Safety of Experimental Gene and Cell Therapies Under Scrutiny: What Recent Trial Fatalities and Post-Marketing Experience Tell Us

Gene and cell therapies are moving rapidly from experimental research into clinical practice. Advances in viral-vector gene delivery, CRISPR-based gene editing, base editing, RNA technologies and in-vivo cell therapies are creating therapeutic possibilities for diseases that have historically had few treatment options.

Several of these approaches have already demonstrated meaningful clinical benefit. The U.S. Food and Drug Administration (FDA) now lists numerous licensed cellular and gene therapy products, including AAV-based therapies such as Luxturna (voretigene neparvovec), Zolgensma (onasemnogene abeparvovec), Hemgenix (etranacogene dezaparvovec), Roctavian (valoctocogene roxaparvovec), Elevidys (delandistrogene moxeparvovec) and Itvisma (onasemnogene abeparvovec).

But the rapid expansion of advanced therapies has also exposed difficult safety questions. Some platforms can trigger intense immune responses, organ toxicity or other serious complications, while the long-lasting biological effects of certain gene therapies mean that safety cannot necessarily be assessed fully during a conventional clinical-development program.

Those challenges have recently come into sharper focus in China, where several deaths associated with experimental gene and cell therapies conducted through investigator-initiated studies have been reported. Investigations by Science, Retraction Watch, STAT and other specialist outlets have raised questions about preclinical assessment, independent oversight, adverse-event reporting and transparency.

The cases should not be interpreted as evidence that investigator-initiated trials or advanced therapies are inherently unsafe. Nor should different technologies be grouped together simply because they fall under the broad category of gene or cell therapy.

Instead, they raise a broader regulatory question:

Do high-risk investigator-initiated trials involving experimental gene and cell therapies require enhanced, risk-proportionate safeguards that match the biological and clinical uncertainties of these technologies?

Answering that question requires looking beyond the recent Chinese cases and examining the wider clinical experience, particularly with AAV-based gene therapies.

What the Recent Chinese Cases Actually Show

The recently reported Chinese cases involve different experimental technologies and should not be treated as one clinical event.

One case involved a six-year-old girl treated at Xinhua Hospital in Shanghai with an experimental base-editing therapy for Snijders Blok-Campeau syndrome, a rare neurodevelopmental disorder. According to an investigation published by Science in collaboration with Retraction Watch, the therapy used AAV-based delivery into the cerebrospinal fluid. The child died seven days after treatment following a severe immune-related complication. The investigation also raised questions about how preclinical safety findings were evaluated before the trial proceeded.

A second case involved a boy with Duchenne muscular dystrophy who received HuidaGene Therapeutics’ experimental CRISPR-based therapy HG302 in a first-in-human investigator-initiated study. According to the company’s August 5, 2026 disclosure, the participant died in August 2025 after receiving a high dose of HG302 delivered using systemic AAV administration. HuidaGene said the patient developed acute respiratory distress syndrome in the setting of severe complement and cytokine activation.

STAT had investigated the case before the company publicly disclosed the fatal serious adverse event, making the timing and transparency surrounding the event a major part of the subsequent debate.

A third recently reported Chinese death involved an experimental in-vivo CAR-T therapy for systemic sclerosis. The therapy used circular RNA delivered through lipid nanoparticles rather than an AAV vector. It therefore belongs to the broader advanced-therapy safety discussion but should not be presented as another AAV-related fatality.

The cases consequently have different scientific characteristics. Their common feature is the use of highly experimental therapies in patients, accompanied by serious safety outcomes and questions about how adverse events and trial information were handled.

That distinction is important. The regulatory lesson is broader than AAV, while the AAV cases provide an especially useful body of evidence for examining how biological risk can be identified and managed.

AAV: A Clinically Validated Platform with Significant Safety Challenges

AAV-based gene therapy provides one of the clearest examples of the balance between therapeutic benefit and biological risk.

The platform has produced clinically meaningful therapies for inherited retinal disease, spinal muscular atrophy, hemophilia and Duchenne muscular dystrophy. Its clinical success demonstrates that AAV can deliver therapeutic genetic material effectively enough to produce meaningful benefits in selected diseases.

At the same time, AAV is not a single product or a uniform safety profile. Risk can vary according to the vector capsid, dose, route of administration, transgene, target tissue, manufacturing characteristics and patient factors.

