Glioblastoma angiogenesis: VEGF resistance solutions and new strategies based on molecular mechanisms of tumor vessel formation

Abstract

Glioblastomas are highly vascular tumors. Recent preclinical and clinical investigations have revealed that agents targeting angiogenesis may have efficacy against this type of tumor. Antibodies to vascular endothelial growth factor are being studied in this patient population. Unfortunately, treatment inevitably fails. This review provides an update on recent research on the mechanisms by which tumor cells acquire resistance, and discusses recent preclinical and experimental development of novel new-generation anti-angiogenic agents that overcome this problem, especially those based on the molecular mechanisms of tumor vessel formation. The tumor vasculature not only nourishes glioblastomas, but also provides a specialized microenvironment for tumor stem-like cells and for the brain tumor. The factors, pathways, and interactions described in this review provide information about the cell biology of glioblastomas which may ultimately result in new modes of treatment.

Full text

REVIEW ARTICLE
Glioblastoma angiogenesis: VEGF resistance solutions
and new strategies based on molecular mechanisms of tumor
vessel formation
Shingo Takano
Received: 7 December 2011 / Accepted: 9 December 2011 / Published online: 6 January 2012
Ó The Japan Society of Brain Tumor Pathology 2011
Abstract Glioblastomas are highly vascular tumors.
Recent preclinical and clinical investigations have revealed
that agents targeting angiogenesis may have efficacy
against this type of tumor. Antibodies to vascular endo-
thelial growth factor are being studied in this patient pop-
ulation. Unfortunately, treatment inevitably fails. This
review provides an update on recent research on the
mechanisms by which tumor cells acquire resistance, and
discusses recent preclinical and experimental development
of novel new-generation anti-angiogenic agents that over-
come this problem, especially those based on the molecular
mechanisms of tumor vessel formation. The tumor vascu-
lature not only nourishes glioblastomas, but also provides a
specialized microenvironment for tumor stem-like cells
and for the brain tumor. The factors, pathways, and inter-
actions described in this review provide information about
the cell biology of glioblastomas which may ultimately
result in new modes of treatment.
Keywords Resistance to VEGF blockade 
Normal vessel formation  Tumor vessel formation 
Vasculogenic mimicry  Cancer stem-like cell
transdifferentiation
Introduction
Glioblastomas are among the most angiogenic human tumors,
and endothelial proliferation is a hallmark of the disease [1].
A better understanding of glioblastoma vasculature is needed
to optimize anti-angiogenic therapy, which has high but
transient efficacy [2, 3]. Although it is known that neural stem
cells interact closely with the vascular microenvironment [4],
the origins of the endothelial cells in this malignancy are
poorly understood. Two papers [5, 6], following an initial
paper [7], show that stem-like cells in the tumor can differ-
entiate into endothelial cells, thereby generating the tumor
vasculature. Both Ricci-Vitiani et al. [5] and Wang et al. [6]
began by examining the angiogenic endothelium in glioblas-
toma tissue from patients, using a range of endothelial and
glioblastoma markers. Soda et al. [7] performed similar
studies with an experimental model. They found that some of
the endothelial cells had the same genomic alterations as the
tumor cells, indicating they were of neoplastic origin.
However, this phenomenon, ‘‘the tumor cell itself gen-
erates its own vasculature’’ is not new. It has been reported
there are several modes of vessel formation including
normal and tumor vasculature. Study of this phenomenon,
including normal vessel formation, is directly connected
with solving the problem of glioblastoma resistance to
vascular endothelial growth factor (VEGF) blockade. To
understanding the molecular mechanism of glioblastoma
angiogenesis and its clinical application, normal vessel
development, tumor vessel development, the mechanism of
resistance against VEGF blockade, two new anti-angio-
genesis targets (vasculogenic mimicry and transdifferenti-
ation), and future directions are reviewed.
Mechanisms of resistance against VEGF (receptor)
blockade and its solutions
A fraction of patients with cancer, including glioblastoma,
are refractory to VEGF-inhibitor treatment. For glioblas-
toma patients treated with bevacizumab, tumor recurrence
S. Takano (&)
Department of Neurosurgery, Institute of Clinical
Medicine, University of Tsukuba, 1-1-1 Tennoudai,
Tsukuba, Ibaraki, Japan
e-mail: shingo4@md.tsukuba.ac.jp
123
Brain Tumor Pathol (2012) 29:73–86
DOI 10.1007/s10014-011-0077-6

inevitably occurs. Furthermore, recurrent tumors after
bevacizumab failure are reported to be more aggressive,
with rebound edema. Potential mechanisms of resistance to
targeted VEGF therapy in cancer are closely related to the
mechanism of tumor vascular formation and to tumor
biology. Solution of all the mechanisms of resistance is one
promising method for future anti-angiogenic treatment
directed at the glioma microenvironment [8].
1. VEGF-independent vessel growth. Tumors produce
additional proangiogenic molecules besides VEGF,
before or after treatment with VEGF blockers [9, 10].
Elevated FGF-2 levels have been observed in glio-
blastoma patients treated with a VEGFR small-mole-
cule inhibitor [11]. In patients who experienced tumor
progression while receiving treatment, increased
tumor enhancement volume was associated with
significant increases in plasma levels of FGF-2 and
SDF-1. A statistically significant positive correlation
was observed between FGF2 levels and tumor vessel
size measured by MRI. This work provided further
clinical evidence that FGF-2 may be involved in tumor
relapse in patients treated with anti-VEGF agents [12].
In another study, high levels of platelet-derived growth
factor C (PDGF-C) and c-Met expression were
observed in cediranib-treated glioblastoma patients
[13].
2. Increase of invasiveness. In established tumors, VEGF
blockade aggravates hypoxia, which upregulates the
production of other angiogenic factors or increases
tumor cell invasiveness [14–18]. How might tumors
become more invasive during antiangiogenic therapy?
One possibility is that the tumor may increase the
activity of a preexisting invasion program that was not
previously the driving force of expansive tumor growth,
given the capability for angiogenesis. Alternatively,
glioblastomas may switch on an invasive growth
program distinct from that arising spontaneously during
unperturbed tumor development and progression, in
which antiangiogenic therapy induced a phenotypic
change from single-cell infiltration to migration of cell
clusters along normal blood vessels [18]. Tumor cells
that have acquired other mutations can also become
hypoxia-tolerant. There are examples of HIF-1-inde-
pendent tumor angiogenesis. Interestingly, reduced
tumor angiogenesis was achieved only by inhibiting
both NFkB-dependent IL-8 expression and K-ras-
Val12-dependent VEGF-A expression [19]. These
HIF-independent mechanisms also make significant
contributions to the regulation of angiogenesis.
3. Sprouting-independent vessel growth. Tumors have
available, and can switch to, modes of vessel growth
including vessel co-option, intussusception, and
vasculogenic mimicry, and differentiation of putative
cancer stem cells (CSCs) into endothelial cells (ECs)
may be less sensitive to VEGF blockade. (This is
discussed in more detail in a later section.)
4. Stromal cells—1: bone marrow-derived cells
(BMDCs). Recruited pro-angiogenic BMDCs, mac-
rophages, or activated cancer-associated fibroblasts
can rescue tumor vascularization by production of
pro-angiogenic factors [10]. An intracranial glioblas-
toma xenograft model has revealed that irradiation
induces recruitment of BMDCs into the tumor,
restoring the radiation-damaged vasculature via vas-
culogenesis, thereby enabling growth of surviving
tumor cells. BMDCc are recruited to tumors in part
via the interaction between the HIF-1-dependent
stromal cell-derived factor-1 (SDF-1) and its recep-
tor, CXCR4. For a novel approach to treatment of
GBM, in addition to radiotherapy, this vasculogen-
esis pathway must be blocked; this can be accom-
plished by use of the clinically approved drug
AMD3100, a small molecule inhibitor of the SDF-
1/CXCR4 interaction [21].
5. Stromal cells—2: macrophages. Infiltration of mac-
rophages which express VEGF-A is a hallmark of
many tumors. To test the involvement of macrophage
derived VEGF in tumor progression, cell-lineage-
specific, targeted deletion of VEGF has been
achieved. This results in VEGF gene expression in
75% of tumor-associated macrophages. Vasculature
in tumors lacking macrophages was less tortuous,
with increased pericyte coverage, and reduced vessel
length, indicating vascular normalization. However,
deletion of macrophage VEGF-A resulted in accel-
erated tumor progression and increased the suscep-
tibility of tumors to chemotherapeutic cytotoxicity.
This shows that macrophage-derived VEGF-A is
essential for tumorigenic alteration of the vasculature
and acts to retard, not promote, tumor progression
[22]. Tumor-associated macrophages, especially
those polarized to a proangiogenic M2-like pheno-
type, stimulate angiogenesis by releasing PIGF that
also contributes to vessel disorganization [23]. A
recently observed alternative phenotype of macro-
phages in glioma enhanced angiogenesis and glioma
growth [24]. Thus, blockade of macrophage infiltra-
tion could be a novel therapeutic strategy in gliomas.
Given the tendency for tumor-associated macro-
phages to express the M2 phenotype and contribute
to tumor progression by production of angiogenic
factors, targeting these cells could have a significant
therapeutic effect in glioblastoma. One macrophage-
targeting agent is trabectedin, known as Yondelis, an
74 Brain Tumor Pathol (2012) 29:73–86
123

