Microtubule depolymerizing vascular disrupting agents: novel therapeutic agents for oncology and other pathologies

Abstract

Vascular disrupting agents (VDAs) are a relatively new group of ‘vascular targeting’ agents that exhibit selective activity against established tumour vascular networks, causing severe interruption of tumour blood flow and necrosis to the tumour mass. Microtubule depolymerizing agents form by far the largest group of small molecular weight VDAs many of which, including lead compound disodium combretastatin A-4 3-O-phosphate (CA-4-P), are under clinical development for cancer. Although distinct from the angiogenesis inhibitors, VDAs can also interfere with angiogenesis and therefore constitute a potential group of novel drugs for the treatment of pathological conditions characterized by excessive angiogenesis, in addition to cancer. The endothelial cytoskeleton is the primary cellular target of this family of drugs, and some progress in understanding the molecular and signalling mechanisms associated with their endothelial disrupting activity has been made in the last few years. Susceptibility of tumour vessels to VDA damage is ascribed to their immature pericyte-defective nature, although the exact molecular mechanisms involved have not been clearly defined. Despite causing profound damage to tumours, VDAs fail to halt tumour growth unless used together with conventional treatments. This failure is attributed to resistance mechanisms, primarily associated with cells that remain viable within the tumour rim, and enhanced angiogenesis. The focus is now to understand mechanisms of susceptibility and resistance to identify novel molecular targets and develop strategies that are more effective.

Introduction

The development of new blood vessel networks by angiogenic sprouting of existing host capillaries is a characteristic feature of solid tumours (Folkman 1990). Angiogenesis also plays a key role in lesion development in other pathological conditions such as various ocular diseases, psoriasis, rheumatoid arthritis, haemangiomas and endometriosis (Folkman 1995). The importance of vascular networks in the context of establishment and progression of these diseases, and in particular cancer, has led to the development of the concept of ‘vascular targeting’ for therapy (Folkman 2003; Thorpe 2004; Ferrara 2005; Neri & Bicknell 2005). This is mainly achieved by strategies designed to inhibit specific steps of the angiogenic process, using angiogenesis inhibitors, or alternatively by vascular disrupting approaches that aim to cause rapid collapse of existing tumour vessels and indirectly necrosis of the tumour mass (Folkman 2003; Tozer et al. 2005b; Cao 2008). The latter concept arose from the work of Juliana Denekamp in the 1980s in which she described dramatic tumour eradication by interruption of blood flow (Denekamp et al. 1983). The ensuing discovery of low molecular weight drugs with rapid tumour selective vascular disrupting properties, now collectively known as vascular disrupting agents (VDAs), opened up a new wave of interest in vascular targeting as a means of eradicating tumours (Hill et al. 1993; Dark et al. 1997; Tozer et al. 2005b). Today, microtubule depolymerizing agents form by far the largest family of low molecular weight molecules, with established vascular disrupting activity at non-toxic doses (Tozer et al. 2005b; Kanthou & Tozer 2007). Their capacity to target the cytoskeleton and compromise the integrity of endothelial cell junctions, thought to be central to their mechanism of action. Although it is not yet firmly established why VDAs are selective for tumours, current views favour the hypothesis that selectivity relates to the fragile and immature nature of tumour blood vessels (Tozer et al. 2005b). Disodium combretastatin A-4 3-O-phosphate (CA-4-P or Zybrestat™) (Pettit et al. 1987) is the lead microtubule depolymerizing agent in this group and was the first to enter clinical trials for cancer (Dowlati et al. 2002; Galbraith et al. 2003; Rustin et al. 2003). Preclinical studies have concluded that VDAs are ineffective at stopping tumour growth when used as single agents, but they hold great promise when combined with conventional therapies or even anti-angiogenic agents (Horsman & Siemann 2006). Therefore, current efforts centre on evaluating these combinations in both the preclinical and clinical settings. Knowledge of the molecular mechanisms responsible for tumour vascular collapse is only now beginning to accumulate (Tozer et al. 2008a,b) and this is essential to design better strategies to overcome treatment resistance.

