Cancer is one of the most common and serious diseases seen in clinical medicine. There are 14 million new cases of cancer diagnosed each year and over 8 millions deaths from the disease worldwide. Based on the most recent statistics available, cancer treatment costs $80 billion per year in direct health care expenditures in the United States alone. Cancer is invariably fatal if it is not treated. Identification of persons at increased risk for cancer before its development is an important objective of genetics research. And for both those with an inherited predisposition to cancer as well those in the general population, early diagnosis of cancer and its early treatment are vital, and both are increasingly reliant on advances in genome sequencing and gene expression analysis.
Cancer is the name used to describe the more virulent forms of neoplasia, a disease process characterized by uncontrolled cellular proliferation leading to a mass or tumor (neoplasm). The abnormal accumulation of cells in a neoplasm occurs because of an imbalance between the normal processes of cellular proliferation and cellular attrition. Cells proliferate as they pass through the cell cycle and undergo mitosis. Attrition, due to programmed cell death (see Chapter 14), removes cells from a tissue. For a neoplasm to be a cancer, however, it must also be malignant, which means that not only is its growth uncontrolled, it is also capable of invading neighboring tissues that surround the original site (the primary site) and can spread (metastasize) to more distant sites (Fig. 15-1). Tumors that do not invade or metastasize are not cancerous but are referred to as benign tumors, although their abnormal function, size or location may make them anything but benign to the patient.
FIGURE 15-1 General scheme for development of a carcinoma in an epithelial tissue such as colonic epithelium. The diagram shows progression from normal epithelium to local proliferation, invasion across the lamina propria, spread to local lymph nodes, and final distant metastases to liver and lung.
Cancer is not a single disease but rather comes in many forms and degrees of malignancy. There are three main classes of cancer:
• Sarcomas, in which the tumor has arisen in mesenchymal tissue, such as bone, muscle, or connective tissue, or in nervous system tissue;
• Carcinomas, which originate in epithelial tissue, such as the cells lining the intestine, bronchi, or mammary ducts; and
• Hematopoietic and lymphoid malignant neoplasms, such as leukemia and lymphoma, which spread throughout the bone marrow, lymphatic system, and peripheral blood.
Within each of the major groups, tumors are classified by site, tissue type, histological appearance, degree of malignancy, chromosomal aneuploidy, and, increasingly, by which gene mutations and abnormalities in gene expression are found within the tumor.
In this chapter, we describe how genetic and genomic studies demonstrate that cancer is fundamentally a genetic disease. We describe the kinds of genes that have been implicated in initiating cancer and the mechanisms by which dysfunction of these genes can result in the disease. Second, we review a number of heritable cancer syndromes and demonstrate how insights gained into their pathogenesis have illuminated the basis of the much more common, sporadic forms of cancer. We also examine some of the special challenges that such heritable syndromes present for medical genetics and genetic counseling. Third, we illustrate ways in which genetics and genomics have changed both how we think about the causes of cancer and how we diagnose and treat the disease. Genomics—in particular the identification of mutations, altered epigenomic modifications, and abnormal gene expression in cancer cells—is vastly expanding our knowledge of why cancer develops and is truly changing cancer diagnosis and treatment.
Genetic Basis of Cancer
Driver and Passenger Gene Mutations
The application to the study of cancer of powerful new sequencing technologies for genome sequencing (see Chapter 4) and RNA expression studies (see Chapter 3) has brought remarkable new clarity to our understanding of the origins of cancer. By analyzing many thousands of samples obtained from more than 30 types of human cancer, researchers are building The Cancer Genome Atlas, a public catalog of mutations, epigenomic modifications, and abnormal gene expression profiles found in a wide variety of cancers. Although the project is still under way, the results to date from these studies are striking. The number of mutations present in a tumor can vary from just a few to many tens of thousands. Most mutations found through sequencing of tumor tissue appear to be random, are not recurrent in particular cancer types, and probably occurred as the cancer developed, rather than directly causing the neoplasia to develop or progress. Such mutations are referred to as “passenger” mutations. However, a subset of a few hundred genes has been repeatedly found to be mutated at high frequency in many samples of the same type of cancer or even in multiple different types of cancers, mutated in fact far too frequently to simply be passenger mutations. These genes are thus presumed to be involved in the development or progression of the cancer itself and are therefore referred to as “driver” genes, that is, they harbor mutations (so-called driver gene mutations) that are likely to be causing a cancer to develop or progress. Although many driver genes are specific to particular tumor types, some, such as those in the TP53 gene encoding the p53 protein, are found in the vast majority of cancers of many different types. Although the most common driver genes are now known, it is likely that additional, less abundant driver genes will be identified as The Cancer Genome Atlas continues to grow.
Spectrum of Driver Gene Mutations
Many different genome alterations can act as driver gene mutations. In some cases, a single nucleotide change or small insertion or deletion can be a driver mutation. Large numbers of cell divisions are required to produce an adult organism of an estimated 1014 cells from a single-cell zygote. Given a frequency of 10−10 replication errors per base of DNA per cell division, and an estimated 1015 cell divisions during the lifetime of an adult, replication errors alone result in thousands of new single nucleotide or small insertion/deletion mutations in the genome in every cell of the organism. Some environmental agents, such as carcinogens in cigarette smoke or ultraviolet or X-irradiation, will increase the rate of mutations around the genome. If, by chance, mutations occur in critical driver genes in a particular cell, then the oncogenic process may be initiated.
Chromosome and subchromosomal mutations (see Chapters 4 and 5) can also serve as driver mutations. Particular translocations are sometimes highly specific for certain types of cancer and involve specific genes (e.g., the BCR-ABL translocation in chronic myelogenous leukemia) (Case 10); in contrast, other cancers can show complex rearrangements in which chromosomes break into numerous pieces and rejoin, forming novel and complex combinations (a process known as “chromosome shattering”). Finally, large genomic alterations involving many kilobases of DNA can form the basis for loss of function or increased function of one or more driver genes. Large genomic alterations include deletions of a segment of a chromosome or multiplication of a chromosomal segment to produce regions with many copies of the same gene (gene amplification).
The Cellular Functions of Driver Genes
The nature of some driver gene mutations comes as no surprise: the mutations directly affect specific genes that regulate processes that are readily understood to be important in oncogenesis. These processes include cell-cycle regulation, cellular proliferation, differentiation and exit from the cell cycle, growth inhibition by cell-cell contacts, and programmed cell death (apoptosis). However, the effects of other driver gene mutations are not so readily understood and include genes that act more globally and indirectly affect the expression of many other genes. Included in this group are genes encoding products that maintain genome and DNA integrity or genes that affect gene expression, either at the level of transcription by epigenomic changes, at the post-transcriptional level through effects on messenger RNA (mRNA) translation or stability, or at the post-translational level through their effects on protein turnover (Table 15-1). Other driver genes affect translation, for example, genes that encode noncoding RNAs from which regulatory microRNAs (miRNAs) are derived (see Chapter 3). Many miRNAs have been found to be either greatly overexpressed or down-regulated in various tumors, sometimes strikingly so. Because each miRNA may regulate as many as 200 different gene targets, overexpression or underexpression of miRNAs may have widespread oncogenic effects because many driver genes will be dysregulated. Noncoding miRNAs that impact gene expression and contribute to oncogenesis are referred to as oncomirs.
Classes of Driver Genes Mutated in Cancer
mRNA, Messenger RNA; mTOR, mammalian target of rapamycin; PI3, phosphatidylinositol-3.
Figure 15-2 is a diagram outlining how mutations in specific regulators of growth and in global guardians of DNA and genome integrity perturb normal homeostasis (see Fig. 15-2A), leading to a vicious cycle causing loss of cell cycle control, uncontrolled proliferation, interrupted differentiation, and defects in apoptosis (see Fig. 15-2B).
FIGURE 15-2 A, Overview of normal genetic pathways controlling normal tissue homeostasis. The information encoded in the genome (black arrows) results in normal gene expression, as modulated by the epigenomic state. Many genes provide negative feedback (purple arrows) to ensure normal homeostasis. B, Perturbations in neoplasia. Abnormalities in gene expression (dotted black arrows) lead to a vicious cycle of positive feedback (brown dotted lines) of progressively more disordered gene expression and genome integrity.
Activated Oncogenes and Tumor Suppressor Genes
Both classes of driver genes—those with specific effects on cellular proliferation or survival and those with global effects on genome or DNA integrity (see Table 15-1)—can be further subdivided into one of two functional categories depending on how, if mutated, they drive oncogenesis.
The first category includes proto-oncogenes. These are normal genes that, when mutated in very particular ways, become driver genes through alterations that lead to excessive levels of activity. Once mutated in this way, driver genes of this type are referred to as activated oncogenes. Only a single mutation at one allele can be sufficient for activation, and the mutations that activate a proto-oncogene can range from highly specific point mutations causing dysregulation or hyperactivity of a protein, to chromosome translocations that drive overexpression of a gene, to gene amplification events that create an overabundance of the encoded mRNA and protein product (Fig. 15-3).
FIGURE 15-3 Different mutational mechanisms leading to proto-oncogene activation. These include a single point mutation leading to an amino acid change that alters protein function, mutations or translocations that increase expression of an oncogene, a chromosome translocation that produces a novel product with oncogenic properties, and gene amplification leading to excessive amounts of the gene product.
