Larry G. Maxwell
Cancer is a complex disease that arises because of genetic and epigenetic alterations that disrupt cellular proliferation, senescence, and death (Fig. 1.1). The alterations that underlie the development of cancers have a diverse etiology, and loss of DNA repair mechanisms often plays a role in allowing mutations to accumulate. Specific molecular changes that cause a normal cell to become malignant have been identified, but their spectrum varies considerably between cancer types.
The malignant phenotype is also characterized by its ability to invade surrounding tissues and metastasize. The development of a cancer elicits a considerable molecular response in the local microenvironment that is characterized by recruitment of stromal elements such as new blood vessels and by an active immunological response. These secondary events play a critical role in the evolution and progression of cancers. Although the molecular pathogenesis of gynecologic cancers has been only partially elucidated, advances in the understanding of these diseases are providing the opportunity for improvements in diagnosis, treatment, and prevention.
The initial sections of this chapter will outline what is known regarding the basic molecular mechanisms involved in the development of cancers and the evolution of the malignant phenotype. The molecular alterations characteristic of gynecologic cancers will be outlined in the latter sections.
The number of cells in normal tissues is tightly regulated by a balance between cellular proliferation and death. The final common pathway for cell division involves distinct molecular switches that control cell cycle progression from G1 to the S phase of DNA synthesis. These include the retinoblastoma (Rb) and E2F proteins and their various regulatory cyclins, cyclindependent kinases (cdks), and cdk inhibitors. Likewise, the events that facilitate progression from G2 to mitosis and cell division are regulated by other cyclins and cdks (Fig. 1.2).
In some tissues—such as the bone marrow, epidermis, and gastrointestinal tract—the life span of mature cells is relatively short, and high rates of proliferation by progenitor cells are required to maintain the population. In other tissues—such as liver, muscle, and brain—cells are long lived, and proliferation rarely occurs. Complex molecular mechanisms have evolved to closely regulate proliferation. These involve a finely tuned balance between stimulatory and inhibitory growth signals.
Figure 1.1 Role of proliferation, cell death, senescence, and DNA damage in cancer development.
Figure 1.2 Regulation of cell cycle arrest in G1 by cyclin-dependent kinase (cdk) inhibitors. (A) cell cycle arrest (B) cell cycle progression.
Dysregulation of cellular proliferation is one of the main hallmarks of cancer. There may be increased activity of genes involved in stimulating proliferation (oncogenes) or loss of growth inhibitory (tumor suppressor) genes or both. In the past, it was thought that cancer might arise solely because of more rapid proliferation or a higher fraction of proliferating cells. Although increased proliferation is a characteristic of many cancers and is an appealing therapeutic target (1), the fraction of cancer cells actively dividing and the time required to transit the cell cycle is not strikingly different between many cancers and corresponding normal cells of the same lineage. Altered regulation of proliferation is only one of several factors that contribute to malignant transformation.
In addition to being driven by increased proliferation, growth of a cancer may be attributable to cellular resistance to death. At least three distinct types of cell death pathways have been characterized, including apoptosis, necrosis, and autophagy (2). All three pathways may be ongoing simultaneously within a tumor, and methods that distinguish between them are far from perfect.
The term apoptosis derives from Greek and alludes to a process akin to leaves dying and falling off a tree. Apoptosis is an active, energy-dependent process that involves cleavage of the DNA by endonucleases and proteins by proteases called caspases. Morphologically, apoptosis is characterized by condensation of chromatin, nuclear and cytoplasmic blebbing, and cellular shrinkage. The molecular events that affect apoptosis in response to various stimuli are complex and have only been partially elucidated (3), but several reliable markers of apoptosis have been discovered including annexin V, caspase-3 activation, and DNA fragmentation (4).
External stimuli such as tumor necrosis factor, tumor necrosis factor-related apoptosis-inducing ligand, fatty acid synthase (Fas), and other death ligands that interact with cell surface receptors can induce activation of caspases and lead to apoptosis via an extrinsic pathway (Fig. 1.3). The intrinsic pathway is activated in response to a wide range of stresses including DNA damage and deprivation of growth factors. The intrinsic apoptosis pathway is regulated by a complex interaction of pro- and antiapoptotic proteins in the mitochondrial membrane that affect its permeability. Proteins that increase permeability allow release of cytochrome c, which activates the apoptosome complex leading to activation of caspases that affect apoptosis. Conversely, proteins that stabilize mitochondrial membranes inhibit apoptosis. The first major insight that led to the understanding of the intrinsic apoptosis pathway was the finding that an activating translocation of the bcl-2 gene in B-cell lymphomas results in essentially complete inhibition of apoptosis (5). Subsequent studies demonstrated that the antiapoptotic effect of bcl-2 is attributable to stabilization of the mitochondrial membrane. Additional genes related to bcl-2 (such asBAD, BCL-XL, and others) also block apoptosis by inhibiting membrane permeability. Other genes in the BCL family (such as BAX, BAK, and others) increase membrane permeability and are proapoptotic. An increased understanding of the complex system of molecular checks and balances involved in regulation of apoptosis provides opportunities for targeted cancer therapies; several strategies are under development (6).
In addition to restraining the number of cells in a population, apoptosis serves an important role in preventing malignant transformation by allowing the elimination of cells that have undergone genetic damage. Following exposure of cells to mutagenic stimuli, including radiation and carcinogenic drugs, the cell cycle is arrested so that DNA damage may be repaired. If DNA repair is not sufficient, apoptosis occurs so that damaged cells do not survive (Fig. 1.1). This serves as an anticancer surveillance mechanism by which mutated cells are eliminated before they become fully transformed. In this regard, the TP53 tumor suppressor gene is a critical regulator of cell cycle arrest and apoptosis in response to DNA damage.
Necrosis is a type of cell death that is distinct from apoptosis and is the result of bioenergetic compromise (4). Morphologic changes include swollen organelles and rupture of the cell membrane, leading to loss of osmoregulation and cellular fragmentation. Necrosis is a less well regulated process that leads to spillage of protein contents, and this may incite a brisk immune response. This is in contrast to the silent elimination of cells by apoptosis, which typically elicits a minimal immune response. There is evidence that some drugs may enhance necrotic death in tumors, and this may stimulate a beneficial antitumor immune response.
Figure 1.3 Apoptosis pathways.
Autophagy is a potentially reversible process in which a cell that is stressed “eats” itself (4). A wide range of stresses have been identified that may elicit autophagy (some of which may also elicit apoptosis), including growth factor deprivation and accumulation of reactive oxygen species. Unlike necrosis and apoptosis—in which loss of integrity of the cytoplasmic and nuclear membranes, respectively, are defining events—autophagy is characterized by the formation of cytoplasmic autophagic vesicles into which cellular proteins and organelles are sequestered. This may allow for cell survival if damaged organelles can be repaired. Conversely, the process may lead to cell death if these vesicles fuse with lysosomes with resultant degradation of their contents. Several cancer therapeutic agents have been shown to induce autophagy, while targeted disruption of genes such as ATG5that are involved in autophagy can inhibit cell death (4).
Normal cells are only capable of undergoing division a finite number of times before becoming senescent. Cellular senescence is regulated by a biological clock related to progressive shortening of repetitive DNA sequences (TTAGGG) called telomeres that cap the ends of each chromosome. Telomeres are thought to be involved in chromosomal stabilization and in preventing recombination during mitosis. At birth, chromosomes have long telomeric sequences (150,000 bases) that become progressively shorter by 50 to 200 bases each time a cell divides. Telomeric shortening is the molecular clock that triggers senescence (Fig. 1.1). Malignant cells often avoid senescence by turning on expression of telomerase activity to prevent telomeric shortening (7). Telomerase is a ribonucleoprotein complex, and both the protein and RNA subunits have been identified. The RNA component serves as a template for telomeric extension, and the protein subunit catalyzes the synthesis of new telomeric repeats.
Telomerase activity is detectable in a high fraction of many cancers, including ovarian (8,9), cervical (10,11), and endometrial cancers (12). It has been suggested that detection of telomerase might be useful for early diagnosis of cancer, but lack of specificity is a significant issue. In this regard, endometrium is one of the normal adult tissues in which telomerase expression is most common (13). Perhaps this relates to the need for a large number of lifetime cell divisions because of rapid growth and shedding of this tissue each month during the reproductive years. Therapeutic approaches to inhibiting telomerase are under development that focus on reversing the immortalized state of cancer cells to make them susceptible once again to normal replicative senescence (7).
Origins of Genetic Alterations
Human cancers arise because of a series of genetic and epigenetic alterations that lead to disruption of normal mechanisms that govern cell growth, death, and senescence (14,15).Genetic damage may be inherited or arise after birth as a result of either exposure to exogenous carcinogens or endogenous mutagenic processes within the cell (Table 1.1). The incidence of most cancers increases with aging because the longer one is alive, the higher the likelihood that a cell will acquire sufficient damage to become fully transformed. It is thought that at least three to six alterations are required to fully transform a cell. Most cancer cells are genetically unstable, and this leads to an accumulation of a substantial number of secondary changes that play a role in evolution of the malignant phenotype with respect to growth, invasion, metastasis, and response to therapy, among other characteristics. Genetic instability also results in evolution of heterogeneous clones within a tumor. There is some evidence that progenitor cells (stem cells) exist within a tumor that may be relatively resistant to therapy (16).
Inherited Cancer Susceptibility
Although most cancers arise sporadically in the population because of acquired genetic damage, inherited mutations in cancer susceptibility genes are responsible for some cases. Families with these mutations exhibit a high incidence of specific types of cancers. The age of cancer onset is younger in these families, and it is not unusual for some individuals to be affected with multiple primary cancers. Many of the genes involved in hereditary cancer syndromes have been identified. The most common forms of hereditary cancer syndromes predispose to breast or ovarian (BRCA1, BRCA2) and colon or endometrial (HNPCC genes) cancers (Table 1.2). Examples of other hereditary cancer syndromes are also outlined in Table 1.2.
Tumor suppressor genes have been implicated most frequently in hereditary cancer syndromes, followed by DNA repair genes. In only a few instances are germ-line mutations in oncogenes responsible for hereditary cancers. Although affected individuals carry the germ-line alteration in every cell of their bodies, paradoxically, cancer susceptibility genes are characterized by a limited repertoire of cancers. In addition, there is no relationship between expression patterns of these genes in various organs and the development of specific types of cancers. For example, BRCA1 expression is high in the testis, but men who inherit mutations in this gene are not predisposed to develop testicular cancer. The penetrance of cancer susceptibility genes is incomplete because all individuals who inherit a mutation do not develop cancer. The emergence of cancers in carriers depends on the occurrence of additional genetic alterations.
Table 1.1 Origins of Genetic Damage in Human Cancers
Table 1.2 Hereditary Cancer Syndromes
The familial cancer syndromes described above result from rare mutations that occur in less than 1% of the population. In addition, low-penetrance common genetic polymorphisms may also affect cancer susceptibility, albeit less dramatically (17). There are more than ten million polymorphic genetic loci in the human genome. Many of these polymorphisms are common, with the rarer allele occurring in more than 5% of individuals.
Although genetic polymorphisms would not be expected to increase risk sufficiently to produce familial cancer clustering, they could account for a significant fraction of cancers currently classified as sporadic because of their relatively high prevalence. For example, 6% of Ashkenazi Jews carry the rare allele of the I1307K polymorphism in the APC gene, which increases the risk of colorectal cancer by about 50% (18). The recent development of genomic technologies that can assess thousands of polymorphisms simultaneously in large numbers of individuals is fueling the search for additional genetic susceptibility polymorphisms. A more complete understanding of the genetic factors that affect cancer susceptibility could facilitate implementation of screening and prevention approaches in subsets of the population at increased risk.
