Molecular Events Surrounding the Angiogenic Switch of Lung Cancer

Multicellular organisms such as humans require a complex
vascular system to supply cells with oxygen and glucose and
to dispose of waste products. During embryonic development
and at the main phases of organ growth, the vascular
system evolves with the growing organism. However, in
adulthood, the vascular system is quiescent. A well-controlled
angiogenic response (production of blood vessels from existing
blood vessels) to specific transient cues may occur. Other
than the changes that occur in the female reproductive system
or in wound healing, endothelial cells rarely proliferate in
adults. This tight control is the result of a continuous balance
between angiogenic signaling and inhibition. Several proangiogenic
and antiangiogenic molecules have been identified,
many of which are potentially active in creating this balance
(Table 8.1). The exceptions to this balance include pathologic
conditions such as wound healing, inflammation, and cancer,
in which the formation of new blood vessels is vital. 1 The
abnormal state of cancer, in which proangiogenic signaling
does not abate, is reminiscent of what happens in wounds;
however, it occurs in a perpetual manner. In this sense, cancer
can be regarded as a “wound that never heals.” 2
Blood supply is delivered by a highly organized conduit
network that is spread out, designed to reach most cells. The
exchange of nutrients and waste molecules between the blood
and cells occurs through capillary walls, the thinnest vessels,
those that connect the arterial tree to the venous tree. Vascular
capillaries are formed from endothelial cells, which create the
tubal conduit for blood, surrounded by a basement membrane.
Pericytes (mural vascular cells) are embedded in the basement
membrane. These cells provide physical support for the blood
vessel and provide communication ports with endothelial cells,
thus controlling capillary function. Unlike vascular smooth
muscle cells, which are found in the media layer of large blood
vessels, most pericytes are in direct contact with endothelial cells;
cell–cell communications take place between these cell types. 3
As a malignant growth exceeds the size of a few hundred
microns, nutrient diffusion becomes a growth-limiting factor.
Hypoxia and various cancer-specific genetic abnormalities
drive the secretion of proangiogenic factors and suppression
of antiangiogenic factors. In this manner, the tumor microenvironment
becomes proangiogenic. New blood vessels are
formed, and existing blood vessels are modified to provide a
better blood supply to the tumor. The production of blood
vessels from existing blood vessels is called angiogenesis , whereas
production of de novo blood vessels is termed vasculogenesis .
Both of these processes are controlled by the counteracting
effects of proangiogenic and antiangiogenic factors. The tipping
of the balance toward a proangiogenic state is called the
angiogenic switch . This switch seems to be essential to cancer
progression. Accordingly, tumoral angiogenesis was suggested
as a therapeutic target more than 35 years ago. 4 Only in recent
years has this idea entered clinical practice; antiangiogenesis
is used to treat colon, kidney, breast, and other cancers. The
clinical importance of angiogenesis inhibition in the treatment
of non–small cell lung cancer (NSCLC) was demonstrated by
an improved outcome after treatment with an antiangiogenic
agent and classic chemotherapy. 5,6
In this chapter, we will discuss the importance of the angiogenic
switch in cancer, as demonstrated in studies of whole
organs and the major cell types and mechanisms that supply
nutrients to cancer. In vitro studies and the major molecular
players will be presented. The role of vasculogenesis in tumor
perfusion is controversial and sometimes difficult to distinguish
from that of angiogenesis. Thus, both are discussed in this
chapter. We conclude with the available data on the angiogenic
switch and alternative modes of lung cancer vascularization,
and a note about the clinical implications of this information.
ANGIOGENIC SWITCH IN CANCER:
WHOLE-ORGAN STUDIES
A milestone of molecular biology and cancer research is the
development of specific gene silencing and upregulation techniques
in in vivo mouse models. Several mouse models of the
angiogenic switch are described in the succeeding discussion.
Mouse Models The RIP-Tag mouse model was created by
genetically manipulating mouse pancreatic -cells to express the
SV40 large T antigen; a hyperproliferative stage arises in about
half the pancreatic islets, and in a minority of those, pancreatic
islet cell carcinoma develops. The induction of angiogenic activity
demarcates an angiogenic stage, which is found in islets that
have progressed past the proliferative stage and apparently precedes
carcinoma. 7 The isolation and examination of these islets
revealed that an angiogenic phenotype is not a direct outcome
of oncogene expression, nor is it the inevitable effect of a proliferative
mass that requires an increased blood supply. The RIPTag
model demonstrated discrete stages of tumor progression in
which additional genetic manipulation can be used to determine
the roles of various molecules in each phase. Thus, this model
was used to evaluate the roles of vascular endothelial growth factor
(VEGF) in the angiogenic switch, 8 matrix metalloproteases
(MMP9) in the release of VEGF from the extracellular matrix
(ECM), 9 and VEGF receptor 2 (VEGFR-2) expression on vascular
endothelial cells 10 (see succeeding discussion).
