Introduction
The prevalence of peripheral arterial disease (PAD) in the adult population is as high as 12–18%.^1,2 Conservative treatments, including risk factor modification, supervised exercise training and medical therapy are indicated in patients whose lower extremity PAD has not reached an advanced stage. Symptoms of intermittent claudication may improve with these maneuvers.³ However, pharmacologic treatments have had limited impact on the outcome of lower extremity PAD when the disease progresses to a state of critical limb ischemia (CLI). Percutaneous transluminal angioplasty (PTA) and reconstructive arterial surgery have their known limitations, with high restenosis and graft failure rates, especially in infra-popliteal disease. Hence, induction of therapeutic angiogenesis has been attempted in patients with advanced PAD in order to establish a safe method of revascularization that ensures stable, viable and durable vasculature. Most of the information about therapeutic angiogenesis in cardiovascular disease has been gained from its application in coronary heart disease, with significantly less information available on the treatment of PAD.^4,5 This review summarizes current clinical data on patients with PAD who underwent therapeutic angiogenesis in the context of phase I and phase II clinical trials.

Angiogenesis
The term, “angiogenesis,” used liberally in the cardiovascular community, refers to any form of new vessel growth. However, it is useful to distinguish between three different aspects of neovascularization: angiogenesis, arteriogenesis and vasculogenesis. Angiogenesis in its strictest definition describes capillary growth from enlarged venules, and is mainly stimulated by tissue hypoxia involving several mediators, including hypoxia-inducible factor (HIF)-1 alpha, vascular endothelial growth factor (VEGF) and angiopoietin-2.^6,7 Arteriogenesis is a process that produces fully developed arteries which, in contrast to angiogenesis, is usually large enough to be visualized with angiography.^8,9 Among the factors involved in this process are fibroblast growth factor (FGF), platelet-derived growth factor (PDGF) and VEGF. Finally, vasculogenesis is the formation of new vessels from endothelial and vascular progenitor cells, as seen in embryonic development. For the purpose of this review, therapeutic angiogenesis refers to the formation of both capillary structures and arterial conduits unless otherwise noted.

Potential Candidate Genes
Different classes of gene products involved in the physiologic process of angiogenesis are potential candidates for gene therapy. These include growth factors, transcription factors, certain chemokines and extracellular matrix proteins. VEGF and FGF are the best characterized angiogenic growth factors. The VEGF family is comprised of five closely related genes: VEGF-A to –D and PDGF.^10,11 VEGF-A has several isoforms that differ by their amino acid length, of which VEGF-A121 and VEGF-A165 are most widely studied in clinical trials. Of the 23 different FGF family members, FGF-1, FGF-2 and FGF-4 are highly angiogenic and thus have also been studied in several clinical trials.¹² Physiologically, VEGF and FGF are involved in the process of arteriogenesis, the de novo formation of true arterial conduits, but only VEGF appears to be an important factor in angiogenesis, according to its strictest definition.^6,13 HIF-1 alpha activates the transcription of several genes involved in the latter process, including VGEF and inducible nitric oxide synthase (iNOS), and consequently represents another possible target for gene therapy.¹⁴ Hepatocyte-growth factor (HGF) represents another potential candidate gene for therapeutic angiogenesis, which has proven to be a potent angiogenic growth factor.¹⁵ PR39, a peptide regulator for angiogenesis which upregulates HIF-1 alpha, the chemokine monocyte chemoattractant protein-1 (MCP-1), or extracellular matrix proteins such as Cyr61 are further potential candidates, all of which have not yet been evaluated in humans.^13,16,17

Product Formulation and Delivery
There are three factors to be considered in therapeutic angiogenesis: the purified protein versus gene delivery (DNA); the vector for delivery; and the route of delivery, which can be transcatheter/intra-arterial, systemic/intravenous, or application by direct intramuscular injection in the ischemic limb region. The advantages of protein formulations include predictable pharmacokinetics and tissue therapeutic levels, but the short half-life of these proteins limits the duration of exposure. Most of the earliest therapeutic angiogenesis studies employed recombinant formulations of angiogenic growth factors. Following discouraging results, a shift occurred towards gene transfer. Several advantages apply to the concept of gene transfer over the usage of recombinant proteins, including a more prolonged expression of the angiogenic factor. However, low level gene expression, induction of an inflammatory response and absence of regulation of gene expression have been significant limiting factors of this technique.^18,19

