INTRODUCTION — Colorectal cancer (CRC) is a common disease. Approximately 151,030 new cases are diagnosed each year in the United States, of which 106,180 originate in the colon and the rest originate in the rectum [1]. Annually, approximately 52,980 Americans die of CRC, accounting for approximately 9 percent of all cancer deaths; in the United States, CRC ranks third in both incidence and cause of cancer death in both men and women. Global, country-specific data on incidence and mortality are available from the World Health Organization (WHO) GLOBOCAN database.
The risk factors for CRC are both environmental and inherited. The mode of presentation of CRC follows one of three patterns that are reflective of these differing risk factors: sporadic, inherited, and familial:
●Sporadic disease, in which there is no family history, accounts for approximately 70 percent of all CRCs. It is most common over the age of 50, and dietary and environmental factors have been etiologically implicated. (See "Colorectal cancer: Epidemiology, risk factors, and protective factors".)
●Fewer than 10 percent of patients have a true inherited predisposition to CRC, and these cases are subdivided according to whether or not colonic polyps are a major disease manifestation. The diseases with polyposis include familial adenomatous polyposis (FAP), MUTYH-associated polyposis (MAP), and the hamartomatous polyposis syndromes (eg, Peutz-Jeghers, juvenile polyposis [2], phosphatase and tensin homolog [PTEN] hamartoma tumor [Cowden] syndrome), while those without polyposis are referred to as hereditary nonpolyposis CRC (HNPCC; Lynch syndrome). These conditions are all associated with a high risk of developing CRC. In many cases, the causative genetic mutation has been identified, and a test is available. (See "Clinical manifestations and diagnosis of familial adenomatous polyposis" and "MUTYH-associated polyposis" and "Peutz-Jeghers syndrome: Clinical manifestations, diagnosis, and management" and "Juvenile polyposis syndrome" and "PTEN hamartoma tumor syndromes, including Cowden syndrome" and "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis".)
●The third and least well understood pattern is known as "familial" CRC, which accounts for up to 25 percent of cases. Affected patients have a family history of CRC, but the pattern is not consistent with one of the inherited syndromes described above. Individuals from these families are at increased risk of developing CRC, although the risk is not as high as with the inherited syndromes. Having a single affected first-degree relative (ie, parent, child, sibling) increases the risk of developing CRC 1.7-fold over that of the general population. The risk is further increased if two first-degree relatives have CRC or if the index case is diagnosed before age 55. (See "Colorectal cancer: Epidemiology, risk factors, and protective factors".)
Some of these patients may have familial CRC type X, in which clinical criteria are met for Lynch syndrome but in the absence of an identified germline mutation in one of the mismatch repair genes, the genetic hallmark of Lynch syndrome [3]. The term familial CRC syndrome is probably a misnomer as it is likely that these patients have a currently unidentified inherited genetic mutation. However, data suggest that individuals with CRC arising in the context of familial CRC type X do not have outcomes that are as favorable as those seen in individuals with CRC in the setting of Lynch syndrome [4]. (See 'Mismatch repair genes' below.)
In general, the mechanisms underlying familial clustering of CRC in the absence of a discernible inherited syndrome remain incompletely understood.
Our level of understanding of the molecular events underlying CRC is far greater than for other common solid tumors. Specific germline mutations are responsible for the inherited CRC syndromes, while a stepwise accumulation of somatic mutations is thought to underlie most sporadic cases. In contrast, the genetic abnormalities underlying familial CRC remain incompletely understood.
This topic will review the major genetic aspects of colorectal carcinogenesis, with particular emphasis on sporadic CRC. Inherited conditions that significantly increase the risk of CRC (eg, FAP and Lynch syndrome) and the role of genetic testing and screening for patients with an inherited predisposition are discussed elsewhere. (See "Clinical manifestations and diagnosis of familial adenomatous polyposis" and "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis" and "Familial adenomatous polyposis: Screening and management of patients and families" and "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Cancer screening and management" and "Juvenile polyposis syndrome".)
MOLECULAR PATHOGENESIS OF COLORECTAL CANCER — Specific genetic changes are thought to drive the transformation from normal colonic epithelium to invasive cancer. Genetic mutations can be inherited or acquired. Any genetic mutation that occurs at or before fertilization of the ovum is termed a germline mutation and can be transmitted from parent to offspring as an inherited defect. If the mutation occurs spontaneously in the sperm, ovum, or zygote, the affected person's parents do not manifest the cancer phenotype, but future progeny may inherit the de novo mutation. More commonly, a spontaneous mutation appears in a cell during the growth and/or development of a particular tissue or organ; this is called a somatic mutation. Because these mutations often confer a selective growth advantage, they result in preferential proliferation of the cell containing the mutated genetic material (clonal evolution) [5].
The clonal nature of tumors is a critical feature of the somatic mutation/clonal evolution theory of human carcinogenesis. According to this model, the growth advantage acquired by a single mutated cell allows its progeny to outnumber those of neighboring cells. From within this clonal population, a single cell acquires a second mutation, providing an additional growth advantage that allows further clonal expansion. Subsequent waves of clonal expansion are driven by the sequential acquisition of more mutations, further cellular disorganization, and eventually the ability to invade and metastasize.
The adenoma-carcinoma sequence — Most human CRCs are thought to arise from adenomas (adenomatous polyps) that become dysplastic (figure 1). Adenomatous polyps form in the colon when normal mechanisms regulating epithelial renewal are disrupted. Surface cells lining the intestine are continuously lost into the bowel lumen due to apoptosis and exfoliation, and must be continuously replaced. Typically, proliferation occurs exclusively at the crypt base. As cells move towards the luminal surface, they cease proliferating and terminally differentiate. This ordered process is increasingly disrupted as adenomas increase in size, become dysplastic, and eventually attain invasive potential.