agent used to treat soft-tissue sarcomas. Other
potential agents include the immunomodulator lino-
mide, which blocks the angiogenic effects of tumor-
associated macrophages, and the bisphosphonate
zoledronic acid, which inhibits matrix metallopro-
teinases (MMPs) and reduces angiogenesis [25, 26].
6. Stromal cells—3: cancer-associated fibroblast. Glio-
blastoma-associated stromal cells which are diploid,
do not have the genomic alterations typical of
glioblastoma cells and have phenotypic and func-
tional properties in common with the cancer-associ-
ated fibroblast described in the stroma of carcinomas
have recently been isolated [27]. They expressed
alpha-smooth muscle cell actin, suggesting a strong
relationship with angiogenesis, a component of the
vascular microenvironment, and a candidate for
antiangiogenic treatment.
7. Endothelial cell instability. Endothelial cells with
cytogenetic abnormalities or tumor endothelial cells,
which differentiate from CSC-like cells (as in
glioblastoma), may not be as sensitive to VEGF
(receptor) blockade as sprouting endothelial cells.
Treatment with bevacizumab induced expression of
VEGF-C and D in glioma cells and resulted in
increased proliferation of, and tube formation on,
tumor endothelial cells [28]. Treatment with bev-
acizumab may cause alterations in human brain and
tumor endothelial cells leading to escape mechanisms
from anti-VEGF therapy. VEGF-C and D might act
as alternative pro-angiogenic factors during anti-
VEGF therapy.
8. Vascular independence. Mutant tumor clones or
inflammatory cells are able to survive in hypoxic
tumors; their reduced vascular dependence impairs
the antiangiogenic response. Some tumors have a
hypovascular stroma. Tumors can also metastasize
via the lymphatics; their growth may not be blocked
by antiangiogenic therapy. Vascular remodeling
induced by anti-VEGF treatment leads to a more
hypoxic tumor microenvironment. This favors a
metabolic change in the tumor cells toward glycol-
ysis, which leads to enhanced tumor cell invasion of
the normal brain. There is, therefore, a need to
combine anti-angiogenic treatment in glioblastomas
with drugs targeting metabolic changes linked to the
glycolytic phenotype [29].
9. Mature vessels (pericyte coverage). Mature supply
vessels are covered by vascular smooth muscle cells
(pericytes) and are not easily pruned by EC-targeted
treatment. Tumor vessels covered by pericytes are
less sensitive to VEGF blockade [30, 31]. Recruit-
ment of pericytes is controlled by platelet-derived
growth factor (PDGF) receptor-beta (PDGFR-beta).
Sphingosie-1-phosphate receptor signaling is also
involved in the control of endothelial/pericyte inter-
actions. Angiopoietin-1 stabilizes vessels, promotes
pericyte adhesion, and makes the vessels leak-
resistant by tightening endothelial junctions. Eph-
rinB2 and Notch signaling are also involved in
maturation and arterial differentiation of pericytes.
These factors could be candidate targets. Removal of
coverage by pericytes leads to exposed tumor vessels,
which may explain the enhanced effect of anti-
angiogenic inhibitors [32, 33]. Bergers et al. [34]
showed that combined treatment with anti-PDGF-B/
PDGFR-B reduces pericyte coverage and increases
the success of anti-VEGF treatment in the mouse
cancer model. Sennino et al. [35] demonstrated that
treatment with a novel selective PDGF-B blockade
DNA aptamer, AX102, which blocks the action of
PDGF-B, led to progressive reduction of pericytes in
Lewis lung carcinomas. The same effect was
achieved when PDGF inhibitors were combined with
an antiangiogenic chemotherapy regimen that tar-
geted endothelial cells [36].
10. EC radioresistance and vascular permeability. Hyp-
oxic activation of HIF1alpha renders ECs resistant to
irradiation. In endothelial cell cultures, moderate
hypoxia (e.g. 5% oxygen) promotes expression of a
variety of angiogenic molecules, for example VEGF-
A and eNOS, which promotes EC proliferation,
survival, and migration. However, hypoxia may also
reduce EC proliferation and trigger EC apoptosis.
The net outcome may depend on the severity of
hypoxia [37]. Hypoxia also regulates endothelial
cell–cell junctions and vascular permeability. In the
brain, hypoxia disrupts the blood brain barrier, where
elevated VEGF-A levels under hypoxia are respon-
sible for increased vascular leakage. It has been
reported that after radiation-induced damage to tumor
vasculature, the tumor becomes more dependent on
vasculogenesis (recruitment of endothelial precursor
cells or bone marrow-derived hematopoietic cells to
initiate de novo capillary formation) to reinitiate
tumor growth. Kioi et al. [21] explored this hypoth-
esis using a murine intracranial glioblastoma xeno-
graft model and demonstrated that, after radiation
treatment, there was a cascade of events that resulted
in increased vasculogenesis. This cascade was initi-
ated by increased HIF-1 expression that led to
increased levels of SDF-1. Importantly, this process
was inhibited by AMD3100, which blocks the action
of SDF-1.
11. Gene variations of angiogenic factors. Gene varia-
tions in VEGF receptors determine the responsive-
ness to VEGF (receptor) blockade. Genetic variation
Brain Tumor Pathol (2012) 29:73–86 75
123

in VEGF and VEGFR2 was investigated by study of
blood samples from glioblastoma patients. Seventeen
tagging single nucleotide polymorphisms (SNPs) in
VEGF and 27 in VEGFR2, covering 90% of the
genetic variability within the genes, were genotyped
and analyzed. In VEGFR2, two SNPs have been
shown to be significantly associated with survival
[38]. VEGFR2 plays a crucial role in glioblastoma
development, prognosis, and response to therapy.
Genetic variations in this receptor could therefore be
of importance for the outcome of these tumors.
12. Vessel normalization. Transient vessel normalization
can reduce antiangiogenic drug delivery and efficacy;
alternatively, barrier tightening could impede drug
penetration. In normalization of tumor vessels, leak-
age, tortuosity, and remodeling are reduced, whereas
endothelial cell quiescence, barrier tightening, and
vessel maturation are increased—changes that boost
perfusion and reduce hypoxia [39]. A streamlined
monolayer of phalanx endothelial cells is also formed,
and acts as a more impenetrable barrier for intravasat-
ing tumor cell. To overcome this problem, we need a
precise therapeutic window for tumor vessel normal-
ization with antiangiogenic agents which act in a
region-specific and agent-specific manner.
13. Organ-specific differences. Tumors have contrasting
invasive behavior, depending on the organ of inoc-
ulation. Recently, a novel way of generating a mouse
glioblastoma multiforme model by use of Cre-loxP-
controlled lentiviral vectors has been reported [40].
There was little tumor formation from the cortex,
whereas tumor formation from the neurogenic areas,
for example the subventricular zone and hippocam-
pus, was observed in more than 75% of the mice.
There may be unknown growth factors in neurogenic
areas that promote the transformation of GFAP?
cells or GFAP? neural progenitor cells that may be
susceptible to transformation. These could be candi-
date targets.
14. Primary tumor versus invasive site. The signals that
regulate angiogenesis in primary tumors are distinct
from those active in invasive tumors. Recently, a
novel glioma model has been used to demonstrate,
histologically, two distinct patterns of invasion,
namely angiogenesis-dependent invasion with over-
expression of angiogenesis related genes, for example
VEGF, MMP9, HIF1, and PDGF, and angiogenesis-
independent invasion with overexpression of inva-
sion-related genes, for example integrin alphavbeta3,
MMP2, nestin, and secreted protein acidic and rich in
cysteine [41]. During initiation of the angiogenesis-
independent invasion phenotype, angiogenic signals
could be different.
Glioblastoma vessel formation
Normal vessel formation
Before anti-angiogenic treatment of angiogenic tumor
vessels can be understood, the precise mechanism of
normal vessel formation should be re-evaluated, because
anti-angiogenic therapy uses some of these vessel forma-
tion methods, with the eventual expectation of vessel
normalization.
Normal vessel equipment, branching, maturation,
and quiescence [20]
1. Quiescent endothelial cells. In a healthy adult, qui-
escent endothelial cells have long half-lives and are
protected against insults by the autocrine action of
maintenance signals, for example VEGF, Notch, an-
giopoietin-1, and FGFs.
2. Oxygen sensor (PHD2). Because vessels supply
oxygen, endothelial cells are equipped with oxygen
sensors and hypoxia-inducible factors—for example
prolyl hydroxylase domain 2 (PHD2) and hypoxia-
inducible factor-2alpha (HIF-2alpha), respectively—
which enable the vessels to adjust their shape to
optimize blood flow.
3. Phalanx cells. Quiescent endothelial cells form a
monolayer of phalanx cells with a streamlined
surface, interconnected by junctional molecules, for
example VE-cadherin and claudins.
4. Pericytes. These endothelial cells are ensheathed by
pericytes, which suppress endothelial proliferation
and release cell-survival signals. for example VEGF
and ANG-1 [31]. Endothelial cells and pericytes at
rest produce a common basement membrane.
5. Detachment of pericytes. When a quiescent vessel
senses an angiogenic signal, for example VEGF,
VEGF-C, ANG-2, FGFs or chemokines, released by
a hypoxic, inflammatory, or tumor cell, pericytes first
become detached from the vessel wall then liberate
themselves from the basement membrane by prote-
olytic degradation, which is mediated by (MMPs).
6. Tip cells. To build a perfused tube and prevent
endothelial cells from moving en mass towards an
angiogenic signal, one endothelial cell, known as the
tip cell, is selected to lead the tip in the presence of
factors such as VEGF receptors, neuropilins, and the
Notch ligands DLL4 and Jagged1.
7. Stalk cells. The neighbors of the tip cell assume
subsidiary positions as stalk cells, which divide to
elongate the stalk and establish the lumen [42].
8. Notch signal in vessel branching. The vessel-branch-
ing model postulates that, in general, tip cells migrate
76 Brain Tumor Pathol (2012) 29:73–86
123