Acquired Genetic Damage
The etiology of acquired genetic damage seen in cancers also has been elucidated to some extent. For example, a strong causal link exists between cigarette smoke and cancers of the aerodigestive tract and between ultraviolet radiation and skin cancer. For many common forms of cancer (colon, breast, endometrium, ovary) a strong association with specific carcinogens does not exist. It is thought that the genetic alterations responsible for these cancers may arise mainly because of endogenous mutagenic processes such as methylation, deamination, and hydrolysis of DNA. In addition, spontaneous errors in DNA synthesis may occur during the process of DNA replication associated with normal proliferation. Likewise, free radicals generated in response to inflammation and other cellular damage may cause DNA damage. These endogenous processes produce many mutations each day in every cell in the body. Several families of highly effective DNA damage surveillance and repair genes exist, but some mutations may elude them. The efficiency of these DNA damage-response systems varies between individuals because of genetic and other factors and may affect susceptibility to cancer.
Epigenetics changes are heritable changes that do not result from alterations in DNA sequence (15). Methylation of cytosine residues that reside next to guanine residues is the primary mechanism of epigenetic regulation, and this process is regulated by a family of DNA methyltransfereases. Most cancers have globally reduced DNA methylation, which may contribute to genomic instability. Conversely, selective hypermethylation of cytosines in the promoter regions of tumor suppressor genes may lead to their inactivation, and this may contribute to carcinogenesis.
There is a family of imprinted genes in which either the maternal or paternal copy is normally completely silenced because of methylation. Loss of imprinting in genes that stimulate proliferation, such as insulin-like growth factor 2 (IGF2), may provide an oncogenic stimulus by increasing proliferation.
Acetylation and methylation of the histone proteins that coat DNA represent another level of epigenetic regulation that is altered in cancer. The underlying cause of these epigenetic alterations remains poorly understood, but they may represent appealing therapeutic targets.
Alterations in genes that stimulate cellular growth (oncogenes) can cause malignant transformation (14,15). Oncogenes can be activated via several mechanisms. In some cancers, amplification of oncogenes with resultant overexpression of the corresponding protein has been noted. Instead of two copies of one of these genes, there may be many more copies. Some oncogenes may become overactive when affected by point mutations. Finally, oncogenes may be translocated from one chromosomal location to another and then come under the influence of promoter sequences that cause overexpression of the gene. This latter mechanism frequently occurs in leukemias and lymphomas but not in gynecologic cancers or other solid tumors.
In cell culture systems in the laboratory, many genes that are involved in normal growth regulatory pathways can elicit transformation to overactive forms when altered to overactive forms via amplification, mutation, or translocation. On this basis, a large number of genes have been classified as oncogenes. Studies in human cancers have suggested that the actual spectrum of genes altered in the development of human cancers is more limited. A number of genes that elicit transformation when activated in vitro have not been documented to undergo alterations in human cancers. In this section, the various classes of oncogenes will be summarized and particular attention paid to those that are altered in gynecologic cancers.
Cell Membrane Oncogenes: Peptide Growth Factors and Their Receptors
Peptide growth factors—such as those of the epidermal growth factor (EGF), plateletderived growth factor (PDGF), and fibroblast growth factor (FGF) families—stimulate a cascade of molecular events that leads to proliferation by binding to cell membrane receptors (Fig. 1.4). Growth factors are involved in normal cellular processes such as development, stromal-epithelial communication, tissue regeneration, and wound healing. Growth factors in the extracellular space can stimulate a cascade of molecular events that leads to proliferation by binding to cell membrane receptors. Unlike endocrine hormones, which are secreted into the blood stream and act in distant target organs, peptide growth factors typically act in the local environment where they have been secreted.
The concept that autocrine growth stimulation may be a key strategy by which cancer cell proliferation becomes autonomous has received considerable attention. In this model, it is postulated that cancers secrete stimulatory growth factors that then interact with receptors on the same cell. Although peptide growth factors provide a growth stimulatory signal,there is little evidence to suggest that overproduction of growth factors is a precipitating event in the development of most cancers. Increased expression of peptide growth factors likely serves in most cases as a cofactor rather than as the driving force behind malignant transformation.
Cell membrane receptors that bind peptide growth factors are composed of an extracellular ligand-binding domain, a membrane spanning region, and a cytoplasmic tyrosine kinase domain (19). Binding of a growth factor to the extracellular domain results in aggregation and conformational shifts in the receptor and activation of the inner tyrosine kinase. This kinase phosphorylates tyrosine residues both on the growth factor receptor itself (autophosphorylation) and on molecular targets in the cell interior, leading to activation of secondary signals.
Figure 1.4 Mitogenic signal transduction pathways.
Growth of some cancers is driven by overexpression of receptor tyrosine kinases receptors. Therapeutic strategies that target receptor tyrosine kinases have been an active area of investigation. Trastuzumab is a monoclonal antibody that blocks the HER-2/neu receptor, and it is widely used in the treatment of breast cancers that overexpress this tyrosine kinase (20). Cetuximab is a monoclonal antibody that targets the epidermal growth factor receptor (EGFR), whereas gefitinib is a direct inhibitor of the EGFR tyrosine kinase (21).Lapatinib is a dual EGFR/HER-2 kinase inhibitor. Imatinib antagonizes the activity of the BCR-ABL, c-kit, and PDGF receptor tyrosine kinases and has proven effective in treatment of chronic myelogenous leukemias and gastrointestinal stromal tumors.
Following the interaction of peptide growth factors and their receptors, secondary molecular signals are generated to transmit the growth stimulus to the nucleus (Fig. 1.4). This function is served by a multitude of complex and overlapping signal transduction pathways that occur in the inner cell membrane and cytoplasm. Many of these signals involve phosphorylation of proteins by enzymes known as nonreceptor kinases (22). These kinases transfer a phosphate group from ATP to specific amino acid residues of target proteins. The kinases that are involved in growth regulation are of two types: those that are phosphorylate tyrosine residues on proteins, including those of the SRC family (23); and others that are specific for serine or threonine residues such as AKT (24). The activity of kinases is regulated by phosphatases such as PTEN, which act in opposition to the kinases by removing phosphates from the target proteins.
Guanosine-triphosphate-binding proteins (G proteins) represent another class of molecules involved in transmission of growth signals (Fig 1.4). They are located on the inner aspect of the cell membrane and have intrinsic GTPase activity that catalyzes the exchange of guaninetriphosphate (GTP) for guanine-diphosphate (GDP). In their active GTP-bound form, G proteins interact with kinases that are involved in relaying the mitogenic signal, such as those of the MAP kinase family. Conversely, hydrolysis of GTP to GDP, which is stimulated by GTPase-activating proteins (GAPs), leads to inactivation of G proteins. The ras family of G proteins is among the most frequently mutated oncogenes in human cancers (e.g., gastrointestinal and endometrial cancers). Activation of ras genes usually involves point mutations in codons 12, 13, or 61 that result in constitutively activated molecules. The BRAFgene encodes a kinase that interacts with ras proteins in activating the MAP kinase pathway. BRAF mutations occur in many cancers that lack ras mutations, and most of these mutations involve codon 599 in the kinase domain (25). Therapeutic approaches to interfering with ras signaling are being developed, including farnesyltransferase inhibitors that block attachment of ras to the inner cell membrane, antisense oligonucleotides, and RNA interference (26).
If proliferation is to occur in response to signals generated in the cell membrane and cytoplasm, these events must lead to activation of nuclear transcription factors and other genetic products responsible for stimulating DNA replication and cell division. Expression of several genes that encode nuclear proteins increases dramatically within minutes of treatment of cells with peptide growth factors. Once induced, the products of these genes bind to specific DNA regulatory elements and induce transcription of genes involved in DNA synthesis and cell division. Examples include the fos and jun oncogenes, which dimerize to form the activator protein 1 (AP1) transcription complex. When inappropriately overexpressed, however, these transcription factors can act as oncogenes. Among the nuclear transcription factors involved in stimulating proliferation, amplification or overexpression of members of the myc family has most often been implicated in the development of human cancers (27). Many of the nuclear regulatory genes such as myc that control proliferation also affect the threshold for apoptosis. Thus, there is overlap in the molecular pathways that regulate the opposing processes of proliferation and apoptosis.
Genes involved in chromatin remodeling also that have been implicated as oncogenes, but primarily in hematologic malignancies rather than solid tumors. Finally, as discussed previously, genes encoding nuclear proteins that inhibit apoptosis (e.g., bcl-2) can act as oncogenes when altered to constituitively active forms.
Tumor Suppressor Genes
Loss of tumor suppressor gene function also plays a role in the development of most cancers. This usually involves a two-step process in which both copies of a tumor suppressor gene are inactivated. In most cases, there is mutation of one copy of a tumor suppressor gene and loss of the other copy because of deletion of a segment of the chromosome where the gene resides. There is also evidence that some tumor suppressor genes may be inactivated because of methylation of the promoter region of the gene (15). The promoter is an area proximal to the coding sequence that regulates whether the gene is transcribed from DNA into RNA. When the promoter is methylated, it is resistant to activation, and the gene is essentially silenced despite remaining structurally intact.
This two-hit paradigm is relevant to both hereditary cancer syndromes, in which one mutation is inherited and the second acquired, and sporadic cancers, in which both hits are acquired. Tumor suppressor gene products are found throughout the cell, reflecting their diverse functions. With the recognition that inactivation of tumor suppressor genes is a defining feature of cancers, genetic therapy strategies have been developed that aim to deliver functional copies of these lost genes to cancer cells.
Nuclear Tumor Suppressor Genes
The retinoblastoma gene was the first tumor suppressor gene discovered (28). It was found in the context of a rare hereditary cancer syndrome, as have many other tumor suppressor genes (Table 1.2). The Rb gene plays a key role in the regulation of cell cycle progression (Fig. 1.2). In the G1 phase of the cell cycle, Rb protein binds to the E2F transcription factor and prevents it from activating transcription of other genes involved in cell cycle progression. G1 arrest is maintained by cyclin-dependent kinase inhibitors that prevent phosphorylation of Rb, such as p16, p21, and p27 (29). When Rb is phosphorylated by cyclin-cdk complexes, E2F is released and stimulates entry into the DNA synthesis phase of the cell cycle. Other cyclins and cdks are involved in progression from G2 to mitosis. Mutations in the Rb gene have been noted primarily in retinoblastomas and sarcomas but rarely in other types of cancers. By maintaining G1 arrest, the cdk inhibitors p16, p21, p27, and others act as tumor suppressor genes. Loss of p16 tumor suppressor function as a result of genomic deletion or promoter methylation occurs in some cancers, including familial melanomas. Likewise, loss of p21 and p27 has been noted in some cancers.
Mutation of the TP53 tumor suppressor gene is the most frequent genetic event described thus far in human cancers (Fig. 1.5) (30,31). The TP53 gene encodes a 393 amino acid protein that plays a central role in the regulation of both proliferation and apoptosis. In normal cells, p53 protein resides in the nucleus and exerts its tumor suppressor activity by binding to transcriptional regulatory elements of genes, such as the cdk inhibitor p21, that act to arrest cells in G1. The MDM2 gene product degrades p53 protein when appropriate, whereas p14ARF down regulates MDM2 when up regulation of p53 is needed to initiate cell cycle arrest.
Figure 1.5 Inactivation of the p53 tumor suppressor gene by “dominant negative” missense mutation or by truncation mutation and deletion.
Many cancers have missense mutations in one copy of the TP53 gene that result in substitution of a single amino acid in exons 5 through 8, which encode the DNA binding domains (Fig 1.5). Although these mutant TP53 genes encode full-length proteins, they are unable to bind to DNA and regulate transcription of other genes. Mutation of one copy of the TP53gene often is accompanied by deletion of the other copy, leaving the cancer cell with only mutant p53 protein. If the cancer cell retains one normal copy of the TP53 gene, mutant p53 protein can complex with wild-type p53 protein and prevent it from oligimerizing and interacting with DNA. Because inactivation of both TP53 alleles is not required for loss of p53 function, mutant p53 is said to act in a “dominant negative” fashion. Although normal cells have low levels of p53 protein because it is rapidly degraded, missense mutations encode protein products that are resistant to degradation. The resultant overaccumulation of mutant p53 protein in the nucleus can be detected immunohistochemically. A smaller fraction of cancers have mutations in the TP53 gene that encode truncated protein products. In these cases, loss of the other allele occurs as the second event as is seen with other tumor suppressor genes.