The hypothesis that invasive CRCs develop from intermediate precancerous precursors is supported by pathologic, epidemiologic, and observational clinical data (see "Overview of colon polyps", section on 'Adenomatous polyps'). Summarized briefly:
●Early carcinomas are frequently seen within large adenomatous polyps, and areas of adenomatous change can often be found surrounding human CRCs.
●Adenomas and carcinomas are found in similar distributions throughout the large bowel, and adenomas are typically observed 10 to 15 years prior to the onset of cancer in both sporadic and familial cases.
●In animal models, adenomas develop before carcinomas, and carcinomas develop exclusively in adenomatous tissue.
●The ability to reduce the incidence of CRC through removal of polyps has been shown in controlled trials in humans [6].
In 1990, Fearon and Vogelstein described the molecular basis for CRC as a multistep process in which each accumulated genetic event conferred a selective growth advantage to the colonic epithelial cell [5]. Later studies have served to further refine their hypothesis.
According to the Vogelstein model, germline or somatic mutations are required for malignant transformation, and the accumulation of multiple genetic mutations rather than their sequence determines the biological behavior of the tumor (figure 2). Germline mutations underlie the common inherited syndromes (eg, familial adenomatous polyposis [FAP], Lynch syndrome), while sporadic cancers result from the stepwise accumulation of multiple somatic mutations. Mutations in the adenomatous polyposis coli (APC) gene, a feature common to both inherited and sporadic CRCs, occur early in the process, while p53 tumor suppressor gene mutations generally occur late.
In addition to point mutations, other genetic changes that are implicated in human tumorigenesis include altered DNA methylation, and gene rearrangements, amplifications, overexpression, and deletions.
Serrated polyp pathway — The majority of CRCs are believed to progress through an adenoma-carcinoma sequence. However, more recent evidence increasingly supports the existence of an alternative route for colorectal carcinogenesis through serrated polyps, a group that encompasses a morphological spectrum, including hyperplastic polyps, mixed hyperplastic polyp/adenoma, and serrated adenomas. (See 'Hypermethylation phenotype (CIMP+) pathway' below and "Overview of colon polyps", section on 'Sessile serrated polyps and traditional serrated adenomas'.)
Molecular pathways to colorectal tumorigenesis — There appear to be at least three molecular pathways leading to colorectal tumorigenesis (table 1): the chromosomal instability (CIN) pathway, which is typified by the inherited condition FAP; the mutator-phenotype/DNA mismatch repair pathway, which is implicated in the inherited condition Lynch syndrome as well as in a proportion of sporadic CRCs in which there is loss of DNA mismatch repair protein function; and the hypermethylation phenotype hyperplastic/serrated polyp pathway, which is characterized by a high frequency of methylation of some CpG islands (CpG island hypermethylation phenotype [CIMP]-positive) [7].
The chromosomal instability (APC) pathway — CIN results from "gain of function" mutations [5]. These may result either in activation of growth promoting pathways including oncogenes or diminished activity of tumor suppressor genes or apoptotic pathways.
These tumors may be inherited (as with FAP) or sporadic, and they are characterized by gross chromosomal abnormalities including deletions, insertions, and loss of heterozygosity.
The mutator phenotype/mismatch repair pathway — Another pathway leading from adenomatous polyps, and probably from serrated adenomas, to invasive cancer has also been described, the mismatch repair pathway. This pathway is involved in CRCs arising in association with the inherited condition Lynch syndrome. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'Genetics'.)
The key element of this pathway is dysfunction of DNA mismatch repair (MMR) enzymes, resulting from germline mutations in one of several different DNA mismatch repair genes, most commonly MLH1 or MSH2. Cells with deficient DNA repair capacity due to silencing of MMR genes accumulate DNA errors throughout the genome. The biologic "footprint" is the accumulation of abnormalities in short sequences of nucleotide bases that are repeated dozens to hundreds of times within the genome; these are called microsatellites, and the tumors are described as having the phenotype of high levels of microsatellite instability (MSI-H).
Besides being the biologic hallmark of Lynch syndrome, high levels of MSI are also found in approximately 15 percent of sporadic CRCs. However, in most of these cases, gene silencing is not due to a specific MMR mutation, but to an epigenetic phenomenon, hypermethylation of the gene promoter for the MMR enzyme (usually MLH1), which leads to transcriptional silencing of gene expression [8]. (See 'Mismatch repair genes' below.)
Hypermethylation phenotype (CIMP+) pathway — Epigenetic alterations such as DNA hypomethylation and loss of imprinting, as well as DNA hypermethylation, can silence the expression of certain genes, including MMR enzymes [8-11]. CRCs that have a particularly high frequency of methylation of some CpG islands (in which a cytosine [C] base is followed immediately by a guanine [G] base that are linked with a phosphodiester bond [CpG]) are referred to as CIMP+ tumors [8]. The defect may result in hypermethylation of the promoter region of MMR enzymes such as MLH1 and silencing of gene expression [12].
Activating mutations in the BRAF gene (most of which are in the V600E codon) occur almost exclusively in MSI-H, CIMP+ CRCs that do not carry mutations in KRAS [13,14]. Lynch-related CRCs present only with KRAS and not BRAF mutations [13,14]. BRAF V600E mutations are particularly prevalent in smokers with sporadic CRCs [15]. The presence of a BRAF V600E mutation appears to abrogate the favorable prognosis that is typically associated with MSI-H tumors [16,17]. Whether the adverse prognosis associated with V600E BRAF mutations is also seen in individuals with non-V600 BRAF mutations is unclear. (See "Pathology and prognostic determinants of colorectal cancer", section on 'RAS and BRAF'.)
