FAT, BILE ACIDS, AND BACTERIA

Several lines of evidence suggest that diets containing large percentages of fat predispose to CRC, especially in the descending and sigmoid colon. Colon cancer rates are high in populations whose total fat intake is high and are lower in those who consume less fat. On average, fat (saturated plus unsaturated) constitutes 40% to 45% of total calorie intake in Western countries with high rates of CRC, whereas in low-risk populations, fat accounts for only 10% to 15% of dietary calories. Case-control and cohort studies also have suggested that the incidence and mortality rates from colon cancer, and in some cases rectal cancer, are positively correlated with dietary fat, but these findings are less convincing than data from descriptive epidemiologic studies.

Early trials also often failed to take into account total energy intake. A prospective study assessed the relationship of meat, fat, and fiber intake among 88,751 women aged 34 to 59 years. After adjustment for total energy intake, the intake of animal fat was significantly correlated with the risk of colon cancer. The intake ratio of red meat to chicken and fish was strongly associated with increased incidence of colon cancer, perhaps owing to differences in their fat composition. Similar findings have been reported that correlate the intake of saturated fat and the ratios of red meat to chicken and fish intake with both the incidence and recurrence of colorectal adenomas in women. Cohort studies and a combined analysis of 13 case-control studies that adjusted for total energy intake, however, failed to provide clear-cut evidence for the association between dietary fat and CRC observed in earlier studies. Studies that specifically examine the association between intake of saturated or animal fat and CRC suggest a stronger association than with total fat.

Animal studies lend additional support for the role of dietary fat in the development of colon cancer. These studies usually involve the injection of carcinogens such as 1,2-dimethylhydrazine (DMH) or azoxymethane into rodents fed various diets. Animals fed a variety of polyunsaturated and saturated fats develop greater numbers of carcinogen-induced colonic adenocarcinomas than do those on low-fat diets. The amount and source of dietary fat might affect tumor development in such studies; fatty acids derived from polyunsaturated fish oils (ω-3 fatty acids) and monosaturated olive oil might not promote tumors to the extent that other polyunsaturated fats do.

It has been proposed that dietary fat enhances cholesterol and bile acid synthesis by the liver, thereby increasing the amounts of these sterols in the colon. Colonic bacteria convert these compounds to secondary bile acids, cholesterol metabolites, and other potentially toxic metabolic compounds. Population studies demonstrate increased excretion of sterol metabolites and fecal bile acids in groups that consume a high-fat, low-fiber Western diet compared with other groups, and high fecal bile acid levels are found in some patients with CRC. Dietary fat also has been shown to increase the excretion of secondary bile acids in carcinogen-treated rats; secondary bile acids do not act as primary carcinogens but as potent promoters of colon carcinogenesis in such animal models. Little is known about how lipid and sterol metabolites promote tumors, but both bile acids and free fatty acids have been shown to damage the colonic mucosa and increase the proliferative activity of its epithelium. Dietary consumption of high amounts of corn oil and beef fat increase colonic ornithine decarboxylase levels, which are associated with rapidly proliferating mucosa.

Activation of protein kinase C (PKC) by bile acids in colonic mucosa also might represent an important intracellular event by which bile acids provoke a proliferative response. Mucin alterations are a common feature of colonic neoplasia, and alterations in MUC2 mucin have been associated with tumor progression in the colon. Bile acids induce mucin expression in human colon carcinoma cells by increasing MUC2 transcription through a process involving mitogen-activated protein (MAP) kinase-independent, PKC-dependent activation of the transcription factor AP-1.⁴ Bile acids can, in addition, induce release of arachidonate and conversion of arachidonic acid to prostaglandins in the mucosa, which can enhance cell proliferation.⁵ Preclinical and clinical evidence indicate that nonsteroidal anti-inflammatory drugs (NSAIDs), which reduce prostaglandin synthesis, reduce the incidence of large bowel cancer (discussed later); inhibition of the inducible enzyme cyclooxygenase (COX)-2 may be particularly important in this regard.⁶

Certain fatty acids could promote carcinogenesis by altering membrane fluidity after being incorporated into cell membranes. Bacterial enzymes such as 7-α-dehydroxylase (which converts cholic to deoxycholic acid), β-glucuronidase, nitroreductase, and azoreductase may be induced by a high-fat diet and also could convert compounds ingested in the diet to active carcinogens (see later).

An inverse relationship has been reported between physical activity and risk for CRC in men; obesity is associated with elevated risk of CRC. Serum cholesterol and β-lipoprotein levels have been positively correlated with the development of colorectal adenomas and CRC, but this association has not been demonstrated consistently, and serum cholesterol levels can decline before the development of colon cancer. Adiponectin is a hormone secreted by adipose tissue, serum levels of which are inversely correlated with obesity and hyperinsulinemia. Variants of adiponectin and adiponectin receptor genes have been correlated with differences in CRC risk.⁷

FIBER

Epidemiologic, case-control, and animal studies suggest that dietary fiber protects against the development of colon cancer. Dietary fiber is plant material that resists digestion and that is composed of a heterogeneous mix of carbohydrates (e.g., cellulose, hemicellulose, pectin) and noncarbohydrates (e.g., lignin). Although the protective role of fiber is not completely clear, epidemiologic studies correlate high-fiber intake with a lower incidence of colorectal neoplasia.^8,9 The majority of observational-epidemiologic and case-control studies support the protective effect of fiber-rich diets; however, these data do not define the relationship between fiber-rich food and the importance of nonfiber vegetable components, nutrients, and micronutrients in fruits and vegetables. The effect of fiber components on different portions of the large bowel also can vary. This might explain, in part, the inability to demonstrate a protective effect of fiber in several randomized, controlled trials that have examined the ability of fiber supplementation to prevent adenoma recurrence.^10,11

Two large prospective observational studies, including one from the Prostate, Lung, Colon and Ovarian Cancer Trial (PLCO) found that increased fiber intake is significantly associated with reduced risk of colorectal neoplasia.⁸ Another analysis of 13 prospective cohort studies included in the Pooling Project of Prospective Studies of Diet and Cancer demonstrated that dietary fiber intake was inversely associated with risk of CRC in age-adjusted analyses, but that after accounting for other dietary factors, high intake of dietary fiber was not associated with a reduced risk of CRC.¹²

Investigators postulate that fibers such as cereal bran exert their protective role by increasing stool bulk, thereby diluting carcinogens and promoters of carcinogenesis, enhancing their elimination, and minimizing their duration of mucosal contact by decreasing intestinal transit time. Increased fiber intake, in the form of whole wheat and rye bread, also reduces the concentration of fecal secondary bile acids and fecal mutagens in healthy subjects. Animal studies also have demonstrated a decreased incidence of colonic tumors in DMH-treated rats fed diets high in fiber and fiber components, such as wheat bran, cellulose, and hemicellulose. Cellulose and hemicellulose decrease the levels of bacterial metabolic enzymes, such as β-glucuronidase, in experimental animals and can diminish the activation of carcinogens or cocarcinogens. Furthermore, some fiber components can bind to toxic or carcinogenic substances, perhaps decreasing their contact with the colonic mucosa. Fiber components also are fermented by fecal flora to short-chain fatty acids, thereby decreasing colonic pH and potentially inhibiting carcinogenesis.

CARCINOGENS AND FECAL MUTAGENS, VITAMINS, AND MICRONUTRIENTS

The possibility that specific genotoxic carcinogens might play a role in the genesis of CRC was raised when it was noted that the stools of certain persons exhibited mutagenic activity for bacteria in vitro. Mutagenic activity often is present in the feces of populations at high risk for large bowel cancer and is low or absent in low-risk populations. A specific group of highly unsaturated reactive compounds synthesized by colonic bacteria, fecapentaenes, might play a role in large bowel carcinogenesis. It also has been recognized that “charbroiled” meat and fish, and to a lesser extent fried foods, contain powerful mutagenic compounds. The structures of these compounds resemble a heterocyclic amine known to cause colon cancer in rodents. Metabolites similar to those of fried meat and fish are being sought as mutagens in human stools. A possible association between rectal cancer and beer- and ale-drinking also has been noted. A two-fold-to-three-fold increase in CRC also has been observed in automotive pattern and model makers, but the specific carcinogenic agent in this industrial environment has not yet been identified.

The exact nature of genotoxic carcinogens that might act in the human colon remains speculative, but the identification of such compounds could provide a basis for intervention and primary prevention of CRC. Limited data suggest that foods rich in carotene (vitamin A) and vitamin C could act as antioxidants in the chemoprevention of colon cancer, but prospective trials have failed to demonstrate such an effect. Other areas that merit further exploration in the prevention of CRC include the role of yellow-green cruciferous vegetables and the role of micronutrients including selenium salts, vitamin E, and folic acid. A good deal of attention has been given to a possible role for dietary calcium in preventing colon cancer (discussed next).

