Gene therapy for the prevention of vein graft disease

Abstract

Ischemic cardiovascular disease remains the leading cause of death worldwide. Despite advances in the medical management of atherosclerosis over the past several decades, many patients require arterial revascularization to reduce mortality and alleviate ischemic symptoms. Technological advancements have led to dramatic increases in the use of percutaneous and endovascular approaches, yet surgical revascularization (bypass surgery) with autologous vein grafts remains a mainstay of therapy for both coronary and peripheral artery disease. Although bypass surgery is highly efficacious in the short-term, long-term outcomes are limited by relatively high failure rates as a result of intimal hyperplasia, which is a common feature of vein graft disease. The supply of native veins is limited, and many individuals require multiple grafts and repeat procedures. The need to prevent vein graft failure has led to great interest in gene therapy approaches to this problem. Bypass grafting presents an ideal opportunity for gene therapy, as surgically harvested vein grafts can be treated with gene delivery vectors ex vivo, thereby maximizing gene delivery while minimizing the potential for systemic toxicity and targeting the pathogenesis of vein graft disease at its onset. Here we will review the pathogenesis of vein graft disease and discuss vector delivery strategies and potential molecular targets for its prevention. We will summarize the preclinical and clinical literature on gene therapy in vein grafting and discuss additional considerations for future therapies to prevent vein graft disease.

INTRODUCTION

Atherosclerosis results in obstruction of the coronary and peripheral arteries and is a major cause of morbidity and mortality worldwide¹. Pharmacological therapies help to control symptoms and slow disease progression in mild to moderate or chronic stable disease, but more invasive interventions, including percutaneous and surgical revascularization, are required in acute or severe cases. Although advances in percutaneous and endovascular approaches over the past two decades have resulted in substantial increases in utilization of these procedures, surgical revascularization remains a mainstay of therapy for atherosclerosis of the coronary and peripheral arteries. In the United States, over 400,000 patients per year still undergo coronary artery bypass grafting (CABG)^1–4. Peripheral artery disease is now recognized to be nearly as common as coronary disease, with over 8 million affected individuals in the U.S.¹ Although the rate of peripheral bypass surgery has fallen significantly, over 40,000 of these procedures are performed annually⁵. Arterial conduits such as the internal mammary artery are preferred when available, but venous conduits, primarily the saphenous vein, are still the most commonly used vascular grafts.

Post-operatively, vein grafts undergo "arterialization", in which the vessel wall thickens due to proliferation of vascular smooth muscle cells (SMC) and increases in extracellular matrix (ECM) deposition. This process is thought to be necessary for the vein to adapt to the high wall stress and shear force within the arterial circulation, but arterialization may also contribute to subsequent stenosis, accelerated atherosclerosis, and ultimate failure of the graft. In the coronary circulation the vein graft failure rate at 1 and 10 years is reported to be 15% and 50%, respectively^{6, 7}. In the peripheral vasculature failure rates may be as high as 50% within 5 years⁸. Vein graft failure is a substantial health problem, as it leads to recurrent ischemic events, such as angina, myocardial infarction, and gangrene, necessitating repeat revascularization procedures or amputation. Vein graft failure is not only devastating to individual patients but it also contributes significantly to the healthcare costs of cardiovascular disease, which exceeded $286 billion dollars in the United States in 2007 alone⁹.

Medical therapies, including aspirin and other anti-platelet agents and aggressive lipid-lowering therapy, have proven successful in reducing saphenous vein graft (SVG) atherosclerosis, the need for revascularization procedures, and the incidence of major adverse cardiovascular events^10–12. However, despite these advances, overall vein graft failure rates have remained fairly constant since the introduction of coronary artery bypass grafting in 1968^13–15. The inability of traditional approaches to effectively reduce vein graft failure rates has led to the investigation of alternative strategies to maintain long-term vein graft patency, including gene therapy. Here we will review the pathophysiology of vein graft disease (VGD) and discuss vector and gene delivery strategies for the prevention of VGD. We will then discuss targets for vein graft gene therapy based on the molecular mechanisms of VGD and address additional considerations for future studies, with a focus on translating the wealth of pre-clinical data into clinical applications.

PATHOPHYSIOLOGY OF VEIN GRAFT DISEASE

Although the vasculature is often perceived as a system of passive channels, blood vessels are themselves complex organs that consist of several cell types working in concert to regulate vascular function. Vessels larger than capillaries can be divided into three layers: the intima, media, and adventitia. The composition and thickness of some of these layers depends on the size of the vessel and varies considerably between arteries and veins. The intima is the innermost layer of the blood vessel and in all normal vessels is composed of a single, continuous layer of endothelial cells (EC) and basal lamina. Outside of the intima is the media, which, in veins, consists primarily of SMCs surrounded by ECM. The adventitia is the outermost layer of the vessel and contains connective tissue, nerves, and the vasa vasorum, the small vessels that vascularize the vessel wall.

The molecular mechanisms that lead to vein graft disease are initiated as soon as the vein graft is surgically harvested (Fig 1). Removal of the vein from the circulation cuts off blood supply to the vasa vasorum, resulting in ischemia and subsequent cellular hypoxia, leaving the vein vulnerable to ischemia-reperfusion (IR) injury after engraftment (reviewed in^{16, 17}). Physical manipulation during removal contributes to injury as well, and a “no-touch” technique for vein graft harvesting has been shown to improve patency rates (reviewed in^{18, 19}). Once the vein is grafted into the arterial circulation, increased wall stretch and shear stress result in additional EC injury and, in some cases, death²⁰. These altered hemodynamic forces induce gene expression changes in the remaining cells, which promote inflammation and cellular proliferation^21–23. Damage to the endothelium results in platelet adhesion, activation, and aggregation; leukocyte recruitment; and activation of the coagulation cascade^24–27. All of these events represent targets for molecular intervention and will be discussed in detail below. The common result of the vascular injury is local production of growth factors and hormones that stimulate a phenotypic change in SMCs in the vessel wall. Within 24 hours of vein graft implantation, vascular SMCs undergo a well-characterized shift from quiescent, contractile cells to proliferative, synthetic, and motile cells^{28, 29}. By post-operative day 7, SMCs have migrated into the intima and begun to express ECM genes³⁰. The thickening of the intima as a result of SMC proliferation, migration, and ECM protein production is known as intimal hyperplasia (IH) and is the hallmark of VGD. The so-called "neointima" also forms the nidus for accelerated atherosclerosis, which enhances VGD and can lead to plaque rupture and subsequent acute graft occlusion³¹.

Fig 1.

