Combination of antiangiogenesis with chemotherapy for more effective cancer treatment Free

Angiogenesis is a hallmark of tumor development and metastasis and is now a validated target for cancer treatment. However, the survival benefits of antiangiogenic drugs have thus far been rather modest, stimulating interest in developing more effective ways to combine antiangiogenic drugs with established chemotherapies. This review discusses recent progress and emerging challenges in this field; interactions between antiangiogenic drugs and conventional chemotherapeutic agents are examined, and strategies for the optimization of combination therapies are discussed. Antiangiogenic drugs such as the anti-vascular endothelial growth factor antibody bevacizumab can induce a functional normalization of the tumor vasculature that is transient and can potentiate the activity of coadministered chemoradiotherapies. However, chronic angiogenesis inhibition typically reduces tumor uptake of coadministered chemotherapeutics, indicating a need to explore new approaches, including intermittent treatment schedules and provascular strategies to increase chemotherapeutic drug exposure. In cases where antiangiogenesis-induced tumor cell starvation augments the intrinsic cytotoxic effects of a conventional chemotherapeutic drug, combination therapy may increase antitumor activity despite a decrease in cytotoxic drug exposure. As new angiogenesis inhibitors enter the clinic, reliable surrogate markers are needed to monitor the progress of antiangiogenic therapies and to identify responsive patients. New targets for antiangiogenesis continue to be discovered, increasing the opportunities to interdict tumor angiogenesis and circumvent resistance mechanisms that may emerge with chronic use of these drugs. [Mol Cancer Ther 2008;07(12):3670–84]

Angiogenesis is a highly regulated process, whereby new blood vessels form from preexisting ones (1). In adult mammals, physiologic angiogenesis is largely limited to the ovaries, uterus, and placenta, with the turnover rate of vascular endothelial cells being very low in most other tissues. Pathophysiologic angiogenesis is a characteristic of wound healing and diseased states, particularly cancer, where the number of proliferating endothelial cells increases significantly and the morphology of the vasculature is altered in multiple ways (2). For many types of cancer, as tumor cells undergo dysregulated proliferation, the tumor mass initially expands beyond the support capacity of the existing vasculature, leading to decreased levels of oxygen and nutrients and the accumulation of metabolic wastes. Tumor cells respond to this deterioration of the tumor microenvironment by up-regulating several proangiogenic factors, including vascular endothelial growth factor (VEGF)-A, basic fibroblast growth factor, placental growth factor, and platelet-derived endothelial growth factor, which collectively activate quiescent endothelial cells and promote their migration into the tumor. This shift of the tumor microenvironment to an angiogenic state, or “angiogenic switch” (3), is an important rate-limiting factor in tumor development. Despite the active angiogenesis induced by tumor cell-derived proangiogenic factors, structural defects associated with the tumor vasculature often lead to inefficient blood perfusion in established tumors, which contributes to tumor hypoxia. Tumor metastasis is also regulated by angiogenesis, as well as by lymphangiogenesis, where new lymphatic vessels are formed from preexisting ones (4). Tumor cell dissemination, the first step in tumor metastasis, requires access to both blood and lymphatic circulation. Once successfully extravasated, the survival and further colonization of the disseminated tumor cells is dependent on angiogenesis at the secondary site. Angiogenesis is thus a key factor in the development and metastasis of a variety of tumor types and is an important hallmark of malignant disease. Moreover, angiogenesis presents unique opportunities for therapeutic intervention in cancer treatment, as first proposed by the late Judah Folkman more than 35 years ago (5).

The control of tumor angiogenesis is an integral part of the host defense response to tumor growth. Loss of endogenous angiogenesis inhibitors, such as endostatin and thromobospondin-1, leads to increased tumor angiogenesis and accelerates tumor growth in transgenic mouse models (6). In contrast, when angiogenesis is impaired and the expression of endogenous angiogenesis inhibitors is increased, tumors may enter a period of prolonged dormancy (7). Dormant tumors may be found in autopsy samples from trauma victims and in cancer patients that relapse after being disease-free for months or even years (8), indicating that tumors can exist as microscopic lesions for long periods without any clinical manifestation of disease. This dormancy may reflect the inability of these in situ tumors to disrupt or circumvent host antiangiogenic defenses (9).

The inhibition of tumor growth by antiangiogenic drugs has been achieved in both preclinical studies and clinical trials, where promising antitumor responses have been reported for a variety of antiangiogenic agents (ref. 10; Table 1). Bevacizumab, an anti-VEGF antibody and the first U.S. Food and Drug Administration-approved antiangiogenesis drug, significantly increases overall survival or progression-free survival of patients with metastatic colorectal cancer, non-small cell lung cancer, and breast cancer when given in combination with conventional chemotherapeutic regimens (refs. 11–13; Table 2). Renal cell carcinoma is a highly vascularized tumor that is associated with inactivation of the von Hippel-Lindau tumor suppressor gene and up-regulation of VEGF expression (14). Sunitinib, an antiangiogenic drug that inhibits the VEGF receptor (VEGFR) tyrosine kinase, shows superior activity in patients with advanced renal cell carcinoma when compared with the standard-of-care IFN-α treatment (15). Sunitinib is a multi-receptor tyrosine kinase inhibitor (RTKI); it also provides significant clinical benefit for patients with advanced gastrointestinal stromal tumors, which relates, at least in part, to its c-KIT-inhibitory activity (16). Sorafenib, an antiangiogenic RTKI that also has Raf kinase-inhibitory activity, has been approved for the treatment of renal cell carcinoma and liver cancer (17, 18). Many other antiangiogenic drugs are progressing through preclinical and clinical development, with >800 clinical trials presently under way.¹

Overall, however, the survival benefits of antiangiogenic drugs have thus far been rather modest, leading to increased interest in developing more effective ways to combine antiangiogenic drugs with traditional, cytotoxic chemotherapies. In this review, we discuss recent progress and some emerging challenges in the development of antiangiogenic drugs for cancer treatment. Interactions between these novel drugs and conventional chemotherapeutic agents are examined, and strategies for the optimization of combination therapies are discussed.