Beyond simply inhibiting proliferation, normal p53 is thought to play a role in preventing cancer by stimulating apoptosis of cells that have undergone excessive genetic damage. In this regard, p53 has been described as the “guardian of the genome” because it delays entry into S phase until the genome has been cleansed of mutations. If DNA repair is inadequate, then p53 may initiate apoptosis, thereby eliminating cells with genetic damage. Likewise, other genes that repair damage to the DNA nucleotide sequence or strand breakage also sometimes are classified as tumor suppressors. These other genes will be discussed in the next sections in the context of hereditary gynecologic cancer syndromes.
Extranuclear Tumor Suppressor Genes
Although many tumor suppressor genes—including TP53, Rb, and p16—encode nuclear proteins, some extranuclear tumor suppressors have been identified. Theoretically, any protein that normally is involved in inhibition of proliferation has the potential to act as a tumor suppressor. In this regard, appealing candidates include phosphatases such as PTEN that normally oppose the action of the tyrosine kinases by dephosphorylating tyrosine residues. In addition to its phosphatase activity, PTEN is homologous to the cytoskeleton proteins tensin and auxin, and it has been postulated that PTEN might act to inhibit invasion and metastasis through modulation of the cytoskeleton. The APC tumor suppressor gene encodes a cytoplasmic protein involved in the wnt signaling pathway that regulates both cellular proliferation and adhesion. Inactivation of APC leads to malignant transformation, and inherited mutations in this gene are responsible for familial adenomatous polyposis syndrome. The transforming growth factor-beta (TGF-β) family of peptide growth factors inhibit proliferation of normal epithelial cells and serve as a tumor suppressive pathway (32). It is thought that TGF-β causes G1 arrest by inducing expression of cyclin-dependent kinase inhibitors such as p27. Three closely related forms of TGF-β have been discovered that are encoded by separate genes (TGF-β1, TGF-β2, and TGF-β3). TGF-β is secreted from cells in an inactive form bound to a portion of its precursor molecule from which it must be cleaved to release biologically active TGF-β. Active TGF-β interacts with type I and type II cell surface TGF-β receptors and initiates serine or threonine kinase activity. Prominent intracellular targets include a class of molecules called Smads that translocate to the nucleus and act as transcriptional regulators. Although mutations in the TGF-β receptors and Smads have been reported in some cancers, this does not appear to be a feature of gynecologic cancers.
In addition to primary disregulation of oncogenes and tumor suppressor genes, altered expression of microRNAs that regulate the expression of these genes occurs in many cancers (33). MicroRNA genes consist of a single RNA strand of approximately 21 to 23 nucleotides that does not encode proteins. They bind to messenger RNAs that contain complementary sequences and can block protein translation.
Invasion and Metastasis
Metastasis is a process by which cancer cells spread from the primary tumor to distant sites (34). Cancer metastasis can proceed only if a series of sequential steps are completed, including proliferation, angiogenesis, invasion, embolism or circulation, transport, adherence in organs, adherence to vessel wall, and extravasation (Fig. 1.6) (34).
It has long been recognized that most types of cancer have an organ-specific pattern of metastasis. The propensity of various types of cancer to form metastases in specific organs was first proposed by Paget, who hypothesized that these patterns resulted from the “dependence of the seed (cancer cell) on the soil (the metastatic site)” (35). This hypothesis was suggested by the nonrandom pattern of metastasis. Paget concluded that metastases formed only when the seed and soil were compatible. It is now appreciated at a molecular level that metastasis is dependent on a balance between stimulating factors from both the tumor and host cells versus inhibitory signals. To produce metastasis, the balance must be weighted toward the stimulatory signals.
Cancer progression is a product of an evolving crosstalk between different cell types within the tumor and its surrounding supporting tissue, the tumor stroma (36). The tumor stroma contains a specific extracellular matrix as well as cellular components such as fibroblasts, immune and inflammatory cells, and blood-vessel cells. The interactive signaling between tumor and stroma contributes to the formation of a complex multicellular organ. The organ microenvironment can markedly change the gene-expression patterns of cancer cells and therefore their behavior and growth potential (36). Recent studies regarding chemokines and their receptors provide important clues regarding why some cancers metastasize to specific organs. For example, breast cancer cells frequently express chemokine receptors CXCR4 and CCR7 at high levels. The specific ligands for these receptors, CXCL12 and CCL 21, are found at high levels in lymph nodes, lung, liver, and bone marrow, which are common sites for breast cancer metastasis.
Figure 1.6 Molecular pathways involved in invasion and metastasis. Redrawn from Fidler IJ. The pathogenesis of cancer metastasis: the “seed and soil” hypothesis revisited. Nature Reviews Cancer 2003;3:453-458.
All cells require oxygen and other nutrients for survival and growth, and cells must reside within 100 µm of a capillary in order to receive oxygen (37). Therefore, growth of new vessels, termed angiogenesis, is required for sustained malignant growth beyond approximately 1 mm in diameter. Angiogenesis occurs as a result of a shift in balance toward proangiogenic factors within the tumor microenvironment along with down regulation of antiangiogenic influences. One of the primary mediators of angiogenesis is vascular endothelial growth factor A (VEGF-A) (38), which increases vascular permeability, stimulates endothelial cell proliferation and migration, and promotes endothelial cell survival (39). Other mediators of angiogenesis include tumor-derived factors and host stromal factors including interleukin-8, alpha v-beta 3 integrin, the tyrosine kinase receptor EphA2, and matrix metalloproteinases (40). From a translational perspective, patient-specific tumor microenvironmental characteristics may influence the response to antiangiogenic therapy (41). Antiangiogenesis therapies such as Bevicizumab that target the VEGF pathway are showing promise in ovarian cancer and have already entered phase III clinical trials (40). Approaches targeting additional angiogenic pathways are also under clinical development.
A critical first step in metastasis, and the primary feature that defines malignancy, is invasion through the basement membrane. This requires interplay between cancer cells and a permissive underlying stroma (42). Invasion of malignant cells through the basement membrane and endothelial cell migration for angiogenesis require degradation of the extracellular matrix. This process is facilitated by a group of enzymes called matrix metalloproteinases (MMPs), which are a family of zinc-dependent endopeptidases that digest collagen and other extracellular matrix components. They also stimulate proliferation and induce release of VEGF. Ovarian tumors overexpress MMP-2 and MMP-9, and this increased expression correlates with aggressive clinical features (43).
Tumor cell adhesion to the extracellular matrix within tissues greatly influences the ability of a malignant cell to invade and metastasize (44). Given the shedding nature of ovarian cancer, adhesion molecules such as focal adhesion kinase, integrins, and E-cadherin have been evaluated for their role in peritoneal metastasis (45). The proteins of the extracellular matrix consist of type I and IV collagens, laminins, heparin sulfate proteoglycan, fibronectin, and other noncollagenous glycoproteins (46). Cell adhesion to these proteins is mediated in part by a group of heterodimeric transmembrane proteins called integrins, which are composed of a noncovalently associated α and β subunit that define the integrin-ligand specificity (47). Approximately 18 β subunits and eight α subunits have been identified, and at least 24 receptor combinations exist (48). The intracellular domains of integrins interact with cytoskeletal components and are actively involved in generating intracellular signals.
Cadherins are another group of cell-cell adhesion molecules that are involved in development and maintenance of solid tissues. E-cadherins are the subgroup predominantly found in epithelial cells (49). These transmembrane proteins mediate cell-cell adhesion: Cadherins on neighboring cells preferentially bind to the same types of cadherins on adjacent cells. E-cadherin is uniformly expressed in ovarian cancer, in low-malignant-potential tumors, in benign neoplasms, and—notably—in inclusion cysts of normal ovaries, but not in the normal surface epithelium (50). Cadherin dysfunction is associated with loss of cell-cell cohesion, altered cellular motility, and increased invasiveness and metastatic potential. Changes in the composition of the cadherin-catenin complex, phosphorylation of components in the complex, and alterations in the interactions with the actin cytoskeleton have all been suggested as playing a role in regulating adhesion.
E-cadherin mutations occur only rarely (51), but cadherin expression may also be down regulated in the absence of mutations. The cytoplasmic tails of cadherins exist as a macromolecular complex with β-catenin, which is involved in the wnt signaling pathways that regulate both adhesion and growth. Regulation of β-catenin activity also depends on the APC gene product and others in the wnt pathway (Fig. 1.7). Mutations in the APC gene that abrogate its ability to inhibit β-catenin activity are common in both the hereditary adenomatous polyposis coli syndrome and sporadic colon cancers (52). Likewise, mutations in the β-catenin gene that result in constitutively activated molecules also have been observed in some cancers, including endometrial cancers (53).
Figure 1.7 The wnt signaling pathway. Redrawn from Spannuth WA, Sood AK, Coleman RL. Angiogenesis as a strategic target for ovarian cancer therapy. Nature Clinical Practice Oncology 2008;5:194-204.
Within the tumor microenvironment, other cell types also play a critical role in tumor growth and progression. For example, recent studies indicate that certain types of inflammatory cells, including macrophages and mast cells, and their associated cytokines confer an unfavorable prognosis and increased tumor growth. Conversely, the presence of an adaptive immune response characterized by cytotoxic T cells is associated with improved clinical outcome. In addition, cancer cells may evade immune recognition and destruction by various means, such as Fas ligand production to induce lymphocytice apoptosis and HLA-G secretion to inhibit natural-killer cell activity (54). Moreover, cytokine production by cancer cells promotes growth and inhibits apoptosis. The mechanistic relationships between the microenvironment and tumor growth remain only partially understood, but immunomodulating strategies that target the cancer-promoting properties of both innate and adaptive immune cell populations are being developed.
Gynecologic cancers vary with respect to grade, histology, stage, response to treatment, and survival. It is now appreciated that this clinical heterogeneity is attributable to differences in underlying molecular pathogenesis. Some cancers arise in a setting of inherited mutations in cancer susceptibility genes, but most occur sporadically in the absence of a strong hereditary predisposition. The spectrum of genes that are mutated varies between cancer types. For each type of cancer, there are a few genes that are frequently mutated, while a wider spectrum are altered in a small fraction of cases (55).
There also is significant variety with respect to the spectrum of genetic changes within a given type of cancer. Cancers with a similar microscopic appearance may differ greatly at the molecular level. In some instances, molecular features may be predictive of clinical phenotypes such as stage, histologic type, and survival. As we gain a more complete understanding of the clinical implications of various genetic alterations in gynecologic cancers, the molecular profile may prove valuable in predicting clinical behavior and response to treatment.
Epidemiologic and clinical studies of endometrial cancer have suggested that there are two distinct types of endometrial cancer (56). Type I cases are associated with unopposed estrogen stimulation and often develop in a background of endometrial hyperplasia. Obesity is the most common cause of unopposed estrogen and is part of a metabolic syndrome that also includes insulin resistance and overexpression of insulin-like growth factors that may also play a role in carcinogenesis. Type I cancers are well differentiated, endometrioid, early stage lesions and have a favorable outcome. In contrast, type II cancers are poorly differentiated or nonendometrioid (or both) and are more virulent. They often present at an advanced stage, and survival is relatively poor.
In practice, not all cancers can be neatly characterized as either pure type I or II lesions, and endometrial cancers can also be viewed as a continuous spectrum with respect to etiology and clinical behavior. However, as the genetic events involved in the development of endometrial cancer have been elucidated, it has been found that specific alterations often, but not always, are seen primarily in either type I or II cases (Table 1.3).