Schematic representation of the phases of vein graft injury and timeline of different gene therapy effects. Surgical harvesting induces injury of the vein graft endothelium, which is compounded by arterial shear force and wall stress after anastomosis. Rapid endothelial damage results in exposure of tissue factor and upregulation of adhesion molecules, increasing the risk of thrombosis and leading to adhesion and migration of inflammatory leukocytes into the vessel media. Upregulation of growth factors, cytokines, and matrix metalloproteinases (MMP) in the vein graft wall leads to SMC migration into the neointima, SMC proliferation, and resultant intimal hyperplasia and increased susceptibility to development of atherosclerosis. Onset and magnitude of transgene expression or effect varies depending on the gene delivery modality. Oligodeoxynucleotide (ODN) decoys and siRNA, which do not require gene expression, can act almost immediately; adenoviruses express high levels of transgene for brief periods; and adeno-associated viruses express moderate levels of transgene for extended periods.

The molecular mechanisms that underlie IH are complex and incompletely understood. A more detailed review of this process is beyond the scope of this article, however the reader is referred to several excellent reviews for more information^{28, 30, 32}.

VECTORS AND DELIVERY STRATEGIES

Vein graft disease is an ideal candidate for gene therapy for several reasons. First, because vein graft disease begins with the surgical dissection and removal of the vein graft from its native location, intraoperative gene therapy allows molecular reprogramming to occur at the time of the initial insult (Fig 2). Unlike many other conditions, gene delivery for prevention of VGD is one of the few scenarios in which pre-treatment is possible. Second, the time between harvesting the vein and surgical anastomosis into the arterial circulation provides a natural window for gene delivery. Third, gene delivery to vein grafts can be performed ex vivo, which substantially diminishes some of the concerns inherent in systemic gene therapy. After vector delivery to the vein graft lumen, unincorporated vector can be flushed from the graft, markedly decreasing the chances of systemic delivery. Accordingly, our group used ex vivo adenovirus delivery in a canine model of aortocoronary bypass in which the vein graft was gently flushed with heparinized saline following gene delivery, and the transgene was undetectable in multiple organs at the time of graft harvest 90 days later (Fig 3D)³³.

Fig 2.

Vector delivery strategy to prevent vein graft disease. Prior to surgical harvesting of the vein graft, both the endothelium and underlying SMCs are in a quiescent state. Injury occurs within minutes as a result of ischemia and mechanical trauma and is exacerbated upon anastomosis into the arterial circulation. Transgene delivery is performed ex vivo and occurs rapidly and with a low risk of systemic transgene toxicity. Endothelial injury, thrombosis, inflammation, and SMC activation have been targeted for the prevention of vein graft intimal hyperplasia.

Fig 3.

Targeting vein graft intimal hyperplasia using PTEN overexpression. (A) Numerous growth factors and hormones activate different isoforms of phosphoinositide 3-kinase (PI3K) via receptor tyrosine kinases and G protein-coupled receptors (GPCR). PI3K activates numerous downstream effector molecules, which promote the cellular responses that collectively lead to intimal hyperplasia and vein graft failure. PTEN is a lipid phosphatase that specifically hydrolyzes the phospholipid product of PI3K, thereby inhibiting PI3K-mediated SMC growth and intimal thickening. (B–C) Adenovirus-mediated overexpression of PTEN (C) in a canine aorto-coronary bypass graft model resulted in inhibition of intimal hyperplasia at 90 days compared to saline-treated grafts (adapted from Hata et al³³). (D) The PTEN transgene was detectable by PCR only in the AdPTEN-treated vein graft (VG) and not in other tissues at the time of harvest. RV, right ventricle; LV, left ventricle.

The development of gene transfer vectors is a major challenge in the field of gene therapy, but it is important to consider features of existing vectors and gene targeting modalities as they pertain to the pathophysiology of IH (Fig 1). For additional information on gene transfer strategies, the reader is referred to other reviews in this series.

Non-viral vectors

Non-viral vectors are attractive options for gene delivery because they are relatively safer than viral vectors and easier to prepare. However, non-viral vectors have historically been plagued by poor transfection efficiency. In the following section, we review non-viral vector-based gene delivery approaches to prevent VGD, including naked plasmid DNA, oligodeoxynucleotides (ODN), small interfering RNA (siRNA), and polymeric nanoparticles.

Naked plasmids, ODN, siRNA

Non-viral vectors are a promising strategy for VGD for several reasons. First, they can be optimized to efficiently transfect both ECs and SMCs^34–37. Second, they can effect very early transgene expression. For example, Santel et al demonstrated uptake of fluorescently labeled siRNA in mouse endothelium 20 minutes following systemic delivery³⁴. Intervention at the earliest stages of endothelial injury may promote EC integrity and help preserve the anti-inflammatory, anti-thrombotic effects of the healthy endothelium. In addition to targeting the early phases of VGD, non-viral vectors have induced sustained transgene expression for up to 8 weeks in a rabbit vein graft model³⁶. Whereas siRNA and short hairpin RNA (shRNA) can be used to silence expression of a target gene within the vein graft wall, ODNs represent a unique approach to gene therapy. These gene-targeting tools are typically composed of short double-stranded DNA sequences that mimic transcription factor binding sites in the genome, thereby acting as "decoys" to block transcription of genes involved in cellular proliferation. Importantly, ex vivo delivery of ODNs and siRNA has been shown to be safe in both pre-clinical and clinical studies³⁸.

Nanoparticles

Nanoparticles (NPs) are submicrometer-sized polymeric colloidal particles³⁹. The physicochemical characteristics of NPs are ideally suited for genetic modulation of vein grafts, as they allow for efficient intracellular delivery of transgenes, can effect sustained gene delivery, and have an excellent safety profile. Successful therapeutic intervention for VGD is highly dependent on efficient intracellular gene delivery. NPs have demonstrated increased intracellular uptake compared to microparticles in both in vitro and in vivo models^{40, 41}. Furthermore, certain polymeric formulations are able to escape lysosomal degradation via a reversal of surface charge, enabling entry into the cytosolic compartment⁴². Once within the cytosol the therapeutic agent of interest, such as a transgene, is slowly released from the polymeric matrix. In fact, polymer properties can be manipulated to regulate the release of their cargo over a predetermined time frame, thereby achieving the desired therapeutic effect^{43, 44, 41} Sustained gene delivery would be particularly helpful in suppressing SMC proliferation prior to re-endothelialization. Finally, unlike viral vectors, there is much less concern about the safety of NPs for gene delivery. The most extensively studied NP polymers are hydrolyzed to biologically compatible byproducts, which are eliminated by the citric acid cycle⁴⁵.

