ROLE OF ANGIOGENIC SIGNALS IN LUNG CANCER

Prognostic Markers of Angiogenesis MVD is commonly
used as an immunohistochemically derived indication
of angiogenesis in tumors. MVD was found to be correlated
with prognosis in NSCLC in several studies. 207–215 However,
MVD has not been found to be correlated with prognosis in
other studies. 205,216,217 In one of the larger studies that reported
no correlation of MVD with prognosis, 216 MVD was
assessed in tissue microarrays, which are comprised of small
cores taken from each tumor. The manner in which tumor
areas were chosen for the tissue microarray probably differed
from those normally chosen for MVD scoring (see previous
discussion). This difference might explain the discordant results
regarding the prognostic value of MVD. In another study
of 106 NSCLC patients, 205 MVD was correlated with clinical
disease stage, but not with survival.
MVD, as assessed immunohistochemically with CD105,
is suggested to mark only active blood vessels. MVD assessed
by CD105 was more strongly correlated with VEGF tumor
levels than was MVD assessed by CD34 (panendothelial
marker). MVD by CD105 was also more strongly correlated
with clinical outcome in a study of 236 NSCLC patients. 15
CD34 apparently marks all blood vessels, including vessels
that are not functional. Overall, the evidence suggests that angiogenic
activity as assessed by tumor MVD is prognostically
important in lung cancer. Fine tuning and standardization of
the MVD measurement methods is needed.
Tumor expression of VEGF, as evaluated by immunohistochemical
analysis, was found to have a negative prognostic value
in NSCLC patients, 216,218–221 although a lack of correlation has
also been reported 103 (see Table 8.2 for a summary of the prognostic
angiogenic factors in lung cancer). Regarding a different
type of lung cancer, small cell lung cancer (SCLC), tumor VEGF
expression was also found here to be a poor prognostic factor in
a study of 75 surgically resected patients. 222 In another study of
SCLC, serum levels of VEGF of 69 patients were found to be associated
with poor outcome and poor survival, 223 attesting to the
importance of angiogenesis also for this type of lung cancer.
Levels of VEGF (and bFGF) in NSCLC tissue extracts, including
both tumor and stromal components, have prognostic
implications. 224 Interestingly, VEGF-C expression in stromal
cells of NSCLC was found to have a positive prognostic value. 216
Given the important role of the stroma in tumor progression
and regression, this finding entails further investigation.
VEGF mRNA levels were investigated in NSCLC tissue
specimens and were found to be associated with a grim prognosis.
225 An in-depth study of VEGF splice variants in tumor
tissues revealed that the 189 isoform was associated with a poor
prognosis in resected NSCLC patients. 133 Other VEGF splice
variants were not associated with prognosis in that cohort of
patients, suggesting that VEGF levels in tissues studies should
differentiate between the various VEGF splice variants.
Serum VEGF levels have been found to be prognostic in
some studies 227,228 but not others. 229,230 E4599, a study of
878 NSCLC patients with advanced disease, was retrospectively
analyzed to determine the prognostic and predictive
significance of VEGF plasma levels. All patients had received
carboplatin and paclitaxel, and some were randomly assigned
to bevacizumab. VEGF plasma levels were predictive of response
to bevacizumab but were not prognostic of survival. 191
Another study evaluated 462 early stage lung cancer patients
for polymorphisms in the VEGF gene, and reported them to
be prognostically important. The examined polymorphisms
are expected to correlate with reduced VEGF serum levels, 231
but actual serum level evaluations were not available. It can
be speculated that VEGF levels within a cancer are important
whereas serum levels are modulated by additional mechanisms,
irrelevant for tumor biology.
VEGF receptor levels have also been evaluated. One large
study evaluated VEGF and VEGFR isoform levels by immunohistochemical
analysis in 335 NSCLC patients (stages I to
IIIa). 216 These levels were evaluated separately in tumor and
adjacent stroma tissue. Although several factors were found to
be associated with poor prognosis in univariate analysis, this
association was found only for VEGFR-3 tumor expression in
multivariate analysis. 216 In another study of NSCLC patients,
real-time RT-PCR analysis revealed that VEGFR-2 mRNA
blood levels were correlated with response to treatments and
with clinical outcome. 226
Tumor expression of Ang-2, a context-dependent modulator
of angiogenesis, was found to have a negative effect on
survival in a study of 236 resectable NSCLC patients. This
effect was evident only among tumors that expressed high levels
of VEGF and was not significant in those with low VEGF
levels. 137 This result is consistent with the known molecular
mechanisms involved; Ang-2 positively controls angiogenesis
only when VEGF is abundant (see previous discussion).
