36_01_02

Protooncogenes and Growth Stimulation

The activation of protooncogenes requires mechanisms such as gene
amplification or point mutation of a single allele leading to
constitutive overexpression. In some cases, however, the mechanisms
causing overexpression are still unclear. Protooncogene products
include several growth factor receptors, such as epidermal growth
factor receptor (EGFR), ERBB2, KIT, and MET. Indeed, many growth
factor/receptor systems are expressed by either the lung tumor or
adjacent normal cells, thus providing autocrine or paracrine growth
stimulatory loops (Fig. 31.1_1). For example, overexpression of the
EGFR encoded by the ERBB1 gene is more common in NSCLC than SCLC and
may be related to tumor stage and differentiation. Coexpression of EGFR
and their ligands, especially transforming growth factor-alpha (TGF-
alpha), by lung cancer cells indicates the presence of an autocrine
(self-stimulatory) growth factor loop. Overexpression of EGFR occurred
in 70% of NSCLCs, and overexpression of both EGFR and TGF-alpha
occurred in 38%.[ref: 6] Clinically, the presence of this loop had no
significant impact on overall survival in early-stage NSCLCs and thus
seems to play a role in lung tumor formation rather than tumor
progression. [ref: 6] ERBB2 (HER2/neu) is highly expressed in more than
one-third of NSCLCs, especially adenocarcinomas. Its expression
correlates with a shorter survival in lung adenocarcinomas and may also
be a marker for intrinsic multidrug resistance in NSCLC cell lines.
[ref: 7] KIT and its ligand, stem cell factor (SCF), are both
preferentially expressed in many SCLCs. Activation of this putative
autocrine loop may provide a growth advantage, or it may mediate
chemoattraction. The SCF/KIT signal transduction pathway has been shown
to be associated with Lck, a src-related tyrosine kinase. [ref: 8] MET
and its ligand, hepatocyte growth factor (HGF), are involved in fetal
lung development. Coexpression of this putative loop is observed in
most NSCLCs, and high HGF levels were associated with a poor outcome in
resectable NSCLC patients. [ref: 9] There are several clinical
applications of these growth factor receptor abnormalities that have
occurred in other cancers, which also need to be explored as new
treatments for lung cancer. These include treatment with humanized
monoclonal anti-HER-2/neu receptor antibody Herceptin alone or in
combination with chemotherapy, treatment with monoclonal anti-EGF
receptor antibody combined with radiotherapy, and new drugs that would
inhibit the tyrosine kinase activity of these receptors.
Apart from protooncogene products, other growth stimulatory loops are
found in lung cancer. The best known is that governed by gastrin-
releasing peptide (GRP) and other bombesin-like peptides, together with
their receptors, which participates in lung development and repair as
well as promoting SCLC growth via an autocrine loop. This loop
represents a future therapeutic target because a clinical trial of an
anti-GRP monoclonal antibody showed some antitumor activity in patients
with previously treated SCLC. [ref: 10]
Because downstream effectors are needed to transduce incoming growth
factor/receptor signals to the nucleus, it is not surprising that
cytoplasmic signal transduction cascades are also implicated in
carcinogenesis. For instance, the receptor tyrosine kinases initially
signal the guanosine triphosphate-binding RAS protein. The RAS gene
family (KRAS, HRAS, and NRAS) can be activated by point mutations at
codons 12, 13, or 61, and one member of this family is mutated in
approximately 20% to 30% of NSCLC (particularly adenocarcinomas) but