Although no studies to date have used polymeric NPs for gene delivery to prevent VGD, there is evidence to support their use in this context. First, investigators have demonstrated efficient uptake of polymeric NPs by SMCs and ECs, the primary cell types responsible for development of the neointimal lesion^{42, 46}. Additionally, Cohen-Sacks et al demonstrated that NP delivery of a PDGF receptor B antisense RNA inhibited neointimal hyperplasia in a rat carotid artery injury model⁴⁷. Taken together, these studies support the use of NPs as a promising approach for gene delivery to prevent VGD.

Viral vectors

Viral vectors used in vein graft models include adenovirus (Ad), adeno-associated virus (AAV), and lentivirus. Each vector has distinct cell tropism, transgene expression profiles (Fig 1), and safety concerns that impact its potential utility for prevention of VGD.

Adenovirus

Adenoviruses (Ad) are non-enveloped double-stranded DNA viruses that are, in many respects, ideally suited for use in VGD. Ad vectors have large packing capacities, efficiently infect both dividing and quiescent cells, and begin to express the transgene within hours after infection^{33, 48}. The ability to target non-dividing cells is a feature common to all of the vectors discussed here and is particularly important given that target cells within the vessel wall are typically quiescent and do not become activated until the time of graft harvest, when the vector will be delivered. However, because SMCs become activated rapidly after removal of the vein graft from its native location, the ability to infect proliferating cells is an advantage. However, there are some aspects of adenovirus biology that may limit its clinical use. Unlike the mostly seronegative animal models used in pre-clinical vein graft studies, approximately 60% of the human population is seropositive for antibodies against adenovirus⁴⁹. Thus, if Ad vectors are delivered systemically in humans, it is likely that infection efficiency would be compromised. First-generation adenoviruses are also quite immunogenic as a result of their large (~30 kb) genome and concomitant expression of viral genes in infected cells. Cells expressing adenoviral proteins are aggressively eliminated by cytotoxic T cells, resulting in limitation of transgene expression, endothelial injury, and potential exacerbation of intimal hyperplasia^50–52. Moreover, the viral promoters commonly used in gene therapy vectors are rapidly methylated, resulting in promoter silencing and inhibition of transgene expression^53–55. In vivo gene expression with adenoviral vectors typically peaks by day 7 and is gone by day 28⁵⁶. Recently, attempts to improve vascular gene therapy have utilized third-generation or helper-dependent adenoviruses (HDAd), in which all viral genes are deleted, thereby decreasing immunogenicity^{57, 58}. HDAd vectors have demonstrated expression for up to 8 weeks after infection of rabbit carotid arteries but with significantly less inflammation and intimal hyperplasia than first-generation Ad⁵⁹. To our knowledge, however, HDAd has not been tested in a vein graft model. Finally, there are legitimate safety concerns regarding the use of adenovirus in human patients stemming from a patient death during a clinical trial of gene therapy using direct intra-hepatic adenovirus administration, although there are important differences in the mode of gene delivery to vein grafts (discussed below)^{60, 61}.

Adeno-associated virus

Adeno-associated viruses are single-stranded DNA viruses of the parvovirus family⁶². Several features of AAV suggest that it would be preferable to adenovirus for vascular gene therapy^{63, 64}. AAVs effect long-term gene expression (months to years), making them particularly useful in targeting SMC proliferation, which continues for weeks to months following engraftment^{30, 65}. Persistent AAV-mediated transgene expression and inhibition of SMC proliferation and migration may help prevent late graft failure, provided the target cells are infected at the time of gene delivery. Recombinant AAVs are deleted of all viral genes and therefore have substantially lower immunogenicity than adenovirus, and they have shown an excellent safety profile in pre-clinical and clinical trials^66–69. Despite these qualities, most currently used AAV vectors are suboptimal for targeting vein grafts, primarily due to their lack of native tropism for ECs and SMCs, the primary cellular elements of the intimal lesion^{70, 71}. This issue can be addressed by modifications of the AAV capsid, which dramatically enhance transduction of ECs and SMCs^70–72. A potential limiting factor in AAV-mediated gene delivery for vein graft disease is the relatively slow onset of transgene expression compared to that of Ad vectors. Low levels of transgene expression during the first few weeks following EC injury may be insufficient to prevent progression of SMC growth and proliferation. This delay in gene expression arising from recombinant AAV vectors is due to the time required to convert its single-stranded DNA genome into a double-stranded form amenable to transcription. Delayed onset of AAV-mediated transgene expression can be circumvented by the use of self-complementary AAV vectors (scAAV, reviewed in⁷³), which contain a DNA motif that enables folding into a double-stranded DNA genome without the requirement for DNA synthesis. To date, scAAV has not been evaluated for vascular gene delivery.

Lentivirus

Lentiviruses are derived from the retrovirus family. Lentiviral vectors efficiently transduce vascular tissues, have low immunogenicity, and yield long-lasting transgene expression^74–76. However, the potential for insertional mutagenesis is a major safety concern, which dampens enthusiasm for the use of lentiviral vectors in humans⁷⁷. Non-integrating vectors have been developed, but their safety and efficacy have yet to be evaluated in the vasculature⁷⁸.

Vector delivery strategies

It is important to note that many of the concerns and/or weaknesses related to the use of viral vectors are circumvented in the case of vein graft treatment simply by virtue of the vector deliver strategy (Fig 2). Because vein grafts are typically treated ex vivo after removal of blood from the vein, there is a low likelihood of antibody-mediated clearance of the vector prior to infection of the vessel wall. Furthermore, systemic gene therapy is likely to require in vivo delivery of high titers of virus. Because vein graft treatment involves highly localized tissue infection, lower titers of virus can be used to effect efficient transgene expression, thus there is a reduced probability of systemic toxicity.

Delivery of gene transfer vectors to vein grafts has been accomplished in a variety of ways. They have been applied extra-luminally, intra-luminally, or both, and delivered with or without pressurization of the vein. The primary approach to extra-luminal vector delivery is via application of a gel (e.g., poloxamer) containing the transgene and a transfection reagent to the adventitia. Diffusion of the transfection complex into the media and intima allows expression throughout the vessel wall. This technique has been used successfully in animal models of vein grafting to deliver siRNA targeting the proliferative proteins STAT-3 and c-myc and to deliver cDNA encoding platelet-derived endothelial cell growth factor (PD-ECGF)^{37, 79, 80}. Extra-luminal delivery appears to be safe in animal models, and the application of gel lacking transgene does not exacerbate IH⁸⁰. Intra-luminal delivery has been used for both viral and non-viral vectors. Following vein harvest, the gene transfer solution is infused into the vein lumen and the graft is typically incubated in buffer at room temperature for 20–30 minutes with a distension pressure of approximately 10 mmHg. Following infection, gentle flushing of the vein with heparinized saline removes the majority of unincorporated viral particles^{33, 81, 82}. This strategy has demonstrated excellent transduction efficiency and safety, however there is some concern that higher distending pressures may worsen IH⁸³. Non-distending pressure-mediated transfection is a hybrid approach that delivers viral or non-viral vectors both intra- and extra-luminally. The gene transfer solution is introduced into the lumen and also bathes the vein, which is incubated in a chamber that is pressurized to 300 mmHg for 10 minutes without distention. This method has been investigated extensively and was used to deliver an ODN decoy in the PREVENT trials, the only clinical trials to date of gene therapy for VGD^{14, 38, 84, 85}. The PREVENT trials are discussed more extensively below.

