Microtubule-targeting agents in angiogenesis: Where do we stand?

2.1. Tumor angiogenesis 

After an initial avascular state, which can last several years following first acquisition of mutations (Folkman and Kalluri, 2004), tumors enter a second phase that involves a switch to the angiogenic phenotype through the constant recruitment of new blood vessels. This angiogenic switch crucially depends on the balance between pro- and anti-angiogenic signals and is greatly influenced by hypoxia. Schematically, when the angiogenic switch turns on in tumors – due to hypoxic stress, low pH, oncogene activation and infiltration of immune cells – the secretion of pro-angiogenic factors increases, leading to the activation of neighbouring endothelial cells. This activation results in the degradation of the basement membrane (BM), migration of endothelial cells and invasion of the extracellular matrix (ECM), proliferation of endothelial cells and formation of a new capillary network (Fig. 1, top). In physiological angiogenesis, this activation phase is followed by a resolution phase, corresponding to the maturation and stabilization of the newly formed microvasculature by pericytes, inhibition of endothelial proliferation, and BM reconstitution. However, this resolution phase is generally incomplete in tumor angiogenesis, resulting in the formation of abnormal vessels developing unique features (i.e. disorganized, chaotic, heterogeneous and leaky, with irregular blood flow and irregular association with perivascular cells). Although little is known about the final steps of tumor angiogenesis, especially how vascular projections fuse with each other to form loops to enable blood to flow in newly vascularized areas, this pathological angiogenesis has been extensively investigated and reviewed (Carmeliet and Jain, 2000, Kerbel, 2000, Bergers and Benjamin, 2003).

View Large Image Download to PowerPoint

Fig. 1. Mechanisms involved in the anti-angiogenic effects of MTDs. Scheme representing the mechanisms involved in tumor angiogenesis (top) as well as those involved in the anti-angiogenic activity of MTDs (bottom). Once the angiogenic switch is turned on in tumors, the secretion of pro-angiogenic factors increases, leading to the activation of neighbouring endothelial cells. This activation results in endothelial cell proliferation and migration, basement membrane (BM) and extracellular matrix (ECM) degradation, and finally capillary tube formation. Bone-marrow-derived circulating endothelial progenitor cells (CEPs) are also involved in tumor angiogenesis. The anti-angiogenic effects of MTDs can be schematically classified in direct effects when they act on endothelial cells and indirect effects when they act on tumor cells. The effect on TSP1 can be viewed as a mixed effect as it is both mediated by endothelial and cancer cells. These cellular effects all result in anti-angiogenesis through the inhibition of endothelial cell migration, proliferation and differentiation as well as ECM and BM degradation. The impairment of CEPs mobilization and viability is also involved, but through mechanisms which remain to be determined.

Recent reports have demonstrated that tumor angiogenesis is also supported by the mobilization and the functional incorporation of bone-marrow-derived circulating endothelial progenitor cells (CEPs) (Rafii et al., 2002, Ribatti, 2004). In some cancers, CEP mobilization has actually been shown to correlate with tumor volume and VEGF production (Mancuso et al., 2001, Monestiroli et al., 2001) and to be required for tumor angiogenesis (Lyden et al., 2001). However, the percentage of CEPs incorporated into tumor blood vessels varies with tumor type, ranging from 95% in B6RV2 lymphoma or 55% in Lewis lung carcinoma (Lyden et al., 2001) to scattered incorporation in other cancers such as neuroblastoma and colon carcinoma (Rafii et al., 2002).

Tumor angiogenesis thus appears as a multi-step process, involving both cancer and endothelial cells, that can be inhibited at several levels. Indeed, each of these steps constitutes a putative target for developing anti-angiogenic therapeutic strategies. In this way, there is a wide variety of molecules in development or even in clinical evaluation that aim to specifically target either the activation and proliferation of endothelial cells or the degradation of BM and ECM and subsequent endothelial cell migration (Kerbel and Folkman, 2002, Eichhorn et al., 2004). It has been shown that MTAs can inhibit tumor angiogenesis at different levels and their anti-angiogenic effects can be direct or indirect, affecting endothelial or cancer cells, respectively.