The second, and more common, category of driver genes includes tumor suppressor genes (TSGs), mutations in which cause a loss of expression of proteins necessary to control the development of cancers. To drive oncogenesis, loss of function of a TSG typically requires mutations at both alleles. There are many ways that a cell can lose the function of TSG alleles. Loss-of-function mechanisms can range from missense, nonsense, or frame-shift mutations to gene deletions or loss of a part or even an entire chromosome. Loss of function of TSGs can also result from epigenomic transcriptional silencing due to altered chromatin conformation or promoter methylation (see Chapter 3), or from translational silencing by miRNAs or disturbances in other components of the translational machinery (see Box).
Genetic Basis of Cancer
Regardless of whether a cancer occurs sporadically in an individual, as a result of somatic mutation, or repeatedly in many individuals in a family as a hereditary trait, cancer is a genetic disease.
• Genes in which mutations cause cancer are referred to as driver genes, and the cancer-causing mutations in these genes are driver mutations.
• Driver genes fall into two distinct categories: activated oncogenes and tumor suppressor genes (TSGs).
• An activated oncogene is a mutant allele of a proto-oncogene, a class of normal cellular protein-coding genes that promotes growth and survival of cells. Oncogenes facilitate malignant transformation by stimulating proliferation or inhibiting apoptosis. Oncogenes encode proteins such as the following:
• Proteins in signaling pathways for cell proliferation
• Transcription factors that control the expression of growth-promoting genes
• Inhibitors of programmed cell death machinery
• A TSG is a gene in which loss of function through mutation or epigenomic silencing directly removes normal regulatory controls on cell growth or leads indirectly to such losses through an increased mutation rate or aberrant gene expression. TSGs encode proteins involved in many aspects of cellular function, including maintenance of correct chromosome number and structure, DNA repair proteins, proteins involved in regulating the cell cycle, cellular proliferation, or contact inhibition, just to name a few examples.
• Tumor initiation can be caused by different types of genetic alterations. These include mutations such as the following:
• Activating or gain-of-function mutations, including gene amplification, point mutations, and promoter mutations, that turn one allele of a proto-oncogene into an oncogene
• Ectopic and heterochronic mutations (see Chapter 11) of proto-oncogenes
• Chromosome translocations that cause misexpression of genes or create chimeric genes encoding proteins with novel functional properties
• Loss of function of both alleles, or a dominant negative mutation of one allele, of TSGs
• Tumor progression occurs as a result of accumulating additional genetic damage, through mutations or epigenetic silencing, of driver genes that encode the machinery that repairs damaged DNA and maintains cytogenetic normality. A further consequence of genetic damage is altered expression of genes that promote vascularization and the spread of the tumor through local invasion and distant metastasis.
Cellular Heterogeneity within Individual Tumors
The accumulation of driver gene mutations does not occur synchronously, in lockstep, in every cell of a tumor. To the contrary, cancer evolves along multiple lineages within a tumor, as chance mutational and epigenetic events in different cells activate proto-oncogenes and cripple the machinery for maintaining genome integrity, leading to more genetic changes in a vicious cycle of more mutations and worsening growth control. The lineages that experience an enhancement of growth, survival, invasion, and distant spread will come to predominate as the cancer evolves and progresses (see Box). In this way, the original clone of neoplastic cells evolves and gives rise to multiple sublineages, each carrying a set of mutations and epigenomic alterations that are different from but overlap with what is carried in other sublineages. The profile of mutations and epigenomic changes can differ between the primary and its metastases, between different metastases, and even between the cells of the original tumor or within a single metastasis. A paradigm for the development of cancer, as illustrated in Figure 15-4, provides a useful conceptual framework for considering the role of genomic and epigenomic changes in the evolution of cancer, a point we emphasize throughout this chapter. It is a general model that applies to all cancers.
FIGURE 15-4 Stages in the evolution of cancer. Increasing degrees of abnormality are associated with sequential loss of tumor suppressor genes from several chromosomes and activation of proto-oncogenes, with or without a concomitant defect in DNA repair. Multiple lineages, carrying different mutations and epigenomic profiles, occur within the primary tumor itself, between the primary and metastases and between different metastases.
Although the focus of this chapter is on genomic and epigenomic changes within the tumor, the surrounding normal tissue also plays an important role by providing the blood supply that nourishes the tumor, by permitting cancer cells to escape from the tumor and metastasize, and by shielding the tumor from immune attack. Thus cancer is a complex process, both within the tumor and between the tumor and the normal tissues that surround it.
Cancer in Families
Although essentially all individuals are at risk for some cancer at some point during their lives, many forms of cancer have a higher incidence in relatives of patients than in the general population. In some cases, this increased incidence is due primarily to inheritance of a single mutant gene with high penetrance. These mutations result in hereditary cancer syndromes (see, for examples,Cases 7, 15, 29, 39, and48)following mendelian patterns of inheritance that were presented in Chapter 7. Among these syndromes, we currently know of approximately 100 different genes in which deleterious mutations increase the risk for cancer many-fold higher than in the general population. There are also many dozens of additional genetic disorders that are not usually considered to be hereditary cancer syndromes and yet include some increased predisposition to cancer (Case 6) (for example, the ten- to twenty-fold increased lifetime risk for leukemia in Down syndrome [see Chapter 6]). These clear examples notwithstanding, it is important to emphasize that not all families with an apparently increased incidence of cancer can be explained by known mendelian or clearly recognized genetic disorders. These families likely represent the effects of both shared environment and one or more genetic variants that increase susceptibility and are therefore classified as multifactorial, with complex inheritance (see Chapter 8), as will be explored later in this chapter.
Although individuals with a hereditary cancer syndrome represent probably less than 5% of all patients with cancer, identification of a genetic basis for their disease has great importance both for clinical management of these families and for understanding cancer in general. First, the relatives of individuals with strong hereditary predispositions, which are most often due to mutations in a single gene, can be offered testing and counseling to provide appropriate reassurance or more intensive monitoring and therapy, depending on the results of testing. Second, as is the case with many common diseases, understanding the hereditary forms of the disease provides crucial insights into disease mechanisms that go far beyond the rare hereditary forms themselves. These general concepts are illustrated in the examples discussed in the sections that follow.
Activated Oncogenes in Hereditary Cancer Syndromes
Multiple Endocrine Adenomatosis, Type 2
The type A variant of multiple endocrine adenomatosis, type 2 (MEN2) is an autosomal dominant disorder characterized by a high incidence of medullary carcinoma of the thyroid that is often but not always associated with pheochromocytoma, benign parathyroid adenomas, or both. Patients with the rarer type B variant, termed MEN2B, have, in addition to the tumors seen in patients with MEN2A, thickening of nerves and the development of benign neural tumors, known as neuromas, on the mucosal surface of the mouth and lips and along the gastrointestinal tract.
The mutations responsible for MEN2 are in the RET gene. Individuals who inherit an activating mutation in RET have a greater than 60% chance of developing a particular type of thyroid carcinoma (medullary), although more sensitive tests, such as blood tests for thyrocalcitonin or urinary catecholamines synthesized by pheochromocytomas, are abnormal in well above 90% of heterozygotes for MEN2.
RET encodes a cell-surface protein that contains an extracellular domain that can bind signaling molecules and a cytoplasmic tyrosine kinase domain. Tyrosine kinases are a class of enzymes that phosphorylate tyrosines in proteins. Tyrosine phosphorylation initiates a signaling cascade of changes in protein-protein and DNA-protein interactions and in the enzymatic activity of many proteins (Fig. 15-5). Normally, tyrosine kinase receptors must bind specific signaling molecules in order to undergo the conformational change that makes them enzymatically active and able to phosphorylate other cellular proteins. The mutations in RET that cause MEN2A increase its kinase activity even in the absence of its ligand (a state referred to as constitutive activation).
FIGURE 15-5 Schematic diagram of the function of the Ret receptor, the product of the RET proto-oncogene. Upon binding of a ligand (L), such as glial-derived growth factor or neurturin, to the extracellular domain, the protein dimerizes and activates its intracellular kinase domain to autophosphorylate specific tyrosine residues. These then bind the SHC adaptor protein, which sets off multiple cascades of complex protein interactions involving other serine-threonine and phosphatidylinositol kinases and small G proteins, which in turn activate other proteins, ultimately activating certain transcription factors that suppress apoptosis and stimulate cellular proliferation. Mutations in RET that result in type A variant of multiple endocrine adenomatosis, type 2 (MEN2A) cause inappropriate dimerization and activation of its own intrinsic kinase without ligand binding.
The RET gene is expressed in many tissues of the body and is required for normal embryonic development of autonomic ganglia and kidney. It is unclear why germline activating mutations in this proto-oncogene result in a particular cancer of distinct histological types restricted to specific tissues, whereas other tissues in which the oncogene is expressed do not develop tumors. Interestingly, RET is the same gene implicated in Hirschsprung disease (Case 22) (see Chapter 8), although those mutations are usually loss-of-function, not activating, mutations. There are, however, some families in which the same mutation in RETcan act as an activated oncogene in some tissues (such as thyroid) and cause MEN2A, while not having sufficient function in other tissues, such as the developing enteric neurons of the gastrointestinal tract, resulting in Hirschsprung disease. Thus even the identical mutation can have different effects on different tissues.
The Two-Hit Theory of Tumor Suppressor Gene Inactivation in Cancer
As introduced earlier, whereas the proteins encoded by proto-oncogenes promote cancer when activated or overexpressed, mutations in TSGs contribute to malignancy by a different mechanism, the loss of function of both alleles of the gene. The products of many TSGs have now been isolated and characterized, some of which are presented in Table 15-2.