Sporadic CRCs with a high degree of microsatellite instability and BRAF mutations are a clinically distinct subgroup that is widely considered to develop from serrated polyps (table 1) [7,18,19]. (See "Overview of colon polyps", section on 'Sessile serrated polyps and traditional serrated adenomas'.)
SPECIFIC MOLECULAR ABNORMALITIES — The following sections will describe the major abnormalities in oncogenes, tumor suppressor genes, mismatch repair (MMR) genes, and epigenetic phenomena such as DNA hypomethylation and hypermethylation that underlie CRC tumorigenesis, including the importance of each factor in cell cycle control and the molecular and clinical consequences of individual mutations. This discussion is meant as an overview of the molecular pathogenesis of CRC rather than an exhaustive survey, which can be found elsewhere [20].
Oncogenes — Oncogenes are homologs of normal cellular genes that participate in cell growth pathways and cell cycle regulation. A mutational change in an oncogene leads to constitutive activation of the gene, which then results in uncontrolled cellular proliferation [21,22]. Because the normal gene function is activated, these are referred to as gain of function mutations.
Among the oncogenes implicated in sporadic CRC are RAS, SRC, MYC, and the human epidermal growth factor receptor 2 (HER2; previously called HER2/neu or ERBB-2); the most important is RAS (table 2) [23-27].
RAS — The RAS oncogene exists as three cellular variants, HRAS, KRAS, and NRAS. Although all three oncogenes, when mutated, have the ability to transform normal cells, KRAS is the most frequently mutated in human CRC [28-30]. The importance of RAS to colorectal tumorigenesis is underscored by the finding that CRC cells in which a mutated RAS gene has been removed or replaced lose their ability to form tumors in nude mice [31].
The RAS oncogenes encode a family of small proteins with homology to G-proteins that regulate cellular signal transduction by acting as a one-way switch for the transmission of extracellular growth signals to the nucleus [32]. These proteins normally cycle between an inactive guanosine diphosphate (GDP)-bound state and an active guanosine triphosphate (GTP)-bound state. RAS mutations, typically point mutations, leave the protein resistant to GTP hydrolysis by GTPase, resulting in a constitutively active GTP-bound protein and a continuous growth stimulus. Data from animal models suggest that RAS mutations may contribute to colorectal tumorigenesis by activating cancer stem cells that have already been activated by adenomatous polyposis coli (APC) mutations [33].
RAS mutations are found in up to 50 percent of sporadic CRCs and 50 percent of colonic adenomas larger than 1 cm; they are rarely seen in smaller adenomas [28,34]. At least two reports suggest that they are more common in proximal colon cancers than in more distal colorectal primaries [35,36]. KRAS has also been implicated in the process of tumor invasion and metastasis [37,38].
The lack of mutations in smaller adenomas suggests that RAS mutations are acquired during later adenoma progression [39]. However, RAS mutations are not limited to dysplastic colonic lesions. Up to 100 percent of nondysplastic aberrant crypt foci (ACF, believed to be the first intermediate between normal colonic mucosa and the adenomatous polyp) and 25 percent of hyperplastic polyps have RAS mutations, but their significance is unclear [28,31,32,40].
The identification of RAS mutations in CRC is of potential clinical relevance for both screening and therapy:
●The detection of RAS mutations in fecal material is a potentially sensitive screening method for the early diagnosis of CRC. Detection of KRAS mutations, in combination with aberrantly methylated BMP3 and NDRG4 promoter areas, beta-actin, and a test for stool hemoglobin are included in current multi-target stool DNA testing for colon cancer (Cologuard) [41]. The utility of this test for CRC screening is discussed elsewhere. (See "Tests for screening for colorectal cancer", section on 'Multitarget stool DNA tests with fecal immunochemical testing'.)
●Posttranslational modification of the RAS protein by the enzyme farnesyl transferase is necessary for activation, a fact that has been exploited for therapeutic purposes. The therapeutic potential of agents that target the RAS signal transduction pathway (eg, farnesyl transferase inhibitors) is being explored in patients with CRC whose tumors contain RAS mutations.
●The presence of a RAS mutation in CRC is significantly associated with the absence of response to agents targeting the epidermal growth factor receptor (EGFR) such as cetuximab. (See "Systemic chemotherapy for metastatic colorectal cancer: General principles", section on 'RAS mutations'.)
Tumor suppressor genes — In contrast to oncogenes, tumor suppressor genes normally have an inhibitory influence on the cell cycle. Once these genes are deleted or their function reduced, normal control mechanisms are no longer operative, and growth proceeds unchecked. At the cellular level, tumor suppressor genes act in a recessive fashion, meaning that the function of the normal protein is lost only when both copies (alleles) of the gene are inactivated by point mutations, rearrangements, or deletions.
Tumor suppressor genes were first described by Knudson in the context of childhood retinoblastoma (RB), which is caused by mutational inactivation of the retinoblastoma (RB1) gene, and presents either as hereditary or sporadic disease [42,43]. A "two-hit" model was proposed to explain the different clinical features of these two presentations (figure 3). (See "Retinoblastoma: Clinical presentation, evaluation, and diagnosis", section on 'Pathogenesis'.)
●The inherited form of RB requires a germline mutation that can be either inherited or de novo (ie, the result of a new germline mutation), plus a second somatic mutation, occurring later in development, that affects the remaining RB1 allele within retinal cells [22,42]. Affected individuals are at risk for multifocal and bilateral tumors.
●By contrast, in the nonhereditary form of RB, both allelic mutations arise spontaneously in a single retinal cell (ie, both are somatic mutations). The resulting phenotype is that of a unifocal, unilateral tumor that presents at a later age than the inherited variant and without heritable transmission to later offspring [43].