Modified viruses, plasmids and liposomes, as well as entire cells, can potentially be used as gene transfer vehicles. The most commonly used viral vector are modified replication incompetent adenoviruses, which deliver the specific gene into the host cell by its inherent mechanisms.^20,21 Other viral vectors include retroviruses, lentiviruses and adeno-associated viruses, which all integrate the gene into the host genome compared to adenoviruses where the therapeutic DNA remains extrachromosomal. When compared to adenovirus, the gene transfer efficiency of these viral vectors is very low.²² Although naked plasmid DNA or liposome complexes result in only a limited amount of cellular transfection, they are much easier to manufacture in large quantities than viral vectors, and thus have been used frequently as vehicles for therapeutic angiogenesis.^20,22 Altogether, the safety profile of adenoviral vectors is favorable, although inflammatory reactions, antibody formation, transient fever and elevated liver enzymes have been described in human trials. The same applies to plasmid DNA and liposomal complexes, which potentially induce inflammation, but to a lesser degree.²³ Finally, in a cell-based gene transfer approach, autologous cells are used as vectors after in vitro transfection with the target gene.²⁴ This promising concept has not yet been evaluated in humans.

The delivery route also affects the potency of therapeutic angiogenesis: the specific delivery of the gene or recombinant protein to the area of interest — the ischemic extremity, in the case of PAD. Thus, it is intuitive to assume that systemic intravenous application may be the least effective. Further, that systemic exposure of viral vectors should be avoided in order to prevent transfection and potential harmful induction of angiogenesis in non-target tissue. Successful selective intra-arterial, transcatheter gene transfer to the lower extremity circulation has been documented in humans.²⁵ Although cumbersome in the case of myocardial ischemia, intramuscular injection of angiogenic factors in PAD is a simple, efficient and specific way to induce vascular growth, which entirely avoids exposure of the gene or vehicle to the systemic circulation.²⁶ Another approach on the horizon is the injection of biodegradable microspheres coated with vectors of recombinant growth factors that can be targeted with ultrasonography techniques.^27,28

Phase I trials
The first case of therapeutic angiogenesis in a patient with PAD was reported in 1996 by Isner et al.²⁹ This particular patient with CLI presented with gangrene of the right great toe and failed surgical revascularization. The patient underwent transcatheter delivery of VEGF165-plasmid gene with initial improvement which manifested as improved flow to the right lower extremity, an increase in collateral vessels as seen on digital subtraction angiography (DSA), and by development of spider angiomas in the treated extremity. Despite these initial findings, the patient underwent below-knee amputation 5 months following the gene therapy.

Following this description of gene therapy in a patient with PAD, several phase I trials have been conducted, mostly using genes to induce angiogenesis. These studies are mostly small, non-controlled and non-blinded, simply serving the purpose of proof-of-concept and establishing the safety profile of the study agent. In essence, all trials revealed therapeutic angiogenesis to be safe, regardless of genes or recombinant proteins used (VEGF121, VEGF165, FGF-1, or FGF-2), vehicles utilized (plasmid or adenoviral), or route of delivery chosen (intra-arterial or intra-muscular).^30–35 The most frequent adverse events have been development of edema at the injection site or more generalized in the treated limb, primarily with administration of VEGF, as well as transient proteinuria or dose-dependent hypotension with intra-arterial FGF. Additionally, myasthenia and paresthesias have been described.³² The inclusion criteria, as well as the outcome parameter differ among all trials, and considering the primary purpose of phase I studies, conclusions on efficacy cannot be made based on these studies. Nevertheless, it is important to mention these early findings, as they have led to larger phase II trials.

Two groups reported their experience with a VEGF165-plasmid vector given as intra-muscular injections in 9 and 21 patients with chronic CLI, respectively.^30,34 An example of the response in one of these patients is illustrated in Figure 1. Both studies revealed subjective improvement with reduction in frequency of rest pain, as well as more objective evidence of a treatment effect as defined by improved ankle-brachial-indices (ABIs), pain-free walking time on a treadmill, and quantitative scores of vascularity derived from magnetic resonance angiography (MRA).