Microtubule depolymerizing agents also display anti-angiogenic activities with endothelial cells being particularly sensitive to these drugs in this respect (Pasquier et al. 2006). The idea of using compounds with VDA activities to target angiogenesis is now beginning to become an attractive alternative option not only for cancer but also for other non-cancer pathologies characterized by excessive angiogenesis (Randall & Young 2006). Whether VDAs can target angiogenesis effectively is likely to depend on drug type, dose as well as fine-tuning of administration schedules. Here, we present an overview of the vascular effects of microtubule depolymerizing VDAs in tumours, with special emphasis on underlying mechanisms and CA-4-P. We also evaluate the role of VDAs in the treatment of pathologies other than cancer, which are characterized by aberrant angiogenesis.

Early effects of the combretastatins and other VDAs on tumours

The combretastatins are natural compounds extracted from the tree Combretum caffrum (Pettit et al. 1987, 1989) and were the first microtubule depolymerizing agents identified to have tumour vascular disrupting activity at well tolerated doses (Chaplin et al. 1996; Dark et al. 1997). Combretastatin A-4 phosphate (CA-4-P or Zybrestat) (http://www.oxigene.com), a soluble prodrug of the natural parent CA-4 molecule,(Pettit et al. 1995) is by far the most widely studied VDA of the microtubule depolymerizing family and was the first such agent to enter into clinical trial (Dowlati et al. 2002). Combretastatin A-1-P (CA-1-P or OXi4503) (http://www.oxigene.com) the prodrug of the natural parent CA-1 is an even more potent VDA than CA-4-P (Hill et al. 2002b) and is also now being clinically tested. The tumour vascular disrupting activities of many natural or synthetic microtubule depolymerizing compounds have been evaluated in preclinical models (Tozer et al. 1999; Davis et al. 2002; Hori et al. 2002; Micheletti et al. 2003; Shi & Siemann 2005; Kim et al. 2007; Malcontenti-Wilson et al. 2008). To date, more than 10 such agents have progressed to clinical trials for cancer (Table 1).

Table 1.

Low molecular weight VDAs in clinical trials

Drug name	Company	Drug type	Developmental phase
CA-4-P (Zybrestat)	OxiGene http://www.oxigene.com/	Tubulin binding agent	Phase I/II/III
CA-1-P (Oxi4503)	OxiGene http://www.oxigene.com/	Tubulin binding agent	Phase I
ZD6126	First developed by AstraZeneca http://www.astrazeneca.com/Now licensed to Angiogene http://www.angiogene.co.uk/	Colchicine analogue	Reached Phase II
AVE8062	Aventis http://www.sanofi-aventis.com/	Combretastatin prodrug	Phase I/II
ABT-751	Abbott http://www.abbott.com/	Sulfonamide, β-tubulin inhibitor	Phase I
NPI-2358	Nereus http://www.nereuspharm.com/	Tubulin binding agent, derived from marine fungus	Phase I
MPC-6827 (Azixa)	Myriad Pharmaceuticals http://www.myriad.com/	Tubulin binding agent	Phase I/II
CYT997	Cytopia http://www.cytopia.com.au/	Tubulin binding agent	Phase I/II
BCN105	Bionomics http://www.bionomics.com.au/	Tubulin binding agent	Phase I

Disodium combretastatin A-4 3-O-phosphate and other VDAs cause an almost instant drop in blood flow, which generally becomes maximal within 1–4 h (Tozer et al. 2001; Goertz et al. 2002; Hori et al. 2002). Although a wide range of responses have been described, which are primarily dependent on drug type, dose and tumour model, blood flow stops almost completely within one hour of VDA administration, in sensitive tumours, and remains low for more than 24 h, after which it is gradually re-established (Dark et al. 1997; Tozer et al. 2001). The drop in blood flow is brought about by a rapid and dramatic collapse of the tumour vessels, as visualized by techniques such as intravital microscopy (Tozer et al. 2005a). Tumours become necrotic within 24 h of a single VDA dose (Dark et al. 1997), and using vascular casts among other techniques, several studies showed that blood vessels are eradicated primarily within the necrotic areas (Malcontenti-Wilson et al. 2001, 2008; Salmon et al. 2006). Necrosis induction is a feature of the central tumour region and the extent of necrosis correlates with both extent and duration of blood flow interruption (Tozer et al. 2005a,b). In general, in tumours where the vascular response is robust and sustained, necrosis is also extensive and can affect more than 90% of the tumour mass (Tozer et al. 2001). Haemorrhage and often coagulation are also observed several hours after the drug is administered (Prise et al. 2002; Tozer, et al., 1999) When tested at similar doses, Oxi4503 was found to be at least four-fold more effective at reducing tumour perfusion and inducing necrosis, than CA-4-P (Salmon et al. 2006; Salmon & Siemann 2006). Furthermore, tumours treated with Oxi4503 also generally recover from treatment at a slower rate than those treated with CA-4-P, which reflects the higher potency of Oxi4503 (Salmon & Siemann 2006; Salmon et al. 2006). A rapid increase in tumour vascular permeability to macromolecules is also a prominent feature of tumours treated with CA-4-P and other VDAs (Tozer et al. 2001, 2005a; Reyes-Aldasoro et al. 2007), and current hypotheses suggest that this is important for vascular collapse caused by VDAs in general. Normal tissues do not become necrotic at doses that are highly effective in tumours, although they do suffer from some modest transient blood flow reductions (Tozer et al. 1999; Prise et al. 2002; Hori & Saito 2003).

