A 6-year-old died in a Chinese gene-editing trial. The lever was a hospital ethics board.

A six-year-old girl with a rare neurodevelopmental condition died in March 2025 after receiving an experimental, brain-directed gene-editing therapy at Shanghai Xinhua Hospital, affiliated with Shanghai Jiao Tong University School of Medicine. Her death was never publicly disclosed. The investigation that surfaced it, a joint inquiry by Science magazine and Retraction Watch, has put a different question on the table: not whether the underlying technology is safe, but how a hospital ethics committee was allowed to authorize a one-time gene-editing experiment in a child in the first place.

The answer, investigators found, is a Chinese regulatory shortcut that most outside the field have never heard of: the Investigator-Initiated Trial, or IIT, pathway. Under IIT rules, a hospital's own ethics board can greenlight a clinical study of an unapproved therapy, provided the institution is responsible for sponsoring it. No application to the National Medical Products Administration (NMPA), China's national drug regulator, is required. The same pathway is how Chinese academic hospitals push ambitious biotech projects forward quickly; it is also the loophole that made a brain-directed base-editing therapy in a child possible without any national-level review.

The therapy itself used base editing, a newer cousin of CRISPR that changes a single letter of DNA rather than cutting the strand. The preclinical work had been done in mice and reported in a Nature paper published earlier in 2026. That paper, the joint Science/Retraction Watch investigation found, did not mention the child's death. Earlier monkey studies, which independent experts say should have been cautionary, also showed liver and kidney issues that the published preclinical work did not foreground. Lead investigators Qiu Zilong of Shanghai Jiao Tong University and clinician Yongguo Yu did not respond to questions from the joint reporting team, according to the Straits Times wire summary of the investigation.

The family's role sharpens the picture. To support the therapy's development, the family contributed roughly $860,000 of their own savings and money borrowed from relatives, according to the joint investigation. Together, the family's request to withdraw the paper and the lead investigators' lack of response to questions show the line between a desperate family and a research subject is one the current Chinese system does not draw.

The IIT pathway becomes a global problem, not a Chinese one, because of how it interacts with the rest of the field. Base editing is now a worldwide clinical pipeline, with the first human therapies in the United States and Europe already in early-stage trials. The Chinese hospital pathway that authorized an experimental brain-directed therapy in a child has no direct counterpart at the FDA, where any first-in-human gene-editing study would require a formal investigational new drug filing. But the publication record is international. A Nature paper that omits a fatal adverse event in a related human study can be read, cited, and built upon by labs on every continent. Outside experts quoted in the joint investigation said the omission may be grounds for retraction.

The category of event the case forces into the open is not "CRISPR is dangerous." Base editing has produced its first approved human therapy, and pediatric gene editing for serious single-gene disorders is an active and legitimate area of clinical research. The category is the absence of an international rulebook for experimental gene editing in children. A hospital ethics board in Shanghai was the only gate that mattered, and the gate was not connected to any regulator with the standing to ask what the preclinical data did and did not show. A STAT+ write-up of the investigation frames the same case as an ethics and oversight story rather than a safety one, an indication that the responsible read of the case is converging around the regulatory gap rather than the technology.

What changes next depends on two actors the field already knows. The Nature paper's authors can be asked, formally, whether the preclinical record they published was the same record the hospital ethics board saw, and the editors of the journal can decide whether omission of a fatal adverse event in a related human study meets the threshold for retraction. China's NMPA, which has been silent so far, can clarify whether hospital-initiated gene-editing studies in children are within its existing jurisdiction or whether the rules need to be rewritten. The first of those is a question any journal can answer in the next 30 days. The second is a question the global gene-therapy field cannot answer for Beijing, but it can stop treating it as a domestic curiosity.