Rajagopalan, Mohler and colleagues report their results with the VEGF121 adenoviral vector separately for patients with CLI and intermittent claudication.^31,33 Combined, they included 33 patients and showed a significant increase in walking time on the treadmill, but no improvement in ABIs. Neutralizing antibody titers to adenoviruses doubled in 70% of the patients within one week. During one year follow up, two patients were incidentally detected to have bladder cancer. Although it is impossible to exclude the possibility that this occurrence was related to the drug being studied, the natural history of bladder cancer suggests simple coincidence of this finding.

One phase I trial evaluated FGF-1 as a plasmid vector preparation.³² The FGF-1 gene was administered as intra-muscular injections to 51 patients with ischemic rest pain, ulceration or tissue necrosis. Pain scores, transcutaneous oximetry, semi-quantitatively analyzed ulcer healing and ABIs significantly improved. Although the latter parameter improved statistically, the absolute value was < 0.1, failing to satisfy recommended outcome criteria used to judge the results of revascularization procedures.³⁶

A single study evaluated the safety and effects of FGF-2 administered as a recombinant protein by intra-arterial infusion.³⁵ This small phase I trial is the only study with a placebo group evaluating 19 patients with a history of intermittent claudication. Mild transient proteinuria was noted in both the placebo and treatment group. Patients treated with FGF-2 had marked symptomatic improvement and favorable changes in resting calf blood flow compared to those in the placebo group.

Aside from the phase I studies evaluating VEGF and FGF, one open-label phase I trial reports results of HGF.37 This small, 6-patient study used HGF plasmid DNA administered intramuscularly to patients with chronic CLI. The injections were well tolerated and the investigators were able to document a statistically significant increase in ABIs from 0.42 to 0.63 within 4 weeks. In addition, 8 of 11 ulcers improved and the vascularity of the treated lower extremity increased as qualitatively assessed by DSA.

Phase II Trials
To date, three phase II trials of therapeutic angiogenesis in patients with PAD have been completed and published (Table 1). One additional study was prematurely terminated because of the development of significant proteinuria caused by the study drug.

The TheRapeutic Angiogenesis with recombinant Fibroblast growth Factor-2 for Intermittent Claudication (TRAFFIC) trial was conducted following earlier experiences with recombinant FGF-2 protein reported by Lazarous et al.^35,38 This randomized, double-blind, placebo-controlled trial included 190 patients with exercise-limiting claudication and ABIs < 0.8. Patients received placebo, one or two doses of recombinant FGF-2, as intra-arterial infusions 4 weeks apart, and were followed for 6 months. Both treatment groups had a statistically significant change compared to placebo in the primary outcome parameter of peak walking time (change in peak walking time: 0.6 minutes, 1.77 minutes, and 1.54 minutes in the placebo, single-dose and double dose groups, respectively). Further, a change in ABIs at 3- and 6- month follow up was noted in the FGF-2 group, but not in participants receiving placebo, although this difference was not statistically significant. Subjective outcome measures evaluated by quality-of-life questionnaires did not differ among the three groups. Thus, although the secondary outcome parameters did not reach statistical significance, the TRAFFIC study was the first phase II trial showing evidence of clinical therapeutic angiogenesis.

Transient hypotension occurred in 3% of patients in the placebo group and 6 and 8% in the active treatment arm. Proteinuria was judged to occur infrequently in 9 and 11% of the FGF-2 treatment groups and 3% in the placebo group. Another phase II trial evaluating recombinant FGF-2 protein for the treatment of PAD had been terminated prematurely after enrollment of 24 subjects prior to the TRAFFIC trial because of the development of severe proteinuria.³⁹ The main difference between these two trials is the route FGF-2 is given. According to Cooper et al., the protocol FGF-2 was given intravenously six times on a weekly basis, while in the TRAFFIC trial, the FGF-2 was selectively infused twice intra-arterially into the lower extremities. This difference may account for the high rate of severe proteinuria (defined as > 1g/24h) observed in Cooper’s study, which was 25% in the treatment group.