Many genes, proteins, and pathways have been identified as potential targets for antiangiogenic agents, of which the VEGF/VEGFR signaling pathway has been studied most extensively. The binding of VEGF to VEGFR-2 is a critical step that stimulates the major proangiogenic activities of VEGF, including endothelial cell proliferation, migration, tube formation, and capillary sprouting (19). More acute responses to VEGF include increased microvascular permeability and vasodilation (20). VEGF and VEGFR play an essential role in vascular development as exemplified by the embryonic lethality of targeted disruptions in the genes coding for VEGF and VEGFR-2 (21). Other VEGF and VEGFR family members are important for lymphangiogenesis (22). VEGF is often overexpressed by tumors, where the vasculature shows greater sensitivity to inhibition of VEGF signaling than in normal tissues (23). Major proangiogenic VEGF isoforms (splicing variants) include VEGF121, VEGF165, and VEGF189, which bind with similar affinity to VEGFR but differ in their angiogenic activities (24, 25); VEGF165b is an endogenous antiangiogenic isoform whose down-regulation in cancer is a marker of poor prognosis and metastatic potential (26, 27).

Several strategies have been employed to block VEGF/VEGFR signaling. In one approach, antibodies to VEGF or its receptor inhibit angiogenesis by blocking VEGF–VEGFR binding at the cell surface. Examples include bevacizumab, a humanized anti-VEGF monoclonal antibody (28); IMC-1121B, a human anti-VEGFR-2 monoclonal antibody; the anti-VEGF F(ab) fragment ranibizumab; and the VEGF RNA aptamer pegaptanib (Table 1). In a second approach, the tyrosine kinase activity that is intrinsic to activated VEGFR and related proangiogenic receptors is inhibited by small-molecule RTKIs, such as sunitinib, sorafenib, vatalanib, and axitinib (Table 1). Because of the structural similarities between VEGFR and other receptor tyrosine kinases, antiangiogenic RTKIs often inhibit multiple tyrosine kinases (29). For example, sorafenib inhibits the tyrosine kinase activities of VEGFR-1 and VEGFR-2, as well as those of platelet-derived growth factor (PDGF)-B receptor, Raf, Flt-3, and c-KIT, albeit with different affinities (30). Whereas the VEGFRs and PDGF-B receptor are important for angiogenesis, Raf kinase is central to the Raf/MEK/ERK signaling pathway, which is constitutively activated in several human tumors, including renal cell carcinoma, hepatocellular carcinoma, and non-small cell lung cancer (31). Flt-3, a receptor tyrosine kinase that regulates hematopoiesis, is highly expressed in acute leukemia and may be a therapeutically useful target (32). c-KIT inhibition is important for the treatment of gastrointestinal stromal tumors (33). Thus, the multiple receptor tyrosine kinase targeting activity of sorafenib provides an opportunity to interdict tumor growth by several independent mechanisms.

The tumor vasculature is frequently characterized by loss of intimate contact between pericytes and endothelial cells (34). This morphologic defect may contribute to the selective vulnerability of tumor blood vessels to VEGF inhibition (35). A similar deficiency in pericyte-endothelial cell association is observed in mice with a disruption of the gene coding for PDGF-B, an endothelial cell-derived proangiogenic factor required for the recruitment of pericytes to immature blood vessels (36), suggesting that the PDGF-B signaling pathway may be a target for antiangiogenesis (37). However, RIP-Tag2 pancreatic tumors, which are typically nonmetastatic, develop distant metastases when grown in Pdgfb-deficient mice (38), raising the possibility that PDGF-B pathway-selective inhibitors might actually enhance tumor metastasis. Coinhibition of VEGF and PDGF-B signaling, however, decreases leakiness and regresses tumor blood vessels (39), which reduces the access of metastatic cells to the circulatory system and limits the potential of PDGF-B inhibition to elicit a prometastatic response. Further studies are required to investigate these potential differences between PDGF-B inhibition and VEGF/PDGF-B coinhibition and their effect on tumor metastasis.

VEGF induces endothelial cell expression of delta-like ligand 4 (Dll4), a Notch receptor ligand that activates a negative feedback mechanism to restrain the sprouting and branching of new blood vessels (40). Soluble Dll4-IgG fusion protein blocks Dll4 activity and increases tumor vascularity. Nevertheless, this treatment inhibits tumor growth, as a majority of the newly formed tumor vessels are not perfused with blood. Thus, tumor blood flow can be inhibited when excessive angiogenesis leads to the formation of dysfunctional blood vessels.

