Background

Adeno-Associated Virus (AAV) vectors have been regarded as the safest gene delivery vehicles due to their characteristic of remaining in the nucleus as episomes without integrating into the host cell genome. The confidence that retrovirus- or lentivirus-based therapies could avoid the risk of oncogenic mutations highlighted in earlier studies has served as the foundation for the development of numerous therapeutics. Most successfully commercialized drugs, such as Novartis' spinal muscular atrophy treatment Zolgensma or Roche's hereditary retinal disease treatment Luxturna, are based on the AAV platform.

Particularly in the field of rare, incurable central nervous system (CNS) diseases where crossing the blood-brain barrier (BBB) is difficult, local administration methods—injecting vectors directly into the ventricles or the cisterna magna (ICM)—have been actively applied in clinical trials. This strategy aimed to maximize therapeutic efficiency and reduce systemic side effects by delivering high concentrations of the therapeutic agent to the lesion site. Although preclinical reports intermittently emerged from rodent or primate models suggesting that AAV genes could integrate into hepatocyte genomes and cause liver cancer, the possibility of tumorigenesis in the CNS, centered on non-dividing neurons, was considered extremely low. Furthermore, research tracking genomic mutations associated with long-term engraftment had not yet accumulated sufficiently at the institutional level.

Key Findings

According to a letter report published in the New England Journal of Medicine (NEJM), a neuroepithelial tumor induced by AAV vector insertion was identified in the brain tissue of a pediatric patient who received gene therapy. The patient was a 5-year-old boy with Hurler syndrome, the severe form of Mucopolysaccharidosis type I (MPS I), a congenital metabolic disorder. To supply the deficient alpha-L-iduronidase (IDUA) enzyme, he was administered RGX-111, an AAV serotype 9 (AAV9) vector-based candidate from RegenXbiotech, via the cisterna magna route.

Following treatment, the patient showed improved developmental milestones and neurological improvement; however, a brain tumor was discovered during a routine magnetic resonance imaging (MRI) follow-up conducted in the fourth year post-treatment. Medical staff immediately performed a craniotomy to achieve total tumor resection, and performed whole-genome sequencing and single-cell transcriptome analysis on the excised tumor tissue.

Precise analysis captured a clonal genomic insertion mutation where the entire recombinant AAV vector sequence had integrated into the PLAG1 proto-oncogene locus within tumor cells. The AAV-PLAG1 chimeric transcript was continuously produced due to the abnormal fusion of the vector's inverted terminal repeats (ITR) and promoter regions with the host gene. This chimeric transcript induced excessive expression of the PLAG1 protein by bypassing normal regulatory mechanisms, ultimately triggering the malignant transformation of brain cells. This is the first clinical case in which the molecular biological causal relationship—where AAV, previously considered a non-integrating vector, physically integrated into the human brain cell genome to induce cancer—has been clearly demonstrated.

Meaning and Prospects

The U.S. Food and Drug Administration (FDA) immediately issued a clinical hold on the same series of clinical trials using the cerebellar-tonsillar injection method shortly after confirming this case report. As it becomes evident that the possibility of therapeutic gene vectors randomly integrating into the patient's genome cannot be ruled out, there is a growing demand for re-verification of safety across the entire pipeline under development. There is growing persuasiveness within and outside regulatory authorities for extending the long-term follow-up observation period from around five years to more than 10 years, and for mandating periodic genomic testing.

On the other hand, there is no small measure of cautious opinion that it is premature to completely deny the utility of treatment itself. The infant survived tumor resection without significant neurological sequelae, fully preserving cognitive function by preventing the fatal brain damage that would have resulted from the genetic deficiency. Considering that thousands of patients worldwide have received AAV therapies, the frequency of tumor occurrence itself remains extremely low. The industry is seeking solutions through capsid engineering to reduce random genomic insertion frequency, gene expression control technologies to minimize residual vector amounts, and the advancement of delivery technologies to block off-target tissue penetration.