The vascular wall consists of three layers: the intima, a single layer of endothelial cells (ECs) that lie on the luminal surface of the vessel, overlying a thin layer of connective tissue; the media, consisting of vascular smooth muscle cells (SMCs) and connective tissue; and the adventitia which consists predominantly of loose connective tissue, but also contains fibroblasts. These layers are separated by the internal and external elastic laminae respectively. Gene transfer into the vascular wall was demonstrated first in 1989 [1]. Porcine primary ECs were transduced ex vivo using a murine amphotropic retrovirus and subsequently reintroduced into isolated segments of porcine ileo-femoral arteries using a double-balloon catheter. Since proof of principle was established the main clinical problems that have been investigated as potential targets for vascular gene therapy are prevention of restenosis post-coronary angioplasty (which now occurs principally in the guise of in-stent restenosis following coronary stent deployment), saphenous vein graft (SVG) degenerative disease following coronary artery bypass grafting (CABG), and induction of therapeutic angiogenesis within the peripheral and coronary arterial trees.

Gene delivery to cardiomyocytes, principally within the left ventricle, offers the potential to treat several conditions. As discussed above, pro-angiogenic genes may improve blood supply to the myocardium, and localized delivery of genes involved in the generation and propagation of cardiac electrical activity offers the potential for “biopacemaking” as an alternative to permanent implantable electronic pacemakers [19]. Myocardial gene therapy may also be useful for the cardiovascular manifestations of genetic conditions such as Fabry disease [20] and ion channel disorders (for instance single gene long QT syndrome). At the present, however, heart failure is the most actively investigated potential target for myocardial gene therapy.

Other putative therapeutic applications of cardiovascular gene therapy include risk factor modifications such as cholesterol lowering or antihypertensive gene therapy. Both high blood pressure and hypercholesterolaemia typically require lifelong oral therapy at the present and while such therapies are often very effective, the potential for gene therapy to act as a one-shot treatment for these chronic pathologies makes them attractive targets for investigation. However we shall confine ourselves for the remainder of this review to a consideration of those viral gene transfer vectors that have been applied to or have potential application to clinical cardiovascular gene transfer, and to a discussion of the current state of clinical virally-mediated gene therapy in the CVS.

Adenoviruses (Ad) were first described in 1953 [34]. Adenovirus is a natural human pathogen and over 50 serotypes of human adenovirus are known to exist [35,36]: wild-type infection most commonly causes respiratory tract infections, but can also result in pharyngitis, gastroenteritis, conjunctivitis, haemorrhagic cystitis and, most importantly from our perspective, myocarditis. Indeed, adenovirus infection (including infection by serotype 5 adenovirus, which is the basis of the most commonly used recombinant adenovirus vectors) is one of the commonest viral causes of acute myocarditis in children and young adults [37]. As anyone familiar with virus-mediated gene transfer will be aware, adenoviruses have several features which make them attractive for gene therapy: they have a broad natural tropism (reflected in the variety of illnesses that they can cause in wild-type guise); their high nuclear transfer efficiency ensures a rapid onset of transgene expression; they do not integrate into the host genome and do not, therefore, carry an appreciable risk of oncogenesis; they can infect both dividing and quiescent cells, and they can easily be produced in large quantities. The principal disadvantage of adenoviruses is their potent pro-inflammatory nature. This is largely a consequence of the hit-and-run fashion of wild-type adenovirus infection: adenoviruses have no mechanisms of cellular persistence and rely upon infecting and rapidly producing large quantities of new virus from host cells before they are killed by host inflammatory responses. The E3 region of the adenovirus genome encodes proteins that assist in evading host immunity, but these do so only to such an extent that will allow infected cells to survive long enough post-infection for the adenovirus lytic cycle to complete. The pro-inflammatory nature of adenoviruses also results in a limited duration of transgene expression as a consequence of clearing of infected cells by host inflammatory and immune mechanisms (although this may, in fact, be advantageous in certain applications where the pathological process is transient, such as neointima formation following coronary stent deployment). Their widespread prevalence as pathological agents in human communities also means that the majority of human adults have pre-existing adenovirus-neutralising antibodies [38]. Adenoviruses are also liable, consequent upon their broad tropism, to transduce non-target organs.