Selected Tumor Suppressor Genes
The existence of TSG mutations leading to cancer was proposed some five decades ago to explain why certain tumors can occur in either hereditary or sporadic forms (Fig. 15-6; see discussion later in this section). It was suggested that the hereditary form of the childhood cancer retinoblastoma (see next section) might be initiated when a cell in a person heterozygous for a germline mutation in the retinoblastoma TSG, required to prevent the development of the cancer, undergoes a second, somatic event that inactivates the other retinoblastoma gene allele. As a consequence of this second somatic event, the cell loses function of both alleles, giving rise to a tumor. In the sporadic form of retinoblastoma, both alleles are also inactivated, but in this case, the inactivation results from two somatic events occurring in the same cell.
FIGURE 15-6 Comparison of mendelian and sporadic forms of cancers such as retinoblastoma and familial polyposis of the colon. See text for discussion.
This so-called two-hit model is now widely accepted as the explanation for many hereditary cancers in addition to retinoblastoma, including familial polyposis coli, familial breast cancer, neurofibromatosis type 1 (NF1), Lynch syndrome, and Li-Fraumeni syndrome.
Tumor Suppressor Genes in Autosomal Dominant Cancer Syndromes
Retinoblastoma is the prototype of diseases caused by mutation in a TSG and is a rare malignant tumor of the retina in infants, with an incidence of approximately 1 in 20,000 births (Fig. 15-7) (Case 39). Diagnosis of a retinoblastoma must usually be followed by removal of the affected eye, although smaller tumors, diagnosed at an early stage, can be treated by local therapy so that vision can be preserved.
FIGURE 15-7 Retinoblastoma in a young girl, showing as a white reflex in the affected left eye when light reflects directly off the tumor surface. SeeSources & Acknowledgments.
Approximately 40% of cases of retinoblastoma are of the heritable form, in which the child (as just discussed and as represented generally by the family shown in Figure 15-6) inherits one mutant allele at the retinoblastoma locus (RB1) through the germline from either a heterozygous parent or, more rarely, from a parent with germline mosaicism for an RB1 mutation (see Chapter 7). In these children, retinal cells, which like all the other cells of the body are already carrying one inherited defective RB1 allele, suffer a somatic mutation or other alteration in the remaining normal allele, leading to loss of both copies of the RB1gene and initiating development of a tumor in each of those cells (Fig. 15-8).
FIGURE 15-8 Chromosomal mechanisms that could lead to loss of heterozygosity for DNA markers at or near a tumor suppressor gene in an individual heterozygous for an inherited germline mutation. The figure depicts the events that constitute the “second hit” that leads to retinoblastoma with loss of heterozygosity (LOH). Local events such as mutation, gene conversion, or transcriptional silencing by promoter methylation, however, could cause loss of function of both RB1 genes without producing LOH. +, normal allele, rb, the mutant allele.
The disorder appears to be inherited as a dominant trait because the large number of primordial retinoblasts and their rapid rate of proliferation make it very likely that a somatic mutation will occur as a second hit in one or more of the more than 106 retinoblasts already carrying an inherited RB1 mutation. Because the chance of a second hit is so great, it occurs frequently in more than one cell, and thus heterozygotes for the disorder often have tumors arising at multiple sites, such as multifocal tumors in one eye, in both eyes (bilateral retinoblastoma), or in both eyes, as well as in the pineal gland (referred to as “trilateral” retinoblastoma). It is worth emphasizing, however, that the occurrence of a second hit is a matter of chance and does not occur 100% of the time; the penetrance of retinoblastoma therefore, although greater than 90%, is not complete.
The other 60% of cases of retinoblastoma are sporadic; in these cases, both RB1 alleles in a single retinal cell have been mutated or inactivated independently by chance, and the child does not carry an RB1mutation inherited through the germline. Because two hits in the same cell is a statistically rare event, there is usually only a single clonal tumor, and the retinoblastoma is found at one location (unifocal) in one eye only. Unilateral tumor is no guarantee that the child does not have the heritable form of retinoblastoma, however, because 15% of patients with the heritable type develop a tumor in only one eye. Another difference between hereditary and sporadic tumors is that the average age at onset of the sporadic form is in early childhood, later than in infants with the heritable form (see Fig. 15-6), reflecting the longer time needed on average for two mutations, rather than one, to occur.
In a small percentage of patients with retinoblastoma, the mutation responsible is a cytogenetically detectable deletion or translocation of the portion of chromosome 13 that contains the RB1 gene. Such chromosomal changes, if they also disrupt genes adjacent to RB1, may lead to dysmorphic features in addition to retinoblastoma.
Nature of the Second Hit.
Typically, for retinoblastoma as well as for the other hereditary cancer syndromes, the first hit is an inherited mutation, that is, a change in the DNA sequence. The second hit, however, can be caused by a variety of genetic, epigenetic, or genomic mechanisms (see Fig. 15-8); although it is most often a somatic mutation, loss of function without mutation, such as occurs with epigenetic silencing (see Chapter 3), has also been observed in some cancer cells. Although a number of mechanisms have been documented, the common theme is loss of function of RB1. The RB1 gene product, p110 Rb1, is a phosphoprotein that normally regulates entry of the cell into the S phase of the cell cycle (see Chapter 2). Thus loss of the RB1 gene and/or absence of the normal RB1 gene product (by whatever mechanism) deprives cells of an important checkpoint and allows uncontrolled proliferation (see Table 15-2).
Loss of Heterozygosity.
In addition to mutations and epigenetic silencing, a novel genomic mechanism was uncovered when geneticists made an unusual but highly significant discovery when they compared DNA polymorphisms at the RB1 locus in DNA from normal cells to those in the retinoblastoma tumor from the same patient. Individuals with retinoblastoma who were heterozygous at polymorphic loci flanking the RB1 locus in normal tissues (see Fig. 15-8) had tumors that contained alleles from only one of their two chromosome 13 homologues, revealing a loss of heterozygosity (LOH) in tumor DNA in and around the RB1 locus. Furthermore, in familial cases, the retained chromosome 13 markers were the ones inherited from the affected parent, that is, the chromosome with the abnormal RB1 allele. Thus, in these cases, LOH represents the second hit of the remaining allele. LOH may occur by interstitial deletion, but there are other mechanisms as well, such as mitotic recombination or monosomy 13 due to nondisjunction (see Fig. 15-8).
LOH is the most common mutational mechanism by which the function of the remaining normal RB1 allele is disrupted in heterozygotes, although each of the mechanisms shown in Figure 15-8 have been documented in different patients. LOH is a feature of tumors in a number of cancers, both heritable and sporadic, and is often considered evidence for the existence of a TSG in the region of LOH.
Familial Breast Cancer due to Mutations in BRCA1 and BRCA2
Breast cancer is common. Among all cases of this disease, a small proportion (≈3% to 5%) appears to be due to a highly penetrant dominantly inherited mendelian predisposition that increases the risk for female breast cancer fourfold to sevenfold over the 12% lifetime risk observed in the general female population. In these families, one often sees features characteristic of hereditary (as opposed to sporadic) cancer: multiple affected individuals in a family, earlier age at onset, frequent multifocal, bilateral disease or second independent primary breast tumor, and second primary cancers in other tissues such as ovary and prostate.
Although a number of genes in which mutations cause highly penetrant mendelian forms of breast cancer have been discovered from family studies, the two genes responsible for the majority of all hereditary breast cancers are BRCA1 and BRCA2 (Case 7). Together, these two TSGs account for approximately one half and one third, respectively, of autosomal dominant familial breast cancer. Numerous mutant alleles of both genes have now been catalogued. Mutations in BRCA1 and BRCA2 are also associated with a significant increase in the risk for ovarian and fallopian duct cancer in female heterozygotes. Moreover, mutations in BRCA2 and, to a lesser extent, BRCA1, also account for 10% to 20% of all male breast cancer and increase the risk for male breast cancer ten to sixtyfold over the 0.1% lifetime risk observed among males in the general population (Table 15-3).
Lifetime Cancer Risks in Carriers of BRCA1 or BRCA2 Mutations Compared to the General Population
Data from Petrucelli N, Daly MB, Feldman GL: BRCA1 and BRCA2 hereditary breast and ovarian cancer. Updated September 26, 2013. In Pagon RA, Adam MP, Bird TD, et al, editors: GeneReviews[Internet], Seattle, University of Washington, Seattle, 1993-2014, http://www.ncbi.nlm.nih.gov/books/NBK1247/.
The gene products of BRCA1 and BRCA2 are nuclear proteins contained within the same multiprotein complex. This complex has been implicated in the cellular response to double-stranded DNA breaks, such as occur normally during homologous recombination or abnormally as a result of damage to DNA. As might be expected for any TSG, tumor tissue from heterozygotes for BRCA1 and BRCA2 mutations frequently demonstrates LOH with loss of the normal allele.
Penetrance of BRCA1 and BRCA2 Mutations.