and stalk cells proliferate. Recent studies have
implicated Notch signaling in this model. In response
to VEGF, activation of VEGFR2 upregulates DLL4
expression in tip cells. In neighboring stalk cells,
DLL4 then activates Notch, which downregulates
VEGFR-2 but upregulates VEGFR-1. Specification
of endothelial cells into tip and stalk cells is
controlled by the Notch pathway. Analysis of Notch
signaling revealed high Notch activity in stalk cells
but low levels of Notch signaling in tip cells. In
contrast, tip cells express higher levels of DLL4.
9. Tip cells migrate and stalk cells elongate. Tip cells
are equipped with filopodia to sense environmental
guidance cues, for example ephrins and semaphorins,
whereas stalk cells release molecules, for example
EGFL7, so that the stalk elongates. For a vessel to
become functional, it must become mature and
stable. Overall, DLL4 and Notch signaling restricts
branching but generates perfused vessels.
10. Jagged-1. Jagged1, another Notch ligand expressed
by stalk cells, promotes tip-cell selection by interfer-
ing with the reciprocal DLL4 and Notch signaling
from the stalk cell to the tip cell [43].
11. Resume quiescence. Endothelial cells resume their
quiescent phalanx state and signals such as PDGF-B,
ANG-1. TGF-B, ephrin-B2, and Notch cause the cells
to become covered by pericytes.
12. In summary, an unexpected complexity is that
endothelial cells continuously compete for the tip-
cell position by fine-tuning their expression of
VEGFR-2 versus VEGFR-1, indicating that this
signaling circuit is constantly re-evaluated as cells
meet new neighbors. Normal endothelial cells change
into several different types—phalanx cells, tip cells,
and stalk cells—as a result of cell signaling induced
by angiogenic and chemokine stimulation. During
anti-angiogenic therapy, each of these cycles is first
interrupted or inhibited, and then resumes. The final
objective of antiangiogenic therapy is to achieve
vessel normalization, with endothelial cells, phalanx
cells, tip cells, and stalk cells in their normal
morphological and biological states.
Several modes of vessel formation: difference
between tumor and normal tissue [20]
1. Several modes of vessel formation have been
identified.
2. Vasculogenesis. In the developing mammalian
embryo, angioblasts differentiate into endothelial
cells, which assemble into a vascular labyrinth—a
process known as vasculogenesis.
3. Postnatal vasculogenesis. Although debated, the
repair of healthy adult vessels or the expansion of
pathological vessels can be aided by recruitment of
bone-marrow-derived cells (BMDCs) and/or endo-
thelial progenitor cells to the vascular wall. The
progenitor cells then become incorporated into the
endothelial lining in a process known as postnatal
vasculogenesis.
4. Angiogenesis (sprouting). Subsequent vessel sprouting
ensures expansion of the vascular network that
remodels into arteries and veins, known as
angiogenesis.
5. Arteriogenesis. Arteriogenesis then occurs, in which
endothelial cell channels become covered by peri-
cytes or vascular smooth muscle cells, which provide
stability and control perfusion.
6. Intussusception. Pre-existing vessels can split by a
process known as intussusception, giving rise to
daughter vessels. There are distinct differences
between sprouting-type and intussusception-type
angiogenesis. In the sprouting type of angiogenesis
related to hypoxia, there is no blood flow in the rising
capillary sprout. In contrast, it has been shown that an
increase of wall shear stress initiates the intussuscep-
tion type of angiogenesis in skeletal muscle. Other-
wise, driving development, both sprouting and
intussusception act in parallel in building a vascular
network, although with differences in spatiotemporal
distribution. Thereby, in addition to regulatory mol-
ecules, flow dynamics support the patterning and
remodeling of the rising vascular tree [44]. These two
angiogenesis patterns have been shown to be used
differently in cancer models. Both irradiation and anti-
angiogenic therapy cause a switch from sprouting to
intussusceptive angiogenesis, indicative of an escape
mechanism and accounting for the development of
resistance [45].
7. Vessel co-option. Vessel co-option occurs in which
tumor cells take control of the existing vasculature.
Experimental animal models of glioma have suggested
that the initial, preangiogenic stages of neoplastic
growth actually involve regression of native blood
vessels. New tumor cells first gain access to a vascular
supply by co-option [46, 47]. Vaso-occlusion could
also result from angiopoietin-2 mediated endothelial
cell apoptosis and vascular regression, which follows
neoplastic co-option of native vessels in an animal
model of glioma [48]. In animal models, transmission
electron microscopic study shows co-option vessels
totally surrounded by neoplastic cells [49].
8. Vascular mimicry. Tumor cells can line vessels. (This
is discussed in a later section.)
Brain Tumor Pathol (2012) 29:73–86 77
123

9. Cancer stem-like cell transdifferentiation. Putative
cancer stem-like cells can even generate tumor
endothelium. (This is discussed in a later section.)
10. Differences between normal and tumor vessel
formation. In normal tissues, modes of vascular
formation are limited to vasculogenesis, sprouting,
and intussusception whereas in tumors, 6 modes of
vascular formation, including co-option of preexist-
ing vessels, vascular mimicry, and transdiffentia-
tion, are available. To target tumor vasculature,
preventing co-option of pre-existing vessels, vascu-
lar mimicry, and transdifferentiation should be
promising strategies. Vascular mimicry and trans-
differentiation are predominantly important in glio-
blastoma (Fig. 1).
Tumor vessels and endothelial cells
Morphological change of tumor vessels
In general, tumor vessels have abnormal structure and
function with seemingly chaotic organization. Highly
dense regions neighbor vessel-poor areas, and vessels vary
from a normally wide, irregular, and tortuous serpentine-
like shape to thin channels with small or compressed
lumens. Every layer of the tumor vessel wall is abnormal.
Endothelial cells lack a cobblestone appearance, are poorly
interconnected, and are occasionally multilayered. Also,
arterio-venous identity is ill defined, and shunting com-
promises flow. The basement membrane is irregular in
thickness and composition, and fewer, more loosely
Fig. 1 Modes of vessel
formation [20]. There are
several known methods of blood
vessel formation in normal
tissues and tumors. Vessel
formation can occur by
sprouting angiogenesis (a), by
recruitment of bone-marrow
derived and/or vascular-wall-
resident endothelial progenitors
(EPCs) that differentiate into
endothelial cells (ECs; b), or by
a process of vessel splitting
known as intussusception (c).
Tumor cells can co-opt pre
existing vessels (d), or tumor
vessels can be lined by tumor
cells (vasculogenic mimicry;
e) or by endothelial cells, with
cytogenetic abnormalities in
their chromosomes, derived
from putative cancer stem cells
(f). Unlike normal tissues,
which use sprouting
angiogenesis, vasculogenesis,
and intussusception (a–c),
tumors can use all six modes of
vessel formation (a–f)
78 Brain Tumor Pathol (2012) 29:73–86
123

