A six-year-old died in a brain gene-editing trial. A Shanghai medical school is investigating

In March 2025, a six-year-old girl in Shanghai received an experimental treatment that her doctors described as the world’s first brain-targeted gene-editing therapy. A week later she was dead. The case stayed out of public view for more than a year, until Science and the integrity watchdog Retraction Watch reported it on 23 July. Three days afterwards, Shanghai Jiao Tong University School of Medicine, whose affiliated Xinhua Hospital carried out the procedure, said it had formed a working group for a comprehensive investigation and promised ‘serious action’ if the facts warranted it.

The girl, referred to in the reports as Mei, had Snijders Blok-Campeau syndrome, a rare neurodevelopmental disorder caused by a mutation in the CHD3 gene. Fewer than three hundred cases are known worldwide. The condition slows language and intellectual development, can come with autistic behaviour and low muscle tone, but it is not progressive and does not usually shorten life. Mei’s parents worried she might never live independently, so they turned to Zilong Qiu, a senior researcher at the medical school’s Songjiang Research Institute and a well-known figure in Chinese efforts to treat rare childhood brain disorders with gene editing.

Qiu’s approach was base editing: the faulty letter in one gene is swapped for the correct one while the brain is still developing. Because the editing machinery is too large for a single virus to carry, his team used two adeno-associated virus (AAV) vectors and delivered them into Mei’s spinal fluid. Two vectors meant a higher dose of viruses overall.

The procedure failed within days. Mei developed a fever, her kidneys began to fail, her platelet count fell to dangerous levels and she was moved to intensive care. She died seven days after the infusion. The hospital’s ethics committee later concluded the death was linked to the treatment, caused by thrombotic microangiopathy, a complication that can follow a violent immune reaction to high doses of AAV.

What has unsettled researchers and bioethicists is less the failure itself, which any first-in-human trial can produce, than what the people who approved it knew. A toxicology report from a Chengdu contract research organization, dated 17 February 2025, recorded moderate to severe liver damage in four monkeys across all dose levels, and kidney damage in one animal in the high-dose group. The Xinhua ethics committee had approved the trial about a month before that report was completed. Several experts who reviewed the case said the panel apparently never saw the monkey data, and that the trial should not have gone ahead in its final form, or at least should have been scrutinized far more closely.

Money complicates the picture. Mei’s parents paid more than $860,000 towards developing the therapy, including preparing the viruses, and shared the documents and recordings with the reporters. Critics argue that a family financing its own child’s experimental treatment blurs the line between research and purchase, and makes fully informed consent hard to claim. The consent form explained human genes to a child by comparing them to a book with a spelling mistake in a single letter of DNA; bioethicists and medical law scholars have questioned whether that framing conveyed the real risk of a procedure that carried a known chance of death. Some have also asked why a non-life-threatening disorder was treated with a therapy whose safety was unproven, when research involving children is meant to follow the principle of minimal harm and controlled risk.

In February this year Qiu’s group published a paper in Nature describing preclinical work in which the base editor corrected Mei’s mutation in human cells and in the brains of mice and monkeys. The paper did not mention her death or the side effects seen in the primates. Her parents say they asked him to withdraw the paper after she died, and were upset that it did not acknowledge their financial contribution. Several gene therapy researchers have since questioned parts of the efficacy data. Qiu did not answer questions from Science; he told the South China Morning Post that the medical school’s leadership would provide responses.

The medical school’s statement said it attached great importance to the reports and firmly opposed medical research conducted in breach of ethics and regulation. Mei’s parents said they were willing to help the investigation and provide whatever information was needed.

I found one detail harder to shake than the rest. Mei’s parents paid more than eight hundred thousand dollars towards a treatment that no regulator, in the ordinary sense, had cleared. They were not investors and they were not sponsors. They were parents who had been told there might be a way to fix their child, and they reached for it with everything they had. I am not going to call that decision foolish, because I am not in their position, and neither is anyone who is judging it from outside.

What the money reveals is more uncomfortable. When a family has to fund the research that might save their child, the usual checks fall away. The family has every reason to want the treatment to happen, and the team has every reason to try it. The distance that is supposed to exist between a patient and a researcher, the careful disagreement about whether the risk is worth taking, largely disappears. A trial built on shared hope has already lost part of its judgement.

I do not think this case is an argument against gene editing. The field is young, the promise is real, and the world will keep needing it. The case is an argument about who watches the people who build these treatments. A child died in March 2025 and the fact stayed hidden for sixteen months. That tells you something about the pressure to be first, and about how little pressure exists to stop and ask whether a trial should be happening at all.

I hope the investigation is thorough and its findings are published in full. The family deserves that much, and so does every family who will one day be handed the same impossible choice.