Similar to other human cancers, endometrial cancers are believed to arise because of a series of genetic alterations. A small minority of endometrial cancers occur in women with a strong hereditary predisposition because of germ-line mutations in DNA repair genes in the context of HNPCC syndrome. A central unresolved issue in the understanding of endometrial carcinogenesis is the role of unopposed estrogenic stimulation. It has long been thought that estrogens may contribute to the development of endometrial cancer by virtue of their mitogenic effect on the endometrium. A higher rate of proliferation in response to estrogens may lead to an increased frequency of spontaneous mutations. In addition, when genetic damage occurs, regardless of the cause, the presence of estrogens may facilitate clonal expansion. It also has been postulated that estrogens may act as “complete carcinogens” that not only promote carcinogenesis by stimulating proliferation but also act as initiating agents by virtue of their carcinogenic metabolites. In contrast,progestins oppose the action of estrogens by both down regulating estrogen receptor levels and decreasing proliferation and increasing apoptosis.
Hereditary Endometrial Cancer
Approximately 3% to 5% of endometrial cancers arise because of inherited mutations in DNA repair genes in the context of hereditary nonpolyposis colon cancer (HNPCC) syndrome. HNPCC typically manifests as familial clustering of early onset colon cancer (57,58). There is also an increased incidence of several other types of cancers—most notably, endometrial cancer in women. The risk of ovarian, stomach, biliary tract, and other cancers also is somewhat increased. The identification of the DNA mismatch repair genes responsible for HNPCC has facilitated the development of genetic testing (59). Most HNPCC cases result from alterations in MSH2 and MLH1. MSH6 mutations also are associated with an increased incidence of endometrial cancer (60). PMS1 and PMS2 have been implicated in a small number of these cancers as well. Loss of mismatch repair leads to a “mutator phenotype” in which genetic mutations accumulate throughout the genome, particularly in repetitive DNA sequences called microsatellites. Examples of microsatellite sequences include mono-, di-, and trinucleotide repeats (AAAA, CACACACA, and CAGCAGCAGCAG). The propensity to accumulate mutations in microsatellite sequences is referred to asmicrosatellite instability (MSI). Some microsatellite sequences are in noncoding areas of the genome, whereas others are within genes. It is thought that the accumulation of mutations in microsatellite sequences of tumor suppressor genes may inactivate them and accelerate the process of malignant transformation.
Table 1.3 Characteristics of Type 1 versus Type 2 Endometrial Cancers
The Amsterdam and Bethesda criteria have been developed to provide clinical guidelines for the diagnosis of HNPCC based on the spectrum of cancers noted in a family. These criteria are inexact, and genetic testing should be considered in all families suspected of having HNPCC based on family history. Involvement of genetic counselors is useful in facilitating this process. Analysis of cancers for microsatellite instability has been proposed as a genetic screening test for HNPCC. Among families with germ-line mutations in mismatch repair genes, MSI is seen in greater than 90% of colon cancers and approximately 75% of endometrial cancers (61,62). However, MSI is found in 20% to 25% of endometrial cancers (63) and 15% to 20% of colorectal cancers overall (64), and most of these cases are attributable to silencing of the MLH1 gene because of promoter methylation (65,66).Another screening approach for HNPCC is immunohistochemical staining of tumors to determine where there has been a loss of MSH2 or MLH1 protein (67). In cancers with MSI or loss of expression of one of the mismatch repair proteins, these genes can be sequenced to identify the disease-causing mutations, most of which cause truncated protein products.Currently, mutational analysis of the responsible genes remains the gold standard for diagnosis of HNPCC (59). Mutational analysis typically involves analysis of only MSH2 and MLH1,which may overlook mutations in the other mismatch repair genes. Approximately half of families for which HNPCC is strongly suspected will be found to have a germ-line MLH1 orMSH2 mutation.
Endometrial cancer is the most common extracolonic malignancy is women with HNPCC. The risk of a woman developing endometrial cancer has ranged from 20% to 60% in various reports (68,69,70), and in some studies this exceeds the risk of colon cancer. In addition, the risk of ovarian cancer is increased to approximately 5% to 12%.
The most striking clinical feature of HNPCC-related cancers is early onset, typically at least ten years earlier than sporadic cases. The average age of women with sporadic endometrial cancers is in the early 60s, whereas cancers that arise in association with HNPCC are often diagnosed before the menopause (average age in the 40s) (68,71,72). The clinical features of HNPCC-associated endometrial cancers are similar to those of most sporadic cases (well differentiated, endometrioid, early stage), and survival is approximately 90% (71,73).
The mean age of onset of ovarian cancer in HNPCC families is in the early 40s, and the clinical features of these cancers generally are more favorable than sporadic cases (74). They are usually are early stage and well or moderately differentiated, and approximately 20% occur in the setting of synchronous endometrial cancers. However, analysis of groups of patients with synchronous cancers of the ovary and endometrium has revealed that few of these exhibit microsatellite instability, and most probably are not attributable to HNPCC syndrome (75).
The optimal strategy for prevention of HNPCC-associated mortality is unclear. Screening and surgical prophylaxis are both employed for colonic and extracolonic malignancies. Surveillance and prophylactic surgery should be considered early (between ages 25 and 35), generally ten years before the earliest onset of cancer in other relatives who had an HNPCC-related malignancy (72). Transvaginal ultrasound has been proposed as a screening test for endometrial and ovarian cancer, but it appears to be relatively ineffective (76). There is no evidence that CA125 or other blood markers facilitate early detection of endometrial cancer, but CA125 can be justified as a means of screening for HNPCC-associated ovarian cancer. Endometrial biopsy may be the only screening test with sufficient sensitivity, and it has been suggested that this should be employed periodically beginning around age 30 to 35. However, no data have been published demonstrating that this approach is superior with respect to decreasing mortality compared to simply performing biopsies in response to abnormal uterine bleeding.
The rationale for prophylactic hysterectomy in women with HNPCC is based on the high lifetime risk of endometrial cancer and the fact that the uterus does not serve a vital function once childbearing is complete. One study demonstrated that there were no cases of endometrial cancer in 61 HNPCC carriers who underwent prophylactic hysterectomy, whereas endometrial cancer developed in 69 of 210 (33%) who did not undergo surgery (77). On the other hand, because survival of women with HNPCC-associated endometrial cancers is approximately 90%, it is conceivable that prophylactic hysterectomy may not appreciably decrease mortality.
Some women in HNPCC families elect to undergo prophylactic colectomy. This provides an opportunity to remove the uterus as well. Hysterectomy in concert with colectomy, either via laparoscopy or laparotomy, should not greatly increase operative time or complications. If an endometrial biopsy has not been performed before prophylactic hysterectomy, the uterus should be opened intraoperatively and examined carefully. If a visual suspicion of cancer is confirmed by frozen section, then surgical staging can be performed. In view of the increased risk of ovarian cancer in HNPCC syndrome, concomitant prophylactic salpingo-oophorectomy should be strongly considered. Estrogen-replacement therapy following oophorectomy is not contraindicated in women with HNPCC because there is no evidence that this adversely affects the incidence of other cancers. In fact, postmenopausal estrogen-replacement therapy in the general population substantially decreases colon cancer risk (78).
Sporadic Endometrial Cancer
Cytogenetic studies have described gross chromosomal alterations in endometrial cancers, including changes in the number of copies of specific chromosomes (79). More recently, comparative genomic hybridization (CGH) studies have demonstrated areas of chromosomal loss and gain in both endometrial cancers and atypical hyperplasias (80,81). The most common sites of chromosomal gain are 1q, 8q, 10p, and 10q (82,83,84). Chromosomal losses also are frequently observed using CGH and in loss of heterozygosity studies (85). A correlation has been noted between higher numbers of chromosomal alterations on CGH and more virulent clinical features (86). The overall number of chromosomal alterations detected using CGH is lower in endometrial cancers relative to other cancer types.
Ploidy analysis simply measures total nuclear DNA content. Approximately 80% of endometrial cancers have a normal diploid DNA content as measured by ploidy analysis. Aneuploidy occurs in 20% and is associated with advanced stage, poor grade, nonendometrioid histology and poor survival (87). The frequency of aneuploidy (20%) is relatively low in endometrial cancers relative to ovarian cancers (80%). One might speculate that endometrial cancers more often present at an early stage than ovarian cancers because they usually have a lower level of genetic aberrations, as opposed to the conventional wisdom that attributes the favorable outcome of endometrial cancers to earlier diagnosis.
Finally, patterns of genetic expression have been described using microarrays that distinguish between normal and malignant endometrium and between various histologic types of cancer (85,88,89). Different types of microarrays have been employed by various groups, and study results often cannot be compared directly. In addition, selection of normal controls for comparison to cancers has not been uniform (90), and microdissection methods aimed at maximizing isolation of cancer have not always been employed (91). Global gene-expression profiles associated with both lymph node metastasis (92) and recurrence (93) have been identified in endometrial cancer. Studies to validate these biomarker panels and molecular-based prediction models are ongoing. This approach has the potential to dramatically increase our understanding of the molecular pathogenesis of endometrial cancer and to enhance prediction of clinical phenotypes.
Tumor Suppressor Genes
Inactivation of the TP53 tumor suppressor gene is among the most frequent genetic events in endometrial cancers (30). Overexpression of mutant p53 protein occurs in approximately 20% of endometrial adenocarcinomas and is associated with several known prognostic factors, including advanced stage, poor grade, and nonendometrioid histology (87,94). Overexpression occurs in some 10% of stages I and II and 40% of stages III and IV cancers (94). Numerous studies have confirmed the strong association between p53overexpression and poor prognostic factors and decreased survival (95,96,97,98,99,100,101). In some of these studies, p53 overexpression has been associated with worse survival even after controlling for stage (102,103). This suggests that loss of p53 tumor suppressor function confers a particularly virulent phenotype. Although little is known regarding molecular alterations in uterine sarcomas, overexpression of mutant p53 occurs in a majority of mixed mesodermal sarcomas of the uterus (74%) and in some leiomyosarcomas (104,105).
Endometrial cancers that overexpress p53 protein usually harbor missense mutations in exons 5 through 8 of the gene that result in amino acid substitutions in the protein (94,106,107,108,109). These mutations lead to loss of DNA binding activity. Because TP53 mutations rarely, if ever, occur in endometrial hyplerplasias (106,110), this likely represents a late event in the development of type I endometrioid endometrial cancers. Alternatively, it is possible that acquisition of a TP53 mutation leads to development of a virulent poorly differentiated or serous “type II” endometrial cancer that does not transition through a phase of hyperplasia, and it is associated with rapid spread of disease. In studies of papillary serous carcinoma, TP53 mutation and p53 protein overexpression have been observed in the vast majority of cases, as well as in its putative dysplastic glandular precursor lesion (111,112).
Mutations in the PTEN tumor suppressor gene occur in approximately 30% to 50% of endometrial cancers (113,114,115), and this represents the most frequent genetic alteration described thus far in these cancers. Deletion of the second copy of the gene is also a frequent event, which results in complete loss of PTEN function. Most of these mutations are deletions, insertions, and nonsense mutations that lead to truncated protein products, whereas only about 15% are missense mutations that change a single amino acid in the critical phosphatase domain. The PTEN gene encodes a phosphatase that opposes the activity of cellular kinases. For example, it has been shown that loss of PTEN in endometrial cancers is associated with increased activity of the PI3 kinase with resultant phosphorylation of its downstream substrate Akt (116).
Mutations in the PTEN gene are associated with endometrioid histology, early stage and favorable clinical behavior (117). Well differentiated, noninvasive cases have the highest frequency of mutations. In addition, PTEN mutations have been observed in 20% of endometrial hyperplasias, suggesting that this is an early event in the development of some endometrioid type I endometrial cancers (118). It has been reported that loss of PTEN may occur in endometrial glands that appear normal, and it is proposed that this may represent the earliest event in endometrial carcinogenesis (119,120).
