Histologic Features

Adenomatous polyps are tumors of benign neoplastic epithelium that can either be pedunculated (i.e., attached by a stalk) or sessile (i.e., attached by a broad base with little or no stalk). The neoplastic nature of adenomas is apparent by histologic examination of their glandular architecture. Tubular adenomas are the most common subgroup and are characterized by a complex network of branching adenomatous glands (Fig. 122-1A). In villous adenomas, the adenomatous glands extend straight down from the surface to the center of the polyp, thereby creating long, finger-like projections (see Fig. 122-1B). Tubulovillous (villoglandular) adenomas manifest a combination of these two histologic types.

Figure 122-1. Comparison of tubular and villous histology. A, Tubular adenomas consist of branched, crowded glands arranged in a complex cerebriform pattern. B, Villous adenomas consist of glands that are long, finger-like fronds typically projecting from the polyp stroma to the surface without much branching.

(Courtesy of Noam Harpaz, MD, PhD, New York.)

A polyp is assigned a histologic type on the basis of its predominant glandular pattern, and in practice, pure villous adenomas are quite rare. According to the World Health Organization, adenomas are classified as tubular if at least 80% of the glands are of the branching, tubule type and as villous if at least 80% of the glands are villiform.¹ Of all adenomatous polyps, tubular adenomas account for 80% to 86%, tubulovillous for 8% to 16%, and villous adenomas for 3% to 16%.^2,3 Tubular adenomas usually are small and exhibit mild dysplasia, whereas villous architecture is more often encountered in large adenomas and tends to be associated with more severe degrees of dysplasia (Table 122-2).

By definition, all colorectal adenomas are dysplastic. Adenomatous epithelium is characterized by abnormal cellular differentiation and renewal, resulting in hypercellularity of colonic crypts with cells that possess variable amounts of mucin and that are hyperchromatic, with elongated nuclei arranged in a picket-fence pattern. These cytologic alterations confer an increased basophilic appearance to the adenomatous epithelium on conventional hematoxylin-eosin staining. Although the predominant cell type is an immature goblet cell or columnar cell, adenomas can contain other cell types, such as neuroendocrine cells, Paneth cells, squamous morules, and, rarely, melanocytes. On cross section, the inner contour of an adenomatous gland lumen usually is smooth, in contrast to the serrated appearance of a hyperplastic gland lumen (see later).

The dysplasia exhibited by all adenomas can be graded subjectively on the basis of certain cytologic and architectural features into three categories: mild, moderate, and severe. Some polyps contain the entire spectrum from mild to severe dysplasia, but in all cases, the adenoma is classified according to its most dysplastic focus. In cells that exhibit mild dysplasia, the nuclei in the cell maintain their basal polarity but are hyperchromatic, slightly enlarged, and elongated, yet uniform in size, without prominent nucleoli (Fig. 122-2A). There often is loss of goblet cell mucin. Architecturally, the glands manifest branching and budding and become more crowded. With moderate dysplasia, nuclei become stratified and pleomorphic, with prominent nucleoli, along with further loss of goblet cell mucin and increased glandular crowding. Severe dysplasia (see Fig. 122-2B) is characterized by further stratification and pleomorphism of nuclei, more numerous and prominent nucleoli, increased nucleus-to-cytoplasm ratios, and extreme glandular crowding.

Figure 122-2. Comparison of low-grade dysplasia (LGD) and high-grade dysplasia (HGD). A, LGD is characterized by branching crypts lined by cells with long, thin nuclei that begin to stratify, resulting in an increased nucleus-to-cytoplasm ratio and a loss of normal goblet cells. These changes occupy the entire crypt, including the surface epithelium. B, Adenoma with HGD and intramucosal carcinoma. This photomicrograph shows a section of adenoma with crypts demonstrating HGD (right, arrowheads). There is nuclear pleomorphism and abnormal cellular polarity resulting in a cribriform appearance (i.e., glands within glands). A focus of intramucosal carcinoma is seen in the center (arrows), characterized by cells with pleomorphic nuclei and poor gland formation that are located in the lamina propria.

(Courtesy Noam Harpaz, MD, PhD, New York.)

