Three deaths, three companies, one trial pathway: the regulatory fix came after the harm but before the disclosure
Three patients died in separate China-based gene and cell therapy trials. None was disclosed.
Days before the third surfaced, the FDA cleared an IND relying partly on the same trial-data type.
In three weeks this July and August, three patients died in China's gene and cell therapy trials. None of the deaths was announced by the company or hospital that ran the trial. All three came out only after journalists pushed for answers.
Days before the third death became public, the U.S. FDA cleared a new trial for the very same company. The FDA was leaning in part on data from the same kind of unmonitored trials now under fire.
What happened
Three patients died in three separate China-based trials for gene editing and cell therapy. All three trials ran under China's "investigator-initiated trial" system, called an IIT for short. An IIT is led by a hospital and its own doctors, not by a drug company running a formal regulatory submission.
None of the three deaths was announced by the company or hospital that ran the trial. All three came out only after outside pressure: two from investigative journalism, one from a news outlet's direct questions to the company involved.
Three deaths. Three companies. None was disclosed voluntarily.
The three deaths, side by side
- Death 1, March 2025. A 6-year-old girl died seven days after an experimental base-editing gene therapy at Shanghai Xinhua Hospital, led by neuroscientist Zilong Qiu of Shanghai Jiao Tong University. The therapy aimed to fix a single mutated DNA letter causing a rare, neurodevelopment-affecting mutation. Her family paid more than $860,000 privately for the treatment. The death was not disclosed until a Science and Retraction Watch investigation, published July 30, 2026, about 16 months later. The hospital was fined roughly $3,600. No patient compensation has been reported.
- Death 2, August 2025. A boy, one of four children ages 4 to 8 enrolled, died of acute respiratory distress syndrome (a severe, whole-body immune reaction) after HuidaGene Therapeutics' CRISPR gene-editing trial for Duchenne muscular dystrophy at Shanghai Children's Medical Center. The therapy used HuidaGene's own gene-editing tool, delivered through the body by an AAV virus-based carrier. The trial had no independent safety monitoring board watching for problems. The trial's public registry entry was marked "complete" in February 2026 with no mention of a death. It surfaced only after STAT News pressed the company for answers, reported around August 5-8, 2026, about 12 months later.
- Death 3, date not disclosed. A patient with systemic sclerosis died after RiboX Therapeutics' RXIM002. The disease hardens skin and internal organs. RXIM002 is an "in vivo" CAR-T therapy that builds CAR-T cells inside the patient's body rather than in a lab. RiboX confirmed the death to Endpoints News, which reported it August 18-19, 2026. The exact date the patient died has not been made public.
How China's investigator-initiated trials work
China runs two parallel tracks for testing new cell and gene therapies in people. One is the familiar path: a company files a formal application with the National Medical Products Administration, China's version of the FDA, before dosing a single patient.
The other is the IIT path: a hospital and its own doctors run the trial themselves, historically without the national regulator reviewing it first. Oversight was left to the hospital's own ethics committee, not a national regulator.
This second path is not a small side door. China's clinical trial registry had logged more than 40,000 investigator-initiated trial registrations by April 2026.
Companies increasingly use IITs as an early, faster proving ground. They get human safety and efficacy signals from a hospital-run study, then bring that data to the FDA or NMPA later to support a formal registration trial. That is exactly what RiboX did with RXIM002, and exactly the pattern behind both gene-editing deaths.
China has already moved to close some of this gap. State Council Order No. 818 took effect in May 2026, requiring independent safety monitoring boards and review committees for cell and gene therapy IITs.
It restricts them to hospitals certified to national trial standards. That is real, meaningful reform. But the timing is worth sitting with: it took effect after both gene-editing deaths had already happened, and before either one had been made public.
Why disclosure took so long
None of the three deaths was disclosed by the party that actually knew about it first. The first surfaced only because outside journalists went looking.
The second surfaced only after a news outlet directly pressed the company. It came only after the trial had already been marked "complete" in a public registry with no mention of what happened. The company's chief executive quietly left soon after presenting inconclusive data at a conference.
For the third, RiboX confirmed the death once asked. In no case did the death lead to a voluntary public statement.
That pattern, not any single death, is the real story. A therapy can fail safely and still be worth pursuing.
Bad outcomes happen in early human trials everywhere, including in the United States and Europe. What is different here is the gap between when something goes wrong and when the people relying on that trial's data actually find out. Those people include patients, regulators, investors, and future licensing partners.
The FDA leaned on this same kind of data days before the third death surfaced
Around August 8-9, 2026, the FDA cleared an IND application for RiboX Therapeutics for RXIM002. The platform is the same in vivo CAR-T technology, now targeting immune thrombocytopenia. The FDA granted two real concessions: faster dose escalation and simpler under-the-skin injection.