Treatment resistance and approaches to overcome it

Despite the fact that a single VDA dose can cause substantial necrosis, only moderate tumour growth delay is achieved, unless repeated dosing schedules are applied (Horsman & Siemann 2006). Even with repeated dosing regimens, tumours almost always re-grow when treatment stops and this failure has been attributed to a few remaining layers of viable cells, in the peripheral tumour rim. The rim appears to be resistant both in terms of initial blood flow reduction and subsequent necrosis induction (Dark et al. 1997; El-Emir et al. 2005). The vascular network in the tumour rim is often more dense than in the tumour centre and vessels tend to be of larger calibre. Therefore, a relatively greater vascular reserve and more effective perfusion in the tumour rim are likely to contribute to its resistance. Several preclinical models have demonstrated that the outer rim resistance can be overcome by combining VDAs with conventional chemotherapeutics, radiotherapy or even anti-angiogenic agents (Horsman & Siemann 2006). Although interactions are complex, enhanced responses of combined treatments are thought to be at least in part, due to targeting of both the tumour and vascular cell compartments, and this could certainly be the case for chemotherapy and radiotherapy. It is also possible that such combination treatments work better because of spatial co-operation since well-oxygenated tissues respond better to chemotherapy and radiation, and the tumour periphery is likely to be better oxygenated than the centre. Many investigators have tested combinations of various VDAs with conventional modalities, with specific emphasis on dose, timing, and sequence of administration (Horsman & Siemann 2006). Generally, administering chemotherapy first, followed by the VDA ensures that the chemotherapeutic drug reaches the tumour before blood flow is interrupted. This sequence has resulted in enhanced responses and in some instances, evidence has been provided for drug retention within the tumour (Morinaga et al. 2003; Nelkin & Ball 2001; Yeung et al. 2007). Nevertheless, others demonstrated significant therapeutic benefit between chemotherapy and VDAs in the absence of any drug entrapment (Grosios et al. 2000). Precise sequencing appears to be particularly important when combining taxanes with microtubule depolymerizing VDAs as potential antagonism between these two types of agents has been reported on the basis of their opposing effects on the stability of the endothelial microtubule cytoskeleton (Taraboletti et al. 2005). Indeed, in preclinical models ZD6126 failed to cause vascular shutdown if given shortly after paclitaxel (Martinelli et al. 2007). An interval of at least 24 h after paclitaxel was found to be necessary to observe an enhanced response with the combination, at which time presumably the cytoskeleton had recovered from the stabilizing actions of paclitaxel.