The first molecular evidence for the involvement of tumor suppressor genes in CRC came from the study of allelic loss, in which large chromosomal deletions were detected using polymorphic markers that distinguish the two alleles present in the germline. When comparing tumor alleles with those present in normal tissue, deletions were identified as "loss of heterozygosity" (LOH).
In early studies of CRC, LOH for chromosomes 5q, 8p, 17p, or 18q was detected in 36, 50, 73, and 75 percent of cases, respectively [34,44]. Presumptive tumor suppressor genes were subsequently identified on 5q (the location of the APC gene), 18q (the location of the deleted in colon cancer [DCC] gene, and the SMAD4 and SMAD2 genes), and 17p (location of the TP53 gene).
APC gene — Perhaps the most critical gene in the early development of CRC is the adenomatous polyposis coli (APC) gene (figure 4). Somatic mutations in both alleles are present in 80 percent of sporadic CRCs, and a single germline mutation in this gene is responsible for familial adenomatous polyposis (FAP), a dominantly inherited syndrome characterized by the development of hundreds to thousands of colorectal polyps by the second or third decade of life. (See "Clinical manifestations and diagnosis of familial adenomatous polyposis", section on 'Genetics'.)
A germline APC mutation is also thought to contribute to the development of familial CRC in Ashkenazi Jews [22,45,46]. A thymine to adenine transversion at nucleotide 3920 in the APC gene that resulted in a substitution of lysine for isoleucine at codon 1307 (I1307K) is found in 6 percent of all persons of Ashkenazi Jewish descent but in a higher frequency of Ashkenazi Jews with both a personal and family history of CRC (28 percent) [45]. This mutation was previously thought to represent a polymorphism.
The recognition of the importance of the APC gene began with genetic studies linking inheritance of the FAP syndrome to chromosome 5q21 and the subsequent identification of germline mutations involving a gene at this locus, the APC gene. The earliest malignant lesions in these patients, dysplastic aberrant crypt foci (microadenomas) and small adenomatous polyps, have lost the second APC allele (through deletion or somatic mutation), suggesting that loss of APC expression is a very early event in CRC tumorigenesis [47,48].
The function of the APC gene product and the mechanism whereby the abnormal gene promotes tumor formation are beginning to be understood. An important clue was the observation that most sporadic CRCs with normal or wild-type APC had mutations in the CTNNB1 gene, which encodes beta-catenin, a protein involved in the same signaling cascade as the APC gene product, the Wnt (Wingless-type) signaling pathway [49,50]. It is now hypothesized that most sporadic CRCs are initiated by activating Wnt pathway mutations, characterized by the stabilization of beta-catenin and constitutive transcription by a beta-catenin/T-cell factor (Tcf)-4 complex.
The Wnt pathway is an evolutionarily conserved signal transduction pathway that is necessary for embryonic development [50-52]. It also plays a central role in supporting intestinal epithelial renewal, an important fact since CRC is thought to originate in the expansion of colonic crypt cells.
The basic features of the Wnt signaling pathway are depicted in the figure (figure 5). The normal APC protein appears to prevent the accumulation of cytosolic and nuclear beta-catenin by mediating its phosphorylation and resultant degradation. The majority of mutations in the APC gene (both germline and somatic) lead to premature truncation of the APC protein and loss of its beta-catenin regulatory domains (figure 4). Loss of functional APC (as well as mutations in the beta-catenin gene) results in the nuclear accumulation of beta-catenin, which binds and activates the transcription factor Tcf-4 [53-55].
It is proposed that beta-catenin/Tcf-4 acts as a switch controlling proliferation versus differentiation in the intestinal crypt epithelial cells [56,57]. Activation of this pathway prevents the cells from either entering G1 arrest or undergoing terminal differentiation and induces resistance to apoptosis [58]. The end result is cellular proliferation. In addition, because several other cell signaling pathways converge with the Wnt pathway, it represents a "final common pathway" through which multiple abnormalities affecting other cellular signaling pathways may result in the same carcinogenic result.
In addition to the typical loss of function tumor suppression mechanism of carcinogenesis, more recent evidence suggests that C-terminally truncated APC may also have gain of function properties; this mechanism of carcinogenesis promotion is typically attributed to oncogenes and not tumor suppressor genes (figure 4). Studies have reported that a truncated APC gene promotes cell survival through regulation of the BCL2 gene [59,60], stimulates cell migration through mediation of a guanine nucleotide exchange factor termed Asef [61], and activates proliferation of human colonic epithelial cells in culture [62,63].
Other mechanisms may also contribute to the tumorigenic potential of APC mutations. Mutations in APC (but not CTNNB1) are associated with chromosomal instability [64], predisposing the cell to "hits" in other genes that may contribute to tumor progression and malignant transformation.
TP53 gene — The TP53 gene on chromosome 17p is the most commonly mutated gene in human cancer. In approximately 50 to 70 percent of CRCs, TP53 inactivation occurs by a mutation of one allele followed by loss of the remaining wild type gene [34,65-68]. 17p sequences are lost in as many as 75 percent of CRCs, while they are rarely lost in adenomas and aberrant crypt foci, suggesting that loss of p53 function represents a relatively late event in colorectal tumorigenesis [5,34,69]. In keeping with this hypothesis, a large international study of 3583 CRCs found an increase in the frequency of TP53 mutations with advancing disease stage [68].
The normal "wild-type" TP53 gene produces a DNA-binding protein p53 that acts as a transcriptional activator of growth inhibitory genes. Wild-type p53 may be particularly critical when cells are under stress. Normally, cells arrest their growth in response to DNA damaging agents and other stressors (eg, hypoxia) via induction/activation of p53 [70-72]. Once activated, p53 induces a variety of growth-limiting responses, including cell cycle arrest (in order to facilitate DNA repair), apoptosis, senescence, and differentiation. p53 produces these responses largely by altering the expression of a number of target genes, at least 20 of which have been described as being under transcriptional control of p53. Because of its central role in preventing the propagation of cells with DNA damage, p53 has been referred to as the "guardian of the genome" [73].
