Background Previous investigations in live animals convincingly established that arterial gene transfer, while feasible, was compromised by a low transfection efficiency. More recent studies have shown that transfection efficiency may be substantially augmented by the use of recombinant adenoviral vectors. Most in vivo transfections reported to date, however, have used direct (operative) administration of the adenoviral vector. Clinical applications of arterial gene transfer (such as prevention of restenosis), however, would require local percutaneous delivery of the transgene. The present study was designed to extend in vivo intraoperative findings to percutaneous delivery systems and to assess whether gene transfer remains site specific.Methods and Results A recombinant, replication-defective adenovirus modified to include an expression cassette for nucleus-targeted beta -galactosidase was introduced into rabbit iliac arteries in vivo using either a double-balloon catheter (DBC, n=27) or a hydrogel-coated balloon catheter (HBC, n=27). Contralateral arteries--normal, endothelium- denuded, or sham-transfected with a control adenoviral vector--served as controls. beta -Galactosidase expression was assessed by X-Gal staining. Cell-transduction efficiency was measured by morphometric analysis. Polymerase chain reaction (PCR) and histochemistry were used to detect the presence and/or expression of viral DNA in remote organs. Transgene expression was detected in all cases (46 of 46) between 3 and 14 days after transfection but was in no case detectable 28 days after transfection. In the DBC group, transgene expression was limited to endothelial cells when the endothelium was left intact and to rare medial cells (<2.2%) when it had been removed. In contrast, HBC delivery resulted in transduction of up to 9.6% of medial smooth muscle cells (P=.0001). Optimized PCR and histochemistry failed to detect evidence of extra-arterial transfection except in a small number of cells (between 1 in 3 x 102and 1 in 3 x 10 sup 5 cells) in the livers of 2 animals in the DBC group.Conclusions (1) Efficient, adenovirus-mediated, arterial gene transfer to endothelial and/or smooth muscle cells is feasible by percutaneous, clinically applicable techniques. (2) Consistent transfection of medial smooth muscle cells may be achieved when the endothelial layer is abraded. (3) Medial transfection is more efficient when an HBC, rather than a DBC, is used. (4) Percutaneous delivery of the adenoviral vector via HBC results in site-specific arterial gene transfer. Very-low-level extra-arterial transfection may occur, however, when the DBC is used. (Circulation. 1994;90:1648-1656.)