POTENTIAL MOLECULAR THERAPEUTIC TARGETS IN VEIN GRAFT DISEASE

The pathogenesis of VGD presents many potential molecular targets for gene therapy. Vascular quiescence is driven largely by a healthy endothelium, which inhibits vascular inflammation, thrombosis, and pathological vascular SMC growth through the production of vasoprotective factors, and these processes are adversely affected as a result of vein graft harvesting and subsequent endothelial injury (Fig 1). Accordingly, pre-clinical and clinical studies have utilized gene delivery approaches to target molecular mediators of many different steps in the development and progression of vein graft disease. The primary cellular targets in most of these studies have been endothelial and vascular SMCs, however the relative importance of each cell type in the prevention of VGD remains unclear. In theory, maintenance of endothelial integrity would prevent thrombosis, inflammation, and subsequent SMC injury and neointima formation. Alternatively, interventions that target SMCs directly would prevent the injury-induced shift toward a more proliferative, synthetic phenotype, preventing neointima formation irrespective of endothelial damage and until re-endothelialization of the vein graft can occur. Although it is not known whether one approach is superior, it is important to consider the dynamic nature of different cells within the remodeling vein graft. At least one study has suggested that loss of Ad vector-mediated transgene expression is due in part to SMC proliferation within the neointima and gradual replacement of transduced cells⁸⁶, which could affect the functional outcome of gene delivery. This section will review both pre-clinical and clinical studies of gene therapy for VGD in the context of these mechanistic considerations, and these studies are summarized in Table I.

Table I.

Summary of pre-clinical and clinical studies of gene delivery modalities for the prevention of vein graft disease.

Pre-Clinical Studies
Target	Approach	Vector	Model	Delivery Modality	Summary of Results	Reference
EC Health	eNOS overexpression	HVJ-liposome	Canine femoral artery bypass	Intraluminal distending pressure-mediated transfection	>50% reduction in intimal thickness at 4 weeks	94
	eNOS overexpression	HVJ-liposome	Hypercholesterolemic rabbit jugular vein-carotid artery bypass	Intraluminal distending pressure-mediated transfection	30% reduction in intimal thickness at 4 weeks	95
					Reduction in neointimal macrophages at 4 weeks, decreased PCNA positive cells at 2 weeks
	COX-1 (PGI2) overexpression	Adenovirus	Hypercholesterolemic rabbit jugular vein-carotid artery bypass	Intraluminal without distension	No difference in IH	100, 101
					Increased luminal area, blood flow, and PGI2 at 4 weeks
	EC-SOD overexpression (in combination with 35K and TIMP-1)	Adenovirus	Rabbit jugular vein-carotid artery bypass	Intraluminal distending pressure-mediated transfection	Reduced I/M ratio at 2 weeks in combination with 35K and at 2 and 4 weeks with TIMP-1	108
					Reduced macrophage infiltration and BrDU staining in combination with 35K at 4 weeks
					Slight reduction in superoxide anion production at 2 weeks with EC-SOD alone
Thrombosis	Thrombomodulin overexpression	Adenovirus	Rabbit jugular-carotid artery bypass	Intraluminal distending pressure-mediated transfection	No difference in IH	118
					Reduced thrombin generation at 7 days
	tPA overexpression	Adenovirus	Porcine saphenous-carotid artery bypass	Intraluminal without distension	Reduction in flow-restricting thrombi at 1 day	122
Inflammation	MCP-1 inhibition (7ND-MCP-1)	Naked plasmid	Hypercholesterolemic mouse donor IVC-carotid artery bypass	Systemic expression (IM injection and electroporation)	51% decrease in IH at 4 weeks	135
					No change in macrophage infiltration
	MCP-1 inhibition (7ND-MCP-1)	Adenovirus	Canine jugular-carotid artery bypass	Submersion in adenoviral solution without pressurization	Decreased IH and lumen stenosis at 4 weeks	136
					Decreased PCNA positive cells and macrophages at 4 weeks
	MCP-1 inhibition (CCR2 silencing)	Lentiviral shRNA	Hypercholesterolemic mouse donor IVC-carotid artery bypass	Adventitial application of bioprotein gel	38% reduction in IH at 4 weeks	137
	CC-Chemokine inhibition (35K)	Adenovirus	Hypercholesterolemic mouse donor IVC-carotid artery bypass	Systemic expression (IV injection 2 days pre-op)	Reduced wall thickness, Ki67 positive cells, SMA positive cells, and macrophage infiltration at 2 and 4 weeks	144
	CC-Chemokine inhibition (35K)	Adenovirus	Rabbit jugular-carotid artery bypass	Intraluminal distending pressure-mediated transfection	Reduced I/M ratio at 2 weeks	145
					Reduced macrophage infiltration and BrdU incorporation at 4 weeks
	NFκB inhibition	ODN	Canine aortocoronary bypass	Non-distending pressure-mediated transfection	Decreased intimal area and PCNA positive cells at 4 weeks	139
	NFκB inhibition	ODN	Rabbit hypercholesterolemic jugular vein-carotid artery bypass	Non-distended; pressure-mediated transfection	Decrease in IH at 4 weeks	35
					Improved endothelium-mediated vasorelaxation
	PD-ECGF overexpression	Naked plasmid	Rabbit jugular-carotid artery bypass	Adventitial application of bioprotein gel	Reduced intimal thickness at 8 weeks	79
					Reduced PCNA positive SMC at 2 weeks
VSMC	Gβγ inhibition	Adenovirus	Canine aortocoronary bypass	Intraluminal pressure-mediated delivery	Reduced intimal area and I/M ratio at 3 months	82
	PTEN overexpression	Adenovirus	Canine aortocoronary bypass	Intraluminal pressure-mediated delivery	Reduced intimal area and I/M ratio at 3 months	33
	TIMP-3 overexpression	Adenovirus	Porcine saphenous-carotid artery bypass	Intraluminal delivery without distention	58% reduction in IH at 28 days	152
	TIMP-3 overexpression	Adenovirus	Porcine saphenous-carotid artery bypass	Intraluminal delivery without distention	Reduced IH at 3 months	153
	PCNA inhibition; cdc2 inhibition	HVJ liposome-ODN	Rabbit jugular-carotid artery bypass	Intraluminal; pressure-mediated transfection	Increased vasoresponsiveness to acetylcholine at 6 weeks	156
	E2F inhibition	ODN	Rabbit hypercholesterolemic jugular vein-carotid artery bypass	Non-distended; pressure-mediated transfection	Decrease in IH at 6 months	166–169
Clinical Studies
Trial	Approach	Vector	Model	Delivery Modality	Summary of Results	Reference
PREVENT I (Phase I)	E2F inhibition	ODN	Human infrainguinal arterial bypass with autologous vein	Non-distended; pressure-mediated transfection	Decrease graft occlusions, revisions, or critical stenoses at 12 months	38
PREVENT II (Phase IIb)	E2F inhibition	ODN	Human coronary artery bypass grafting with autologous vein	Non-distended; pressure-mediated transfection	30% relative reduction in critical stenosis at 12 months	170
PREVENT III (Phase III)	E2F inhibition	ODN	Human infrainguinal arterial bypass with autologous vein	Non-distended; pressure-mediated transfection	No significant difference in primary graft patency or limb salvage at 12 months	171
PREVENT IV (Phase III)	E2F inhibition	ODN	Human coronary artery bypass grafting with autologous vein	Non-distended; pressure-mediated transfection	No significant difference in critical graft stenosis at 12 months	14