Although inactivation of the p53 pathway appears to be a late event in the majority of human CRCs, it may represent an earlier event in inflammatory bowel disease-related CRC. (See "Surveillance and management of dysplasia in patients with inflammatory bowel disease", section on 'Molecular pathogenesis'.)
Given the association of p53 pathway inactivation with CRC, it is unclear why patients with the Li Fraumeni syndrome (a condition caused by a germline mutation in TP53 in which patients frequently develop carcinomas, sarcomas, and leukemias) are not at particularly increased risk of developing CRCs. However, germline TP53 mutations may be associated with early onset CRC. In a report derived from the population- and clinic-based Colon Cancer Family Registry, germline TP53 mutations were identified in 6 of 457 individuals (1.3 percent) diagnosed with CRC at age 40 or younger [74]. This frequency is roughly comparable to the prevalence of germline APC mutations in CRC. (See "Li-Fraumeni syndrome".)
The identification of TP53 mutations in an individual CRC is of potential clinical significance, prognostically and therapeutically. In many but not all studies, patients whose tumors harbor TP53 mutations have worse outcomes and shorter survival than those without such mutations. At least some of the discordant results may be due to the fact that the prognostic influence of TP53 abnormalities appears to depend on tumor site, type of mutation, and the use of adjuvant therapy [68]. This topic is discussed in detail elsewhere. (See "Pathology and prognostic determinants of colorectal cancer", section on 'Molecular factors'.)
From a therapeutic standpoint, it is hoped that p53 may prove to be a highly selective and effective target for intervention. Several new therapies under study for advanced disease specifically target TP53-mutant cells, while others seek to correct the TP53 mutations directly or to restore the integrity of the p53 pathway [60,75-77].
Chromosome 18q: The DCC, SMAD4, and SMAD2 genes — As with the TP53 and APC genes, the first evidence of a tumor suppressor gene on chromosome 18q came from studies of allelic loss in CRC. In an early study, one copy of 18q was lost in 73 percent of sporadic CRCs and 47 percent of large adenomas with foci of invasive cancer, but in fewer than 15 percent of less advanced adenomas [34]. In 1989, a candidate gene termed the "deleted in colorectal carcinoma" (DCC) gene was identified at 18q21 [78], and point mutations in the DCC gene have been identified in CRCs [79,80].
Gene mutations presumably lead to a loss of expression of the DCC protein, which is thought to have a role in cell-cell or cell-matrix interactions [78,81,82]. DCC is normally expressed in many tissues, including the colonic mucosa, although its normal function has been difficult to elucidate because of its large size and the lack of expression in CRCs [81].
Loss of DCC expression may have prognostic value, particularly in patients with early stage CRC. Five-year survival rates seem to be worse for patients with stage II (node-negative (table 3)) CRCs that lack DCC expression compared with those that express it [83]. For patients with DCC-negative stage II disease, prognosis more closely approximates that of patients with more advanced stage III (node-positive) disease [84].
The ultimate benefit of this information may be the identification of a subgroup of patients with stage II colon cancer who might benefit from adjuvant chemotherapy. While appealing, there are no prospective data that currently support the validity of this strategy. (See "Adjuvant therapy for resected stage III (node-positive) colon cancer".)
A second tumor suppressor gene at 18q was identified during the course of investigation of allelic losses in pancreatic cancer, termed the DPC4 (deleted in pancreatic cancer) gene, now redesignated SMAD4 [34]. (See "Molecular pathogenesis of exocrine pancreatic cancer", section on 'Tumor suppressor genes'.)
The SMAD4 gene encodes a protein that may be important to the signaling pathway of the transforming growth factor beta (TGF-beta) superfamily of signaling polypeptides. TGF-beta suppresses the growth of most normal cells by binding to type I (TGFBR1) and type II (TGFBR2) transmembrane receptors, but many cancer cells are resistant to this growth-suppressive effect. In CRC cells, SMAD4 is required for TGF-beta signaling, and at least in vitro, reintroduction of the gene (via transfection of an intact chromosome 18) is associated with restoration of TGF-beta sensitivity [85,86].
Mutations in SMAD4 or a third putative tumor suppressor gene that also maps to 18q (SMAD2) have been found in a subset of sporadic CRCs [63,87-89]. Perhaps more importantly, germline mutations in SMAD4 (and in BMPR1A [ALK3], a gene that codes for a member of the TGF-beta receptor superfamily, which is located upstream from SMAD4 [90]) have also been identified in patients with juvenile polyposis. These patients develop multiple juvenile polyps that are distinct from adenomas, and they are at an increased risk for invasive CRCs. (See "Juvenile polyposis syndrome", section on 'Genetics'.)
TGF-beta signaling — As noted above, one mechanism by which CRC cells escape the normal inhibitory influence of transforming growth factor beta (TGF-beta) is through SMAD4 mutations, which interfere with the production of a protein that is required for TGF-beta signaling. Other possible mechanisms of interference with normal TGF-beta signaling in CRC cells have also been identified, including inactivating mutations in the TGF-beta receptor, type II (TGFBR2) gene [91] and molecular changes that result in the redirection of TGF-beta growth inhibitory signals into growth stimulatory signals.