Presymptomatic detection of women at risk for development of breast cancer as a result of any of these susceptibility genes relies on detecting clearly pathogenic mutations by gene sequencing. For the purposes of patient management and counseling, it would be helpful to know the lifetime risk for development of breast cancer in individuals, whether male or female, carrying particular mutations in the BRCA1 and BRCA2genes, compared with the risk in the general male or female population (see Table 15-3). Initial studies showed a greater than 80% risk for breast cancer by the age of 70 years in women heterozygous for deleterious BRCA1 mutations, with a somewhat lower estimate for BRCA2 mutation carriers. These estimates relied on estimates of the risk for development of cancer in female relatives within families ascertained because breast cancer had already occurred many times in family members; that is, families in which the particular BRCA1 or BRCA2 mutation was highly penetrant.
When similar risk estimates were made from population-based studies, however, in which women carrying BRCA1 and BRCA2 mutations were not selected because they were members of families in which many cases of breast cancer had already developed, the risk estimates were lower and ranged from 40% to 50% by the age of 70 years. The discrepancy between the penetrance of mutant alleles in families with multiple occurrences of breast cancer and the penetrance seen in women identified by population screening and not by family history suggests that other genetic or environmental factors must play a role in the ultimate penetrance of BRCA1 and BRCA2 mutations in women heterozygous for these mutations.
In addition to mutations in BRCA1 and BRCA2, mutations in other genes can also cause autosomal dominantly inherited breast cancer syndromes, albeit less commonly. These syndromes, which include the Li-Fraumeni, hereditary diffuse gastric cancer, Peutz-Jeghers, and Cowden syndromes, demonstrate lifetime breast cancer risks that approach those seen in carriers of BRCA1 or BRCA2 mutations, as well as risks for other forms of cancer such as sarcomas, brain tumors, and carcinomas of the stomach, thyroid, and small intestine.
Clinicians faced with a family with multiple affected individuals with breast cancer often look for distinguishing signs in the patient and in the family history to help guide the choice of which genes to analyze (see Box). However, the rapid decline in the cost of gene or even genome-wide sequencing has allowed the development of gene panels in which a dozen or more candidate genes can be accurately and simultaneously tested for mutations, often at a cost that is equivalent or even less than what was charged previously to analyze just one or two genes.
Diagnostic Criteria for Hereditary Cancer Syndromes
Li-Fraumeni Syndrome (LFS): Chompret Criteria*
• Proband with tumor belonging to LFS tumor spectrum (e.g., soft tissue sarcoma, osteosarcoma, brain tumor, premenopausal breast cancer, adrenocortical carcinoma, leukemia, lung bronchoalveolar cancer) before age 46 years AND at least one first- or second-degree relative with LFS tumor (except breast cancer if proband has breast cancer) before age 56 years or with multiple tumors; OR
• Proband with multiple tumors (except multiple breast tumors), two of which belong to LFS tumor spectrum and first of which occurred before age 46 years; OR
• Patient with adrenocortical carcinoma or choroid plexus tumor, irrespective of family history
Hereditary Diffuse Gastric Cancer Syndrome
• Family history of diffuse gastric cancer with two or more cases of gastric cancer, with at least one diffuse gastric cancer diagnosed before age 50 years
• Family with multiple lobular breast cancer
• Peutz-Jeghers–type hamartomatous polyps in the small intestine as well as in the stomach, large bowel, and extraintestinal sites, including the renal pelvis, bronchus, gallbladder, nasal passages, urinary bladder, and ureters
• Pigmented macules on the face, around oral mucosa and the perianal region, most pronounced in childhood
• Early-onset breast cancer, particularly before age 40
• Macrocephaly, especially 63 cm or larger in males, 60 cm or larger in females
• Thyroid cancer, particularly follicular type, before age 50
• Goiter, Hashimoto thyroiditis
• Dysplastic gangliocytoma of the cerebellum (Lhermitte-Duclos disease)
• Intestinal hamartomas
• Esophageal glycogenic acanthosis
• Skin findings of tricholemmomas or penile freckling
• Papillomas of oral cavity
*From Tinat J, Bougeard G, Baert-Desurmont S, et al: 2009 Version of the Chompret criteria for Li Fraumeni syndrome, J Clin Oncol 27:e108, 2009.
Hereditary Colon Cancer
Colorectal cancer, a malignancy of the epithelial cells of the colon and rectum, is one of the most common forms of cancer. It affects approximately 1.3 million individuals worldwide per year (150,000 of whom are in the United States) and is responsible for approximately 10% to 15% of all cancer. Most cases are sporadic, but a small proportion of colon cancer cases are familial, among which are two autosomal dominant conditions: familial adenomatous polyposis (FAP) and Lynch syndrome (LS), along with their variants.
Familial Adenomatous Polyposis.
FAP (Case 15) and its subvariant, Gardner syndrome, together have an incidence of approximately 1 per 10,000. In FAP heterozygotes, benign adenomatous polyps numbering in the many hundreds develop in the colon during the first two decades of life. In almost all cases, one or more of the polyps becomes malignant. Surgical removal of the colon (colectomy) prevents the development of malignancy. Because this disorder is inherited as an autosomal dominant trait, relatives of affected persons must be examined periodically by colonoscopy. FAP is caused by loss-of-function mutations in a TSG known as the APC gene (so-named because the condition used to be called adenomatous polyposis coli). Gardner syndrome is also due to mutations in APC and is therefore allelic to FAP. Patients with Gardner syndrome have, in addition to the adenomatous polyps with malignant transformation seen in FAP, other extracolonic anomalies, including osteomas of the jaw and desmoids, which are tumors arising in the muscle of the abdominal wall. Although the relatives of an individual affected with Gardner syndrome who also carry the same APC mutation tend to also show the extracolonic manifestations of Gardner syndrome, the same mutation in unrelated individuals has been found to cause only FAP in one individual and Gardner syndrome in another. Thus whether or not an individual has FAP or Gardner syndrome is not simply due to which mutation is present in the APC gene but is likely affected by genetic variation elsewhere in the genome.
Approximately 2% to 4% of cases of colon cancer are attributable to LS (Case 29). LS is characterized by autosomal dominant inheritance of colon cancer in association with a small number of adenomatous polyps that begin during early adulthood. The number of polyps is generally quite small, in contrast to the hundreds to thousands of adenomatous polyps seen with FAP. Nonetheless, the polyps in LS have high potential to undergo malignant transformation. Heterozygotes for the most commonly mutated LS gene have an approximately 80% lifetime risk for development of cancer of the colon; female heterozygotes have a somewhat smaller risk (approximately 70%) but also have an approximately 40% risk for endometrial cancer. There are also additional risks of 10% to 20% for cancer of the biliary or urinary tract and the ovary. Sebaceous gland tumors of the skin (Muir-Torre syndrome) may be the first presenting sign in LS; thus the presence of such tumors in a patient should raise suspicion of a possible hereditary colon cancer syndrome.
LS results from loss-of-function mutations in one of four distinct but related DNA repair genes (MLH1, MSH2, MSH6, and PMS2) that encode mismatch repair proteins. Although all four of these genes have been implicated in LS in different families, MLH1 and MSH2 are together responsible for the vast majority of LS, whereas the others have been found in only a few patients and are often associated with a lesser degree of mismatch repair deficiency and lower penetrance. Like the BRCA1 and BRCA2 genes, the LS mismatch repair genes are TSGs involved in maintaining the integrity of the genome. Unlike BRCA1 and BRCA2, however, the LS genes are not involved in double-stranded DNA break repair. Instead, their role is to repair incorrect DNA base pairing (i.e., pairing other than A with T or C with G) that can arise during DNA replication.
At the cellular level, the most striking phenotype of cells lacking mismatch repair proteins is an enormous increase in both point mutations and mutations occurring during replication of simple DNA repeats, such as a segment containing a string of the same base, for example (A)n, or a microsatellite, such as (TG)n (see Chapter 4). Microsatellites are believed to be particularly vulnerable to mismatch because slippage of the strand being synthesized on the template strand can occur more readily when a short tandem repeat is being synthesized. Such instability, referred to as the microsatellite instability-positive (MSI+)phenotype, occurs at two orders of magnitude higher frequency in cells lacking both copies of a mismatch repair gene. The MSI+ phenotype is easily seen in DNA as three, four, or even more alleles of a microsatellite polymorphism in a single individual's tumor DNA (Fig. 15-9). It is estimated that cells lacking both copies of a mismatch repair gene may carry 100,000 mutations within simple repeats throughout the genome.
FIGURE 15-9 Gel electrophoresis of three different microsatellite polymorphic markers in normal (N) and tumor (T) samples from a patient with a mutation in MSH2 and microsatellite instability. Although marker #2 shows no difference between normal and tumor tissues, genotyping at markers #1 and #3 reveals extra alleles (blue arrows), some smaller, some larger, than the alleles present in normal tissue.
Because of the increase in mutation rate in these classes of sequence, loss of function of mismatch repair genes will lead to somatic mutations in other driver genes. Two such driver genes have been isolated and characterized. The first is APC, whose normal function and role in FAP were described previously. The second is the gene TGFBR2, in which mutations also cause an autosomal dominant hereditary colon cancer syndrome. TGFBR2 encodes transforming growth factor β receptor II, a serine-threonine kinase that inhibits intestinal cell division. TGFBR2, is particularly vulnerable to mutation when mismatch repair proteins are lost because it contains a stretch of 10 adenines encoding three lysines within its coding sequence; deletion of one or more of these As results in a frameshift and loss-of-function mutation. LS is an excellent example of how a gene, like MLH1, which has a global effect on mutation rate throughout the genome, can be a driver gene through its effect on other genes, such as TGFBR2, that are more specifically involved in driving the development of a cancer.
