Gene transfer of growth factors to promote development of collateral blood vessels may provide new approaches to therapy of coronary artery and peripheral vascular disease. In either case, reestablishing blood flow to ischemic tissues is intended to provide a biologic bypass around occluded arteries. This strategy, known as therapeutic angiogenesis, promises new options for patients who may not be optimal candidates for surgical revascularization or angioplasty, procedures whose long-term results are often limited by restenosis. In early clinical studies, direct myocardial injection of the gene encoding a vascular growth factor reduced the severity and frequency of symptoms in patients with angina, and similar therapy was shown to stimulate blood vessel growth in patients with peripheral vascular insufficiency.

As is well known, the pathologic counterpart of therapeutic angiogenesis underlies such diseases as diabetic retinopathy, rheumatoid arthritis, and--most of all--cancer. Both the therapeutic and the pathologic implications of angiogenic growth factors were identified in the 1970s by Judah Folkman and colleagues at Harvard University. Their work documented the dependence of tumor development on neovascularization and suggested that angiogenic factors specific for neoplasms might be involved. Therapy to limit or even reverse tumor growth is now under investigation and is showing promise (see "Angiogenesis Inhibitors as Cancer Therapy" by S. Gail Eckhardt).

Beginning in the mid-1980s, a series of human growth factors was purified, sequenced, and shown to be responsible for physiologic as well as pathologic angiogenesis. These growth factors, all of which act as mitogens for endothelial cells, selectively induce neovascularization in ischemic tissues. Various studies documented upregulation of vascular endothelial growth factor (VEGF, also known as vascular permeability factor and vasculotropin) after severe tissue ischemia. VEGF may thus be viewed as a prototype of vascular growth factors.

Mechanisms

Angiogenesis is thought to begin with the activation of endothelial cells in a parent vessel. The cells break free of the basement membrane, migrate into interstitial space (perhaps in the direction of an ischemic stimulus), and then proliferate, generating new vessels (Figure 1). In this classic paradigm, migration is more critical than proliferation. The concept is perhaps best supported by experiments with rat cornea reported by M.M. Sholley and associates in 1984. Using x-rays to block endothelial cell proliferation, they showed that vascular sprouting nonetheless occurred in response to an inflammatory stimulus. This laid the basis for subsequent demonstrations that angiogenesis is promoted by growth factors that either primarily affect migration, such as some forms of VEGF, or exclusively do so, such as angiopoietin (Table 1).

Studies of ischemia have shown that endothelial cell proliferative activity is virtually absent in normal arteries; even a relatively low level of proliferation in response to arterial occlusion or administration of growth factors may therefore represent a considerable enhancement of such activity. In addition, studies have shown that smooth muscle cells are needed to form media of developing blood vessels, and that proliferative activity for endothelial as well as smooth muscle cells is highest in the smallest collateral vessels, the so-called midzone collateral segments.

In particular, the proliferative activity of midzone-collateral smooth muscle cells was found to increase threefold in response to VEGF in experimental limb ischemia. Such proliferation probably did not represent a direct effect. Since VEGF increases vascular permeability, for example, extravasation of other angiogenic growth factors from circulating blood might have activated smooth muscle cell proliferation. Alternatively, endothelial cells stimulated by VEGF might have secreted factors that promote smooth muscle cell proliferation, such as tissue-type plasminogen activators. Experiments in our group's laboratory have shown that VEGF induces up-regulation of platelet-derived growth factor (PDGF), which is also a mitogen for vascular smooth muscle cells. Furthermore, a high-affinity VEGF receptor on endothelial cells is expressed by smooth muscle cells, which could be a mechanism by which VEGF recruits smooth muscle cells to form media in developing blood vessels.

Site-Specificity of Angiogenic Cytokines

Teleologically, VEGF and other endothelial cell mitogens should not promote angiogenesis indiscriminately but rather limit it to sites of wound healing and tissue ischemia, where vascular growth may be beneficial. In fact, this may be the case. Studies in animal models using fibroblast growth factor (FGF) or VEGF have shown that administration of angiogenic cytokines produces neovascularization in the region of ischemia. For example, when recombinant VEGF was injected into the normal or ischemic limb in rabbits studied at this center, angiogenesis was observed only in the ischemic limb. There have been similar reports from clinical trials: when patients with peripheral vascular disease were treated with recombinant FGF, circulating FGF levels were evident, yet angiogenesis was limited to the ischemic extremities; and when circulating VEGF levels were documented in patients with myocardial and limb ischemia after intramuscular injection of the gene encoding the growth factor, angiogenesis occurred only in ischemic territory.