attached, hypocontractile mural cells cover tumor vessels,
although tumor-type-specific differences exist.
In glioblastoma, microvascular proliferation (glomeru-
loid vessels) and pseudopalisading are hallmarks of the
histology and have been intensively investigated. Glom-
eruloid structure of newly formed vessels is frequently
observed [50]. In glioblastoma, the transformation and
proliferation of vascular smooth muscles may accompany
neovascularization and may also be induced by growth
factors [51]. Intense immunostaining with chemokine
CXCR12, particularly in the pseudopalisading cells and the
proliferating microvessels, was observed for Glioblasto-
mas [52]. Electron microscopic study has revealed the
morphology of the mitochondrial network and the vascular
component of the glioma [53], and angioarchitectural het-
erogeneity in human glioblastoma has quantified by use of
a new means of assessment [54].
Are tumor endothelial cells abnormal or normal?
An important concept in tumor angiogenesis is that tumor
endothelial cells are assumed to be genetically normal even
though they are structurally and functionally abnormal.
Tumor-associated endothelial cells express typical endo-
thelial cell markers, for example CD31. Hida et al. [55] and
Akino et al. [56] reported, unexpectedly, that tumor
endothelial cells were cytogenetically abnormal. FISH
analysis showed that freshly isolated uncultured tumor
endothelial cells were aneuploid and had abnormal multi-
ple chromosomes. Tumor endothelial cells, unlike normal
endothelial cells, are inherently genetically unstable, and
the extent of tumor endothelial cell aneuploidy was exac-
erbated in culture. They concluded that tumor endothelial
cells can acquire cytogenetic abnormalities while in the
tumor microenvironment.
It has been reported that tumor vessels may be derived
from intratumor embryonic-like vasculogenesis. This con-
dition might be because of normal stem and progenitor
cells of hematopoietic origin or resident in the tissues. For
tumor endothelial cells it is plausible that normal host
endothelial cells, on invading the tumor, acquire their
cytogenetic abnormalities from the tumor microenviron-
ment. One possibility is that mouse endothelial cells fuse
with human tumor cells [57] or take up human tumor
oncogenes, for example H-ras, by phagocytosis of apop-
totic bodies [58] or after mRNA and microRNA transfer by
exosomes and microvesicles [59]. Another possibility is
that the tumor cells transdifferentiate into endothelial cells,
as has been reported in leukemia [60]. A recent publication
reports that a median of 37% of endothelial cells in B-cell
lymphomas harbor lymphoma-specific chromosomal
translocations, and that the lymphoma cells and endothelial
cells share chromosomal abnormalities. A genetic rela-
tionship between lymphoma cells and endothelial cells has
been suggested [65].
Tumor cells may therefore promote genetic instability in
tumor endothelial cells by two distinct mechanisms—giv-
ing rise to them directly or by sending a signal to a nearby
endothelial cell. These results suggest the possible devel-
opment of an ideal anti-angiogenic therapy. First, not only
the tumor compartment, but also genetically unstable
tumor endothelial cells, may contribute to drug resistance.
Moreover, there seems to be a link between increased
activity of the signaling cascades that promote blood-vessel
formation and chromosome abnormalities in endothelial
cells [61].
Vasculogenic mimicry
Three papers [5–7] show that stem-like cells in the tumor
can differentiate into endothelial cells, thereby generating
the tumor vasculature. This is not the first time researchers
have suggested that cancer cells can make their own blood
vessels. In 1999, Maniotis et al. [62] reported a similar
effect in melanoma cells which they called ‘‘vascular
mimicry’’. Aggressive melanoma cells may generate vas-
cular channels that facilitate tumor perfusion independent
of tumor angiogenesis. The generation of microvascular
channels by genetically deregulated aggressive tumor cells
was termed ‘‘vasculogenic mimicry’’ to emphasize their de
novo generation without the participation of endothelial
cells and independent of angiogenesis [63].
Another team [64] has also investigated the source of
cells contributing to tumor vessels, and has shown that
tumor stem-like cells cultured from human glioma tumors
form endothelial cells in vitro. The authors detected
channels lined with tumor-derived cells in mice trans-
planted with human tumors—a process they classify as
vasculogenic mimicry. However, their analysis of the ori-
ginal human tumors was limited to marker expression, and
so they could draw no firm conclusion about the relation-
ship between the tumor cells and the endothelial cells.
Similarly, other groups [65, 66] have presented evidence of
genetic abnormalities common to tumor cells and endo-
thelial cells, but their data did not distinguish among sev-
eral potential mechanisms for the observations.
Tumor-associated endothelial vessels have been repor-
ted in human neuroblastoma [66], human B-cell lymphoma
[67], and breast cancer [68]. Human ovarian cancer cells
expressing CD31 and factor VIII of vascular endothelial
markers, had the plasticity to engage in vasculogenic
mimicry in vivo [69]. These findings suggest that the
plasticity of cancer cells enables them to mimic the
activities of endothelial cells and participate in the process
Brain Tumor Pathol (2012) 29:73–86 79
123

of vasculogenic mimicry. A study of breast cancer found
CSCs in endothelial differentiating medium were capable
of differentiating into endothelial cells, which were able to
form both vessels and tumor [68]. Vasculogenic mimicry is
currently regarded as one of mode of tumor vessel for-
mation [20].
Vasculogenic mimicry in glioblastoma
Hallani et al. [70] analyzed human glioblastoma tissues and
found non-endothelial cell-lined blood vessels that were
formed by tumor cells (vasculogenic mimicry of the
tubular type). They hypothesized that CD133? glioblas-
toma cells with stem-cell properties may express pro-vas-
cular molecules enabling them to form blood vessels de
novo. They demonstrated in vitro that glioblastoma stem-
like cells were capable of vasculogenesis and expression of
endothelial-associated genes. Moreover, some of these
glioblastoma stem-like cells could transdifferentiate into
vascular smooth muscle-like cells. They described a new
mechanism of glioblastoma vascularization which might
lead to a new strategy for antivascular treatment.
Vasculogenic mimicry has been reported in a study of
malignant astrocytomas [71]. Recently, Liu et al. [72]
reported the association between vasculogenic mimicry
and the clinical characteristics of 101 glioma patients.
Survival of patients with vasculogenic mimicry-positive
gliomas was less than for those with vasculogenic mim-
icry-negative gliomas.
Cancer stem cell and vasculogenic mimicry
Cancer stem cells are closely associated with tumor vas-
culogenic mimicry. Yao et al. [73] hypothesize that CSCs
might participate in vasculogenic mimicry to induce
angiogenesis. This hypothesis is based on the observation
that many CSC biomarkers are co-expressed with angio-
genic markers. Vasculogenic mimicry produces vascular-
like structures, mimicking the pattern of the embryonic
vascular network, through which tumor tissues nourish
themselves. Tumor cells with high differentiation plasticity
may contribute to the de novo formation of tumor cell-lined
blood channels. Zhang et al. [74] proposed three-stage
blood supply patterns in tumor—vasculogenic mimicry,
mosaic vessels [75], and endothelium-dependent vessels.
All three patterns supplied blood to the tumors. The model
proposes that vasculogenic mimicry is the dominant blood-
supply pattern in the early stage of tumor growth.
It has been demonstrated that endothelial cells sur-
rounding CSCs directly generate a specific microvascular
microenvironment and/or secrete factors that promote the
formation and/or maintenance of brain CSCs [4]. VEGF in
the microenvironment, which may be derived mainly from
CSCs, directly affects the phenotype of CSCs and promotes
CSC-associated vasculogenic mimicry. On the basis of
these findings, it is plausible that the microenvironment
CSC compartment controls the differentiation plasticity of
CSCs which is responsible for tumor vasculogenesis,
including vasculogenic mimicry.
Vasculogenic mimicry-targeted anti-angiogenic therapy
Yao et al. recently showed that expression of MMP9 and
MMP2 is upregulated in CSCs derived from the U87 cell
line [76]. They observed that migration-associated mole-
cules, including two G-protein coupled (FPR) and CXCR4
were overexpressed in CSC isolated from human glio-
blastoma and from the U87 cell line. Traditional anti-
angiogenic drugs, for example angiostatin and endostatin,
which target normal endothelial cells, have little effect on
vasculogenic mimicry, because of the absence of normal
endothelial cells. Vasculogenic mimicry targeted therapy
may destroy the microenvironment that maintains CSCs,
block the metastasis passage tumor cells, and reduce the
recurrence of cancer.
Cancer stem-like cell transdifferentiation
To grow, solid tumors need a blood supply [77]. They
recruit new blood vessels mainly by inducing the sprouting
of endothelial cells from external vessels and promoting
the cells’ migration into the tumor. This ability, called the
angiogenic switch, is required for tumor cells to invade
surrounding tissue and metastasize to distant sites—the
deadly hallmarks of cancer [78]. Recent papers [5–7] show
that, in addition to recruiting vessels from outside, brain
tumors produce endothelial cells for vessel formation from
within.
The concept of existence of CSCs in solid tumors
including glioblastoma has revolutionized understanding of
tumor biology and gives a plausible explanation of treat-
ment failure [79]. Circulating progenitor cells, for example
endothelial progenitors recruited into the tumor stroma,
emphasize the important contribution of vasculogenesis to
tumor neoangio-architecture formation. In addition, Ricci-
Vitiani et al. [5], Wang et al. [6], and Soda et al. [7] suggest
the idea of de novo vasculogenesis in glioblastomas. These
three groups of investigators provided evidence showing
there is a population of cancer stem-like cells within gli-
omas that have endothelial progenitor properties with the
capacity to form functional endothelial cells and to par-
ticipate actively in neovascular formation.
80 Brain Tumor Pathol (2012) 29:73–86
123