Synchronous endometrioid cancers are sometimes encountered in the endometrium and ovary that are indistinguishable microscopically. In some of these cases, identical PTENmutations have been identified, suggesting that the ovarian tumor represents a metastasis from the endometrium (121). In other cases, the PTEN mutation seen in the endometrial cancer was not found in the ovarian tumor, suggesting that these represent two distinct primary cancers. PTEN mutations also have been observed in approximately 20% of endometrioid ovarian cancers that arise in the absence of endometrial cancers (122). As noted in the section on hereditary endometrial cancer, it has been shown that inherited mutations in DNA mismatch repair genes are responsible for the HNPCC syndrome. Endometrial cancer is the second most common malignancy observed in women with HNPCC.
Cancers that arise in these women with HNPCC syndrome are characterized by mutations in multiple microsatellite repeat sequences throughout the genome. This microsatellite instability also has been seen in approximately 20% of sporadic endometrial cancers (123,124). Endometrial cancers that exhibit microsatellite instability tend to be type I cancers.Loss of mismatch repair in these cases usually results from silencing of the MLH1 gene by promoter methylation (65,66). Methylation of the MLH1 promoter also has been noted in endometrial hyperplasias (124,125) and normal endometrium adjacent to cancers, suggesting that this is an early event in the development of some of these cancers (126). It is thought that global changes in methylation that result in decreased expression of a number of tumor suppressor and DNA repair genes may be a characteristic of some endometrial cancers, particularly type I cases (127,128). Loss of DNA mismatch repair may accelerate the process of malignant transformation by facilitating accumulation of mutations in microsatellite sequences present in genes involved in malignant transformation.
Several other tumor suppressor genes may play a role in the development of some endometrial cancers. The Par-4 gene is a proapoptotic factor, and loss of expression of this gene occurs in some human cancers. Reduced expression occurs in approximately 40% of endometrial cancers and may be attributable to methylation of the promoter region of the gene (129). The Cables gene is a putative tumor suppressor involved in regulating phosphorylation of cyclin-dependent kinase 2 in a manner that restrains cell cycle progression. Cablesmutant mice develop endometrial hyperplasia at an early age, and exposure to low levels of estrogen causes endometrial cancer (130). Cables expression is up regulated by estrogen and decreases following progestin treatment. Loss of Cables expression also occurs in human endometrial hyperplasias and cancers. Finally, mutations in the CDC4 gene, which is involved in regulating cyclin E expression during cell cycle progression, have been noted in 16% of endometrial cancers (131). Mutations were accompanied by loss of the wild-type allele and were more common in cancers with poor prognostic factors such as high-grade and lymph node metastases. It is postulated that CDC4 may act as a tumor suppressor by restraining the activity of cyclin E in promoting progression from G1 to S phase.
Alterations in oncogenes have been demonstrated in endometrial cancers, but these occur less frequently than inactivation of tumor suppressor genes (Table 1.3). Increased expression of the HER-2/neu receptor tyrosine kinase initially was noted in only 10% of endometrial cancers (99,132,133,134,135) and was associated with advanced stage and poor outcome. Recently, it has been suggested that HER-2/neu overexpression may be more prevalent in patients with papillary serous endometrial cancers (136,137,138). In a tissue microarray study of 483 endometrial cancers using immunohistochemical analysis and fluorescent in situ-hybridization (FISH), the highest rate of HER-2/neu overexpression and amplification was found in serous carcinomas (43% and 29%), whereas grade 1 endometrioid adenocarcinomas showed the lowest levels (3% and 1%) (138). These data also suggest that therapies that target HER-2/neu may have a role in the treatment of papillary serous endometrial carcinomas. The levels of HER-2/neu overexpression in endometrial cancers are much less striking than in breast cancers. With rare exception, trastuzumab (anti-HER-2/neu antibody) generally has not been a useful therapy in endometrial cancer.
The fms oncogene encodes a tyrosine kinase that serves as a receptor for macrophage-colony stimulating factor (M-CSF). Expression of fms in endometrial cancers was found to correlate with advanced stage, poor grade, and deep myometrial invasion (139,140). Subsequently, it was shown that fms and its ligand (M-CSF) usually were coexpressed in endometrial cancers, and it was proposed that this receptor-ligand pair might mediate an autocrine growth stimulatory pathway (141). In support of this hypothesis, M-CSF serum levels are increased in patients with endometrial cancer. In addition, M-CSF increases the invasiveness of cancer cell lines that express significant levels of fms, but it has no effect on cell lines with low levels of the receptor (142).
The ras oncogenes undergo point mutations in codons 12, 13, or 61 that result in constitutively activated molecules in many types of cancers. Initially, these codons of the K-ras, H-ras, and N-ras genes were examined in 11 immortalized endometrial cancer cell lines (143). Mutations in codon 12 of K-ras were seen in four cell lines, whereas three mutations were found in codon 61 of H-ras. Subsequent studies of primary endometrial adenocarcinomas have confirmed that codon 12 of K-ras is mutated in approximately 10% of U.S. cases and 20% of Japanese cases (106,144,145,146,147,148,149,150). These mutations occur more often, but not exclusively, in type I endometrial cancers. K-ras mutations also have been identified in some endometrial hyperplasias (145,150,151), which suggests that this may be a relatively early event in the development of some type I cancers.
As noted previously, the PTEN tumor suppressor gene, which normally acts to restrain PI3K activity, is frequently inactivated in type I endometrial cancers. Conversely, the PIK3CAgene is oncogenically activated in some cases. The catalytic subunit of PI3K (PIK3CA) is located on chromosome 3q26.3, and activating mutations in this gene have been described in several types of cancers. In an initial study, PIK3CA mutations were seen in 36% of endometrial cancers, and 24% of cases had mutations in both PTEN and PIK3CA (152). This suggested that there is an additive effect of two mutations in the same pathway. In a subsequent study, 39% of endometrial cancers and 7% of atypical endometrial hyperplasias were found to harbor mutations in PIK3CA (153). This study also implied that PIK3CA mutation occurred at the time of tumor invasion and could serve as a marker for invasion. As in the initial study, a high fraction of cases had mutations in both PTEN and PIK3CA. These and other studies confirm that PIK3CA activating mutations are common in endometrial cancers. Both inactivation of PTEN or unrestrained PIK3CA can lead to activation of AKT, which in turn leads to up regulation of the mammalian target of rapamycin (mTOR). Recent studies have suggested that mTOR inhibitors may have a role in the in the management of progesterone refractory hyperplasia and treatment of type I endometrial cancer (154,155).
Alterations in the wnt pathway involving E-cadherin, APC, and β-catenin, the product of the CTNNB1 gene, have been noted in some endometrial cancers. E-cadherin is a transmembrane glycoprotein involved in cell-cell adhesion, and decreased expression in cancer cells is associated with increased invasiveness and metastatic potential (Fig. 1.7). E-cadherin mutations occur only rarely in endometrial cancers (51), but cadherin expression may also be down regulated in the absence of mutations (156,157). The cytoplasmic tail of E-cadherin exists as a macromolecular complex with the β-catenin and APC gene products, which link it to the cytoskeleton. It appears that a critical function of the APC tumor suppressor gene is to regulate phosphorylation of serine and threonine residues (codons 33, 37, 41, 45) in exon 3 of β-catenin, which results in degradation of β-catenin. Mutational inactivation of APC allows accumulation of β-catenin, which translocates to the nucleus and acts as a transcription factor to induce expression of cyclin D1 and perhaps other genes involved in cell cycle progression (157).
Germ-line APC mutations are responsible for the adenomatous polyposis coli syndrome, and somatic mutations are common in sporadic colon cancers, but APC mutations have not been described in endometrial cancers (52,158). The APC gene may be inactivated in some endometrial cancers because of promoter methylation. In addition, it has been shown that missense mutations in exon 3 of β-catenin lead to the same end result—namely, abrogation of the ability of APC to induce β-catenin degradation—which results in abnormal transcriptional activity. In view of this, the β-catenin gene is considered an oncogene (159). β-catenin mutations have been observed in several types of cancers, including hepatocellular, prostate, and endometrial cancers. Mutation of β-catenin occurs in approximately 10% to 15% of endometrial cancers, but abnormal accumulation of β-cateninprotein occurs in approximately one-third of cases, suggesting that mechanisms other than mutation might be involved in some cases (53,158).
Mutations have also been observed in the fibroblast growth factor receptor 2 (FGFR2) gene in approximately 10% of endometrial cancers (160). FGFR2 mutations were found almost exclusively in endometrioid cancers, and there was no association with clinical outcome. Further studies will be needed to evaluate the significance of the identified mutations and other mechanisms of increased FGF signaling in endometrial cancer as well as any potential clinical utility by drug targeting of the receptor.
Among nuclear transcription factors involved in stimulating proliferation, amplification of members of the myc family has most often been implicated in the development of human cancers. It has been shown that c-myc is expressed in normal endometrium (161) with higher expression in the proliferative phase. Several studies have suggested that myc may be amplified in a fraction of endometrial cancers (135,162,163).
Approximately 10% of ovarian cancers arise in women who carry germ-line mutations in cancer susceptibility genes—predominantly BRCA1 or BRCA2. The vast majority of ovarian cancers are sporadic and arise because of accumulation of genetic damage.
The causes of acquired genetic alterations in the ovarian epithelium remain uncertain, but exogenous carcinogens, with the possible exception of talc, have not been strongly implicated. Some mutations may arise spontaneously because of increased epithelial proliferation required to repair ovulatory defects. Oxidative stress and free radical formation as a result of inflammation and repair at the ovulatory site may also contribute to accumulation of DNA damage. Regardless of the mechanisms involved, reproductive events that decrease lifetime ovulatory cycles (e.g., pregnancy and birth control pills) are protective against ovarian cancer (164). The protective effect of these factors is greater in magnitude than one would predict based on the extent that ovulation is interrupted, however. Five years of oral contraceptive use provides a 50% risk reduction while only decreasing total years of ovulation by less than 20%. There is evidence to suggest that the progestagenic milieu of pregnancy and the pill might also protect against ovarian cancer by increasing apoptosis of ovarian epithelial cells, thereby cleansing the ovary of cells that have acquired genetic damage (165). The action of other reproductive hormones such as estrogens, androgens, and gonadotropins also may contribute to the development of ovarian cancers.
Epithelial ovarian cancers are heterogeneous with respect to behavior (borderline versus invasive) and histologic type (serous, mucinous, endometrioid, clear cell). Although the strongest epidemiologic risk factors generally affect risk of all disease subsets, differences have been observed with respect to etiology and molecular alterations. For example, although it is thought that serous cancers arise from epithelial cells on the surface of the ovary or in underlying inclusion cysts or in the fallopian tube, many endometrioid and clear cell cancers likely develop in deposits of endometriosis.
Likewise, differences in the pattern of genetic alterations have been noted between histological types (Table 1.4). It has been proposed that ovarian tumors can be classified as low or high grade based on histology, clinical behavior, and molecular phenotypes (166). Low-grade tumors are generally confined to the ovary at diagnosis and include low-grade serous carcinoma, mucinous, endometrioid, and clear cell carcinomas. They are genetically stable and characterized by mutations in a number of genes including K-ras, BRAF, PTEN, and β-catenin. High-grade cancers typically present at an advanced stage and are predominantly serous but also include carcinosarcoma and undifferentiated cancers. This group of tumors has a high level of genetic instability and is characterized by mutation of TP53.
Table 1.4 Clinical and Molecular Characteristics of Histological Types of Ovarian Cancers
As our understanding of the molecular pathogenesis of ovarian cancer continues to mature in the future, it is likely that the various disease subsets will be increasingly thought of as distinct entities that are defined by characteristic patterns of molecular signatures. Elucidation of the molecular basis for the clinical heterogeneity of ovarian cancer has the potential to facilitate future improvements in diagnosis, treatment, and prevention.