Experiments performed at this center on endothelial cells in vitro suggest that the basis for localized VEGF bioactivity is hypoxia-induced upregulation of VEGF receptors (Figure 2). In these experiments, we observed an increase in mRNA levels of KDR, the principal VEGF receptor, in endothelial cells incubated for three hours in conditioned medium from hypoxic myocytes, compared with the levels in endothelial cells incubated in medium from normoxic myocytes. Analysis of I¹²⁵VEGF-binding indicated no substantial increase in KDR affinity in normoxic or hypoxic endothelial cells. Instead, increased binding during hypoxia was associated with a 13-fold increase in the number of KDR receptors. VEGF site-specificity is thus a function of KDR expression by endothelial cells. Factors secreted from hypoxic myocytes in ischemic tissues upregulate VEGF expression on adjacent endothelial cells, which then attract circulating VEGF into the ischemic tissue and amplify its effects.

Although single-bolus administration of VEGF stimulates formation of new collateral vessels in several days, the circulating half-life of VEGF is less than three minutes. One possible explanation for the apparent disparity is that, as a heparin-binding protein, VEGF is rapidly cleared from the circulation, binding avidly to endothelial heparin sulfate proteoglycans. An autocrine loop activated by hypoxia might also serve to amplify and thereby protract the response of endothelial cells stimulated by administration of VEGF. Moreover, VEGF also inhibits endothelial cell apoptosis (programmed cell death). This effect of VEGF, in conjunction with its mitogenic effect, might be expected to increase the total number of endothelial cells that survive.

In native collateral blood vessels, abnormal vascular reactivity may limit VEGF's effects on perfusion. Studies have shown that perfusion through native coronary collateral vessels is associated with endothelial dysfunction in downstream vasculature. Therapeutic angiogenesis appears to promote recovery of endothelium-dependent flow. In a rabbit model of limb ischemia, endothelium-dependent and -independent flows were essentially restored 30 days after administration of VEGF. At least three mechanisms could explain the endothelium-dependent responses--a VEGF-induced increase in perfusion pressure helped to repair dysfunctional endothelium, VEGF directly repaired damaged endothelial cells, or newly formed blood vessels induced by VEGF accounted for some of the improved endothelial function.

Four homodimeric species of VEGF have been identified, each containing 121, 165, 189, or 206 amino acids. The principal isoforms--VEGF₁₂₁, VEGF₁₆₅, VEGF₁₈₉--differ markedly with regard to heparin avidity, but all include a secretory signal sequence. This contrasts with other angiogenic cytokines such as acidic FGF and basic FGF, which lack such a signal peptide and therefore are typically not secreted from intact cells. The secretory feature has been critical to gene transfer using naked DNA vaccines--that is, DNA unassociated with viral or other vectors.

Promoting Angiogenesis

A series of investigations established the feasibility of using recombinant growth factors to induce neovascularization in animal models of myocardial and peripheral ischemia. The growth factors first used for this purpose were members of the FGF family. Evidence that VEGF stimulates angiogenesis first was developed in experiments performed in a rabbit model of unilateral hindlimb ischemia. Our group showed that doses of 500 to 1,000 gm of VEGF produced a statistically significant augmentation of angiographically visible collateral vessels and histologically identifiable capillaries, and that the hemodynamics of the ischemic limb improved (measured as calf blood-pressure ratio) in VEGF-treated as compared with control animals.

Comparable results were obtained in another series of experiments in the same model. Recombinant VEGF was administered by intramuscular injection once daily for 10 days. Thereafter, blood flow at rest, as well as maximum flow velocity and maximum flow provoked by papaverine, were all significantly elevated. In similar rabbit experiments using arterial constriction to create an area of myocardial ischemia, recombinant VEGF increased myocardial blood flow.

Despite these encouraging results, we were unable to persuade commercial manufacturers to develop a recombinant protein formulation of VEGF for use in humans. We therefore turned to the possibility of exploiting the VEGF gene's secretory signal sequence to accomplish therapeutic angiogenesis. We had previously observed that arterial gene transfer of DNA coding for human growth hormone (HGH) resulted in circulating HGH levels equivalent to those in the physiologic range, even though immunohistochemical staining of sections taken at necropsy disclosed evidence of successful gene transfer or transfection in less than 1% of cells in the transfected arterial segment. It therefore appeared possible that secreted gene products might have biologic effects even when the number of transfected cells was low.