CALCIUM AND VITAMIN D

Epidemiologic, clinical, and laboratory evidence suggest that calcium intake might protect against carcinogenesis in the colon. Calcium has numerous biological effects that might reduce colon carcinogenesis, including actions on the cell cycle, cyclic adenosine monophosphate (cAMP), calmodulin, tyrosine kinases, ornithine decarboxylase, and e-cadherin. The calcium-sensing receptor expressed in the intestine senses extracellular calcium with effects on differentiation and proliferation. The potential chemopreventive activity of calcium was suggested originally by epidemiologic studies reporting an inverse relationship between CRC and intake of vitamin D and calcium. Dietary calcium supplementation in the form of low-fat dairy foods can affect a variety of intermediate biomarkers thought to be associated with tumor progression in the colon, and supplemental calcium plus vitamin D alters preneoplastic features of colorectal adenomas.¹³ Although the relationship between calcium intake and a lower incidence of colonic adenomas and carcinomas has not been uniformly demonstrated, observational studies overall suggest a protective effect. A pooling project of 10 cohort studies strongly suggested that calcium intake is inversely related to the risk of CRC.¹⁴

Further support for the beneficial effect of calcium in preventing large bowel cancer comes from numerous animal studies. Abnormal proliferation occurs in neoplastic and preneoplastic lesions in the colon. The increase in colonocyte proliferation stimulated by intrarectal instillation of deoxycholate and free fatty acids or by dietary supplementation with cholic acid may be ameliorated by oral calcium supplementation in laboratory animals. Studies in rodents fed high-fat diets also demonstrate a reduction in the number of carcinogen-induced tumors in animals receiving supplemental calcium in their diet. This reduction may be true especially for tumors containing K-ras mutations, a finding also suggested by a study in humans. Ornithine decarboxylase, an enzyme involved in polyamine biosynthesis and elevated in preneoplastic states, is reduced in rat colonic mucosa incubated with calcium in vitro, and supplemental calcium suppresses elevated levels of this enzyme in the mucosa of elderly patients with adenomatous polyps. It has been suggested that dietary calcium binds to ionized fatty acids and bile acids in the intestine, converting them to insoluble calcium compounds incapable of stimulating epithelial proliferation. Calcium increases fecal excretion of both phosphate and bile acids and modifies relative amounts of bile acids in bile. In addition, calcium in milk products is capable of precipitating luminal cytotoxic surfactants, thereby inhibiting their effects on colonic mucosa. These potential beneficial effects of calcium have not been observed uniformly, however, and studies of the effects of calcium on the rectal mucosa have not always demonstrated a reduction in the proliferation rates. In other studies, calcium supplementation normalized the distribution of proliferating cells in the colonic crypt without affecting the rate of proliferation in the colorectal mucosa.

Vitamin D₃ metabolites and analogs have been shown to play an important role in the regulation of a number of important cellular processes, including proliferation, differentiation, and apoptosis, in addition to their established role in mineral homeostasis. These steroid compounds have rapid effects that do not involve gene transcription or protein synthesis, as well as genomic effects involving the vitamin D receptor and other transcription factors.^15,16 Vitamin D can modulate more than 200 genes involved in cell cycle regulation, growth factor signaling, protection against oxidative stress, bile acid and xenobiotic metabolism, cell adhesion, DNA repair, angiogenesis, inflammation, and immune function. Effects of vitamin D and its metabolites have been demonstrated in normal and malignant colonocytes, and several potential mechanisms have been suggested by which these compounds might prevent carcinogenesis in the colon. Dietary supplementation with calcium and vitamin D in rodents fed colon tumor-inducing Western diets significantly reduced tumor incidence and multiplicity in addition to altering expression of a variety of genes linked to initiating formation of colon tumors.¹⁷

ARACHIDONIC ACID, EICOSANOIDS, AND CYCLOOXYGENASE-2

Clinical case-control and cohort studies have shown a 40% to 50% reduction in CRC-related mortality in persons taking aspirin and other NSAIDs on a regular basis compared with those not taking these agents. The exact mechanism for cancer protection with these agents is unknown, but it might relate to altered synthesis of arachidonic acid metabolites (eicosanoids) including prostaglandins, thromboxanes, leukotrienes, and hydroxy-eicosatetraenoic acids. These compounds modulate a number of signal transduction pathways that affect cellular adhesion, growth, and differentiation. COX (or prostaglandin-endoperoxide synthase) oxidizes arachidonic acid to prostaglandin G₂, reduces prostaglandin G₂ to prostaglandin H₂, and is the key enzyme responsible for production of prostaglandins and other eicosanoids.

This enzyme exists in two isoforms: COX-1 and COX-2. COX-1, the constitutive form of the enzyme, is present in most tissues and is involved in the physiologic production of prostaglandins to maintain normal homeostasis. COX-2 is induced by cytokines, mitogens, and growth factors, and its level has been shown to be elevated in both murine and human CRCs.^18–22 Expression of COX-2 is increased markedly in 85% to 95% of CRCs and in experimental models of CRC. COX-2 inhibition prevents cancer from developing during the initiation and the promotion and progression stages of carcinogenesis.²¹ Knockout of COX-2 results in suppression of intestinal polyposis in animal models of familial adenomatous polyposis (FAP).¹⁹ 15-Hydroxyprostaglandin dehydrogenase (15-PGDH) is a prostaglandin-degrading enzyme that is lost in human colon cancers and has been shown to be a physiologic antagonist of the prostaglandin-synthesizing activity of COX-2.²² It has been speculated that NSAIDs might reduce formation of colon tumors by inhibiting prostaglandin-mediated proliferation, but other evidence suggests that part of their effect might result from inducing apoptosis. Overexpression of COX-2 has been demonstrated to decrease apoptosis, whereas inhibition of COX-2 leads to an increase in apoptosis.

One potential mechanism by which NSAIDs can induce apoptosis is through elevating the prostaglandin precursor arachidonic acid. Increases in arachidonic acid after NSAID inhibition of COX stimulate conversion of sphingomyelin to ceramide, which is a mediator of apoptosis. NSAIDs also might inhibit the activation of genes by the nuclear hormone receptor peroxisome-proliferator-activated receptor δ (PPARδ) by disrupting the ability of this receptor to bind DNA. PPARδ expression is elevated in CRCs and repressed by the APC gene product, which is altered in CRC cells. The inhibition of PPARδ function enhances the ability of NSAIDs to induce apoptosis in colon cancer cells. PPARδ activates a variety of genes, including those involved in cellular growth and differentiation after exposure to a variety of ligands, such as eicosanoids. COX-2 inhibition could prevent production of these ligands and thereby prevent activation of PPARδ.

Other potential mechanisms by which COX-2 inhibition might affect tumor formation include alterations of cell adhesion to extracellular matrix proteins, inhibition of tumor neovascularization (angiogenesis), and reduction in carcinogen activation. A study using the Apc^Δ716 mouse, an animal model of FAP, demonstrated that treatment with the COX-2 specific inhibitor rofecoxib (Vioxx) was associated with a significant dose-dependent reduction in size and number of polyps as well as alterations in polyp morphology. COX-2 inhibition was associated with decreased levels of vascular endothelial growth factor (VEGF) and with lower rates of DNA replication.²⁰

In summary, environment and diet might affect the genesis of CRC, but their exact roles remain unclear. Their complex nature renders definition of the influence of individual environmental and dietary components difficult.

CHEMOPREVENTION

Chemoprevention refers to the use of natural or synthetic agents to reverse, suppress, or prevent progression or recurrence of cancer^23,24; this is a cornerstone of primary prevention. Data on chemoprevention of CRC come from studies in laboratory animals (see earlier), observational epidemiologic studies, case-control studies, and randomized controlled trials (RCTs). Because the natural history of CRC is protracted, clinical RCTs often have concentrated on preventing colorectal adenomas, which represent a form of intraepithelial neoplasia and are the precursors to carcinoma. The duration of the studies required, sample sizes necessary, cost, and ethical considerations make the use of cancer as an end point impractical. This difficulty has led to an increasing use of surrogate biomarkers to study chemoprevention of CRCs,²⁵ with the hope that use of such markers will lead to shorter, smaller, and less expensive trials.²⁶ To be valid, however, such biomarkers need to accurately represent the events involved in the process of carcinogenesis. When an intervention such as a chemopreventive agent is tested, there should be a clear relationship among the agent, modulation of the biomarker, and the development of cancer. Surrogate end points for cancer ideally should be validated in the context of clinical studies that use cancer as the ultimate end point. This is a difficult task because these are the very trials that such markers are designed to complement or replace.

There has been interest in the use of magnifying endoscopy to study aberrant crypt foci of the colon as possible markers in chemoprevention trials.²⁴ These foci consist of large, thick crypts that can be detected by chromoendoscopy using agents such as methylene blue or indigo carmine (Fig. 123-4). Aberrant crypt foci, particularly large crypts with dysplastic features, are thought to be precursors of adenomas in the colon. Standardization of techniques to identify and quantify these lesions is crucial to the successful interpretation of data from these trials, because it is unclear whether these lesions, which appear to be precursors to neoplasia in animal models, play a similar role in the human colon.²⁷ Studies suggest that the majority of aberrant crypt foci in the human colon may be hyperplastic rather than dysplastic (see Chapter 122).

Figure 123-4. A-C, Three magnification views of aberrant crypt foci at different magnifications. Aberrant crypt foci consist of large thick crypt, and are thought to be precursors of adenomas in the colon.

