A Single Injection Reversed Muscle Loss in a Disease With Zero Treatments. It Didn't Cut Any DNA.

Zero point four kilograms. Less than a pound of muscle, spread across the entire body, measured by MRI at six months. In the world of muscular dystrophy clinical trials, where the best outcomes are usually described as "slower decline" or "stabilization," those 400 grams represent something that has never happened before in a controlled study: patients with a degenerative muscle disease actually gained tissue.

Epicrispr Biotechnologies, a South San Francisco company that raised $68 million in its Series B round, announced the results at the International Research Congress on FSHD in late June 2026 in Chicago. The three patients in their Phase 1/2 trial had facioscapulohumeral muscular dystrophy, a genetic disorder that progressively destroys skeletal muscle, starting with the face and shoulders and working its way down. About 870,000 people worldwide live with FSHD. None of them has a disease-modifying drug.

The therapy, called EPI-321, does not work the way most gene therapies do, because it does not cut DNA at all.

The Mixer Board, Not the Scissors

Standard CRISPR gene editing sends a Cas9 enzyme to a precise location on a chromosome and slices through both strands of DNA, after which the cell repairs the break, and if the edit lands correctly the disease-causing sequence is disabled or corrected. This is what Vertex Pharmaceuticals' Casgevy does for sickle cell disease, what Zolgensma does for spinal muscular atrophy: cut the bad part out, fix the code, hope the scissors did not land somewhere they were not supposed to.

EPI-321 takes a fundamentally different approach, one that Fyodor Urnov, a biologist at the University of California, Berkeley, who co-founded the epigenetic editing company Tune Therapeutics, compared to using an audio mixing board instead of scissors: rather than cutting a track out of a song, you turn the volume down, and the DNA stays intact while the gene stays present but stops being expressed.

FSHD is peculiarly suited to this approach because it is, at root, an epigenetic disease in which the problem is not a broken gene but a missing set of instructions about whether to read one. Healthy people carry ten or more repeats of a DNA region called D4Z4 on chromosome 4, all heavily tagged with methyl groups, chemical markers that tell the cell to keep the gene called DUX4 silent. In people with FSHD, the D4Z4 region is truncated, meaning fewer repeats, fewer methyl groups, and without that chemical silence the cell starts producing DUX4 protein, which is toxic to muscle fibers, killing them slowly over years in a pattern that typically begins in the face and scapular muscles and progresses downward through the trunk to the legs until the patient cannot lift their arms above their head, cannot smile, cannot walk.

EPI-321 uses a deactivated version of a CRISPR enzyme called Cas12F, derived from archaea rather than bacteria, which at about 500 amino acids is less than half the size of standard Cas9's 1,300. It binds to a target sequence upstream of D4Z4, recruits methylation machinery to restore the missing methyl tags, and DUX4 goes quiet while the muscle stops being poisoned.

What the Data Actually Shows

Three adults with genetically confirmed FSHD received a single intravenous dose of EPI-321 and were evaluated by whole-body MRI at six months. All three showed a statistically significant increase in lean muscle mass, averaging approximately 0.4 kg. "We were startled," CEO Amber Salzman told Scientific American. "That's the first time we do an MRI, and the patients were actually gaining muscle mass."

That gain exists against a specific backdrop, one constructed from two years of prospective data collection across multiple countries. Epicrispr's external comparator is the ReSolve natural history study, a prospective, multinational dataset tracking 240 symptomatic FSHD patients, of whom 171 met the enrollment criteria for EPI-321's trial, creating a well-matched cohort in which lean muscle does not hold but declines.

Quantifying the rate of decline is complicated by FSHD's variable progression. MRI studies published in the Journal of Neurology (2025) measured fat fraction increases of 2.0 ± 0.6 percent per year at the thigh and 1.9 ± 0.7 percent per year at the leg across 30 patients. A longitudinal Nature Communications Medicine study found an overall yearly fat fraction increase of 1.3 ± 1.6 percent across all lower extremity muscles, with the fastest-progressing muscles (those with 30 to 40 percent baseline fat) deteriorating at 4.9 ± 2.7 percent per year. Fat replacing muscle is the disease's signature, and it does not pause.