Vascular disrupting agents also enhance the activity of radiotherapy when administered either immediately after or a few hours later (Horsman & Siemann 2006). This scheduling ensures that the radiation is effective in oxygenated tissues before they become hypoxic by the actions of the VDA. In addition to spatial co-operation and targeting of different cell populations by the radiotherapy and the VDA, interactions are likely to be more complex based on the fact that both modalities have vascular targets (Tozer et al. 2008b). In a clinical study, non-invasive dynamic computed tomography demonstrated that vascular interactions occurred between radiation and CA-4-P in patients with advanced non-small cell lung cancer (Ng et al. 2007). In this study, radiotherapy caused vascular changes in the tumours rendering them more prone to subsequent vascular shutdown by CA-4-P. Radiotherapy, except when given at very high doses, generally improves blood flow to tumours (Tozer et al. 1991). Recently, Hori et al. (2008) demonstrated that the combretastatin derivative AC7700 resulted in greater anti-tumour activity when given 48 rather than 2 h after a single 5 Gy dose. These authors attributed this synergism on the basis of a VDA-mediated disruption of the improved blood flow caused by the radiotherapy. Significant enhancement in tumour responses has also been demonstrated by combining VDAs with anti-angiogenic agents (Shaked et al. 2006; Siemann & Shi 2008). One study also reported the eradication of the tumour rim with this approach (Shaked et al. 2006). The success of this combination is attributed to inhibition of pro-angiogenic adverse effects, which can occur as a consequence of VDA treatment. Indeed, vascular shutdown causes hypoxia, (El-Emir et al. 2005) which is itself, a strong stimulus of angiogenic gene expression through stabilization of the hypoxia-inducible factor HIF-1α. Elevated levels of both VEGF and basic FGF have been observed in tumours after exposure to VDAs (Boehle et al. 2001; Sheng et al. 2004) thus substantiating this hypothesis. In addition, VDAs themselves can stabilize HIF-1α in tumour cells even in the absence of hypoxia (Dachs et al. 2006). Vascular disrupting agents have also been shown to cause mobilization of endothelial progenitor cells which home to the tumour viable rim area and these cells may be responsible for initiating angiogenesis thus contributing to treatment resistance (Shaked et al. 2006).

Vascular disrupting agents and clinical trials

Many VDAs are now undergoing clinical testing (Table 1). Phase I trials of CA-4-P established that when used as single agent the drug is well tolerated, with myocardial ischaemia, reversible neurological events and tumour pain as the main limiting toxicities (Dowlati et al. 2002; Rustin et al. 2003; Stevenson et al. 2003). These trials also demonstrated selective reductions in blood flow and their findings were consistent with data obtained from preclinical studies (Anderson et al. 2003; Galbraith et al. 2003). Phase I/II clinical trials are now being conducted by OxiGene, in which CA-4-P is tested in combination with radiotherapy, carboplatin, paclitaxel and the anti-VEGF antibody bevacizumab (Avastin). Recent reports show that CA-4-P results in some profound and sustained vascular changes in the presence of bevacizumab and the combination is safe and well tolerated (Nathan et al. 2008). Interestingly, results from this trial also provide evidence for an acute rise of circulating bone marrow progenitors likely to be mediated by VEGF, and this is also consistent with the preclinical findings. Oxi4503 is also being tested as a single agent and early results point to vascular activity at well-tolerated doses (Patterson et al. 2007).

VDA mechanism of action: cellular and molecular mediators

Despite the fact that many VDAs are now being tested in clinical trials, their molecular targets and the mechanisms through which they cause selective reduction of blood flow and the collapse of tumour vessel networks remain largely unexplored. In vitro models have helped establish that the endothelial cytoskeleton is the initial target of microtubule depolymerizing VDAs (Galbraith et al. 2001; Kanthou & Tozer 2002). Damage to microtubules becomes the rapid trigger for further morphological and cytoskeletal changes, through activation of Rho-GTPase, Rho kinase and mitogen activated protein kinase signalling pathways (Kanthou & Tozer 2002). These pathways are responsible for driving CA-4-P-mediated actin remodelling, contractility, blebbing and disruption of cell-to-cell junctions as well as a rise in monolayer permeability in endothelial cells (Kanthou & Tozer 2002). Microtubule integrity itself is crucial for maintaining the stability of tubule-like structures that resemble capillary networks formed by endothelial cells in semi-solid media. Vascular disrupting agents such as CA-4-P and ZD6126 cause rapid collapse of such networks (Ahmed et al. 2003; Bayless & Davis 2004; Vincent et al. 2005) and several studies using various other microtubule targeting drugs showed that capillary-like collapse was Rho-dependent as it could be prevented by Rho inhibitors (Bayless & Davis 2004). Therefore, activated Rho signalling by VDAs may well be the important link between microtubule disruption and vessel collapse, although the details of downstream mechanisms are still not clearly established. Rho proteins control permeability through their effects on actin-myosin contractility and integrity of cell-to-cell VE-cadherin junctions (Kanthou & Tozer 2002; Vincent et al. 2005). VE-cadherin antagonists, when combined with CA-4-P, synergize in disrupting tumour blood vessels (Vincent et al. 2005). A rapid permeability rise is also a feature of early tumour VDA activity in vivo and it is envisaged that this could cause blood flow reduction because fluid loss to the tissues is likely to increase viscous resistance (Tozer et al. 2005a,b). Blebbing and rounding up of endothelial cells, an early morphological feature of their in vitro CA-4-P response, would also exacerbate resistance of flow in vivo. Recently, Yeung et al. (2007) performed ultrastructural analysis in tumour xenografts, 6 h after exposure to CA-4-P and reported morphological changes in endothelial cells including formation of blebs. Changes in interstitial fluid pressure (IFP) as a result of elevated permeability, have also been considered as possible causes of blood flow shutdown (Tozer et al. 2005a,b). However, IFP does not increase after CA-4-P (Ley et al. 2007), although high baseline IFP in tumours is likely to be a determining factor for blood flow shut-down, if intravascular pressure decreases significantly, as is likely following vasoconstriction of up-stream arterioles (Tozer et al. 2001; Hori & Saito 2003). Active vasoconstriction probably occurs through contractile Rho-mediated mechanisms. As blood flow drops, then red cells begin to stack together and contribute to further flow stagnation. Further events such as haemorrhage and clotting, which occur at later time points are then likely to contribute to sustained blood flow shutdown in vivo.