Hereditary Ovarian Cancer
It had long been suspected based on epidemiologic and family studies that approximately 10% of epithelial ovarian cancers are attributable to inheritance of mutations in high-penetrance cancer susceptibility genes. The BRCA1 gene was identified on chromosome 17q in 1994, and BRCA2 was identified on chromosome 13q in 1995. Inherited mutations in these two breast and ovarian cancer susceptibility genes are responsible for approximately 6% and 3% of ovarian cancers, respectively (167). Inherited mutations in the DNA mismatch repair genes involved in HNPCC that are described in the section on hereditary endometrial cancer are responsible for some 1% of ovarian cancer cases. The vast majority of BRCA-associated ovarian cancers are papillary serous (168), as are most peritoneal and fallopian tube cancers and some uterine cancers. Although there are conflicting reports regarding whether BRCA mutations increase the risk of serous cancers of the uterus (169,170), the evidence to support inclusion of serous fallopian tube cancers in this syndrome is stronger (171). In two studies, BRCA mutations were found in 28% (172) and 17% (173) of women with fallopian tube cancer. Likewise, germ-line BRCA mutations have been reported in some studies in approximately onethird of those with primary peritoneal cancer (173,174).
The BRCA1 and BRCA2 gene products complex with Rad51 and other proteins involved in repair of double-stranded DNA breaks (DSBs) by homologous recombination (175,176). BRCA1and BRCA2 have been classified as tumor-suppressor genes because the nonmutated copy is invariably deleted in breast and ovarian cancers that arise in women who inherit a mutant gene. In some studies, survival of BRCA carriers with ovarian cancer was better than that of sporadic cases that were matched for age, stage, and other prognostic factors (177). It had been suggested that loss of DNA repair because of mutation of BRCA1 or BRCA2 might improve survival by rendering cancer cells more susceptible to chemotherapy. Poly(ADP-ribose) polymerase (PARP) is an enzyme involved in base excision repair, a key pathway in the repair of DNA single-strand breaks. Inhibition of PARP leads to the persistence of DNA lesions normally repaired by homologous recombination. Inhibitors of PARP recently have been shown to be highly selective for tumor cells with defects in the repair of DSBs by homologous recombination, particularly in the context of BRCA1 or BRCA2 mutation (178). Clinical trials are currently ongoing to determine the efficacy of PARP inhibitors in BRCA mutation carriers with recurrent ovarian cancer.
BRCA1 and BRCA2 mutations are associated with 60% to 90% lifetime risks of breast cancer, and this begins to manifest before age 30. BRCA2 also increases the risk of breast cancer in men. Screening, prophylactic mastectomy, and chemopreventives such as tamoxifen all play a role in decreasing breast cancer incidence and mortality. The lifetime risk of ovarian cancer ranges from 20% to 40% in BRCA1 carriers and 10% to 20% in BRCA2 carriers, but this increased risk is not manifest until the late 30s (179,180,181,182,183). The median age of sporadic epithelial ovarian cancer is in the early to mid-60s, compared to the mid-40s and early 50s for BRCA1- and BRCA2-associated cases.
It is unclear why only a fraction of women who carry BRCA1 mutations develop ovarian cancer. It has been postulated that incomplete penetrance may result from the effect of modifying genes or gene-environment interactions (e.g., birth control pill use) (184). In some series, mutations in the carboxy terminus of BRCA1 have been associated with a higher frequency of breast cancer relative to ovarian cancer (185). Conversely, mutations in the proximal amino end of the gene resulted in a higher likelihood of developing ovarian cancer. Likewise, some studies have suggested that ovarian cancer may occur more often in families with truncation mutations in exon 11 of BRCA2 (186). Further studies are needed to examine whether a genotype-phenotype correlation exists.
BRCA1 and 2 mutations are rare and carried by fewer than one in 500 individuals in most populations, but there are some notable exceptions (187). Founder mutations that presumably arose thousands of years ago in a single ancestor have been identified in some ethnic groups. The most common founder mutations described thus far are the BRCA1185delAG and BRCA2 6174delT mutations that occur in approximately 1.0% and 1.4% of Ashkenazi Jews, respectively (180,188). The high frequency of these mutations implies that they likely arose some 100 generations ago. A third less common founder mutation, BRCA1 5382insC, also has been noted in the Ashkenazi population. Because approximately one in 40 Ashkenazi individuals carries a BRCA founder mutation and testing for this panel of specific mutations is much less expensive, the threshold for genetic testing is much lower in this population.
Because mutations in BRCA1 and BRCA2 in the general population occur throughout the entire coding sequence, the most reliable method of detecting mutations is complete gene sequencing. The effort and cost involved in sequencing these large genes are relatively high, however, and currently it remains impractical to perform mutational analysis in low-risk individuals. The probability of finding a BRCA1 or BRCA2 mutation in a woman over age 50 who is the only individual in her family with ovarian or breast cancer is less than 3%. At the other extreme, in families with two cases of breast cancer and two cases of ovarian cancer, the probability of finding a mutation may be as high as 80% (186). Testing generally has been advocated when the family history suggests at least a 5% probability of finding a mutation. In practical terms, this translates into two first- or second-degree relatives with either ovarian cancer at any age or breast cancer before age 50 (Fig. 1.8). However, some mutations are found in families with less impressive histories, particularly in families that contain few women.
It is preferable to begin by testing individuals in a high-risk family who already have been affected by cancer because a negative test in an unaffected individual may reflect failure to inherit the mutant allele even though others in the family carry a mutation. When a specific mutation is identified in an affected individual, others in the family can be tested much more rapidly and inexpensively for that specific mutation. Most deleterious BRCA mutations encode truncated protein products, but missense mutations that alter a single amino acid have been found to segregate with breast or ovarian cancer in some families (189). In a significant fraction of high-risk families, BRCA testing reveals sequence variants of uncertain significance or no detectable alterations, and these results represent a counseling dilemma. The search for BRCA genetic alterations may also involve sequencing of introns that lie between the coding exons. Intronic mutations may affect RNA splicing and can result in deletion of adjacent exons. In addition, genomic rearrangements may occur that inactivate BRCA1 or 2 and identification of such alterations requires molecular testing beyond sequencing. Failure to identify a BRCA1 or BRCA2 mutation in a family may be reassuring, but it must be tempered by the realization that BRCA mutational analysis may miss some mutations and that other undiscovered hereditary ovarian cancer genes may exist.
Figure 1.8 Familial ovarian cancer pedigree with BRCA1 mutation. The age of family members and type of cancers are noted. Solid circles represent individuals affected with cancer, and slashes denote those who have died of cancer. Individuals denoted M have a mutation in BRCA1, whereas those denoted WT have normal BRCA1genes.
Women should receive educational material and extensive counseling that explains the rationale and potential risks and benefits before they decide to undergo testing. Involvement of genetic counselors in this process is helpful because they have specialized training in nondirective counseling techniques that guide patients toward decisions that reflect their own beliefs and values. In addition, posttest counseling and follow-up are crucial to help women work through various issues, including decisions about prophylactic surgery and other interventions designed to decrease cancer mortality. Because it was feared that misuse of genetic information could have devastating consequences including difficulty in securing employment and life, health, or disability insurance, clinicians initially were hesitant to record genetic testing results in the medical record. Because BRCA testing is now widely accepted and insurance companies generally cover the costs, results should be acknowledged in the medical record chart because they form the basis for the decision to perform prophylactic surgery.
The value of screening for early stage ovarian cancer with CA125 or ultrasound is unproven but seems reasonable until controlled studies are available. Use of birth control pills as a chemopreventive also has been advocated because this is strongly protective against ovarian cancer in the general population. Oral contraceptives are a particularly attractive option for young women who have not yet completed childbearing, but the efficacy of this approach in BRCA carriers remains uncertain (184). Because ovarian cancer has a 70% mortality rate, prophylactic bilateral salpinoophorectomy (BSO) should be discussed with all women who carry germ-line BRCA1 or BRCA2 mutations. Fortunately, the incidence of ovarian cancer in mutation carriers does not begin to rise dramatically until the late 30s (190). This allows women time to bear children before considering BSO as they approach the end of their reproductive life span. The ovaries are internal organs, and most women experience only modest feelings of altered body image and self-esteem after they are removed. Insurance companies will almost always pay for prophylactic BSO in mutation carriers. Highrisk women who lack BRCA mutations or those who have sequence alterations of uncertain significance may encounter greater reimbursement obstacles.
Prophylactic BSO is widely viewed as the most effective currently available means of decreasing ovarian cancer mortality in BRCA mutation carriers. There is strong evidence that this approach significantly decreases ovarian cancer mortality. In addition, prophylactic BSO can be performed laparoscopically in most women with an acceptably low incidence of serious complications. Discussion of the potential for adverse outcomes is particularly important in a setting in which a healthy disease-free individual is subjected to the risks inherent in abdominal surgery.
Patients who undergo prophylactic BSO may experience surgical menopause. In premenopausal women who do not have a personal history of breast cancer, estrogen replacement can be safely administered. Although historically there has been concern that estrogen replacement might increase the risk of breast cancer, this risk already is exceedingly high. In addition, systemic estrogen levels are lower in oophorectomized premenopausal women who are taking hormone replacement than if the ovaries had been left in place. The therapeutic benefit of BSO in women with breast cancer has long been appreciated, and more recent studies support the contention that this intervention significantly reduces breast cancer risk in BRCA carriers (191,192). Many BRCA carriers are identified after developing early onset breast cancer, and this group represents the most difficult in which to balance the risks and benefits of estrogen-replacement therapy.
Many patients elect to have the uterus removed as part of the surgical procedure because they have completed their family. Furthermore, the likelihood of future exposure totamoxifen, which increases endometrial cancer risk two- to threefold, in the context of breast cancer prevention or treatment, also argues for concomitant hysterectomy. The more problematic issue in performing prophylactic BSO is whether the risk of malignant transformation is increased solely in the ovaries and fallopian tubes or in the entire field of mullerian-derived epithelia. Peritoneal papillary serous carcinoma that is indistinguishable histologically or macroscopically from ovarian cancer has been described in rare instances following prophylacticsalpingo-oophorectomy (193,194). These reports preceded the identification of BRCA1 and BRCA2, however, and it is unclear what fraction of these women were mutation carriers. The origin of primary peritoneal cancers after prophylactic BSO is uncertain, and case reports have been published in which retrospective examination of the ovaries has revealed occult ovarian cancers that were not recognized by the pathologist (195). Thus, some cancers thought to originate in the peritoneal cavity may actually represent recurrences of occult ovarian cancer.
Some reports have noted an increased frequency of abnormalities in the ovarian epithelium (invaginations, inclusion cysts, stratification, and papillations) in BRCA carriers (196), but other studies have not confirmed the presence of a consistent pattern of premalignant histologic features (197,198). Careful examination of prophylactic salpingo-oophorectomy specimens has led to the identification of occult cancers in as many as 12% of women in some series (199,200). This adds support to the theory that primary peritoneal cancers that occur years after BSO may represent recurrences of ovarian cancer. Early stage fallopian tube cancers also have been found in BRCA1 carriers undergoing prophylactic BSO (201). Thus, in view of these data, it seems reasonable to recommend that cytologic washings of the pelvis be obtained routinely in concert with prophylactic BSO and. Finally, the pathologist must be informed of the indication for prophylactic BSO and the surgery. Multiple sections of the fallopian tubes and ovaries should be examined to exclude the presence of an occult carcinoma. In this regard, there is some evidence to suggest that the tubal fimbria rather than the ovarian epithelium may be the preferred site of cancer development in BRCA1 and BRCA2 mutation carriers (202).