probably never in SCLC.[ref: 11] KRAS accounts for 90% of the RAS
mutations in lung adenocarcinomas, with approximately 85% of the KRAS
mutations affecting codon 12. Characteristically, approximately 70% of
KRAS mutation are G to T transversions, with the substitution of the
normal glycine (GGT) with either cysteine (TGT) or valine (GTT).
Similar G to T transversions also affect the P53 gene in lung cancer
and represent the type of DNA damage expected from bulky DNA adducts
caused by the polycyclic hydrocarbons and nitrosamines in tobacco
smoke. Further evidence for a causative role for tobacco smoke is the
correlation of KRAS mutations with cigarette consumption. The presence
of KRAS mutations portends a poor prognosis in both early- and late-
stage NSCLCs, [ref: 12] although the data conflict.[ref: 13]
Nonetheless, a metaanalysis of eight studies of 217 (of 881) NSCLC
patients positive for KRAS mutations suggested a negative prognostic
role.[ref: 14] A prospective study has shown that neither
chemosensitivity or survival correlated with KRAS mutation in advanced
lung adenocarcinomas.[ref: 15] New drugs, farnesyltransferase
inhibitors, were developed to specifically kill or inhibit the growth
of tumor cells with RAS mutations. However, these drugs appear to be
active against tumors with and without RAS mutations and are coming
into clinical trials against lung cancer.
A direct downstream effector of RAS is the RAF1 protooncogene
product. Unexpectedly, the experimental growth arrest of SCLC by
activated RAF1 suggests that it has more of a TSG function. [ref: 16]
Although one copy of RAF1 is frequently lost; however, mutations in the
RAF1 gene have not been detected in human lung cancers. Molecules
downstream of RAF1 in the signal transduction pathway, such as MEKK,
MEK, and mitogen-activated protein kinase/extracellular-regulated
kinase (MAPK/ERK), may also be involved occasionally in lung
carcinogenesis.[ref: 3] The PPP2R1B gene, encoding the beta isoform of
protein phosphatase 2A (PP2A), which regulates the RAS/MAPK cascade, is
also infrequently mutated in lung cancers.[ref: 17]
Ultimately, signal transduction cascades result in the activation of
nuclear protooncogene products such as those encoded by the myc family
genes (MYC, MYCN, MYCL). MYC, when heterodimerized with a protein
called MAX, functions as a transcription factor, necessary for normal
cell-cycle progression, differentiation, and programmed cell death. MYC
is most frequently activated via gene amplification or transcriptional
dysregulation in both SCLC and NSCLC, whereas abnormalities of MYCN and
MYCL generally occur in SCLC. Indeed, MYCL was originally isolated from
a SCLC cell line. The myc family gene amplification has been
differentially observed in the major lung cancer subtypes. In SCLC, one
member of the MYC family was amplified in 18% of tumors and 31% of cell
lines, compared to 8% of NSCLC tumors and 20% of cell lines,
respectively. [ref: 11] Amplification appears more frequent in patients
previously treated with chemotherapy, the "variant" subtype of SCLC,
and its presence correlates with adverse survival. In terms of therapy,
all-trans-retinoic acid (RA) treatment inhibited the in vitro growth of
a SCLC cell line overexpressing MYC, a process associated with
increased neuroendocrine differentiation and increased MYCL and
decreased MYC expression. [ref: 18] Lastly, there also have been
reports of MYCL amplification with rearrangement in which MYCL fuses to
the RLF gene causing a chimeric protein.[ref: 19]
Occasional reports have implicated other oncogenes, such as ERBA,
MYB, JUN, and FOS in lung cancer, but data are relatively few.

