Antiangiogenesis: biology and utility in the treatment of gliomas

The growth of all cancers is dependent upon the formation of new tumor blood vessels, a process that includes several mechanisms, including angiogenesis [1–5]. Vasculogenesis is the creation of new blood vessels by recruiting circulating viable endothelial cells and endothelial progenitor cells. This process entails differentiation of endothelium from circulating endothelial stem cells [6–10]. Vasculogenesis is most pronounced during embryonic development;however, initial tumor neovascularization is thought to require circulating endothelial cells. Furthermore, circulating endothelial cells are increased in the circulation in patients with cancer (discussed later). Laboratory models demonstrate that tumor angiogenic vessels are dynamic with endothelial cell turnover. In addition, the tumor luminal vasculature appears to represent a mosaic of tumor and endothelial cells. By contrast, angiogenesis is the process by which new capillaries sprout from pre-existing blood vessels by endothelial cell proliferation and migration [1–5]. Angiogenesis is a physiologic process utilized, for example, in wound healing, menstruation and creation of the fetal–placental vascular network. Angiogenesis is felt to be the dominant process supporting cancer growth. Neovascularization by angiogenesis is a prerequisite for expansion of a tumor beyond 0.125 mm due to limits of oxygen and nutrient diffusion. Another process of tumor neovascularization is co-option [11]. Brain tumors co-opt existing host vessels to form an initially well-vascularized tumor. Subsequent regression of these co-opted vessels leads to secondary avascularization and massive tumor cell death, seen as necrosis, a neuropathologic characteristic of glioblastoma multiforme (GBM). Surrounding central necrosis are proliferating contrast­enhancing tumors that manifests robust angiogenesis.

Angiogenesis represents a balance between proangiogenic stimuli (acidic and basic FGF, PlGF, VEGF, IL-8, HGF/scatter factor [SF], granulocyte colony-stimulating factor and angiopoietin) and angiogenic inhibitors (TS1, angiostatin, IFN-α, metalloproteinase inhibitors, TGF-β and endostatin) [1–5]. In cancer, proangiogenic stimuli dominate and, in particular, VEGF expression is upregulated, resulting in activation of the ‘angiogenic switch’ [12,13]. VEGF is a multifunctional cytokine with six isoforms (VEGF-A, -B, -C, -D and -E, and PlGF) that promotes formation of new blood vessels, renders vessels hyper­permeable, stimulates endothelial cells to divide and migrate, protects endothelial cells from apoptosis and senescence and alters endothelial cell gene expression. VEGF-A is one of the most potent vascular permeability agents known. It affects postcapillary venules and small veins and promotes extravasation pathways by way of vesiculovacuolar organelles (trans­endothelial transcellular process), loosening endothelial fenestrations (transcellular process) and opening interendothelial cell pathways (opening of tight junctions; intercellular process). VEGF is essential for normal development (both vasculogenesis and angiogenesis) and, as stated above, increases vascular permeability to plasma and plasma proteins, is selective for vascular endothelium (as it’s cognate receptors, VEGF receptors [VEGFRs], are expressed only on endothelium) and is overexpressed in most cancers. VEGF-A expression is regulated by a number of factors, including the proangiogenic cytokines mentioned earlier and tissue oxygenation, wherein tissue hypoxia upregulates VEGF-A by way of the transcription factors, hypoxia-inducible factor (HIF)-α and -β. Dexamethasone and activation of p53 or vHL (a gene product that targets HIF-α for ubiquintinylation and proteasome degradation) tumor-suppressor genes downregulate VEGF-A expression.

VEGF-A (ligand) binds with two high-affinity transmembrane tyrosine kinase receptors, VEGFR-1 and VEGFR-2, the latter mediating permeability and proliferation. Binding of VEFG-A to VEGFR-2 results in autophosphorylation of both receptor tyrosine kinases followed by activation of downstream signaling pathways, including PI3-kinase, RAS and MAPK.

Glioma tumor blood vessels are abnormal and manifest numerous perturbations, including defective vascular function, thereby impairing delivery of oxygen and chemotherapeutics, defective endothelium and basement membrane (limited coverage), impaired pericyte coverage and increased vascular permeability [14–17]. Brain tumor blood vessels are tortuous, have a large diameter, are disorganized, have increased and enlarged endothelial pores, have loss of pericyte coverage, a thicker basement membrane and glomeruloid tufts. As a consequence, tumor vascular permeability is variably increased spatially and temporally resulting in peritumoral vasogenic edema, demonstrates nonuniform blood flow, heterogeneous delivery of oxygen and blood-borne drugs and increased interstitial fluid pressure due to extracellular edema [18–23]. As a consequence, tumor vasculature is disorganized, unevenly distributed, twisted and leaky, shows equalization of intraluminal and interstitial pressure, is dependent upon cell survival factors and shows focal hypoxia.

