Preparing buffers? Try Buffer Calculator
Child Dies After Experimental Gene-Editing Treatment in China
A six-year-old girl died in Shanghai in March 2025, seven days after receiving an experimental gene-editing treatment intended to correct a rare mutation affecting brain development, according to a joint investigation published by Science and Retraction Watch on July 23, 2026.
The girl, identified under the pseudonym “Mei,” was the sole participant in an early-stage clinical study conducted at Xinhua Hospital, affiliated with Shanghai Jiao Tong University School of Medicine. Her death was not publicly disclosed for more than a year, even though researchers later published closely related preclinical findings in Nature. Her family had contributed approximately $860,000 toward the development of the individualized treatment, the investigation reported.
The research effort was associated with Zilong Qiu—written as Qiu Zilong in Chinese name order—a neuroscientist affiliated with Shanghai Jiao Tong University’s Songjiang Research Institute and Xinhua Hospital. Qiu was a corresponding author of the subsequent Nature paper and was listed as one of the researchers who designed the study, provided experimental funding and wrote the manuscript.
A treatment designed for one child
Mei had Snijders Blok-Campeau syndrome, a rare neurodevelopmental disorder caused by pathogenic variants in the CHD3 gene. CHD3 encodes a protein involved in chromatin remodeling, a process that helps regulate gene activity during development.
The syndrome is associated with developmental delay, intellectual disability, delayed speech, low muscle tone and characteristic facial features. Its severity varies substantially between patients. The journalistic investigation described Mei’s condition as non-life-threatening, making the risk-benefit calculation especially important for a first-in-human treatment carrying substantial uncertainty.
The registered study, NCT06860672, was designed as an open-label, early Phase 1 trial with an estimated enrollment of one child between two and ten years old. Its stated purpose was to evaluate the safety, tolerability and preliminary effectiveness of a single intrathecal injection targeting the CHD3 c.3073C>T, p.Arg1025Trp mutation. The primary outcome was the incidence of treatment-related serious adverse events during the first 26 weeks.
How the experimental therapy worked
The treatment combined base editing with adeno-associated virus, or AAV, delivery.
Base editors are modified CRISPR-derived systems capable of changing one DNA letter into another without producing the double-strand DNA breaks associated with conventional CRISPR-Cas9 editing. In this case, the researchers developed an adenine base editor intended to reverse the disease-causing point mutation.
Because the complete editor was too large to fit inside one AAV particle, the system was divided between two AAV9 vectors. Both components had to reach the same cells and reassemble into a functional editor. The vectors were administered intrathecally, meaning they were injected into the cerebrospinal fluid surrounding the spinal cord and brain.
In the preclinical study later published in Nature, the researchers reported that the treatment restored CHD3 protein levels and improved behavioral abnormalities in a mouse model. They also reported that intrathecal administration of the dual-vector system produced broad brain transduction and editor reconstitution in nonhuman primates, findings they said supported the approach’s “translational feasibility.”
Death seven days after treatment
According to the Science and Retraction Watch investigation, Mei received the experimental treatment in late March 2025. She subsequently developed fever, falling platelet levels, loss of urine production and severe kidney injury.
Her condition deteriorated rapidly, and she died seven days after the infusion.
Documents reviewed by the investigators reportedly identified the fatal complication as thrombotic microangiopathy, or TMA, and classified it as related to the experimental treatment. TMA involves damage to small blood vessels and the formation of microscopic blood clots, potentially causing platelet depletion, destruction of red blood cells, kidney failure and injury to multiple organs.
Publicly available information does not demonstrate that Mei died because the editor made an unintended DNA change. The reported timing and clinical presentation instead point toward a severe immune or vascular reaction associated with the treatment platform, particularly the high quantity of AAV particles used to deliver the editor.
Experts commenting on the investigation emphasized that gene editing and viral-vector delivery are distinct sources of risk. An editor may create on-target, bystander or off-target genetic changes, while an AAV vector can provoke complement activation, inflammation, liver toxicity or TMA independently of whether the intended edit occurs. A complete causal assessment would therefore require the clinical record, dosing calculations, immune-monitoring data, laboratory findings and any available postmortem analysis.
Questions about warning signals in animal studies
One of the most serious issues raised by the investigation concerns the interpretation and disclosure of preclinical safety data.