Abbreviations: CCR2, C-C chemokine receptor-2; COX-1, cyclooxygenase-1; EC, endothelial cell; EC-SOD, extracellular superoxide dismutase; eNOS, endothelial nitric oxide synthase; HVJ, hemagglutinating virus of Japan; IH, intimal hyperplasia; I/M, intima to media; IVC, inferior vena cava; MCP-1, monocyte chemoattractant protein-1; NFκB, nuclear factor κB; ODN, oligodeoxynucleotide; PCNA, proliferating cell nuclear antigen; PD-ECGF, platelet-derived endothelial cell growth factor; PGI₂, prostaglandin I₂; PREVENT, Project of Ex Vivo Vein Graft Engineering via Transfection; PTEN, phosphatase and tensin homology deleted on chromosome 10; SMA, smooth muscle actin; TIMP, tissue inhibitor of metalloproteinase; tPA, tissue plasminogen activator; VSMC, vascular smooth muscle cell.

Endothelial injury and loss of endothelial vasoprotective factors

Endothelial injury and denudation during vein graft harvest and arterial engraftment results in loss of the vasoprotective substances that are normally produced by healthy ECs. Two of these factors, the gas nitric oxide (NO) and the eicosanoid prostacyclin (PGI₂), have been targeted using gene therapy approaches. Pharmacological approaches to modulating these pathways are reviewed in⁸⁷ and⁸⁸.

Endothelial nitric oxide synthase (eNOS), one of the enzymes that synthesizes NO, is constitutively expressed in normal vascular endothelium and is critical in modulating several aspects of vascular physiology. NO released from the endothelium induces vasodilation, prevents SMC proliferation, and inhibits adhesion of platelets and leukocytes to the endothelium^89–91. Impaired release of NO from the damaged vein graft endothelium is therefore thought to contribute to the initiation of IH. This hypothesis is supported by studies of adenovirus-mediated overexpression of bovine eNOS (Ad-eNOS) in the endothelium and adventitia of human saphenous vein segments. Saphenous veins treated with Ad-eNOS demonstrated increased NO release and improved vascular function in a ring relaxation test⁹². In another study from the same group, human saphenous vein segments were cultured ex vivo and IH was assessed histologically. Vein segments treated with Ad-eNOS had reduced intima to media ratios compared to untreated vein segments⁹³. These findings were applied in animal models of vein graft disease with moderate success. Intraluminal transfection of bovine eNOS cDNA encapsulated in hemagglutinating virus of Japan (HVJ)-liposomes reduced intimal thickness at 4 weeks post-graft in both canine and hypercholesterolemic rabbit models^{94, 95}. However the reduction in intimal thickness was modest, particularly in the rabbit model (30% reduction vs. control). This may be due to damage to the endothelium and media resulting from the transfection process, which utilized distending pressure⁸³.

PGI₂ is also constitutively produced by healthy endothelium. PGI₂ is a potent vasodilator and inhibitor of platelet aggregation, leukocyte adhesion, and SMC proliferation^96–98. Cyclooxygenase-1 (COX-1), the rate-limiting enzyme in the PGI₂ synthesis pathway, must be continuously expressed, as its catalytic mechanism results in its rapid degradation⁹⁹. Adenoviral overexpression of COX-1 in a porcine carotid angioplasty model increased PGI₂ production and prevented arterial thrombosis. However, treatment of rabbit vein grafts with adenovirus encoding human COX-1 had no effect on intimal thickness at four weeks despite improved PGI₂ production and blood flow through the graft^{100, 101}. Although COX-1 failed to prevent IH, these results highlight that, although intimal thickness is a useful indicator of vein graft disease, it does not always correlate with graft function (i.e., perfusion).

Even in intact ECs, endothelial function can be impaired due to generation of reactive oxygen species (ROS) following graft reperfusion (reviewed in^{16, 17, 88}). The principle ROS studied in this context is superoxide (O₂⁻), which reacts with NO to form the weaker vasodilator peroxynitrite (ONOO⁻). ONOO⁻ formation further disrupts NO production by contributing to eNOS uncoupling and impairing PGI₂ production through inhibitory tyrosine nitration of PGI₂ synthase^{102, 103}. The degradation of O₂⁻ is initiated by superoxide dismutases (SOD), which catalyze the conversion of O₂⁻ to hydrogen peroxide (H₂O₂), which is then degraded by the enzyme catalase. Several groups have demonstrated that, compared to SVGs, arterial grafts have greater SOD activity and NO release at the time of harvest^{104, 105}. In a porcine jugular-carotid interposition model, vein grafts had increased O₂⁻ production, decreased SOD activity, and greater lipid accumulation compared to arterial grafts at 1 month¹⁰⁶. These differences in ROS processing may contribute to the increased patency of arterial grafts. Intraoperative PEG-SOD treatment in a rabbit jugular-carotid interposition model reduced the I/M ratio and increased luminal area of the graft at 28 days, but no improvement in vasomotor function was observed¹⁰⁷. Intraoperative delivery of adenovirus encoding extracellular-SOD (AdEC-SOD) reduced macrophage accumulation but not I/M ratio or BrdU incorporation at 4 weeks in a rabbit model¹⁰⁸. However, combination of AdEC-SOD with other genes was more effective at reducing I/M ratio and proliferation in the graft wall than either therapy alone (see Table I). Catalase gene therapy improves liver function following IR injury in a mouse model, but has not been applied to models of VGD¹⁰⁹. Likewise, other anti-oxidant enzymes (such as heme oxygenase-1 and glutathione peroxidase) have been used to reduce atherosclerosis and arterial neointimal hyperplasia in a number of animal models but have not been tested in VGD (reviewed in^{110, 111}).

