Three Children Dead. Twelve Months of Silence.
In July and August 2026, journalists at Science magazine and STAT News broke open three separate deaths in Chinese gene-editing and cell-therapy trials. In each case the company or hospital had stayed silent for months, sometimes longer than a year, while applying for or holding regulatory approval in the United States. The deaths were documented, fatal immune reactions in children and young patients, treated under a loophole that allows Chinese hospitals to run experiments without government oversight. The tally: three patients dead, two of them children. The longest gap between a death and its public disclosure was twelve months. One company announced its death only after journalists pressed it directly. Another had already secured FDA clearance for a US trial while the fatal adverse event from its China study remained unannounced.
"Mei": A Six-Year-Old Who Died in an Unapproved Trial
The first case came from a joint investigation by Science and Retraction Watch, published July 23, 2026. A six-year-old girl, given the pseudonym "Mei," died in March 2025 at Shanghai's Xinhua Hospital, affiliated with Shanghai Jiao Tong University. She had received an experimental base-editing treatment for Snijders Blok-Campeau syndrome, a rare condition caused by a single mutation in the CHD3 gene. Seven days after the infusion, she was dead from a severe immune reaction. Her family had contributed approximately $860,000 toward the cost of developing the individualised therapy. This was not a drug on a shelf. It was built for her alone.
The Ethics Committee Approved the Trial Before Reading the Toxicology Report
The pre-clinical red flags were present and documented. A toxicology report dated February 17, 2025 found that all four monkeys given the therapy, at both low and high doses, developed moderate to severe liver damage. The Xinhua Hospital ethics committee approved the single-patient human trial before reviewing that report. When the researchers later published associated preclinical work in Nature on February 18, 2026, the paper said nothing about Mei's death. Shanghai Jiao Tong University announced an internal review after Science's investigation appeared. As of late July 2026, no findings had been released.
Deaths disclosed voluntarily
0 of 3
Longest gap: death to public disclosure
12 months (HuidaGene)
Mei's family contribution to therapy development
$860,000
HG302 trial patients enrolled
4 boys aged 4 to 8
Monkey liver damage in pre-trial toxicology
4 of 4 at all doses
Independent data safety monitoring board in HG302
none
HuidaGene: The Boy Who Died and the Trial That Had No Watchdog
The second case involved a boy with Duchenne muscular dystrophy enrolled in a trial run by Shanghai startup HuidaGene Therapeutics. The therapy, HG302, used a proprietary CRISPR variant called hfCas12Max, delivered via an adeno-associated virus vector. The boy, the fourth and final patient enrolled, died in August 2025 from acute respiratory distress syndrome after a severe immune reaction to a high dose. HuidaGene did not acknowledge his death for twelve months, until STAT News asked. The trial had no independent data safety monitoring board. Three months before the boy died, HuidaGene's then-CEO Alvin Luk presented early data at the American Society for Gene and Cell Therapy's Presidential Symposium in New Orleans. He described the first two patients as showing a response. Neither Luk nor the company's CTO remain.
A System Designed to Make Transparency Optional
Under China's investigator-initiated study pathway, a hospital can run an experiment without submitting it to the national drug regulator, the NMPA. The trial does not need to appear in the national registry. There is no mandatory adverse-event reporting to a central body. Oversight sits entirely with the institution running the study. HuidaGene's trial registry entry was updated to "complete" in February 2026. The death was disclosed in August. By that point, the company had submitted peer-reviewed findings from the trial to a journal in January 2026. Whether those findings described the death is not publicly known. HuidaGene has said results will be disclosed upon publication.
"The thing that bothers me is the lack of transparency. Not just about the death but about everything that surrounded it.
"Expert quoted by STAT News (August 2026)
A Third Death. Then Congress Wrote to the FDA.
The third case came on August 9, 2026. RiboX Therapeutics disclosed that a patient with systemic sclerosis had died during an investigator-initiated trial of RXIM002 in China, an experimental circular RNA-based in vivo CAR-T therapy. The disclosure came one day after the FDA had already cleared RiboX's application to begin a US trial, using data from the China study as part of its submission package. On August 21, two Republican members of Congress wrote to the FDA asking why it was accepting data from Chinese investigator-initiated trials, given that those trials operate without independent safety oversight. Researchers have drawn parallels to Jesse Gelsinger, the eighteen-year-old who died in a University of Pennsylvania gene therapy trial in 1999. His death froze the entire field for nearly a decade.
The Journalism Is What Changed It
None of the three companies faced a legal obligation to make the deaths public before a journal or regulator compelled disclosure. That is how the system works. What changed it was Science publishing an exclusive, STAT News pressing a company until it answered, and Endpoints News finding a third case days later. The debate now is whether the FDA should require independent safety oversight in any foreign investigator-initiated trial whose data enters a US regulatory submission. The agency has not said publicly whether it will change its approach.