BRCA1 and BRCA2 genes — Breast cancer associated (BRCA) 1 and 2 are tumor suppressor genes known to function in homologous recombination [92,93]. BRCA1 operates in both DNA repair and checkpoint activation, and it has a role that is upstream to that of BRCA2, which functions in the primary mechanism of homologous recombination. Germline mutations in BRCA1 and BRCA2 are predominantly associated with breast and ovarian cancers. BRCA2 and, to a lesser extent, BRCA1 have been associated with pancreatic cancers, prostate cancers, gastric cancers, and CRCs. Three specific founder mutations in BRCA1/2 have been shown to be frequent in individuals of Ashkenazi descent. (See "Cancer risks and management of BRCA1/2 carriers without cancer", section on 'Cancer risks in BRCA1/2 carriers' and "Genetic testing and management of individuals at risk of hereditary breast and ovarian cancer syndromes", section on 'Population-based testing for those of Ashkenazi Jewish descent'.)
Although the data regarding the risk of CRC in BRCA carriers have been inconsistent, a prospective study of BRCA1 mutation carriers suggested an increased risk of CRC in women age 30 to 49, which returned to population-level risk after age 50; no increased risk was noted for BRCA2 mutation carriers [94]. Similarly, a meta-analysis of 18 cohort and case-control studies suggested that carriers of BRCA1 (but not BRCA2) have a higher risk for CRC (odds ratio [OR] 1.49, 95% CI 1.14-1.85) [94-96].
It is often difficult to determine genotype/phenotype causation, as opposed to association, unless molecular genetic analysis is undertaken [97]. A conclusion to support BRCA1/2 causation, as opposed to association, with CRC, as of yet, has no supportive molecular data. This issue will become more common as multigene genetic testing becomes predominant. One study of multigene panel testing in individuals suspected of having Lynch syndrome by National Comprehensive Cancer Network (NCCN) criteria revealed that 5.6 percent of individuals (71 of 1260) had non-Lynch cancer predisposition germline mutations [98,99]. Fifteen of these individuals had BRCA1/2 mutations; 93 percent met NCCN criteria for Lynch syndrome genetic testing, yet only 33 percent met NCCN criteria for BRCA1/2 genetic testing.
A decision regarding referral for genetic testing in individuals with CRC should take cancer-related risk factors into account, including personal history of cancer, familial history of cancer, and any prior genetic testing of family members. (See "Genetic testing and management of individuals at risk of hereditary breast and ovarian cancer syndromes", section on 'Criteria for genetic risk evaluation'.)
Given the lack of information on the age of onset of CRC in at-risk individuals, it is difficult to make recommendations about whether or not BRCA1 mutation carriers should initiate CRC screening at an earlier age than is typically recommended. This subject is discussed in detail elsewhere. (See "Colorectal cancer: Epidemiology, risk factors, and protective factors", section on 'Hereditary CRC syndromes'.)
Mismatch repair genes — MMR genes are responsible for correcting the ubiquitous nucleotide base mispairings and small insertions or deletions that occur during DNA replication [100-103]. Several of these genes exist, including hMSH2 (human mutS homolog 2), hMLH1 (human mutL homolog 1), hPMS1 and hPMS2 (human postmeiotic segregation 1 and 2), hMSH6 (human mutS homolog 6), and hMLH3, an MMR gene that interacts with MLH1.
Germline mutations in one of the MMR genes appear to be the underlying genetic defect in most kindreds with Lynch syndrome, and loss of expression of MMR genes can also be found in approximately 15 percent of sporadic CRCs [104-106]. The biologic footprint of tumors that have deficient MMR is a high level of microsatellite instability (MSI-H). (See 'Microsatellite instability high versus low' below.)
However, sporadic tumors with defective expression of MMR genes do not contain MMR gene mutations; instead, they have epigenetic changes (acquired hypermethylation of the promoters of both alleles of MLH1 gene) that silence gene expression. (See 'Epigenetic alterations affecting mismatch repair genes' below.)
Another group is described as having "Lynch-like" syndrome; these are patients who have MSI-H tumors and loss of expression of one or more of the MMR genes but have neither a germline MMR mutation or promoter hypermethylation of the MLH1 gene. At least some of these patients have biallelic somatically acquired MMR gene mutations [107,108]. The risk of CRC appears to be lower in these families than it is in families with Lynch syndrome but higher than that of families with sporadic CRC [109]. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'Differential diagnosis'.)
These patients are not to be confused with another subgroup of patients who meet clinical criteria for Lynch syndrome (eg, the Amsterdam criteria (table 4)) but do not have a discernible mutation in one of the MMR genes (ie, they are MMR-proficient). This subset of patients is sometimes referred to as familial colorectal cancer syndrome type X (FCCTX), to distinguish them from Lynch syndrome [3,110-112]. In contrast to Lynch syndrome, these patients do not carry a risk for extracolonic tumors, and the histopathologic features that characterize Lynch-related colon cancers are absent [112,113]. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'Differential diagnosis'.)
Although FCCTX is termed "familial," this may be a misnomer, as these patients typically present as isolated cases, likely caused by an as-of-yet undefined germline mutation. At least some data suggest that gain of genetic material on chromosome 20q and loss on chromosome 18q may serve to discriminate between CRCs associated with FCCTX and Lynch syndrome [114,115].
Testing strategies for Lynch syndrome in patients with CRC are discussed in detail elsewhere. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'Identification of individuals at risk for Lynch syndrome'.)
Microsatellite instability high versus low — Cells that are MMR deficient accumulate DNA errors throughout the genome [116]. The biologic "footprint" of an MMR defect is the accumulation of abnormalities in short sequences of nucleotide bases that are repeated dozens to hundreds of times within the genome; these are called microsatellites [116]. Several critical growth regulatory genes (eg, the TGFRB2, BAX, insulin-like growth factor 2 receptor [IGF2R]) contain microsatellites in the promoter region and are therefore susceptible to frameshift mutations. This leaves the cell vulnerable to mutations in these genes controlling cell growth, and this phenomenon is termed MSI [91,116,117].