Several established drugs have been found to have antiangiogenic activity. These include therapeutic agents that are ligands of the nuclear receptors peroxisome proliferator-activated receptor α (fibrate hypolipidemic drugs) and peroxisome proliferator-activated receptor γ (thiazolidinedione antidiabetics), both of which are expressed in endothelial cells. Ligands of peroxisome proliferator-activated receptor γ induce antitumor responses associated with a decrease in tumor microvessel density and VEGFR expression and an increase in endothelial cell expression of CD36, a receptor for the endogenous angiogenesis inhibitor thrombospondin-1 (41). Ligands of peroxisome proliferator-activated receptor α inhibit endothelial cell proliferation and tumor angiogenesis by down-regulating the expression of cytochrome P450 2C family epoxygenases, which convert arachidonic acid to epoxyeicosatrienoic acids, which are potent angiogenic lipids (42). In addition, drugs directed against oncogenes can induce antiangiogenic responses by down-regulating proangiogenic factors or by directly targeting the altered expression of proto-oncogenes in tumor-associated endothelial cells (43, 44). This crosstalk between angiogenic and oncogenic signaling pathways provides a rationale for combining angiogenesis inhibitors with other targeted anticancer agents (45, 46). Additional therapeutic targets and strategies for antiangiogenesis will likely be identified as our understanding of the cellular and molecular basis for angiogenesis increases.

A large number of endogenous antiangiogenic factors have been identified, including angiostatin, endostatin, and thrombospondin-1 (47). Angiostatin is a 38-kDa fragment of plasminogen that can be secreted by a primary tumor and suppresses the growth of metastases in experimental animal models (48). Endostatin, a 20-kDa fragment of collagen XVIII and the most extensively studied endogenous angiogenesis inhibitor, regulates a variety of proangiogenic and antiangiogenic factors (49) and a large downstream signaling network (50). Individuals with Down syndrome have elevated levels of circulating endostatin along with the increased gene dosage on chromosome 21, and this overexpression is associated with a very low incidence of solid tumors (51).

Thrombospondin-1 is a disulfide-linked homotrimeric adhesive glycoprotein that mediates cell-cell and cell-matrix interactions (52). It is a major component of platelet α-granules, and its release from activated platelets significantly increases plasma thrombospondin-1 levels (53). Metronomic chemotherapy (see below) substantially increases plasma thrombospondin-1 levels in tumor-bearing mice (54). Thrombospondin-1 protein can be deposited in the extracellular matrix by endothelial cells, smooth muscle cells, fibroblasts, macrophages, monocytes, and some tumor cells (55). The antiangiogenic activity of thrombospondin-1 is at least in part mediated by its endothelial cell surface receptor protein, CD36, and involves the induction of Fas ligand in proliferating endothelial cells. The latter cells undergo apoptosis when the corresponding cell surface receptor, Fas protein, is up-regulated by proangiogenic factors, such as VEGF, basic fibroblast growth factor, and interleukin-8 (56). New tumor blood vessels, which are continuously formed via angiogenesis, have elevated levels of Fas protein compared with resting blood vessels, which helps to explain the tumor blood vessel selectivity of thrombospondin-1. Thrombospondin-1 also inhibits the mobilization of VEGF from its extracellular matrix reservoir, suppresses endothelial cell migration, and decreases blood flow by blocking nitric oxide-induced relaxation of vascular smooth muscle cells. A peptide mimetic of thrombospondin-1, ABT-510, is currently in clinical development as an angiogenesis inhibitor (57). A related endogenous angiogenesis inhibitor, thrombospondin-2, also binds to CD36 and suppresses tumor growth via an antiangiogenic mechanism (52).

Tumor metastasis is an important but poorly understood target of antiangiogenesis therapy. The growth of tumor metastases, like that of the primary tumor, requires angiogenesis and can be targeted at multiple steps (ref. 9; Fig. 1). Tumors shed millions of cells into blood and lymphatic circulation in a process that requires penetration through a multilayer barrier composed of pericytes, base membrane, and endothelial cells. Antiangiogenesis prunes immature blood vessels and reduces vascular permeability, which, in turn, may limit the shedding of metastatic cells from the primary tumor. In patients with colorectal cancer, the number of intravasated tumor cells is positively correlated with tumor vascularity (58), suggesting that a decrease in tumor blood vessel density may translate into decreased access of tumor cells to the general circulation. Tumor cell intravasation is inhibited by low concentrations of endostatin in a chicken chorioallantoic membrane intravasation assay (59). Human tumor xenografts induce the remodeling of zebrafish vasculature and open holes in blood vessels through which tumor cells intravasate, and this process can be blocked by the antiangiogenic RTKI semaxanib (60).

The “seed and soil” hypothesis posits that tumor cells that have entered the circulatory system need to extravasate and colonize at predetermined locations (61). Bone marrow-derived, VEGFR-1-positive progenitor cells are a critical factor in the assembly of the premetastatic niche, where VEGFR-2-positive endothelial progenitor cells (EPC) may also be involved (62). The inhibition of VEGFR-1 and VEGFR-2 by the antiangiogenic RTKI sunitinib limits the attachment of tumor cells to these premetastatic sites (62). In sentinel lymph nodes of tumor-bearing mice, vascular reorganization and enrichment of blood vessels also occur before the arrival of metastatic cells (63). As noted above, micrometastases can remain dormant at secondary sites over a long period. Angiogenesis inhibition is proposed as the underlying mechanism that blocks the progression of these avascular micrometastases to macrometastases (64). Indeed, angiostatin was first discovered because of its antimetastatic activity (48). EPCs are essential in providing proangiogenic factors critical for metastasis development as indicated by the suppression of micrometastasis growth upon inhibition of EPC mobilization in preclinical studies (65). Moreover, selective inhibition of the vascular remodeling genes matrix metalloproteinase-1 and -2, using small interfering RNA or small-molecule inhibitors, significantly reduces the metastatic potential of tumor cells (66). These antimetastatic effects of antiangiogenic agents need to be confirmed in human patients with the incorporation of metastasis monitoring into clinical studies.