RiboX said both were granted because its China IIT data was accepted as solid enough to shape the U.S. trial's design from day one.
Ten days after the FDA's clearance, Endpoints News reported a death in a different RiboX trial.
The two trials are not the same study. But they share the same company, the same core technology, and the same trial system that the FDA had just leaned on to justify a faster, lighter U.S. path.
That raises the sharpest question in this story. How much visibility do regulators really have behind a "clean" IIT safety dataset when the company has no legal duty to disclose an event?
What this means for patients
None of this means gene and cell therapy research in China should stop, or that every IIT hides a problem. China runs a large share of the world's cell and gene therapy trials, and most of that work moves real science forward for patients with few other options.
The families in these three trials, including one that paid over $860,000 out of pocket, took real risk hoping for a real benefit. That hope is not the problem here.
The problem is clear. A patient or family weighing whether to join a similar trial today has no reliable way to learn what happened to the patients who came before them. A registry marked "complete" told families nothing.
A gap of 12 to 16 months between a death and its disclosure means anyone deciding today is deciding on old, incomplete information. That is a fixable problem, and China's new rule is a first step, not a finished one.
What this means for drug developers and investors
For anyone evaluating a China-sourced cell or gene therapy asset, this pattern changes what "clean safety data" should mean in diligence—whether for in-licensing, investment, or partnership. A dataset with no reported deaths is not the same as a dataset with no deaths.
Before trusting IIT-sourced safety data at face value, it is worth asking directly: was there an independent safety monitoring board? Is the public trial registry complete and current? Has the company disclosed every adverse event, or only the ones a journalist found?
These are not exotic questions. They are the same kind of question a good due-diligence process asks about any asset, made sharper by a system where disclosure is not guaranteed by law.
This is exactly the kind of risk that is cheap to check for before a term sheet and expensive to discover after one. An avoidable hurdle, in other words: a real cost that a few direct questions, asked early, could have avoided.
If you've solved a problem like this, Commercial & Strategy experts get matched to paid consulting work through GTC's Expert Network — join free.
The wider impact on the field
China's investigator-initiated trial system has become a genuine engine for global cell and gene therapy innovation, not a footnote. Companies worldwide now use it to generate early human data faster and cheaper than a Western Phase 1 trial allows. Then they bring that data to the FDA or European regulators to support faster development.
That model only works if the data behind it can be trusted. Three undisclosed deaths surfacing in three weeks, across three different companies and two different technology types, is the kind of pattern that tests whether it can be.
China's reform is a real answer, but it is untested. Independent safety boards and multidisciplinary review committees only work if they are actually independent and actually empowered to slow a trial down.
Whether that holds in practice, especially at hospitals eager to be first with a new therapy, is not yet known. For now, the honest read is that the rules improved before the world found out why they needed to.
What this means for a CGT program
Whether China's IIT pathway can still be trusted as a regulatory shortcut is now an open question industry-wide, not just a question about these three trials.
GTC analysis: that is a regulatory decision, not only a diligence one. A sponsor building a US or EU regulatory strategy on IIT-sourced China data should expect regulators to weigh disclosure completeness, not just data volume or trial size. Document that check inside the regulatory submission strategy itself, not as an informal diligence step that never reaches the filing.
Frequently asked questions
What happened in China's cell and gene therapy trials?
Three patients died in separate China-based investigator-initiated trials (IITs) for gene editing and cell therapy, and none of the deaths was disclosed by the company or hospital running the trial. All three surfaced within about three weeks in July and August 2026, each after outside pressure from journalists rather than a voluntary announcement.
How many patients have died, and from what therapies?
A 6-year-old girl died in March 2025, seven days after an experimental base-editing gene therapy at Shanghai Xinhua Hospital. A boy died in August 2025 of ARDS after HuidaGene Therapeutics' AAV-delivered CRISPR gene-editing trial for Duchenne muscular dystrophy. A patient with systemic sclerosis died after RiboX Therapeutics' in vivo CAR-T therapy, reported by Endpoints News in August 2026. The exact date of that death has not been made public.
Why weren't these deaths disclosed sooner?
The first death was disclosed about 16 months later, only after a joint Science and Retraction Watch investigation. The second was disclosed about 12 months later, only after STAT News pressed the company directly. The trial's own registry entry had been marked "complete" months earlier with no mention of a death. Neither hospital nor company made a voluntary public disclosure.
Has China changed its rules in response?
Yes, but the timing matters. State Council Order No. 818 took effect in May 2026. It requires independent safety monitoring boards and review committees for cell and gene therapy IITs, restricting them to NMPA-certified tertiary hospitals. That reform came after both gene-editing deaths had already happened, though before either was made public.