Tumor fragments transplanted onto the irises of rabbits
were used in a classic model that demonstrated the importance
of the angiogenic switch. 11 In some cases, the tumor fragments
induced angiogenesis and progressed; in others, they remained
clinically dormant. Active proliferation and apoptosis were
found at the cellular level in clinically dormant tumors. Similar
apparent dormancy has also been described in other conditions
in which angiogenesis was inhibited. 12
The angiogenic switch is required not only for the progression
of primary tumors but also for the growth of metastatic deposits.
In a mouse model of metastatic disease, removal of the primary
tumor induced rapid growth in previously dormant metastatic deposits.
An angiogenesis inhibitor produced by the primary tumor
was later identified. The rapid growth of the micrometastasis
was accompanied by angiogenesis induction and a concomitant
apoptosis reduction, with no change in the proliferative rate. 12
Both activators and inhibitors of angiogenesis are produced in
intact organisms, and the balance between them determines the
vascularity of the tumor and its metastasis and progression.
Angiogenesis Evaluation in Human Cancer In
clinical studies, the most useful method for evaluating tumor
vascularity is counting microvessel density (MVD) in tumor
sections by immunohistochemically staining for one of several
molecular markers of endothelial cells, namely factor VIII, von
Willebrand factor, CD31, or CD34. Under low magnification,
areas of high MVD can be observed, usually at the tumor periphery.
The actual MVD is determined in those areas by counting
the number of capillaries per high-power field. Various
protocols for counting or subjective visual evaluation exist. 13
MVD has been correlated with imaging-determined tumor perfusion
14 and clinical outcome in many tumor types. However,
many vessels visualized by immunohistochemical staining may
not be functional. CD105 staining was recently suggested to be
specific for active blood vessels as opposed to general endothelial
markers that stain also nonfunctional vessels. 15 Proliferating
endothelial cells can also be assessed by immunohistochemical
analysis, because they are a more reliable marker of ongoing angiogenesis
than MVD. Double-staining for endothelial- specific
proteins and Ki-67 is used to estimate the proportion of proliferating
cells; this method was better correlated with stage than
counting MVD in colorectal cancer. 16 An important caveat of
these methods is the high variability within tumors 17 ; thus,
sampling errors may be significant. Although highly useful and
clinically prognostic in many studies (see later), these methods
provide no information about tumor perfusion, which is the
biologically relevant end point of the angiogenic switch. For
this, other approaches must be used.
An indirect method of evaluating tumor angiogenesis is the
assessment of tumor hypoxia. Carbonic anhydrase IX, a transcriptional
target of hypoxia-inducible factor 1 (HIF-1), may be a surrogate
marker of tumor hypoxia. High levels are found in hypoxic
areas in several cancer types, including NSCLC, and its expression
is associated with a poor prognosis. 18 Pimonidazole is an exogenous
marker of hypoxia 19 that can be used in immunohistochemical
analyses of biopsy samples 20 ; however, this method is not commonly
used, because it requires pimonidazole to be intravenously
infused to patients prior to biopsy. Interestingly, tumor cell necrosis,
a plausible indicator of tumor hypoxia, has not been reported
to be a prognostic factor in lung cancer. Nuclear medicine allows
molecular imaging, including the ability to detect hypoxia. The
positron emission tomography tracer 18F-misonidazole 21 and
other tracers are being assessed as prognostic or predictive markers
in several cancer types. However, such tools require further validation
and are not yet available in most cancer centers.
Major inducers and inhibitors of angiogenesis can be used as
surrogate markers for angiogenesis in cancer. The levels of VEGF,
its receptors, and various endogenous facilitators and inhibitors of
angiogenesis can be evaluated in patients’ tumor tissues or serum.
Importantly, such studies can be performed in large-scale setups
and may be useful clinically. Alternatively, immunohistochemical
analyses of protein expression in tumors can be performed, but
these have the same potential sampling bias of MVD studies.