2.2. Direct effects of MTAs on endothelial cells 

As mentioned above, the anti-angiogenic properties of MTAs were discovered over a decade ago in in vitro studies (Ettenson and Gotlieb, 1992, Belotti et al., 1996, Vacca et al., 1999). These works showed that both MT-stabilizing agents, such as paclitaxel, and MT-depolymerizing agents, such as vinblastine, can inhibit several functions of endothelial cells involved in angiogenesis: proliferation, migration, capillary-like structure formation on Matrigel™, and degradation of BM and/or secretion of metalloproteases (Belotti et al., 1996, Vacca et al., 1999). These anti-angiogenic properties were then confirmed (Ribatti et al., 2003, Pasquier et al., 2004) and extended to most of the MTAs, including docetaxel (Vacca et al., 2002, Hotchkiss et al., 2002), epothilones (Bocci et al., 2002, Woltering et al., 2003), 2-methoxyestradiol (Klauber et al., 1997, Mooberry, 2003, Stafford et al., 2005), vincristine, vindesine, vinblastine, vinorelbine (Hayot et al., 2002) and more recently vinflunine (Pourroy et al., 2006). Some of the MTAs, mainly the depolymerizing agents, also exhibit “vascular-targeting” properties leading to a rapid collapse of existing tumor vessels through MT depolymerization. These vascular-targeting MTAs include Vinca alkaloids, such as vinblastine (Bayless and Davis, 2004) and vinflunine (Holwell et al., 2001), as well as new tubulin-binding agents which were developed for this purpose, such as combretastatin A4 (Galbraith et al., 2001, Tozer et al., 2001), its analogues (Hua et al., 2003, Peifer et al., 2006) and the phosphate prodrug of N-acetylcolchinol ZD-6126 (Davis et al., 2002, Micheletti et al., 2003). In contrast to the general anti-angiogenic activity, this “vascular-targeting” activity involves rapid and dramatic changes in endothelial cell morphology that occur at relatively high drug concentrations, ultimately leading to vascular shutdown. This process has been well reviewed (Thorpe, 2004, Eichhorn et al., 2004, Tozer et al., 2005). Here, therefore, we will focus on the anti-angiogenic activity of MTAs.

Interestingly, the MTA-induced anti-angiogenic effects appeared at very low concentrations, as compared to those required to affect cancer cells (Bocci et al., 2002, Grant et al., 2003). Comparing fibroblast, endothelial and cancer cells on the one hand, and, short and long-term exposure on the other, Bocci et al. have demonstrated that protracted low concentrations of paclitaxel and epothilones specifically inhibited endothelial cell proliferation (Bocci et al., 2002). This selectivity of MTAs for human endothelial cells has been then confirmed in keratinocytes, fibroblasts, epithelial and endothelial cells by comparing paclitaxel with 5-fluorouracil, doxorubicin and camptothecin (Wang et al., 2003). Moreover, it has been shown that MTAs can inhibit both the migration of endothelial cells and the formation of capillary-like structures on Matrigel™, at very low concentrations that affect neither endothelial cell proliferation nor the structural organization of the MT network (Hotchkiss et al., 2002, Wang et al., 2003, Pourroy et al., 2006). Thus, even if anti-angiogenic effects can occur at cytotoxic concentrations (Belotti et al., 1996, Vacca et al., 1999), these effects are often maximal at very low non-cytotoxic concentrations (Belotti et al., 1996, Hayot et al., 2002, Pasquier et al., 2004, Pourroy et al., 2006). Finally, several in vitro studies suggested that chemotherapeutic drugs that do not target MTs are less effective at inhibiting angiogenesis than are MTAs (Farinelle et al., 2000, Hayot et al., 2002, Wang et al., 2003, Drevs et al., 2004). Taken together, these results demonstrate that human endothelial cells are extremely sensitive to MTAs and, conversely, that MTAs act on these cells through specific mechanisms, which are likely to be distinct from the ones involved in cancer cells.

2.3. Mechanisms proposed for direct anti-angiogenic effects of MTAs 

Recently, several mechanisms have been suggested to explain the high sensitivity of endothelial cells to MTAs. Hotchkiss et al. demonstrated that low concentrations of docetaxel impaired the repositioning of the MT-organizing center (MTOC)—structure involved in cell chemotaxis and migration. They suggested that the anti-migratory effects of MTAs on endothelial cells may be mediated by changes in MT plasticity rather than in the global structure of the MT network (Hotchkiss et al., 2002). In agreement with this hypothesis, we have recently demonstrated that low anti-angiogenic concentrations of MTAs – both MT-stabilizing and depolymerizing agents – induced an unexpected increase in MT dynamic instability in human endothelial cells (Pasquier et al., 2005, Pourroy et al., 2006). This increase in MT dynamics was associated with an inhibition of cell motility (Pourroy et al., 2006), which could be due to impairment of cell polarization and/or of MT-induced focal adhesion turnover. These studies revealed dramatic changes in interphase MT dynamics that had no effect on the mitotic process but rather on MT interphase functions, highlighting for the first time a putative role of interphase MT dynamics in angiogenesis (Pasquier et al., 2005, Pourroy et al., 2006).