Systemic administration is particularly important. When the therapeutic target is distributed throughout the body, intravenous delivery may require high vector doses. A substantial proportion of systemically administered AAV can reach the liver, making hepatotoxicity an important safety concern. Immune responses against the vector can also contribute to systemic inflammation and organ injury.

Pre-existing immunity can further complicate treatment. Some patients have antibodies against naturally occurring AAV serotypes, which can affect vector performance and influence treatment eligibility.

The resulting safety spectrum ranges from laboratory abnormalities and transient inflammatory reactions to severe hepatotoxicity, thrombocytopenia, thrombotic microangiopathy and other potentially life-threatening complications. Complement activation has also been implicated in serious immune-mediated toxicity following high-dose AAV administration.

This does not make AAV an inherently unsafe platform. It means that the benefit-risk profile depends heavily on how the vector is designed, dosed, administered and monitored.

What AAV Toxicity Has Taught the Field

AAV safety management begins before the first patient is dosed.

Preclinical studies are intended to characterize biodistribution, tissue exposure, dose-response relationships, immune activation and organ toxicity. Animal models cannot perfectly predict human outcomes, but significant findings can influence whether a program proceeds, how the dose is selected and what safety monitoring is incorporated into the clinical protocol.

This makes the interpretation of preclinical toxicity particularly important.

In the Chinese base-editing case, the Science investigation reported concerns surrounding preclinical findings and their consideration before the child received treatment. The reporting described serious liver toxicity signals in the relevant animal studies and raised questions about whether the available safety evidence had been adequately reviewed before the clinical study proceeded.

The broader lesson is not that every adverse animal finding should automatically prevent a clinical trial. Translational research inevitably involves uncertainty.

The critical question is whether known safety signals were adequately identified, independently evaluated and incorporated into the clinical risk-benefit assessment.

Once an advanced therapy enters human testing, safety management becomes continuous.

Key safety considerations for high-risk AAV programs

• Vector dose: higher systemic exposure can increase the risk of immune-mediated and organ-specific toxicity.
• Route of administration: systemic delivery can create substantially greater whole-body exposure than localized administration.
• Liver exposure: hepatic accumulation makes liver monitoring particularly important for many systemic AAV therapies.
• Immune activation: complement, cytokine and adaptive immune responses can contribute to severe complications.
• Patient factors: baseline liver function, immune status, age and disease characteristics can influence risk.
• Dose escalation: first-in-human studies require careful escalation and predefined criteria for pausing or stopping treatment.
• Clinical monitoring: early detection of liver injury, inflammatory syndromes and other serious adverse events is essential.
• Long-term follow-up: some gene therapies produce persistent biological effects and therefore require extended surveillance.

The exact safety-monitoring strategy should be tailored to the product rather than treated as a universal protocol.

Successful AAV Therapies Also Matter

A balanced safety discussion must recognize the successes of the platform.

Luxturna demonstrated the feasibility of treating inherited retinal dystrophy associated with biallelic RPE65 mutations. Zolgensma changed the treatment landscape for spinal muscular atrophy, while Hemgenix and Roctavian introduced one-time gene-therapy approaches for hemophilia B and hemophilia A. Itvisma subsequently expanded AAV-based treatment for spinal muscular atrophy through intrathecal administration for adults and children aged 2 years and older.

These products demonstrate that serious safety concerns do not invalidate the therapeutic platform.

Instead, the field has to work within a more realistic framework:

AAV gene therapy can provide substantial clinical benefit while still carrying clinically important risks that require active management.

That distinction is essential when interpreting recent fatalities.

Elevidys: Why the Safety Question Extends Beyond IITs

The experience with Elevidys provides an important comparison because it demonstrates that serious AAV toxicity can emerge outside the investigator-initiated setting.

In June 2025, the FDA reported two deaths from acute liver failure in non-ambulatory pediatric patients with Duchenne muscular dystrophy after treatment with Elevidys, with the agency stating that the deaths appeared related to treatment. The reports came from both clinical-trial and post-marketing data.

In July 2025, the FDA reported that it had received three reports of fatal acute liver failure following treatment with Sarepta’s AAVrh74 gene therapies and said the deaths appeared to have been caused by acute liver failure associated with the gene-therapy products.

The FDA subsequently added a boxed warning for acute serious liver injury and acute liver failure and revised the approved use of Elevidys.

The significance of this experience is not that the regulatory system simply “failed.” Rather, it demonstrates why safety evaluation continues after approval.