Recent research in tumor endothelial biology has
focused on two main concepts. According to the first
concept—vasculogenic mimicry—some tumor cells take
on specific characteristics of vascular endothelial cells and
line the tumor’s blood vessels as described above [80]. The
origin of such tumor cells is ill-defined: whereas one study
[70] suggested that tumor stem cells participate in vascul-
ogenic mimicry, it is generally believed that tumor cells in
the immediate environment of the nascent vessel are co-
opted for the purpose. The co-opted cells are thought to
retain most of their tumor-cell characteristics while
acquiring a limited number of endothelial-cell features.
The second concept is that a CSC microenvironment
provides the environment for CSC self-renewal and
maintenance, stimulating essential signaling pathways on
CSCs and leading to secretion of factors that promote
angiogenesis [81]. For glioblastomas, CSCs have been
reported to promote tumor angiogenesis while being
resistant to chemotherapy and radiotherapy. Specifically,
glioma stem cells have been shown to affect angiogenesis
and vasculogenesis by increased expression of VEGF and
SDF1 [82]. The deadly brain tumor glioblastoma is thought
to arise from the perivascular microenvironment of tumor
stem cells [83]. Cancer stem-like cell transdifferentiation is
currently regarded as one of mode of tumor vessel for-
mation [20].
Transdifferentiation experiments
Ricci-Vitiani et al. [5] analyzed chromosomal aberration
within the vascular tree in 15 glioblastomas, by use of
immunofluorescence and in situ hybridization, and found a
significant amount of endothelial cells (CD31?/CD144?)
originated from the tumor and shared the same chromo-
somal aberration as the tumor cells.
They gained further insight by generating undifferenti-
ated cell aggregates from human tumor-derived CD133-
expressing cells and grafting them into mice. When they
implanted freshly isolated CD31?/CD144? cells from
glioblastomas in recipient mice, they were able to reca-
pitulate tumorigenesis that produced anaplastic vascula-
ture. Undifferentiated cell populations from human
glioblastoma tissue also formed vascularized tumors in
immunodeficient mice, and a high percentage of the
endothelial cells in the vessels in the middle of these
tumors were of human origin. When these glioblastoma-
derived endothelial cells in the xenografts were selectively
killed, the tumors decreased in size and became necrotic,
showing that vessels containing these cells are essential for
survival of specific regions of the tumor. Human endo-
thelial cells were found exclusively along the vascular wall
within the core of the tumor mass when neurospheres
derived from CD133?/CD31- cells were grafted into
mice, whereas murine-derived endothelial cells were
detected peripherally mainly along the vascular tree in the
capsule of the tumor.
Despite this topographic segregation between the host
and recipient endothelial progenitors, the neoangioarchi-
tecture was noted to be functional and with circulating
erythrocytes. The internal vessels of the resulting tumors
expressed human vascular markers, whereas more external
vessels carried mouse-specific endothelial-cell markers.
The authors also found human endothelial cells in tumor
vessels linked to the mouse vessels and delivering blood to
the tumor.
Wang et al. [6] provided evidence that CD133?/CD144-
and CD133?/CD144? cells fractionated from human glio-
blastoma specimens were capable of developing capillary
structures in vitro.
Although CD133-/CD144? and CD133-/CD144-
cells both were capable of forming tumors in immunode-
ficient mice, only CD133?
/CD144? cells were found to
form highly vascular tumors. The blood vessels in these
tumors expressed human markers, confirming that they had
developed from the xenografted cells.
Wang et al. focused on the CD133? stem cell-like
fraction of tumor cells and found a subset of cells
expressing vascular endothelial cadherin (also known as
CD144) that behaved like endothelial progenitors. These
cells could mature into endothelial cells. Lineage analysis
and single-cell clonal studies showed that a subpopulation
of this CD133? stem-like cell fraction is multipotent,
differentiating into both tumor and endothelial cell types.
They showed that a clone of cells derived from a single
tumor cell, which expressed CD133 but not VE-cadherin,
was multipotent: in vitro, the cells differentiated into both
neural cells (which eventually form tumor cells) and
endothelial cells. Many of the cells from these fractionated
populations also had the genomic aberrations seen in the
parent tumor. Importantly, blocking VEGF A pharmaco-
logically with bevacizumab (Avastin; Genentech/Roche) or
using small hairpin RNAs did not inhibit the differentiation
of CD133? cells into endothelial cell types.
Soda et al. [7] examined the vessels of glioblastomas in
tumors induced by transduction of p53
?/-
heterozygous
mice with lentiviral vectors containing oncogenes and the
marker GFP in the hippocampus of GFAP-Cre recombi-
nase (Cre) mice. Transplantation of mouse GBM cells
revealed that the tumor-derived endothelial cells (TDECs)
originated from tumor-initiating cells and did not result
from cell fusion of endothelial cells and tumor cells.
An in-vitro differentiation assay suggested that hypoxia
is an important factor in the differentiation of tumor cells to
endothelial cells and is independent of VEGF. TDEC for-
mation was not only resistant to an anti-VEGF receptor
Brain Tumor Pathol (2012) 29:73–86 81
123

inhibitor in mouse GBMs but also led to an increase in their
frequency. A xenograft model of human glioblastoma
spheres from clinical specimens and direct clinical samples
from patients with glioblastoma also showed the presence
of TDECs. The TDECs are functional because blood flows
through them.
Although TDECs share common endothelial markers,
for example CD31, CD34, vVW, and CD144, their special
feature is the lack of expression of VEGFR2, the major
tyrosine kinase receptor of VEGF-A. The absence of
VEGFR is further illustrated by the negligible effect of
inhibition of all VEGF receptors on either EC-dependent
tube formation in Matrigel or in-vivo tumor growth.
TDECs have a unique VEGF-A and bFGF-independent
angiogenic mechanism that potentially accounts for the
resistance of glioblastoma to anti-VEGF-A therapy. Fur-
thermore, cord formation by TDECs continues to occur
under hypoxia conditions, when VEGF-A autocrine func-
tion is blocked by neutralizing antibody or when auto-
phosphorylation of VEGFR tyrosine kinase is inhibited by
an antagonist, suggesting that transdifferentiation of tumor
cells into ECs is VEGF-independent. This observation is
confirmed in vivo by showing that survival does not change
for animals treated with this antagonist. In the treated
animals, TDECs increase, more at the tumor margins than
in deep areas, suggesting TDECs are involved in resistance
to anti-VEGF treatment.
Further questions about transdifferentiation
The findings of these three research groups reveal a novel
link between glioblastoma cells and endothelial progenitors
and thus, a new mechanism for tumor vascularization. This
work may lead to development of therapeutic strategies for
this aggressive malignancy. They found that a glioblastoma
cell population that could differentiate into endothelial
cells and form blood vessels in vitro was enriched in cells
expressing the tumor-stem-cell marker CD133.
Hypoxia is an important factor in the differentiation of
tumor cells to endothelial cells and is independent of
VEGF. Although TDECs share some common endothelial
markers, for example CD31, CD34, vVW, and CD144,
their special feature is lack of expression of VEGFR2, the
major tyrosine kinase receptor of VEGF-A.
The articles highlight the importance of CSCs and their
potential to differentiate into endodermal lineage cells that
are essential to the establishment of a tumor and its
microenvironment. However, the importance of other naive
progenitor cells derived from the host environment is a
question that needs further investigation.
Compared with vasculogenic mimicry, endothelial
transdifferentiation has not been fully investigated. First,
how general is the differentiation of tumor stem-like cells
into endothelial cells? What are the conditions promoting
differentiation of tumor stem-like cells into endothelial
cells and determining the prevalence of this process within
a given tumor environment. For example, does local
shortage of oxygen trigger this differentiation? Soda et al.
[7] demonstrated that an in-vitro differentiation assay
suggested that hypoxia is an important factor in the dif-
ferentiation of tumor cells to ECs and is independent of
VEGF. Are there other important factors? What is the
functional significance of tumor origin to the vascular
endothelium? Finally, it will be crucial to determine how
TDECs and vessels differ from their non-tumor counter-
parts in both morphology and function.
Current studies examine the molecular pathways that
regulate the formation of tumor-derived endothelium at a
superficial level. These studies focused on glioblastomas,
and so the relevance of this pathway in other tumors of
suspected stem-cell origin must also be determined.
Recently, the possibility of EC differentiation of tumor
cells has been suggested in lymphoma, myeloma, chronic
myeloid leukemia (CML), breast cancer, and neuroblas-
toma [60, 65–68].
Other cell types of the underlying support tissue
(stroma), for example fibroblasts, are also involved in
tumor formation and progression. Do tumor stem cells
contribute to these non-endothelial stromal lineages, and, if
so, under what conditions?
Defining the relevant mechanisms thoroughly is an
essential prerequisite for design of new therapy. In contrast
with the conventional theory of tumor angiogenesis in
which endothelial cells are derived from mesodermal bone
marrow progenitor cells, the presence of TDECs in GBM
suggests that the endothelial cells transdifferentiated from
the neuroectoderm and that tumor cells can also be
involved in tumor angiogenesis. The endothelial transdif-
ferentiation of the tumor cells may result from the aberrant
stem cell character of the tumor progenitor cells. Another
possible mechanism is that the endothelial cell differenti-
ation of GBM cells is not the result of transdifferentiation
but reflects the normal differentiation pathway of NSC,
which have previously been reported as differentiating into
endothelial cells [84].
Transdifferentiation targeted anti-angiogenic therapy
Using these experimental models, Wang et al. [6] further
illustrated the therapeutic implication of manipulating the
Notch-1 signaling pathway and blocking VEGF receptor
signaling in the differentiation of CD133?/CD144- cells
to double-positive cells and their maturation along the
endothelial phenotype. They suggest that differentiation of
82 Brain Tumor Pathol (2012) 29:73–86
123