The increase in MT dynamic instability seems to be specific for endothelial cells, since dynamic instability decreased at all tested MTA concentrations in tumor cells (Pasquier et al., 2005). So far, MTAs have been described to inhibit MT dynamics in tumor cells, resulting in an impairment of the metaphase-to-anaphase transition and in a subsequent mitotic block and apoptosis induction (Jordan, 2002, Jordan and Wilson, 2004, Honore et al., 2005). The increase in MT dynamic instability induced by MTAs in endothelial cells has never been described in any other cell type, suggesting that endothelial cells possess special mechanisms for regulating MT dynamics, which could in part explain the sensitivity of these cells to MTAs (Pasquier et al., 2005, Pourroy et al., 2006). Therefore, the endothelial cell MT network probably differs in its composition in tubulin isotypes, tubulin post-translational modifications and/or proteins that regulate MT dynamics. Identification of these specific differences may identify novel targets for cancer therapy.

In addition to the unusual sensitivity of the MT network in endothelial cells, several other mechanisms may explain the direct anti-angiogenic effects of MTAs. For example, paclitaxel uptake has recently been shown to be increased up to five-fold in endothelial cells as compared with fibroblasts and cancer cells (Merchan et al., 2005). In addition, it has been shown that docetaxel can induce CD95 presentation by endothelial cells in vitro, suggesting that CD95 could also be involved in angiogenesis inhibition by MTAs (Yap et al., 2005). However, this modest increase in CD95 appeared, in vitro, at concentrations where apoptosis was already prominent and it only resulted in a mild 1.5-fold increase in vivo. We have previously demonstrated that low anti-angiogenic concentrations of paclitaxel exert a cytostatic effect on human endothelial cells through a slowing down of the cell cycle (Pasquier et al., 2004). This cytostatic effect was associated with the initiation, without completion, of the mitochondrial apoptotic pathway, as shown by the early and transient increase in Bax/Bcl-2 ratio as well as in the mitochondrial reducing and membrane potentials. It is thus reasonable to think that mitochondria could also be involved in the anti-angiogenic activity of MTAs. Accordingly, several anti-mitochondrial agents, such as betulinic acid or a peptide trivalent arsenical, have been shown to exhibit potent anti-angiogenic properties in vitro (Kwon et al., 2002, Don et al., 2003). In addition, mitochondria are well known as key actors in the cytotoxic effects of MTAs in cancer cells (Andre et al., 2002, Varbiro et al., 2001, Carre and Braguer, in press). Mitochondria may thus be involved in the anti-angiogenic effects of MTAs, but the rationale for their involvement in the specific sensitivity of endothelial cells is still lacking.

Several recent studies suggest that mobilization and functional incorporation of CEPs are involved in tumor angiogenesis (see above) (Rafii et al., 2002, Ribatti, 2004). It has recently been shown that metronomic treatment of mice bearing tumor cells with vinblastine, vinorelbine, or cyclophosphamide can induce a persistent decrease in CEP mobilization and viability (Bertolini et al., 2003, Shaked et al., 2005). Thus, the impairment of CEPs could also be involved in the anti-angiogenic activity of MTDs, but this possibility requires further investigation.

2.4. Indirect effects of MTAs on tumor cells 

The most important explanation for the indirect anti-angiogenic effect exerted by MTAs may be the antitumor effect itself, because it ultimately results in a decrease in the secretion of pro-angiogenic factors. Besides their growth factor activity, VEGF and bFGF are known to protect endothelial cells from apoptosis in several ways, including both the activation of MAPK, JNK and PI3K survival pathways, and the up-regulation of Bcl-2 and survivin (Nor et al., 2001, Tran et al., 2002, Miller et al., 2005a). In particular, one study showed that VEGF was more effective in protecting endothelial cells against MTAs than against DNA-damaging agents (Tran et al., 2002). This may be a consequence of survivin up-regulation, since survivin is an important MT-binding apoptosis inhibitor involved in mitotic spindle regulation (Li et al., 1998, Giodini et al., 2002). Thus, by directly killing tumor cells, MTAs may decrease the protection of endothelial cells by the tumor micro-environment.