Tumor hypoxia not only induces angiogenesis but also is an important regulator of tumor cell mobility (4). Deletion of the gene encoding hypoxia-inducible factor-1α significantly reduces the metastatic potential of tumor cells in a transgenic mouse model (67). Counterintuitively, the increase in tumor hypoxia and hypoxia-inducible factor-1α expression that accompanies sustained angiogenesis inhibition may promote tumor cell invasion and migration as seen in some clinical studies (68, 69). This finding suggests it may be beneficial to coinhibit hypoxia-inducible factor-1α signaling in combination with antiangiogenesis. Considering the dominant role of metastasis in cancer mortality, long-term administration of antiangiogenesis drugs that are effective in maintaining an antimetastatic state could provide important clinical benefit.

Despite their promising activity in patients with a variety of cancers, current antiangiogenic treatments have provided only a modest survival benefit. Thus, there is increasing interest in combining antiangiogenic drugs with existing therapeutic modalities. As antiangiogenics are generally cytostatic rather than cytoreductive, combinations involving conventional cytotoxic chemotherapies may be useful for maximizing therapeutic activity. Of note, many antiangiogenic drugs can be administered over extended periods safely and with manageable toxicity compared with standard maximum tolerated dose (MTD) chemotherapies, which are often accompanied by severe adverse effects.

The low vascularity, poor organization, and abnormal morphology of the tumor vasculature leads to inefficient transport of oxygen and therapeutic drugs into tumors (70). This problem is exacerbated by the high interstitial fluid pressure within tumors, which is largely a consequence of the increased deposition of fibrin and other plasma proteins in the tumor stroma in response to the increased microvascular permeability induced by VEGF, coupled with the absence of efficient lymphatic drainage (19). The high interstitial fluid pressure not only promotes the dissemination of tumor cells to the peritumoral space but also may limit the delivery of chemotherapeutics into the tumor. Following antiangiogenesis treatment, the tumor vasculature may undergo morphologic normalization, whereby immature blood vessels are pruned, blood vessel tortuosity and dilation decrease, and a closer association between pericytes and endothelial cells is induced (71). Tumor blood vessel leakage, vascular permeability, and interstitial fluid pressure all decrease, which alleviates edema in cancer patients and provides an important clinical benefit (72–75). Many angiogenesis inhibitors induce these morphologic and permeability changes, suggesting that they represent a general response to the inhibition of VEGF signaling.

Counterintuitively, in preclinical studies of bevacizumab and certain other angiogenesis inhibitors, an increase in vascular patency has been observed, with an increase in tumor blood perfusion and drug uptake and a decrease in tumor hypoxia (refs. 76–80; Supplementary Table S1).²

^,3 These improvements in overall tumor vascular function indicate that blood vessels that survive antiangiogenic drug treatment have increased transport capability, which more than compensates for the decrease in the total number of patent blood vessels (71). This increase in vascular patency is transient, however, corresponding to a “window of opportunity” during which antiangiogenics may be combined with classic chemotherapeutics to increase overall tumor cell exposure to cytotoxic drugs. In this scenario, the cytotoxic drug is the major determinant of the overall therapeutic response. The optimal dosing and scheduling of the antiangiogenic agent becomes critical, as excessive suppression of the tumor vasculature may prematurely close the normalization window. Indeed, improved clinical responses are observed when conventional chemotherapy is combined with low-dose bevacizumab compared with high-dose bevacizumab (ref. 81; Table 2). In another example, in preclinical studies of sunitinib, interstitial fluid concentrations of the cancer chemotherapeutic drug temozolomide were increased when tumors were pretreated with sunitinib at 10 mg/kg but not at 40 mg/kg (79). To optimize the benefit of vascular normalization-enhanced tumor drug delivery, the duration of the open window during antiangiogenesis treatment needs to be better defined, for example, using noninvasive imaging techniques that monitor tumor blood flow (82). Chronic treatment with neutralizing antibodies to VEGF and VEGFR eventually reduces tumor blood perfusion and increases tumor hypoxia in experimental animal studies (83, 84), suggesting that uninterrupted treatment with the antiangiogenic drug, although perhaps maximally effective as a monotherapy, may not be optimal for tumor vascular normalization-enhanced combination chemotherapy (Fig. 2A). Intermittent antiangiogenesis treatment schedules need to be investigated to ascertain whether, and under which conditions, repeated cycles of vascular normalization and increased drug uptake might be achieved (Fig. 2B). The potential of such cycles of renormalization of tumor vasculature to facilitate tumor cell recovery from cytotoxic drug treatment during the chemotherapeutic drug-free period would need to be carefully considered. Furthermore, studies are needed to verify that the functional normalization and increase in drug uptake seen in preclinical studies also occurs in the clinic and contributes to the enhanced responses seen in patients treated with chemotherapy in combination with bevacizumab.