The rationale for antiangiogenic therapy in the treatment of cancer is multidimensional [13,18,19,24–26]. Endothelial cells are easily accessible to antiangiogenic therapy relative to tumor owing to lack of physical barriers (i.e., the blood–brain barrier), extracellular matrix and elevated tumor interstitial pressure. Endothelial cells are genetically stable and, therefore, less likely to develop resistance to therapy. Since angiogenesis is absent in adults except during menstruation, pregnancy and wound healing, antiangiogenic therapy permits therapeutic selectivity. Nonetheless, antiangiogenic therapy appears to be cytostatic and, consequently, requires long-term continuous treatment. However, it is problematic that the leading edge of infiltrating glioma cells are not dependent upon angiogenesis and, therefore, may be unaffected by antiangiogenic therapy [27–32].

Why use antiangiogenic therapies in high-grade gliomas (HGG)? Evidence for likely benefit is derived from several observations [33–42]. Multiple experimental studies use GBM as a tumor model of angiogenesis. Additionally, GBM cells have been shown to produce high levels of VEGF. Within GBM, VEGF levels are highest in areas of necrosis and regions of endothelial proliferation. GBM pathologically is characterized by endothelial proliferation (i.e., tumor angiogenesis that results in blood–brain barrier disruption and contrast enhancement). Finally, in patients with HGG, high circulating levels of endothelial progenitor cells (CD34⁺, CD133⁺ and VEGFR2⁺) are seen.

Several mechanisms of action have been suggested for the antiglioma effect of antiangiogenic therapies, including a direct antiangiogenic effect on tumor vasculature, direct antiglial effect on VEGFR-expressing tumor cells, forced vascular normalization, thereby improving drug delivery, and a direct anticancer stem cell effect [33–42]. Antiangiogenic therapy, by improving vascular function (i.e., vascular normalization), improves tumor oxygenation, which correlates with increased response to radiation and chemotherapy. Antiangiogenic therapy also reduces tumor vessel permeability by endothelial maturation, thereby decreasing tumor interstitial pressure and improving drug delivery. Antiangiogenic therapy may disrupt the cancer stem cell vascular microniche and, thereby, injure glioma stem cells. Lastly, antiangiogenic therapy may also sensitize endothelial cells to cytotoxic therapies (radiotherapy and chemotherapy).

Although multiple antiangiogenic strategies are being explored, with respect to antiglioma-based therapy, only two have entered clinical practice: ligand-based antagonist therapy utilizing monoclonal antibodies, such as bevacizumab, and receptor-based antagonist therapy with tyrosine kinase inhibitors, such as AZD2171 (cediranib) (Table 1). The use of bevacizumab has been predominantly with cytotoxic chemotherapy (irinotecan [CPT-11] or carboplatin) based upon two concepts; interruption of VEGF signaling leads to sensitization or reversal of cytotoxic drug resistance and improvement in cytotoxic drug vascular access through vascular normalization and decrease in tumor interstitial pressure. This combinatorial rationale is compelling for non-neural cancers, for example colorectal, non-small-cell lung and breast cancer, although it has not been convincingly demonstrated for HGG.

The available data for antiangiogenic therapy in HGG are most robust for ligand-based VEGF antagonism and are based on several single institution studies, many of which are retrospective (Table 1)[43–49]. With one exception, these studies have utilized combinatorial therapy and predominantly treated GBM (total of 344 patients treated, among whom 45 [13%] were anaplastic glioma [AG]). Response rates vary with complete response to bevacizumab plus therapy in 0–5%, partial response in 21–59% (median: 38%) and stable disease in 21–59% (median: 31%) of patients. Progression-free survival also varies (median: 23.9 weeks; 6 months: 39%; and 12 months: 20%). Median overall survival was 8.7 months (range: 8.2–9.5) with bevacizumab plus therapy. These results with bevacizumab can be compared with temozolomide for recurrent temozolomide-naive GBM (progression-free survival at 6 months: 21%) and an aggregate series of eight Phase II trials for recurrent HGG (pretemozolomide) from the MD Anderson Cancer Center (GBM progression-free survival at 6 months: 15%) (Table 1). Bevacizumab therapy therefore appears to increase response of recurrent GBM by sevenfold, progression-free survival by twofold and overall survival by 2.5–3.0 months. Based on these data and comparisons, bevacizumab plus therapy has become the current treatment of choice for recurrent GBM. In a provocative Phase II study comparing bevacizumab with or without CPT-11 (the only randomized trial in HGG), results suggest similar outcomes between treatment arms, raising the question as to what added value is provided by CPT-11 [49].