Furthermore, low non-cytotoxic concentrations of paclitaxel exhibit in vivo anti-angiogenic properties in mice, in part through VEGF down-regulation (Lau et al., 1999). Docetaxel and vincristine can also directly down-regulate VEGF in human leukaemia T-cells lines even when they are drug-resistant (Avramis et al., 2001). This effect on VEGF may be mediated by the inhibition of hypoxia inducible factor (HIF)-1α synthesis—HIF-1α being the major hypoxia-induced transcription factor of VEGF. In fact, Giannakakou and co-workers have recently found that both MT-stabilizing and depolymerizing agents specifically inhibit HIF-1α expression in tumor cells (Mabjeesh et al., 2003, Escuin et al., 2005). This inhibition occurs at the translational level and is dependent on significant MT disturbances—either depolymerization or stabilization. Since 15–30% of cellular mRNA is associated with the cytoskeleton (Jansen, 2001), it is possible that this includes the mRNA for HIF-1α, which could thus ensure an effective and targeted translation. As suggested by Giannakakou et al., the anti-HIF-1α effect of docetaxel may be of crucial importance since this agent is currently used for the treatment of advanced hormone-refractory prostate cancers, which overexpress HIF-1α. Moreover, the mitotic index is about 2% in these cancers, suggesting that the effectiveness of docetaxel in treating this kind of cancer could stem from its inhibition of HIF-1α rather than from its anti-mitotic effects (Escuin et al., 2005).

Another interesting mechanism for the anti-angiogenic activity of MTAs involves thrombospondin 1 (TSP1) as a potential mediator of the effects of low-dose metronomic chemotherapy. TSP1 is a well-known endogenous inhibitor of angiogenesis, mainly acting through binding to CD36 receptors that are selectively expressed by endothelial cells (Lawler, 2002). In addition, TSP1 inhibits the pro-angiogenic and pro-survival activities of VEGF and FGF in part through sequestrative binding (Gupta et al., 1999, Rodriguez-Manzaneque et al., 2001, Margosio et al., 2003). In vivo evidence for the involvement of TSP1 as a secondary mediator has been provided by investigating metronomic treatment with cyclophosphamide. So far, thanks to its high water solubility and in vitro anti-angiogenic properties, cyclophosphamide is one of the chemotherapeutic agents most studied in metronomic schedules in mice (Bocci et al., 2003, Hamano et al., 2004) and in the clinic (Colleoni et al., 2006). For example, the anti-angiogenic and antitumor effects of a low-dose metronomic treatment with cyclophosphamide have been studied in Tsp1-null mice bearing subcutaneously transplanted Lewis lung carcinoma tumors or B16 mouse melanoma (Hamano et al., 2004). This study demonstrated that, in sharp contrast to wild-type mice, the treatment had no effect in Tsp1-null mice. This involvement of TSP1 may explain the selectivity of MTA-based metronomic chemotherapy for endothelial cells. Indeed, Bocci et al. have demonstrated that long-term exposure to low concentrations of paclitaxel or cyclophosphamide induced a marked increase in both the expression and the secretion of TSP1 by endothelial cells in vitro (Bocci et al., 2003). The role of TSP1 was further confirmed by the use of anti-TSP1 antibodies that increased the survival of endothelial cells after metronomic treatment (Bocci et al., 2003). This effect on TSP1 may be both direct and indirect, since this induction occurs in endothelial cells in vitro (Bocci et al., 2003) and in tumor and perivascular cells (Hamano et al., 2004). TSP1 is down-regulated in some cancers such as prostate cancer (Vallbo et al., 2004). Significantly, Damber et al. showed that low-doses of paclitaxel and cyclophosphamide can re-induce TSP1 expression in rat prostate tumors not expressing TSP1 (Damber et al., 2005). Further, Bocci et al. suggested that the increase in TSP1 expression could be involved in the impairment of CEP mobilization and viability (Bocci et al., 2003), observed after metronomic chemotherapy (see above) (Bertolini et al., 2003, Shaked et al., 2005).

Thus, the anti-angiogenic activity of MTAs appears to be mediated through several mechanisms, either direct or indirect, which act complementarily and which all result in angiogenesis inhibition in vivo, as summarized in Table 1 and Fig. 1. Elsewhere, it is clear that these mechanisms are not exclusive and that there are still missing pieces. The complete elucidation of these mechanisms will improve our understanding of how MTAs are anti-angiogenic in vivo and will help to develop new therapeutic schedules.