Clinical trials involve selected patient populations and relatively limited numbers of participants. Broader clinical use can expose a product to patients with different characteristics, disease severity and risk factors. Rare adverse events may therefore become more visible after approval.

Elevidys consequently provides a critical counterpoint to the Chinese IIT cases:

The underlying AAV safety challenge is not created by the IIT model.

IIT governance may influence how effectively risks are identified, monitored and communicated, but the biological risks of AAV can exist regardless of who sponsors the study.

IIT Governance: Should Oversight Depend on the Risk of the Therapy?

Investigator-initiated trials are an important part of biomedical research. They can generate evidence in rare diseases, test innovative approaches and explore scientific questions that may not immediately attract conventional industry-sponsored development.

The issue is therefore not the existence of IITs.

The more important question is whether the level of oversight should be determined partly by the risk of the intervention rather than by the administrative category of the study.

A first-in-human, high-dose systemic AAV therapy is fundamentally different from a low-risk investigator-led study involving an established intervention.

The same principle applies to experimental gene-editing and in-vivo cell-therapy platforms where human safety experience may be extremely limited.

A risk-proportionate model could therefore place additional requirements on the highest-risk studies.

Potential safeguards for high-risk advanced-therapy IITs

These should not necessarily become identical requirements for every IIT. A more practical approach would be to match the intensity of oversight to the degree of biological uncertainty and potential harm.

China’s Regulatory Framework Is Already Evolving

The recent Chinese cases also need to be viewed in the context of a changing regulatory environment.

China’s investigator-initiated research system has supported rapid development of innovative biomedical technologies. At the same time, concerns have emerged over consistency of oversight, preclinical requirements, manufacturing standards, safety monitoring and adverse-event transparency.

China has since strengthened its framework for emerging biomedical technologies. Order No. 818, issued in October 2025 and effective May 1, 2026, established a more formal framework for clinical research and the translation and application of new biomedical technologies. Associated rules address areas including institutional and personnel requirements, investigational-product preparation and quality control, clinical-study filing, preclinical safety evaluation, risk management and subject screening.

The timing matters. The two deaths highlighted by Science and STAT occurred before Order 818 took effect, while their public disclosure came later.

The cases therefore should not simply be used to argue that China’s current framework is inadequate. The more relevant question is whether the new rules will be implemented with sufficient consistency, independence and transparency to address the weaknesses exposed by earlier cases.

Regulation on paper is only one component of a safety system. Institutional accountability, independent review, accurate clinical documentation, rapid adverse-event reporting and transparency determine how effectively those rules work in practice.

Post-Marketing Safety: The Other Half of the Lifecycle

Safety oversight does not end with regulatory approval.

The FDA recommends long-term follow-up for certain gene-therapy products because their effects can be long-lasting and delayed adverse events may not become apparent during active clinical development. The agency also notes that post-licensure surveillance may be appropriate for products presenting long-term risks.

The FDA’s 2025 draft guidance on post-approval methods for cell and gene therapies further emphasizes the importance of collecting safety and efficacy information after approval because clinical-development programs often involve relatively small patient populations and these products may have long-lasting effects.

This creates a continuous safety pathway:

Preclinical toxicology → first-in-human studies → clinical development → regulatory review → post-marketing surveillance → long-term follow-up

Each stage answers different questions.

Preclinical studies help identify potential hazards.

Early clinical studies reveal whether those risks translate to humans.

Larger studies provide additional information about frequency and clinical management.

Regulatory review evaluates whether the benefits justify the identified risks.

Post-marketing surveillance can reveal uncommon or population-specific adverse events that were not apparent during development.

For advanced therapies, therefore, pharmacovigilance is not simply a final regulatory requirement. It is part of the product’s continuing safety assessment.

What Recent Events Should Teach the Field

The Chinese cases and broader AAV experience point to several lessons.

First, technology-specific risk must remain central to trial design. AAV, gene-editing and in-vivo cell therapies have different mechanisms and different safety concerns. They should not be regulated as though they were interchangeable.

Second, IIT status should not determine the level of safety oversight. The risk of the intervention should be a major determinant of monitoring requirements.

Third, high-risk first-in-human studies may warrant independent safety oversight. An independent monitoring structure can provide an additional layer of review when decisions about continuation involve investigators with substantial scientific or institutional interests in the program.

