Gene Therapy · asia
Six-Year-Old Girl Dies After Brain Gene Editing: The Regulatory Gaps Exposed by a Single-Patient Trial
The therapy, tailored to a rare CHD3 mutation, had shown potential in mouse and primate studies; now, an undisclosed human death, animal toxicity signals, and the family’s financial involvement are forcing the gene-editing field to reassess the safety thresholds for first-in-human trials.
An experimental therapy intended to directly correct a disease-causing gene in the brain might have become an important milestone for rare neurodevelopmental disorders. Instead, following the death of a six-year-old girl, it has become the subject of serious questions about research ethics and regulatory systems. According to a joint investigation by *Science* and Retraction Watch, the girl received treatment at Shanghai Xinhua Hospital in March 2025 and died seven days later from a severe immune reaction; the incident did not become widely known until more than a year later.
The girl had Snijders-Blok-Campeau syndrome, a rare disorder caused by variants in the CHD3 gene that may result in developmental, speech, and motor delays, as well as intellectual disability and other symptoms. A team led by neuroscientist Qiu Zilong designed an adenine base editor targeting her CHD3 p.R1025W variant. The editing components were carried in two sets of adeno-associated viral vectors, and trillions of viral vectors were injected into the cerebrospinal fluid in the hope that the tool would enter neurons and correct a single base.
The investigative report said that the girl developed severe inflammation and kidney damage after treatment. A report by the ethics committee of Shanghai Xinhua Hospital reportedly determined that the death was “definitely related to the treatment infusion.” However, the complete medical records, dose-selection decisions, and ethics-review documents have not been made public. The available account currently relies mainly on records the family provided to the media and on investigative reporting, making it impossible to independently reconstruct every medical detail.
More troubling were the safety warnings that had emerged before treatment. Independent experts interviewed by the media said that trials in four nonhuman primates had shown severe liver and kidney damage, and questioned whether the research team had adequately disclosed, explained, or ruled out these risks to the family. Although adeno-associated viruses are commonly used in gene therapy, they may trigger strong immune reactions at high doses. When treatment also requires delivering two types of vectors into the central nervous system, dose, tissue distribution, and systemic toxicity cannot be inferred solely from efficacy in mice.
The paper the research team published in *Nature* in February this year presented a more optimistic scientific picture: the novel base editor restored some CHD3 protein levels in mice carrying the human CHD3 variant and improved several behavioral abnormalities; after intrathecal injection of the dual AAV system, it was also broadly expressed in neurons in nonhuman primates. The paper reportedly did not mention the girl who had received the treatment and died, the family’s approximately $860,000 in funding for the therapy’s development, or fully present the disputed animal toxicity information. The family has requested retraction, but as of the publication of the relevant reports, the paper had not been retracted.
Transparency concerns also extend to clinical-trial registration and conflict-of-interest management. The relevant ClinicalTrials.gov record reportedly has not been updated since 2025. At the same time, having the participant’s family bear substantial development costs may entangle decisions about research recruitment, informed consent, and whether to proceed with treatment. This does not mean that family-funded personalized therapies are inherently improper, but it does require ethics-review bodies to establish stricter independent assessment, financial-disclosure, and trial-stopping mechanisms.
The central issue in this case is not whether therapies should be developed for individual patients with rare diseases. Rather, it reminds researchers that a first-in-human trial cannot be judged solely by whether editing is technically possible. When animal toxicity, irreversible administration to the nervous system, a child participant, and family funding are all present, the thresholds for safety evidence and informed consent should be higher. Regardless of whether the outcome is success or failure, serious adverse events must be promptly registered, disclosed, and subjected to independent review; otherwise, the entire field of precision gene therapy will bear the cost in lost trust.