The potential benefit of low-fat, high-fiber diets based on descriptive epidemiology and case-control studies already has been discussed, but current data from prospective human trials are thus far equivocal or negative. Two large RCTs examined the effects of fiber supplementation on recurrence of adenomas. The Polyp Prevention Trial¹¹ randomized 2079 subjects with a history of colorectal adenomas to receive counseling together with a low-fat, high-fiber diet rich in fruits and vegetables or to receive their usual diet alone. The incidence of recurrent adenomas at one and four years, as determined by colonoscopy, was similar in both groups. In a study conducted by the Phoenix Colon Cancer Prevention Physicians’ Network,¹⁰ 1429 patients with a history of colorectal adenoma were randomized to receive 2.0 g or 13.5 g of supplemental wheat bran per day. Colonoscopy failed to show a difference in the incidence of recurrent adenomas at a median follow-up of 34 to 36 months.

A large body of observational and laboratory studies suggests a role for dietary calcium supplementation in chemoprevention. A prospective double-blind placebo-controlled trial showed that supplemental calcium (3000 mg of calcium carbonate per day, equivalent to 1200 mg of elemental calcium) reduced the incidence and number of recurrent adenomas in subjects chosen for a recent history of such lesions.²⁶ The effect of calcium was modest: 19% reduction in recurrence of adenomas and 24% reduction in the number of adenomas over three years, independent of age, sex, or dietary intake of calcium, fat, or fiber. The protective effect of calcium supplementation on the risk of colorectal adenoma recurrence extended up to five years after cessation of active treatment, even in the absence of continued supplementation.²⁸ Analysis of subjects’ serum vitamin D status suggested that calcium supplementation and vitamin D status appear to act together to reduce the risk of adenoma recurrence.²⁹ The results of a Japanese study, the Fukuoka Colorectal Cancer Study,³⁰ support the joint action of calcium and vitamin D in preventing CRC. Human trials using antioxidant vitamins A, C, and E have provided equivocal results, and current data do not support their routine use for colon cancer prevention in average-risk persons.

Folic acid and its metabolites play an important role in DNA synthesis, strand integrity, and methylation. Epidemiologic studies have found a lower incidence of CRC among those with high compared with low dietary intake of folate.³¹ This protective effect also was suggested by the Nurses’ Health Study, in which high doses of folate (as part of multivitamin supplementation) given over several years were protective against CRC. A large prospective RCT^32,33 failed, however, to demonstrate a protective effect of 1 mg/day of folate supplementation on recurrence of adenoma compared with placebo and suggested that folate supplementation in persons with prior adenomas actually might increase adenoma risk. Analysis of baseline dietary and serum folate levels in these subjects supports the idea that although moderate doses of folate may be protective compared with deficiency, at some point of sufficiency, supplementation provides no additional benefit.³⁴ The lack of efficacy of folate (0.5 mg/day) supplementation to prevent adenoma recurrence also was found in another RCT.³⁵

Epidemiologic, case-control, and prospective cohort trials suggest a protective effect against the development of CRC in women taking hormone (estrogen) replacement therapy. It has been postulated that estrogen might protect against colon cancer by decreasing production of secondary bile acids, by decreasing levels of insulin-like growth factor-1, or through as yet undetermined direct effects on colonic mucosal epithelial cells. Data from the Women’s Health Initiative demonstrated that short-term use of estrogen plus progestin was associated with a decreased risk of CRC; however, CRCs in women who took estrogen plus progestin were diagnosed at a more advanced stage than those in women who took placebo.³⁶

Cigarette smoking has been associated with incidence of and mortality from CRC in observational studies,³⁷ but the long-term effects of smoking cessation on CRC have not been studied.

The most promising results for CRC prevention come from trials using aspirin and NSAIDs. Case-control and cohort studies have suggested that the risk for development of adenoma and carcinoma may be reduced substantially (40% to 50%) among aspirin and NSAID users, compared with controls.^6,24,38 A prospective cohort study among male health professionals demonstrated that persons who take aspirin more than twice per week were at lower risk for CRC (relative risk [RR], 0.68) than were controls. An RCT that assessed the effect of low-dose aspirin in an average-risk population demonstrated no significant reduction in the number of CRC cases during the first six years of follow-up; longer follow-up may be necessary to demonstrate a significant effect of aspirin on development of cancer, because the Nurses’ Health Study demonstrated that the benefits of aspirin might not be evident until after at least a decade of regular aspirin consumption.⁶ Three prospective adenoma prevention trials (see later) now provide compelling evidence that aspirin use reduces the risk of developing colorectal adenoma in persons with a personal history of adenoma or carcinoma.^39–41

Given the long natural history of CRC, preventing recurrence of adenoma after endoscopic removal often is used as an intermediate or surrogate end point in chemoprevention trials.²⁴ In FAP, in which hundreds of adenomas occur in the colon and rectum, chemoprevention trials often use reductions of the number and size of adenomas as end points in short-term studies. Such trials have suggested a potential role for NSAIDs as chemopreventive agents in this setting. There is a significant decrease in the mean number and size of polyps in patients treated with the NSAID sulindac. In a small, double-blind RCT of 22 patients with FAP, treatment with sulindac reduced the number and mean diameter of colorectal polyps by 44% and 35% of respective baseline values after nine months; three months after treatment was stopped, however, the number and size of polyps had increased. A subsequent prospective cohort study⁴² confirmed that long-term use of sulindac is effective in reducing the number of adenomas in patients with FAP. A 76% reduction in the number of polyps was seen at one year and was sustained (74% reduction) through 63 months of follow-up. Sulindac might not be effective, however, for primary prevention of development of adenomas in genetically disposed persons with FAP who have not yet developed macroscopic adenomas.⁴³

A double-blind RCT studied the effects of celecoxib (Celebrex), a selective COX-2 inhibitor, on colorectal polyps in patients with FAP.⁴⁴ Treatment with high doses of celecoxib for six months was associated with a significant reduction (28%) in the number of colorectal polyps compared with placebo (4.5%). This drug is now approved in the United States as an adjunct to standard therapy in patients with FAP.

Data are now available on the role of nonselective NSAIDs and COX-2 inhibitors as chemopreventive agents for preventing sporadic adenomas. Four published trials have demonstrated a reduction in recurrence of adenoma in chemoprevention trials involving aspirin. An RCT of aspirin versus placebo⁴⁰ demonstrated that daily use of 81 mg of aspirin was associated with a 19% reduction in occurrence of adenoma and a 41% reduction in occurrence of advanced neoplasms (adenomas measuring at least 1 cm or with tubulovillous or villous features, severe dysplasia, or invasive cancer) compared with results of placebo treatment three years after removal of an index sporadic adenoma. In a similar trial, aspirin (300 mg/day) was associated with a 21% overall risk reduction for all adenomas and a 37% reduction in advanced adenomas at three years.³⁵ A prospective RCT in a higher-risk group for adenoma recurrence, those with a previous history of sporadic CRC, demonstrated a 45% risk reduction in those taking 325 mg of aspirin daily compared with those taking placebo with a mean follow-up of almost three years.³⁹

Given biological plausibility, preclinical in vitro and animal data, and data on regression of adenoma in patients with FAP, three randomized trials were undertaken to examine the effect of COX-2 selective inhibitors on formation of new adenomas in patents with a history of sporadic adenomas. The Adenoma Prevention with Celecoxib (APC) trial⁴⁵ randomized 2035 subjects to celecoxib at a dose of 200 mg or 400 mg twice daily. Use of celecoxib was associated with a dose-dependent 33% to 45% reduction in the development of new adenomas by three years, with a 57% to 66% reduction in the number of patients developing advanced adenomas. The Prevention of Sporadic Adenomatous Polyps (PreSAP) trial⁴⁶ randomized 1561 patients at a ratio of two to three to receive 400 mg celecoxib once daily or placebo. Use of celecoxib was associated with a 36% overall reduction in formation of new adenomas and a 51% reduction in advanced adenomas over three years. The Adenomatous Polyp Prevention on Vioxx (APPROVe) Trial^47,48 was a double-blind RCT of the efficacy of oral rofecoxib, 25 mg daily, to prevent colorectal adenomas in 2587 subjects. The risk of adenoma recurrence over three years was lower for subjects on rofecoxib than on placebo, with a 24% overall reduction and a 30% reduction in advanced adenomas. The effect of drug was more pronounced in the first year (RR 0.65; 95% confidence interval [CI]: 0.57-0.73) than in the subsequent two years.

Thus, three well-conducted prospective RCTs demonstrated a significant reduction in formation of new adenomas with the use of a COX-2 selective inhibitor in those with a history of colorectal adenomas. Unfortunately, adverse thrombotic cardiovascular events were associated with COX-2 inhibition in at least two of these trials. Data suggest that an increased risk of cardiovascular events may be associated with most NSAIDs, and not just COX-2 inhibitors.

Other agents currently undergoing study for chemoprevention of colorectal neoplasia include the ornithine decarboxylase inhibitor difluoromethylornithine (DFMO); the bile acid ursodiol; 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors, such as pravastatin and lovastatin; epidermal growth factor receptor (EGFR) inhibitors; and matrix metalloproteinase (MMP) inhibitors.²⁴ The combination of DFMO and the NSAID sulindac were studied in an RCT to assess their efficacy in preventing sporadic adenoma recurrence in 375 subjects. Use of this regimen was associated with a 70% reduction in new adenomas at three years compared with placebo.⁴⁹ Larger studies are needed to confirm this result and to fully assess the toxicity of this combination. A population-based case-control study of persons with CRC and matched controls demonstrated a 47% relative reduction of CRC associated with statin use, but further investigation is needed to assess this benefit.⁵⁰ Table 123-2 and Figure 123-5 summarize the status of current studies that examine the effect of chemopreventive agents on colorectal neoplasia.