Mutations in Tumor Suppressor Genes Causing Autosomal Recessive Pediatric Cancer Syndromes
As expected from the important role that DNA replication and repair enzymes play in mutation surveillance and prevention, inherited defects that alter the function of repair enzymes can lead to a dramatic increase in the frequency of mutations of all types, including those that lead to cancer.
Mutations in the LS mismatch repair genes are frequent enough in the population for there to be rare individuals with two germline mutations in one of the LS genes. Although much rarer than autosomal dominant forms of LS just discussed, this condition, known as constitutional mismatch repair syndrome, results in a markedly elevated risk for many cancers during childhood, including colorectal and small bowel cancer, as well as some cancers not associated with LS, such as leukemia in infancy and various types of brain tumors in childhood.
Several other well-known autosomal recessive disorders, including xeroderma pigmentosum (Case 48), ataxia-telangiectasia, Fanconi anemia, and Bloom syndrome, are also due to loss of function of proteins required for normal DNA repair or replication. Patients with these rare conditions have a high frequency of chromosome and gene mutations and, as a result, a markedly increased risk for various types of cancer, particularly leukemia or, in the case of xeroderma pigmentosum, skin cancers in sun-exposed areas. Clinically, radiography must be used with extreme caution, if at all, in patients with ataxia-telangiectasia, Fanconi anemia, and Bloom syndrome, and exposure to sunlight must be avoided in patients with xeroderma pigmentosum.
Although these syndromes are rare autosomal recessive disorders, heterozygotes for these gene defects are much more common and appear to be at increased risk for malignant neoplasia. For example, Fanconi anemia, in which homozygotes have a number of congenital anomalies, bone marrow failure, leukemia, and squamous cell carcinoma of the head and neck, is a chromosome instability syndromeresulting from mutations of at least 18 different loci involved in DNA and chromosome repair. In the aggregate, Fanconi anemia has a population frequency of approximately 1 to 5 per million, which translates to a carrier frequency of approximately 1 to 2 per 500. One of these Fanconi anemia loci turns out to be the known hereditary cancer gene BRCA2. Others include BRIP1, PALB2, and RAD51C (discussed in the next section), which are known to increase susceptibility to breast cancer in heterozygotes. Similarly, female heterozygotes for certain ataxia-telangiectasia mutations have overall a twofold increased risk for breast cancer compared with controls and a fivefold higher risk for breast cancer before the age of 50 years. Thus heterozygotes for these chromosome instability syndromes constitute a sizeable pool of individuals at increased risk for cancer.
Testing for Germline Mutations Causing Hereditary Cancer
As introduced earlier, although some sporadic cancers will be truly sporadic and due entirely to somatic mutation(s), others likely reflect a predisposition to a specific cancer due to familial variants in one or more genes. This raises the possibility of using genetic testing or even whole-genome sequencing to screen for germline mutations that might inform risk estimates for members of the general population or for families with insufficient family history to implicate a hereditary cancer syndrome. Here we illustrate the issues involved in the case of two common neoplasias, breast cancer and colorectal cancer.
BRCA1 and BRCA2 Testing
Identification of a germline mutation in BRCA1 or BRCA2 in a patient with breast cancer is of obvious importance for genetic counseling and cancer risk management for the patient's children, siblings, and other relatives, who may or may not be at increased risk. Such testing is, of course, also important for the patient's own management. For instance, in addition to removal of the cancer, a woman found to carry a BRCA1mutation might also choose to have a prophylactic mastectomy on the unaffected breast or a bilateral oophorectomy simultaneously to minimize the number of separate surgeries and anesthesia exposures. Finding a mutation in the proband or a first-degree relative would also allow mutation-specific testing in the rest of the family.
Importantly, however, the fraction of all female breast cancer patients whose disease is caused by a germline mutation in either the BRCA1 or BRCA2 gene is small, with estimates that vary between 1% and 3% in populations unselected for family history of breast or ovarian cancer, or for age at onset of the disease. Male breast cancer is 100 times less common than female breast cancer, but when it occurs, the frequency of germline mutations in hereditary breast cancer genes, particularly BRCA2, is 16%.
Until quite recently, the cost of mutation analysis in BRCA1 and BRCA2 was used to justify limiting gene sequencing to those patients most likely to be carrying a mutation, such as all male breast cancer patients and all women younger than 50 years with breast cancer, women with bilateral breast cancer, or women with first- and second-degree relatives with ovarian cancer or breast cancer. However, as the cost of sequencing falls, and large gene panels of breast cancer susceptibility genes, including BRCA1 and BRCA2, can now be analyzed for less than it cost previously to sequence just BRCA1 and BRCA2 alone, the guidelines of just a few years ago will inevitably undergo reevaluation.
Colorectal Cancer Germline Mutation Testing
Only 4% of patients with colon cancer, not selected for a family history of cancer, carry a mutation in one of the four mismatch repair genes MLH1, MSH2, MSH6, and PMS2 causing LS; an even smaller fraction contain APC mutations causing FAP. As with breast cancer, geneticists need to balance the cost and yield of sequencing hereditary colorectal cancer genes in every patient with colon cancer against the obvious importance of finding such a mutation for the patient and his or her family.
For LS, clinical factors such as the presence of multiple polyps, an early age at onset (before the age of 50 years), the location of the tumor in more proximal portions of the colon, the presence of a second tumor or history of colorectal cancer, a family history of colorectal or other cancers (particularly endometrial cancer), and cancer in relatives younger than 50 years of age, all boost the probability that a patient with colon cancer is carrying a mutation in a mismatch repair gene. Molecular studies of the tumor tissue, to look for evidence of the MSI+ phenotype (as discussed earlier in this chapter) or evidence of absent MSH2 and/or MSH6 protein by antibody staining in the tumor, also increase the probability that an individual patient with colorectal cancer carries a germline mismatch repair mutation. Unfortunately, loss of MLH1 protein staining in tumors due to promoter methylation is a frequent epigenetic finding in sporadic colon cancers and is therefore much less predictive of a germline LS mutation.
Combining clinical and molecular criteria allows the identification of a subset of all colorectal cancer patients in whom the probability of finding a mismatch repair mutation is much greater than 4%. These patients are clearly the most cost-effective group in which sequencing could be recommended. However, as with all such attempts at cost-effectiveness, limiting the number of patients studied to increase the yield of patients with positive sequencing inevitably results in missing a sizeable minority (20%) of patients with germline mismatch repair mutations. Again, the cost of mutation analysis must be reevaluated as the technology gets less expensive. More detailed discussions of gene testing will be presented in Chapter 18.
For FAP, the presence of hundreds of adenomatous polyps developing at an early age, multiple sebaceous adenomas, or the extracolonic signs of Gardner syndrome are sufficient to trigger germline testing for an APC mutation. There are, however, certain APC mutations that result in many fewer polyps and no extracolonic features (referred to as “attenuated FAP”). Attenuated FAP can be confused clinically with LS, but the tumors generally lack mismatch repair defects or microsatellite instability.
Familial Occurrence of Cancer
Cancer can also show increased incidence in families without fitting a clear-cut mendelian pattern. For example, it is estimated that as many as 20% of all breast cancer cases occurring in families that lack a clear, highly penetrant mendelian disorder nonetheless have a significant genetic contribution, as revealed by twin and family studies (see Chapter 8). The observed increase in cancer risk when relatives are affected may be due to mutations in a single gene but with penetrance that is sufficiently reduced to obscure any mendelian inheritance pattern. For example, in breast cancer, mutations in a gene such as PALB2 can increase lifetime risk for breast cancer to approximately 25% by age 55 and approximately 40% by age 85. A lack of obvious breast cancer risk in men with PALB2 mutations further obscures the inheritance pattern, although there is a significant increased risk for pancreatic cancer in men with these reduced penetrance alleles. Mutations in BRIP1 and RAD51C have similar effects.
The bulk of familial cancer is, however, likely to be a complex disorder caused by both genetic and shared environmental factors (see Chapter 8). The degree of complex familial cancer risk can be assessed by epidemiological studies that compare how often the disease occurs in relatives versus the general population. The age-specific incidence of many forms of cancer in family members of probands is increased over the incidence of the same cancer in an age-matched cohort in the general population (Fig. 15-10). This increased risk has been observed in individuals whose first-degree relatives (parent or sibling) are affected by a wide variety of different cancers, with an even greater increase in incidence when an individual's parent and sibling are both affected. For example, population-based epidemiological studies have shown that approximately 5% of all individuals in North America and Western Europe will develop colorectal cancer in their lifetime, but the lifetime risk for colorectal cancer is increased twofold to threefold over the average population risk if one first-degree relative is affected.
FIGURE 15-10 Standardized incidence ratios (SIRs) for cancers at various sites in first-degree relatives (child or sibling) of an affected person. A SIR is similar to the relative risk ratio (λr) that is based on prevalence of disease (as described in Chapter 8), except SIR is the ratio of the incidence of cases of cancer in relatives divided by the number expected from the incidence in an age-matched group in the general population. Error bars reflect 95% confidence limits on the SIRs. SeeSources & Acknowledgments.
In agreement with the likely complex inheritance of cancers, genome-wide association studies (see Chapter 10) have identified more than 150 mostly common variants associated with a variety of cancers. Prostate cancer, in particular, shows multiple associations with single nucleotide polymorphisms located in the intergenic or intronic regions of over a dozen loci. Unfortunately, odds ratios for most of these associations are all less than 2.0, and most are less than 1.3, therefore accounting for at most 20% of the observed familial risk for prostate cancer. Overall, then, although the role of inherited variants in the genome is clear, we cannot yet explain in detail the increased familial tendencies of most cancers. Whether common variants do not capture all of the risk or there are unrecognized environmental exposures in common between family members remain nonexclusive possibilities.