Tumor Suppressor Genes and Growth Suppression

p53 Pathway

p53 (or TP53) maintains genomic integrity in the face of cellular
stress from DNA damage (for example caused by gamma- and UV
irradiation, carcinogens, or chemotherapy). It functions as a
transcription factor to activate the expression of genes that control
cell-cycle checkpoints (e.g., p21**WAF1/CIP1), apoptosis (BAX), DNA
repair (GADD45), and angiogenesis (thrombospondin). The p53 gene is the
most frequently mutated TSG in human malignancies, and mutations affect
approximately 90% of SCLCs and 50% of NSCLCs. [ref: 20,21] Most
mutations occur in the evolutionarily conserved p53 exons 5 to 8. In
NSCLCs, p53 alterations occurred more frequently in squamous cell
(51.2%) and large cell (53.7%) carcinomas than adenocarcinomas (38.8%).
[ref: 21] p53 mutations correlate with cigarette smoking and most are
of the type of G to T transversions expected from tobacco smoke
carcinogens. More evidence linking smoking damage with p53 mutations is
the finding that a major cigarette smoke carcinogen, benzo(a)pyrene,
selectively forms adducts at the p53 mutational hot spots.[ref: 22] The
types of p53 mutations are varied and include missense, nonsense, and
splicing abnormalities as well as larger deletions. Missense mutations
(the most common type of mutation) often prolong the half-life of the
p53 protein to several hours, leading to increased levels detectable by
immunohistochemistry and thus the use of immunohistochemistry as a
surrogate assay for p53 mutation. [ref: 23] Whether the occurrence of
p53 mutations (detected by either immunohistochemistry or molecular
analysis) in a patient's tumor affect survival is controversial. [ref:
24] Approximately 15% of lung cancer patients develop antibodies to the
p53 protein, raising the possibility that mutant p53 protein
overexpression can lead to a humoral immune response. Although p53
antibodies have been proposed as a marker for early diagnosis [ref: 25]
and chemoresponsiveness,[ref: 26] the development of these antibodies
does not appear to improve prognosis in lung cancer.[ref: 27] There
have been several promising gene therapy clinical trials with objective
response rates of approximately 10% to 15% in which lung cancers are
treated by intratumoral injection [endobronchially, or by computed
tomography (CT)-guided needle injection] with a normal (wild-type) p53
gene using retroviral [ref: 28] or adenoviral vectors.[ref: 29] These
local injections are now being combined with chemo- and radiotherapy to
test if gene therapy increases sensitivity to conventional treatments.
Also, systemic methods (such as with lipid vesicles) of delivering p53
gene therapy to disseminated tumors are being developed. In another
approach, clinical trials are being conducted to immunize patients
against either mutant p53 or RAS proteins occurring in patients' tumors
in attempts to generate a tumor-specific cytotoxic T-cell response.
p53 functions in a biochemical pathway; thus it is reasonable to
consider that other components of this pathway may be mutated in lung
tumors that are wild-type for p53. One of the upstream components is
the kinase that phosphorylates p53 encoded by the ataxia telangiectasia
(ATM) gene. However, this gene has not yet been found to be mutated in
lung cancer. Other components are the MDM2 and p14**ARF genes (see
p16**INK4A, later in this chapter), which regulate the levels of p53
protein, but so far they have not been implicated in lung cancer. Two
proteins homologous to p53--p51 and p73--have been discovered, leading
to the hypothesis that they may be mutated in p53 wild-type tumors.
However, they are only infrequently, if ever, mutated in lung cancers.
Finally, the Li-Fraumeni syndrome of inherited susceptibility to cancer
determined by an inherited germline mutation of p53 may also lead to
increased susceptibility to lung cancer in adults in these pedigrees.

The p16**INK4A-Cyclin D1-Cyclin-Dependent Kinase-4-Retinoblastoma
Protein Pathway

p16**INK4A

The p16**INK4A-cyclin D1-cyclin-dependent kinase-4 (CDK4)-
retinoblastoma (RB) protein pathway is a key cell-cycle regulator at
the G(1)/S phase transition, and one of the components of this pathway
is abnormal in the majority of lung cancers. The activation of CDKs by
cyclins eventually leads to the phosphorylation (inactivation) of the
growth-suppressive RB protein (see Fig. 31.1_1). As p16**INK4A is an
inhibitor of CDKs, especially CDK4 or CDK6 (which phosphorylate and
keep RB in an "inactive" state), its normal role is to positively
regulate RB's growth-controlling function by keeping RB
unphosphorylated. However, if p16**INK4A is inactivated by mutation, RB
remains chronically phosphorylated and thus cannot function to regulate
growth (see Fig. 31.1_1).
The p16**INK4A (also called CDKN2) gene locus on chromosome 9p21 is
frequently abnormal in human malignancies. In lung cancer, p16**INK4A
abnormalities are frequent in NSCLC but rare in SCLC (in which, by
contrast, RB is mutated in 90% of cases). Thus, these two major
histologic lung cancer types have this pathway inactivated by mutation
of one or the other gene, and double mutants in the same tumor are very
rare. A summary of 20 studies showed that p16**INK4A point mutations in
NSCLCs were observed in only 14% of primary tumors. [ref: 3] However,
homozygous deletions or aberrant promoter methylation can also down-
regulate p16**INK4A. Indeed, aberrant methylation of p16**INK4A may be
the most frequent as well as an early preneoplastic event in the
pathogenesis of squamous cell carcinomas. [ref: 4] Taken together,
these mechanisms ultimately result in absent p16**INK4A expression in
approximately 40% of primary NSCLCs, [ref: 3] indicating that this may
be the most common way to inactivate the p16**INK4A-cyclin D1-CDK4-RB
pathway in NSCLC. Because of the frequency of the abnormalities,
p16**INK4A is an attractive clinical trials candidate for replacement
gene therapy or induction of re-expression with antimethylation drug
therapy.
Complicating matters is the discovery of an alternative reading frame
at the p16**INK4A locus, p14**ARF, which encodes a protein that binds
p53/MDM2 complex, leading to p53 protein stabilization. If p14**ARF is
missing because of mutations in the p16 locus, p53 is less stable and
its function diminished. It thus emerges that inactivation of the p53
pathway may also be triggered by abnormalities of the p16**INK4A gene
locus. Another CDK inhibitor gene, p15**INK4B is situated close to
p16**INK4A and can be co-deleted with p16**INK4A in NSCLC. However, it
appears that the majority of lung cancer abnormalities focus on
p16**INK4A and not on p15**INK4B.