Fourth, serious adverse events must be reported promptly. Delayed public disclosure can limit the ability of regulators, investigators and other developers to learn from emerging safety signals, even where the event has been reported through formal institutional channels.

Fifth, preclinical findings must influence clinical decisions. Toxicology studies have value only when concerning findings are adequately evaluated and incorporated into dose selection, patient selection and monitoring plans.

Finally, safety surveillance must continue after approval. The Elevidys experience demonstrates why post-marketing monitoring can remain important even after a therapy has completed conventional development and received regulatory approval.

Stricter Oversight Does Not Have to Mean Slower Innovation

Advanced-therapy development depends on the ability to test new technologies rapidly enough to benefit patients with serious diseases. Excessive or poorly designed regulation could make academic innovation more difficult and increase development costs.

But stronger safety requirements do not necessarily have to eliminate flexibility.

A risk-proportionate model could allow lower-risk investigator-led research to proceed under conventional requirements while applying enhanced safeguards to first-in-human, high-dose or otherwise high-uncertainty therapies.

For example, a first-in-human systemic AAV therapy could require more intensive independent oversight than an investigator-led study involving an established gene-therapy platform with substantial human experience.

The objective should therefore not be to slow innovation for its own sake.

It should be to ensure that the speed of clinical translation does not exceed the ability of investigators and regulators to identify and respond to emerging safety risks.

Conclusion

Recent fatalities reported in Chinese investigator-initiated studies have brought renewed attention to the safety and governance of experimental gene and cell therapies. But the cases should not be interpreted as evidence that one technology, one country or one clinical-trial model is inherently unsafe.

They instead highlight a broader challenge.

Advanced therapies can produce substantial therapeutic benefits while also introducing biological risks that may be difficult to predict fully before human exposure. AAV-based gene therapies illustrate this tension particularly well. The platform has produced transformative treatments, yet clinical and post-marketing experience has also demonstrated potentially severe immune-mediated and organ-specific toxicities.

The Chinese cases have raised questions about how preclinical evidence, independent oversight and serious adverse events are evaluated and communicated in high-risk investigator-initiated research. The Elevidys experience demonstrates that serious AAV-related safety signals can also emerge within conventional development and after regulatory approval.

The appropriate response is therefore unlikely to be a blanket tightening of all IIT requirements.

Instead, high-risk gene and cell therapies may require risk-proportionate safeguards across the entire product lifecycle, including rigorous preclinical assessment, appropriate dose selection, independent safety monitoring, predefined stopping criteria, rapid serious-adverse-event reporting and long-term pharmacovigilance.

The central lesson is not that innovation should move more slowly.

It is that the systems designed to protect patients must evolve as quickly as the technologies being tested in them.

References and Source Framework

Primary regulatory and company sources

1. US. FDA. Approved Cellular and Gene Therapy Products.
2. US. FDA. FDA Investigating Deaths Due to Acute Liver Failure Following Treatment of Sarepta’s AAVrh74 Gene Therapies. July 18, 2025.
3. US. FDA. FDA Investigating Deaths Due to Acute Liver Failure in Non-Ambulatory Duchenne Muscular Dystrophy Patients Following ELEVIDYS. June 24, 2025.
4. US. FDA. FDA Takes Action on New Boxed Warning for Acute Serious Liver Injury and Acute Liver Failure Following Treatment with Elevidys. November 2025.
5. US. FDA. Long Term Follow-up After Administration of Human Gene Therapy Products. January 2020.
6. US. FDA. Post-approval Methods to Capture Safety and Efficacy Data for Cell and Gene Therapy Products. September 2025 draft guidance.
7. US. FDA. Itvisma approval and product information.
8. HuidaGene Therapeutics. Update on the HG302-01 First-in-Human Trial. August 5, 2026.
9. China Order No. 818 and associated implementation framework for new biomedical technologies.

 Investigative and independent reporting

10. Science/Retraction Watch. Investigation into the death of a child following an experimental gene-editing trial in China. July 2026.
11. Reporting on the HuidaGene HG302 trial death and transparency concerns. August 2026.
12. Fierce Biotech. Analysis of two undisclosed gene-therapy deaths and China’s evolving IIT framework. August 17, 2026.
13. Reporting on the third death involving an experimental in-vivo CAR-T study in China. August 2026.

Scientific background

14. Review of AAV vector toxicity, immune responses, dose-related toxicity, hepatic burden and complement-mediated complications.