tumor stem-like cells into endothelial cells might be
mediated by signaling pathways involving two proteins—
VEGF and Notch. The authors propose that Notch regu-
lates the initial differentiation of tumor stem-like cells into
endothelial progenitor cells, whereas VEGF selectively
affects the differentiation of endothelial progenitors into
TDECs. Notch signaling is also critical in pathways of
CSCs and glioblastoma angiogenesis [85]. Notch inhibitor
DAPT (r-secretase inhibitor) not only leads to reduction
of the self-renewal ability of CSCs and the number of
CD133? tumor cells, it also reduces the expression of
vascular markers such as CD105, CD31, and Von Wille-
brand factor [85].
Future directions of anti-angiogenic therapy
for glioblastoma
1. In the short term. An important question is how anti-
angiogenic medicine can be improved in the short
term, and how the use of current anti-VEGF agents
should be optimized. It is not known whether the
approved antiangiogenic regimens are optimally used
in terms of dosing, duration, and combination therapy.
Another matter of high priority, given the low
response, is the discovery of predictive biomarkers to
identify responders among the large patient group of
non-responders, and identification of treatment tailored
for particular tumors and stage [86–88]. Little is
understood about the mechanisms of vascularization of
micrometastatic lesions, and agents that can block
other modes of tumor vascularization (for example
vasculogenesis, intussusception co-option, vasculo-
genic mimicry, transdifferentiation) are needed. Fur-
thermore, understanding the mechanistic differences
between VEGFR TKIs and anti-VEGF antibodies will
help to optimize the design of anticancer treatments.
2. In the intermediate term. Anti-VEGF agents could be
combined with agents that the escape pathways
detected in clinical studies. Examples are ANG-2,
PIGF, SDF-1, and CXCR4 [13, 86]. The challenge will
be when to add these second agents—before, during,
or after anti-VEGF therapy.
3. In the long term. the therapeutic potential of sustained
vessel normalization to suppress metastasis and
enhance chemotherapy will need to be evaluated
clinically, and additional studies are required to estab-
lish how this could be combined best with available
vessel pruning therapy [89]. Development of additional
antiangiogenic drugs [90], independent of VEGF
signaling, and evaluation of their potential in clinical
trials, in particular as combination therapy with current
VEGF (receptor) inhibitors, is likely to expand the
antiangiogenic armamentarium [91].
4. New strategies. Finally, additional strategies, with
specific targets, could be based on relatively new
information about modes of tumor vascularization,
vasculogenic mimicry, and endothelial cell transdif-
ferentiation. Another new concept of vascular normal-
ization is unique. Endothelial cells are equipped with
the oxygen sensor PHD2, which senses shortage of
oxygen and starts a feedback loop to read vessel
perfusion and oxygenation. Partial loss of PHD2
induces endothelial cell normalization, resulting in
streamlined phalanx formation of quiescent, tightly
interconnected, and orderly arranged endothelial cells,
which improves perfusion and oxygenation. Thus,
blocking PHD2 in endothelial cells may be a concep-
tually novel strategy to convert tumors into a more
benign phenotype, by improving—not impairing—
Fig. 2 New antiangiogenic
strategies. These target
vasculogenic mimicry,
endothelial cell
transdifferentiation, oxygen
sensor PHD2 inhibition, and
epigenetic control using
antiangiogenic miRNA
Brain Tumor Pathol (2012) 29:73–86 83
123

tumor vessel perfusion [92, 93] and by disabling a
tumor’s ability to adjust to hypoxic conditions [94].
There is increasing evidence of epigenetic control of
angiogenesis, particularly by non-coding microRNAs
(miRNAs) which induce messenger RNA degradation
or block translation. Because miRNAs target multiple
genes, they are well positioned to regulate complex
processes such as angiogenesis. Endothelial cells
express several miRNAs that are induced by hypoxia
or VEGF. Most of these stimulate angiogenesis by
taking control of pro-angiogenic cascades while sup-
pressing angiostatic pathways [95]. In human brain
tumors it has been shown that miR-296 is elevated in
tumor-related endothelial cells of new vessels and this
miRNA may govern growth factor receptor expression
[96]. MicroRNAs are involved in multiple hallmark
biological characteristics of glioblastoma, including
angiogenesis [97]. Inhibition of miR-93 function may
be a feasible approach to suppressing angiogenesis and
glioma growth [98]. Crosstalk between microRNA and
Notch signaling, which is important in glioma angio-
genesis, endothelial transdifferentiation, and normal
vessel formation has been identified [99]. MicroRNA
is also involved in blood vessel development, for
example sprouting angiogenesis, intussusception angi-
ogenesis, and flow-driven remodeling [100]. Epige-
netic regulation of angiogenesis is a promising new
strategy (Fig. 2).
Summary
The recent failure of VEGF treatment for glioblastomas
does not mean the strategy of targeting the blood supply is
inherently faulty. There are many more ways of affecting
the vasculature than by use of such drugs. We must identify
either more than one drug to shut down blood vessels of
tumor and non-tumor origin, or a common molecular
mechanism affecting both the tumor and the vasculature.
New molecular mechanisms of tumor vascular formation,
for example vasculogenic mimicry and transdifferentiation,
and new antiangiogenic strategies, for example oxygen
sensor and epigenetic control, have been discovered. These
should lead to new targets for anti-angiogenesis.
Acknowledgments Ministry of Education, Culture, Sports, Science
and Technology of Japan (grant nos 21390403 and 21659338) (S.
Takano)
References
1. Jain RK, di Tomaso E, Duda DG et al (2007) Angiogenesis in
brain tumours. Nat Rev Neurosci 9:610–622
2. Beal K, Abrey LE, Gutin PH (2011) Antiangiogenic agents in
the treatment of recurrent or newly diagnosed glioblastoma:
analysis of single-agent and combined modality approaches.
Radiat Oncol 6:2. doi:10.1186/1748-717x-6-2
3. Norden AD, Drappatz J, Wen PY (2008) Novel anti-angiogenic
therapies for malignant gliomas. Lancet Neurol 7:1152–1160
4. Calabrese C, Poppleton H, Kocak M et al (2007) A perivascular
niche for brain tumor stem cells. Cancer Cell 11:69–82
5. Ricci-Vitiani L, Pallini R, Biffoni M et al (2010) Tumour vas-
cularization via endothelial differentiation of glioblastoma stem-
like cells. Nature 468:824–828
6. Wang R, Chandalavada K, Wilshire J et al (2010) Glioblastoma
stem-like cells give rise to tumour endothelium. Nature 468:
829–833
7. Soda Y, Marumoto T, Friedmann-Morvinski D et al (2011)
Transdifferentiation of glioblastoma cells into vascular endo-
thelial cells. Proc Natl Acad Sci USA 18:4274–4280
8. Charles NA, Holland EC, Gilbertson R et al (2011) The brain
tumor microenvironment. Glia 59:1169–1180
9. Bergers G, Hanahan D (2008) Modes of resistance to anti-
angiogenic therapy. Nature Rev Cancer 8:592–603
10. Ferrara N (2010) Pathways mediating VEGF-independent tumor
angiogenesis. Cytokine Growth Factor Rev 21:21–26
11. Batchelor TT, Sorensen AG, di Tomaso E et al (2007)
AZD2171, a pan-VEGF receptor tyrosine kinase inhibitor, nor-
malize tumor vasculature and alleviates edema in glioblastoma
patients. Cancer Cell 11:83–95
12. Lieu C, Heymach J, Overman M et al (2011) Beyond VEGF:
inhibition of the fibroblast growth factor pathway and antian-
giogenesis. Clin Cancer Res 17:6130–6139
13. di Tomaso E, Snudel M, Kamoun WS et al (2011) Glioblastoma
recurrence after cediranib therapy in patients: lack of rebound
revascularization as mode of escape. Cancer Res 71:19–28
14. Ebos JML, Lee CR, Cruz-Hunoz W et al (2009) Accelerated
metastasis after short-term treatment with a potent inhibitor of
tumor angiogenesis. Cancer Cell 15:232–239
15. de Groot JF, Fuller G, Kumar AJ et al (2010) Tumor invasion
after treatment of glioblastoma with bevacizumab: radiographic
and pathological correlation in humans and mice. Neuro-Oncol
12:233–242
16. Kenig S, Alonso MBD, Mueller MM et al (2010) Glioblastoma
and endothelial cells cross-talk, mediated by SDF-1, enhances
tumour invasion and endothelial proliferation by increasing
expression of cathepsins B, S, and MMP9. Cancer Lett 289:
53–61
17. Lucio-Etervic AK, Piao Y, de Groot JF et al (2009) Mediators of
glioblastoma resistance and invasion during antivascular endo-
thelial growth factor therapy. Clin Cancer Res 15:4589–4599
18. Paez-Ribes M, Allen E, Hudock J et al (2009) Antiangiogenic
therapy elicits malignant progression of tumors to increased
local invasion and distant metastasis. Cancer Cell 15:220–231
19. Mizukami Y, Li J, Zhang X et al (2004) Hypoxia-inducible
factor-1-independent regulation of vascular endothelial growth
factor by hypoxia in colon cancer. Cancer Res 64:1765–1772
20. Carmeliet P, Jain RK (2011) Molecular mechanisms and clinical
applications of angiogenesis. Nature 473:298–306
21. Kioi M, Vogel H, Schultz G et al (2010) Inhibition of vascu-
logenesis, but not angiogenesis, prevents the recurrence of
glioblastoma after irradiation in mice. J Clin Invest 120:694–705
22. Stockmann C, Doedens A, Weidermann A et al (2008) Deletion
of vascular endothelial growth factor in myeloid cells accelerate
tumorigenesis. Nature 456:814–818
23. Potente M, Gerhardt H, Carmeliet P (2011) Basic and thera-
peutic aspects of angiogenesis. Cell 146:873–887
24. Gabrusiewiczk K, Ellert-Miklaszewska A, Lipko M et al (2011)
Characteristics of the alternative phenotype of microglia/
84 Brain Tumor Pathol (2012) 29:73–86
123