The use of adenoviruses for gene therapy has been reviewed numerously over the last 15 years and it would not serve us well to spend much of the present review discussing the basics of adenovirus biology from this perspective. However, briefly, the adenovirus virion consists of a non-enveloped icosahedral capsid particle containing a 30–40 kb linear dsDNA genome. Located at each of the twelve vertices of the icosahedron is a trimeric fibre shaft terminating in a globular knob domain. The length and flexibility of the fibre shaft varies significantly between adenovirus subtypes and can influence both binding and virus uptake [39]. The primary cell surface receptor for Ad5 is the Coxsackie-Adenovirus receptor (CAR) which binds to the knob domain and greatly enhances infection of cells upon which it is expressed [40]; heparan sulphate proteoglycans (HSPGs) and integrins act as co-receptors for certain cell-types. Not all cardiovascular cells express CAR, and while it is present on the surface of cardiomyocytes [41] it does not occur (or occurs only at low levels) on vascular SMCs and ECs [42], which are the commonest cell types in the vascular wall. The CAR is not essential for Ad5 infection however, and it is now appreciated that Ad cell-binding is more complex than previously thought. Alternative mechanisms of Ad5 transduction have recently been demonstrated in vivo: blood factors including coagulation protein IX and complement protein C4BP have been shown to bind the adenoviral fibre and promote localisation of adenovirus to the liver via cellular HSPGs and the LDL receptor proteins [43,44]. Other adenovirus serotypes have different primary receptors which are not as well characterised (see [45]). Following cell binding, the adenovirus virion enters the cell via clathrin-mediated endocytosis and undergoes endosomal processing prior to cytoplasmic release and delivery of the virus genome to the nucleus. Binding to the nuclear pore complex allows rapid transfer of the genome to the host nucleus [46]. Adenovirus is non-integrative and the genome remains in the nucleus in linear episomal form following successful infection.

Despite its broad natural tropism, when given systemically virtually all Ad5-mediated transduction occurs in the liver, predominantly within resident Kupffer cells [47]. Combined with the low-level of CAR expression on ECs, this makes Ad5-derived vectors poor candidates for systemic administration to the vasculature. However, Ad5 can transduce ECs in vivo if administered locally [48,49] and, although under normal circumstances the endothelium represents a barrier that adenoviruses cannot easily cross (except in the liver) [50], Ad5 can transduce medial SMC effectively if there is endothelial denudation [49], which occurs in association with advanced atherosclerosis and at sites of PCI (as a result of the physical process of intervention). Adenovirus is also capable of very effective myocardial transduction after local delivery; almost 80% of cardiomyocytes were transduced following combined simultaneous transfusion of first generation vector into the left anterior descending coronary artery and great cardiac vein in juvenile pigs [51]. As a consequence, Ad5 has found use in cardiovascular gene transfer in studies of localized delivery of gene therapies to the vessel wall, to the myocardium and into skeletal muscle in ischaemic limbs.

Recombinant vectors derived from serotype 5 adenovirus (Ad5 – a subgoup C adenovirus) are by far the best characterised and have been used in the majority of clinical trials. First-generation recombinant Ad5 have typically had the E1 and E3 regions (which contain genes that are essential for viral assembly and for evasion of host immunity respectively) removed. Several second-generation recombinant Ad5 have been described that include additional deletions of the adenovirus genome from the E4 locus [52], or the E2A region [53]. Other second-generation modifications include functional mutations in the E2A region or the inclusion of an immunomodulatory transgene from the serotype 2 adenovirus [53]. Some evidence exists to suggest that modest benefits in transgene expression might be obtained by use of such second-generation vectors in preference to first-generation adenoviruses within the vasculature [52], although greater transgene expression was manifest only at 10 days after infection of rabbit carotids. No difference in transgene expression was observed at 3 or 28 days post-infection in this study, and other studies of second-generation vectors have provided no evidence at all of benefit in magnitude or duration of transgene expression after arterial gene transfer [53]. As a consequence, first-generation recombinant adenoviruses, despite (or perhaps because of) their relative simplicity, have remained the mainstay of clinical studies of gene therapy within the vasculature.