Reversal vs. Stabilization: A New Category

No approved gene therapy for a muscular dystrophy has demonstrated net muscle mass gain. Consider what the existing treatments actually measure:

Therapy	Disease	Primary Endpoint	What It Measures	Price
Zolgensma	SMA	Motor milestone achievement	Can the infant sit, stand, walk?	$2.1M
Elevidys	DMD	Micro-dystrophin expression	Is the protein present?	$3.2M
Casgevy	SCD/TDT	Transfusion independence / VOC-free	Does the patient still need blood?	$2.2M
EPI-321	FSHD	Lean muscle mass (MRI)	Did the muscle grow?	TBD

Zolgensma measures what babies can do that they could not do before, Elevidys measures whether a therapeutic protein shows up in muscle biopsies, and Casgevy measures whether bone marrow produces enough functional hemoglobin to end transfusion dependence. These are legitimate and meaningful endpoints, every one of them hard-won and clinically significant, but none asks the question EPI-321 data is answering: did the patient end up with more muscle than they started with?

If EPI-321's results hold across the full 12-patient cohort, they would establish a new endpoint category for muscular dystrophy trials. Every drug that follows would face the question: does your therapy reverse, or does it merely slow the decline?

The Treatment Experience Gap

Beyond efficacy endpoints, the patient experience of epigenetic editing differs starkly from DNA-cutting gene therapy in ways that reshape the entire treatment economics conversation. Casgevy, the only approved CRISPR therapy, requires myeloablative conditioning, a chemotherapy regimen that destroys the patient's bone marrow to make room for edited cells. The treatment protocol unfolds over months: harvest blood stem cells, edit them ex vivo, administer busulfan conditioning (which carries risks of veno-occlusive disease, infertility, and secondary malignancies), then reinfuse the edited cells, all of which keeps patients hospitalized for approximately six weeks.

EPI-321 is a single intravenous infusion with no cell harvest, no chemo, no weeks-long hospitalization. The therapy is delivered by an adeno-associated virus (AAV) vector in an outpatient or short-stay setting, and the patient goes home.

This difference matters beyond convenience, and a manufacturing cost analysis by Parexel illustrates why. A 200-liter AAV batch costs approximately $2 million to produce under good manufacturing practices, yielding roughly 200 doses at a manufacturing cost of about $10,000 each, which means manufacturing accounts for just 0.235 percent of the final sticker price while the remainder recovers clinical development costs. At typical AAV gene therapy pricing of $1.5 million to $3.5 million per dose, an outpatient EPI-321 infusion would carry a total cost of care of approximately $1.5 million to $2 million including the drug and infusion. Casgevy's total cost of care, by contrast, runs approximately $2.5 million when hospitalization, conditioning, and cell processing are included alongside the $2.2 million list price.

Epigenetic editing delivers a simpler procedure at comparable or lower total cost while eliminating the most dangerous step in the existing gene therapy playbook, and that combination of reduced risk and reduced complexity could prove more important than any single efficacy number.

Seven Companies, Five Diseases, One Year

Epicrispr is not alone in the space. 2026 has become the year epigenetic editing moved from animal models into human patients across multiple indications.

Company	Target	Status (Mid-2026)	Key Result
Epicrispr	FSHD (DUX4 silencing)	Phase 1/2, 6+ dosed	+0.4 kg muscle in 3 patients
Tune Therapeutics	Hepatitis B	Human data (EASL, May 2026)	Viral markers undetectable
nChroma Bio	Hepatitis B	Phase 1 dosing (Jan 2026)	Pending
nChroma Bio	Cholesterol (PCSK9)	Preclinical (primates)	70% LDL reduction in monkeys
Scribe Therapeutics	Cholesterol (PCSK9)	Preclinical (ELXR platform)	Pending
Epigenic Therapeutics	Hep B + cholesterol	Preclinical	Pending
General Control	Age-related diseases	Early R&D	Undisclosed

Tune Therapeutics' hepatitis B data is particularly striking. At the European Association for the Study of the Liver Congress in Barcelona in late May, the company showed that its epigenetic silencer drove hepatitis B RNA intermediates and viral proteins to undetectable levels in some recipients. Hepatitis B, which chronically infects an estimated 240 million people globally, has resisted cure because fragments of the virus integrate into the host genome and produce proteins that suppress immune clearance. Epigenetic silencing offers a path that existing antivirals cannot: muting both the free virus and the integrated copies without cutting the patient's DNA.

Between Epicrispr's FSHD results and Tune's hepatitis B data, 2026 is the first year in which two separate epigenetic editing programs reported meaningful clinical efficacy in humans. That does not mean the field is mature. It means the field exists.