macrophages and its modulation in experimental gliomas. PLos
One 6:e23902
25. Mukhtar RA, Nseyo O, Campbell MJ et al (2011) Tumor-
associated macrophages in breast cancer as potential biomarkers
for new treatments and diagnostics. Expert Rev Mol Diagn 11:
91–100
26. Cimini E, Piacentini P, Sacchi A et al (2011) Zoledronic acid
enhances V2 T-lymphocyte antitumor response to human glioma
cell lines. Int J Immunopathol Pharmacol 24:139–148
27. Clavreul A, Etcheverry A, Chassevent A et al. (2011) Isolation
of a new cell population in the glioblastoma microenvironment.
J Neurooncol. doi:101007/s11060-011-0701-7
28. Grau S, Thorsteinsdottir J, von Baumgarten L et al (2011)
Bevacizumab can induce reactivity to VEGF-C and -D in human
brain and tumour derived endothelial cells. J Neurooncol 104:
103–112
29. Keunen O, Johansson M, Oudin A et al (2011) Anti-VEGF
treatment reduces blood supply and increases tumor cell
invasion in glioblastoma. Proc Natl Acad Sci USA 108:
3749–3754
30. Gaengeal K, Genove G, Armulik A et al (2009) Endothelial-
mural cell signaling in vascular development and angiogenesis.
Arterioscler Thromb Vasc Biol 29:630–638
31. Greenberg JI, Shields DJ, Barillas SG et al (2008) A role for
VEGF as a negative regulator of pericyte function and vessel
maturation. Nature 456:809–813
32. Bergers G, Sons S (2005) The role of pericytes in blood-vessel
formation and maintenance. Neuro-Oncol 7:452–464
33. Ribatti D, Nico B, Crivellato E (2011) The role of pericytes in
angiogenesis. Int J Dev Biol 55:261–268
34. Bergers G, Song S, Meyer-Morse N et al (2003) Benefits of
targeting both pericytes and endothelial cells in the tumor vas-
culature with kinase inhibitors. J Clin Invest 111:1287–1295
35. Sennino B, Falcon BL, Mggauley D et al (2007) Sequential loss
of tumor vessel pericytes and endothelial cells after inhibitor of
platelet derived growth factor B by selective aptamer AX102.
Cancer Res 67:6367–7358
36. Pietras K, Hanahan D (2005) A multi targeted, metronomic and
maximum-tolerated dose ‘‘chemo-switch’’ regimen is antian-
giogenic, producing objective responses and survival benefit in a
mouse model of cancer. J Clin Oncol 23:939–952
37. Fong GH (2008) Mechanisms of adaptive angiogenesis to tissue
hypoxia. Angiogenesis 11:121–140
38. Sjostrom SW, Wibom C, Andersson U et al (2011) Genetic
variations in VEGF and VEGFR2 and glioblastoma outcome.
J Neurooncol 104:523–527
39. Mazzone M, Dettori D, de Oliveira RL et al (2009) Heterozy-
gous deficiency of PhD2 restores tumor oxygenation and inhibits
metastasis via endothelial normalization. Cell 136:839–851
40. Marumoto T, Tashiro A (2009) Friedmann-Morvinski D et al
(2009) Development of a novel mouse glioma model using
lentiviral vectors. Nat Med 15:110–116
41. Inoue S, Ichikawa T, Kurozumi K et al (2011) Novel animal
glioma models that separately exhibit two different invasive and
angiogenic phenotypes of human glioblastoma. World Neuro-
surgery. doi:10.1016/j.wneu.2011.09005
42. Phng LK, Gerhardt H (2009) Angiogenesis: a team effort
coordinated by NOTCH. Dev Cell 16:196–208
43. Benedit R, Roca C, Sorensen I et al (2009) The Notch ligands
Dll4 and jagged1 have opposing effects on angiogenesis. Cell
137:1124–1135
44. Styp-Rekouska B, Hlushchuk R, Pries AR et al (2011) Intus-
susceptive angiogenesis: pillars against the blood flow in the
rising capillary sprout. Acta Physiol (oxf) 202:213–223
45. Hlushchuk R, Riesterer O, Baum O et al (2008) Tumor recovery
by angiogenic switch from sprouting to intussusceptive
angiogenesis after treatment with PTK787/ZK222584 or ioniz-
ing radiation. Am J Pathol 173:1173–1785
46. Holash J, Maisonpierre PC, Compton D et al (1999) Vessel
cooption, regression, and growth in tumors mediated by angio-
poietins and VEGF. Science 284:1994–1998
47. Zagzag D, Amirnovin R, Greco MA et al (2000) Vascular
apoptosis and involution in glioma precede neovascularization:
a novel concept for glioma and angiogenesis. Lab Invest 80:
837–849
48. Brat DJ, van Meir EG (2004) Vaso-occlusive and prothrombotic
mechanisms associated with tumor hypoxia, necrosis, and
accelerated growth in glioblastoma. Lab Invest 84:397–405
49. Arismerdi-Morillo G, Castellano A (2005) Tumoral micro-blood
vessels and vascular microenvironment in the astrocytic tumors.
A transmission electron microscopy study. J Neuro-oncol 73:
211–217
50. Sharma S, Sharma MC, Gupta DK et al (2006) Angiogenic
patterns and their quantitation in high grade astrocytomas.
J Neurooncol 79:19–30
51. Takeuchi H, Hashimoto N, Kitai R et al (2010) Proliferation of
vascular smooth muscle cells in glioblastoma. J Neurosurg 113:
218–224
52. Komatsu H, Sugita Y, Arakawa F et al (2009) Expression of
CXCL12 on pseudopalisading cells and proliferating micro-
vessels in glioblastomas: an accelerated growth factor in glio-
blastomas. Int J Oncol 34:665–672
53. Arismendi-Morillo G (2011) Electron microscopy morphology
of the mitochondrial network in glioma and their vascular
component. Biochim Biophys Acta 1807:602–608
54. Dilera A, Grizzi F, Sherif C et al (2011) Angioarchitectural
heterogeneity in human glioblastoma multiforme: a fractal-
based histopathological assessment. Microvasc Res 81:222–230
55. Hida K, Hida Y, Amin DN et al (2004) Tumor-associated
endothelial cells with cytogenetic abnormalities. Cancer Res 64:
8249–8255
56. Akino T, Hida K, Hida Y et al (2009) Cytogenetic abnormalities
of tumor-associated endothelial cells in human malignant
tumors. Am J Pathol 175:2657–2667
57. Storchova Z, Pellman D (2004) From polyploidy to aneuploidy,
genome instability and cancer. Natl Rev Mol Cell Biol 5:45–54
58. Bergsmedh A, Szeles A, Henriksson M et al (2001) Horizontal
transfer of oncogenes by uptake of apoptotic bodies. Proc Natl
Acad Sci USA 98:6407–6411
59. Bussolati B, Grange C, Camussi G (2011) Tumor exploits
alternative strategies to achieve vascularization. FASEB J 25:
2874–2882
60. Gunsllius E (2003) Evidence from a leukemia model for
maintenance of vascular endothelium by bone-marrow-derived
endothelial cells. Adv Exp Med 522:17–24
61. Taylor SM, Nevis KR, Plate HL et al (2010) Angiogenic factor
signaling regulates centrosome duplication in endothelial cells
of developing blood vessels. Blood 116:3108–3117
62. Maniotis AJ, Folberg R, Hess A et al (1999) Vascular channel
formation by human melanoma cells in vivo and in vitro: vas-
culogenic mimicry. Am J Pathol 155:739–752
63. Folberg R, Hendrix MJC, Maniotis AJ (2000) Vasculogenic
mimicry and tumor angiogenesis. Am J Pathol 156:361–381
64. Dong J, Zhang Q, Huang Q et al (2010) Glioma stem cells
involved in tumor tissue remodeling in a xenograft model.
J Neurosurg 113:249–260
65. Sneubel B, Chott A, Huber D et al (2004) Lymphoma specific
genetic abnormalities in microvascular endothelial cells i8n B
cell lymphomas. N Engl J Med 351:250–259
66. Pezzolo A, Parodi F, Corrias MV et al (2007) Tumor origin of
endothelial cells in human neuroblastoma. J Clin Oncol 25:
376–383
Brain Tumor Pathol (2012) 29:73–86 85
123