Further modification has given rise to third-generation recombinant adenoviruses, from which all wild-type adenoviral coding sequences have been deleted. These viruses have an increased cloning capacity of approximately 35 kb (compared to around 8 kb in first- and second-generation viruses) and produce no viral proteins in infected cells [54], as a result of which they give rise to markedly reduced host adaptive immune responses and longer durations of transgene expression [55]. These helper-dependent (or “gutless” if you prefer the more colourful nomenclature) vectors have been applied to gene transfer within the vasculature of animals with impressive medium-term results [56,57]. Transgene expression persisted for at least eight weeks in rabbit carotids infected with a helper-dependent adenovirus expressing rabbit urokinase-type plasminogen activator, with stable expression from day 14 to day 56, which contrasted with complete loss of transgene expression by day 14 from arteries infected with first-generation viruses. Helper-dependent adenoviruses also elicited a significantly reduced inflammatory response within rat myocardium compared with first-generation viruses, which was associated with evidence of prolonged transgene expression [58]. On these bases, it seems that gutless adenoviruses are superior to their first- and second-generation forebears as vectors for cardiovascular gene transfer. However, despite these reports, very few pre-clinical studies of cardiovascular gene transfer have used helper-dependent adenoviruses as their mode of gene transfer. They have been used in a murine model of hypertension in which tail vein injection of vector achieved long term (> 120 day) regulatable hepatic expression of atrial natriuretic peptide with concomitant reduction of heart weights and systolic BP in infected animals [59]. They have also been applied to a rabbit model of hindlimb ischaemia, in which intramuscular injection of a gutless adenovirus expressing sphingosine kinase resulted in improved limb perfusion 20 days post-delivery [60]. No more long-term observations were reported in this study however and, as no comparison was made with first- or second-generation adenoviruses, we will never know if this effect was greater than what might have been achieved by a more simple vector, or sustained for the months (or even years) that would be required to elicit an effect of genuine clinical value.

It is almost certain that part of the reason for the surprisingly limited uptake of gutless adenoviruses in pre-clinical studies of cardiovascular gene transfer lies in the relative difficulty of making the helper-dependent vectors. Most such pre-clinical studies are of a relatively short-term nature and it seems that the advantages in duration of transgene expression and reduced host inflammatory responses that are offered by gutless adenoviruses do not outweigh the extra effort of manufacture, particularly as Wen et al reported that peak transgene expression following helper-dependent virus-mediated gene transfer was only around 10% of that observed after a first-generation virus was used to deliver the same transgene [57]. In addition, gutless adenoviruses still induce an innate immune response to the viral caspid [61] (and possibly to CpG motifs within the viral genome itself [62-64]) and most adult humans still possess pre-existing antibodies to the serotype 5 virus particles: the absence of viral protein expression does not confer any greater capacity to evade pre-existing humoral immunity [57]. Until the advantages that third-generation adenoviruses undoubtedly possess in immunologically naïve experimental animals are shown to translate into clinical benefits by comparison to first-generation adenoviruses, it is likely that researchers will persevere with the old technology.

AAV is a small member of the parvovirus family with a 4.7 kb single-stranded DNA genome. Wild-type AAV has three unique, potentially beneficial characteristics which distinguish it from other gene therapy vectors: firstly it cannot replicate without the assistance of a helper virus, such as adenovirus or herpes simplex virus; secondly the AAV genome is capable of long-term persistence within the nucleus, either by site-specific integration into the AAVS1 locus on the long arm of chromosome 19 or in episomal form; thirdly, despite the fact that a large proportion of the world’s population is seropositive for a variety of AAV serotypes, AAV has never been shown to cause human disease. Since the first infectious clone of AAV serotype 2 was established in 1982 [82], a total of 12 serotypes [83,84] and over 100 variants have been identified from human and non-human primate tissues [85]. Recombinant AAV (rAAV) vectors have had almost the entire viral genome removed, leaving only two regions of inverted terminal repeats (ITRs) in between which the transgenic DNA is inserted. The AAV Rep and Cap genes which are required for viral replication and packaging are supplied by a helper plasmid during the production process [86].

As a consequence of loss of the Rep gene, rAAV lose the capacity for site-specific integration into chromosome 19 and acquire the potential for random integration with the risk of oncogenesis, although the available evidence suggests that integration of AAV genomes is inefficient even in wild-type form and nuclear persistence is usually a consequence of episomal maintenance [87]. The actual risk of oncogenesis arising from random integration of AAV genomes is likely to be small therefore, although this is obviously a matter for future studies to clarify. Advances in the development of rAAV vectors have been reviewed in recent years [83].

