The short answer
Viruses spend their existence solving one problem: getting genetic material into cells. So we take a virus, remove the genes that make it dangerous and let it replicate, and put our cargo in instead. The main ones are AAV, which does not integrate into the genome, and lentivirus, which does.
Adeno-associated virus delivers cargo that persists mainly as an episome, giving durable expression in non-dividing tissue but diluting out in dividing cells; capacity is about 4.7 kb and pre-existing neutralising antibodies exclude many patients. Lentivirus integrates into the genome, giving durable expression through division at the cost of insertional-mutagenesis risk. Both provoke immune responses, and high systemic AAV doses have caused severe and fatal hepatotoxicity.
Where the risk actually lies
AAV is often described as safe because it does not integrate and causes little disease naturally. At the doses needed to reach a large tissue like muscle, that reassurance breaks down: severe liver injury, complement activation and deaths have occurred in gene-therapy trials at high systemic doses. This is a dose problem, not a virus problem, and it is the main reason muscle indications remain so difficult.
Sources
- National Human Genome Research Institute
Talking Glossary of Genomic Terms ↗