Currently, a large variety of chemotherapeutic drugs are being used to treat cancer. Unfortunately, many compounds hold limited efficacy, due to problems of delivery and penetration and a moderate degree of selectivity for the tumor cells, thereby causing severe damage to healthy tissues. However, the activity of these compounds is mainly limited by the development of drug resistance. Tumor cells are a rapidly changing target because of their genetic instability, heterogeneity and high rate of mutation, leading to selection and overgrowth of a drug-resistant tumor cell population (85, 127).

In 1971, J. Folkman hypothesized that tumors are angiogenesis-dependent and that anti-angiogenic therapy might represent a good alternative for the treatment of solid tumors (78). Since then, numerous reports have pointed to the crucial role of neovascularization in the malignancy of tumors and different non-neoplastic pathologies referred to as angiogenic diseases (29, 66). Anti-angiogenic therapy, which targets activated endothelial cells, presents several advantages over therapy directed against tumor cells. First, endothelial cells are a genetically stable, diploid and homogenous target and spontaneous mutations rarely occur. Also, turnover of tumor endothelial cells may be 50 times higher than that of endothelium in normal quiescent tissues, and activated blood vessels express specific markers, like integrin avb3, E-selectin, TIE and VEGF receptors. Because anti-angiogenic therapy is directed at activated endothelial cells, its target should be easily accessible by systemic administration. Finally, different tumor cells are sustained by a single capillary and tumor-associated endothelial cells contribute both to endothelial and tumor cell growth by releasing autocrine and paracrine factors. Consequently, the activated endothelium presents a more specific target than the tumor cells and inhibition of a small number of tumor vessels may affect the growth of many tumor cells (127).

An increasing number of anti-angiogenic compounds have been identified (Table 5), many of which have been shown to hold anti-angiogenic activity in a particular assay, like the CAM. More recently, research has focused on the search for compounds with a specific effect on an individual step of the angiogenic process. As each step in the angiogenic cascade involves a great variety of enzymes, cytokines and receptors, angiogenesis presents different possible targets for therapeutic intervention.

In the following section, we will only discuss the more recently discovered promising compounds or drugs undergoing clinical evaluation.

INHIBITION OF MMP ACTIVITY

The activity of matrix metalloproteinases (MMPs) is tightly regulated during physiological tissue remodeling. However, a large body of evidence suggests that this regulation is lost during tumor growth and metastasis (191). Excessive MMP activity has been detected in colorectal, lung, breast, gastric, cervical, bladder, prostate cancer and malignant glioblastoma. Moreover, in a number of these studies, a good correlation was seen between the amount of MMPs and the aggressiveness/invasiveness of the tumor (44, 84, 265).

Recently, a naturally occurring noncatalytic fragment PEX of MMP-2 was found to prevent binding of the enzyme to the integrin avb3 receptor, leading to inhibition of enzymatic activity at the cell surface. A recombinant form of PEX was shown to block angiogenesis and tumor growth in vivo (26). Despite the latest discovery of PEX and other endogenous MMP inhibitors (TIMP-1, -2, -3, -4) (90), research has focused on synthetic, orally available inhibitors. Several MMP inhibitors (MMPI) have been developed, from broad-spectrum inhibitors, which block most of the MMPs, to selective inhibitors, which interfere with the activity of a particular MMP. One general structural feature of MMPIs is the presence of a metal-binding group, often a carboxyl, thiol or hydroxamate, that chelates the Zinc atom in the active site of the enzyme (191).
 
 

The first synthetic MMPI evaluated in the clinic was batimastat (BB-94). Batimastat possesses potent activity against most of the major MMPs (MMP-1, -2, -3, -7, -9) by reversible competition with the MMP substrate. Despite its ability to suppress or prevent the growth of various tumors in animal models (220), clinical studies with batimastat have been suspended because of its insolubility and hence, low oral bioavailability. The related compound marimastat (BB-2516) has an enzyme activity spectrum similar to batimastat, but with a more favorable pharmacological profile in humans, since it is orally available (56, 257). Marimastat has now entered Phase III trials in patients with small cell lung, non-small cell lung, and breast cancer and is undergoing Phase II studies for pancreas and brain cancer (29).Other MMPI in the stage of clinical development include AG3340, an orally active hydroxamic acid derivative (217), and Neovastat, an endogenous inhibitor isolated from cartilage (29).

