ADENO-ASSOCIATED VIRUS (AAV) VECTORS were first devel-oped 20 years ago (Carter, 2004) and an AAV vector was first administered to a human subject in November 1995 (Flotte et al., 2003). Initially, clinical development of AAV vectors was limited to a few indications for monogenic diseases such as cystic fibrosis and hemophilia B, but at least 11 different AAV vectors have now entered into clinical trials for in vivo delivery, although only one, for cystic fibrosis, has progressed to controlled phase II trials (Moss et al., 2004). The increased momentum reflects several events, but most importantly, the anticipated safety profile of AAV vectors based on preclinical studies was borne out in the early clinical trials and the overall safety profile of all the AAV vectors tested to date generally appears to be impressive. More than 200 subjects have been enrolled, some with repeated doses, with more than 8 years of follow-up in some patients. In parallel with the clinical development of AAV vectors, expanded studies of AAV biology have led to additional insights that will inform future development. The initial reasons for developing AAV vectors were that the parental AAV is not a pathogen and is able to persist in cells cultured in vitro. The lack of pathogenicity has been borne out thus far in the good safety profiles of AAV vectors. The wild-type AAV serotype 2 persists in cultured cells because it integrates at high frequency at a specific site on human chromosome 19 in a reaction that requires the AAV rep gene. AAV vectors do not contain the rep gene, but exhibit remarkable long-term expression in animal models, particularly in nondividing cells. There is an emerging consensus that the genomes of such AAV vectors generally persist as nonintegrated episomal entities that are mainly circular, duplex concatemers. AAV vectors integrate, if at all, at only low frequencies and this decreases the risk of insertional mutagenesis. For one AAV vector being developed as a human immunodeficiency virus (HIV) vaccine, the integration frequency was examined quantitatively after delivery to muscle in rodents and rabbits and was shown to be less than 3 10–7 (Munson et al., 2003; Schnepp et al., 2003a, b), which is significantly lower than the generally accepted rate of approximately 10–5 for spontaneous mutations in human genes (Cole and Skopek, 1994).

AAV vectors also do not contain viral genes that could elicit undesirable cellular immune responses and generally appear not to induce inflammatory responses. The primary host response that might impact the use of AAV vectors is a neutralizing antibody response against the viral capsid, either preexisting in the human population or induced by vector administration. Reinfection of humans by AAV is not prevented by serum neutralizing antibodies (Blacklow, 1988) but the route of delivery may be important. For instance, immune responses are greatly reduced after airway administration of AAV (Hernandez et al., 1999) and vector transduction occurred after repeated administrations to the lungs of rodents, rabbits, and rhesus macaques (Beck et al., 1999; Fischer et al., 2003; Sandalon et al., 2004). Whether immune responses might impact applications of AAV vectors may be difficult to assess in animal models (Flotte, 2004) and will most likely require studies in humans. However, in view of their long-term gene expression in vivo, particularly in terminally differentiated cells, which may be the preferred cellular targets, AAV vectors may be most suited for application in patients with chronic diseases, to whom the vector is delivered infrequently and in whom any potential host immune response to the AAV capsid protein may be less …