Figure 123-5. Diagram depicting the relative risk of colorectal adenoma recurrence after a clearing colonoscopy in different colorectal adenoma prevention studies using various potential chemopreventive agents. DFMO, difluoromethylornithine; vit., vitamin.

(Courtesy of Asad Umar, DVM, PhD, Bethesda, Md.)

ABNORMAL CELLULAR PROLIFERATION

Abnormal cellular proliferation is a hallmark of neoplasia (see Chapter 3). Actively proliferating cells are more susceptible to initiators of carcinogenesis (primary carcinogens) and genetic alterations than are resting cells. In the normal colon, DNA synthesis occurs and cells divide and proliferate only in the lower and middle regions of the crypts. As cells normally migrate upward from deeper in the crypt, the number of cells that continue to proliferate decreases, and, on reaching the upper crypt region, cells become terminally differentiated and can no longer divide. A substantial body of literature indicates that this sequence of events is disordered during the evolution of neoplastic lesions in the colon. Increased proliferative activity and characteristic differences in the distribution of tritiated thymidine-labeled cells (i.e., those that actively synthesize DNA) within the colonic crypts have been demonstrated and distinguish at-risk and affected members of kindreds with familial polyposis, as well as nonpolyposis-inherited colon cancer from lower-risk groups. Correlations between rectal mucosal proliferation and the clinical and pathologic features of nonfamilial colorectal neoplasia also have been demonstrated. Conversely, populations at low risk for developing colon cancer, such as Seventh Day Adventist vegetarians, have relatively quiescent proliferative activity in their colonic mucosa.

Disordered proliferative activity can be found in the colonic mucosa of rodents treated with a variety of chemical carcinogens, and increased proliferative activity is seen in animals whose colonic or rectal mucosa is exposed to tumor promoters such as secondary bile acids. Colonic epithelial cells also fail to repress DNA synthesis during epithelial renewal in ulcerative colitis (UC), a condition associated with an increased risk of CRC. Ornithine decarboxylase, an enzyme marker of rapid cellular proliferation, is present at high levels in the mucosa of members of familial polyposis kindreds, and levels increase in the colonic mucosa during chemically induced colonic carcinogenesis in rats. Ornithine decarboxylase increases in the colonic mucosa with age and is elevated in elderly patients with colonic adenomas. Experimental evidence also suggests that PKC may be involved in the stimulation of colonic epithelial cell proliferation by tumor promoters. These findings have led to speculation that inhibitors of cellular proliferation might prove useful as anticancer drugs. NSAIDs and COX-2 inhibitors, for example, decrease proliferation and increase apoptosis.

An explosion of knowledge in the field of molecular genetics has demonstrated how alterations in proto-oncogenes and tumor suppressor genes might lead to disruption of mechanisms that regulate the normal cell cycle and cell proliferation. In some cases, the cell is predisposed to abnormal proliferation by germline mutations, such as FAP, whereas in other cases, somatic mutations occur as the result of complex interactions with environmental factors,⁵¹ as detailed earlier.

MOLECULAR BIOLOGY AND BIOCHEMICAL CHANGES

Molecular Genetics

Tumor cells in the colon, as elsewhere, are characterized by heritable phenotypic changes that are the result of quantitative or qualitative alterations in gene expression (see Chapters 3 and 122). A large body of evidence demonstrates that CRCs are associated with an accumulation of such genetic alterations (Table 123-3; Fig. 123-6). Genetic changes that might lead to the development of CRC traditionally have been categorized into three major classes: alterations in proto-oncogenes, loss of tumor suppressor gene activity, and abnormalities in genes involved in DNA mismatch repair (MMR).^52–54 Adenomas and carcinomas arise in the context of genomic instability, by which epithelial cells acquire the number of mutations needed to attain a neoplastic state. Destabilization of the genome is a prerequisite to tumor formation and most commonly involves chromosomal instability (CIN), which is found in 80% to 85% of colorectal tumors with subsequent allelic loss; chromosomal amplifications and translocations; or increased rates of intragenic mutation in tandemly repeated DNA sequences known as microsatellites (microsatellite instability [MSI]).⁵⁴

Table 123-3 Genes Altered in Sporadic Colorectal Cancer

Figure 123-6. Proposed sequence of molecular genetic events in the evolution of colon cancer. Carcinomas arise from an accumulation of events, the sequence of which has been defined. Alterations in APC or DNA mismatch repair genes may be inherited in the germline (familial adenomatous polyposis, hereditary nonpolyposis colorectal cancer) or may be acquired after birth (somatic mutations). Upper part, The top row (A-C) shows colonoscopic photographs and the bottom row (D-F) shows the histology. From left to right: Dysplastic aberrant crypt focus (A, with methylene blue staining; D, E, and E with hematoxylin and eosin staining). adenomatous polyp (B and E); and invasive carcinoma (C and F).

(Aberrant crypt focus [A and D] reproduced from Takayama T. Katsucki S, Takahashi Y, et al. Aberrant crypt foci of the colon as precursors of adenoma and cancer. N Engl J Med 1998; 339:1277.)

Cellular proto-oncogenes are evolutionarily conserved human genes that contain DNA sequences homologous to those of acute transforming retroviruses. Many of these genes play a role in signal transduction and the normal regulation of cell growth. Inappropriate activation of these genes leads to abnormal transmission of regulatory messages from the cell surface to the nucleus that results in abnormal proliferation and, eventually, tumor formation. Three human ras genes—K-ras, N-ras, and H-ras—encode guanine nucleotide-binding proteins that regulate intracellular signaling pathways. Approximately 65% of sporadic CRCs have activating point mutations in a ras gene, most in K–ras. Most ras mutations appear to occur during intermediate stages of adenoma growth (see Chapter 122). Ras gene mutations occur in 47% of carcinomas, in 58% of adenomas larger than 1 cm, but in only 10% of adenomas smaller than 1 cm, suggesting that earlier events must contribute to formation of neoplasms. Alterations in signal transduction might lead to abnormal cell growth and thus participate in neoplastic transformation, but activation of ras alone is not sufficient for progression to carcinoma. The exact functional relationship between ras mutations and carcinogenesis remains to be established, but understanding the role of ras mutations in stimulating proliferation might lead to development of anti-tumor therapies aimed at interrupting signals that alter tumor cell growth.

Chromosomal abnormalities have been reported in CRCs for more than a decade, and abundant evidence has shown that allelic losses, particularly at chromosome locations 5q, 17p, and 18q, play major roles in the genesis of large bowel tumors.^52–55 A deletion within chromosome 5 in patients with FAP led to the identification of the APC gene on the long arm of this chromosome (5q21). Positional cloning identified a single tumor suppressor gene, which is mutated in both the germline of FAP patients and in sporadic colorectal tumors. The protein encoded by APC consists of 2843 amino acids and is located at the basolateral membrane of colorectal epithelial cells; expression is increasingly pronounced as cells migrate up through the colonic crypt.

Somatic mutations of the APC gene occur in 60% to 80% of sporadic CRCs and adenomas, including the smallest dysplastic lesions. These mutations result in truncation of the APC protein in more than 98% of cases, a finding that has led to the development of clinically useful tests for genetic screening of FAP families. Inactivation of both copies of the APC gene appears to be the gatekeeping event for the initiation of colorectal neoplasia. The APC gene product interacts with at least six other proteins, including glycogen synthetase 3β (GSK-3β) and axin in the cytoplasm. Inactivation of this gene is required for net cellular proliferation and initiation of neoplasia in the colon.

APC functions to modulate extracellular signals that are transmitted to the nucleus through the cytoskeletal protein β-catenin, as part of the Wnt signaling pathway (Fig. 123-7). Nuclear β-catenin binds to transcription factors in the nucleus that are members of the lymphoid enhancer factor/T-cell factor (LEF/TCF) family including Tcf-4, which in turn activate various target genes (e.g., c-myc, cyclin D₁)⁵⁶ that affect cell cycling and growth. APC is a tumor suppressor gene that binds to β-catenin and causes its degradation through phosphorylation. Loss of APC function, therefore, leads to accumulation of β-catenin and unopposed stimulation through the Wnt-Tcf signaling pathway, which in turn leads to increased and unregulated proliferation and decreased programmed cell death (apoptosis). APC gene abnormalities also might lead to disruption of normal cell-cell adhesion through altered association with the cellular adhesion molecule E-cadherin. Disruption of APC-mediated regulation of transcriptional activation is critical for colorectal tumorigenesis and is achieved most commonly through inactivating mutations of both APC alleles; disruption also can occur through dominant mutations of the β-catenin gene that render β-catenin-Tcf-regulated transcription insensitive to the regulatory effects of normal wild-type APC.

Figure 123-7. A model of Wnt signaling in normal (A and B) and cancer (C) cells. A, In the absence of Wnt signaling, APC, axin, and GSK3-β form a complex that results in β-catenin phosphorylation and degradation by a ubiquitin-dependent mechanism. B, Binding of Wnt to its cell surface receptor results in stabilization of β-catenin. Unphosphorylated β-catenin is able to translocate to the nucleus to form a complex with members of the LEF/TCF (lymphoid enhancer factor/T-cell factor) family and activates Wnt target genes. Frizzled and Dishevelled refer to gene products that participate in this pathway. C, Loss or mutation of APC results in lack of β-catenin degradation and high levels of this protein in the cytoplasm and nucleus. Strong evidence exists that misregulation of Wnt target gene expression is crucial to transformation in colon cells. GBP, glycogen synthase kinase-3 binding protein; LRP, lipoprotein receptor-related protein.