Recombinant AAV offers some very advantageous features as a gene therapy vector. As wild-type AAV is not pathogenic it represents the safest of the viral vectors being considered and is significantly less immunogenic than Ad. Recombinant AAV elicit long-term gene expression as the genome persists in the nucleus, largely as circularised dsDNA episomes [88]. A single intramuscular injection of rAAV containing the factor IX gene to treat haemophilia B has been shown to result in continuing gene expression at 3.7 years in humans [89]. AAV are not without their drawbacks however. The onset of transgene expression is substantially delayed compared with other vectors, as a result of slow nuclear transport and the need for the single-stranded genome to be converted to dsDNA prior to expression [90]. As a consequence of this, early studies of AAV-mediated arterial gene transfer found no transgene expression within the first week following vessel infection, although transduction was manifest in the second week post-exposure [91]. AAV have been generated that contain a self-complementary double-stranded DNA genome. These elicit a significantly more rapid onset of transgene expression and enhanced cellular transduction by comparison to the parent ssDNA vector; however this improvement comes at the cost of a halving of the packaging capacity of the resulting vectors [92,93].

The small packaging capacity of AAV containing an ssDNA genome (approximately 4.6kb) imposes modestly severe limits upon the size of transgene expression cassette that can be inserted. Having said that, 4.6 kb still offers substantial scope for therapeutic gene transfer, and sterling work has been done in minimizing the size of therapeutic gene sequences in order to allow packaging into AAV. This is exemplified by AAV-mediated transfer of dystrophin: the full-length dystrophin cDNA at ≈14kb is far too large for packaging into AAV, yet a functional micro-dystrophin cDNA of 3.8kb has been packaged into a rAAV and used to elicit potentially therapeutic effects in mice [94]. Nonetheless, there are some genes that will probably never be suitable for AAV-mediated gene transfer, large ion-channels with multiple transmembrane regions for example, and physiological regulation of transgene expression from AAV by inclusion of genomic promoter sequences is likely to prove challenging. As with adenovirus, immune clearance of transduced cells can be a major problem too, particularly given the high prevalence of neutralising antibodies in the general population [95].

AAV2 was the first adeno-associated virus to be developed as a gene therapy vector and represents the most extensively investigated of the AAV serotypes. Infection is thought to be primarily mediated by membrane-associated HSPG [96], although other pathways for cellular uptake exist in non-hepatic tissue including the heart [97]. Removal of the HSPG primary receptor reduces liver transduction whilst cardiac transduction is preserved [98]. AAV2 is tropic for arterial SMCs and elicited transgene expression in 10–20% of medial SMCs 21 days after infection of rabbit carotid arteries in vivo. Endothelial transduction was poor however [49]. Comparison in cultured cells confirmed that AAV2 elicited modestly greater transduction of human saphenous vein SMCs than AAV3-8 and none of these alternative serotypes elicited substantial transduction of ECs either [99,100]. AAV2 targeting to increase SMC transduction has been achieved using the heptapeptide that was effective in targeting recombinant adenoviruses [76]: an increase of up to 70-fold in transgene expression was seen in human coronary artery SMCs exposed to targeted AAV by comparison with non-targeted vectors, although an 18-hour period of exposure was required for this magnitude of effect. A significant enhancement of transduction was observed in coronary artery SMCs after only one hour of exposure, although no enhancement of transduction of human saphenous vein SMCs was observed after this shorter period of exposure. Use of AAV for vascular gene transfer has been very limited however (the authors are aware of only one study that has ever attempted to elicit a ‘therapeutic’ effect by localized AAV-mediated vascular gene transfer [101]), and most interest in AAV within the cardiovascular system has been directed towards its use for myocardial gene transfer. In that respect AAV2 is not the most efficacious serotype for potential therapeutic application. AAV2 vectors pseudotyped with capsid proteins from other AAV serotypes have been studied to establish whether myocardial delivery can be improved by such means. AAV2 pseudotyped with AAV1, AAV6 and AAV8 capsid proteins all elicited greater myocardial transduction in rats than AAV2 after direct intramyocardial injection, at all time points up to 24 week post-infection [102]. AAV1 and 6 gave rise to transgene expression that maximized at four weeks and remained stable until the final 24-week time point. However, greatest expression at all time points was achieved by AAV8, which manifested an increase in transgene expression at each consecutive time point. In a different study, recombinant AAV2 pseudotyped with AAV1 (AAV2/1) increased transgene expression in human and adult murine cardiomyocytes by approx 2- to 3-fold when compared with AAV2 [103], but AAV2/8 and AAV2/9 were subsequently shown to elicit ≈20-fold and ≈200-fold greater myocardial transgene expression than AAV2/1 following intravenous injection into 1-day old mouse pups [27]. The cardiotropism of AAV9 was confirmed following intrapericardial injection into neonatal mice and adult rats, in which AAV9 produced global myocardial transduction that was stable for up to one year and significantly greater than AAV1, 6, 7 or 8 [104].