Findings from the randomized, double-blind, controlled Regional Angiogenesis with Vascular Endothelial growth factor (RAVE) were published in 2003.⁴⁰ The investigators randomized 105 patients with chronic, stable unilateral intermittent claudication and ABI < 0.8 to either placebo, a low or high dose of VEGF121 with an adenoviral vector via 20 intra-muscular injections. Although all patients demonstrated improvements in change in peak walking time, no difference between the three groups at 3- and 6- month follow up was seen. Further, treatment with VEGF121 had no effect on claudication onset time, ABI, and quality-of-life compared to placebo. The placebo effect on the change in peak walking time was remarkable and more pronounced with a 36 and 54% improvement at 3- and 6-month follow up, respectively, compared to a 14% improvement at 3 months in the TRAFFIC trial. This excessive placebo effect has potential implications for study endpoints selected in future trials, since change of peak walking time does not seem to be a specific marker for therapeutic angiogenesis. As in prior studies of VEGF, localized edema was the most frequent adverse event occurring in 20–28% of the treatment groups, and 9% in the placebo arm.

The smallest phase II trial reported by Mäkinen et al. included 54 patients with claudication or CLI.41 Three groups of patients received either placebo or VEGF165 with a plasmid or adenoviral vector system administered as intra-arterial infusions. The trial design is considerably different from the TRAFFIC or RAVE studies, in that all patients underwent PTA and received the study drug immediately following the transcatheter intervention. In addition, the primary outcome was a change in vascularity in the treated limb assessed by DSA at 3 months follow up. The VEGF-treated groups had significantly improved vascularity, which was determined by two radiologists conducting a blind review. However, although ABIs and Rutherford grade significantly improved during follow up, this finding was not confined to the active treatment arms, suggesting again primarily a placebo effect. Also, clinical outcomes regarding need for amputation, ulcer healing or resolution of rest pain did not differ among the treatment groups. The procedure was well tolerated with only mild edema was reported as an adverse event.

Shortcomings of Prior Trials
The initial euphoria of therapeutic angiogenesis as a novel treatment option for patients with PAD triggered by the phase I trials partially subsided after publication of the TRAFFIC and RAVE studies with their inconclusive clinical benefits. This raises the question why the remarkably positive results from animal models as well as phase I trials did not translate into clinical practice. Further, in light of the profound placebo effect found, the chosen endpoints for future trials in this field need to be revisited. The first explanation for treatment failures relates to patient selection. Patients selected for trials of therapeutic angiogenesis tend to be older, often having previously undergone multiple revascularization attempts, and usually have extensive disease and have exhausted standard therapeutic options. One can argue that these patients suffer from a failure of natural angiogenic responses, and thus represent the group least likely to respond to gene therapy or recombinant growth factors. A possible mechanism would be a concept of growth factor resistance with down-regulated receptors, leading to a blunted or absent response to the administration of angiogenic factor, regardless if in protein or gene form.

The choice of therapeutic agent and route of administration is another important aspect to consider in this context. Pharmacokinetics and pharmacodynamics of a variety of vectors and genes are largely unknown.42 Since several cofactors affect the optimal dose of the agents, such as the vehicle used or the route of delivery chosen, identifying the optimal combination of agent, vehicle and route will take some time.

The ability to monitor the outcome of therapeutic angiogenesis has been a long-standing challenge. Current trials used a wide range of different end points, some purely subjective and some semi-objective. Although surrogate endpoints are helpful in suggesting clinical efficacy, they do not substitute for hard endpoints, such as mortality, need for revascularization or amputation; all of which, if included in prior trials, showed no difference between placebo and active treatment groups. Exercise tolerance testing as a primary endpoint may be questioned in general for angiogenesis trials in treatment of PAD, as exercise capacity and walking time are significantly confounded by coronary heart disease and congestive heart failure, conditions which are prevalent in patients with advanced PAD.43 Semi-quantitative analysis of leg vascularity by DSA or MRA seems more reasonable in this setting, but may miss pure angiogenic responses, leading only to the development of capillaries not visualized with these methods.