INHIBITION OF CELL ADHESION MOLECULES

The avb3 integrin, an adhesion receptor for extracellular matrix components with an exposed RGD sequence is an attractive target for anti-angiogenic therapy since it is exclusively present on the cell surface of activated endothelial cells, but absent on quiescent endothelium or other cell types (58). An RGD-containing peptide antagonist of avb3 and an anti-avb3 monoclonal antibody, LM609, were found to inhibit adhesion-dependent signal transduction by angiogenic factors, leading to apoptosis of the activated endothelium. Consequently, these compounds blocked endothelial tube formation in vitro andangiogenesis during development, arthritis (224) and in growing tumors (25, 27). Based on these convincing data in different animal models, the clinical potential of integrin antagonists is currently being evaluated in Phase I and II trials for patients with late-stage cancer. Vitaxin, the humanized form of the anti-avb3antibody LM609, has successfully completed Phase I clinical trials (29).

ENDOGENOUS INHIBITORS

Several natural inhibitors of angiogenesis have been detected. Among them, thrombospondin-1 (TSP-1) is considered to be the main physiological inhibitor of angiogenesis, being constitutively produced by normal cells. Its expression is inversely correlated with angiogenesis, i.e. during tumorigenesis, TSP-1 is downregulated while the angiogenic activity is increased. Accordingly, it was shown that TSP-1 production is regulated by the p53 tumor suppressor gene (95). Mutation of p53 results in the loss of TSP-1 production and a switch to the angiogenic phenotype (46, 115). Consequently, overexpression of TSP-1 caused a decrease in angiogenesis and inhibition of tumor growth (225).

However, the most promising tumor-shrinking anti-angiogenic drugs thus far are derived from an unlikely source: the tumor cells themselves. Angiostatin and endostatin are examples of endogenous inhibitors that are generated by the proteolysis of inactive circulating precursors.

Angiostatin, which encompasses the first four disulfide-linked kringle domains of plasminogen (175), was originally purified from the serum and urine of mice bearing LLC. These mice did not suffer from metastases until the primary tumor was removed which resulted in rapid growth of the previously dormant lung metastases. This observation indicated that the primary tumor produced a potent angiogenesis inhibitor or, alternatively, released an enzyme that generates angiostatin from plasminogen. The mediator of angiostatin production in LLC was identified as a tumor-infiltrating macrophage, expressing metalloelastase (54). However, tumor cells themselves also have been shown to produce proteolytic activity that generates angiostatin from plasminogen and this enzymatic activity was found to differ from that released by tumor-infiltrating macrophages (86). In vivo experiments in mice have shown that angiostatin suppresses the growth of a number of human tumors and their metastases (33). Immunohistochemical analysis revealed that the rate of tumor cell proliferation was identical in growing and dormant metastases. However, the apoptotic rate was threefold higher in the dormant metastases. Thus, tumor dormancy may depend upon a balance between tumor cell growth and death (174). A current report showed that ATP synthase binds angiostatin, implying that angiostatin interferes with ATP production resulting in the inhibition of endothelial cell growth (158). Finally, several data suggest that different kringle domains may contribute to the overall anti-angiogenic function of angiostatin by their distinct anti-migratory and anti-proliferative activities (34, 121).

Endostatin, a carboxy-terminal fragment of collagen XVIII, derived through elastase-mediated cleavage (254), was isolated from the conditioned media of hemangioendothelioma (EOMA) cells (173). Endostatin specifically suppresses endothelial cell proliferation in vitro and increases the apoptotic rate in tumors seven-fold without affecting the proliferation rate of the tumor cells. In vivo, endostatin showed potent inhibitory activity against EOMA, Lewis lung, T241 fibrosarcoma and B16F10 tumor cell lines. Interestingly, endostatin does not seem to induce drug resistance (21). Moreover, repeated cycles of systemic endostatin administration in tumor-bearing mice caused sustained tumor dormancy in the absence of further treatment (21).