Table 1. Cellular effects of MTAs associated with their anti-angiogenic activity

	MTA	Models	Cellular effects	References
	Direct effects (against endothelial cells)
	Docetaxel	HUVEC	Reduction in the reorientation of the MTOC toward the leading edge of migrating cells in vitro	Hotchkiss et al. (2002)
	Paclitaxel	HUVEC	Initiation, without completion, of the mitochondrial apoptotic pathway in vitro, leading to a slowing down of the cell cycle	Pasquier et al. (2004)
		HMEC-1
	Paclitaxel; Vinflunine	HUVEC	Increase in interphase MT dynamics in vitro	Pasquier et al. (2005)
		HMEC-1		Pourroy et al. (2006)
	Paclitaxel	HUVEC	Increase in the drug cellular uptake in human endothelial cells as compared with fibroblasts and tumor cells in vitro	Merchan et al. (2005)
		HMVEC-L
		HMVEC-D
	Docetaxel	Human endothelial cells, Matrigel plug in C57BL/6 mice	Modest induction of endothelial CD95 in vitro and in vivo	Yap et al. (2005)
	Vinblastine	SCID mice bearing murine or human breast cancer	Rapid decline in CEPs viability in vivo	Shaked et al. (2005)
	Vinorelbine (three times/week)
	Indirect effects (against tumor cells)
	Paclitaxel; Docetaxel	Nude mice bearing murine breast cancer; human leukemia cell lines	VEGF down-regulation in vitro (even in drug resistant cells) and in vivo	Lau et al. (1999)
	Vincristine			Avramis et al. (2001)
	2-ME2; Docetaxel	Human cancer cell lines (breast, glioblastoma, lung, ovarian, prostate, etc.)	Inhibition of HIF-1α in vitro at the translational level and downstream MT disruption, leading to VEGF down-regulation	Mabjeesh et al. (2003)
	Epothilone B Vincristine			Escuin et al. (2005)
	Vinblastine Colchicine
	Mixed effects
	Paclitaxel	HUVEC	Increase in TSP-1 expression in vitro	Bocci et al. (2003)
	Epothilone B	HDMEC
	BMS-275183 (new oral taxane)
	Paclitaxel (continuous s.c. injections)	Rats bearing syngeneic prostate cancer not expressing TSP-1	Re-induction of TSP-1 expression in vivo	Damber et al. (2005)

3.1. Metronomic taxane therapy: weekly versus daily schedules 

Numerous studies support weekly taxane dosing as an active regimen (Eniu et al., 2005, Seidman, 2005) and this kind of schedule is becoming increasingly common in the treatment of human cancers. Interestingly, patients with metastatic breast or ovarian cancers who no longer respond to every-3-weeks administration can respond to the same drug when given at lower doses on a much more frequent basis (i.e. weekly schedule) (Fennelly et al., 1997, Greco, 1999, Burstein et al., 2000, Gately and Kerbel, 2001). These results strongly suggest that such therapeutic schedules could overcome drug resistance by switching from one target to another, namely from tumor to the endothelial compartment. Moreover, a significant therapeutic improvement has been shown with weekly administration of paclitaxel as compared to once every-3-weeks administration for metastatic breast cancer in recent randomized clinical trials (Seidman, 2005, Green et al., 2005). Interestingly, the combination of this treatment schedule with bevacizumab significantly prolongs time-to-tumor progression in women with metastatic breast cancer (Miller et al., 2005b). This clinical study clearly demonstrated the synergism between weekly paclitaxel and “pure” anti-angiogenic therapy, as it had been previously shown in vitro for docetaxel (Sweeney et al., 2001). Metronomic therapy can also be combined with standard chemotherapy regimens. For example, a phase III randomized trial combining standard carboplatin- and paclitaxel-based chemotherapy with continued weekly low-dose paclitaxel in ovarian cancer is currently recruiting patients [Gynecologic Oncology Group, NCI, NCT 00003644].

However, most of these weekly regimens are still dose-intense (80–100mg/m2/week versus 50–60mg/m2/week for weekly and every-3-weeks administration, respectively) and delayed toxicities, while greatly reduced, are common. Moreover, such regimens may not be fully optimized for anti-angiogenic activities, as suggested by some in vitro data. For example, it has been shown that endothelial growth arrest persists for 3 days after paclitaxel withdrawal, suggesting incomplete inhibition of angiogenesis for injection intervals longer than 3 days (Kunstfeld et al., 2003). This may explain the poor outcome of some clinical trials involving weekly paclitaxel (Nabholtz et al., 1996, Chang et al., 1997, Forastiere et al., 2001, Kahl et al., 2005) and the promising results of clinical trials involving low dose-intensity and/or daily paclitaxel administration (Rathmann et al., 1999, Massacesi et al., 2005, Jeremic et al., 2005a; Jeremic et al., 2005b).