Previously, we introduced the concept of activation of oncogenes by a variety of mutational mechanisms (see Fig. 15-3). Here, we explore these mechanisms and their effects in greater detail, particularly in the context of sporadic cancers.
Activation of Oncogenes by Point Mutation
Many mutated oncogenes were first identified by molecular studies of cell lines derived from sporadic cancers. One of the first activated oncogenes discovered was a mutant RAS gene derived from a bladder carcinoma cell line. RAS encodes one of a large family of small guanosine triphosphate (GTP)–binding proteins (so-called G proteins) that serve as molecular “on-off” switches to activate or inhibit downstream molecules. Remarkably, the activated oncogene and its normal counterpart proto-oncogene differed at only a single nucleotide. The alteration led to synthesis of an abnormal Ras protein that was able to signal continuously, thus stimulating cell division and changing it into a tumor. RAS point mutations are now known in many tumors, and the RAS genes have been shown experimentally to be the mutational target of known carcinogens, a finding that supports a role for mutated RAS genes in the development of many cancers.
To date, nearly 50 human proto-oncogenes have been identified as driver mutations in sporadic cancer. Only a few of these proto-oncogenes have also been found to be inherited in a hereditary cancer syndrome.
Activation of Oncogenes by Chromosome Translocation
As pointed out previously (see Fig. 15-3), oncogene activation is not always the result of a DNA mutation. In some instances, a proto-oncogene is activated by a subchromosomal mutation, typically a translocation. More than 40 oncogenic chromosome translocations have been described to date, primarily in sporadic leukemias and lymphomas but also in a few rare connective tissue sarcomas. Although originally detected only by cytogenetic analysis, such chromosome alterations can be detected now by whole-genome sequence analysis (see Fig. 5-7), even using cell-free DNA in plasma samples from cancer patients.
In some cases, translocation breakpoints lie within the introns of two genes, thereby fusing two genes into one abnormal gene that encodes a chimeric protein with novel oncogenic properties. The best-known example is the translocation between chromosomes 9 and 22, the so-called Philadelphia chromosome that is seen in chronic myelogenous leukemia (CML) (Fig. 15-11) (Case 10). The translocation moves the proto-oncogene ABL1, a tyrosine kinase, from its normal position on chromosome 9q to a gene of unknown function, BCR, on chromosome 22q. The translocation results in the synthesis of a novel, chimeric protein, BCR-ABL1, containing a portion of the normal Abl protein with increased tyrosine kinase activity. The enhanced tyrosine kinase activity of the novel protein encoded by the chimeric gene is the primary event causing the chronic leukemia. New, highly effective drug therapies for CML, such as imatinib, have been developed, based on inhibition of this tyrosine kinase activity.
FIGURE 15-11 The Philadelphia chromosome translocation, t(9;22)(q34;q11). The Philadelphia chromosome (Ph1) is the derivative chromosome 22, which has exchanged part of its long arm for a segment of material from chromosome 9q that contains the ABL1 oncogene. Formation of the chimeric BCR-ABL1 gene on the Ph1 chromosome is the critical genetic event in the development of chronic myelogenous leukemia.
In other cases, a translocation activates an oncogene by placing it downstream of a strong, constitutive promoter belonging to a different gene. Burkitt lymphoma is a B-cell tumor in which the MYC proto-oncogene is translocated from its normal chromosomal position at 8q24 to a position distal to the immunoglobulin heavy chain locus at 14q32 or the immunoglobulin light chain genes on chromosomes 22 and 2. The function of the Myc protein is still not entirely known, but it appears to be a transcription factor with powerful effects on the expression of a number of genes involved in cellular proliferation, as well as on telomerase expression (see later discussion). The translocation brings enhancer or other transcriptional activating sequences, normally associated with the immunoglobulin genes, near to the MYC gene (Table 15-4). These translocations allow unregulated MYC expression, resulting in uncontrolled cell division.
Characteristic Chromosome Translocations in Selected Human Malignant Neoplasms
Based on Croce CM: Role of chromosome translocations in human neoplasia, Cell 49:155-156, 1987; Park M, van de Woude GF: Oncogenes: genes associated with neoplastic disease. In Scriver CR, Beaudet AL, Sly WS, Valle D, editors: The molecular and metabolic bases of inherited disease, ed 6, New York, 1989, McGraw-Hill, pp 251-276; Nourse J, Mellentin JD, Galili N, et al: Chromosomal translocation t(1;19) results in synthesis of a homeobox fusion mRNA that codes for a potential chimeric transcription factor, Cell 60:535-545, 1990; and Borrow J, Goddard AD, Sheer D, Solomon E: Molecular analysis of acute promyelocytic leukemia breakpoint cluster region on chromosome 17, Science 249:1577-1580, 1990.
Telomerase as an Oncogene
Another type of oncogene is the gene-encoding telomerase, a reverse transcriptase that is required to synthesize the hexamer repeat, TTAGGG, a component of telomeres at the ends of chromosomes. Telomerase is needed because, during normal semiconservative replication of DNA (see Chapter 2), DNA polymerase can only add nucleotides to the 3′ end of DNA and cannot complete the synthesis of a growing strand all the way out to the very end of that strand on the chromosome arm; thus, in the absence of a specific mechanism to allow replication of telomeres, the end of each chromosome arm would shorten each and every cell division.
In human germline cells and embryonic cells, telomeres contain approximately 15 kb of the telomeric repeat. As cells differentiate, telomerase activity declines in all somatic tissues; as telomerase function is lost, telomeres shorten, with a loss of approximately 35 bp of telomeric repeat DNA with each cell division. After hundreds of cell divisions, the chromosome ends become damaged, leading cells to stop dividing and to enter G0 of the cell cycle; the cells will ultimately undergo apoptosis and die.
In contrast, in highly proliferative cells of tissues such as bone marrow, telomerase expression persists, allowing self-renewal. Similarly, telomerase persistence is observed in many tumors, which permits tumor cells to proliferate indefinitely. In some cases, telomerase activity results from chromosome or genome mutations that directly up-regulate the telomerase gene; in others, telomerase may be only one of many genes whose expression is altered by a transforming oncogene, such as MYC.
Loss of Tumor Suppressor Gene in Sporadic Cancer
TP53 and RB1 in Sporadic Cancers
Although Li-Fraumeni syndrome, caused by a dominantly inherited germline mutation in the TP53 gene, is a rare familial syndrome, somatic mutation causing a loss of function of both alleles of TP53 is one of the most common genetic alterations seen in sporadic cancer (see Table 15-2). Mutations of TP53, deletion of the segment of chromosome 17p that includes TP53, or loss of the entire chromosome 17 is frequently and repeatedly seen in a wide range of sporadic cancers. These include breast, ovarian, bladder, cervical, esophageal, colorectal, skin, and lung carcinomas; glioblastoma of the brain; osteogenic sarcoma; and hepatocellular carcinoma.
The retinoblastoma gene RB1 is also frequently mutated in many sporadic cancers, including breast cancer. For example, 13q14 LOH in human breast cancers is associated with loss of RB1 mRNA in the tumor tissues. In still other cancers, the RB1 gene is intact and its mRNA appears to be at or near normal levels, yet the RB1 protein is deficient. This anomaly has now been explained by the recognition that RB1 can be down-regulated in association with overexpression of the oncomir miR-106a, which targets RB1 mRNA and blocks its translation.
Cytogenetic Changes in Cancer
Aneuploidy and Aneusomy
As introduced in Chapter 5, cytogenetic changes are hallmarks of cancer, whether sporadic or familial, particularly in later and more malignant or invasive stages of tumor development. Cytogenetic alterations suggest that a critical element of cancer progression includes defects in genes involved in maintaining chromosome stability and integrity and ensuring accurate mitotic segregation.
Initially, most of the cytogenetic studies of tumor progression were carried out in leukemias because the tumor cells were amenable to being cultured and karyotyped by standard methods. For example, when CML, with the 9;22 Philadelphia chromosome, evolves from the typically indolent chronic phase to a severe, life-threatening blast crisis, there may be several additional cytogenetic abnormalities, including numerical or structural changes, such as a second copy of the 9;22 translocation chromosome or an isochromosome for 17q. In advanced stages of other forms of leukemia, other translocations are common. In contrast, a vast array of chromosomal abnormalities are seen in most solid tumors. Cytogenetic abnormalities found repeatedly in a specific type of cancer are likely to be driver chromosome mutations involved in the initiation or progression of the malignant neoplasm. A current focus of cancer research is to develop a comprehensive cytogenetic and genomic definition of these abnormalities, many of which result in enhanced proto-oncogene expression or the loss of TSG alleles. Whole-genome sequencing is replacing cytogenetic analysis in many instances, because it provides a level of sensitivity and precision well beyond detection of cytologically visible genome changes.