Additional parameters of angiogenic activity that can be assessed
from a blood sample include circulating endothelial cells
(CECs) and circulating endothelial progenitor cells ( circulating
EPC; CEPs). CECs were first reported more than 30 years ago
by Hladovec and Rossmann. 22,23 Mature CECs probably originate
from cells shed from vessel walls. Some of such cells possess
progenitor characteristics and are referred to as CEPs; they
are thought to originate from the bone marrow. Both CECs and
CEPs may be useful surrogate markers for angiogenic activity and
for tumors’ response to antiangiogenic treatment. 24 However,
the assessment of CECs or CEPs in patients’ blood is technically
challenging, and no consensus exists about assessment methods
or even CEPs’ significance (see discussion later).
Tumor perfusion can be assessed in vivo using various imaging
studies. Most of these methods are investigational, although
promising. Microscopic bubbles can be used as contrast material
in sonography studies to demonstrate blood flow in vivo.
Computed tomography and magnetic resonance imaging using
intravenous contrast material can also be used to measure blood
flow and volume. Positron emission tomography nuclear imaging
using 11 C- or 15 O-marked carbon monoxide can be used for the
same purpose. Molecular imaging, in which the contrast material
is conjugated to a molecule that binds to a cancer endothelialspecific
epitope, is currently being evaluated. 25
Angiogenic Switch in Human Cancer Stepwise progression
is evident in several cancer types, and discrete stages
can be differentiated in pathologic specimens. For example,
breast cancer is preceded by carcinoma in situ, in which vessel
density is correlated with several poor prognostic factors. 26
Additional indicators of angiogenic activity, such as messenger
RNA (mRNA) expression levels of VEGF and its receptors,
were upregulated in breast in situ carcinoma, similar to what
was found in invasive breast cancer. 27 Cervical squamous cell
intraepithelial neoplasia is the precursor of cervical carcinoma
and is graded according to the proportion of dysplasia. Vessel
density in the stroma below the basement membrane of the
dysplastic epithelium was correlated with the grade of epithelial
dysplasia. 28 Evaluation of dysplastic bronchial epithelium,
lung premalignant lesions, revealed increased MVD and increased
levels of VEGF levels compared with normal controls.
A characteristic pattern of VEGFR and VEGF isoform expression
comparable to invasive lung cancer was found. Apparently
normal lungs of heavy smokers harbored enhanced VEGF
mRNA levels. 29 Abnormal microvasculature structure was
described to appear near dysplastic squamous bronchial epithelium.
30 The results of these studies suggest that, similar to
what was demonstrated in mouse models, human cancer must
acquire a vascular supply in order to progress. At least in some
types of cancer, including lung cancer, the angiogenic switch
occurs prior to the invasive phase of cancer progression.
Angiogenic Signaling in Response to Hypoperfusion
Oncogene activation and tumor suppressor gene
inactivation can activate the angiogenesis switch (see later).
However, hypoperfusion is an alternative, and conceptually
antagonizing, source of proangiogenic signals. Tumors that
outgrow their blood supply experience reduced oxygen and
glucose levels. Hypoxic regions are commonly found in several
cancer types. These regions may indicate highly dysregulated
cell growth; they are associated with a poor prognosis. 31,32
Hypoxia may select for more aggressive tumor cells 33 or activate
signaling pathways that lead to increased invasion and
metastasis. For example, in NSCLC cells, hypoxia induced in
a HIF-1 -dependent manner CXCR4 expression. 34 CXCR4
is a cytokine receptor involved in invasion and metastasis. 35
A direct consequence of hypoxia and hypoglycemia is HIFdependent
induction of VEGF expression in malignant cells. 36
VEGF activates an angiogenic response, which results in new
blood vessels, improved tumor vascularization, and potentially
relief of hypoxia and hypoglycemia. VEGF induction by hypoxia
or hypoglycemia involves an apparently normal feedback
mechanism, in which reduced perfusion activates a corrective
mechanism. Normally, once perfusion has improved, VEGF
secretion is reduced. 36 Hypoperfusion-induced signaling is
not perpetual, unlike the angiogenic switch. However, newly
formed blood vessels of tumors are not as functional as mature
vessels of normal tissues: They are only partially covered by
pericytes and are highly permeable, torturous, and chaotic. 25
The inefficiency of the newly formed tumor blood supply
causes the hypoxia-induced signaling to prevail and paradoxically
contributes to cancerous growth.