Many but not all tumors that contain MMR mutations can be identified by the presence of MSI-H. The majority of patients with Lynch syndrome have MSI-H tumors. In addition, approximately 15 percent of sporadic tumors are MSI-H. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis".)
In contrast to microsatellite-stable CRCs, sporadic tumors with MSI-H have characteristic clinicopathologic features. They tend to occur in the proximal colon, have a greater mucinous component, contain lymphocytic infiltration, and are more often poorly differentiated. Interestingly, the tendency to have a lymphocytic infiltrate likely reflects immune activation from T-cells directed against tumor-specific carboxy-terminal frameshift peptides that are associated with microsatellite instability [118].
Although tumors in Lynch syndrome tend to be poorly differentiated, the presence of MSI mitigates the adverse prognostic impact of this feature, and in fact, MSI-H tumors are associated with longer survival in both Lynch syndrome and sporadic cases, for unclear reasons. This topic is discussed in detail elsewhere. (See "Pathology and prognostic determinants of colorectal cancer", section on 'Histologic type, grade of differentiation, and presence of mucin' and "Pathology and prognostic determinants of colorectal cancer", section on 'Mismatch repair deficiency'.)
Most laboratories use a panel of several microsatellite loci when testing for MSI [119]. A panel consisting of three dinucleotide repeats and two mononucleotide repeats has been proposed as a standard test for MSI; a tumor is called MSI-H when at least two (40 percent) are affected by instability [104]. While most tumors show either a high degree of instability or no unstable markers, a minority display instability in <40 percent of the markers studied. These tumors are referred to as MSI-low (MSI-L). While almost all MSI-H tumors are MMR deficient, most MSI-L tumors have no MMR defect [120,121]. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'Genetics'.)
The significance of the MSI-L phenotype is incompletely understood [120,122,123]. It is not associated with silencing of MLH1 or any of the other known DNA MMR genes, but there is evidence to suggest that MSI-L is a nonrandom phenomenon with a biologic basis [123-125]:
●Some data suggest that MSI-L in sporadic CRC reflects loss of the MutS Homologue 3 (MSH3) gene [123].
●The MSI-L phenotype has also been associated with methylation of the promoter for the DNA repair gene O-6 methylguanine DNA methyltransferase (MGMT) [126]. It has been hypothesized that loss of expression of the MGMT gene results in accumulation of methyl G:T mismatches and excess stress on the MMR system, which ultimately leads to MSI-L [126,127]. While rare in Lynch syndrome, MGMT promoter methylation or loss of MGMT gene expression occurs in up to 25 percent of sessile serrated adenomas, 78 percent of dysplastic serrated adenomas, and 50 percent of serrated adenocarcinomas [127,128]. These findings suggest that the factors leading to the MSI-L phenotype may be important in the serrated neoplasia/CIMP pathway (table 1).
In any case, accumulating evidence suggests that the MSI-L phenotype is associated with a poor clinical outcome in both stage II and III CRC [123,129-131].
Epigenetic alterations affecting mismatch repair genes — As noted above, mutations and allelic loss of one of the MMR genes are responsible for the MSI-H phenotype in most cases of Lynch syndrome. In contrast, hypermethylation of the promoter region of some MMR genes and/or DNA hypomethylation with loss of imprinting (ie, silencing of gene expression) is thought to underlie cases of sporadic CRC that display the MSI-H phenotype [100,132-137]. Epigenetic inactivation of the second normal MMR gene allele may also play a role in individuals with Lynch syndrome, in whom the second allele must be inactivated in order to progress to cancer [138,139]:
●DNA hypermethylation specifically targets CpG dinucleotides, which are present in the promoters of many genes (including the MMR gene hMLH1). Although the stimulus that drives hypermethylation remains unknown, methylated CpG is bound by a family of proteins known as methyl-CpG binding domain proteins. These proteins, in turn, form a multiprotein complex that alters chromatin conformation and silences expression [140-142]. (See 'Hypermethylation phenotype (CIMP+) pathway' above.)
●Another mechanism of epigenetic inactivation is DNA hypomethylation and loss of imprinting (LOI) [11]. Imprinting refers to the selective loss of expression of parent lineage-specific genes due to selective methylation of one allele. (See "Inheritance patterns of monogenic disorders (Mendelian and non-Mendelian)", section on 'Parent-of-origin effects (imprinting)'.)
This pattern is set in the zygote and maintained during development to suppress the expression of the maternal or paternal copy of an allele. LOI for the insulin-like growth factor 2 gene (IGF-2) has been highly associated with MMR-deficient CRC [136,137,143]. However, LOI for IGF2 is found more often in the normal colonic mucosa and peripheral blood lymphocytes of patients with CRC compared with those without the disease [136,144]. These findings have led some to hypothesize that LOI for IGF2 represents a risk factor for CRC rather than a somatic defect underlying tumorigenesis [144].
●Large deletions in the 3' end of the epithelial cell adhesion molecule (EPCAM) gene leads to hypermethylation of the adjacent MSH2 gene on chromosome location 2p21, which has the effect of silencing expression of the MSH2 gene [145-147]. Up to 6 percent of Lynch patients have an EPCAM epigenetic mutation. (See "Lynch syndrome (hereditary nonpolyposis colorectal cancer): Clinical manifestations and diagnosis", section on 'Genetics'.)
MUTYH defects and familial colorectal cancer — A small proportion of patients with multiple colorectal adenomas and a family history of CRC have germline mutations (often biallelic) in the base excision repair gene mutY homolog (MUTYH), sometimes in conjunction with somatic mutations in the APC gene [148-150]. These mutations predispose patients to recessive inheritance of multiple colonic adenomas and a variant form of FAP, frequently referred to as MUTYH-associated polyposis (MAP). In one series of 152 patients with multiple adenomas seen at one institution, 7.5 percent of those without a germline APC mutation were found to have two separate germline MUTYH mutations [148]. These findings have implications for screening strategies in patients suspected of having FAP, which in most cases is inherited in an autosomal dominant pattern. (See 'APC gene' above.)