At the present, rAAV are the vector of choice for myocardial gene transfer and the capacity of serotypes 1, 6, 8 and 9 for effective transduction of cardiomyocytes offers the prospect of genuinely effective therapeutic myocardial gene transfer in the clinical setting. Unanswered questions remain about the prospect of integrational oncogenesis, and it is likely that the usefulness of rAAV as therapeutic agents will ultimately be confined by their limited capacity to deliver transgenic material. Nevertheless, rAAV offer the best prospect of breakthrough successes in the field of clinical virus-mediated cardiovascular gene therapy.

Lentiviruses are part of the retrovirus family and consist of a ssRNA genome enveloped in a lipid bilayer; most currently investigated lentiviruses are derived from HIV-1. The primary receptor for lentivirus is the T-cell CD4 receptor and, as opposed to Ad and AAV, cellular entry occurs via membrane fusion. The viral capsid is subsequently released into the cytoplasm where uncoating and reverse transcription of the viral ssRNA to dsDNA occurs followed by nuclear transport via the microtubuli [105]. A major advantage of lentiviruses is that, unlike Ad and AAV, they are not inherently immunogenic. Unlike other retroviruses, which cannot readily cross the nuclear membrane, lentiviruses are able to transduce non-dividing cells, which is an attractive characteristic for cardiovascular gene therapy as vascular cells and cardiomyocytes are quiescent in their resting state. Lentivirus possesses an 8 kb packaging capacity.

Two major developments were required to make lentivirus a possible gene therapy vector. Firstly, self-inactivating lentivirus vectors were generated in which the U3 promoter region of the long terminal repeat had been inactivated [106], reducing the chance that homologous recombination and generation of wild-type HIV-1 can occur. Secondly, given that wild type lentivirus only infects CD4+ immune cells, pseudotyping with glycoproteins derived from other enveloped viruses is required to improve tropism for other cells. Lentiviruses pseudotyped with the attachment glycoprotein of the vesicular stomatitis virus (VSV-G) have been the most extensively investigated vectors. These vectors demonstrate significantly broadened tropism and high stability (reviewed by: [107]) and have been used to demonstrate efficient transgene delivery in vitro into SMCs and ECs from human saphenous vein [100], human coronary artery SMCs and ECs [108], and cardiomyocytes [109]. Comparison with Ad5 and AAV2-6 confirmed greater transgene expression in lentivirus-infected SMCs, although Ad5 was a more effective transducer of ECs [100]. Pseudotyping of lentivirus with Hantavirus glycoprotein has been shown to result in greater levels of transgene expression in the balloon-injury rabbit carotid model, and the delivery of human extracellular superoxide dismutase resulted in a reduction in neointima formation [110].

Potential clinical uses of lentivirus have been demonstrated in vivo in animal models. Expression of TIMP-3 resulted in reduced SMC migration and increased SMC apoptosis [100], while administration of a VSV-G pseudotyped lentivirus encoding VEGF resulted in increased angiogenesis in an in vivo rabbit hindlimb ischaemia model [111]. Study of direct intraventricular injection of lentivirus encoding for alpha-galactosidase in a mouse model of Fabry disease showed short-term correction of cardiac abnormalities but this benefit was lost by three months [20]. Direct intraportal injection of a third generation liver-specific lentivirus encoding for the low-density lipoprotein receptor resulted in significant reductions in serum cholesterol in a hyperlipidaemic rabbit model which were maintained up to two year follow-up [112].