With further cell proliferation within the crypt, cells pile up, lose polarity, and create glands within glands, giving a disorderly cribriform appearance termed carcinoma in situ (see Fig. 122-2B). Most pathologists group severe dysplasia and carcinoma in situ, considering them both high-grade dysplasia²; one reason for this grouping is to avoid using the term carcinoma for these lesions because they often can be managed endoscopically rather than surgically (see later). Indeed, it is now common practice to categorize dysplasia in colorectal adenomas into only two grades: low-grade dysplasia, which includes mild and moderate dysplasia, and high-grade dysplasia, which comprises severe dysplasia and carcinoma in situ.

Carcinoma in situ is characterized by intracryptal cell proliferation that leaves intact the basement membrane surrounding the gland. If a focus of neoplastic cells grows beyond the basement membrane and into the lamina propria of the mucosa, the lesion is termed intramucosal carcinoma (see Fig. 122-2B). Both carcinoma in situ and intramucosal carcinoma are noninvasive lesions without metastatic potential, because lymphatics are not present in the colonic mucosa above the level of the muscularis mucosae.⁴ Because clinical confusion often arises on encountering these two entities, it has been suggested that both carcinoma in situ and intramucosal carcinoma be reported as “noninvasive carcinoma” to avoid unnecessarily aggressive management. Only when a focus of neoplastic cells has spread through the muscularis mucosae is the lesion considered invasive carcinoma (Fig. 122-3). An adenoma that contains a focus of invasive carcinoma commonly is referred to as a malignant polyp (see later).

Figure 122-3. Histopathology of a malignant polyp. This pedunculated adenoma demonstrates tubulovillous histology on the surface. The entire stalk of the polyp (center, arrows) contains numerous malignant glands representing well-differentiated adenocarcinoma. Unlike the adenomatous glands on the surface, many of the invasive glands demonstrate sharply angulated edges.

(Courtesy of Noam Harpaz, MD, PhD, New York.)

Of all adenomatous polyps, mild dysplasia is found in 70% to 86% moderate dysplasia in 18% to 20%, severe dysplasia (carcinoma in situ) in 5% to 10%, and invasive carcinoma in 5% to 7%.^3,5,6 Higher grades of dysplasia are more common in adenomas of larger size and greater villous content,² and adenomas with severe dysplasia are more likely to contain foci of invasive cancer.

Adenoma Size

Adenomas are categorized into three size groups: less than 1 cm, 1 to 2 cm, and greater than 2 cm.⁵ Overall, most adenomas are smaller than 1 cm, but the size distribution of adenomas can vary greatly among studies, depending on study design, age of the study population, and location of the adenomas within the colon. Thus, in autopsy series, which describe a presumably asymptomatic population dying of other causes, only 13% to 16% of adenomas are larger than 1 cm,^7–9 whereas surgical and colonoscopic series that include symptomatic or higher-risk patients report a higher prevalence (26% to 40%) of adenomas larger than 1 cm.^2,3,5 In countries where the prevalence of colon cancer is high, adenomas tend to be larger than in low-prevalence countries.^10,11 Adenoma size increases as a function of age,^8,12,13 even in low-prevalence countries,¹⁰ and larger adenomas are more common in distal colonic segments.^2,5,8

Diminutive Polyps

Diminutive polyps measure 5 mm or less in diameter and are commonly encountered during endoscopy. An earlier concept that these lesions were almost always non-neoplastic has been revised based on several flexible sigmoidoscopic and colonoscopic studies in which 30% to 50% of diminutive polyps were found to be adenomatous^14–18; despite the frequency of adenomatous change, however, they represent little if any threat of cancer. Earlier studies found that less than 1% of diminutive polyps were villous or contained a focus of severe dysplasia and that they almost never harbored invasive carcinoma.^14–16,18 In an analysis of 4381 diminutive polyps, 4.4% contained severe dysplasia or villous components, although still only 0.1% had invasive carcinoma.¹⁹ Moreover, in a retrospective study of predominantly asymptomatic people with diminutive adenomas found on flexible sigmoidoscopy, full colonoscopy identified a synchronous proximal adenoma in only 33% of subjects, and most of the proximal lesions were smaller than 5 mm.²⁰ Likewise, prospective colonoscopic studies confirm only a 24% to 34% prevalence of proximal adenomas in asymptomatic patients with distal diminutive polyps (of all histologic types)^21,22; the likelihood of finding proximal adenomas is greater when the distal polyp is larger than 5 mm.²¹ Diminutive adenomas manifest little, if any, appreciable growth over time.^23,24 A population-based study that involved fulgurating small polyps (even those up to 1 cm) without obtaining initial histologic identification reported that the subsequent risk for colorectal cancer and overall survival was no worse than in the general population.²⁵