The efficacy of BSO in reducing breast and ovarian cancer in mutation carriers has been suggested by several retrospective studies. One study that examined the effect of prophylactic BSO revealed a 75% lower rate of breast and ovarian cancer (192). A separate study of 551 BRCA1 and BRCA2 carriers from various registries also found evidence of efficacy (191). Among 259 women who had undergone prophylactic oophorectomy, 2.3% were found to have stage I ovarian cancer at the time of the procedure, and two women subsequently developed papillary serous peritoneal carcinoma. Among controls, 58 women (19.9%) developed ovarian cancer after a mean follow-up of 8.8 years. With the exclusion of the six women whose cancer was diagnosed at surgery, prophylactic oophorectomy reduced the risk of coelomic epithelial cancer by 96%. Most recently, an international registry study of more than 1,800 subjects with median follow-up of 3.5 years found that prophylactic BSO reduced ovarian, tubal, and peritoneal cancer risk by only 80%, partly because of an estimated 6% residual lifetime risk of primary peritoneal cancer (203).
Sporadic Ovarian Cancer
Global Genomic Changes
Invasive epithelial ovarian carcinoma generally is a monoclonal disease that develops as a clonal expansion of a single transformed cell in the ovary (204). There is evidence, however, that some serous borderline tumors (134) as well as cancers that arise in the peritoneum of patients with BRCA1 mutations may be polyclonal (205). Most ovarian cancers are characterized by a high degree of genetic damage that is manifest at the genomic and molecular levels. Gains and losses of various segments of the genome have been demonstrated using comparative genomic hybridization (206). Likewise, loss of heterozygosity (LOH), indicative of deletion of specific genetic loci, also has been demonstrated to occur at a high frequency on many chromosomal arms (207). It is unclear whether the wide range of genetic alterations in ovarian cancers reflects the need to alter several genes in the process of malignant transformation or results from generalized genomic instability. Both CGH and LOH studies have shown that advanced stage, poorly differentiated cancers have a higher number of genetic changes than early stage, low-grade cases (208,209,210).
It is estimated that the human genome contains some 25,000 genes. Microarray chips that contain sequences complementary to thousands of genes have been created that allow global assessment of the level of expression of each gene. Expression arrays have proven useful in predicting clinical phenotypes in several types of solid tumors. Many genes have been identified that appear to be up or down regulated in the process of malignant transformation (211,212). In addition, microarrays have demonstrated patterns of gene expression that distinguish between histologic types (213), borderline and invasive cases (214) and between early and advanced stage disease (212,215). Molecular signatures also have been identified that are predictive of response to therapy (216, 217) and survival (218). Further validation of genomic signatures is needed, but genomic approaches hold the exciting potential to guide selection of therapy in the future. Patients identified as having a “poor prognosis” molecular profile might be the best candidates for investigational trials of new therapies. Microarray analyses have also demonstrated that dysregulation of oncogenic molecular pathways varies considerably between ovarian cancers (219). This may provide the opportunity to incorporate biological therapies that target oncogenic pathways associated with src, ras, EGFR, and others based on an understanding of the underlying molecular alterations in a patient's cancer (216).
Tumor Suppressor Genes
Alteration of the TP53 tumor suppressor gene is the most frequent genetic event described thus far in ovarian cancers (Table 1.4) (220,221,222,223,224,225). The frequency of overexpression of mutant p53 is significantly higher in advanced stage (40% to 60%) relative to early stage cases (10% to 20%). The histologic distribution of early and advanced stage cases varies significantly, however, and may account for the difference in TP53 mutation rate. In this regard, approximately two-thirds of early stage serous ovarian cancers were found to have TP53 mutations compared to only 21% of nonserous cases (226). There is a suggestion that overexpression of mutant p53 protein may be associated with slightly worse survival in advanced stage ovarian cancers (220,222,224,225,227,228,229,230). Finally, although there is a high concordance between TP53 missense mutations in exons 5 through 8 and protein overexpression, approximately 20% of advanced ovarian cancers contain mutations that result in truncated protein products, which usually are not overexpressed (221,230). Some of these mutations may lie outside of exons 5 to 8. Overall, some 70% of advanced ovarian cancers have either missense or truncation mutations in the TP53 gene.Most TP53 missense mutations are transitions rather than transversions or microdeletions (231,232), which suggests that these mutations occur spontaneously rather than resulting from exogenous carcinogens.
It has been suggested that loss of functional p53 might confer a chemoresistant phenotype because of its role in chemotherapy-induced apoptosis. In this regard, several studies have examined the correlation between chemosensitivity and TP53 mutation in ovarian cancers (233,234,235,236). Some have suggested a relationship between p53 mutation and loss of chemosensitivity and resistance to therapy, but in other studies such a relationship has not been observed. It is likely that the status of the TP53 gene is just one of a multitude of factors that determine chemosensitivity.
Overexpression of p53 is rare in stage I serous borderline tumors and well-differentiated serous cancers, but it does occur in approximately 20% of advanced stage borderline cases (237,238). In a study of advanced serous borderline tumors, p53 overexpression was associated with a sixfold higher risk of death (238). In some cases, invasive serous cancers may arise following an earlier diagnosis of borderline tumor. Ortiz et al. showed that TP53 mutational status was not concordant between the original borderline tumor and the subsequent invasive cancer (239). This suggests that the invasive cancer either arises independently or as a clonal outgrowth within the original tumor.
Although mutations in the Rb tumor suppressor gene are not a common feature of ovarian cancers, recent evidence suggests that inactivation of Rb greatly enhances tumor formation in ovarian cells with p53 mutations (240). In a mouse model in which these genes were inactivated in the ovarian epithelium, few cancers developed in response to loss of either TP53 or Rb alone. When both genes were inactivated, epithelial ovarian cancers with serous features developed in almost all cases. Given that Rb mutations are rare in ovarian cancers, it is possible that inactivation of one of a number of genes in the Rb pathway can initiate transformation cooperatively with TP53. Inactivation of Rb itself may not be requisite. This mouse model of ovarian cancer has the potential to add greatly to our understanding of epithelial ovarian carcinogenesis.
The cyclin-dependent kinase (cdk) inhibitors act as tumor suppressors by virtue of their inhibition of cell cycle progression from G1 to S phase. Expression of several cdk inhibitors appears to be decreased in some ovarian cancers. In approximately 15% of ovarian cancers, p16 undergoes homozygous deletions (241). There is evidence to suggest that p16 (242), CDKN2B (p15) (243), and some other tumor suppressor genes such as BRCA1 (244,245) may be inactivated via transcriptional silencing because of promoter methylation rather than mutation or deletion. Likewise, decreased expression of the p21 cdk inhibitor has been noted in a significant fraction of ovarian cancers despite the absence of inactivating mutations (246,247). Loss of CDKN1B (p27) also may occur and correlates with poor survival in some studies (248,249,250,251). It has been suggested that abberant expression of p27in the cytoplasm may be most associated with poor outcome (252).
Normal ovarian epithelial cells are inhibited by the growth inhibitory peptide TGF-β, whereas most immortalized ovarian cancer cell lines are unresponsive (253,254). The effect of TGF-β on primary ovarian cancer cells obtained directly from patients is less straightforward. In studies conducted using ovarian cancer cells grown in monolayer culture, most remain sensitive to the growth inhibitory effect of TGF-β (254). In contrast, when ovarian cancer cells are grown in collagen matrix, they are unresponsive (255). There is some evidence that mutations may occur in cell surface TGF-β receptors or in the Smad family of genes that are involved in downstream signaling (256); in other studies, these signaling pathways were found to be intact (255). Thus far, it has not been convincingly demonstrated that derangement of the TGF-β pathway plays a role in the development of ovarian cancers.
Ovarian cancers produce and are capable of responding to various peptide growth factors. For example, epidermal growth factor (257) and transforming growth factoralpha (TGF-α)(258) are produced by some ovarian cancers that also express the receptor that binds these peptides (EGF receptor) (259,260). Some cancers produce insulin-like growth factor-1 (IGF-1), IGF-1 binding protein, and express type 1 IGF receptor (261). Plateletderived growth factor also is expressed by many types of epithelial cells, including human ovarian cancer cell lines, but these cells usually are not responsive to PDGF (253,262,263). In addition, ovarian cancers produce basic fibroblast growth factor and its receptor, and basic FGF acts as a mitogen in some ovarian cancers (264). Ovarian cancers produce macrophagecolony stimulating factor, and serum levels of M-CSF are elevated in some patients (265). Because the M-CSF receptor (fms) is expressed by many ovarian cancers, this could constitute an autocrine growth stimulatory pathway in some cancers.(266). Ascites of patients with ovarian cancer also contains phospholipid factors such as LPA that stimulate proliferation and invasiveness of ovarian cancer cells (267). The edg-2 G-protein coupled receptors act as functional receptors for LPA. The finding that neutralization of LPA activity decreases growth and increases apoptosis of ovarian cancers suggests that manipulation of this pathway may be therapeutically beneficial (268).
Several groups also have demonstrated that normal ovarian epithelial cells produce, and are responsive to, many of the same peptide growth factors as malignant ovarian epithelial cells (260,269,270,271). Thus, despite cell culture data demonstrating autocrine and paracrine growth regulation of ovarian cancer cells by peptide growth factors, it remains unclear whether alterations in expression of growth factors are critical early events involved in the development of ovarian cancers. Alternatively, it is possible that growth factors may primarily act as “necessary but not sufficient” cofactors that support growth and metastasis following malignant transformation.
The HER-2/neu tyrosine kinase is a member of a family of related transmembrane receptors that includes the EGF receptor (272). Approximately 30% of breast cancers express increased levels of the HER-2/neu (273), often as a result of gene amplification. Overexpression of HER-2/neu in breast cancer has been associated with poor survival. Expression of HER-2/neu is increased in a fraction of ovarian cancers and overexpression has been associated with poor survival in some studies (273,274), but not all (275). Unlike breast cancers, ovarian cancers that exhibit HER-2/neu overexpression rarely have high-level gene amplification. Monoclonal antibodies that interact with HER-2/neu can decrease growth of breast and ovarian cancer cell lines that overexpress this receptor (276,277). Anti-HER-2/neu antibody therapy (trastuzumab) has demonstrated efficacy in the treatment of breast cancer and often is administered in concert with paclitaxel. A study performed by the Gynecologic Oncology Group found only 11% of ovarian cancers exhibit significant HER-2/neuoverexpression (278). The response rate to single-agent trastuzumab therapy was disappointingly low (7%), but perhaps some benefit may be found in the future using combination regimens that also include taxanes or other cytotoxics.
Activating mutations in codons 12 and 13 the K-ras gene are rare in high-grade invasive serous ovarian cancers (279,280). Some types of cancers that lack K-ras mutations have activating mutations in codon 599 of the downstream BRAF gene, but this is not the case in high-grade serous ovarian cancers. In contrast, K-ras mutations are common in borderline serous ovarian tumors, occurring in approximately 25% to 50% of cases (281). In addition, mutations in BRAF occur in some 20% of serous borderline cases lacking K-ras mutations(282). Mutations in K-ras and BRAF have also been noted in cystadenoma epithelium adjacent to serous borderline tumors, suggesting that this is an early event in their development (25). K-ras mutations have been noted in approximately 50% of mucinous ovarian cancers, butBRAF mutations have not been found (283). These findings highlight the distinct differences in the molecular pathology between various histological types and between borderline tumors and invasive ovarian cancers (Table 1.4).
Similar to endometrial cancers, activation of the PIK3CA and AKT2 oncogenes occurs in some ovarian cancers. The region of chromosome 3p26 that includes the phosphatidylinositol 3-kinase (PIK3CA) is amplified in some ovarian cancers (284). In addition, activating mutations in PIK3CA occur in about 10% of ovarian cancers, and are much more common in endometrioid and clear cell cancers (20%) as compared to serous cancers (2%) (285). Likewise, the AKT2 serine/threonine kinase that is downstream of PIK3CA also has been shown to be amplified and overexpressed in some ovarian cancers (286). PIK3CA and AKT2 kinase activity is opposed by the PTEN phosphatase, and this tumor suppressor gene also is inactivated in about 20% of endometrioid ovarian cancers (287).