Despite significant links between CA-4-P-induced Rho signalling and morphological and functional changes in endothelial cells, definitive evidence that Rho signalling is associated with vascular collapse by VDAs in vivo is still lacking. However, we have recently showed that the dramatic drop in perfusion of tumour vessels caused by CA-4-P in SW1222 human colorectal carcinoma xenografts, was attenuated if Rho kinase inhibitor Y27632 (Ishizaki et al. 2000) was administered shortly before CA-4-P (G.M. Tozer, C. Kanthou, L. Williams, unpublished data). Furthermore, in these tumours, the Rho kinase inhibitor resulted in a dramatic protective effect against CA-4-P mediated necrosis induction (unpublished data) thus providing the first evidence of the involvement of Rho signalling in CA-4-P mechanisms in vivo.

Tumour blood vessel susceptibility to VDAs

Tumour vessels differ significantly from those of normal tissues, both in terms of morphology and function, and these differences are considered important in determining susceptibility to VDAs. Tumour vessels are fragile, with poorly developed and unstable leaky junctions (Baluk et al. 2005) and their endothelial proliferation index is significantly higher than normal tissues (Denekamp 1982; Eberhard et al. 2000). Already unstable vessels with defective junctions are probably easier to disrupt further by a VDA and this hypothesis indeed appears to be supported by a study which used magnetic resonance imaging to demonstrate greater responses to CA-4-P in tumours which had more permeable vessels before treatment began (Beauregard et al. 2001). Proliferating endothelial cells within tumours (Denekamp 1982; Eberhard et al. 2000) have been suggested to be more sensitive to VDAs than their non-proliferating counterparts in normal tissues, although the mechanisms for such susceptibility have not been defined. It is possible that the cytoskeleton of tumour endothelial cells is particularly sensitive to disruption by VDAs due to expression of specific tubulin isotypes or post-translational modifications to microtubule associated regulatory proteins (Honore et al. 2005). However, to date no evidence for any such differences has been put forward.

The tumour microenvironment is heterogeneous and erratic in terms of blood flow (Tozer et al. 1990) and this may mean that a further reduction in flow becomes more catastrophic in the tumour than in normal tissues. Tumour hypoxia, which is a consequence of the defective blood flow, could also be responsible for sensitizing blood vessels to further VDA damage. This could occur at the cytoskeletal level and indeed there is now significant evidence to suggest that hypoxia directly affects and sensitizes signalling pathways associated with the remodelling of the cytoskeleton of endothelial cells (Wojciak-Stothard et al. 2005). Oxidative stress, which is a major consequence of transient hypoxia in the tumour, could also exacerbate cytoskeletal damage and ‘blebbing’ in endothelial cells induced by VDAs. Endothelial blebbing, is a known oxidative stress response, driven by stress activated protein kinase p38 (SAPK-2) (Houle & Huot 2006). CA-4-P itself activates p38 to induce blebbing (Kanthou & Tozer 2002) and this could be through induction of oxidative stress, which has been reported to occur as a consequence of VDAs and microtubule depolymerizing agents in general (Kurata 2000; Fang et al. 2007; Folkes et al. 2007; Petit et al. 2008). Blebbing of endothelial cells could initially contribute to not only a rapid rise in permeability in the tumour but also to a subsequent loss of endothelial cells from vessels through reduction of cell adherence and induction of necrosis (Kanthou & Tozer 2002).