According to the reporting and outside experts who reviewed the available materials, all four primates involved in relevant toxicity testing showed some degree of liver injury, and one reportedly exhibited kidney injury. Critics argued that these findings should have led to additional toxicology studies, a lower starting dose or a delay in treating a child.
The published Nature paper described two intrathecal dose levels in monkeys and emphasized brain delivery, editor reconstitution and the potential for clinical translation. Its methods state that the high-dose animals received two AAV9 preparations, each at a concentration of (1 \times 10^{14}) vector genomes per milliliter, with 1.2 milliliters administered from each vector preparation.
The central unanswered question is not merely whether an abnormal laboratory result occurred in an animal. Investigators must determine whether the complete toxicology data were available to the hospital ethics committee, how the human dose was selected, whether the animal findings were adequately communicated to the parents and whether the projected likelihood of meaningful neurological benefit justified the immediate risks.
The family’s $860,000 contribution
Mei’s parents reportedly used personal savings and contributions from relatives to provide approximately $860,000 toward the therapy’s development. The money was used to support activities including editor design, animal testing, manufacturing and clinical implementation, according to the investigation.
Patient-funded research is not inherently unethical, particularly in ultra-rare diseases for which conventional drug development may be commercially unrealistic. However, an arrangement in which the same family is simultaneously the principal funder, the legal decision-maker and the source of the trial’s only participant creates significant ethical concerns.
After investing a life-changing amount of money and years of hope, parents may find it psychologically difficult to withdraw. Researchers may also face financial, reputational or competitive pressure to proceed despite uncertain data. Such circumstances require unusually strong independent oversight and a consent process that clearly separates scientific possibility from evidence-based expectations.
The investigation alleges that the parents were not adequately informed about the severity of the animal toxicity findings or the possibility of a fatal immune reaction. Whether the consent documents and discussions met applicable legal and ethical standards remains a matter for formal investigation.
What the Nature paper disclosed—and omitted
The related preclinical paper, titled “In vivo base editing of Chd3 rescues behavioural abnormalities in mice,” was submitted to Nature in December 2024, before Mei received the treatment. It was accepted in January 2026 and published on February 18, 2026—nearly 11 months after her death.
The article reported the mouse and nonhuman-primate findings but did not mention that a child had already received a treatment based on the platform or that the child had died. It also did not identify the family’s reported financial contribution.
In its declarations, the paper stated that the authors had no competing interests. Its acknowledgments listed numerous government and institutional grants, while its author-contribution statement said that Qiu and two other authors “provided the experimental funds.”
Because the paper was formally a preclinical animal study, the fact that it did not contain a human case report does not by itself prove a violation of publication rules. The more consequential questions are whether the human treatment was sufficiently connected to the submitted research to require disclosure, whether the family’s funding represented a relevant financial relationship, and whether knowledge of the fatal event would have materially affected peer review or readers’ interpretation of the article’s claims about clinical translation.
Retraction Watch reported that the family has asked for the paper to be withdrawn and that outside experts have called for a full examination of the underlying data. Nature was reportedly unaware of the controversy before being contacted by the investigators and began reviewing the matter. No final editorial determination had been announced as of July 28.
A clinical-trial record that remained “Recruiting”
The public trial record also raised transparency concerns.
When the investigation was released, the ClinicalTrials.gov entry reportedly continued to identify the study as “Recruiting,” even though its planned enrollment was one participant and that participant had died. The registry showed a last update dated March 6, 2025, before the treatment and death.
Timely trial-registration updates are not merely administrative formalities. They allow patients, ethics boards, clinicians and other researchers to identify serious safety events and avoid exposing additional participants to an inadequately understood risk.
University investigation begins
Following publication of the joint investigation, Shanghai Jiao Tong University announced that it was examining the allegations. The university had not released final findings as of July 28, 2026.
Xinhua Hospital had previously paid a fine of approximately 24,000 yuan, or about $3,500, to local health authorities in September 2025, according to subsequent reporting. The publicly reported penalty did not initially disclose the complete circumstances of the child’s death, and Qiu faced no publicly announced sanction at that time.