Perhaps more importantly, an increasing number of reports suggest that germline MUTYH mutations may account for a substantial fraction of familial CRCs that occur in the absence of a dominantly inherited familial syndrome [150-155]. In a population-based series that screened 1238 patients diagnosed with CRC over a three-year period and 1255 healthy age- and gender-matched control subjects without a personal history of cancer for germline MUTYH mutations (Y165C and G382D), mutation carriers were significantly more likely to develop CRC and were more likely to have first- or second-degree relatives with CRC [152].
The phenotype is variable, ranging from 10 or more polyps to widespread polyposis with associated CRC. The risk of CRC among carriers of biallelic mutations appears to be high; it was close to 100 percent by age 60 in one study [156], and 43 percent in another [157]. Some data suggest that heterozygotes have a slightly increased risk of CRC [152,158,159], while other series report no increased risk in those with monoallelic mutations [157].
The prognosis may be more favorable in CRCs that develop in the setting of MAP compared with those that arise in the general population [160]. In a retrospective European cohort study comparing 147 patients with MAP-related CRC and 272 population-based matched control patients with CRC, survival was significantly better for the patients with MAP-associated CRC, even after adjustment for differences in age, stage, sex, subsite, country, and year of diagnosis (hazard ratio [HR] for death 0.48, 95% CI 0.32 to 0.72). Though compelling, these findings need to be replicated in independent, prospective studies.
Modifier genes — In addition to the genes described, several other genes seem to be important in colorectal carcinogenesis, although their exact roles and mechanisms of tumorigenesis have not been fully determined.
COX-2 — A substantial body of evidence supports a protective effect of aspirin and other cyclooxygenase (COX) inhibitors on the development of CRC. Furthermore, one nonspecific COX inhibitor, sulindac, can cause polyp regression in patients with FAP. The mechanism underlying these effects is not well understood, but they suggest a role for the COX-2 gene, which is upregulated in CRC cells, in colorectal tumorigenesis. This is an area of active investigation and is discussed in detail elsewhere. (See "NSAIDs (including aspirin): Role in prevention of colorectal cancer".)
PPAR gene — The peroxisome proliferator-activating receptor (PPAR) gene has been implicated in colorectal carcinogenesis. The PPAR gene encodes a family of nuclear receptors that function as transcriptional regulators for proteins controlling lipid metabolism and cell growth. Activation of these receptors inhibits cell growth and promotes differentiation in a variety of epithelial cell types, including CRC cells [161]. Preliminary studies demonstrate that PPAR gene is downstream from the APC gene and may also be involved in the COX pathway [162-165].
Loss of function mutations in PPAR has been described in sporadic CRCs [166]. Furthermore, there is some evidence that abnormalities in the PPAR genes are responsible for the increased frequency of CRCs and adenomas in patients with acromegaly. (See "Colorectal cancer: Epidemiology, risk factors, and protective factors", section on 'Other risk factors'.)
SUMMARY
●Colorectal cancer (CRC) represents an ideal model for the study of the molecular pathogenesis of cancer due to the accessibility of tissue for biopsy and the clear progression from normal colonic epithelium to invasive cancer via an intermediate precursor, the adenomatous polyp. A multistep process of specific genetic changes is thought to drive the transformation from normal colonic epithelium to invasive cancer. Single, specific germline mutations underlie the common inherited syndromes (eg, adenomatous polyposis coli [APC], Lynch syndrome), while sporadic cancers result from the stepwise accumulation of multiple somatic mutations. Mutations in the APC gene occur early, while others, such as mutations of the TP53 suppressor gene, generally occur late in the process (figure 2). (See 'Molecular pathogenesis of colorectal cancer' above.)
The concept of microsatellite instability (MSI) provides an additional layer of complexity to this model. The MSI-high (MSI-H) phenotype is associated with Lynch syndrome and is also observed in approximately 10 to 15 percent of sporadic CRCs. (See 'The mutator phenotype/mismatch repair pathway' above.)
●The available data suggest that most colorectal neoplasias begin with inactivation (through a germline or sporadic mutation) of the APC gene, whether or not MSI is present. However, the genetic events may thereafter diverge, depending on the mechanism underlying the genetic instability (ie, chromosomal instability, germline mutations in DNA mismatch repair [MMR] enzymes, CpG island hypermethylation phenotype [CIMP+]). This has led to the molecular classification of CRC into three carcinogenic pathways (table 1) [7]. (See 'Molecular pathways to colorectal tumorigenesis' above.)
Compared with patients with chromosomal instability (which includes those with familial adenomatous polyposis [FAP] and the majority of sporadic CRCs), the MSI-H or "mutator" pathway involves a unique destabilizing mechanism and inactivation of a different set of genes (which encode the DNA MMR proteins). The CIMP+ pathway involves hypermethylation of the gene promoter for a DNA MMR gene. Regardless of the molecular pathway involved, the final result is the same.
●In addition to providing insight as to the biology of all CRCs, the identification of specific genetic mutations responsible for colorectal tumorigenesis has had a direct influence on clinical care. Patients at highest risk for developing CRC can be identified via genetic testing for specific germline mutations. (See 'Tumor suppressor genes' above and 'Mismatch repair genes' above and 'MUTYH defects and familial colorectal cancer' above.)
Furthermore, new molecular screening methods for early detection of CRC via detection of mutations in fecal material are under study. In addition, these mutations are also being examined as prognostic markers and as potential therapeutic targets. (See 'RAS' above.)
ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges Paula M Calvert, MD, who contributed to an earlier version of this topic review.