Recently, a third fragment with potent anti-angiogenic activity was purified from small lung cell cancer. AaAT (anti-angiogenic anti-thrombin) results from the cleavage of anti-thrombin III by a yet unidentified enzyme (176).

SYNTHETIC COMPOUNDS
 

TNP-470 (AGM-1470), a synthetic derivative of the antibiotic fumagillin, is perhaps the most studied inhibitor of angiogenesis (113). However, its molecular target has only been recently identified. TNP-470 was shown to bind and subsequently inhibit type 2 methionine aminopeptidase (94), resulting in the abrogation of amino-terminal processing of methionine, which may lead to the inactivation of as yet unidentified proteins essential for endothelial cell growth (242). TNP-470 inhibits endothelial cell proliferation and migration in vitro (3, 260). In animal models TNP-470 is effective in the treatment of a wide variety of tumors and their metastases (207, 235, 259). Its antitumor activity together with its moderate side effects has led to Phase II-III clinical trials for a variety of solid tumors (29). TNP-470 is administered at a dose of 60 mg/m² as a 60-min i.v. infusion three times a week, the dose-limiting complication being neurotoxicity (136).

Thalidomide is a well-known teratogen with anti-inflammatory and anti-angiogenic activity (45). The exact mechanism of the drug is not yet known. Thalidomide was inactive in the CAM assay, but showed potent inhibitory activity upon oral administration in the FGF-2-induced cornea assay, which may reflect the need for metabolic activation in the liver (15). The anti-tumor efficacy of thalidomide has been demonstrated in a large study comprising 84 patients with multiple myeloma. An oral dose of 200 mg Thalidomide, per day, induced marked and durable responses in some patients, including those who relapsed after high-dose chemotherapy (219).

The tubulin-binding drug Combretastatin A-4 exhibits a selective toxicity for proliferating endothelial cells in vitro by induction of apoptosis (117). In vivo, systemic drug administration causes vascular shutdown within experimental and human cancer models at doses that are 10% of the maximum tolerated dose. Histologically, the reduction in blood flow is associated with extensive necrosis of the tumor (48, 239). These actions against tumor vasculature and the broad therapeutic window demonstrate the clinical potential of this drug.

COMPOUNDS THAT INTERFERE WITH ANGIOGENIC GROWTH FACTORS OR THEIR RECEPTORS

Of the long list of growth factors involved in the angiogenic process, VEGF and FGF-2 are considered the most important mediators of tumor angiogenesis. Consequently, different strategies have been developed to inhibit the production or release of these growth factors or to interfere with their receptor interactions. Specific targeting of VEGF using anti-VEGF antibodies, soluble VEGF receptors or dominant negative Flk-1 (89, 154, 190) decreased the vessel density and reduced the growth rate of several tumors in animal models.

Anti-VEGF antibodies and soluble Flk-1 and Flt-1 receptors have also proven successful in the treatment of ischemia-associated iris neovascularization in primates and retinal neovascularization in a murine model for ischemic retinopathy, respectively (67). Accordingly, a humanized anti-VEGF antibody, that has completed phase I trials without significant systemic toxicity, is now being tested in phase II studies involving patients with metastatic renal cell cancer (29). Similarly, a soluble recombinant extracellular domain of the Tie-2 receptor has been constructed, that substantially impaired angiogenesis, tumor growth and metastases (140).
 