Thus, taken together, these observations indicate that diminutive polyps, even when they prove to be adenomas, have little biologic or clinical significance. Nonetheless, the fact that these tiny polyps often are missed by computed tomographic (CT) colonography (or virtual colonoscopy; see later) has provided some fuel to the debate over the safety of leaving these lesions undetected. An important exception to the rule of the innocuous nature of diminutive adenomas is in the setting of hereditary nonpolyposis colorectal cancer (HNPCC), in which even small adenomas can display advanced pathologic features such as villous histology or high-grade dysplasia (see later).

Malignant Potential of Adenomatous Polyps

The three principal features that correlate with malignant potential for an adenomatous polyp are size, histologic type, and degree of dysplasia (Table 122-3). Although higher rates of malignant transformation are found when the source of the pathologic material is mainly from surgical polypectomies⁵ rather than colonoscopic polypectomies,⁶ the malignant potential is correlated directly with larger adenoma size, more villous histology, and higher degrees of dysplasia. These three histopathologic criteria usually are interdependent, so it is difficult to assign a primary premalignant role to any one of them. For example, although only 1.3% of all adenomas smaller than 1 cm harbor a cancer (see Table 122-3), if these small lesions have a predominant villous component or contain a focus of severe dysplasia, the cancer rate rises to 10% or 27%, respectively (Table 122-4). A small (<1 cm), tubular, mildly dysplastic adenoma is highly unlikely to harbor a focus of invasive cancer. Nonetheless, although this type of lesion is innocuous in itself, once removed, it often is considered a marker of a person who is at (low) risk for developing a recurrent adenoma (discussed later). Because adenomas that are larger than 1 cm, have villous architecture, or manifest high-grade dysplasia or carcinoma represent a more biologically hazardous group, the term adenoma with advanced pathology (AAP) often is applied to adenomas that display any of these features.

Other Adenoma Variants

Histogenesis

Adenomatous polyps are thought to arise from a failure in a step, or steps, of the normal process of cell proliferation and cell death (apoptosis). The initial aberration appears to arise in a single colonic crypt in which the proliferative compartment, instead of being confined to the crypt base, is expanded throughout the entire crypt. This disturbance results in a unicryptal adenoma. The DNA-synthesizing cells at the surface are not sloughed into the lumen, as normally occurs, and they accumulate in a downward infolding manner, interposing themselves between normal preexisting crypts. New adenomatous glands then are created either by further infolding or by branching. Thus, the unicryptal adenoma is believed to arise from a monoclonal expansion of an abnormal cell, and as the adenoma enlarges, the adenomatous cell population becomes polyclonal. Evidence for this concept comes from studying intestinal tissues from an extremely rare patient with FAP who was an XO/XY mosaic.⁴⁴ Analysis of Y chromosome expression in the intestinal mucosa of this patient revealed that normal crypts of the small and large intestine and even unicryptal adenomas were monoclonal (either XO or XY), whereas at least 76% of very small microadenomas were polyclonal. Whether the same situation also applies to sporadic adenoma development is not clear at present.

Adenoma-Carcinoma Hypothesis

It is generally accepted that most, if not all, colon cancers originate within previously benign adenomas. Rarely, colon cancers develop de novo in apparently flat, nonadenomatous epithelium although, as noted earlier, even these lesions might conceivably arise from preexisting flat adenomas, or serrated polyps. Evidence in support of the adenoma-carcinoma sequence comes from epidemiologic, clinical, pathologic, and molecular studies.