Thrombosis

In addition to the loss of endothelial vasoprotective substances, endothelial injury and subsequent denudation exposes the basement subendothelial matrix to circulating platelets, which initiates the coagulation cascade. The release of coagulation factors results in thrombosis and graft occlusion, and release of certain factors exacerbates IH by promoting SMC proliferation and migration^112–114. Anticoagulation reduces the risk of early graft thrombosis, but does not eliminate it. Furthermore, systemic anticoagulation is associated with side effects, primarily bleeding complications^{115, 116}, which is particularly dangerous in the post-operative period. Gene therapy to inhibit vein graft thrombosis would circumvent these systemic side effects.

There have been comparatively few studies using gene therapy to target thrombosis. The two primary targets have been the anti-coagulant proteins thrombomodulin (TM) and tissue plasminogen activator (tPA). TM binds to thrombin, inactivating its pro-coagulant effects and activating the antithrombotic protein C pathway. TM activity in vein grafts decreases dramatically within 90 minutes of engraftment into the arterial circulation^{117, 118}. Cationic liposome-mediated delivery of TM improved graft thromboresistance in a rat model¹¹⁹. However, adenoviral delivery of human TM in a rabbit model had no effect on graft wall thickness at 6 weeks¹¹⁸. Similar results were obtained using pharmacological coagulation inhibitors. For example, aspirin applied locally to vein grafts attenuated acute thrombosis in a mouse model, and although systemic aspirin reduced 1-year clinical vein graft failure rates, it had no effect on long-term (3-year) graft patency^{120, 121}. The other model tested the ability of tPA, the primary intravascular activator of plasminogen, to promote graft thrombolysis. Like TM, tPA expression is decreased after engraftment. Adenoviral overexpression of tPA in a porcine model decreased flow-restricting thrombi 1 day after engraftment¹²². However, the effect of tPA on IH is unclear. In a rabbit arterial injury model, Ad-tPA treatment promoted IH, possibly due to ECM degradation by tPA¹²³. Conversely, in a rabbit vein graft model Ad-tPA reduced IH at 4 weeks¹²⁴. These results may be explained by differences in the pathogenesis of IH in arteries and veins, and reinforce the importance of studying gene delivery in disease-specific models. However, further studies are required to ensure that tPA overexpression does not have adverse effects on IH in VGD models.

Inflammation

In the week following graft implantation, platelet activation at sites of endothelial damage initiates an inflammatory cascade in the vein graft wall that promotes SMC proliferation, intimal thickening, and accelerated atherosclerosis³¹. P-selectin glycoprotein ligand-1 (PSGL-1) on the surface of rolling leukocytes interacts with P-selectin on adherent platelets, activating leukocyte migration into the vessel wall¹²⁵. Migration is mediated by the engagement of leukocyte transmembrane proteins with the endothelial counter-receptors ICAM-1 and VCAM-1. Activated leukocytes, ECs, and SMCs secrete a number of cytokines and growth factors that potentiate the inflammatory response, including MCP-1, IL-1, IL-6, IL-8, and TNFα. The molecular mechanisms of endothelial cell interaction with the adaptive and innate immune system are reviewed in detail elsewhere^126–128. Many of these adhesion molecules and cytokines have been shown to play a role in the development of VGD, and several of these genes have been used in VGD gene therapy approaches.

In a rat VGD model, pharmacological macrophage depletion after engraftment decreased MCP-1 and TGFβ expression in the graft and decreased intimal thickness at 4 weeks¹²⁹. This led to several approaches to block macrophage recruitment, with a focus on MCP-1, a potent monocyte/macrophage chemokine secreted by SMCs, ECs, fibroblasts, and leukocytes¹³⁰. In humans, elevated MCP-1 (both circulating and localized to atherosclerotic lesions) correlates with increased risk of restenosis in patients undergoing percutaneous angioplasty^{131, 132}. In rat vein graft models, MCP-1 is upregulated within hours of engraftment, peaks during the first week, and persists for at least 8 weeks^{129, 130}. 7ND-MCP-1 is a dominant negative N-terminal deletion mutant of MCP-1 that binds to the MCP-1 receptor, CCR2, but does not promote monocyte/macrophage migration^{133, 134}. Treatment of veins with naked plasmid DNA encoding 7ND-MCP-1 inhibited SMC proliferation in vitro and resulted in an approximately 50% reduction in IH in an ApoE3 Leiden mouse model and in cultured segments of human saphenous veins¹³⁵. These results were confirmed in a canine model in which Ad-7ND-MCP-1 treatment reduced IH without impairing re-endothelialization¹³⁶. Another innovative approach to blocking the MCP-1 pathway used lentivirus to deliver CCR2 shRNA to grafts in an ApoE3 Leiden mouse model. Treated veins had decreased SMC proliferation, macrophage recruitment, and graft wall thickness¹³⁷.

The transcription of many of the inflammatory cytokines and adhesion molecules described above is upregulated by NFκB, a transcription factor that is activated by inflammatory stimuli and other cell stressors. The role of NFκB in vascular inflammation (with a focus on atherosclerosis) has been reviewed previously¹³⁸. Pressure-mediated transfection of the lumen and adventitia of vein grafts with an NFκB decoy ODN in canine and hypercholesterolemic rabbit models decreased intimal thickness and VSMC proliferation^{35, 139}. The reduction in the intima to media (I/M) ratio in the rabbit study was modest, particularly in light of the increased medial area in the NFκB ODN-treated group, but the treatment did significantly inhibit macrophage recruitment to the graft wall. Interestingly, the NFκB decoy-treated grafts also displayed improved acetylcholine-induced vasodilation at 4 weeks post-implantation. This relaxation was inhibited by the potent nitric oxide synthase inhibitor L-NAME, indicating that the NFκB ODN may preserve some of the vasoprotective functions of the surviving endothelium³⁵.

Another novel approach to broadly inhibiting the inflammatory response utilized the vaccinia virus protein, 35K. 35K is a soluble poxvirus protein that binds to CC chemokines such as members of the MCP family and RANTES, blocks their interactions with chemokine receptors, and prevents their physiological effects in vitro and in vivo^140–142. Intravenous injection of adenovirus encoding 35K (Ad35K) decreased systemic atherosclerosis in ApoE−/− mice fed a high fat diet and reduced vein graft wall thickness, proliferation, and macrophage infiltration in ApoE−/− mice fed standard chow^{143, 144}. The results of the vein graft study are striking at 2 weeks post-implantation but seem to diminish between 2 and 4 weeks, possibly due to the use of an adenoviral expression vector. Similar results were reported in a rabbit vein graft model, in which intraluminal infection of the graft with Ad35K resulted in modest decreases in I/M ratio but dramatic decreases in macrophage infiltration¹⁴⁵. A membrane-targeted mutant of 35K, m35K, improved CC chemokine inhibition in murine peritonitis and hepatitis models, but has not been tested in a vein graft model, in which sequestration of m35K to the membrane may prove particularly useful¹⁴⁶.