Despite the potential that pseudotyped lentiviruses offer as vectors for cardiovascular gene transfer, their use in the clinical setting is very substantially hindered by concerns over their safety. The risk of insertional mutagenesis with integrative vectors has been confirmed in a clinical trial of a gammaretrovirus for the treatment of X-linked severe combined immunodeficiency. Two out of ten patients in this trial developed T-cell leukaemia as a result of integration of the vector in proximity to the LMO2 proto-oncogene [113,114]. Unlike haematological precursor cells, the cellular targets of cardiovascular gene therapy are very infrequently associated with primary neoplasia. Nonetheless there is a largely comprehensible reluctance to take risks with potentially oncogenic gene transfer vectors in any clinical setting until the potential for generation of malignancies can be shown to be within acceptable limits. The generation of replication-competent recombinant lentiviruses is also a theoretical safety concern. Non-integrating lentiviruses, created by mutation of the integrase gene, have been developed recently and offer the potential for safer gene therapy with a much lower risk of insertional oncogenesis and generation of replication-competent recombinants, whilst maintaining a broad tropism and high transduction efficiency. Despite the lack of genomic integration, long-term gene expression can occur in quiescent cells as a result of episomal nuclear retention, although the virus is inevitably lost in dividing cells. Sustained transgene expression with non-integrating lentivirus has been demonstrated in vivo in the rodent brain [115], retina [116], skeletal muscle [117] and liver [118]. However efficient cardiovascular gene transfer has yet to be demonstrated with integrase-deficient lentiviruses: a study with an earlier generation of integrase-defective lentivirus did not result in sustained transgene expression in cardiomyocytes [109]. For those seeking greater enlightenment, non-integrating lentiviral vectors are reviewed by Ravet et al. in another article in this issue.

The induction of angiogenesis as a therapeutic strategy for both coronary and peripheral arterial disease has been investigated in a series of clinical randomised clinical trials, primarily using vascular endothelial growth factor (VEGF) or fibroblast growth factor (FGF) as the proangiogenic transgene. Most research on PAD has employed plasmid vectors for gene transfer, although studies of viral gene therapy have been published: in a phase II study of intra-arterial injection of AdVEGF₁₆₅ following peripheral angioplasty an increase in new vessels distal to the site of vector delivery was demonstrated, but this was not accompanied by improved healing of ischaemic ulcers, resolution of rest pain or increased ankle-brachial index by comparison with controls [143]. Interestingly, in the same study, a similar effect was elicited by plasmid/liposome delivery of VEGF₁₆₅ as was achieved by AdVEGF₁₆₅._{In the RAVE study of intramuscular injection of AdVEGF121, there was no improvement in measures of ischaemia or clinical outcomes, although the therapy was well-tolerated [144]. The lack of demonstrable, clinically beneficial effect from adenovirus-mediated gene transfer in the periphery appears to have largely silenced further interest in virus-mediated gene transfer in this setting, although an ongoing phase II study of Ad2-mediated delivery of hypoxia inducible factor-1α (entitled WALK) is expected to deliver results in 2010 (http://clinicaltrials.gov/ct2/show/NCT00117650), and it will be of enormous interest to see if this study offers something of greater therapeutic substance than has been reported previously in this setting.}

In contrast to studies of PAD, the majority of studies of angiogenic gene transfer for myocardial ischaemia have used virus-mediated (specifically, adenovirus-mediated) gene transfer – presumably it is easier to convince regulatory authorities of the life-saving potential of viruses in this setting than in the limb vasculature (patients with PAD do not typically die of their peripheral arterial problems, but of myocardial infarction or stroke).