The original Science and Retraction Watch investigation stated that Qiu, Xinhua Hospital and the university had not responded to multiple requests for comment before publication. The university’s later announcement of an investigation represents an institutional response, but it is not a finding that Qiu or any other individual committed misconduct.
What an independent investigation must establish
A credible inquiry will need to examine the complete preclinical dataset, including individual-animal results rather than selected averages; the scientific justification for the human dose; manufacturing and quality-control records for both AAV preparations; ethics-committee submissions and meeting records; the full informed-consent process; financial transactions between the family and researchers; treatment-emergent laboratory data; the management of the immune reaction; reporting to regulators and trial registries; and disclosures made to Nature during peer review.
It must also distinguish among institutional responsibility, clinical responsibility and individual scientific responsibility. Qiu was central to the underlying research program, but the registered trial listed Xinhua Hospital physician Yongguo Yu as sponsor-investigator and principal investigator. Treatment authorization, product release, dosing, clinical administration and emergency management may have involved multiple researchers, physicians, committees and institutions.
A test of transparency in individualized gene therapy
Individualized genome editing could eventually create treatment options for patients whose mutations are too rare to attract conventional drug-development investment. But an n-of-1 therapy does not justify an n-of-1 safety standard.
When the participant is a child, the underlying condition is not immediately fatal, the family is financing the project and no previous human has received the platform in the brain, the threshold for proceeding should be exceptionally high. Independent review and complete disclosure become more important—not less important—when a treatment is unprecedented.
Mei’s death does not establish that base editing as a field is inherently unsafe. Nor does it prove, by itself, that a particular researcher committed misconduct. It does show the potential consequences of moving a complex delivery platform into a child before uncertainties about dose, immune toxicity and clinical benefit have been resolved.
The scientific community now needs a transparent accounting of what was known before the treatment, what the parents were told, what happened after administration and why the death remained outside the public scientific record for more than a year.
References
- Science. “Exclusive: Death of girl in Chinese gene-editing trial was never made public.”
https://www.science.org/content/article/exclusive-death-girl-chinese-gene-editing-trial-was-never-made-public - Qiu, Zilong, et al. “In vivo base editing of Chd3 rescues behavioural abnormalities in mice.” Nature.
https://www.nature.com/articles/s41586-026-10113-6 - MedlinePlus Genetics. “CHD3 gene.”
https://medlineplus.gov/genetics/gene/chd3/ - ClinicalTrials.gov. “Study of AAV-Delivered Base Editing for CHD3-Related Developmental Disorders,” NCT06860672.
https://clinicaltrials.gov/study/NCT06860672 - Science Media Centre Spain. “Investigation alleges death of girl in China was covered up after she was given experimental gene therapy.”
https://sciencemediacentre.es/en/investigation-alleges-death-girl-china-was-covered-after-she-was-given-experimental-gene-therapy - Retraction Watch. “Exclusive: Death in gene-editing trial in China raises questions about Nature paper.”
https://retractionwatch.com/2026/07/23/exclusive-death-gene-editing-trial-china-nature-science-investigation/ - The Straits Times. “Death of girl in Chinese gene-editing trial kept secret: Report.”
https://www.straitstimes.com/world/death-of-girl-in-chinese-gene-editing-trial-kept-secret-report - MedPath. “AAV-CHD3-R1025W for Developmental Disorders,” NCT06860672.
https://trial.medpath.com/clinical-trial/fb295cfd3e8e8a99/nct06860672-aav-chd3-r1025w-developmental-disorders - South China Morning Post. “Chinese scientist faces probe after girl, 6, dies following experimental gene therapy.”
https://www.scmp.com/news/china/science/article/3361898/chinese-scientist-faces-probe-after-girl-6-dies-following-experimental-gene-therapy - CRISPR Medicine News. “Snijders Blok-Campeau Syndrome Clinical Trial,” NCT06860672.
https://crisprmedicinenews.com/clinical-trial/snijders-blok-campeau-syndrome-nct06860672/ - El País. “Científicos chinos ocultan la muerte de una niña tras recibir una terapia génica experimental y publican el estudio como un éxito.”
https://elpais.com/ciencia/2026-07-24/ocultada-en-china-la-muerte-de-una-nina-de-seis-anos-tras-recibir-una-terapia-genica-experimental.html