Cyclin D1 and Cyclin-Dependent Kinase-4

Because cyclin D1/CDK4 complex inhibits RB activity by stimulating its
phosphorylation, cyclin D1 or CDK4 overexpression is an alternative way
to disrupt this pathway. Immunohistochemically, cyclin D1 is
overexpressed in 12% to 47% of primary NSCLCs and, in some cases,
cyclin D1 overexpression is associated with a poor prognosis.[ref: 30]
How this overexpression occurs in lung cancer is unknown.

Retinoblastoma Protein

The RB gene (RB), located at chromosomal region 13q14, encodes a
growth-suppressive nuclear phosphoprotein. When active (i.e.,
hypophosphorylated), RB binds and inactivates proteins such as
transcription factor E2F-1, which is essential for G(1)/S transition of
the cell cycle (see Fig. 31.1_1). RB mutations (truncation by
deletion, nonsense mutation, or splicing abnormalities), together with
loss of the wild-type RB allele, have been demonstrated in lung
cancers, particularly SCLC. The RB protein is absent or structurally
abnormal in more than 90% of SCLCs and 15% to 30% of NSCLCs. [ref: 31]
Absent RB expression may be associated with poor prognosis in NSCLCs,
although this is not a consistent finding. The relatively low frequency
of RB abnormalities in NSCLC is consistent with the frequent disruption
of the p16**INK4A-cyclin D1-CDK4-RB pathway in these histologic types.
Essentially, lung cancers can be characterized as having either RB
mutation (mostly SCLC) or p16**INK4A inactivation (mostly NSCLC). Gene
therapy with replacement of RB function has been considered, but
because of the large size of the RB coding region and the necessity for
systemic delivery in, for example, typical widely metastatic SCLCs,
such therapy has not been actively pursued. Mutations of two other RB
related genes, p107 and RB2/p130, also have been implicated in lung
cancer. In the case of RB2, loss of protein expression and restoration
of growth control by genetically re-introducing a normal copy of RB2
have been demonstrated. [ref: 32] Finally, retinoblastoma patients or
their relatives who carry a mutant RB in the germline have an excess
risk of developing small cell lung cancer if they survive into adult
life. [ref: 33]

PTEN, FHIT, RAR-beta, Putative 3p Tumor Suppressor Genes, and Other
Tumor and Growth-Suppressive Gene Sites in Lung Cancer Pathogenesis