In the earliest study of intracoronary injection of AdFGF-4 (AGENT-2), delivery of active vector by subselective coronary catheterization of culprit arteries resulted in a borderline significant reduction in the size of the region of myocardium demonstrating stress hypoperfusion eight weeks post-delivery [145]. This was not accompanied however, by a significant clinical effect. At around the same time, the KAT study investigated localized intracoronary delivery of AdVEGF₁₆₅. This study differed from AGENT-2 in employing an intracoronary balloon catheter to attempt to restrict vector delivery to the coronary artery wall at the site of stent deployment, rather than simply injecting adenoviruses down the coronary artery. As was the case in the AGENT-2 study however, no clinical effect was observed, and no effect was evident on in-stent restenosis either although an improvement was claimed in myocardial perfusion as assessed by cardiac SPECT imaging at six months post-delivery when compared with the pre-PCI myocardial perfusion in the AdVEGF₁₆₅-treated group [146]. The data presented however do not actually suggest that there was a significant difference in myocardial perfusion between the treated and control groups at 6 months after PCI and vector delivery.

Further studies of intracoronary injection of AdFGF-4, in the guise of the AGENT-3 and AGENT-4 studies, have given rise to the largest experience to date of cardiovascular gene therapy, with over 500 patients with chronic angina having undergone enrolment [147]. Intracoronary administration of Ad5-FGF-4 failed to improve the primary end-point of total exercise time. However further analysis of these trials has identified unusual gender-specific results, with a significant improvement in clinical outcomes (exercise treadmill endurance and angina class) for women in a vector-dose-related fashion at both six and twelve months post-delivery, with greater improvements observed in those exposed to a greater vector dose. No significant improvement was observed in men. This finding is being investigated further in the phase III AWARE trial (http://clinicaltrials.gov/ct2/show/NCT00438867).

In addition to intracoronary administration of viruses, direct intramyocardial injection of adenovirus has been investigated following thoracotomy in human studies. Administration of Ad5-VEGF₁₂₁ hinted at clinical efficacy in a phase I study, with “suggested” improvements in regional ventricular wall motion, severity of angina and exercise tolerance [148]. This led to the phase II REVASC study which reported improvements in the primary end-point of exercise time as compared to a medical therapy control group who did not undergo thoracotomy. However myocardial perfusion was significantly worse in the treatment group and it is likely that the improvement in symptoms in this trial was largely attributable to a placebo effect related to the thoracotomy [149].

All considered, adenovirus-mediated myocardial gene transfer of pro-angiogenic genes has provided little cause for enthusiasm about the potential for widespread clinical application of such therapy. However, the fact that it has been possible to elicit clinically useful effects (albeit restricted to women) by delivery of adenoviruses, which – it must be remembered – are likely to give rise to appreciable transgene expression for only two or three weeks post-delivery, does raise the possibility that more useful clinical effects may arise from the use of vectors that will elicit transgene expression of longer duration. In this respect, it will be interesting to see whether studies of angiogenic gene transfer using AAV, gutless adenoviruses or lentivirus will ever see the light of day.

A vast number of potentially therapeutic transgenes have been studied in animal models of restenosis [91] but very few clinical trials have been performed. The KAT trial, which was discussed in the previous section, assessed restenosis as a secondary endpoint following local virus delivery to the site of coronary angioplasty (92/103 of patients received stents) [146]. Administration of AdVEGF₁₆₅ had no effect on angiographic restenosis at six months. Similarly, in the study by Laitinen and colleagues, AdVEGF₁₆₅ improved angiogenesis but did not reduce restenosis at the site of peripheral angioplasty [143].

Gene-eluting stents using both adenovirus and plasmid vectors have been investigated in vivo [6,7,150,151]. Reductions in neointima formation have been demonstrated following adenovirus-coated stent implantation in the rat carotid [6,7] and rabbit iliac arteries [150] but no human trials of this method of vector delivery have been performed at the time of writing. In truth, the relative success of drug-eluting stents (DES) at preventing (or at least delaying the onset of) in-stent restenosis really means that the technical demands of virus-mediated gene therapy are never going to be suited to stent-mediated delivery in the clinical setting. DES can be deployed into a patient straight out of the packet without any time-consuming virus loading (and concomitant biological safety paraphernalia) and any competing technology must offer a similar ease of use for the clinician or some very substantial clinical benefit. In this respect, virus-mediated stent-based gene transfer is akin to intra-coronary brachytherapy in that potential benefits exist (albeit in the case of virus-mediated gene transfer, never proved in the clinical setting), but those benefits are offset by the technical difficulties incumbent upon actually delivering the therapy. Any gene therapeutic approach to ISR must be as easy to deliver clinically as a DES, which means plasmid-eluting stents may be a preferable option.