It is noteworthy that daily low doses of paclitaxel appeared to be feasible in phase I trials when employed with radiation therapy in patients with non-small cell lung cancer (Gobitti et al., 2005). Moreover, phase II studies analysing the safety and efficacy of daily low dose paclitaxel and carboplatin, with concurrent hyperfractionated radiotherapy in either stage I/II or III non-small cell lung cancer showed results that were among the best ever reported for this type of cancer (Jeremic et al., 2005a, Jeremic et al., 2005b)—with objective response rates about 80% and very promising results for the median survival time, the median time-to-local progression and the median time-to-distant metastasis. A phase I clinical trial of daily metronomic docetaxel is also currently planned for treatment of refractory or recurrent advanced gynecologic cancer [University of Minesota, NCI, NCT 00287885]. Finally, it has been also recently shown, in SCID mice bearing human ovarian cancer xenografts, that the anti-vascular properties of metronomic paclitaxel could be enhanced when combined with SU6668, an inhibitor of the VEGFR2, FGFR1 and PDGFRβ tyrosine kinase (Garofalo et al., 2003; Naumova et al., 2006). Altogether, these studies strongly suggest that metronomic taxane therapy could be a promising approach to treat cancer, especially when used in combination strategies.

3.2. Metronomic Vinca alkaloid therapy 

The other major class of MTAs, namely the Vinca alkaloids, also exhibit potent anti-angiogenic activity in vivo when administered metronomically. The treatment of mice bearing childhood neuroblastoma xenografts with 1.5mg/m2 vinblastine twice weekly – which is about one tenth to one twentieth of the maximum tolerated dose in mice – resulted in growth delays through a strong anti-angiogenic process (Klement et al., 2000). However, Klement et al. demonstrated that tumor growth was not stopped indefinitely and resumed after an initial response. These limitations were also observed when mice were treated with a “pure” angiogenesis inhibitor (anti-VEGFR2 antibody (DC101)) as monotherapy (Klement et al., 2000). Interestingly, the anti-angiogenic effects of metronomic low-dose vinblastine were enhanced by combination with DC101 antibody, resulting in long-term remission (Klement et al., 2000, Klement et al., 2002). These results confirmed the synergistic effect of MTA-based metronomic therapy and “pure” anti-angiogenic therapy. Moreover, these studies showed that MTA-based metronomic therapy can provide a stable and safe way to circumvent drug resistance as long as MTAs are used in combination with a second anti-angiogenic drug.

In addition, a combination of anti-angiogenic chemotherapies (i.e. weekly vinorelbine and daily low-dose cyclophosphamide) was found to be feasible and effective against recurrent sarcomas (Casanova et al., 2004). A phase II trial evaluating metronomic oral vinorelbine (three times a week continuously) in recurrent breast cancer, non-small cell lung cancer and metastatic prostate cancer is currently recruiting patients [Hellenic Cooperative Oncology Group, University of Ioannina, Medical School, NCT 00278070].

4.1. Formulation problems 

The concept of anti-angiogenic chemotherapy implies long-term treatment with drug administrations that must be as frequent as possible. For this reason, there is an obvious need to develop oral drug formulations. Such formulations of MTAs are not presently available except for vinorelbine (oral Navelbine®), which is unfortunately one of the least anti-angiogenic MTAs (Hayot et al., 2002, Kruckzynski et al., 2003). However, several oral formulations of MTAs, such as oral taxanes, are currently in development or in clinical evaluation (Malingre et al., 2001, Pratesi et al., 2003, Taraboletti et al., 2003, Kamath et al., 2005a). In particular, two new oral formulations of taxane have recently cleared phase I clinical trials (Bröker et al., 2006; Novagli Pharma, unpublished data). These new compounds and formulations may enable patients to receive the drug metronomically as outpatients, potentially improving the antitumor effectiveness while reducing treatment restraints, toxic side effects and even overall costs. For example, a new tubulin-binding compound (EM015) has been recently rationally designed (Aneja et al., 2006). This new MTA appears to be effective in vitro against many breast tumor cell lines even when they are drug resistant and, in vivo, against breast tumor xenografts. Furthermore, this compound is orally available and exerts no detectable toxicity in mice, which makes it a good candidate for MTA-based anti-angiogenic therapeutic strategies. Further investigations are now required to determine the anti-angiogenic potential of EM015 as well as others new orally available MTAs.

Ng et al. recently showed that clinically relevant concentrations of vehicles cremophor EL and polysorbate 80 nullified the anti-angiogenic activity of both paclitaxel and docetaxel (Ng et al., 2004). These results suggest that the traditionally formulated drugs (Taxol® and Taxotere®) might need to be administered at higher doses than anticipated to achieve anti-angiogenic chemotherapy, possibly cancelling any benefit of reduced toxicity (Ng et al., 2004). For this reason, novel cremophor-free paclitaxel formulations are currently under development. Albumin-bound paclitaxel (Abraxane®), for example, has been shown to have potent in vitro anti-angiogenic properties, inhibiting endothelial cell proliferation and tube formation on Matrigel™ at very low concentrations (Ng et al., 2005), similarly to cremophor-free paclitaxel (Belotti et al., 1996, Grant et al., 2003, Pasquier et al., 2004). Interestingly, optimal metronomic low doses of albumin-bound paclitaxel, as determined by measuring CEP levels in peripheral blood, were sufficient to inhibit tumor growth in SCID mice bearing human breast and prostate cancer xenografts to the same extent as maximun tolerated doses of Taxol®. In addition, albumin-bound paclitaxel demonstrated in vivo anti-angiogenic activity as shown by a decrease in measurable FGF in a Matrigel™ plug assay. In this model, neither metronomic nor maximum tolerated dose of Taxol® inhibited in vivo angiogenesis (Ng et al., 2005).