Safety Considerations
Therapeutic angiogenesis with recombinant proteins or genes of vascular growth factors has the potential to cause several serious adverse effects. Much concern has been expressed, primarily because of the possible detrimental pro-angiogenic effect on atherosclerosis and malignancies, which are both diseases where pathologic angiogenesis plays a critical role in the development and progression of disease.^44,45 Neither of these concerns was confirmed by human in vivo data, albeit patients with a history of cancer or active malignancy were excluded from the previously discussed trials. Another area of possible “off-target” angiogenesis is the development of proliferative retinopathy by angiogenic growth factors.⁴⁶ Clinical studies to date have not shown that exacerbation of proliferative retinopathy is a risk of therapeutic angiogenesis for treatment of PAD or coronary heart disease. VEGF augments vascular permeability, and was initially called vascular permeability factor.⁴⁷ As expected from its physiological consequences, administration of VEGF has led to the development of localized edema, which responds well to diuretic therapy. Baumgartner et al. found the risk of developing edema with VEGF therapy directly corresponded with the degree of tissue ischemia. None of their 90 patients evaluated following treatment with VEGF165 developed edema if they were treated for claudication, but 24% developed edema if they had rest pain, and 60% of patients with gangrene developed edema.⁴⁸ Despite the fact that VEGF and FGF stimulate nitric oxide production, potentially leading to vasodilatation and hypotension, this complication has never been described after gene transfer in humans, regardless of the gene or vector.^49,50 The development of transient proteinuria following systemic application of recombinant FGF-2 has been discussed above.

Summary and Outlook
Therapeutic angiogenesis is a promising therapy for patients with PAD, but it is still in its infancy. Prior clinical trials of therapeutic angiogenesis are limited to mostly uncontrolled phase I trials and a few phase II studies. The current data support safety of the agents and vectors used. The clinical efficacy as seen in the phase II trials, however, is at most modest and inconsistent so far. Obviously, trials with more participants, longer follow up times and reproducible outcome parameters are needed to define the potential role of therapeutic angiogenesis in patients with PAD.

Nevertheless, the excitement for therapeutic angiogenesis in patients with PAD is ongoing, as seen by the number of ongoing trials. These predominantly phase II trials are evaluating some new agents, which are believed to be more potent than prior tested factors VEGF and FGF.^51,52 These agents, tested in ongoing trials in the United States, Europe and Japan include HGF (hepatocyte-growth factor), DEL-1 (developmentally regulated endothelial locus 1), and HIF-1 alpha-VP16. Although the final results of the initial phase I trial evaluating HIF-1 alpha-VP16 in patients with CLI have not been published, preliminary data of the 28 participants appear promising, as rest pain resolved in most patients, and pre-existing ulcers healed in 3 of 5 cases.⁵³

In addition, previously evaluated angiogenesis factors such as VEGF165 and FGF-1 are currently undergoing phase II trials in new formulations with different vectors than used in prior studies. In this context, the poloxamer 188 VLTS 934 needs to be mentioned. Used as an angiogenesis factor vehicle for Del-1, it was found to have a beneficial effect on waking distance in a prior phase I trial, even without the actual angiogenetic active agent Del-1. VLTS 934 is also known to have anti-inflammatory features and inhibits neutrophilic migration.^54,55 A proposed mechanism explaining the potential benefit of VLTS 934 in patients with PAD is improved microvascular flow due to an inhibition of the atherosclerotic inflammatory process. Thus, although it is not considered to have angiogenetic potential in itself, VLTS 934 is undergoing a phase II trial comparing intramuscular application of the agent versus placebo.

Eventually two general questions need to be answered, which go beyond study design issues or the simple evaluation of various potential candidate proteins or genes. First, to be effective in patients with PAD, should gene therapy induce angiogenesis or arteriogenesis? In other words, is the development of capillaries sufficient to improve blood flow and symptoms in patients with PAD, or are newly formed arterial conduits necessary to achieve this goal? Second, are we able to achieve clinical meaningful results in therapeutic angiogenesis with a single agent in general? Considering the complexity of physiologic angiogenesis, future research may need to focus on multiagent therapy to mimic the complex process of angiogenesis in humans.