Finally, several studies have demonstrated that inflammation of the adventitia contributes to neointimal thickening in arterial balloon injury models^147–149 (reviewed in¹⁵⁰). Using an "outside-in" approach, genes that antagonize growth factor signaling and ECM remodeling were delivered directly to the adventitia and successfully reduced IH after arterial injury. In a rabbit vein graft model, poloxamer hydrogel-mediated delivery of the gene encoding PD-ECGF to the graft adventitia resulted in efficient transgene expression and decreased intimal thickness at 8 weeks⁷⁹. Several reports discussed in this review utilized protocols that involve submersing grafts in vector-containing solution during gene delivery, resulting in gene transfer to the adventitia as well as the lumen, but to our knowledge no studies have utilized perivascular gene transfer to target adventitial inflammation exclusively for the prevention of vein graft IH.

Smooth muscle cell proliferation and migration

SMCs are the predominant cells within the neointimal lesion. The goal of gene therapy targeting SMCs is to restore cellular quiescence and the contractile phenotype. In the quiescent vasculature, SMC proliferation and migration are inhibited in part by integrin-mediated contacts with the ECM¹⁵¹. Following SMC activation, matrix-degrading metalloproteinases (MMPs) digest the ECM, facilitating SMC proliferation and migration into the neointima. George et al demonstrated that overexpression of tissue inhibitor of metalloproteinase (TIMP)-3 significantly reduced IH in both short-term (28-day) and long-term (3-month) porcine vein graft models.^{152, 153}

There is a significant body of literature demonstrating that, following vein graft anastomosis, increased wall stress upregulates fibroblast growth factor (FGF) and platelet-derived growth factor (PDGF)¹⁵⁴, both of which are potent SMC mitogens and chemoattractants¹⁵⁵. Increased arterial wall stress and shear forces also directly activate growth factor receptors²¹. The ultimate effect of increased growth factor signaling is to promote cell cycle progression and SMC proliferation, thereby enhancing intimal hyperplasia. Mann et al targeted SMC proliferation in a rabbit model of jugular vein-carotid artery interposition grafting by inhibiting expression of the cell cycle regulatory genes proliferating cell nuclear antigen (PCNA) and cdc2 kinase using antisense ODN¹⁵⁶. HVJ-liposomes were used to deliver ODN complexes to the vein graft wall by pressure-mediated transfection, resulting in a reduction in intimal hyperplasia and inhibition of diet-induced atherosclerosis. Many of the mitogens responsible for SMC proliferation act via G protein-coupled receptors (GPCRs). Members of our group used a first-generation adenovirus to express a dominant negative peptide inhibitor derived from the carboxyl terminus of the β-adrenergic receptor kinase (βARKct) in a canine model of aortocoronary bypass surgery. Inhibition of Gβγ signaling with this approach significantly reduced IH at 90 days.⁸²

Recently, our group has focused on the phosphoinositide 3-kinase (PI3K) pathway as a target for inhibition of intimal hyperplasia in models of both arterial injury and vein graft disease^{33, 157, 158}. PI3K is activated by most growth factor receptors and GPCRs and acts through multiple downstream effector molecules to promote cellular growth, proliferation, migration, and survival^{159, 160}, all of which play a role in intimal hyperplasia and vein graft disease (Fig 3A). The therapeutic benefit of targeting the PI3K pathway has been proven in clinical trials of sirolimus-eluting stents, which potently inhibit intimal hyperplasia and restenosis following percutaneous coronary intervention^{161, 162}. Sirolimus (rapamycin) inhibits mammalian target of rapamycin (mTOR), which is an important PI3K effector molecule that regulates protein translation and cell growth¹⁶³. We reasoned that inhibition of PI3K signaling more broadly would have greater effects on intimal hyperplasia than targeting only one of PI3K's many effector molecules. Therefore, we generated a first-generation adenovirus encoding PTEN (phosphatase and tensin homology deleted on chromosome 10), a lipid phosphatase that is a natural PI3K antagonist (Fig 3A)¹⁶⁴. PTEN hydrolyzes the 3-phosphoinositide lipid products of PI3K, which are potent signaling molecules, thereby abrogating all PI3K signaling and cellular responses involved in intimal hyperplasia. We first demonstrated that adenovirus-mediated overexpression of PTEN markedly inhibited intimal hyperplasia in a rat carotid injury model as a result of increased apoptosis and decreased proliferation of SMCs¹⁵⁸. We subsequently showed that AdPTEN delivery to vein grafts significantly inhibited intimal hyperplasia in a 90-day canine aorto-coronary bypass graft model (Fig 3B–C)³³. Notably, the PTEN transgene could be detected by PCR only in the treated vein graft but not in any of a number of other tissues, demonstrating that ex vivo delivery of AdPTEN poses a low risk of systemic infection (Fig 3D).

THE PREVENT TRIALS

As we have highlighted in this review, there is a substantial body of work demonstrating pre-clinical efficacy of a variety of gene therapy approaches in different animal models. However, only one strategy, the E2F decoy edifoligide, has been evaluated in clinical trials. E2F is a transcription factor responsible for the upregulation of several genes associated with cell cycle progression¹⁶⁵. Edifoligide is a 14-base pair double-stranded ODN designed from the E2F binding site in its target genes. When delivered to cells or tissues, edifoligide acts as a decoy binding site, sequestering E2F from genomic DNA and thereby preventing target gene expression and cellular proliferation³⁶. Pre-clinical studies in rabbit models and human vein segments demonstrated that edifoligide treatment inhibited SMC proliferation, preserved endothelial function, limited the development of atherosclerosis, and significantly reduced IH^166–169. These exciting data led to the development of a series of clinical trials that were part of the Project of Ex Vivo Vein Graft Engineering via Transfection (PREVENT) trials. PREVENT I was a phase I trial that demonstrated the safety and biologic efficacy of E2F decoy gene delivery in 41 patients undergoing lower extremity bypass surgery³⁸. In PREVENT II, a phase IIb trial of 200 patients undergoing coronary artery bypass grafting, the treatment group demonstrated a 30% relative reduction in vein graft critical stenosis by angiography and a 30% reduction in total wall volume by intravascular ultrasound¹⁷⁰. Based on these positive results, the E2F decoy strategy was investigated in two separate phase III trials of vein graft disease: PREVENT III in lower extremity bypass surgery and PREVENT IV in coronary artery bypass grafting. Unfortunately, neither study demonstrated a significant difference in the primary endpoints of primary graft patency and critical graft stenosis at 1 year^{14, 171}.