bFGF is considered an important angiogenesis inducer
and prognostic factor in tumors: Its expression in NSCLC tissue
is associated with a poor prognosis. However, data on the
prognostic value of bFGF blood levels vary. 229
Interleukin-8 (IL-8, also called CXCL8) was shown to be
highly expressed in bronchiogenic lung carcinomas. Its expression
in lung cancer cells was significantly induced by coculturing
with macrophages. Tumor-infiltrating macrophages and
IL-8/CXCL8 mRNA levels in NSCLC tumors were found to
be correlated with microvascular density and patient survival. 82
A study of specific neutralizing antibodies demonstrated that
IL-8/CXCL8 was responsible for endothelial cell migration
and for the angiogenic response elicited by NSCLC tumor extracts
in a corneal neovascularization assay. 232
PlGF, a VEGFR-1 ligand was also found to be correlated
with stage in NSCLC. PlGF protein levels (as measured by
immunostaining) and gene transcript levels (as measured by
RT-PCR) were correlated with poor prognosis. 233 Because
VEGFR-1 is a negative regulator of the VEGF pathway, PlGF’s
correlation with poor prognosis might be through a different
pathway. Possibly related is an in vitro study of NSCLC cell
lines, where PlGF was found to influence cell motility, through
ROCK1, a major regulator of the cytoskeleton. 234
Aberrant expression of p53, commonly used as an indicator
of mutant p53, was correlated with increased mRNA
levels of VEGF, IL-8/CXCL8, and MVD and poor prognosis
in 65 NSCLC patients. 235 A larger study reported poor prognosis
for patients with an aberrant p53 expression, after tumor
resection with no further treatment. However, such patients
gained significant benefit when given adjuvant chemotherapy.
236 It can be speculated that tumors with an aberrant p53
have higher levels of angiogenic factors in their microenvironment;
chemotherapy agents might thus be better delivered to
micrometastasis disease.
A transcriptional target of HIF-1, BNIP3, was evaluated
in 105 NSCLC patients. BNIP3 is a pro-apoptotic mitochondrial
protein that can activate necrosis-like cell death and may
be important in the necrotic response to hypoxia in tumors.
BNIP3 levels were found to be highly correlated with poor
prognosis in patients with resectable disease. 237 HIF-1 protein
levels, as assessed by immunohistochemical analysis, were
correlated with poor prognosis in 172 NSCLC tumors (stages
I to IIIa). 238 On the other hand, HIF-1 mRNA levels, as
measured by RT-PCR, were not correlated with clinical outcome
in a study of 54 NSCLC patients. 239 Therefore, HIF-1
regulation at a posttranslational level, by VHL-mediated degradation,
might be its dominant regulatory mechanism in
NSCLC. Regarding another surrogate marker of hypoxia, carbonic
anhydrase IX was also found to be associated with poor
prognosis in early NSCLC. 18
A more direct method of quantifying hypoxia in NSCLC is
with the use of nuclear medicine tracers. An 18F- misonidazole
evaluation of 14 NSCLC patients predicted recurrence after
curative radiotherapy in those patients that had evidence of
significant tumor hypoxia. 240 Additional hypoxia tracers exist,
such as copper-60 derivate ([60]Cu-ATSM), which also has
prognostic value in NSCLC. 241
Trx-1 is a small redox protein that modulates the activity
of various enzymes, including DNA binding and transactivation
by transcription factors. Trx-1 was shown to increase
the protein level and activity of HIF-1 in cancer cells. 242
Interestingly, Trx-1 levels were found to be correlated with
lymph node invasion and a poor prognosis in a study of 102
early stage NSCLC patients. 243 Although possibly related to
several other pathways, Trx increased levels could have activated
the HIF pathway and thus disease progression.
HGF-Met signaling is also implicated in the progression
of lung cancer. HGF levels, as quantified in lung tumor
tissue, were associated with a poor prognosis. 244 Recently,
somatic mutations were discovered in the c-Met receptor of
NSCLC specimens. The mutations were concentrated in a
region close to a splice junction and led to an alternatively
spliced protein. This protein was defective in Cbl-mediated
degradation, demonstrating prolonged ligand-induced activation.
245 The HGF-Met pathway affects multiple cellular
pathways, including tumor angiogenesis. 92 In many cases,
HGF production is enhanced in tumor-associated fibroblasts,
demonstrating another tumor-stroma interaction that
promotes cancer progression.
Blood levels of CEPs were investigated in 53 patients with
various stages and histologic subtypes of NSCLC. CEPs were
detected in this study by a flow cytometric CD34 VEGFR2
analysis. In the multivariate analysis, the CEP count was a statistically
significant prognostic marker, whereas surprisingly,
disease stage had no prognostic value. 226 The blood mRNA
transcript levels of several possible CEPs’ molecular markers
(CD34, CD-133, and VE-cadherin) were not correlated with
prognosis or with the CEP count, whereas high mRNA levels of
VEGFR2 were correlated with a lack of response to therapy. 226
Another small study reported the feasibility of magnetic bead
separation for CEP identification in lung cancer. 246 The lack
of consensus in the field about the importance of CEP and the
lack of reliable methods for CEP detection hinder our understanding
of their role in lung cancer.