Ex vivo delivery of virus to vein grafts at the time of CABG offers a more immediately tempting milieu for virus-mediated vascular gene therapy than ISR [152] and efficacy in pre-clinical models has been demonstrated using many transgenes [153,154]. However, in spite of this, no clinical studies using ex vivo viral gene delivery to human saphenous vein bypass conduits have been reported yet. The only clinical trial of nucleic acid therapy aimed at amelioration of bypass graft disease (using an elongation factor 2 transcription decoy oligonucleotide) reported negative results [155]. Part of the difficulty underlying translation of pre-clinical work into the real world of cardio-thoracic surgery is the short-term nature of those pre-clinical studies that have been performed: vein-graft disease is a phenomenon that manifests clinically over years rather than weeks or months, and it is almost certain that to be effective gene therapy strategies will have to elicit transgene expression for years too. Low-generation adenoviruses, the vector used in the large majority of pre-clinical studies of vein graft NIH, are entirely unsuitable for long-term gene transfer, and those vectors that are suited for this purpose (gutless adenovirus, AAV, lentivirus [100,156]) have scarcely been studied in this setting; only one study of “therapeutic” gene transfer employing any of these vectors in a model of vein graft disease has made it to press as far the authors are aware [157]. Furthermore, means of delivering plasmid DNA with sufficient efficacy to be of potential clinical value after clinically-pertinent periods of exposure are surfacing [158], so it is distinctly possible that non-viral gene transfer will eventually displace virus-mediated gene transfer as the most clinically-relevant method of gene transfer in this setting too.

The only completed clinical trials of myocardial gene therapy to date have been of proangiogenic factors, and as already discussed, the outcomes of these studies have been less than impressive. However myocardial gene therapy has recently become the focus of renewed interest due to the initiation of clinical studies using an AAV vector for the treatment of heart failure. Two excellent recent reviews discuss this field in more detail [24,159]. Briefly, the SERCA2a gene encodes for the sarcoplasmic reticulum calcium^{ATPase pump which transfers cytoplasmic calcium back into the sarcoplasmic reticulum during cardiomyocyte relaxation. A decrease in activity of SERCA2a and subsequent impaired calcium reuptake has been shown to be present in human heart failure [160] and animal models have demonstrated that transgene expression of SERCA2a using viral vectors (adenovirus and AAV) can improve left ventricular function [161]. As a result of promising preclinical data, two studies of gene therapy in heart failure have received approval using AAV vectors containing the human CMV promoter and the SERCA2a transgene. The CUPID study is a phase 1/2 placebo-controlled clinical trial and randomised patients with severe heart failure of either ischaemic or non-ischaemic aetiology to receive either a ten minute intracoronary infusion of AAV1-CMV-SERCA2a or placebo. The study has finished recruitment and preliminary results are expected in 2010 [162] . A second study in the UK is investigating gene delivery of the same transgene using a different AAV serotype (AAV6-CMV-SERCA2a) in patients with end-stage heart failure who have already undergone left ventricular assist device (LVAD) implantation. This trial has received approval and is due to commence recruitment in early 2010 (http://clinicaltrials.gov/ct2/show/NCT00534703).}

It is expected that these studies will provide important information on both the suitability of AAV as a vector and whether SERCA2a is a beneficial transgene in human heart failure. Although the results are eagerly awaited there are several reasons to be cautious. The vectors chosen for these studies may not be the optimal vectors for myocardial gene delivery; as discussed earlier, the AAV8 and AAV9 serotypes have been shown to exhibit greater myocardial tropism than the AAV1 and AAV6 serotypes, and the human CMV promoter, whilst efficacious in cardiomyocytes, runs the risk of being rendered quiescent in the long-term as a consequence of DNA methylation. There are also some safety concerns with the use of SERCA2a. Overexpression of SERCA2a in rat myocardium leads to an increased rate of fatal arrhythmia [163] and for this reason all patients in the CUPID trial are required to have an implantable cardiac defibrillator prior to enrolment. Still, this is a therapeutic area where there is a great (and growing) clinical need and it represents one of the cardiovascular targets where gene therapy can hope to offer something entirely new, although it is important not to pin too many hopes on a successful outcome in what is effectively the first therapeutic iteration in this area.