Pathways of Colon Carcinogenesis

It is useful to consider the process of colon carcinogenesis in two general stages: the formation of the adenoma, termed tumor initiation, and the progression of the adenoma to carcinoma, termed tumor progression (Fig. 122-4). It is believed that most, if not all, adenomas arise from an initial loss of APC gene function, and for that to happen, epithelial cells must lose the function of both APC alleles (two hits). In patients with FAP, one APC allele is inherited in a mutated form (germline mutation) from the affected parent. Adenomas arise when the second, normal copy of the APC gene from the unaffected parent either is lost or mutated (somatic mutation). Because persons with FAP are born with the first hit, they develop polyps at a much younger age and in much greater number than does the general population; thus, FAP can be considered a condition of accelerated tumor initiation. Despite this abnormal initiation rate, once adenomas form in patients with FAP, it is believed that each adenoma tends to display a normal progression to carcinoma. Thus, the inevitable progression to cancer in FAP is more a consequence of the numerous polyps than of any increased premalignant potential of the individual adenoma. In the general population, sporadic adenomas arise as a consequence of two acquired somatic mutations of the APC gene. Because two acquired hits are statistically less likely than one acquired hit, sporadic adenomas tend to occur later in life and to be fewer than the polyps in patients with FAP.

Figure 122-4. Pathways of colon carcinogenesis. Adenomas develop as a consequence of factors involved in initiating tumors, and they progress to carcinomas because of factors that act as tumor promoters. A simplified theory comparing the two hereditary colon cancer syndromes suggests that the adenoma-initiation phase is accelerated in patients with familial adenomatous polyposis (FAP), accounting for numerous polyps. In contrast, the adenoma progression phase is accelerated in hereditary nonpolyposis colorectal cancer (HNPCC) patients, accounting for the often rapid progression of adenomas to carcinoma. APC, adenomatous polyposis coli; DCC, deleted in colon cancer.

Another major molecular pathway for colon carcinogenesis involves mutations in DNA MMR genes (see Chapter 123); this is the predominant pathway in patients with HNPCC. Mutations in these enzymes result in a characteristic molecular phenotype termed microsatellite instability (MSI), a phenomenon that is observable in colon cancer cells from approximately 85% of HNPCC colon cancers but in only 15% of sporadic colon cancers. Although its name implies a lack of polyps, HNPCC colon cancers do arise from preexisting adenomas. It is believed that the numbers of adenomas that occur in patients with HNPCC are similar to those in the general population but that HNPCC is marked by an accelerated tumor progression stage, so the few adenomas that do arise often manifest advanced pathology (villous features, high-grade dysplasia) even at small sizes.⁵³ Indeed, adenomas in patients with HNPCC often manifest MSI⁵⁴ even in their earliest stages of formation. Because these adenomas tend to progress more rapidly to carcinoma,^55,56 surveillance intervals for colonoscopy following removal of adenomas in HNPCC patients should be shortened (see later; see also Chapter 123).

Prevalence

The prevalence of adenomatous polyps is affected by four major factors: the inherent risk for colon cancer in the population, age, sex, and family history of colorectal cancer. The frequency of colon adenomas varies widely among populations, but it tends to be higher in populations at greater risk for colon cancer (Table 122-5).¹⁰ One illustrative example is to compare the very high adenoma prevalence in Japanese living in Hawaii (a very high risk area for colon cancer) with the much lower adenoma prevalence in Japanese who still reside in Japan, an area of much lower risk. Even within Japan itself, adenoma prevalence correlates quite well with colon cancer prevalence in different prefectures of the country. Data from autopsy series provide an approximation of adenoma prevalence. In populations at low risk for colon cancer, adenoma prevalence rates are lower than 12% (see Table 122-5), whereas in most intermediate- and high-risk populations, adenomas are found in 30% to 40% of the population, and rates as high as 50% to 60% have been observed.^7,57 One half to two thirds of people older than 65 years in high-risk areas can harbor colonic adenomas.^7,57

The true prevalence rate of adenomatous polyps within an asymptomatic population is only now being elucidated because, until recently, colonoscopy was not performed on healthy persons in the absence of gastrointestinal symptoms. Approximately 27% to 32% of asymptomatic average-risk persons older than 50 years undergoing screening colonoscopy will have an adenoma, and 6% to 10% will have an AAP (Table 122-6).^58–61 By comparison, colonoscopic screening of asymptomatic persons between 40 and 49 years of age revealed prevalence rates of only 8.7% for tubular adenomas and 3.5% for AAP or cancer.⁶² Colonoscopic series indicate that men have a 1.5 relative risk of adenomas compared with age-matched women,^62,63 thus confirming earlier observations in autopsy series.^8,11 Men also have slightly higher rates of advanced adenomas than do women, with a relative risk of approximately 1.5.⁶⁴ A large database study found a higher rate of polyps larger than 9 mm in African Americans undergoing screening colonoscopy compared with age-matched whites.⁶⁵

The prevalence of adenomas is higher in older people, particularly those older than 60 years. In fact, age is the single most important independent determinant of adenoma prevalence^{7,11,12,57,62,63,66,67} in high-risk and low-risk regions of the world (see Table 122-5). Not only is advancing age associated with a higher prevalence rate of adenomas, but it also correlates with a greater likelihood for multiple polyps, adenomas with more severe degrees of dysplasia, and, in some studies, larger adenomas.