 

In order to interfere with receptor signaling, synthetic low molecular weight inhibitors of tyrosine kinase receptors have been designed. The first receptor antagonist to enter clinical trials was SU5416, which selectively blocks VEGF-induced phosphorylation of Flk-1. SU5416 displayed potent anti-tumor activity in animal models and was found to induce endothelial cell apoptosis (80, 215), pointing to the role of VEGF as a survival factor for these cells. The efficacy of SU5416 is currently being evaluated in Phase II trials for KS, colorectal cancer and von-Hippel Lindau disease (29). However, tumor cells may produce several cytokines and, due to their instability, they may switch from the production of one cytokine to another. Therefore, inhibition of only one single growth factor may cause only partial control of tumor growth. This hypothesis has lead to the development of SU6668, a potent inhibitor of VEGF, FGF-2, and PDGF tyrosine kinase receptors (215), that has recently entered a Phase I study for advanced tumors (29).

 
 

A number of drugs originally developed for their capacity to inhibit FGF-2 induced angiogenesis, have been shown to interfere with the biological activities of several other heparin-binding growth factors (253). These molecules may either mimic or bind heparin. The importance of heparin for FGF-2-mediated angiogenesis has been extensively documented. After secretion, FGF-2 binds to heparan sulfate and is subsequently sequestered in the ECM and basement membrane, from where it can be released by matrix-degrading enzymes or soluble heparin. HSPG also serve as low affinity FGF-2 receptors and FGF-2 binding is required for interaction with the FGFR. Thus, the availability and biological activity of FGF-2 on endothelial cells strictly depend on the extracellular heparin concentration, indicating the possibility of modulating the angiogenic activity of FGF-2 in vivo by using exogenous heparin analogues (202). These include among others suramin (231), pentosan polysulfate (PPS) (266), sulfonated distamycin A derivatives (40) and carboxylated compounds (152). Suramin (138) and PPS have been evaluated in patients with various tumors, including KS (213), but the general observation is that high doses of these compounds are required to show activity and that their efficacy is limited by anti-coagulant side effects. However, the introduction of minor structural changes to suramin was found to result in significantly less toxicity without loss of activity (70).

Promising results have also been obtained with IFN-a in the treatment of juvenile, life-threatening hemangiomas (93). IFN-a was shown to downregulate the expression of FGF-2 (52), which is abundantly present in hemangioma lesions and in the urine of patients suffering from proliferating hemangiomas (37).
 

MISCELLANEOUS

Other possible targets for anti-angiogenic therapy include enzymes with important angiogenic properties such as platelet-derived endothelial cell growth factor/Thymidine phosphorylase (PD-ECGF/ TP) and cyclooxygenase (COX). PD-ECGF was found to stimulate endothelial cell migration in vitro and angiogenesis in vivo. Its angiogenic effect is mediated by the release of 2-deoxy-D-ribose, as a result of the reversible phosphorolysis of pyrimidine (deoxy)nucleosides by TP, to 2-deoxyribose 1-phosphate and their respective bases. 2-deoxyribose 1-phosphate is rapidly dephosphorylated and transported out of the cell. However, the mechanism by which this molecule induces angiogenesis has not yet been elucidated (30). TP is overexpressed in many solid tumors, including breast (143), ovarian (193), colorectal (232) and pancreatic cancers (109), and in the endometrium during the menstrual cycle (83), pointing to a role for this enzyme in both physiological and tumor vascularization. Among the very few TP inhibitors  that have been described, 6-aminothymine (6AT), 6-amino-5-bromouracil (6A5BU) and 7-deazaxanthine (7-DX) (12) are the most potent.