Encapsulation of paclitaxel in cationic liposomes has also been shown to improve its antitumor and anti-angiogenic effects in vivo (Kunstfeld et al., 2003, Schmitt-Sody et al., 2003, Strieth et al., 2004). These cationic liposomes selectively accumulate in activated tumor microvessels and thus appear to be very promising vehicles to direct chemotherapeutic substances to tumor endothelium (Campbell et al., 2002, Strieth et al., 2004). However, in one of these studies, the characteristics of impaired tumor microvasculature induced by encapsulated-paclitaxel (reduction of functional vessel density and microcirculatory perfusion index) support the targeting of existing tumor blood vessels (“neovascular targeting”) as the underlying mechanism rather than anti-angiogenesis.

4.2. “Anti-angiogenic resistance” 

As mentioned above, Klement et al. have pointed out the limitations of anti-angiogenic monotherapies that only induced delayed tumor growth but not long-term remission in mice (Klement et al., 2000). These results clearly support a lack of efficacy and even an emergence of resistance to both “pure” anti-angiogenic strategies and metronomic chemotherapy, when used as monotherapy. In accordance with this, Man et al. demonstrated that tumors that re-emerged after initial response to cyclophosphamide-based metronomic therapy responded again to the same treatment protocol upon implantation in a new host (Man et al., 2002). Interestingly, the synergistic effects reported for anti-angiogenic combination strategies strongly suggest that these strategies may be able to delay the appearance of endothelial cell resistance (Klement et al., 2000, Miller et al., 2005b).

Recently, some putative mechanisms responsible for the emergence of anti-angiogenic resistance have been reviewed (Broxterman et al., 2003, Miller et al., 2005a). They include the redundancy of pro-angiogenic factors, the protection of endothelial cells by tumor micro-environment as well as the decrease in tumor-dependence to angiogenesis. Indeed, tumor regrowth can be, in some cases, independent from angiogenesis and tumor cells that are resistant to hypoxia can be selected. Moreover, some tumors have been shown to be able to co-opt pre-existing vessels or even to mimic a vascular network without recruiting endothelial cells.

Endothelial cells, not being themselves tumor cells, were thought to be genetically stable a priori. Hence, anti-angiogenic therapies were first hypothesized to not be susceptible to acquired drug resistance (Kerbel, 1991). However, recent studies suggest that, in some cases, endothelial cells associated with tumors may not be genetically stable and may indeed acquire drug resistance. Hida et al. have thus described cytogenetic abnormalities in murine endothelial cells isolated from human tumor xenografts (aneuploidy, multiple abnormal centrosomes), although the mechanism for the acquisition of such alterations remains unknown (Hida et al., 2004). In addition, it has been reported that a significant percentage of endothelial cells in human B-cell lymphomas could harbour lymphoma-specific chromosomal abnormalities (Streubel et al., 2004). Acquired resistance by endothelial cells to low metronomic MTAs may also involve mechanisms similar to those described in tumor cells. In accordance with this, expression of the major efflux pump, P-gp, has been described in endothelial cells isolated from rat fibrosarcoma (Iwahana et al., 1998).