Pericytes are essential for the maintenance of vessel stability. In tumours, however, pericytes are often difficult to detect, and when present, their contact with endothelial cells is often defective thus contributing to vessel instability and immaturity (Baluk et al. 2005). Tumour abonormalities in pericyte coverage and vessel instability have been put forward as possible explanations for susceptibility to VDAs and several lines of evidence now support this hypothesis (Tozer et al. 2008a). In our laboratory, we developed a series of tumour fibrosarcoma cell lines expressing single VEGF isoforms. The vascular networks formed by these tumours in vivo differ extensively in terms of pericyte recruitment and maturity. VEGF188 isoform expressing tumours display a prominent pericyte coverage and have vessels which are less leaky and less haemorrhagic compared with tumours expressing VEGF164 or VEGF120, which have very few pericytes and leaky and unstable vessels (Tozer et al. 2008a,b). We tested CA-4-P in these models and showed a significantly more profound and sustained vascular damage in the less mature VEGF120 and VEGF164 tumours than in the more mature VEGF188 tumours. The differences in extent of initial vascular damage also translated into increased responsiveness to CA-4-P in terms of tumour growth delay providing compelling evidence for the role of vessel maturity in responsiveness to VDAs (Tozer et al. 2008a). The molecular mechanisms that underlie abnormalities in pericyte recruitment and vessel maturation in tumours are not clearly defined, although factors such as VEGF and PDGF undoubtedly play a significant role. Individual VEGF isoforms clearly have distinct roles in vessel maturation (Tozer et al. 2008a). VEGF was recently shown to act as a negative regulator of pericyte function by disrupting PDGF receptor signalling through interactions with VEGF-R2 (Greenberg et al. 2008). The VEGF188 splice variant is thought to have stronger affinity for VEGF-R1 and may therefore influence maturity through a different mechanism(s) not yet investigated. A clearly defined correlation between VEGF isoform expression and vessel maturity in the tumour would undoubtedly act as a valuable predictive factor of susceptibility to VDAs. Evidence from a spontaneous LH_BETAT_AG retinoblastoma tumour model also supports the hypothesis that maturation protects against VDA-induced damage. In this model, mature vessels invested with pericytes did not show any regression after a local administration of CA-4-P whereas a significant reduction in overall vessel density was evident at 24 h and persisted for at least 1 week after treatment (Jockovich et al. 2007). Resistance to VDA damage has also been modelled in vitro where endothelial cells forming capillary like structures in co-culture with fibroblasts were shown to be resistant to VDA-mediated collapse (Vincent et al. 2005).

Vascular disrupting agents as anti-angiogenic agents in oncology and other pathologies

Microtubule depolymerizing agents such as the Vinca alkaloids have been in clinical practice for several decades, primarily on the basis of their potent anti-mitotic activities on cancer cells (Jordan & Wilson 2004). More recently, these agents have attracted significant interest for their potential as anti-angiogenic agents, in particular when administered at low doses and on a continuous metronomic scheduling (Pasquier et al. 2006). The suitability and clinical efficacy of VDAs in this respect will undoubtedly depend on specific pharmacological characteristics, frequency of administration, dosing as well as combination with other treatments. CA-4-P and other VDAs when administered as single agents but at split doses, result in bigger tumour growth delays, which may suggest some anti-angigogenic activity at this scheduling (Hill et al. 2002a; Ahmed et al. 2003).