67. Rigolin GM, Fraulini C, Ciccone M et al (2006) Neoplastic
circulating endothelial cells in multiple myeloma with 13q
deletion. Blood 107:2531–2535
68. Bussolati B, Grange C, Sapino A et al (2009) Endothelial cell
differentiation of human breast tumor stem/progenitor cells.
J Cell Mol Med 13:309–319
69. Su M, Feng YS, Yao LQ et al (2008) Plasticity of ovarian cancer
cell SKOV3ip and vasculogenic mimicry in vivo. Int J Gynecol
Cancer 18:476–480
70. Hallani SE, Boisselier B, Peglion F et al (2010) A new alter-
native mechanism in glioblastoma vascularization: tubular vas-
culogenic mimicry. Brain 133:973–982
71. Yue WY, Chen ZP (2005) Does vasculogenic mimicry exist in
astrocytoma? J Histochem Cytochem 53:997–1002
72. Liu X, Zhang Q, Mu Y et al (2011) Clinical significance of
vasculogenic mimicry in human gliomas. J Neurooncol. doi:
10.1007/s11060-011-0578-5
73. Yao XH, Ping YF, Blan XW (2011) Contribution of cancer stem
cells to tumor vasculogenic mimicry. Protein Cell 2:266–272
74. Zhang S, Guo H, Zhang D et al (2006) Microcirculation patterns
in different stages of melanoma growth. Oncol Rep 15:15–20
75. Chang YS, di Tomaso E, McDonald DM et al (2000) Mosaic
blood vessels in tumors: frequency of cancer cells in contact
with flowing blood. Proc Natl Acad Sci USA 97:14608–14613
76. Yao XH, Ping YF, Chen JH et al (2008) Glioblastoma stem cells
produce vascular endothelial growth factor by association with a
G-protein coupled formylpeptide receptor FPR. J Pathol 215:
369–376
77. Folkman J (1971) Tumor angiogenesis: therapeutic implications.
N Engl J Med 285:1182–1186
78. Hanahan D, Folkman J (1996) Patterns and emerging mecha-
nisms of the angiogenic switch during tumorigenesis. Cell 86:
353–364
79. Singh SK, Hawkins C, Clarke ID et al (2004) Identification of
human brain tumour initiating cells. Nature 432:396–401
80. Hendrix MJC, Seftor EA, Gess AR et al (2003) Vasculogenic
mimicry and tumour-cell plasticity: lessons from melanoma.
Nature Rev Cancer 3:411–421
81. Zhao Y, Bao Q, Renner A et al (2011) Cancer stem cells and
angiogenesis. Int J Dev Biol 55:477–482
82. Venere M, Fine HA, Dirks PB et al (2011) Cancer stem cells in
gliomas: identifying and understanding the apex cell in cancer
hierarchy. Glia 59:1148–1154
83. Lathia JD, Gallagher J, Heddleston JM et al (2010) Integrin
alpha 6 regulates glioblastoma stem cells. Cell Stem Cell 6:
421–432
84. Wurmser AE, Nakashima K, Summers RG et al (2004) Cell
fusion-independent differentiation of neural stem cells to the
endothelial lineage. Nature 430:350–356
85. Hovinga KE, Shimizu F, Wang R et al (2010) Inhibition of
Notch signaling in glioblastoma targets cancer stem cells via an
endothelial cell intermediate. Stem Cells 28:1019–1029
86. Jain RK, Duda DG, Willett CG et al (2009) Biomarkers of
response and resistance to antiangiogenic therapy. Nat Rev Clin
Oncol 6:327–338
87. Sorensen AG, Batchelor TT, Zhang WT et al (2009) A ‘‘vascular
normalization index’’ as potential mechanistic biomarker to
predict survival after a single dose of cediranib in recurrent
glioblastoma patients. Cancer Res 69:5296–5300
88. Jubb AM, Harris AL (2010) Biomarkers to predict the clinical
efficacy of bevacizumab in cancer. Lancet Oncol 11:1172–1183
89. Jain RK (2005) Normalization of tumor vasculature: an
emerging concept in antiangiogenic therapy. Science 307:58–62
90. Loges S, Schmidt T, Carmeliet Y (2010) Mechanisms of resis-
tance to anti-angiogenic therapy and development of third-
generation anti-angiogenesis drug candidates. Gene Cancer 1:
12–25
91. Saidi A, Hagerdon M, Allain N et al (2009) Combined targeting
of interleukin-6 and vascular endothelial growth factor potently
inhibits glioma growth and invasiveness. Int J Cancer 125:
1054–1064
92. Chan DA, Kawahara TC, Sutphin PD et al (2009) Tumor vas-
culature is regulated by PHD2-mediated angiogenesis and bone
marrow-derived cell recruitment. Cancer Cell 15:527–538
93. Chan DA, Glaccia AJ (2010) PHD2 in tumor angiogenesis. Br J
Cancer 103:1–5
94. Henze AT, Riedel J, Diem T et al (2010) Prolyl hydroxylase 2
and 3 act in gliomas as protective negative feedback regulators
of hypoxia-inducible factors. Cancer Res 70:357–366
95. Ohtani K, Dimmeler S (2010) Control of cardiovascular dif-
ferentiation by microRNAs. Basic Res Cardiol 106:5–11
96. Wurdinger T, Tannous BA, Saydam O et al (2008) miR-296
regulates growth factor receptor overexpression in angiogenic
endothelial cells. Cancer Cell 14:382–393
97. Lawler S, Chiocca EA (2009) Emerging functions of microR-
NAs in glioblastoma. J Neurooncol 92:297–306
98. Fang L, Deng Z, Shatseut T et al (2011) MicroRNA miR-93
promotes tumor growth and angiogenesis by targeting integrin-
beta8. Oncogene 30:806–821
99. Wang Z, Li Y, Kong D et al (2010) Cross-talk between miRNA
and Notch signaling pathways in tumor development and pro-
gression. Cancer Lett 292:141–148
100. Liu D, Kruegar J, LeNoble F (2011) The role of blood flow and
microRNAs in blood vessel development. Int J Dev Biol 55:
419–429
86 Brain Tumor Pathol (2012) 29:73–86
123