In contrast to bevacizumab, the RTKIs sorafenib and sunitinib show significant antitumor activity as monotherapies in phase III clinical trials (15–18). However, combinations of sorafenib or vatalanib with conventional chemotherapy do not increase patient survival (refs. 85, 86; Table 2). This raises the question of whether these RTKIs differ from the anti-VEGF antibody bevacizumab in their antiangiogenic actions and/or in their ability to facilitate cytotoxic drug delivery. Similar to bevacizumab, RTKIs can induce morphologic normalization of the tumor vasculature in both preclinical and clinical studies (75, 87, 88), and in some cases, the functionality of individual surviving blood vessels is improved (73, 89, 90). However, whereas anti-VEGF and anti-VEGFR antibodies can induce the transient increase in tumor drug uptake and oxygenation discussed above, small-molecule antiangiogenic RTKIs frequently elicit the opposite effects, that is, decreased drug uptake and increased tumor hypoxia, as seen in several preclinical studies (refs. 91–95; Supplementary Table S1). Conceivably, the lack of an increase in vascular patency of the RTKI-treated tumors could result from the use of RTKI drug doses and schedules that are designed to maximize antiangiogenesis but are suboptimal with respect to vascular normalization. Alternatively, intrinsic differences between the actions of VEGF-neutralizing antibodies and RTKIs could be a critical factor. Of note, VEGFR-directed RTKIs simultaneously inhibit paracrine and autocrine VEGF signals (96) and, in some cases, could also target PDGF-B receptor tyrosine kinase activity, whereas anti-VEGF and anti-VEGFR-2 antibodies selectively block paracrine VEGF signaling. Rapid decreases in tumor blood perfusion and tumor drug uptake can occur within hours of a single dose of antiangiogenic RTKI treatment, that is, well before the inhibition of VEGF/VEGFR signaling leads to endothelial cell killing and the resultant morphologic changes in tumor blood vessels (92, 94, 97). This suggests that the initial response to antiangiogenic RTKI treatment involves vasoconstriction of existing tumor blood vessels, perhaps due to decreased production of nitric oxide and prostacyclins, which mediate the acute vasodilatory effects of VEGF. Studies are needed to clarify whether the distinct VEGF-inhibitory actions of VEGF/VEGFR-neutralizing antibodies and antiangiogenic RTKIs lead to different effects on endothelial cell nitric oxide synthesis and nitric oxide release (98) and if the coinhibition of PDGF-B receptor by some RTKIs alters the response of smooth muscle cells to a nitric oxide signal. In addition, although decreased vascular permeability following VEGF/VEGFR inhibition reduces tumor interstitial fluid pressure, it is unclear how this affects the extravasation of drugs into the tumor, particularly macromolecular therapeutics.

The level of tumor oxygenation and drug uptake is primarily determined by the total number of functional blood vessels connecting to and within the tumor and by the transport efficiency of each individual blood vessel. As a result, tumor hypoxia and decreased drug uptake are both likely to result when strong antiangiogenic RTKIs induce rapid vasoconstriction and an extensive loss of tumor blood vessels. Nevertheless, the overall therapeutic response may be improved by combining RTKIs with conventional chemoradiotherapies by using appropriately designed treatment schedules. The angiogenesis inhibitor axitinib increases tumor hypoxia without functional normalization of the tumor vasculature, which would be expected to reduce tumor cell sensitivity to radiation therapy. However, antitumor activity is enhanced in tumor-bearing mice when axitinib is combined with radiation treatment (91). Moreover, an improved antitumor response can be achieved by combining axitinib with the cancer chemotherapeutic prodrug cyclophosphamide despite a substantial decrease in tumor uptake of the active chemotherapeutic drug (97). Enhanced antitumor activity is also observed in combination therapies with the antiangiogenic RTKI vandetanib under conditions where tumor blood flow and drug uptake are both decreased (95). Clearly, in these cases, where the RTKI elicits a strong antiangiogenic response, tumor cell exposure to the cytotoxic agent is not the sole determinant of overall antitumor activity. Rather, the antiangiogenic RTKI may induce a direct antitumor response through angiogenesis inhibition-induced tumor cell starvation, which is independent of, but complementary to, the cytotoxic response to chemotherapy/radiotherapy.

The net outcome of a combination therapy is likely to be determined by the balance between the tumor starvation effect of the RTKI and the decrease in tumor cytotoxicity due to the decrease in chemotherapeutic drug exposure (Fig. 3). When antiangiogenesis dominates a tumor's response to a combination therapy (Fig. 3A), optimal antitumor activity can be expected when a strong antiangiogenic agent is used to maximize angiogenesis inhibition. In contrast, when the cytotoxic drug is the predominant therapeutic factor, antiangiogenesis may compromise chemotherapeutic drug uptake and thus decrease antitumor activity (Fig. 3B). In this case, the combination therapy needs to be designed carefully with respect to (a) the choice of antiangiogenic drug (e.g., anti-VEGF antibody versus VEGFR-selective RTKI versus multi-targeted RTKI) and its dose and (b) the schedule of drug administration, which may involve antiangiogenic drug treatment on either a continuous or an intermittent schedule (Fig. 2). Optimization of drug sequencing will also be required, with the choice between neoadjuvant and adjuvant antiangiogenesis treatment likely to vary with the tumor (97) and with the chemotherapeutic drug. As the functional normalization window induced by VEGF or VEGFR-neutralizing antibody ultimately closes with continued antiangiogenesis treatment (Fig. 2A), the balance between increased tumor starvation and decreased cytotoxic drug exposure may also become an important determinant of the effectiveness of combination therapies that use this class of antiangiogenic drugs.

