Successful gene therapy for DMD will require widespread muscle transduction using a replacement gene, presumably a mini- or micro-dystrophin or a surrogate gene. [20, 26] Success in the mdx mouse with tail vein delivery is notable, [9, 10] but key issues for translation to clinical trials have been wanting. In this study we focused on a vascular delivery system that can be adapted for a clinical protocol in an effort to achieve a clinically meaningful outcome, providing safe passage for the virus, and using a dose within the manufacturing constraints for rAAV vectors. We reasoned that a regional vascular delivery approach to the lower limbs through the femoral artery could potentially meet these expectations. This is similar in concept to local versus systemic gene transfer to the liver as demonstrated by Hodges et al. 
The first step in this process was to identify the appropriate AAV serotype that would efficiently cross the vascular barrier using a conditional paradigm of low pressure and low volume. Based on prior studies, there are advocates for AAV6 and AAV8; however, viral loads, vascular transport enhancing agents, and ages of animals [10, 15] preclude direct comparisons between these serotypes in the published literature. AAV1 also has demonstrated activity following vascular delivery in the mouse, [18, 25] but again conditions for delivery cannot be directly compared with AAV6 and AAV8. For clinical trials, we anticipate a regional vascular delivery approach. The femoral artery supplying major muscle groups in the lower extremity represents a potential first target necessitating comparing AAV serotypes (AAV1, AAV6, and AAV8), carrying the same transgene (a modified micro-dystrophin), under control of the same promoter (MCK), using identical delivery conditions appropriate for a clinical setting. The results were unambiguous. AAV6 and AAV8 performed equally well for regional vascular delivery to the lower limb, which was strikingly different from AAV1. We found widespread micro-dystrophin expression in the TA and EDL at 1, 2, and 3 months post gene transfer (Fig. 1B), with insignificant differences in quantitation of transduced myofibers (rAAV8 > 90% and rAAV6 > 80%). These transduction efficiencies through a vascular route of delivery demonstrated a significant increase in isometric force generation and partial protection from contraction-induced injury compared to mdx (Fig. 5).
The transduction efficiency for rAAV1 was dramatically lower (<3%). We attribute the differences between our results with AAV1 and those of prior studies [18, 25] most likely related to the volume of perfusate [1 ml vs 100 μl (current study) per 20 g body weight] and the rate of administration [delivered over 10–39 seconds vs 60–80 seconds (current study)]. Using AAV1 for vascular gene transfer, Gonin et al. demonstrated capillary rupture within the endomysium that more than likely accounts for the high rate of gene expression. They also observed edema of the extremity. Although we did not directly examine the capillary bed in our preparations, we suggest that vascular integrity was maintained based on the following: 1) there was no extremity swelling in our mice; 2) AAV1, AAV6, and AAV8 demonstrated equal transduction efficiencies following intramuscular injections with >94% gene expression, yet impressive differences were seen following vascular gene transfer suggesting a distinction based on vascular endothelial transport, or 3) preferential inhibitory binding within the vasculature by rAAV1.
In the course of testing the serotypes for delivery of micro-dystrophin it was our intent to develop a gene transfer approach that could potentially be adapted to the clinic. In mouse we used isolated limb perfusion, a pre-flush of saline prior to gene delivery, a dwell time with all circulation occluded, and a post flush with an open venous system to rid the extremity of as much vector as possible. We believe that a pre-flush could have particular significance in a clinical trial where pre-existing antibody is a major threat to organ transduction. [27, 28] Even in the absence of pre-existing immunity, delivery of the virus in an environment free of serum and particulate material (complement proteins, platelets, erythrocytes, leucocytes), with its potential viral-binding capacity, provides a clearer path to muscle transduction and an added safety factor because of reduced risk for carrying virus to remote sites. In addition, a prolonged dwell time with occluded circulation in the presence of platelets and other clotting factors could lead to a blood clot in the extremity with potential serious consequences.
In the mouse, attention to the factors outlined resulted in 80% to 90% muscle transduction using rAAV6 and rAAV8 carrying micro-dystrophin. In an attempt to avoid putting too much emphasis on success in the mouse, we extended our studies to the non-human primate employing conditions that would closely simulate a clinical regional vascular delivery model. Thus, in the monkey we directly tested rAAV8 delivery of eGFP through an isolated femoral artery using conditions that would be applicable for a clinical trial (low pressure, low volume, and without pharmacologic agents). We used viral doses appropriate for clinical production runs (2 × 1012 vg/kg), and at the same time achieved levels of muscle gene expression in lower limbs predictive of a therapeutic benefit. 
Our study fulfills stringent criteria for gene delivery with implications for clinical application using an approach and AAV serotype that will potentially accelerate efforts to produce life-altering results in children with DMD, with applicability to other muscular dystrophies. Additional work in the monkey is required, especially using a cassette more analogous to one that will be used in practice. Studies in the non-human primate do provide a number of advantages including a >90% genomic sequence homology to humans, accounting for wide application to vaccine research and organ transplantation. The monkey is also well thought of for translational research for gene therapy owing to its high rate of natural infection by AAV similar to humans.  This is particularly relevant to studies of AAV8 gene transfer where translational paradigms, using monkeys with and without pre-existing immunity, permit study of specific immunosuppressive regimens. To successfully transfer rAAV to a canine dystrophy model, Wang et al.  were required to use anti-thymocyte globulin, cyclosporine and mycophenolate mofetil. The work presented here paves the way for studies of gene expression in a primate host under conditions that will be appropriate for the clinic using regional vascular delivery. Specific muscle groups in the limbs can be perfused and critical sites, like the diaphragm and heart, can be targeted. Studies in the monkey will also permit further sophistication in gene delivery, using fluoroscopy-guided balloon catheters for precise delivery of vector to isolated vascular regions with the advantage of subsequent removal of unbound virus reducing potential for spread to remote sites.