Adenoma prevalence also is higher in persons with a family history of colorectal cancer and adenomas,^67,68 particularly if more than one relative is affected with colorectal neoplasia and if the affected relative is young.

Incidence

Estimating the incidence of new adenomas requires examining the colon at more than one point in time. Two types of endoscopic studies lend themselves to this analysis: surveillance studies following polypectomy (or following cancer resection) and interval examinations in persons who initially had a negative examination. Of course, for both types of studies, the small but measurable miss rate of adenomas can influence the rate of apparent incident adenomas (see “Detection,” later). For the purposes of this discussion, adenomas found in persons after polypectomy are considered recurrences (see “Postpolypectomy Management”), whereas those found in persons after an initial negative colonoscopy are considered incident adenomas.

The incidence of new adenomas varies from 24% to 41%.⁶³ In one study, patients underwent colonoscopy twice on the same day to clear the colon of all potentially missed adenomas, and 38% were found to have new adenomas at colonoscopy two years later.⁶⁹ Three-year follow-up sigmoidoscopy after an initial negative examination in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial found a 3.1% incidence of all adenomas and 0.8% incidence of advanced adenomas or cancer.⁷⁰ A large surveillance study looked at findings on follow-up colonoscopy performed within 5.5 years of the index colonoscopy. Patients who had one or two tubular adenomas less than 10 mm were no more likely to have advanced neoplasia than patients with negative baseline colonoscopies. In average-risk, asymptomatic persons with no adenomas at baseline colonoscopy, repeat colonoscopy within five years detects an adenoma in approximately 16% to 27%^71,72 and an AAP in approximately 1% to 2.4%.^71–73

Anatomic Distribution

The distribution of adenomatous polyps within the colon differs, depending on the method of investigation (Table 122-7). In autopsy series in which the normal distribution is approximated in presumably asymptomatic subjects, adenomas are distributed uniformly throughout the colorectum; this even distribution has been confirmed in colonoscopic investigations of asymptomatic subjects.^23,57 Large adenomas in autopsy series have a distal predominance, in the region where most colon cancers arise, thereby supporting the adenoma-carcinoma hypothesis. Likewise, adenomas detected in surgical and colonoscopic studies of symptomatic people also display a left-sided predominance, indicating that distal adenomas are more likely to come to clinical attention. In older persons, particularly those older than 60 years, distribution of adenomas demonstrates a shift toward more proximal colonic locations. This phenomenon, which is based on autopsy^11,12 and on colonoscopic studies of symptomatic⁷⁴ and asymptomatic^{22,58,59,66,75} subjects, has importance for choosing appropriate colon cancer screening approaches (see Chapter 123). Some data suggest that African Americans have a greater proportion of proximal adenomas compared with whites, especially in persons older than 60 years.⁶⁵

Inherited Susceptibility to Adenomatous Polyps

There is a strong genetic component to the well-defined hereditary polyposis (FAP) and nonpolyposis (HNPCC) colon cancer syndromes that exhibit a Mendelian pattern of inheritance (see later); however, 95% of common (sporadic) adenomas and carcinomas arise in people who do not have these syndromes. In the past, this observation was interpreted to mean that genetic predisposition played only a minor role in most colonic neoplasms. Epidemiologic studies, however, have revealed a two- to three-fold increased risk for colon cancer in probands who have a first-degree relative affected by colonic cancer or adenoma⁶⁷; a similarly increased risk also has been found for adenomas in first-degree relatives of people with adenomas.⁶⁷ Moreover, data from the National Polyp Study indicate that siblings and parents of patients with adenomatous polyps are at an increased risk for colon cancer, particularly when the adenoma proband was younger than 60 years at diagnosis.⁷⁷ There is even a suggestion that adenomas in patients with a family history of colorectal cancer have faster growth rates.⁷⁸ Of patients with HNPCC, 11% were found to have at least one adenoma on first colonoscopy; 5% had at least one hyperplastic polyp; and the frequency of adenomas, but not hyperplastic polyps, increased with older age.⁷⁹