In addition to the potentiation of chemotherapy by antiangiogenesis, discussed above, the activity of antiangiogenesis drugs can be enhanced by cytoreductive treatments. For example, both radiation treatment and chemotherapy can augment the sensitivity of tumor blood vessels to VEGF inhibition, which leads to increased growth delay of human tumor xenografts in combination therapy settings (99, 100). When administered using a MTD schedule, conventional chemotherapeutic drugs not only damage tumor cells but also kill proliferating cells in normal tissues, which mandates the introduction of a drug-free recovery period between treatment cycles. However, during this recovery period, both tumor cells and tumor-associated endothelial cells initiate damage repair, leading to tumor regrowth and, in some cases, the emergence of a population of drug-resistant tumor cells, an important cause of treatment failure (101). This tumor repair process is another potential target for antiangiogenesis. Bone marrow-derived EPCs have been detected in tumor blood vessels, although their precise contribution to tumor angiogenesis is uncertain (102, 103). In preclinical studies, antiangiogenesis decreases the number of circulating EPCs, which are mobilized by bolus administration of cytotoxic drugs or vascular disruption agents (104–106). Blocking EPC mobilization by pretreatment with the antiangiogenic agents DC101 or axitinib increases the antitumor activity of the combination treatment (106, 107). This association between circulating EPC inhibition and the efficacy of the combination treatment warrants further investigation in a clinical setting.

Another important consideration for the combination of antiangiogenesis treatment with conventional chemotherapy is the potential for overlapping toxicities. Hypertension is a frequently observed side effect of antiangiogenesis, which can result from the decrease in nitric oxide production that follows VEGF deprivation (108). Multi-receptor tyrosine kinase targeting could, however, lead to additional toxicities not seen with the VEGF-specific bevacizumab (109, 110), suggesting the utility of RTKIs that are more VEGFR-selective, such as axitinib (111, 112). Neutropenia has been observed in patients treated with the multi-RTKI sunitinib (113, 114), which could be mediated by its non-VEGFR-inhibitory effects. Moreover, an increase in chemotherapy-associated neutropenia is seen with the anti-VEGF antibody bevacizumab (115). Thus, for this and other antiangiogenic agents, the selection of drug and treatment schedule should be carefully considered when designing combination therapies that include myelosuppressive chemotherapeutic drugs.

Pharmacokinetic modulation designed to increase tumor uptake of chemotherapeutic drug has been investigated with several agents, including botulinum toxin, nicotinamide, and various vasodilatory drugs (116). The utility of vasodilators as chemosensitizing agents depends on the functional role of smooth muscle cells in the tumor vasculature. When smooth muscle cells surrounding tumor blood vessels regulate tumor blood flow resistance, transient relaxation of these cells may temporarily increase blood flow through the tumor (117). However, as normal tissue blood vessels are highly sensitive to vasodilation, treatment of tumor-bearing mice with systemic vasodilators, such as hydralzaine, decreases blood flow resistance in normal tissues to a greater extent than in tumors, leading to a reduction in tumor blood supply and an increase in tumor hypoxia (118). Therefore, methods to selectively dilate tumor blood vessels are required, such as local low-dose radiation or the use of endothelin receptor antagonists (116). As tumor vascular patency and drug uptake decrease following chronic antiangiogenesis treatment, intermittent administration of tumor-selective vasodilators before each cycle of chemotherapeutic drug treatment might be useful in transiently increasing tumor blood perfusion, thereby increasing tumor cell exposure to cytotoxic agents, while at the same time retaining the tumor cell starvation effect of continuous antiangiogenesis treatment. In addition, when the resistance to tumor blood flow is not regulated by smooth muscle cells, systemic delivery of a vasoconstrictor, rather than a vasodilator, may be used to indirectly increase blood flow to the tumor based on the premise that normal but not tumor blood vessels remain sensitive to vasoconstrictors (119). These treatments can increase blood pressure, which needs to be monitored carefully.

Metronomic chemotherapy refers to the administration of cytotoxic drugs at a lower dose but increased frequency without prolonged drug-free breaks compared with traditional MTD anticancer drug treatment schedules (120). Metronomic drug treatments have shown promising therapeutic activity using drug administration schedules that range from repeated administration every 6 to 7 days to daily or even continuous drug treatment (120). With several cancer chemotherapeutic drugs, including cyclophosphamide, docetaxel, and vinblastine, metronomic treatment schedules induce a significant antiangiogenic response. Preclinical studies of chemotherapy-resistant tumor models have shown that tumor cell apoptosis is preceded by increased death of tumor-associated endothelial cells, indicating that endothelial cells are a primary target of metronomic chemotherapy (121). This reflects the high intrinsic sensitivity of endothelial cells to certain cytotoxic drugs (122). Toxicities to normal tissues are absent or low-grade, as seen in both preclinical and clinical studies, indicating tumor specificity for the antiangiogenic actions of metronomic chemotherapy (123, 124). In several preclinical studies, expression of the endogenous angiogenesis inhibitor thrombospondin-1 increases significantly during metronomic cyclophosphamide treatment, and antitumor activity is substantially reduced in its absence (54, 92, 125, 126). Metronomic cyclophosphamide also reduces tumor-induced immune tolerance (127), and correspondingly, synergistic antitumor activity results when metronomic cyclophosphamide is combined with a tumor-targeted immunotherapy (128).