The Strongest Case Against

Epigenetic edits may not be permanent, and the biology of why reveals a challenge no amount of clever engineering can fully escape. The cell's own methylation maintenance machinery (DNMT1) copies methyl marks when DNA replicates, which is why researchers believe epigenetic edits can persist through cell division, but muscle tissue is complex in ways that test this assumption. Myonuclei, the nuclei within mature muscle fibers, are post-mitotic and turn over slowly, with estimated half-lives of 10 to 15 years, and if the AAV-delivered editor successfully methylates D4Z4 in existing myonuclei those edits should last for years.

Satellite cells are the problem, and they may prove to be the fatal one. These muscle stem cells activate in response to injury, fusing with existing fibers or forming new ones, and if the AAV does not transduce satellite cells then every newly incorporated nucleus enters the fiber without the therapeutic methylation, gradually producing DUX4 again and eroding the treatment effect over a timeline nobody can yet predict. This is the same biological challenge that limits Zolgensma in SMA: as children grow, un-transduced motor neurons may emerge, and efficacy may wane.

There is also a measurement question that could deflate the entire result. The 0.4 kg gain could reflect reduced inflammation and edema rather than genuine muscle fiber regeneration, because FSHD muscles in active disease phases show elevated water content on MRI (measured by water-T2 signal) and resolving inflammation could change how lean tissue appears without representing true structural recovery. Only biopsy data, or functional outcomes measured over substantially longer follow-up periods, will distinguish regeneration from reduced pathology.

What This Analysis Cannot Determine

Three patients in an open-label trial with an external comparator is a dangerously small dataset, and the history of drug development is full of promising Phase 1 signals that dissolved in larger trials. No placebo control exists, the ReSolve natural history cohort was not enrolled concurrently, and FSHD progresses at wildly different rates between patients (fat fraction change ranged from -0.2 to 4.5 percent per year in published studies), meaning three individuals who happen to gain muscle could represent normal variation rather than treatment effect. Cost-of-care projections for EPI-321 are speculative because the drug has no price, AAV manufacturing costs at commercial scale remain estimates from industry consultants rather than audited figures, the epigenetic editing pipeline table reflects company announcements rather than peer-reviewed data, companies in early clinical stages frequently fail, and durability of epigenetic edits in human muscle has never been measured beyond six months.

The Bottom Line

For 870,000 people with FSHD, there is no approved drug that slows the disease, stabilizes it, or reverses it, and the standard of care remains physical therapy and adaptive equipment. EPI-321's six-month data, from just three patients, suggests that a one-time intravenous injection can add lean muscle to a body that is built to lose it, doing so without cutting DNA, without destroying bone marrow, and without a six-week hospital stay.

If the full 12-patient cohort confirms these findings and the effect proves durable, the implications extend well beyond FSHD. Epigenetic editing becomes a validated therapeutic modality for diseases driven by gene misexpression, which is a vastly larger category than diseases caused by specific DNA mutations. Huntington's disease, certain forms of Parkinson's, autosomal dominant retinitis pigmentosa, and hundreds of conditions caused by genes that are too active rather than absent all become addressable by turning the volume down instead of cutting the wire. If you are a patient with FSHD today, the actionable information is that Epicrispr's trial (NCT identifier available at clinicaltrials.gov) is still enrolling, with sites in the United States, New Zealand, and Australia. If you are a researcher or clinician, the question this data forces is whether "stabilization" should remain the gold standard endpoint for muscular dystrophy when "reversal" may be possible. If you are a gene therapy investor, watch the twelve-month MRI data. Everything rides on whether 0.4 kg at six months becomes 0.8 kg at twelve, stays flat, or fades.

Sources

Scientific American (June 29, 2026). Gene-editing startups are using CRISPR to treat diseases. scientificamerican.com
Epicrispr Biotechnologies (January 8, 2026). Early Clinical Activity and Favorable Safety Profile in First-in-Human Epigenetic Editing Study for FSHD. J.P. Morgan Healthcare Conference. businesswire.com
Epicrispr Biotechnologies (March 25, 2025). $68 Million Series B. businesswire.com
Maranzano, J. et al. (2025). Natural history of facioscapulohumeral muscular dystrophy evaluated by multiparametric quantitative MRI. J Neurol. PMID: 40172709. doi.org
Dahlqvist, J.R. et al. (2021). Whole-muscle fat analysis identifies distal muscle end as disease initiation site in FSHD. Communications Medicine. nature.com
van der Wal, E. et al. (2023). 5-year natural history cohort in FSHD. PubMed
Parexel. Gene therapy: manufacturing costs and pricing analysis. parexel.com
Reuters (June 25, 2026). US prices for new drugs fell in 2025. reuters.com
Eureka/PatSnap (May 2026). AAV Gene Therapy Competitive Landscape Analysis 2026. eureka.patsnap.com