(From Waterman ML. Lymphoid enhancer factor/T cell factor expression in colorectal cancer. Cancer Metast Rev 2004; 23:41.)

Other genetic changes occur later in the adenoma-to-carcinoma sequence. Stepwise tumor progression is associated in more than 75% of cases with loss of the tumor suppressor gene activity located on chromosome 18q. Several candidate genes are present on this chromosome, and loss of chromosome 18 is associated with a poor prognosis.⁵⁷ One gene, DCC (deleted in CRC), originally was thought to be important because its loss from a stage II (Dukes B) cancer was associated with a worse prognosis in some studies; its role as an important tumor suppressor gene has been questioned.

DPC4 (SMAD4) is another candidate tumor suppressor gene, inactivation of which might play a role in development of CRC. DPC4 belongs to the SMAD gene family involved in signal transduction pathways activated through the transforming growth factor (TGF)-β family receptors. Experimental inactivation of the homologue Dpc4 in a mouse model of adenomatous polyposis coli results in malignant progression of intestinal and colonic polyps initiated by loss of the Apc gene (the mouse homologue of APC). Mutations in SMAD4 and a related gene, SMAD2, have been reported in some sporadic CRCs. Deletions of chromosome 17p involve the p53 tumor suppressor gene, the product of which normally prevents cells with damaged DNA from progressing from the G₁ to the S phase in the cell cycle. Deletions within chromosome 17p are present in approximately 75% of CRCs.

Loss of TP53 also may be associated with reduced apoptosis of damaged cells. Inactivation of the TP53 gene mediates the conversion from adenoma to carcinoma, a late and important event in colon carcinogenesis. Distant metastases from CRCs are associated with high fractional allelic loss and deletions of 17p and 18q. A distinct set of metastasis-suppressor genes also has been postulated.⁵⁸

Genomic instability creates a permissive state in which a cell acquires sufficient mutations to be transformed to a cancer cell; this is a common mechanism central to the development of most, if not all, colon cancers.⁵³ Several forms of genomic instability are common in colon cancer, including chromosome instability (CIN) and chromosome translocations, and MSI in which subtle sequence changes, including base substitutions, deletions, or insertions, lead to a hypermutable state (Fig. 123-8). Candidates responsible for CIN include genes responsible for the human mitotic spindle checkpoint (hBUB1 and hBUBR1), genes involved in the DNA damage checkpoint (ATM, BRCA1 and BRCA2, TP53, and hRad17), and genes that control centrosome number. Inactivation of genes involved in base excision repair (BER) resulting from oxidative damage are found in a subset of CRCs. Inactivation of one of the BER genes (MYH) is a cause of an autosomal recessive form of FAP. Although tumors arising as the result of MYH germline mutations demonstrate chromosomal instability, they appear to have a unique pathogenesis compared with sporadic CRC.

Figure 123-8. Model of colon cancer formation in tumors that progress through the adenoma-carcinoma sequence along pathways marked by chromosomal instability (CIN) or microsatellite instability. ACF, aberrant crypt focus; MMR, mismatch repair; TGF, transforming growth factor.

(Modified from Grady WM. Genomic instability and colon cancer. Cancer Metast Rev 2004; 23:11.)

The significance of genomic instability in the pathogenesis of a subset of colon cancers became evident with the discovery of MSI in colon cancers associated with hereditary nonpolyposis colorectal cancer (HNPCC). Alterations in genes that help maintain DNA fidelity during replication are characteristic of patients with HNPCC.^53,59 Alterations in MMR genes designated hMLH1, hPMS1, hPMS2, hMSH2, hMSH3, and hMSH6 might lead to the inability to repair base pair mismatches and result in DNA replication errors or MSI. Inactivation of the MMR system causes genomic instability by increasing the rate of polymerase-generated replication errors and degrading the fidelity of DNA replication, particularly at microsatellite repeat sequences.⁵³ MSI involves mutations or instability in short, tandemly repeated DNA sequences such as (A)ⁿ, (CA)ⁿ, and (GATA)ⁿ. Such DNA sequences are found in several key genes that are important for maintaining normal cellular function (Table 123-4). The receptor for TGF-β (TGF-βRII), for example, often is mutated as the result of MSI.

Table 123-4 Frequency of Some Target Gene Mutations in Colon Cancers with Microsatellite Instability (MSI)

TARGET GENE	FREQUENCY OF MUTATION IN COLON CANCERS WITH MSI (%)	NORMAL FUNCTION OF GENE PRODUCT
TGF-β R2	90	TGF-β signaling
ACVR2	86	Activin signaling
IGFIIR	10	Insulin-like growth factor and TGF-β signaling
BAX	50	Apoptosis
hMSH 3	50	DNA mismatch repair
hMSH 6	33	DNA mismatch repair
E2F-4	65	Cell cycle control
PTEN	19-34	Growth regulation
MBD4 (MED1)	40	DNA repair and binding to methylated DNA
TCF4	39	Growth regulation
CHK1	10	G2 cell cycle checkpoint
STK11 (LKB1)	<2	Signal transduction
BLM	<18	Chromosome stability, DNA repair; helicase
Caspase 5 (ICErel-III)	62	Apoptosis
CDX2	<2	Homeobox protein
TPB	83	TATA binding protein
RIZ	26	Interacts with RB
hRAD50	31	DNA repair
SEC63	49	ER chaperone protein
AIM2	48	Interferon-inducible protein

Note: Most mutations cause frame shifts that prematurely truncate the protein, leading to inactivation of the affected allele.

ER, endoplasmic reticulum; RB, retinoblastoma protein; TGF, transforming growth factor.

Modified from Grady WM, Carethers J. Genomic and epigenetic instability in colorectal cancer progression. Gastroenterology 2008; 135:1079-99.

Multiple lines of evidence suggest that the TGF-β pathway is an important tumor-suppressing pathway in the colon and that alterations in this pathway lead to tumor development. Less frequently targeted genes include the insulin growth factor 2 receptor; Bax and caspase 5, proteins that regulate apoptosis; E2F4, a transcription factor; and MSH3 and MSH6, DNA MMR proteins. β-Catenin mutations are present in up to 25% of MSI colon cancers. MSI, therefore, leads to accumulation of mutations in vulnerable genes, eventually resulting in the acquisition of the malignant phenotype. Although a high frequency of MSI (instability at more than 40% of microsatellite loci) is characteristic of HNPCC, similar alterations can be found in about 15% of sporadic CRCs and also in premalignant lesions. MSI tumors remain diploid. Patients whose tumors demonstrate MSI might have a better prognosis⁶⁰ and respond differently to chemotherapy,^54,61,62 than those whose tumors are characterized by chromosomal instability. Most patients with MSI colon cancers do not possess mutations in the known MMR genes, and evidence indicates that MSI in these tumors might arise through epigenetic mechanisms. Epigenetics is the study of clonal changes in gene expression without accompanying changes in DNA coding sequences.

Epigenetic silencing is recognized now as an important mechanism in the evolution of a subgroup of CRCs. DNA methylation within promoters and alterations in histone modifications appear to be primary mediators of epigenetic inheritance in cancer cells, and hypermethylation of the hMLH1 promoter has been reported in up to 70% of sporadic MSI tumors. In the large intestine, aberrant methylation may be an important early event in the age-related field defect observed in sporadic colorectal neoplasia. Aberrant methylation also contributes to tumor progression through a hypermethylator phenotype (CPG island methylator phenotype [CIMP]) responsible for most cases of MSI related to hMLH1 inactivation (associated with about 15% of sporadic CRCs).^54,63 The hallmark of CIMP is abnormal methylation of several tumor promoters. In one model, hyperplastic aberrant crypt foci may be the initial lesions in a pathway leading to the development of serrated adenomas (see Chapter 122). Methylated promoters of the MGMT, EVL, HLTF, SFRP2, SLC5A8, and MINT1 genes develop during tumor initiation, and hMLH1 promoter hypermethylation corresponds to the development of a serrated adenoma, with methylated TSP1 and TIMP3 helping to drive tumor progression.⁵⁴ DNA methylation and histone H3 lysine 9 hypoacetylation and methylation appear to form a mutually reinforcing loop that contributes to tumor suppressor gene inactivation in CRCs.

The BRAF gene encoding a downstream component of the RAS/RAF/MAPK pathway is often mutated in sporadic MSI tumors but not in tumors from patients with HNPCC (see later).