Several approaches have been investigated to integrate the cytoreductive activities of conventional MTD chemotherapy with the antiangiogenic effects of metronomic chemotherapy. Administration of cyclophosphamide at a MTD/bolus dose followed by cyclophosphamide treatment on a metronomic schedule gives superior antitumor activity compared with either schedule alone in experimental animal studies (129, 130). The antitumor activity of metronomic cyclophosphamide can also be enhanced by intratumoral gene transfer of a cyclophosphamide-activating liver cytochrome P450 enzyme (131, 132), where liver-derived cyclophosphamide metabolites dominate the antiangiogenic activity of the combination treatment, whereas drug-induced tumor cell death is substantially increased by intratumoral prodrug activation (126). Metronomic chemotherapy can also be combined with other antiangiogenic agents. In a phase II clinical trial, coadministration of metronomic cyclophosphamide and bevacizumab resulted in superior antitumor activity when compared with historic controls (133). The outcome of therapies that combine small-molecule antiangiogenic RTKIs with metronomic chemotherapy is more difficult to predict, as either increased or decreased antitumor activity can be achieved, with the latter response perhaps reflecting a decrease in tumor exposure to the cytotoxic drug and/or blocking of thrombospondin-1 induction in host cells (92, 129).

The toxicity-defined approach to drug dose selection, commonly used for cancer chemotherapeutic drugs, may not be suitable for antiangiogenics. For example, endostatin displays a U-shaped dose-response curve, with the conventional MTD dose being ineffective (134). In combination therapies where the efficacy of chemotherapy is enhanced by antiangiogenic drug-induced normalization of tumor blood vessels, excessive inhibition of tumor vasculature function may decrease penetration of the coadministered chemotherapeutic drug, as discussed above. Surrogate markers are thus particularly important for monitoring antiangiogenic activity and could help predict a given patient's response to drug treatment (135).

One marker for tumor angiogenesis, microvessel density, is quantified by counting the number of blood vessels in a tumor section after staining with antibodies to an endothelial cell-specific marker protein, such as CD31, CD34, CD105, CD146, or von Willebrand factor (136). Tumor microvessel density is an independent prognostic factor for a variety of human tumors, where high vascular density is often associated with poor prognosis following surgery or conventional chemoradiotherapy (137). Tumor vascular density reflects the balance between proangiogenic and antiangiogenic factors within the tumor microenvironment and is influenced by many factors including the availability of oxygen and nutrients (138). Highly vascularized tumors are generally considered to be more sensitive to a decrease in blood supply and are therefore expected to be highly responsive to antiangiogenic drugs, whereas hypovascularized tumors are viewed as more hypoxia-tolerant and therefore less sensitive to antiangiogenesis. However, poorly vascularized tumors can respond well to angiogenesis inhibition as seen in several preclinical studies (97, 139, 140). Conceivably, treatment of such poorly vascularized tumors with antiangiogenics may suppress an already low blood supply to the point where continued tumor growth becomes unsustainable. Further study is required to determine whether tumor microvessel density is a useful predictor of tumor response to antiangiogenesis treatment. Changes in tumor cell density that follow drug treatment may further complicate the interpretation of tumor microvessel density measurements (138).

Blood vessel size and vascular coverage area have both been used as surrogate markers for drug-induced antiangiogenesis. However, because tumor vasculature is often poorly perfused, more useful information about vascular function may be obtained from direct measurements of blood vessel patency. This can be achieved by i.v. injection of probe molecules, such as tomato lectin Lycopersicon esculentum and the fluorescent dye Hoechst 33342, which bind to the luminal surface of endothelial cells in perfused blood vessels (tomato lectin) and to tumor cells in close proximity to these blood vessels (Hoechst 33342), respectively (78, 87, 141). High molecular weight tracers, such as fluorescence-labeled dextran, albumin, antibodies, and microspheres, have also been used to detect and measure the leakiness of tumor blood vessels (38, 74, 89).

Tumor oxygenation reflects the balance between oxygen delivered to the tumor by the blood supply and its consumption in local metabolic activities and is an important variable for assessing the functionality of the tumor vasculature. Intratumoral oxygen levels can be measured using polarographic needle electrodes, electron paramagnetic resonance oximetry, and hypoxia-specific dyes, such as pimonidazole (142). However, caution should be applied when using hypoxia-specific dyes to monitor tumor hypoxia induced by antiangiogenesis, which can inhibit penetration of the dye itself (97). Interstitial fluid pressure, which contributes to the reduced penetration of drugs into solid tumors, can be monitored using specific needle probes (143) and may be an indicator of the effectiveness of antiangiogenesis treatments with respect to improving drug delivery (74). Quantification of intratumoral drug concentrations provides a more direct measure of the effect of antiangiogenesis treatments on tumor drug uptake (79, 92). For therapeutic agents with intrinsic autofluorescence (e.g., doxorubicin), intratumoral drug distribution can be visualized directly. Noninvasive imaging techniques can provide real-time monitoring of fluctuations in tumor blood volume and flow rate and the accumulation of tracer molecules following antiangiogenesis. Magnetic resonance imaging, X-ray computed tomography, and ultrasound imaging provide high spatial resolution, and positron emission tomography can be used to detect both tumor blood perfusion and glucose metabolism in both preclinical and clinical studies (82).