An anti-angiogenic scheduling for non-cancer indications should ensure effective and selective actions on the proliferating endothelial cells without induction of necrosis, which would be undesirable in some target organs. Disodium combretastatin A-4 3-O-phosphate has been tested in models of ocular diseases, primarily retinopathies and age related macular degeneration, with some promising outcomes (Randall & Young 2006). Disodium combretastatin A-4 3-O-phosphate when administered on a daily low dose regimen inhibited retinal neo-vascularization in vivo in a neonatal mouse model of oxygen-induced proliferative retinopathy, without affecting the development of normal retinal vasculature suggesting that in this situation the drug was acting primarily as an anti-angiogenic (Griggs et al. 2002). Disodium combretastatin A-4 3-O-phosphate was also found to prevent development of sub-retinal neovascularization in a mouse model of VEGF expression in the retina and to promote regression of established choroidal neovascularization (Nambu et al. 2003). On the other hand, CA-4-P when administered for a period of 6 weeks had no effect on the retinal noevasularization, which develops in galactose-fed dogs (Kador et al. 2007). In this model the disease develops slowly which correlates with the way diabetic retinopathy develops in humans. The results of this study suggest that more long-term administration may be required for targeting slowly proliferating endothelial cells.

Disodium combretastatin A-4 3-O-phosphate is the first VDA to be tested clinically in patients with ocular diseases. Phase I/II trials of CA-4-P in patients with wet age related macular degeneration and of myopic macular degeneration have been conducted. Patients maintained visual acuity in the trials, although optimal doses and delivery schedules have not been established. Further progress in this area now focuses on the development of a topical peri-ocular CA-4-P drug delivery system that will avoid systemic toxicity.

VDA mechanisms associated with inhibition of angiogenesis

Evidence for the anti-angiogenic activities of microtubule depolymerizing drugs was primarily obtained using in vitro models of endothelial cells. These agents are particularly effective against processes associated with angiogenesis, including migration and morphogenesis into 3D capillary-like structures (Vacca et al. 1999; Bocci et al. 2002; Bijman et al. 2006; Pourroy et al. 2006). Vascular disrupting agents were tested in similar models and were found to be active at significantly lower doses than those needed to model vascular disrupting activities. For example, CA-4-P or ZD6126 inhibited endothelial capillary-like formation in matrigel, as well as migration and proliferation at doses in the range of 1–10 nM/l, while significantly higher doses ranging between 0.1 and 10 μM/l were necessary to model VDA activities (Kanthou & Tozer 2002; Ahmed et al. 2003; Vincent et al. 2005). The cytoskeleton directs endothelial migration and morphogenesis and these processes are at least in part, dependent on intact microtubule dynamics (Davis & Camarillo 1995; Connolly et al. 2002; Rodriguez et al. 2003; Honore et al. 2005; Merajver & Usmani 2005; Pasquier et al. 2006). Microtubules alternate between phases of relative stability to phases of alternating rapid shortening and growth, through a process known as dynamic instability. While high doses of microtubule depolymerizing agents cause outright disruption of microtubules, low doses interfere with their dynamic properties and affect motility and morphogenesis without necessarily disrupting their overall structure (Pasquier et al. 2006; Pourroy et al. 2006). Endothelial cells are very sensitive in terms of changes in their microtubule dynamics and therefore, are particularly affected by low doses of microtubule depolymerizing agents. Disruption of mitotic spindles and inhibition of proliferation by microtubule targeting agents is another means by which these compounds target angiogenesis (Bocci, et al., 2002). Exposure of proliferating endothelial cells to low doses of CA-4-P inhibited their growth and induced cell death by mitotic arrest and subsequent cell death by apoptosis-like mechanisms (Iyer et al. 1998; Ahmed et al. 2003; Kanthou et al. 2004). In vivo, efficient targeting of the mitotic endothelial cells would require repeated drug dosing to ensure that proliferating cells are captured in mitosis and this is particularly important with VDAs as these have relatively short half lives in vivo (Prise et al. 2002).

Conclusions

Significant progress has been made in deciphering the mechanisms associated with the vascular activities of small molecular weight VDAs in tumours. Resistance to treatment as a result of failure of VDAs to target the peripheral tumour rim limits their usefulness as single agents. Nevertheless, combination treatments offer avenues for overcoming this resistance. Interactions between endothelial and vascular supporting pericytes undoubtedly influence VDA responsiveness and the focus should now turn towards a better understanding of the molecular mechanisms associated with vessel wall maturity. Undoubtedly this information will allow the development of better strategies for tumour eradication. Vascular disrupting agents also hold great promise as inhibitors of angiogenesis with potential critical applications in cancer and other diseases associated with aberrant vascularization.