The reasons underlying the failure to translate the pre-clinical success of edifoligide into clinically relevant endpoints are not entirely clear. However, there are several factors that may have contributed. First, E2F is not a single transcription factor, but rather a family of 8 isoforms (E2F1-8). E2F1-3 are transcriptional activators, while E2F4-8 are repressors. E2F biology is reviewed in detail in¹⁷². Edifoligide was designed to bind to the activating E2F isoforms, but may have inadvertently bound inhibitory E2F isoforms due to their highly conserved DNA binding domains³⁶. Moreover, even traditionally activating isoforms can act as context-dependent transcriptional repressors¹⁷³. Another surprising result from the PREVENT IV trial was the higher than expected incidence of the primary end point of angiographic vein graft failure, defined as ≥75% stenosis at 12–18 months. Compared to the historical graft failure rate of 15–20% at one year, 45.2% of grafts in the edifoligide group and 46.3% in the placebo group reached the primary end point¹⁷⁴. Interestingly, the study lacked a "standard of care" control group; both the treatment and placebo groups were exposed to 310 mmHg of non-distending pressure, which may have exacerbated endothelial injury¹⁴. The high failure rate in both groups raises concerns that the pressure-mediated delivery system contributed to the development of IH. Finally, there is accumulating evidence that graft extrinsic cells play a large role in formation of the neointimal lesion^{175, 176}. The treatment strategy in the PREVENT trials, and in many of the pre-clinical trials discussed above, primarily targets graft intrinsic cells.

ADDITIONAL CONSIDERATIONS AND CONCLUSIONS

Cardiovascular disease remains the leading cause of morbidity and mortality in the developed world^1,161, and surgical revascularization with autologous veins is likely to remain a mainstay of therapy for patients with arterial occlusive disease until such time as improved therapies for atherosclerosis become available. Therefore, it is critical that advances in our mechanistic understanding of VGD are efficiently translated into the clinic. Although numerous gene therapy approaches have effectively reduced IH and other aspects of VGD in animal models, the pre-clinical success of edifoligide did not translate into clinically meaningful reductions in VGD. In addition to potential explanations for failure of the PREVENT trials already discussed, additional concerns must be addressed before moving promising treatment strategies into human trials of vein graft disease.

First, to properly assess the promise of pre-clinical studies, endpoints in animal models must correlate with clinically relevant endpoints in patients. Many of the published pre-clinical studies assessed IH as the primary endpoint, but this may not be the most clinically meaningful endpoint. Analysis of PREVENT IV demonstrates that critical graft stenosis is associated with an increased need for repeat vascularization but not an increased risk of death and/or myocardial infarction¹⁷⁷. Other readouts such as graft atherosclerosis, measures of systemic inflammation, and plaque stability may be more meaningful, particularly in late graft failure¹⁷⁸. Focusing exclusively on IH in animal models may exclude promising therapies that may have modest effects on IH but dramatic effects on other aspects of VGD (for example macrophage recruitment). Moreover, many of the studies reviewed here demonstrate that increased graft wall thickness does not necessarily reflect poor graft function.

Second, although several large and small animal models recapitulate the pathogenesis of VGD on both gross and histological levels¹⁷⁹, more work must be done to define the molecular pathways responsible for VGD in humans to determine the extent to which animal models accurately represent these molecular mechanisms. For example, while there is a wealth of data on inflammatory pathways in mouse models of VGD, there are significant differences between humans and mice in the cellular expression patterns of adhesion molecules, inflammatory cytokines, and receptors¹²⁶. These and similar issues may lead to disappointing results when animal studies are translated into the clinic.

Finally, even if gene therapy for VGD is successfully applied in the clinic, it is unlikely that targeting a single pathway will be successful in all patients. Moreover, targeting different pathways and cell types simultaneously may be synergistic, as in the case of SOD, TIMP-1, and 35K combination gene therapy discussed above^{108, 145}. Similarly, combining gene therapy with other approaches may yield better outcomes than either approach alone. Potential examples include genetic modification of progenitor cells used to seed bioengineered grafts; genetic modification of grafts to promote progenitor cell homing (used successfully in a rat model of chronic hind limb ischemia¹⁸⁰); the use of sheaths to provide structural support for vein grafts treated while serving as gene delivery vehicles; and the use of biological scaffolds for long-term delivery of gene therapy to engrafted veins.

Although the PREVENT trials did not lead to a clinical treatment, they demonstrated that gene therapy for VGD is feasible and generally safe. Continued efforts to characterize the molecular mechanisms of VGD in animal models and humans, combined with the development of more sensitive and clinically meaningful endpoints, will allow for more efficient translation of pre-clinical studies into successful therapies for VGD.

Abbreviations

Ad

adenovirus

AAV

adeno-associated virus

CABG

coronary artery bypass grafting

CCR2

C-C chemokine receptor-2

COX-1

cyclooxygenase-1

EC

endothelial cell

ECM

extracellular matrix

eNOS

endothelial nitric oxide synthase

HDAd

helper-dependent adenovirus

H₂O₂

hydrogen peroxide

HVJ

hemagglutinating virus of Japan

IH

intimal hyperplasia

IL

interleukin

I/M

intima to media

IR

ischemia-reperfusion

IVC

inferior vena cava

MCP-1

monocyte chemoattractant protein-1

mTOR

mammalian target of rapamycin

NFκB

nuclear factor κB

NO

nitric oxide

NP

nanoparticle

O₂⁻

superoxide

ODN

oligodeoxynucleotide

ONOO⁻

peroxynitrite

PCNA

proliferating cell nuclear antigen

PD-ECGF

platelet-derived endothelial cell growth factor

PDGF

platelet-derived growth factor

PGI₂

prostaglandin I₂

PI3K

phosphoinositide 3-kinase

PREVENT

Project of Ex Vivo Vein Graft Engineering via Transfection

PTEN

phosphatase and tensin homology deleted on chromosome 10

scAAV

self-complementary AAV

ROS

reactive oxygen species

siRNA

small interfering RNA

SMA

smooth muscle actin

SMC

smooth muscle cell

SOD

superoxide dismutase

STAT

signal transducer and activator of transcription

SVG

saphenous vein graft

TGFβ

transforming growth factor β

TIMP

tissue inhibitor of metalloproteinase

TM

thrombomodulin

tPA

tissue plasminogen activator

VCAM

vascular cell adhesion molecule

VEGF

vascular endothelial growth factor

VGD

vein graft disease

VSMC

vascular smooth muscle cell