A child died in a Chinese gene-editing trial. The company went quiet for 15 months.

A 6-year-old girl died during a CRISPR gene-editing trial in Shanghai run by HuidaGene, according to Live Science's reporting on the parallel STAT and Science investigations. The child's disease was Duchenne muscular dystrophy — a fatal, mostly-boys muscle-wasting disorder with very limited treatments — though CRISPR Medicine News has separately characterized the fatal arm as "brain-directed," and the two framings have not been reconciled. Every outlet agrees on the timing: the company did not announce the death for more than a year.

HuidaGene had been presenting its first two-patient data at the ASGCT presidential symposium in New Orleans before the death became public. The early data, the STAT+ investigation found, "was not impressive and it was unclear the therapy worked at all." The company used the same stage to announce a higher-dose cohort. Then it stopped communicating.

For the next 15 months, HuidaGene issued no press releases. The registry listing for the trial was updated in February 2026 to mark the study "complete," according to Science's investigation, with no outcomes posted for the remaining enrolled patients. Two senior executives left during that window: CEO Alvin Luk, a U.S.-based gene-therapy executive, and CTO TJ Cradick, who had been at the company less than a year. Both are former employees per the STAT+ lede; their current roles have not been publicly confirmed.

The trial did not run through China's central drug regulator. In China, gene-therapy studies can run as investigator-initiated studies (IITs), approved by a hospital ethics committee rather than the National Medical Products Administration (NMPA). Under that pathway, there is no centralized rule that forces a serious adverse event or a death to be made public. The hospital committee that greenlit the study is the same body that would have received the death report, and the only external check on its handling is journalism.

Retraction Watch reports that two parallel investigations are now open: one into the associated Nature paper (s41586-026-10113-6), which is preclinical work, not a clinical trial readout, and a second launched by the Chinese university affiliated with the principal investigator. Neither has reported findings. Foreign Policy frames the delay as a cover-up and documents the political backlash inside China. Both point to the same gap: no one outside the hospital and the company knew a child had died until reporters started asking.

HuidaGene's statement on Wednesday, Aug. 5, came only after what STAT describes as "repeated questions" from the newsroom. The company did not announce the death on its own; it responded to inquiries. The contents of that statement have not been independently verified by the reporting outlets, and Science's four-takeaway investigation treats the same-day response as one data point, not resolution.

CRISPR itself carries risk, and gene-therapy trials in the U.S. and Europe have also produced deaths. None of those deaths, however, ran through a pathway that structurally exempts the sponsor from public disclosure. Western regulators require adverse-event reporting; China's IIT pathway, for the studies that use it, does not. The distinction is not about the safety of the molecule. It is about who is required to tell the next family, the next regulator, and the next reader that something went wrong.

The next Chinese hospital-run gene-therapy study that enrolls a child will face the same structural question this case raises: which body, under which rule, is supposed to announce a death, and what happens to the trial if that body chooses not to? Until the IIT pathway is paired with a public adverse-event rule, or until families and investors treat IIT-only studies as a separate risk class from NMPA-reviewed trials, the silence is a feature of the system that produced it.