In addition to translocations and other rearrangements, another cytogenetic aberration seen in many cancers is gene amplification, a phenomenon in which many additional copies of a segment of the genome are present in the cell (see Fig. 15-3). Gene amplification is common in many cancers, including neuroblastoma, squamous cell carcinoma of the head and neck, colorectal cancer, and malignant glioblastomas of the brain. Amplified segments of DNA are readily detected by comparative genome hybridization or whole-genome sequencing and appear as two types of cytogenetic change in routine chromosome analysis: double minutes (very small accessory chromosomes) and homogeneously staining regions that do not band normally and contain multiple, amplified copies of a particular DNA segment. How and why double minutes and homogeneously staining regions develop are poorly understood, but amplified regions are known to include extra copies of proto-oncogenes such as the genes encoding Myc, Ras, and epithelial growth factor receptor, which stimulate cell growth, block apoptosis, or both. For example, amplification of the MYCN proto-oncogene encoding N-Myc is an important clinical indicator of prognosis in the childhood cancer neuroblastoma. MYCN is amplified more than 200-fold in 40% of advanced stages of neuroblastoma; despite aggressive treatment, only 30% of patients with advanced disease survive 3 years. In contrast, MYCNamplification is found in only 4% of early-stage neuroblastoma, and the 3-year survival is 90%. Amplification of genes encoding the targets of chemotherapeutic agents has also been implicated as a mechanism for the development of drug resistance in patients previously treated with chemotherapy.
Applying Genomics to Individualize Cancer Therapy
Genomics is already having a major impact on diagnostic precision and optimization of therapy in cancer. In this section, we describe how one such approach, gene expression profiling, is used to guide diagnosis and treatment.
Gene Expression Profiling and Clustering to Create Signatures
Comparative hybridization techniques can be used to measure simultaneously the level of mRNA expression of some or all of the estimated 20,000 protein-coding genes in any human tissue sample. A measurement of mRNA expression in a sample of tissue constitutes a gene expression profile specific to that tissue. Figure 15-12 depicts a hypothetical, idealized situation of eight samples, four from each of two types of tumor, A and B, profiled for 100 different genes. The expression profile derived from expression arrays for this simple example is already substantial, consisting of 800 expression values. In a real expression profiling experiment, however, hundreds of samples may be analyzed for the expression of all human genes, producing a massive data set of millions of expression values. Organizing the data and analyzing them to extract key information are challenging problems that have inspired the development of sophisticated statistical and bioinformatic tools. Using such tools, one can organize the data to find groups of genes whose expression seems to correlate, that is, increase or decrease together, between and among the samples. Grouping genes by their patterns of expression across samples is termed clustering.
FIGURE 15-12 Schematic of an idealized gene expression profiling experiment of eight samples and 100 genes. Left, Individual arrays of gene sequences spotted on glass or silicon chips are used for comparative hybridization of eight different samples relative to a common standard. Red indicates decreased expression compared with control, green indicates increased expression, and yellow is unchanged expression. (In this schematic, red, yellow, and green represent decreased, equal or increased expression, whereas a real experiment would provide a continuous quantitative reading with shades of red and green.) Center, All 800 expression measurements are organized so that the relative expression for each gene, 1 through 100, is put in order vertically in a column under the number of each sample. Right, Clustering into signatures involves only those 13 genes that showed correlation across subsets of samples. Some genes have reciprocal (high versus low) expression in the two tumors; others show a correlated increase or decrease in one tumor and not the other.
Clusters of gene expression can then be tested to determine if any correlate with particular characteristics of the samples of interest. For example, profiling might indicate that a cluster of genes with a correlated expression profile is found more frequently in samples from tumor A than from tumor B, whereas another cluster of genes with a correlated expression profile is more frequent in samples derived from tumor B than from tumor A. Clusters of genes whose expression correlates with each other and with a particular set of samples constitute a so-called expression signature characteristic of those samples. In the hypothetical profiles in Figure 15-12, certain genes have a correlated expression that serves as a signature for tumor A; tumor B has a signature derived from the correlated expression of a different subset of these 100 genes.
Application of Gene Signatures
The application of gene expression profiles to characterize tumors is useful in a number of ways.
• First, it increases our ability to discriminate between different tumors in ways that complement the standard criteria applied by pathologists to characterize tumors, such as histological appearance, cytogenetic markers, and expression of specific marker proteins. Once distinguishing signatures for different tumor types (e.g., tumor A versus tumor B) are defined using known samples, the expression pattern of unknowntumor samples can then be compared with the expression signatures for tumor A and tumor B and classified as A-like, B-like, or neither, depending on how well their expression profiles match the signatures of A and B. Pathologists have used expression profiling to make difficult distinctions between tumors that require very different management approaches. These include distinguishing large B-cell lymphoma from Burkitt lymphoma, differentiating primary lung cancers from squamous cell carcinomas of the head and neck metastatic to lung, and identifying the tissue of origin of a cryptic primary tumor whose metastasis gives too little information to allow its classification.
• Second, different signatures may be found to correlate with known clinical outcomes, such as prognosis, response to therapy, or any other outcome of interest. If validated, such signatures can be applied prospectively to help guide therapy in newly diagnosed patients.
• Finally, for basic research, clustering may reveal previously unsuspected connections of functional importance among genes involved in a disease process.
Gene Expression Profiling in Cancer Prognosis
Choosing the appropriate therapy for most cancers is difficult for patients and their physicians alike, because recurrence is common and difficult to predict. Better characterization of each patient's cancer as to recurrence risk and metastatic potential would clearly be beneficial for deciding between more or less aggressive courses of surgery and/or chemotherapy. For example, in breast cancer, although presence of the estrogen and progesterone receptors, amplification of the human epidermal growth factor receptor 2 (HER2) oncogene, and absence of metastatic tumor in lymph nodes found on dissection of axillary lymphatics are strong predictors of better response to therapy and prognosis, they are still imprecise. Expression profiling (Fig. 15-13) is opening up a promising new avenue for clinical decision making in the management of breast cancer, as well as in other cancers, including lymphoma, prostate cancer, and metastatic adenocarcinomas of diverse tissue origins (lung, breast, colorectal, uterine, and ovarian).
FIGURE 15-13 Expression patterns for a series of genes (along the vertical axis at left) for series of patient tumors, with the tumors arranged along the horizontal axis at top so that tumors with more similar expression patterns are grouped more closely together. The tumors appear to generally cluster into two groups, which are then correlated with long-term survival. SeeSources & Acknowledgments.
Gene expression profiling of various sets of genes is clinically available for use in the management of breast, colon, and ovarian cancer; which genes and how many are included in the profile depends on the tumor type and vendor. Although the clinical utility and cost-effectiveness continue to be debated (see Chapter 18), there is a general consensus that combinations of clinical and gene expression data in patients newly diagnosed with cancer will provide better prospective estimates of prognosis and improved guidance of therapy. It is hoped that by improving the accuracy of prognosis with tumor expression profiling, oncologists can choose to forgo more vigorous and expensive chemotherapies in patients who do not need and/or will not benefit from them.
The fact that the prognosis of practically every single patient could be associated with a particular combination of clinical features, genome sequence, and expression signatures underscores a crucial point about cancer: each person's cancer is a unique disorder. The genomic and gene expression heterogeneity among patients who all carry the same cancer diagnosis should not be surprising. Every patient is unique in the genetic variants he or she carries, including those variants that will affect how the cancer develops and the body responds to it. Moreover, the clonal evolution of a cancer implies that chance mutational and epigenetic events will likely occur in different and unique combinations in every patient's particular cancer.
Targeted Cancer Therapy
Until recently, most nonsurgical cancer treatment relied on cytotoxic agents, such as chemotherapeutic agents or radiation, designed to preferentially kill tumor cells while attempting to spare normal tissues. Despite tremendous successes in curing such diseases as childhood acute lymphocytic leukemia and Hodgkin lymphoma, most cancer patients in whom complete removal of the tumor with surgery is no longer possible receive remission, not cure, of their disease, usually at the cost of substantial toxicity from cytotoxic agents. The discovery of specific driver genes and their mutations in cancers has opened a new avenue for precisely targeted, less toxic treatments. Activated oncogenes are tempting targets for cancer therapy through direct blockade of their aberrant function. This can include blocking an activated cell surface receptor by monoclonal antibodies, or targeted inhibition of intracellular constitutive kinase activity with drugs designed to specifically inhibit their enzymatic activities.
The proof of principle for this approach was established with the development of imatinib, a highly effective inhibitor of a number of tyrosine kinases, including the ABL1 kinase in CML. Prolonged remissions of this disease have been seen, in some cases with apparently indefinite postponement of the transformation into a virulent acute leukemia (blast crisis) that so often meant the end of a CML patient's life. Additional kinase inhibitors have been developed to target other activated oncogene driver genes in a variety of tumor types (Table 15-5).
Cancer Treatments Targeted to Specific Activated Driver Oncogenes
ALK, Anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; FDA, U.S. Food and Drug Administration; HER2, human epidermal growth factor receptor 2; MEK, mitogen-activated extracellular signal-regulated kinase; PDGF, platelet-derived growth factor.
The initial results with targeted therapies, although very promising in some cases, have not led to permanent cures in most patients because tumors develop resistance to the targeted therapy. The outgrowth of resistant tumors is not surprising. First, as previously discussed, cancer cells are highly mutable, and their genomes undergo recurrent mutation. Even if only a small minority of cells acquire resistance through either mutation of the targeted oncogene itself, or through a compensatory mutation elsewhere, the tumor can progress even in the face of oncogene inhibition. Newer compounds that can overcome drug resistance are being developed and used in clinical trials. Ultimately, combination therapy that targets different driver genes may be required, based on the idea that a tumor cell is less likely to develop resistance in multiple unrelated pathways targeted by a combination of agents.
