## Did a CRISPR Base Editing Trial for a Nonfatal Disorder Kill a Six-Year-Old?

Yes. A six-year-old girl in China died in March 2025, seven days after receiving an experimental [base editing](https://synbiointel.com/glossary/base-editing) treatment for Snijders Blok-Campeau syndrome — a nonfatal neurodevelopmental condition. Her cause of death was thrombotic microangiopathy, which the treating hospital's ethics board judged "definitely related" to the experimental treatment. The case was not publicly disclosed when associated preclinical research appeared in *Nature* on 18 February 2026. It was finally reported by journalist Brendan Borrell in collaboration with Retraction Watch, published in *Science* on 23 July 2026.

The child had an R1025W mutation in *CHD3*. Researchers designed a CRISPR-Cas-derived base editor to convert the mutant adenine to guanine — a single-nucleotide correction that avoids double-strand DNA breaks. Because the editor construct was too large for a single vector, its instructions were split across two [AAV](https://synbiointel.com/glossary/aav)9 vectors administered intrathecally, requiring both vectors to co-transduce the same neurons to produce a functional editor. The child received trillions of viral particles. She was dead within a week.

---

## What the Preclinical Data Actually Showed

This is where institutional failure becomes unambiguous. In the preceding toxicology study in non-human primates, all four monkeys developed moderate-to-severe liver damage. One high-dose animal also showed kidney injury. These are not ambiguous signals — hepatotoxicity at that severity in all animals tested is a stop sign in any standard preclinical framework.

Critically, the hospital's ethics committee approved the single-patient trial before reviewing the final toxicology report. Experts cited in the *Science* investigation questioned three additional dimensions of the program: whether the split dual-AAV9 design could realistically edit enough neurons to provide clinical benefit; whether the primate evidence was adequate given the small cohort and severity of findings; and whether risks were properly communicated to the family during informed consent.

The underlying syndrome — Snijders Blok-Campeau — is nonfatal. That context matters enormously when weighing a risk profile that included universal hepatotoxicity in primates. The risk-benefit calculus for an experimental, high-dose intrathecal AAV intervention looks fundamentally different for a lethal pediatric cancer versus a neurodevelopmental disorder that does not itself kill.

---

## Transparency Failures Compound the Clinical Ones

The death occurred in March 2025. The associated preclinical research was published in *Nature* on 18 February 2026 — with no disclosure of the patient's death. That gap of nearly a year, followed by publication of supporting science without adverse event disclosure, is precisely the transparency failure that erodes public and regulatory trust in the entire gene editing field.

The Chinese Society of Gene Cell Therapy (CSGCT) responded after the *Science* disclosure, acknowledging the case underscored the need for robust preclinical evidence, rigorous ethical governance, transparency, and timely reporting of serious adverse events. The CSGCT also highlighted China's new national framework for investigator-initiated research — State Council Order No. 818 — as a structural response to strengthen oversight and patient protection. Whether that framework would have prevented this specific trial from proceeding, given that an ethics committee approved it without the final tox report, is the harder question.

---

## What This Means for the Base Editing Field

[Base editing](https://synbiointel.com/glossary/base-editing) as a modality is not implicated per se — the approach of making precise nucleotide changes without double-strand breaks remains one of the most promising tools in next-gen therapeutics. Programs targeting conditions with severe unmet need and appropriate preclinical packages continue to advance. The *Science* report separately notes that three boys with Duchenne muscular dystrophy showed broadly stable motor function one year after receiving GEN6050X, an experimental base-editing therapy — a different risk-benefit context for a progressive, fatal disease.

The specific failures here are procedural and institutional: a dual-AAV9 delivery architecture with unresolved co-transduction efficiency questions, universal primate hepatotoxicity that was not treated as a gating finding, ethics approval before final tox data was reviewed, and a nearly year-long delay in adverse event disclosure despite a peer-reviewed publication appearing in that window.

For the broader field, the systemic implications cut across delivery platform, indication selection, and ethics governance:

- **Dual-vector split designs** for large editors carry inherent co-delivery uncertainty. The efficiency assumption that enough neurons will receive both vectors simultaneously — especially in a pediatric patient via intrathecal route — deserves rigorous quantitative validation before human use.
- **Indication appropriateness** is not a peripheral consideration. Nonfatal conditions require a higher preclinical safety bar, not a lower one.
- **Ethics committee sequencing** — approving trials before final toxicology reports are available — is an institutional design flaw, not an individual error.

For enterprise buyers and investors evaluating base editing platforms, this case will accelerate regulatory scrutiny of AAV dose levels and will likely raise the preclinical safety bar for CNS delivery programs globally. Programs with clean, complete NHP tox packages in relevant models will be differentiated; those relying on small cohorts or incomplete data at IND will face harder questions.

---

## Key Takeaways

- A six-year-old died from thrombotic microangiopathy seven days after intrathecal dual-AAV9 base editing for Snijders Blok-Campeau syndrome, a **nonfatal** neurodevelopmental disorder.
- All four monkeys in the preceding tox study developed moderate-to-severe liver damage; one also showed kidney injury — findings that experts say should have halted the program.
- The hospital ethics committee approved the trial before reviewing the final toxicology report.
- The death occurred in March 2025 but was not disclosed when preclinical findings were published in *Nature* in February 2026; disclosure came via a *Science* investigation published 23 July 2026.
- China's CSGCT cited State Council Order No. 818 as a new oversight framework, but its adequacy given the sequence of failures here remains unresolved.
- The base editing modality itself is not the primary failure — delivery architecture, indication selection, preclinical gating criteria, and ethics governance are.

---

## Frequently Asked Questions

**What caused the six-year-old's death in the CRISPR base editing trial?**
The child died from thrombotic microangiopathy seven days after receiving a dual-AAV9 base editor intrathecally in March 2025. The hospital's ethics board judged the death "definitely related" to the experimental treatment.

**Why was the patient treated for a nonfatal disorder with an experimental gene editing therapy?**
Researchers targeted Snijders Blok-Campeau syndrome caused by an R1025W mutation in *CHD3*, aiming to correct the mutation using a CRISPR-Cas-derived base editor. Experts have since questioned whether the risk-benefit profile was appropriate for a non-life-threatening condition.

**What did the preclinical safety data show before the trial proceeded?**
All four non-human primates in the toxicology study developed moderate-to-severe liver damage; one high-dose animal also showed kidney injury. Experts say these findings should have prompted further investigation before human dosing.

**Why was there a split dual-AAV9 design, and why is that problematic?**
The base editor construct was too large to fit in a single AAV vector, so instructions were divided between two AAV9 vectors. Both vectors needed to reach the same neurons to produce a functional editor — a co-transduction challenge whose efficiency at therapeutic scale was questioned by experts cited in the *Science* investigation.

**When was the patient death publicly disclosed?**
The child died in March 2025. Associated preclinical research appeared in *Nature* on 18 February 2026 without disclosure of the death. The case was publicly reported following a *Science* investigation by Brendan Borrell in collaboration with Retraction Watch, published 23 July 2026 — roughly 16 months after the fatal event.

**What regulatory response has followed?**
The Chinese Society of Gene Cell Therapy cited State Council Order No. 818 as a new national framework to strengthen oversight of investigator-initiated research, requiring robust preclinical evidence, rigorous ethical governance, and timely adverse event reporting. The framework's practical enforcement remains to be evaluated against the institutional failures this case exposed.