MicroRNAs are 18- to 25-nucleotide noncoding RNA molecules that regulate translation of many genes. Expression patterns of microRNAs are altered in colon cancers, and they may be associated with survival and therapeutic outcome.⁶⁴

Biochemical and Other Changes

Chapter 3 deals in depth with the biological and biochemical changes that occur during the development of colorectal neoplasia. Alterations in cell surface and secreted proteins and glycoproteins, including a number of important cell adhesion molecules, are characteristic of CRC.⁶⁵ Interactions between tumor cells or between tumor cells and their environment may be homotypic (involving like molecules) or heterotypic (involving different adhesion molecules). Homotypic interactions often maintain the integrity of primary tumors by fostering adhesion between neighboring tumor cells, whereas heterotypic interactions might occur among tumor cells and platelets, lymphocytes, vascular endothelial cells, and components of the basement membrane matrix. Most tumor-associated molecules represent quantitatively or qualitatively altered molecules found either on normal tissues or during development, such as oncofetal antigens—for example, carcinoembryonic antigen [CEA]). Many of these molecules appear to play a role in maintaining normal tissue homeostasis or targeting blood-borne cells to specific sites. Altered expression, therefore, might contribute to tumor invasion and metastasis.

MMPs (matrix metalloproteinases) are a family of enzymes that degrade extracellular matrix. Overexpression of MMP-1, -2, -3, -7, -9, and -13 and MT1-MMP has been demonstrated in human CRCs. The degree of overexpression of some MMPs correlates with stage of disease, prognosis, or both.⁶⁶

Metastasis is a multistage process by which tumor cells escape the primary tumor and establish secondary foci at distant sites (Fig. 123-9). Cells in the primary tumor must become vascularized (angiogenesis via vascular endothelial growth factors). Then they must escape the primary tumor by overcoming adhesive interactions (e.g., loss of E-cadherin) and by disrupting basement membranes (metalloproteinases such as type IV collagenase, matrilysin, loss of tissue inhibitors of collagenase). Finally they must enter lymphatics, the circulation, or both. In the bloodstream, they must survive interactions with blood components and the immune system and be transported to distant organ sites (principally the liver). At distant sites, tumor cells adhere to target endothelia via specific interactions (e.g., tumor-associated sialoglycoproteins and endothelial selectins) (Fig. 123-10), extravasate, interact with the microenvironment (e.g., growth factors), and establish secondary tumor foci.

Figure 123-9. Colon cancer metastasis. Cancer cells metastasize through a complex, multistage process. For tumor cells to form metastatic foci at distant sites, they must complete all stages of this process.

Figure 123-10. Colon cancer metastasis. On reaching the liver, colon cancer cells adhere to the sinusoidal endothelium through specific interactions and then invade the parenchyma. A, Photomicrograph showing tumor cells invading the liver after extravasation from a blood vessel. (Hematoxylin and eosin stain.) B, Electron micrograph showing collagen bundles (c) and tumor cells (T) adherent to sinusoidal endothelium (E) and invading between hepatocytes (H). rbc, red blood cell; S, sinusoid.

Tumor cell subpopulations with different metastatic potentials exist within the same primary tumor, and metastases result from the selective dissemination of tumor cells that possess the ability to participate in all stages of this complex process. Several carbohydrate antigens have been studied in relation to their potential usefulness as markers of metastatic potential and for their possible role in determining prognosis.^65–68

AGE

The risk of developing CRC rises sharply after age 40 years in the general population, 90% of cancers occurring in persons aged 50 years and older (Fig. 123-12). A 50-year-old person has approximately a 5% chance of developing CRC if he or she survives to age 80 years and a 2.5% risk of dying from the disease; this has important implications for screening (see later). Sporadic CRCs arise in other age groups, such as the third and fourth decades, however, and the diagnosis must be considered in younger persons with signs and symptoms characteristic of this disease, especially if they have a family history of colorectal neoplasia.

Figure 123-12. Frequency of colorectal cancer by age at diagnosis in the United States. Ninety percent of cancers occur after age 50 years.

PRIOR ADENOMA AND CARCINOMA

FAMILY HISTORY

The risk of CRC in first-degree relatives of those with sporadic CRC is increased two-fold to three-fold (see earlier). The risk is higher when adenoma or carcinoma has occurred in a relative at an early age or when more than one relative has had carcinoma. These factors have been taken into account in screening guidelines that stratify patients according to potential cancer risk.^72–83

FAP and its variants are inherited in an autosomal dominant manner, and without colectomy virtually all affected members with polyps eventually develop carcinoma. HNPCC also has an autosomal dominant pattern of inheritance (see Chapter 122).

Turcot’s syndrome is a rare combination of inherited adenomatous polyposis and malignant brain tumors. Inheritance is autosomal dominant, and germline mutations in the APC gene or mutations of hMLH1 and hPMS2 characteristic of HNPCC have been described (see Chapter 122).

Muir-Torre syndrome is a rare variant of HNPCC that manifests with multiple skin lesions, including sebaceous adenomas and carcinomas, basal cell and squamous cell carcinomas, and keratoacanthomas in addition to colonic adenomas and adenocarcinomas. MSI and loss of hMLH1 and hMSH2 expression has been noted in a high percentage of tumors occurring in this syndrome.

Peutz-Jeghers syndrome and the familial form of juvenile polyposis both have an increased risk of small and large bowel cancer (see Chapter 122). Peutz-Jeghers syndrome is an autosomal dominant disease, which, in most families, has been mapped to chromosome 19 p13.3 and the STK11 gene (serine threonine kinase 11). Adenomatous changes have been reported in 3% to 6% of hamartomas from these patients. Extracolonic cancers are common and have been reported to occur in 50% to 90% of patients with Peutz-Jeghers syndrome (relative risk for all cancers was 15.2).⁸⁴ A significant increase in a variety of cancers (esophagus, stomach, small intestine, pancreas, lung, breast, uterus, and ovary) has been reported.

Familial juvenile polyposis is a rare autosomal dominant disease associated with polyps that may be limited to the colon or the stomach or that occur throughout the gastrointestinal tract (see Chapter 122).⁸⁵ The genetic basis of this syndrome is not fully understood, but germline mutations in a gene (SMAD4) located on chromosome 18q21.1 that encodes an intracellular mediator in the TGF-β signaling pathway have been identified in some affected patients, as have mutations in the BMPR1A (bone morphogenetic protein receptor type 1A) gene; each accounts for approximately 20% of cases.⁸⁶ The bone morphogenetic receptors (BMPs) are a family of transmembrane serine-threonine kinases whose ligands are members of the TGF-β superfamily. BMPs repress WNT signaling, and BMPR1A plays a role in apoptosis. The PTEN (phosphatase and tensin homologue deleted on chromosome 10) gene located on chromosome 10 also has been linked to some cases of juvenile polyposis. PTEN encodes a dual phosphatase, loss of function of which leads to protection of cells from various apoptotic stimuli. Overexpression of PTEN leads to cell cycle arrest or induction of apoptosis, or both. In patients with juvenile polyposis, the cumulative lifetime risk for CRC might approach 40%.⁸⁷

INFLAMMATORY BOWEL DISEASE

Patients with idiopathic IBD (Crohn’s disease, UC) are at increased risk for developing adenocarcinoma of the colon (see Chapters 111 and 112).⁸⁸ Actuarially derived (life table) cumulative cancer incidences from tertiary referral centers (Fig. 123-13) agree that the risk of cancer in patients with UC begins after a disease duration of seven years and rises about 10% per decade, reaching approximately 30% at 25 years. Nonetheless, difficulties related to sources of referral, sampling, recognition and characterization of disease, differences in follow-up procedures, and methods used to detect neoplastic disease cloud many such studies. The assessed cancer risk for patients with UC followed in one private practice has been reported to be only 6.6% 26 years after disease onset and 11.4% at 32 years. The true risk of developing CRC in UC must lie between predictions from tertiary centers and the primary care setting.

Figure 123-13. Cumulative probability of developing colorectal carcinoma in patients with ulcerative colitis seen at two tertiary referral centers.

Data from primary care settings indicate a similar pattern but lower incidence rates (see text and Chapter 112).

The risk of CRC for patients with UC correlates most closely with duration of disease. In a large group of patients with extensive disease who were followed prospectively, the risk of carcinoma per patient-year was zero before 10 years and one in 86 after 20 years. The risk is greatest with universal colitis. It has been reported that the risk of cancer in left-sided disease (i.e., distal to the splenic flexure) begins approximately a decade later than with universal colitis, but at least one surveillance study found no difference between these groups in the temporal development of preneoplastic dysplasia. The risk for patients with ulcerative proctitis is only slightly increased compared with that in the general population.

The risk of cancer is not related to severity of the initial attack of colitis or age at onset of colitis, independent of disease duration. The relationship to disease activity per se is unclear, although chronic inflammation is a common theme in epithelial carcinogenesis. CRC in persons with UC is a risk factor for CRC in their relatives without colitis and CRC in relatives without UC is a risk factor for those with colitis. There is an association of backwash ileitis with CRC in patients with UC who undergo proctocolectomy.⁸⁹ Cancer arising in the setting of UC traditionally has been thought to be a highly malignant lesion with a poor prognosis, but studies using matched controls from colon cancer populations without colitis have failed to show a significant difference in survival between the two groups.

The increased risk for CRC in patients with Crohn’s disease or ileocolitis has been reported to be as much as 4 to 20 times that in the general population, although one cohort study failed to confirm an increased incidence of colon cancer in these patients. Cancer can arise at an earlier age in these patients than in the general population. Many of these cancers are mucinous carcinomas, and they often occur in surgically bypassed or strictured segments of colon.