Cancer and the Environment
Although the theme of this chapter emphasizes the genetic basis of cancer, there is no contradiction in considering the role of environment in carcinogenesis. By environment, we include exposure to a wide variety of different types of agents—food, natural and artificial radiation, chemicals, even which viruses and bacteria are colonizing the gut. The risk for cancer shows significant variation among different populations and even within the same population in different environments. For example, gastric cancer is almost three times as common among Japanese in Japan as among Japanese living in Hawaii or Los Angeles.
In some cases, environmental agents act as mutagens that cause somatic mutations; the somatic mutations, in turn, are responsible for carcinogenesis. According to some estimates based chiefly on data from the aftermath of the atomic bombings of Hiroshima and Nagasaki, as much as 75% of the risk for cancer may be environmental in origin. In other cases, there appears to be a correlation between certain exposures and risk for cancer, such as the benefits of dietary fiber or low-dose aspirin therapy in lowering colon cancer risks. The nature of environmental agents that increase or reduce the risk for cancer, the assessment of the additional risk associated with exposure, and ways of protecting the population from such hazards are matters of strong public concern.
Ionizing radiation is known to increase the risk for cancer. Everyone is exposed to some degree of ionizing radiation through background radiation (which varies greatly from place to place) and medical exposure. The risk is dependent on the age at exposure, being greatest for children younger than 10 years and for older adults.
Although there are still large areas of uncertainty about the magnitude of the effects of radiation (especially low-level radiation) on cancer risk, some information can be gleaned from events involving large-scale release of radiation into the environment. The data for survivors of the Hiroshima and Nagasaki atomic bombings, for example, show a long latency period, in the 5-year range for leukemia but up to 40 years for some tumors. In contrast, there has been little increase in cancer detectable among populations exposed to ionizing radiation by the more recent nuclear accident at Chernobyl, with the exception of a significant fivefold to sixfold increase in thyroid cancer among the most heavily exposed children living in Belarus. The increase in thyroid cancer is almost certainly caused by the radioactive iodine 131I that was present in the nuclear material released from the damaged reactor and was taken up and concentrated within the thyroid gland.
Interest in the carcinogenic effect of chemicals dates at least to the 18th century, when the high incidence of scrotal cancer in young chimney sweeps was noticed. Today, there is concern about many possible chemical carcinogens, especially tobacco, components of the diet, industrial carcinogens, and toxic wastes. Documentation of the risk of exposure is often difficult, but the level of concern is such that all clinicians should have a working knowledge of the subject and be able to distinguish between well-established facts and areas of uncertainty and debate.
The precise molecular mechanisms by which most chemical carcinogens cause cancer are still the subject of extensive research. One illustrative example of how a chemical carcinogen may contribute to the development of cancer is that of hepatocellular carcinoma, the fifth most common cancer worldwide. In many parts of the world, hepatocellular carcinoma occurs at increased frequency because of ingestion of aflatoxin B1, a potent carcinogen produced by a mold found on peanuts. Aflatoxin has been shown to mutate a particular base in the TP53 gene, causing a G to T mutation in codon 249, thus converting an arginine codon to serine in the critically important p53 protein. This mutation is found in nearly half of all hepatocellular carcinomas in patients from parts of the world in which there is a high frequency of contamination of foodstuffs by aflatoxin, but it is not found in similar cancers in patients whose exposure to aflatoxin in food is low. The Arg249Ser mutation in p53 enhances hepatocyte growth and interferes with the growth control and apoptosis associated with wild-type p53; LOH of TP53 in hepatocellular carcinoma is associated with a more malignant appearance of the cancer. Although aflatoxin B1 alone is capable of causing hepatocellular carcinoma, it also acts synergistically with chronic hepatitis B and C infections.
A more complicated situation occurs with an exposure to complex mixtures of chemicals, such as the many known or suspected carcinogens and mutagens found in cigarette smoke. The epidemiological evidence is overwhelming that cigarette smoke increases the risk for lung cancer and throat cancer, as well as other cancers. Cigarette smoke contains polycyclic hydrocarbons that are converted to highly reactive epoxides that cause mutations by directly damaging DNA. The relative importance of these substances and how they might interact in carcinogenesis are still being elucidated.
The case of cigarette smoking also raises another interesting issue. Why do only some cigarette smokers get lung cancer? The case of cancer and cigarette smoking provides an important example of the interaction between environmental and genetic factors to either enhance or prevent the carcinogenic effects of chemicals. The enzyme aryl hydrocarbon hydroxylase (AHH) is an inducible protein involved in the metabolism of polycyclic hydrocarbons, such as those found in cigarette smoke. AHH converts hydrocarbons into an epoxide form that is more easily excreted by the body but is also carcinogenic. AHH activity is encoded by members of the CYP1 family of cytochrome P450 genes (see Chapter 18). The CYP1A1 gene is inducible by cigarette smoke, but the inducibility is variable in the population because of different alleles at the CYP1A1 locus. People who carry a “high-inducibility” allele, particularly those who are smokers, appear to be at an increased risk for lung cancer, with odds ratios of 4 to 5 compared to individuals without the cancer-susceptibility CYP1A1 alleles. On the other hand, homozygotes for the recessive “low-inducibility” allele appear to be less likely to develop lung cancer, possibly because their AHH is less effective at converting the hydrocarbons to highly reactive carcinogens.
Similarly, individuals homozygous for alleles in the CYP2D6 gene that reduce the activity of another cytochrome P450 enzyme appear to be more resistant to the potential carcinogenic effects of cigarette smoke or occupational lung carcinogens (e.g., asbestos or polycyclic aromatic hydrocarbons). Normal or ultrafast metabolizers, on the other hand, who carry alleles that increase the activity of the Cyp2D6 enzyme, have a four fold greater risk for lung cancer than do slow metabolizers. This risk increases to 18-fold among persons exposed routinely to lung carcinogens. A similar association has been reported for bladder cancer.
Although the precise genetic and biochemical basis for the apparent differences in cancer susceptibility within the normal population remains to be determined, these associations could have significant public health consequences and may point eventually to a way of identifying persons who are genetically at a higher risk for the development of cancer.
Garraway LA, Lander ES. Lessons from the cancer genome. Cell. 2013;153:17–37.
International Agency for Research on Cancer (IARC), World Health Organization. www.cruk.org/cancerstats; 2014.
Schneider L. Counseling about cancer. ed 3. Wiley-Liss: New York; 2011.
Shen H, Laird PW. Interplay between the cancer genome and epigenome. Cell. 2013;153:38–55.
Vogelstein B, Papadopoulos N, Velculescu VE, et al. Cancer genome landscapes. Science. 2013;339:1546–1558.
Chen P-S, Su J-L, Hung M-C. Dysregulation of microRNAs in cancer. J Biomed Sci. 2012;19:90.
Chin L, Anderson JN, Futreal PA. Cancer genomics, from discovery science to personalized medicine. Nat Med. 2011;17:297–303.
Di Leva G, Garofalo M, Croce CM. MicroRNAs in cancer. Annu Rev Pathol Mech Dis. 2014;9:287–314.
Kiplivaara O, Aaltonen LA. Diagnostic cancer genome sequencing and the contribution of germline variants. Science. 2013;339:1559–1562.
Lal A, Panos R, Marjanovic M, et al. A gene expression profile test to resolve head & neck squamous versus lung squamous cancers. Diagn Pathol. 2013;8:44.
Reis-Filho J, Pusztai L. Gene expression profiling in breast cancer: classification, prognostication, and prediction. Lancet. 2011;378:1812–1823.
Watson IR, Takahashi K, Futreal PA, et al. Emerging patterns of somatic mutations in cancer. Nat Rev Genet. 2013;14:703–718.
Wogan GN, Hecht SS, Felton JS, et al. Environmental and chemical carcinogenesis. Semin Cancer Biol. 2004;14:473–486.
Wong MW, Nordfors C, Mossman D, et al. BRIP1, PALB2, and RAD51C mutation analysis reveals their relative importance as genetic susceptibility factors for breast cancer. Breast Cancer Res Treat. 2011;127:853–859.
The Cancer Genome Atlas: http://cancergenome.nih.gov/abouttcga/overview.
1. A patient with retinoblastoma has a single tumor in one eye; the other eye is free of tumors. What steps would you take to try to determine whether this is sporadic or heritable retinoblastoma? What genetic counseling would you provide? What information should the parents have before a subsequent pregnancy?
2. Discuss possible reasons why colorectal cancer is an adult cancer, whereas retinoblastoma affects children.
3. Many tumor types are characterized by the presence of an isochromosome for the long arm of chromosome 17. Provide a possible explanation for this finding.
4. Many children with Fanconi anemia have limb defects. If an affected child requires surgery for the abnormal limb, what special considerations arise?
5. Wanda, whose sister has premenopausal bilateral breast cancer, has a greater risk for developing breast cancer herself than Wilma, whose sister has premenopausal breast cancer in only one breast. Both Wanda and Wilma, however, have a greater risk than does Winnie, who has a completely negative family history. Discuss the role of molecular testing in these women. What would their breast cancer risks be if a pathogenic BRCA1 or BRCA2 mutation were found in the affected relative? What if no mutations were found?
6. Propose a theory for why so few hereditary cancer syndromes, inherited as autosomal dominant diseases, are caused by activated oncogenes, whereas so many are caused by germline mutations in a tumor suppressor gene (TSG).