Reassuringly, toxicity of bevacizumab plus therapy in these various studies have been modest and comprised primarily of low-grade bleeding (epistaxis, vaginal or oral cavity), proteinuria, impaired wound healing and hypertension [50,51]. Rare serious side effects include gastrointestinal perforation, intratumoral hemorrhage and craniotomy or central venous line wound dehiscence (each less than 1% incidence). The high incidence of deep vein thrombosis and pulmonary embolism in patients with HGG confounds separation, as an independent toxicity of antiangiogenic therapies may aggravate this thrombogenic predisposition [52–54]. A retrospective review of bevacizumab plus therapy in patients with recurrent HGG suggests that concurrent use of anticoagulation appears safe without an apparent increased risk of hemorrhage [47].

The only published trial of VEGF receptor antagonists is that of the oral pan-VEGFR tyrosine kinase inhibitor AZD2171 [49]. In this elegant trial of 16 patients with recurrent GBM, several correlative studies were included: tumor assays of VEGFR, MRI determination of vascular permeability, vessel size and vasogenic edema and circulating biomarkers of GBM angiogenesis (VEGF, PlGF, βFGF, HGF/SF and quantification of circulating endothelial cells [CECs]). The study showed normalization of tumor vessels in recurrent GBM by MRI with a rapid onset (as early as 1 day after AZD2171 administration) that was, however, reversible and of comparatively short duration (in most patients by 56 days after AZD2171 administration). Peritumoral edema improvement with concomitant steroid dose lowering was another significant trial observation. Consistent with other reports using antiangiogenic therapies, biomarker analysis revealed increases in VEGF and PlGF and decreases of soluble VEGFR on AZD2171 therapy. Re-recurrence of GBM following failure with AZD2171 was associated with decreased levels of VEGF and PlGF, increased HGF/SF and βFGF, and viable CEC levels.

At present, it is unclear if combination therapy (bevacizumab and CPT-11) is more effective than single-agent bevacizumab therapy. Furthermore there are no trials in HGG that define the best combinatorial therapy that might include either another targeted therapy (e.g., EGF inhibition) or an alternative cytotoxic chemotherapy. Further studies are needed to establish the role for antiangiogenic therapy both as first-line treatment and at recurrence in patients with HGG. Importantly, can risk factors be identified to minimize toxicity (craniotomy dehiscence, intratumoral hemorrhage, hypertension and thrombosis)? Is it possible to identify patients likely to respond to therapy, both before and after initial treatment by using tumor biomarker expression patterns, ¹⁸fluorothymidine PET (Flt-PET) response, ¹⁸fluoromisonidazole (FMISO-PET) hypoxic tumor volume or the ratio of FLAIR volume to the contrast-enhancing tumor volume? [46,47]. Lastly, it is uncertain how best to determine radiographic response to therapy as traditional assessment of contrast-enhancing tumor volume is obviated by antiangiogenic therapy. Future development of targeted therapy, including antiangiogenic therapies, will require identification and validation of appropriate biomarkers for improved patient selection, prognostication and response end points.

Table 1. Bevacizumab and irinotecan (CPT-11) or carboplatin for recurrent high-grade gliomas.

Author	Number treated	Toxicity (%)	Radiographic response (%)				Progression-free survival (%)			Survival (median, months)	Ref.
			Complete	Partial	Stable		Median (weeks)	6 months	12 months
Stark-Vance	21 (11 GBM)	10 Gr 3	5	38	52		NS	NS	NS	NS	[55]
Pope et al.	14 (10 GBM)	7 Gr 3	0	50	21		NS	NS	NS	NS	[43]
Vredenburgh et al.	32 (23 GBM)	12 Gr 3	3	59	34		NS	38 (30% GBM)	NS	9.0	[44]
Vredenburgh et al.	35	26	57	57	24		24.0	46	20	9.5	[45]
Chen et al.	21 (18 GBM)		0	38	29		11.0	10	0%	8.6	[46]
Norden et al.	54 (33 GBM)	41	34	34	59		23.9	42	NS	NS	[47]
Cloughsey et al.	85 (-CPT-11)	45 Gr 3	21	21				36	19	8.2	[48]
	82 (+CPT-11)	63 Gr 3	34	34				51	22	8.7

CPT-11: Irinotecan; GBM: Glioblastoma multiforme; Gr: Grade; NS: Not stated.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.