What does the FDA's reliance on China IIT data have to do with this?
Around August 8-9, 2026, the FDA cleared an IND for RiboX Therapeutics' RXIM002. The agency granted an accelerated dosing scheme partly because China IIT data was accepted as robust enough to shape the U.S. trial design. Within about ten days, a death in a different RiboX China IIT trial, using the same underlying platform, was disclosed.
AAV to AI, Edition 02
Get the weekly resource
Three signals across four published stories, with decision context to help CGT teams act earlier.
Free PDF, 40 pages.
Download Edition 02Sources
- Exclusive: Third death in China's popular but opaque trials revealed, this time in CAR-T — Endpoints News, August 2026
- Patient Death in China Gene-Editing Trial Raises Alarms Over Regulatory Oversight and Transparency — MedPath, summarizing the Science/Retraction Watch investigation
- Girl's undisclosed death in gene-editing trial raises red flags in booming Chinese industry — CNN, July 30, 2026
- Second China Gene-Editing Trial Death: HuidaGene — CASRAI, summarizing the STAT News investigation
- Once again, child dies in gene-editing trial in China, rekindling debate on transparency and safety — STAT News, August 5, 2026
- After 2 undisclosed gene therapy deaths, can China's alternative regulatory path still be trusted? — Fierce Biotech
- Two Deaths In China Reopen Gene Therapy Debates — Forbes, August 10, 2026
- China's New CGT Pathway Explained — Biotech Primer
- China's investigator-initiated trials in cell and gene therapy — VITA Journal (IIT registration statistics, State Council Order No. 818)
- RiboX Therapeutics Announces FDA IND Clearance for RXIM002 — BioSpace / RiboX press release, August 2026
Join the GTC Expert Network
Independent Experts
Independent experts in cell & gene therapy: share knowledge, help build a report, or contribute quietly. Build authority and get called for paid consulting work. Join free.
Early R&D Experts
Early R&D ExpertsContribute target validation, vector engineering, or transgene design expertise. Get matched to paid consulting work through content you help create.
Join FreeCMC Experts
CMC ExpertsContribute scale-up, tech transfer, or GMP manufacturing expertise. Get matched to paid consulting work through content you help create.
Join FreeRegulatory Affairs Experts
Regulatory Affairs ExpertsContribute IND/BLA strategy or regulatory liaison expertise. Get matched to paid consulting work through content you help create.
Join FreeCommercial & Strategy Experts
Commercial & Strategy ExpertsContribute due diligence, licensing, or market access expertise. Get matched to paid consulting work through content you help create.
Join FreeEnabling Technology Experts
Enabling Technology ExpertsContribute bioprocessing hardware, analytical tooling, or supply-chain expertise. Get matched to paid consulting work through content you help create.
Join FreeAI Transformation Experts
AI Transformation ExpertsContribute agentic workflow, LLM, or AI-driven vector design expertise. Get matched to paid consulting work through content you help create.
Join FreeIndustry Representatives
Industry professionals in cell & gene therapy: de-risk your pipeline with expert input on demand, exactly when your program needs it. Join free.
Drug Developers
Drug DevelopersFounders, CSOs, VPs of R&D
Practical guides spanning vector design through commercialization, plus direct access to the expert network that wrote them.
Join FreeInvestors / VCs
Investors / VCsVC Fund Managers, Family Offices, Angels
Due diligence frameworks and strategic intelligence for evaluating CGT assets, from experts who do this diligence for a living.
Join FreeTechnology Developers
Technology DevelopersFounders, Licensing & Commercial Leads
IP protection and platform strategy guides for enabling technologies, plus the network that built them.
Join FreeCDMOs / CROs
CDMOs / CROsBD, Operations & Technical Leads
Partner selection, vendor evaluation, and operational excellence guides, sourced from the CGT organisations that use them.
Join FreeVendors / Suppliers
Vendors / SuppliersCommercial & Technical Leads
Market intelligence and partnership strategies for supply chain companies serving cell and gene therapy.
Join FreeStrategy Consultants
Strategy ConsultantsConsultants, Market Access Advisors
Spot the problems before they cost your clients — join the network that shapes what GTC covers next.
Join FreeAI Transformation Partners
AI Transformation PartnersAI/ML Consultancies, Digital Transformation Firms
AI education, implementation, and consulting for operational excellence in cell and gene therapy — for companies building or buying AI capability.
Join FreeWhere the analysis comes from. Industry teams get expert input on demand; independent experts build authority and get called for paid consulting work.
See more of our work in Google Search
Google lets you pick the sites you want to see more often in Top Stories. Add Gene Therapy Consultancy to your list, and our daily gene therapy news will show up higher for you.