By now we have all heard a lot about retroviral vectors and cancer, but other types of vectors have typically been left out of these discussions. However, results presented at the recent American Society of Gene Therapy meeting and just published in Science1 show that this is a subject that goes beyond retroviruses, hematopoietic cells, and gamma C transgenes. In this study, mice developed hepatocellular carcinomas (HCCs) that contained adenoassociated virus (AAV) vector proviruses at a specific chromosomal locus, raising important questions about the safety of in vivo gene therapy. The experiments were a collaboration among three laboratories: Mark Sands’s group, which generated the mice; the lab of Dan Miller, who isolated the vector: chromosome junctions; and my own (let’s just say I made sure Dan got the samples from Mark).

To summarize the findings, normal newborn mice injected intravenously with an AAV vector containing a cytomegalovirus (CMV) enhancer and human β-glucuronidase gene had an increased incidence of HCC (33–56% vs. 8.6% for untreated controls). Four integrated vector proviruses were identified by inverse PCR methods in the tumors from four mice, and all four mapped to a 6-kilobase region on mouse chromosome 12 (Figure 1), referred to here as the AAV-HCC locus. These specific proviruses were not detected in adjacent normal tissue, and the tumors had upregulated expression of neighboring chromosomal transcripts, including some that encode multiple small nucleolar RNAs (sno-RNAs) and microRNAs. Given that AAV