Besides the p53, p16**INK4A, RB, and loci, cytogenetic and allelotyping
studies show nonrandom, hemiallelic loss at many other chromosome
regions in lung cancer. Such tumor-specific somatically acquired loss
of heterozygosity is a hallmark feature of traditional TSG
inactivation. In other words, the consistent identification of multiple
sites of loss of heterozygosity at various chromosomal regions suggests
the existence of underlying TSGs in these regions. Usually the
remaining allele is silenced by point or small mutations, epigenetic
hypermethylation of the promoter region or, less frequently, by a
larger deletion. These sites of allele loss have been defined at more
than 30 regions dispersed on 21 different chromosomal arms, although
the molecular targets of most of these sites is not known. [ref: 34]
Although several of these chromosomal arms contain known TSGs,
including VHL (3p25), APC (5q21), WT1 (11p13), DCC (18q21), and NF2
(22q12), these genes are not known to be mutated in lung cancer. The
TSG PTEN, which encodes a phosphatase, is located at chromosome region
10q23, another common area of allele loss in lung cancer. However, PTEN
is mutated in only a subset of lung cancers. [ref: 35] Frequent loss
(60% to 80%) of several 4p and 4q regions were found in thoracic
malignancies (particularly SCLC and mesotheliomas); however, the genes
involved are not yet known. [ref: 36]
Among these chromosomal locations, chromosome 3p allele loss
(occurring at more than four different 3p regions) stands out as a very
frequent and early event in lung cancer pathogenesis. 3p loss occurs in
more than 90% of SCLCs and more than 80% of NSCLCs. In addition, it
appears to be the earliest genetic change found in lung cancer
development, occurring at great frequency in patches of normal
epithelium accompanying lung cancer or in smokers, as well as in sites
of hyperplasia, dysplasia, and carcinoma in situ of respiratory
epithelium (see Molecular Changes in Preneoplasia, later in this
chapter). Multiple distinct 3p regions have been identified by
allelotyping, including 3p25-26, 3p21.3-22, 3p14, and 3p12.
Furthermore, homozygous deletions are found in several lung cancer cell
lines (at several 3p21.3 sites, as well as at 3p12 and 3p14.2). Several
candidate TSGs have been identified in an approximately 600-kilobase
3p21.3 region homozygously deleted in three SCLCs, and another 800-
kilobase deletion region at 3p21 also has been described. [ref: 37,38]
The FHIT (fragile histidine triad) gene is found in the 3p14.2
homozygous deletion region. Forty percent to 80% of lung cancer cells
express abnormal FHIT messenger RNA transcripts but paradoxically
almost always also express wild-type FHIT transcripts. [ref: 39]
Regardless of these molecular complexities, the FHIT protein is absent
in many lung cancers, particularly in the squamous cell type (87%)
compared to adenocarcinoma (57%) and may also be lost in some
preneoplastic lesions. The loss of FHIT protein expression is also
strongly associated with smoking.[ref: 40,41] Functionally, FHIT may be
involved in the regulation of apoptosis and in cell-cycle control.[ref:
42] Transfection of wild-type FHIT into lung cancers induces apoptosis
and blocks tumor formation in vivo in mouse models. [ref: 43,44]
Because of the occurrence of FHIT abnormalities early in lung cancer
pathogenesis (e.g., in preneoplastic stages), it is possible to
consider delivering FHIT gene therapy to airways containing multiple
preneoplastic lesions by using aerosols.
TGF-beta1 is a potent inhibitor of proliferation of most epithelial
and hematopoietic cells, and its signal is mediated through TGF-beta
receptors and subsequently SMAD proteins. In SCLC cells, down-
regulation of the type II receptor (TGFbeta-RII) located in chromosome
region 3p, has been shown to correlate with the resistance to growth
inhibition by TGF-beta1. [ref: 45] However, the TGFbeta-RII or SMAD
family genes are rarely mutated in lung cancer.
There is considerable evidence of dysfunction of retinoic acid
receptor beta (RAR-beta), located in chromosome region 3p24, in lung
cancers, leading to resistance of lung cancer cells to retinoids and
making it an excellent candidate 3p TSG. [ref: 46] Although initial
studies did not find RAR-beta mutations, more recent studies have shown
loss of expression of RAR-beta protein in approximately 50% of
clinically overt lung cancers. [ref: 47] It is quite likely that this
loss of expression without genomic changes occurs because of
methylation of the promoter region. Because of the widespread testing
of retinoids as chemoprevention agents, it will be important to
characterize the timing of loss of RAR-beta function in lung cancer
preneoplasia. It is possible that loss of expression of RAR-beta may
occur at such an early stage that chemoprevention with retinoids cannot
succeed.