Immune system cell infiltration affects angiogenesis and
prognosis in NSCLC. In bronchoalveolar carcinoma pathologic
specimens, neutrophil accumulation in the alveolar
lumen was associated with a poor prognosis. Neutrophil count
was correlated with IL-8/CXCL8 levels in the BAL fluid of
these patients. The origin of the secreted IL-8/CXCL8 seemed
to be the cancer cells. 72 Thus, IL-8/CXCL8 can function as
an angiogenic factor either by directly activating endothelial
cells 247 or indirectly by recruiting immune system cells to
tumor sites. IL-8/CXCL8 also has a direct mitogenic effect on
lung cancer cells. 248
Nonangiogenesis Variant of NSCLC In contrast to
the data implicating vascular angiogenesis as a critical requirement
to tumor growth, notable exceptions have been observed.
In NSCLC, a nonangiogenic histologic pattern with a higher
incidence of lymph node metastasis and a poorer prognosis
was described. 249,250 Histologically, it had a nondestructive,
alveolar pattern of malignant cell spread. A detailed analysis
of the tumor vessels suggested that the tumor had co-opted
existing blood vessels and possibly even lymphatic vessels. 251
Although suggestive to be a subtype of bronchoalveolar carcinoma,
analysis of squamous cell lung cancers revealed also
a subgroup of low-vascularity squamous cell tumors. 252 These
squamous cell carcinomas were characterized by a high proliferation
rate, low apoptotic activity, high VEGF expression,
and low bFGF expression, and were associated with a poor
prognosis. In another study, c-ErbB2 overexpression was correlated
with a poor prognosis in a low-angiogenesis subgroup
of NSCLC tumors. 253 These findings indicate that lung cancer
can grow without neoangiogenesis. A lack of new vessel formation
does not necessarily hinder tumor progression, although it
dictates a specific growth pattern. More data is required about
the molecular mechanisms involved in the progression of lowangiogenic
tumors and about potential therapeutic targets in
this subgroup.
Lymphangiogenic Switch in Lung Cancer Most research
on cancerous vascularization has focused on angiogenesis
and blood vessels’ connections to the tumor. The tumor’s
vascular supply is essential to its growth; thus, angiogenesis
induction is a critical turning point. However, the lymphatic
network is also being recognized as important in later stages
of lung cancer progression. Lymphangiogenesis is correlated
with lymph node metastasis and prognosis. 251 VEGF-C controls
lymphangiogenesis and lymph node metastasis in mouse
models of lung cancer. 254 Accordingly, VEGF-C protein levels
in the serum of 116 NSCLC patients were correlated with the
risk of lymph node metastasis. 255 Targeting in parallel VEGF-A
and VEGF-C might be pharmacologically applicable and
could be therapeutically advantageous.
Mouse Models of Lung Cancer Angiogenic Switch
K- Ras is activated by a somatic mutation in 20% to 30% of
NSCLCs. 256 A correlation was found between K- Ras mutation
and high VEGF expression in a group of 181 NSCLC tumors. 257
Another report did not find a correlation between K- Ras mutations
and vascularity or VEGF levels. 258 Mice manipulated to
express active K- Ras in lung epithelium (KRas LA1 mice) developed
atypical adenomatous hyperplasia and adenomas, which
progressed to adenocarcinoma. 259,260 The oncogenic effect of
K- Ras in this model was shown to be Rac1 dependent. 261 PI3K
activation and increased phophatidyleinositole 3–5 triphosphate
levels probably also mediated the oncogenic effect of K- Ras , as
PTEN deletion accelerated K- Ras –induced lung cancer formation.
262 c-Met-HGF signaling contributes to tumor progression
in this model, at least partly through enhanced angiogenesis. 263
Accordingly, an inhibitor of c-Met led to reduced VEGF production
and enhanced thrombospondin-1 expression in lung
cancer cells. 264 In addition, Ras signaling activated expression
of CXCR2 ligands, causing accumulation of inflammatory cells
and vascular endothelial cells in the premalignant lesions of the
KRas LA1 mice. Blockage of CXCR2 signaling prevented the
progression of these lung lesions and caused apoptosis of vascular
endothelial cells within them. 265 CXCR2 inhibition had
a tumor-inhibitory effect in a microenvironment-dependent
manner. This suggests that KRas activation in lung cancer is
proangiogenic and tumorigenic through recruitment of inflammatory
cells and vascular endothelial cells.
A mouse model of lung cancer involving the targeting of
either a wild-type c-Raf kinase or a constitutively active c-Raf
kinase to lung epithelial cells was reported. 266 After a relative
long latency period, isolated foci of lung adenomas developed,
with no evidence of invasion or metastasis. 266 Additional genetic
events are assumed to take place in some of the primed
cells to explain these observations, making this model suitable
for studies of lung cancer progression. In a study using the
c-Raf mouse lung cancer model, disruption of intercellular
adhesions through the downregulation of E-cadherin promoted
an angiogenic switch. 267 E-cadherin disruption led to
nuclear localization of -catenin and secondary upregulation
of VEGF-A, VEGF-C, and VEGFR-3. Phenotypically, this
resulted in a marked increase in MVD, and evidence of increased
permeability, typically seen when the VEGF pathway is
activated. Lymphatic vessel density was also increased, accompanied
by micrometastasis in draining lymph nodes. 267 Loss
of E- cadherin–mediated cell–cell contacts might be a major
regulator of the angiogenic switch in lung cancer.