REACH trial across every endpoint. The drug was tested in 260 patients over 48 weeks and could not distinguish itself from placebo on the primary measure of shoulder and arm mobility (p=0.75) or any secondary measure. Fulcrum's stock fell nearly 70%, and the company suspended the program entirely.
Fulcrum used a small molecule to indirectly reduce DUX4 expression. Epicrispr uses a fundamentally different approach: a dead Cas12F protein, delivered by a single-dose AAV vector, that physically re-methylates the D4Z4 genetic locus where DUX4 lives, silencing the gene at the epigenetic level without cutting the DNA.
| Metric | Fulcrum losmapimod (Phase 3) | Epicrispr EPI-321 (Phase 1/2) |
|---|---|---|
| Patients evaluated | 260 | 3 (evaluable at 6 months) |
| Study design | Randomized, placebo-controlled | Open-label, single-arm |
| Duration | 48 weeks | 6 months |
| Dosing | Oral, twice daily | Single IV infusion |
| Mechanism | Small molecule (p38α/β inhibitor) | Epigenetic editing (dead Cas12F + AAV) |
| Muscle outcome | No significant difference vs placebo | Gain of ~370 mL lean muscle (all 3 patients) |
| Primary endpoint met? | No (p=0.75) | N/A (exploratory) |
This comparison is not apples to apples. Fulcrum ran a rigorous randomized controlled trial at full enrollment and failed definitively. Epicrispr has three patients in an open-label study with no placebo arm. A healthy person's muscle volume can fluctuate by several hundred milliliters over months due to activity changes, hydration, or simple measurement variability. But the direction of the signal matters: FSHD patients in natural history studies consistently lose muscle over time, and the fact that all three evaluable patients moved in the opposite direction, while the biomarker data tracked in the same direction, constitutes a genuine signal worth watching, not yet proof.
Epigenetic Editing Hits a Platform Inflection
Epicrispr is not the only company with an epigenetic editing therapy in humans. As of June 2026, three companies have active clinical programs, each targeting a different disease with a different delivery system:
Epicrispr Biotechnologies: EPI-321 for FSHD, delivered via AAV (adeno-associated virus). Its single-dose approach is designed for permanence: the AAV integrates the dead Cas12F machinery into muscle cells, which are long-lived. Its tradeoff is that AAV delivery triggers an immune response that likely prevents re-dosing.
nChroma Bio: CRMA-1001 for chronic hepatitis B, delivered via lipid nanoparticles, with its first patient dosed in January 2026. Founded by Jonathan Weissman, who co-developed the original dead Cas system. LNP delivery is potentially redosable, which matters for a liver-targeted application where cells turn over. Preclinical data showed a single injection lowered bad cholesterol by about 70% in monkeys via PCSK9 silencing.
Tune Therapeutics: TUNE-401 for chronic hepatitis B, using their TEMPO platform with LNP delivery. Presented Phase 1b/2a data at EASL in May 2026 showing 100% biomarker repression at dose levels 2 through 4, with durable suppression observed out to 17 months from a single dose, and now advancing to Phase 2 in late 2026. Co-founded by Fyodor Urnov, a pioneer of gene-editing technologies.
Three companies, three different diseases, two different delivery modalities, and all reporting biological activity in humans within the same twelve-month window: this is not one company's gamble but the moment when epigenetic editing crosses from a laboratory technique into a clinical platform.
The Timeline Math
CRISPR gene editing, the technology that cuts DNA, went from its foundational papers in 2012-2013 to first-patient dosing for sickle cell disease in 2019 to the first approved therapy (Casgevy, December 2023) in roughly eleven years from concept to approval, four years from first patient to label. That timeline benefited from decades of prior work on zinc-finger nucleases and gene therapy delivery systems.
Epigenetic editing's foundational dead Cas work was published in 2013 by Stanley Qi and Jonathan Weissman, the same year CRISPR gene editing papers appeared. But the clinical timeline lagged: Epicrispr dosed its first patient in mid-2025, nChroma in January 2026, and Tune had a Phase 1b running in late 2025. If the field follows a similar first-patient-to-approval arc as CRISPR cutting, the first epigenetic editing approvals would arrive around 2029 to 2030. That estimate assumes the biology cooperates, which is never a given, but the infrastructure is already proven because it borrows the same delivery vehicles (AAV and LNP) that gene therapy and mRNA vaccines have validated at scale.
The Honest Caveats
Three patients in an open-label design with no placebo arm, no blinding, and six months of follow-up while a company founded by the inventor of the technology reports data on its own drug. None of these are disqualifying, but they set a ceiling on how much confidence any honest reader should carry away from this data.
Several specific weaknesses deserve close attention. First, MRI-measured lean muscle volume is not an established regulatory endpoint for FSHD. The field is still debating what constitutes a clinically meaningful outcome, and the Fulcrum Phase 3 failure demonstrated that even validated functional measures like Reachable Workspace can behave unexpectedly in large trials. Second, AAV delivery carries inherent risks that scale with dose and patient weight, including liver toxicity and immune responses that have proved fatal in other gene therapy programs at high doses. Epicrispr reports no serious adverse events at nine patients, but the safety database is still very small. Third, durability beyond six months is unknown: epigenetic marks can be maintained by the cell's own machinery, but whether a single dose produces genuinely permanent silencing in human muscle tissue is an open question that only years of follow-up will resolve. Fourth, the external comparator from the ReSolve natural history study is informative but does not substitute for a within-trial placebo arm; patient selection effects could account for part or all of the observed signal.
Finally, the commercial comparison to Fulcrum obscures a structural difference in risk: Fulcrum tested a small molecule that needed chronic dosing and showed insufficient target engagement at the molecular level. Epicrispr's single-dose epigenetic approach is mechanistically distinct enough that Fulcrum's failure tells us relatively little about whether EPI-321 will succeed at scale.
What You Can Do
If you or a family member has FSHD, the trial is actively enrolling (ClinicalTrials.gov identifier NCT06907875) in the United States, New Zealand, and Australia. Cohort 2 at the higher dose is underway. The FSHD Society maintains a comprehensive registry of all active clinical trials and can help connect patients with appropriate studies.
If you are an investor or analyst evaluating the epigenetic editing space, the next catalysts to track are: Epicrispr's presentation at the World Muscle Society Annual Congress in September 2026, which should include additional patients and longer follow-up; Tune Therapeutics' Phase 2 initiation for TUNE-401 in late 2026; and nChroma's first efficacy data from the hepatitis B trial, timing not yet disclosed. The field will be meaningfully de-risked or challenged within the next twelve months.
If you work in biopharma or regulatory science, the Fulcrum Phase 3 dataset is now being transferred to the FSHD Society, and the post-mortem analysis by Jeffrey Statland at the University of Kansas contains valuable lessons about endpoint selection, patient heterogeneity, and DUX4 expression variability that will shape every subsequent FSHD trial, including Epicrispr's registrational study design.
The Bottom Line
For thirty years, the therapeutic paradigm in FSHD has been damage control: slow the loss, manage the symptoms, wait. EPI-321's six-month data, while preliminary and undersized, is the first clinical evidence that the direction of travel in this disease might be reversible. Three patients gaining muscle where the entire natural history of the disease says they should be losing it is not proof of a cure. But it is the first sign that editing the epigenome, rather than cutting the genome, can produce measurable changes in human disease biology. Two other companies are seeing signals in different diseases with the same class of technology. If even one of these programs reaches a registrational trial with the effect size intact, epigenetic editing will become the second major clinical platform derived from dead Cas, and possibly the safer one.