Arterial VEGF Gene Transfer

In studies to establish the feasibility of direct arterial gene transfer of VEGF₁₂₁ VEGF₁₆₅, and VEGF₁₈₉, the VEGF gene was delivered to the internal iliac artery in rabbits with unilateral hindlimb ischemia as naked DNA to minimize potential toxicity. The transfer was facilitated by applying solutions of pure plasmid DNA to the hydrogel polymer coating of a standard angioplasty balloon. The hydrogel provided an absorbable surface that retained the highly concentrated DNA. When the balloon was subsequently inflated, the DNA was specifically transferred to the arterial wall. Analysis of transfected internal iliac artery using the reverse transcription polymerase chain reaction (PCR) technique confirmed the presence of reproducible VEGF mRNA for up to 21 days; PCR analyses of other tissues were all negative.

Because the DNA is not delivered into cell nuclei by viruses or liposomes, only a minute fraction of genes enters muscle cells. With arterial gene transfer, we estimate that the transfection efficiency achieved is 0.5% and that the foreign DNA is effective for only a few weeks. Nevertheless, development of new collateral vessels in the ischemic limb following VEGF gene transfer was documented by serial angiography and by increased capillary density at necropsy. The hemodynamic deficit in the ischemic limb was reduced, as indicated by improvement in calf blood pressure ratio. In further studies, limb perfusion was assessed at 30 days using a guidewire with a Doppler crystal at the distal end to measure maximum flow velocity. Maximum flow in response to papaverine was increased, as it had been with use of recombinant protein in the same animal model.

Clinical Studies

We began our clinical studies of VEGF₁₆₅ arterial gene transfer in eight patients with critical limb ischemia (pain at rest or nonhealing ulcers) who were not candidates for surgical revascularization, the only therapy available. A dose-escalating strategy was mandated since VEGF had never been administered to human subjects. The transfer sites selected showed no evidence of atherosclerotic plaque or intimal thickening on intravascular ultrasonography.

Three patients, who had nocturnal rest pain alone, received a 1-mg dose of VEGF₁₆₅. All remained free of rest pain at three months. Both magnetic resonance angiography and contrast angiography showed increased flow distal to preexisting occluded vessels. Although contrast angiography failed to reveal new collateral vessels in these patients, DNA labeling studies in swine and canine models of myocardial ischemia and the rabbit model of hindlimb ischemia had established that increases in flow resulting from development of collaterals are typically associated with proliferation of new vessels that are less than 180 µm in diameter. We thus inferred that blood flow in these patients was increased by an augmented network of collaterals too small to be seen on conventional angiography. In support of that view, direct angiographic evidence of new collateral vessels was observed in a patient receiving 2 mg of plasmid DNA (Figure 3). Unfortunately, angiogenesis was not sufficient to reverse limb gangrene, and the patient underwent amputation five months later.

Intramuscular Gene Therapy

In a subsequent phase I trial, our group studied intramuscular gene therapy in nine patients with critical limb ischemia, most of whom had been advised to undergo amputation. Studies in our laboratory as well as others have shown that transfection efficiency is increased to 5% to 10% with intramuscular delivery to ischemic tissue. In each patient, 2 mg of plasmid DNA encoding VEGF₁₆₅ was injected directly into muscles of the ischemic limb at four sites. Four weeks later, a second 2-mg dose was administered the same way.

Gene expression was shown by a transient increase in serum VEGF levels monitored by an enzyme-linked immunoabsorbent assay (ELISA). Immunohistochemical analysis of tissue specimens revealed foci of proliferating endothelial cells. PCR analysis of these specimens showed persistence of small amounts of plasmid DNA, suggesting that intramuscular injection achieves sufficient expression of VEGF for therapeutic angiogenesis.

Hemodynamically, the mean ankle-brachial index increased from 0.33 at baseline to 0.48 12 weeks later. (An increase >0.1 indicates successful surgical or percutaneous revascularization).Newly visible collateral vessels were documented directly by contrast angiography, and serial magnetic resonance angiography of the ischemic limb added further evidence of improved blood flow. Clinically, pain resolved in the three patients with rest pain alone, and ischemic ulcers healed in four of seven patients.

In several cases, gene therapy accomplished genuine limb salvage. Our first patient had undergone seven unsuccessful surgical revascularizations at another hospital and was taking many analgesics, including methadone, oxycodone-acetaminophen, ami-triptyline hydrochloride, and fentanyl patch. She had a necrotic great toe and an ischemic ulcer at the site of vein harvest in the distal left limb and had been advised by her vascular surgeons to undergo below-knee amputation. Within eight weeks of gene transfer, the ulcer size was reduced sufficiently to permit placement of a split-thickness skin graft, which healed successfully and has remained healed after three years of follow-up (Figure 4).