Carcinomas do not develop de novo from normal mucosa but from mucosa that has undergone a sequence of morphologic changes that culminate in invasive carcinoma. As in precancerous adenomas, dysplasia is a precursor to carcinoma in IBD. Dysplasia comprises abnormalities in crypt architecture and cytologic detail (Figs. 123-14 and 123-15). Epithelial crypts are reduced in number, irregularly branched, and crowded together (“back-to-back” glands). Cell nuclei may be enlarged and hyperchromatic, may have increased numbers of mitoses, and may be located at different levels in the cell, producing a picket-fence appearance (pseudostratification). Dysplasia is classified by grade as mild (or low grade) to severe (or high grade).

Figure 123-14. Photomicrograph of dysplasia in the setting of ulcerative colitis. The number of goblet cells is decreased. Glands are branched, irregular, and crowded together. Cell nuclei are hyperchromatic and occur at different levels, producing a pseudopalisading or picket-fence appearance. (Hematoxylin and eosin stain.)

Figure 123-15. Plaque-like dysplasia–associated lesion or mass in a patient with long-standing ulcerative colitis. A, Lesion as seen through the colonoscope. B, Biopsy specimen revealing high-grade dysplasia. (Hematoxylin and eosin stain.)

Retrospective analyses report that 90% of resected colons from patients with UC and cancer contain dysplastic mucosa somewhere in the colon and that 30% of patients with severe rectal or colonic dysplasia on resection or biopsy have coexisting carcinoma. Colonoscopic studies suggest that 25% of colons that demonstrate high-grade dysplasia on biopsy harbor a carcinoma. Dysplasia often is patchy, and it may be present in the colon but absent from the rectum.

Because the lack of uniformity of the definition of dysplasia can make interpretation of such data difficult, the multidisciplinary Inflammatory Bowel Disease Dysplasia Morphology Study Group developed a standardized classification for dysplasia arising in the setting of IBD. Several large prospective studies have attempted to determine the true risk of cancer in patients with colonic dysplasia and UC and the impact of screening programs for dysplasia.

Results suggest that biopsy surveillance programs can be effective in helping control the risk of carcinoma in patients with long-standing UC. The risk of cancer appears highest in those with high-grade dysplasia that arises in visible plaques or masses (dysplasia-associated lesion or mass [DALM], as illustrated in Fig. 123-15). A computer cohort decision analysis suggested that surveillance should increase life expectancy for patients with UC. Most investigators believe that patients who have UC for more than seven to eight years should undergo colonoscopy with multiple mucosal biopsies annually to identify areas of dysplasia; they advocate colectomy for severe dysplasia or DALM. Because the significance of low-grade or moderate dysplasia is less clear, immediate resection for patients with these levels of dysplasia is controversial. Prophylactic colectomy also has been recommended as an option for those with disease of at least 10 years’ duration. Studies employing flow cytometry have detected aneuploid cell populations in colons resected for UC with dysplasia or early cancer. Chromosomal alterations can occur early in UC-related neoplastic progression and appear to precede the histologic development of dysplasia; relative loss of chromosome 18q also may be important in neoplastic progression.

Both dysplasia and increased risk of colon carcinoma have been reported in patients with Crohn’s disease. As in UC, dysplasia appears in diseased colon segments, and its presence correlates with duration of disease. In one screening trial,⁹⁰ the finding of definite dysplasia was associated with age older than 45 years and increased symptoms. By life-table analysis, the probability of detecting dysplasia or cancer after a negative screening colonoscopy was 22% by the fourth surveillance examination.

OTHER ASSOCIATIONS

Diverting bile and bile acids to the lower small intestine, either surgically or by administration of cholestyramine, increases the yield of proximal colon tumors in carcinogen-treated animals. There is no evidence, however, that chronic use of cholestyramine is associated with an increased risk of colon cancer in humans. The possibility of an increase in colon cancer following cholecystectomy has been raised, although the clinical evidence for such an association is contradictory. Increased proliferative activity in the distal colonic mucosa has been demonstrated in patients who have undergone cholecystectomy, and an increased incidence of tubular adenomas has been observed in patients older than 60 years with a postcholecystectomy interval of more than 10 years, but an increased risk of colon cancer for these patients has been both supported and refuted.

A retrospective population-based cohort study that used the Swedish Inpatient Register evaluated almost 23,000 persons who had had cholecystectomy up to 31 years previously.⁹¹ Patients having had cholecystectomy had an increased risk of proximal intestinal adenocarcinoma, which declined with increasing distance from the common bile duct. The risk was significantly increased for adenocarcinoma of the small bowel and ascending colon but not the remaining colon or rectum.

GROSS PATHOLOGY

The gross morphologic features of CRC depend on the tumor’s site. Carcinomas of the proximal colon, particularly those of the cecum and ascending colon, tend to be large and bulky, often outgrowing their blood supply and undergoing necrosis (Fig. 123-16); this polypoid configuration, however, also may be found elsewhere in the colon and rectum. In the more distal colon and rectum, tumors more commonly involve a greater circumference of the bowel, producing an annular constriction or napkin-ring appearance (Fig. 123-17). The fibrous stroma of these tumors accounts for the constriction and narrowing of the bowel lumen, whereas the circular arrangement of colonic lymphatics is responsible for their annular growth. These tumors also can ulcerate, and occasionally they have a flatter appearance with predominantly intramural spread (Fig. 123-18); the latter are seen most often in the setting of IBD. Morphologic features of CRCs have clinical, diagnostic, and prognostic implications.

Figure 123-16. Carcinomas of the cecum seen at colonoscopy. Carcinomas of the proximal colon are often large and bulky polypoid lesions (A and B), may involve the ileocecal valve (C), and can outgrow their blood supply and become necrotic (D).

Figure 123-17. Obstructing carcinoma of the sigmoid colon. A, Colonoscopic view. B, Apple-core lesion seen on barium enema examination. C, Surgical specimen demonstrating an annular constriction or napkin-ring appearance.

Figure 123-18. Colonoscopic view of flat, plaque-like carcinomas of the colon (A) and rectum (B, arrows) in patients with inflammatory bowel disease.

HISTOLOGY

Carcinomas of the large bowel characteristically are adenocarcinomas that form moderately differentiated to well-differentiated glands and secrete variable amounts of mucin (Fig. 123-19). Mucin, a high-molecular-weight glycoprotein, is the major product secreted by both normal and neoplastic glands of the colon, and may be seen best with histochemical stains such as periodic acid–Schiff (PAS). In poorly differentiated tumors (see Fig. 123-19), gland formation and mucin production are present but less prominent. Signet ring cells, in which a large vacuole of mucin displaces the nucleus to one side, are a feature of some tumors (Fig. 123-20A). In approximately 15% of tumors, large lakes of mucin contain scattered collections of tumor cells (see Fig. 123-20B). Such mucinous or colloid carcinomas are most common in patients with HNPCC or UC and in patients whose carcinomas occur at an early age. Scirrhous carcinomas are uncommon and are characterized by sparse gland formation, with marked desmoplasia and fibrous tissue surrounding glandular structures. Sometimes tumors demonstrate a mixed histologic picture, with glands of varying degrees of differentiation.

Figure 123-19. Histopathology of adenocarcinoma of the colon. A, Well-differentiated adenocarcinoma of the colon. Sections stained with hematoxylin and eosin demonstrate crowded neoplastic glands containing variable amounts of mucin. B, Poorly differentiated adenocarcinoma of the colon.

Figure 123-20. Histopathology of mucinous carcinomas of the colon. Histologic types include signet ring cell carcinoma in which a large vacuole of mucin displaces the nucleus (A) and colloid carcinoma with scattered nests of tumor cells floating in lakes of mucin (B) (Hematoxylin and eosin stain.)

Cancers other than adenocarcinomas account for less than 5% of malignant tumors of the large bowel. Tumors arising at the anorectal junction include squamous cell carcinomas, cloacogenic or transitional cell carcinomas, and melanomas (see Chapter 125). Primary lymphomas and carcinoid tumors of the large bowel account for less than 0.1% of all large bowel neoplasms (see Chapters 29 and 31).

NATURAL HISTORY AND STAGING

CRCs begin as intramucosal epithelial lesions, usually arising in adenomatous polyps or glands. As cancers grow, they become invasive, penetrating the muscularis mucosae of the bowel and invading lymphatic and vascular channels to involve regional lymph nodes, adjacent structures, and distant sites. Although CRCs grow at varying rates, they most often have long periods of silent growth before producing bowel symptoms. The mean doubling time of colon cancers determined radiologically in one study was 620 days. Comparative lesion sequencing using modern molecular techniques combined with clinical observation suggests that it can take approximately 17 years for a large benign tumor to evolve to advanced cancer but less than two years for cells within that cancer to acquire the ability to metastasize.⁹² Patterns of spread depend on the anatomy of the individual bowel segment as well as its lymph and blood supplies.

Cancers of the rectum advance locally by progressive penetration of the bowel wall. Extension of the primary tumor intramurally and parallel to the long axis of the bowel most often is limited, and lymphatic and hematogenous spread is unusual before penetration of the muscularis mucosae. Exceptions appear to be poorly differentiated tumors, which can metastasize lymphatically or hematogenously before completely penetrating the bowel. Because the rectum is relatively immobile and lacks a serosal covering, rectal cancers tend to spread contiguously to progressively involve local structures. Transrectal ultrasonography is useful in staging depth of rectal cancers. Because of the dual blood supply of the lower one third of the rectum, tumors arising here can metastasize hematogenously via the superior hemorrhoidal vein and portal system to the liver or by way of the middle hemorrhoidal vein and inferior vena cava to the lungs. The veins of the upper and middle thirds of the rectum drain into the portal system, and tumors in these segments first spread hematogenously to the liver. Occasionally, lumbar and thoracic vertebral metastases result from hematogenous spread via portal-vertebral communications (i.e., Batson’s vertebral venous plexuses).