Mutations in the β-catenin gene are a feature of some endometrial cancers. Similarly, β-catenin mutations are present in some 30% of endometrioid ovarian cancers (288) but not other histologic types. This provides further evidence of the molecular heterogeneity of the various histologic types of ovarian cancer (Table 1.4). In some endometrioid ovarian cancers with abnormal nuclear accumulation of β-catenin that lacked mutations, the APC, AXIN1, or AXIN2 genes that regulate β-catenin activity were found to be mutated (288). This suggests that alterations in the wnt signaling pathway are a feature of endometrioid ovarian cancers. Mouse models in which the wnt and the PIK3/PTEN pathways are inactivated in the ovarian epithelium leads to the development of endometrioid cancer and endometriosis (289,290).
Increased activity of nuclear transcription factors and cyclins also may enhance malignant transformation. In this regard, amplification of the c-myc oncogene has been reported to occur in approximately 30% of ovarian cancers. Several studies have suggested that the c-myc gene is amplified in approximately 30% of cases (291,292,293,294). The overexpression of c-myc is observed most often in advanced stage serous cancers. Despite these reports of gene amplification, convincing evidence of c-myc protein overexpression has been less convincing. Some ovarian cancers have been reported to have increased expression of cyclin E, which is involved in cell cycle progression (295). In a study of advanced stage suboptimally debulked ovarian cancers, high cyclin E expression was associated with a six-month decrease in median survival (296). In some, but not all, cases, amplification of the cyclin E gene was found to be the underlying cause of overexpression. In a large study using a tissue array, cyclin E overexpression has been was shown to be associated with serous and clear cell histology, advanced stage, and poor outcome (297).
Cervical cancer is the most common gynecologic malignancy worldwide and accounts for more than 400,000 cases annually. Molecular and epidemiologic studies have demonstrated that sexually transmitted human papilloma virus (HPV) infections play a role in almost all cervical dysplasias and cancers (298). HPV infection also is involved in the development of dysplasias and cancers of the vagina and vulva. The peak incidence of HPV infection is in the 20s and 30s, and the incidence of cervical cancer increases from the 20s to a plateau between ages 40 and 50. Although HPV plays a major role in the development of most cervical cancers, only a small minority of women who are infected develop invasive cervical cancer. This suggests that other genetic or environmental factors also are involved in cervical carcinogenesis. For example, individuals who are immunosuppressed because of either HIV infection (299) or immunosuppressive drugs are more likely to develop dysplasia and invasive cervical cancer following HPV infection.
Cervical screening programs in developed nations have dramatically reduced both the incidence of invasive cervical cancer and disease-related mortality. The recent development of vaccines against oncogenic HPV subtypes has the potential to further decrease the incidence of cervical dysplasia and cancer (300). Although cervical cancer mortality is low in the United States and Western Europe, it remains among the leading causes of cancer deaths in women in underdeveloped nations.
Human Papilloma Virus Infection
There are more than 100 HPV subtypes, but not all infect the lower genital tract. HPV 16 and 18 are the most common types associated with cervical cancer and are found in more than 80% of cases. Types 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, 73, and 82 should be considered high-risk types, and types 26, 53, and 66 should be considered probably carcinogenic (298). Low-risk types that may cause dysplasias or condyloma in the lower genital tract, but rarely cause cancers, include types 6, 11, 40, 42, 43, 44, 54, 61, 70, 72, and 81. The advent of HPV typing now allows assessment of whether patients carry high-risk or low-risk HPV types, and this has proven clinically useful in the management of patients with low-grade Pap smear abnormalities.
The HPV DNA sequence consists of 7,800 nucleotides divided into “early” and “late” open reading frames (ORFs). Early ORFs are within the first 4,200 nucleotides of the genome and encode proteins (E1-E8) that are important in viral replication and cellular transformation. Late ORFs (L1 and L2) are found within the latter half of the sequence and encode structural proteins of the virion. In oncogenic subtypes such as HPV 16 and 18, transformation may be accompanied by integration of episomal HPV DNA into the host genome. Opening of the episomal viral genome usually occurs in the E1-E2 region, resulting in a linear fragment for insertion. The location of the opening may be significant because E2 acts as a repressor of the E6-E7 promoter, and disruption of E2 can lead to unregulated expression of the E6/E7 transforming genes. HPV 16 DNA may be found in its episomal form in some cervical cancers, however, and unregulated E6-E7 transcription may occur independently of viral DNA integration into the cellular genome.
Examination of the biological effects of HPV-encoded proteins has shed light on the mechanisms of HPV-associated transformation. Expression of the E4 transcript results in the production of intermediate filaments that colocalize with cytokeratins. E4 proteins of oncogenic subtypes disrupt the cytoplasmic cytokeratin matrix, whereas those of nononcogenic strains do not. It has been suggested that this may facilitate the release of HPV particles in oncogenic subtypes such as HPV 16. The E5 oncogene encodes a 44-amino acid protein that usually forms dimers within the cellular membrane. The transforming properties of E5 appear to involve potentiation of membrane-bound epidermal growth factor receptors or platelet factor growth receptors. The E6 and E7 oncoproteins are the main transforming genes of oncogenic strains of HPV (Fig. 1.9) (301). Transfection of these genesin vitro results in immortalization and transformation of some cell lines. The HPV E7 protein acts primarily by binding to and inactivating the Rb tumor suppressor gene product. E7 contains two domains, one of which mediates binding to Rb while the other serves as a substrate for casein kinase II (CKII) phosphorylation. Variations in oncogenic potential between HPV subtypes may be related to differences in the binding efficacy of E7 to Rb. High-risk HPV types contain E7 oncoproteins that bind Rb with more affinity than E7 from low-risk types. The transforming activity of E7 may be increased by CKII mutation, implying a role for this binding site in the development of HPV-mediated neoplasias.
The E6 proteins of oncogenic HPV subtypes bind to and inactivate the TP53 tumor suppressor gene product (302,303). There also is a correlation between oncogenicity of various HPV strains and the ability of their E6 oncoproteins to inactivate p53. Inactivation of Rb and p53 by E6 and E7 circumvents the need for mutational inactivation of these key growth regulatory genes.
Figure 1.9 Neutralization of p53 and Rb by HPV 16/18 in cervical cancer.
HPV-negative cervical cancers are uncommon but have been reported to exhibit overexpression of mutant p53 protein (304). This suggests that inactivation of the p53 tumor suppressor gene either by HPV E6 or by mutation is a requisite event in cervical carcinogenesis. In some studies, the levels of E6 and E7 in invasive cervical cancers have been found to predict outcome, whereas HPV viral load does not (305).
Comparative genomic hybridization techniques have been used to identify chromosomal loci that are either increased or decreased in copy number in cervical cancers. A strikingly consistent finding of various studies is the high frequency of gains on chromosome 3q in both squamous cell cancers (306,307) and adenocarcinomas (308). Other chromosomes that exhibit frequent gains include 1q and 11q. The most common areas of chromosomal loss include chromosomes 3p and 2q. For the most part, with the exception of the fragile histidine triad (FHIT) gene on chromosome 3p, it has not been proven that these genomic gains and losses result in the recruitment of specific oncogenes and tumor suppressor genes in the process of malignant transformation. It is conceivable that these chromosomal alterations may be frequent sequelae of infection with oncogenic HPVs while playing no significant role in the pathogenesis of cervical cancers. Abnormalities seen in invasive cancers using comparative genomic hybridization also have been identified in high-grade dysplasias, however, suggesting that these are early events in cervical carcinogenesis (307,309,310).
Oncogenes and Tumor Suppressor Genes
Only a small fraction of HPV-infected women develop cervical cancer. This suggests that additional genetic alterations are requisite for progression to high-grade dysplasia and cancer, but little is known regarding these events. Allele loss suggestive of involvement of tumor suppressor genes has been noted at loci on chromosomes 3p, 11p, and others. It is striking that the cyclindependent kinase inhibitor p16 is up regulated in almost all cervical dysplasias and cancers (311). Clinical trials are ongoing to determine whether this will represent a useful adjunct to improve the positive predictive value of high-risk HPV testing for detection of cervical dysplasia.
The role of several oncogenes has been examined in cervical carcinomas, most prominently in the ras and myc genes. Mutant ras genes are capable of cooperating with HPV in transforming cells in vitro. There is some evidence that mutations in either K-ras or H-ras may play a role in a subset of cervical cancers (304,312,313,314,315). Alterations in rasgenes have not been seen in cervical intraepithelial neoplasia, suggesting that mutation of ras is a late event in the pathogenesis of some cervical cancers. In contrast, c-mycamplification and overexpression may be an early event in the development of some cervical cancers (316). Overexpression of c-myc has been demonstrated in one-third of early invasive carcinomas and some CIN 3 lesions, but not in normal cervical epithelium or lower-grade dysplasia. Overexpression of c-myc gene may result from amplification of the gene (four- to 20-fold) in some cases. In some studies, amplification correlated with poor prognosis in early stage cases (317). Other studies have not confirmed the finding of amplification of c-myc in cervical cancers, however. Integration of the HPV genome near c-myc on chromosome 8q may lead to increased expression because of enhanced transcription of the gene rather than amplification. Further studies are needed to clarify the role of ras genes, c-myc, and other oncogenes in cervical carcinogenesis.
The fragile histidine triad gene localized within human chromosomal band 3p14.2 is frequently deleted in many different cancers, including cervical cancer (318,319,320). Decreased expression of this putative tumor suppressor gene is an early event in some cervical cancers (320,321). In one study, FHIT protein expression was markedly reduced or absent in 71% of invasive cancers, 52% of high-grade squamous intraepithelial lesions (HSILs) associated with invasive cancer, and 21% of HSILs without associated invasive cancer (320). In addition, reduced expression is associated with poor prognosis in advanced cervical cancers (322).
As is the case in endometrial and ovarian cancers, it is thought that gene silencing resulting from promoter hypermethylation also may play a role in cervical carcinogenesis(323,324). In this regard, the RAS association domain family 1A (RASSF1A) gene is located on chromosome 3p21.3 in an area that is frequently a site of deletions in cervical cancer. The function of this gene is not completely understood, but it is thought to be involved in ras-mediated signal transduction pathways. Although mutations in RASSF1A do not occur in cervical cancers, inactivation of the gene because of promoter methylation occurs in a fraction of cases, particularly adenocarcinomas (325,326).
Hypermethylation of genes associated with programmed cell death (apoptosis) or tumor suppressor genes have also been described in association with cervical cancer. Likewise, hypermethylation of HPV DNA that has been integrated into the host genome may also play a role in suppressing the transformation associated with viral oncogenes until other molecular alterations overcome this method of epigenetic silencing (327).
Gestational Trophoblastic Disease
The genetic alterations that underlie gestational trophoblastic disease have been elucidated to a great extent. The most prominent feature of these tumors is an imbalance of parental chromosomes. In the case of partial moles, this involves an extra haploid copy of one set of paternal chromosomes, while complete moles generally are characterized by two complete haploid sets of paternal chromosomes and an absence of maternal chromosomes. Although the risk of repeat molar pregnancy is only approximately 1%, women who have had two molar pregnancies have an approximate 25% risk of developing another mole. Although this suggests a hereditary defect that affects gametogenesis, this remains speculative. Thus far, there is no convincing evidence that damage to specific tumor suppressor genes or oncogenes contributes to the development of gestational trophoblastic disease. However, recent microarray studies have identified several genes that are differentially expressed compared to those found in normal villi, particularly genes associated with cellular apoptosis, immunosuppression, and cell invasion (328,329).
Editors: Berek, Jonathan S.; Hacker, Neville F.