Recent reports have implicated COX-2, an enzyme that controls several cellular processes involved in colon cancer development, in the regulation of tumor angiogenesis (241). COX-1, that is constitutively expressed and required to maintain the integrity of gut and kidney, and COX-2, that is inducible, mediate the conversion of arachidonic acid (AA) into prostaglandin G2, which is subsequently converted to prostaglandin H2, and eventually into a number of other prostaglandins and thromboxane A2 (47). Using a coculture of endothelial cells and colon cancer cells, COX-2 was shown to stimulate colon cancer cells to release angiogenic prostaglandins, which induce migration and tube formation of endothelial cells (241). This was blocked by traditional non-steroid anti-inflammatory drugs (NSAID) like aspirin, which inhibit COX-1 and COX-2, and NS-398, a selective COX-2 inhibitor (124). However, addition of exogenous prostaglandins could only partially reverse the inhibitory effect of these drugs, indicating that COX-2 may also exert additional prostaglandin-independent effects on angiogenesis (124). In this context it should be noted that overexpression of COX-2 in intestinal epithelial cell lines resulted in resistance to butyrate-induced apoptosis and downregulation of the adhesion molecule E-cadherin (240). Also, the COX-2 product thromboxane A2 was identified as a mediator of COX-2 dependent endothelial cell migration and angiogenesis (47). Since inhibitors of COX-1 are associated with gastrointestinal and renal toxicity, further studies should focus on specific COX-2 inhibitors for the treatment of malignant colon cancer.

Finally, experimental evidence suggests that a combination of anti-angiogenic drugs with different mechanisms of action may lead to synergistic anti-angiogenic effects. Angiosuppressive therapy has also been shown to increase the efficacy of classical chemotherapeutic agents in anticancer treatment (85). Furthermore, exposure to angiostatin potentiated the antitumor effect of ionizing radiation (149). Interestingly, radiation was found to induce VEGF, resulting in the protection of tumor blood vessels from radiation-mediated cytotoxicity and, hence, tumor radioresistance. Consequently, treatment of tumor-bearing mice with a neutralizing antibody to VEGF prior to irradiation was associated with synergistic antitumor effect (91).

Vascular targeting aims at inhibiting tumor growth by destruction of the tumor vasculature. The main problem so far has been the lack of specific markers for activated, i.e. tumor endothelium. Potential target molecules include the avb3 integrin, VEGF and TIE receptors and E-selectin. Destruction of the tumor vessels may be achieved by the local delivery of peptides or antibodies with direct biological activity or conjugated to toxins. Accordingly, VEGF chemically linked to a truncated diphtheria toxin molecule (DT385) was found to specifically inhibit the proliferation of flk-1 positive endothelial cells in vitro and angiogenesis in vivo (6). Also, an antibody against the cell surface domain of tissue factor (tTF), a human coagulation-initiating protein, caused complete tumor regression in mice with large neuroblastomas. Specific targeting was achieved by transfection of the neuroblastoma cells with IFN-g, which induces the expression of major histocompatibility complex class II antigens (I-Ad and I-Ed) on the tumor endothelium, but not on normal endothelial cells. A bispecific antibody against tTF and I-Ad caused thrombosis of the tumor vessels with extensive necrosis and tumor regression (106).

Recently, an in vivo selection of phage display libraries was used to identify peptides that are present exclusively on blood vessels of specific organs (135). This method was then applied to target tumor blood vessels. Therefore, phage peptide libraries were injected in the circulation of nude mice bearing human breast carcinoma xenografts. Recovery of phages from the tumor led to the identification of three main peptide motifs that target the phage to the tumors. When coupled to the cytotoxic drug doxorubicin, these peptides enhanced the efficacy of the drug against the mammary carcinomas in nude mice and reduced its toxicity (4).

Therapeutic modulation of angiogenesis can be accomplished by the administration of single or multiple doses of angio-regulatory peptides or drugs or by means of gene therapy. Gene therapy offers a potential way to achieve sustained therapeutic release of active substances. For example, adenoviral and retroviral vectors that transduce the cDNA encoding angiostatin (233) or PF-4 (235) or antisense VEGF (112), have been used to inhibit endothelial cell growth in vitro and angiogenesis and tumor growth in vivo. Exact targeting can be attained by the use of endothelial specific (i.e. E-selectin, VEGF or TIE) promoters. However, the most promising results with gene therapy have been obtained in therapeutic angiogenesis. Intramuscular gene transfer of VEGF165 and intramyocardial administration of an adenoviral vector expressing the VEGF121 cDNA improved collateral vessel development in patients with critical limb ischemia (7) and coronary artery disease (201), respectively.