Colon cancers can invade transmurally and involve regional lymphatics and then distant nodes; lymphatic drainage generally parallels the arterial supply to a given bowel segment. The liver is the most common site of hematogenous spread from colon tumors via the portal venous system. Pulmonary metastases from colon cancer result, in general, from hepatic metastases.

On the basis of observations of what he believed to be an orderly progression of locoregional invasion by rectal cancers, Cuthbert Dukes proposed a staging classification in 1929, which has since been modified many times in attempts to increase its prognostic value for cancers of both the rectum and the colon. The most commonly employed modification of the Dukes system is that of Astler and Coller (Table 123-9). This classification uses the following designations:

Stage C tumors are divided further into primary tumors limited to the bowel wall (C1) and those that penetrate the bowel wall (C2). By contrast, in the system proposed by the Gastrointestinal Tumor Study Group (GITSG), C1 lesions are those in which one to four regional lymph nodes contain tumor, and C2 lesions are those in which more than four lymph nodes contain tumor (see Table 123-9). Another modification, by Turnbull and associates, added a D category referring to distant metastases.

In an attempt to provide a uniform and orderly classification for CRCs, the American Joint Committee on Cancer (AJCC) has introduced the tumor-node-metastases (TNM) classification for CRC.⁹³ This system classifies the extent of the primary tumor (T), the status of regional lymph nodes (N), and the presence or absence of distant metastases (M). Cases are assigned the highest value of TNM that describes the full extent of disease and are grouped into five stages (0 through IV). The five stages have become important in uniformly randomizing patients for therapeutic trials (Table 123-10) and have, in most cases, replaced the Dukes classification for CRC.

Table 123-10 AJCC TNM Staging of Colorectal Cancer

AJCC, American Joint Committee on Cancer; N0, no regional lymph node metastasis; M0, no distant metastasis; M1, distant metastasis.

* Stage II is subdivided into IIA (for T3 tumors) and IIB (for T4 tumors). Stage III is subdivided into IIIA (T1-T2 N1 M0), IIB (T3-T4 N1 M0), and IIIC (any T N2 M0). N1 lesions have 1-3 positive nodes; N2 tumors have ≥4 positive nodes. Smooth metastatic nodules in the pericolic or perirectal fat are considered lymph node metastases. Irregularly contoured metastatic nodules in the peritumoral fat are considered vascular invasion.

† Dukes B, which corresponds to stage II, is a composite of better (T3 N0 M0) and worse (T4 N0 M0) prognostic groups, as is Dukes C, which corresponds to stage III (any T N1 M0 and any T N2 M0).

‡ Tis includes cancer cells confined within the glandular basement membrane (intraepithelial) or lamina propria (intramucosal) with no extension through the muscularis mucosae into the submucosa.

§ Direct invasion in T4 includes invasion of other segments of the colorectum by way of serosa; e.g., invasion of the sigmoid colon by a carcinoma of the cecum.

From the American Joint Committee on Cancer. AJCC Cancer Staging Manual: Colon and Rectum, 6th ed. New York: Springer-Verlag; 2002.

SURGICAL-PATHOLOGIC STAGING

The depth of transmural tumor penetration and the extent of regional lymph node spread are the most important determinants of CRC prognosis (Fig. 123-21). The degree of bowel wall penetration also affects prognosis, independently of lymph node status, and correlates with the number of involved nodes as well as with the incidence of local recurrence after surgical resection. The number of involved regional lymph nodes also correlates independently with outcome (see Fig. 123-21B).

Figure 123-21. A, Survival probabilities according to the Dukes stage as modified by Astler-Coller (see Table 123-9) in patients undergoing potentially curative surgery for colorectal cancer. Expected survival among age- and sex-matched general populations is indicated by the straight line. B, Survival probabilities according to the number of nodes involved in patients with stage C colorectal carcinoma. Data using TNM staging confirm a substantial difference in five-year survival among patients with lesions having one to three positive nodes (N1) and four or more positive nodes (N2).

(From Moertel CG, O’Fallon JR, Go VL, et al. The preoperative carcinoembryonic antigen test in the diagnosis, staging, and prognosis of colorectal cancer. Cancer 1986; 58:603.)

TUMOR MORPHOLOGY AND HISTOLOGY

The TNM classification (see Table 123-10) is based in part on the observation that for most cancers, tumor size correlates with local and distant spread and, thus, indirectly with prognosis. Numerous studies suggest that CRC is an exception, however, and that the size of the primary tumor per se does not correlate with prognosis. In fact, patients with exophytic or polypoid tumors appear to have a better prognosis than those with ulcerating or infiltrating tumors.

Tumor prognosis correlates with histologic grade: poor differentiation confers a worse prognosis than does a high degree of differentiation. Mucinous⁶⁵ and scirrhous carcinomas appear to be biologically more aggressive, and patients with these tumors do not survive as long as those who have other types of adenocarcinoma. Mucin-associated antigens might play a role in tumor progression and metastasis of colon cancer cells. Signet ring carcinomas have been reported to present at an advanced stage and to be highly invasive tumors.

Venous invasion by CRC (Fig. 123-22A) usually correlates with local recurrence after resection, visceral metastases, and decreased survival, but it might not have independent prognostic value for tumors confined to the bowel wall. Although lymphatic invasion is associated with decreased survival, it is not clear whether this variable is independent of depth of tumor invasion and regional nodal metastasis. Perineural invasion (see Fig. 123-22B) also is linked to increased local recurrence and decreased survival, but such data are limited. The degree of inflammatory response and lymphocytic infiltration in and around a cancer may be related to outcome; increased inflammation and immune reaction appear to confer a better prognosis, but data are limited.

Figure 123-22. Pathologic features that can influence prognosis adversely include vascular invasion (A) and lymph node invasion (B). The arrow in A points to a lymphatic vessel containing tumor cells. In B, high-magnification microscopy demonstrates a lymph node that contains adenocarcinoma cells. (Hematoxylin and eosin stain.)

The prognosis of patients with CRC may be related to the DNA content of the primary tumor, because survival is shorter for patients with nondiploid or aneuploid tumors than for those whose tumor cells have a normal or diploid DNA content. Although the DNA content of the primary tumor might correlate with the potential for local or distant recurrence after primary resection, the value of routine flow cytometric measurements or image analysis of the DNA content of tumor cells for assessing clinical prognosis and planning treatment remains to be determined. Deletions in chromosomes 18q and 17p (TP53) may be important indicators of prognosis, independent of stage.^55,57 As indicated in Table 123-11, a growing number of other molecular markers also might predict prognosis or response to therapy.⁵⁷

CLINICAL PREDICTORS OF PROGNOSIS

Whereas screening programs for CRC suggest that tumors diagnosed in asymptomatic patients are less advanced, assessment of the impact of early diagnosis on survival of asymptomatic persons awaits the results of prospective randomized, controlled studies such as the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial.⁹⁴ Duration of symptoms might not correlate directly with prognosis, and some presenting symptoms, such as rectal bleeding, may be associated with better rates of survival.

Bowel obstruction or perforation has been linked with poor prognosis. Patients who present with obstructing lesions might not be candidates for curative surgery and have higher rates of operative morbidity and mortality. Recurrence following “curative” surgery also is higher in patients who present with obstruction or perforation.

The location of the primary tumor can influence outcome. Disease-free survival at three years appears to be 2% to 14% higher after surgery for tumors of the left than of the right colon. Some studies suggest a survival advantage for patients with colon compared with rectal cancers.

As many as 3% of CRCs develop before 30 years of age, and only 11% of such persons have a predisposing condition such as FAP or UC. The prognosis is worse than for older patients and is particularly poor in the pediatric age group. Poor prognosis may be related to a higher percentage of more-advanced cancers and mucinous adenocarcinomas in these young patients. Alternatively, patients with tumors demonstrating MSI appear to have a better prognosis irrespective of age.⁶⁰ Thus, whereas CRCs occur at a younger age in those with HNPCC, these patients have a better prognosis than those with microsatellite-stable cancers.

Outcome also is related to preoperative serum CEA levels. Tumor recurrence is higher, and the estimated mean time to recurrence is shorter, in patients with Dukes B and C cancers who have high preoperative CEA levels. The preoperative CEA level may be of prognostic value only in patients with Dukes C CRCs who also have four or more involved lymph nodes (stage C2), but not in patients with Dukes A and B lesions or Dukes C lesions with fewer than four nodes involved. Expression of mucin-associated carbohydrate antigens other than CEA, such as sialyl Lewis^x, also can correlate with prognosis.⁶⁵ Expression of the carbohydrate-binding protein galectin-3 correlates with tumor progression in the colon and can confer a poor prognosis.^65,67

Approximately one fourth of patients with CRC exhibit clinical evidence of hematogenous spread when seen initially, and one half of patients with CRC eventually develop metastases to a distant site, usually the liver; such metastases carry a poor prognosis at all times in the clinical course. The most important determinant of survival time for patients who present with liver metastases is the extent of hepatic involvement by tumor.