Several other surrogate markers for antiangiogenesis have been investigated. Pretreatment plasma levels of VEGF correlate with the survival benefit of bevacizumab treatment in patients with metastatic colorectal cancer (144). Increased VEGF and decreased soluble VEGFR-2 levels in plasma have been observed in patients treated with antiangiogenesis drugs (75, 145). However, changes in the plasma levels of these factors may not be very informative for the measurement of therapeutic responses in tumors, given the significant contributions that normal tissues make to such changes, as revealed by a recent preclinical study (146). Circulating endothelial cells and EPCs have been investigated as alternative blood markers for antiangiogenic activity. Decreases in viable EPC counts in blood occur following treatment with DC101, axitinib, or metronomic chemotherapy in experimental animal models, and these changes directly correlate with antitumor response (106, 107, 147). In rectal cancer patients, the number of viable circulating endothelial cells decreases following bevacizumab treatment (145). Furthermore, in breast cancer patients treated with metronomic chemotherapy, a significant correlation was observed between clinical benefits and an increased fraction of apoptotic circulating endothelial cells (148). Standardized surface markers are needed before the utility of these markers can be evaluated more widely (149).

Tumors may circumvent antiangiogenesis by multiple mechanisms, which include changes in both tumor cells and tumor-associated host stromal cells (150). Tumor-associated endothelial cells and pericytes, both of which can be primary targets of antiangiogenesis treatment, are classically viewed as genetically stable and unlikely to develop drug resistance (151). However, more recent studies have revealed that the expression profile of tumor-associated endothelial cells is distinct from normal endothelial cells (152) and can be tumor type-dependent (153), with many tumor endothelial cells being cytogenetically abnormal (154). Tumor-associated pericytes may also show abnormal morphology and alterations in marker protein expression compared with their normal tissues counterparts (34), although it is unclear how these tumor-associated genetic or epigenetic changes might affect the sensitivity of endothelial cells or pericytes to angiogenesis inhibitors.

Resistance to VEGF inhibition can be mediated by the increased expression of other angiogenic factors. High-level expression of VEGF, VEGF-B, VEGF-C, basic fibroblast growth factor, PDGF-A, transforming growth factor-α, and angiopoietin-2 has been observed in advanced human neuroblastomas (155). Continuous treatment of RIP-Tag2 tumors with antibodies to VEGFR-1 and VEGFR-2 initially leads to stable disease but is followed by the development of resistance to the VEGFR blockade with up-regulation of the angiogenic factors fibroblast growth factor, ephrin, and angiopoietin (156). Interestingly, endostatin suppresses the expression of basic fibroblast growth factor and ephrin-A1 (49), suggesting that the combination of this endogenous angiogenesis inhibitor with inhibitors that target VEGF signaling may suppress the development of drug resistance. Up-regulation of placental growth factor has been observed following VEGF deprivation using either antiangiogenic neutralizing antibodies or RTKIs (75, 145, 146), and the efficacy of anti-VEGFR-2 treatment can be improved by combination with anti-placental growth factor antibody in experimental animal studies (157). Enhanced antiangiogenic effects have also been observed by combining VEGFR inhibitors and PDGF receptor inhibitors (39, 129). In preclinical studies, crosstalk between basic fibroblast growth factor and PDGF-B can synergistically promote tumor angiogenesis and metastasis (158), which further underscores the importance of inhibiting multiple angiogenic pathways in cancer treatment.

In addition to the above changes in the complement of angiogenic factors in response to antiangiogenic drug treatment, the intrinsic tolerance of tumor cells to hypoxia or an acidic microenvironment, as well as hypoxia-induced tumor cell invasion and metastasis, may also increase, further limiting the efficacy of antiangiogenesis (159). Alternatively, in some highly vascularized organs, such as brain, liver, and lung, tumor cells may coopt existing normal blood vessels and grow around them (160–162). Not only does the growth of the tumor become angiogenesis independent, but also the coopted normal blood vessels will be less sensitive to the antivascular actions of antiangiogenesis drugs, which can lead to resistance to antiangiogenesis and can also obscure detection by techniques such as contrast-enhanced magnetic resonance imaging, which depends on the extravasation of contrast agents from leaky tumor blood vessels. Furthermore, as first observed in melanoma and subsequently reported in several other tumor types, tumor cells can form vessel-like structure by vasculogenic mimicry (163). The response of such tumor cell vascular networks to antiangiogenesis agents is unknown, and their effects on drug delivery in combination therapy require further investigation. Given the anticipation that antiangiogenic drugs entering into clinical practice will ultimately be given to patients long-term, it is important to understand the mechanisms of drug resistance that these drugs elicit and whether cross-resistance among antiangiogenic drugs can be anticipated.

Multiple therapeutic approaches have been developed to inhibit tumor angiogenesis, and better, more reliable surrogate markers are needed to assess their effectiveness in individual patients. The projected chronic administration of antiangiogenesis drugs calls for a better understanding of their antitumor and antivascular effects and the mechanisms that may lead to drug resistance. Morphologic normalization of the tumor vasculature is widely observed following angiogenesis inhibition. Preclinical studies indicate that functional improvement of tumor blood perfusion can be induced by some antiangiogenic agents, with the potential to increase tumor cell exposure to coadministered cytotoxic drugs. However, for other antiangiogenic drugs, tumor vascular patency decreases, leading to an increase in tumor hypoxia and a decrease in cytotoxic drug uptake. Nevertheless, antiangiogenic drugs can serve as strong and independent antitumor agents, more than compensating for the decrease in cytotoxic drug exposure by angiogenesis inhibition-induced tumor cell starvation, which may lead to an increase in overall antitumor activity. Antiangiogenics can interact in multiple ways with other anticancer drugs and treatment regimens, making it critical to carefully evaluate and optimize dosing and scheduling in the design of effective drug combinations.

No potential conflicts of interest were disclosed.