Although tubulin has generally been regarded as exclusively a cytoplasmic protein, Walss and colleagues found that the βII isotype of tubulin occurs in non-MT form in the nuclei of cultured cancer cells (Walss et al., 1999; Walss-Bass et al., 2001). Using excised samples of human cancers, Yeh and Ludueña observed that many tumors exhibit nuclear βII tubulin (Yeh and Ludueña, 2004). This was true even for cells that, when not transformed, do not express βII tubulin. The relevant finding, however, was that otherwise normal cells adjacent to a tumor also exhibited nuclear βII tubulin in many cases. This was particularly striking in the case of lymphocytes, which normally do not express detectable βII tubulin, but which exhibited nuclear βII tubulin when they were associated with metastatic breast cancer cells (Yeh and Ludueña, 2004). The phenomenon was observed with endothelial cells as well (Yeh and Ludueña, unpublished observation). One can speculate that the acquisition of drug resistance by endothelial cells and the induction of βII tubulin expression and its localization to the nuclei occur by parallel mechanisms and are somehow induced by the nearby tumor cells. It is even possible that one mechanism causes the other. For example, induction of βII tubulin in a cell will obviously change the relative cellular concentrations of the different tubulin isotypes. Furthermore, these isotypes are known to differ in their affinities for MTAs (Banerjee and Luduena, 1992, Derry et al., 1997, Lobert et al., 1998) and modulations of the MT isotypic composition have been shown to be associated with acquired resistance to MTAs in cancer cells (Burkhart et al., 2001, Kamath et al., 2005b, Seve et al., 2005; Shalli et al., 2005). Finally, there is evidence that the αβII tubulin dimer, in non-MT form, binds to the nuclear membrane and may conceivably play a role in nuclear organization. If induction of βII tubulin somehow disorganizes the nucleus of an endothelial cell, then it is not difficult to imagine that the cell may become more susceptible to mutation and hence become likely to develop drug resistance.

MT dynamics, which are finely regulated in cells, need to be maintained in a narrow range to support the various cell functions in which MTs are involved (Honore et al., 2005). We have previously demonstrated that an increase in MT dynamic instability could be responsible for resistance to continuous low concentrations of paclitaxel in A549 tumor cells (Goncalves et al., 2001). Long-term exposure of tumor cells to low concentrations of paclitaxel induced a modification of basal MT dynamic instability in a way that compensated for the drug effect and maintained MT dynamics in a window compatible with viability. Interestingly, we have also demonstrated that MTAs exert a biphasic effect on MT dynamics in endothelial cells in vitro (Pasquier et al., 2005, Pourroy et al., 2006). While cytotoxic concentrations of MTAs classically suppressed MT dynamics in endothelial cells, low and non-cytotoxic concentrations of MTAs markedly increased dynamics. It is thus conceivable that, in vivo, endothelial cell would adapt differently to a long-term exposure, according to the concentrations (either low and non-cytotoxic or high and cytotoxic) of MTAs and/or of “pure” anti-angiogenic drugs. If that is the case, such “acquired-resistance” may be taken as an advantage, by alternating anti-angiogenic metronomic therapy at cytotoxic and non-cytotoxic concentrations of MTAs. This putative interesting strategy needs to be investigated both in vitro and in animal models.

4.3. What is next? 

The future development of MTA-based metronomic strategies may improve the outcome of cancer treatment, possibly reducing the toxic side effects while increasing treatment efficacy. In addition, these therapeutic strategies may apply to a wide variety of human cancers, even those which are hard to treat or even resistant to chemotherapy. For example, it has been suggested that, despite the initial disappointing experience with taxanes in paediatric oncology, metronomic taxanes may be used to treat childhood cancers (Andre and Meille, 2006). For now, the future development of these anti-cancer strategies will have to critically address the sequence and dosing intervals as well as the potential synergistic combinations. Indeed, even if several synergistic effects of combination strategies have been demonstrated or suggested (see above), an antagonism between paclitaxel and ZD-6126 has been recently described in vitro and in mice when paclitaxel was used at high cytotoxic concentrations (Taraboletti et al., 2005). Doses and schedules thus appear to be very critical in metronomic chemotherapy and anti-angiogenic combinatory strategies.

Additional clinical trials are now required to validate and clinically develop MTAs for their anti-angiogenic activity. Though they have demonstrated interesting therapeutic efficacy, most of the metronomic strategies have so far been clinically evaluated in heavily pre-treated patients. However, it is reasonable to think that such strategies may be more effective in patients with early stage cancer and should be evaluated as such in the clinic. In addition, it is important to accelerate the development of oral MTAs for metronomic treatment.

Last but not least, reliable biomarkers are needed to monitor the activity of anti-angiogenic drugs and low dose metronomic chemotherapy. So far, the absence of such biomarkers may have impaired clinical development of various anti-angiogenic drugs and/or metronomic schedules. Most of clinical trials involving metronomic MTAs evaluate the effectiveness of such therapeutic strategies but, because of the lack of biomarkers, these studies do not investigate the mechanisms responsible for this efficacy. Circulating endothelial cells and their progenitor subset, the vascular permeability and flow (as determined by magnetic resonance imaging dynamic measurement), and circulating VEGF levels are potential candidates (Morgan et al., 2003, Willett et al., 2004, Shaked et al., 2005, Colleoni et al., 2006). Recently, circulating endothelial cell kinetics and viability has been shown to predict survival in breast cancer patients receiving metronomic chemotherapy (Mancuso et al., in press), although their long-term predictive value for other tumors remains to be established.