CRISPR therapy for Duchenne: a death in the HG302 trial

CRISPR therapy for Duchenne is meant to alter a patient's own gene rather than simply supply an additional gene copy. But in August 2025, one of four participants in the first human trial of the experimental therapy HG302 died after receiving the high dose. HuidaGene Therapeutics publicly disclosed the death nearly a year later.

The case matters because HG302 combines two powerful technologies. CRISPR changes DNA inside the patient's cells, while an adeno-associated virus, or AAV, carries the editing machinery into muscle. According to the information released so far, it is this delivery component — rather than demonstrated erroneous genome editing — that may have been central to the fatal complication.

Editing the dystrophin gene

Duchenne muscular dystrophy is caused by mutations in DMD, the gene encoding dystrophin. Dystrophin is a very large structural protein that helps muscle fibers withstand the mechanical stress of contraction. Without sufficient functional dystrophin, muscle fibers are repeatedly damaged and progressively lost.

HG302 is designed for a subset of patients rather than everyone with Duchenne. Its CRISPR/hfCas12Max editor targets a splice-donor site associated with exon 51 of DMD.

Splicing is the process by which cells assemble the final messenger RNA from pieces of an initial RNA transcript. By modifying a splice site, HG302 is intended to make the cell skip a particular exon and restore the gene's reading frame. The resulting dystrophin would still be shorter than normal but could retain useful function.

Unlike drugs that temporarily alter RNA splicing, editing the underlying DNA could in principle produce a durable effect.

Getting the editor into skeletal muscle throughout the body is another problem. HG302 packages its editing system in an AAV vector. These engineered viruses are widely used as gene-delivery vehicles because they can enter human cells without causing a conventional viral infection. But systemic treatment requires enormous numbers of viral particles, and the immune system can react strongly to them.

What happened in the trial

The MUSCLE study, NCT06594094, was the first human trial of HG302. It was an open-label, early-phase dose-escalation study in which safety was the primary outcome. Four boys with Duchenne were ultimately enrolled; the trial eligibility range was 4 to 8 years of age.

The first participants received the lower dose. HuidaGene subsequently reported target-site editing and preliminary functional changes in the first two participants, while reporting no serious adverse events in that cohort. Those observations were early signals from a tiny uncontrolled study and could not establish clinical efficacy.

The trial then moved to a higher-dose cohort. In August 2025, its final participant died following HG302 administration.

In a statement released on August 5, 2026, HuidaGene said the participant developed acute respiratory distress syndrome after high-dose systemic AAV administration, in the setting of severe complement and cytokine activation.

Acute respiratory distress syndrome, or ARDS, involves severe inflammatory injury to the lungs that prevents normal oxygen transfer into the blood. Complement is part of the innate immune system and normally helps defend the body against pathogens. If complement and inflammatory cytokines are activated excessively, however, the response itself can damage blood vessels, lungs and other organs.

HuidaGene says it conducted laboratory, immunological, pathological and post-mortem analyses. The other three participants did not develop the same severe syndrome and remain in long-term follow-up.

Was CRISPR responsible?

This distinction is crucial. The currently available evidence does not establish that the child died because CRISPR edited the wrong DNA sequence.

HuidaGene instead links the fatal syndrome to immune activation following high-dose systemic AAV administration. Serious immune and organ toxicities are already a recognized challenge for systemic AAV gene therapies, particularly when large quantities of vector must be delivered throughout the body.

HG302 had originally been promoted partly as a way to reduce this problem. Because the hfCas12Max editor was designed to work efficiently, its developers hoped to achieve genome editing with a lower viral-vector dose than some AAV therapies that deliver microdystrophin genes.

Lower dose, however, does not mean risk-free.

There is also a major limitation to what can currently be concluded. The detailed investigation of the death has not yet appeared in a peer-reviewed publication. HuidaGene says the findings were submitted for peer review in January 2026. Without those data, independent researchers cannot fully assess the administered dose, sequence of clinical events, laboratory findings, pathology or the proposed causal mechanism.

The second issue is transparency. The participant died in August 2025, but the company publicly disclosed the event on August 5, 2026. HuidaGene says the event was reported through the hospital's ethics and study-oversight processes within the applicable reporting period. That is different, however, from making the information publicly available to researchers, clinicians and families considering experimental therapies.

The HG302 case therefore does not demonstrate that CRISPR therapy as a whole is unsafe or ineffective. It illustrates a more fundamental challenge in genome medicine: making a precise molecular edit is only one part of the treatment.

A genome editor may be highly accurate, but it still has to reach billions of cells safely. In gene therapy, the route to the DNA can be as important as the edit itself.