It is now estimated that as much as 10% to 30% of colon cancers are familial, implying the possibility of susceptibility genes that give rise to common colon cancers.⁷² Despite epidemiologic data suggesting an increased risk in patients with family members who have colon cancer or adenomas, however, it has been difficult to fully elucidate the causative genetic basis. Several genes have been identified that can contribute to common familial risk. These include a germline mutation in the APC gene at codon 1307 (I1307K) that appears to predispose Ashkenazi Jewish populations to colon cancer; mutations of the DNA MMR gene hMSH6, a type I transforming growth factor (TGF)-β receptor allele TbR-I(6A); and polymorphisms of certain genes involved in the metabolism of nutrients and environmental agents, such as methylenetetrahydrofolate reductase and N-acetyltransferase-1 and -2.^67,80 A specific mutation in methylenetetrahydrofolate reductase has been found to be protective against colon cancer risk.⁸¹ Germline mutations of APC and other genes involved in the APC pathway, such as β-catenin and AXIN1, as well as the DNA mismatch repair genes hMLH1 and hMSH2, have been implicated in the predisposition to multiple adenomas.⁸² The identification of genes responsible for common susceptibility to colonic adenoma and carcinoma, particularly using polymorphisms of candidate genes, is an area of considerable research.⁸³

Dietary and Lifestyle Risk Factors

Although genetic predisposition clearly plays a role in colorectal carcinogenesis, diet and life-style factors also contribute. It is estimated that as much as a third to a half of colon cancer risk and a fourth to a third of distal colon adenoma risk might be avoidable by modification of dietary and life-style habits.⁸⁴ For the most part, dietary factors that correlate with a predisposition to colon cancer (see Chapter 123) also are associated with a risk for colonic adenomas.^84,85 Factors that have each been correlated with an increased adenoma risk include excess dietary fat, excess alcohol intake, obesity, and cigarette smoking.⁸⁶ Curiously, low calcium intake, despite being associated with increased risk for colon cancer, does not appear to confer risk for adenoma (although as discussed later, calcium supplementation does seem to lower adenoma recurrences).

Factors that have shown the most consistent protective effect against adenomas in epidemiologic studies include dietary fiber, plant foods, and carbohydrate. Indeed, analysis of dietary questionnaires from patients in the prospective PLCO Trial found that those with the highest fiber intake had a 27% lower risk of adenomas compared with those who had the lowest fiber intake.⁸⁷ Other protective measures include increased physical activity, increased intake of calcium, and high intake of folate. An analysis of an asymptomatic, predominantly male veteran population undergoing screening colonoscopy found that smoking and moderate-to-heavy alcohol use increased risk, whereas cereal fiber intake, vitamin D intake, and use of nonsteroidal anti-inflammatory drugs (NSAIDs) decreased risk for advanced colonic neoplasia (AAPs and colon cancer).⁸⁸

Unfortunately, the rather attractive hypothesis that “proper” diets would reduce colon cancer risk has not been substantiated when tested in prospective, interventional studies.⁸⁹ Dietary changes maintained over two to four years have failed to significantly reduce recurrent or incident adenomas in several studies that tested reductions in fat with increases in fiber, fruits, and vegetables; combinations of low-fat with wheat bran or beta-carotene supplements, or both; supplements of wheat bran fiber with vitamins C and E; and a complex supplement of calcium, vitamin C, vitamin E, and selenium.

Unlike these null studies, four classes of chemopreventive compounds have shown protective effects against colon adenomas or cancers: NSAIDs, calcium, hormone replacement therapy (HRT), and selenium. Of these, NSAIDs including aspirin are the most well-established agents. Greater than 90% of the more than 110 studies of various animal models and more than 35 epidemiologic studies confirm a significant reduction in colorectal adenomas, cancers, and cancer-associated mortality among users of aspirin or NSAIDs.⁸⁹ For example, the Nurses’ Health Study, a prospective cohort study of 27,077 nurses, found that regular use of aspirin was associated with a lower risk of developing colorectal adenomas.⁹⁰

Three prospective randomized trials investigating the use of aspirin to prevent colorectal adenomas have shown decreased rates of adenoma recurrence in the study groups compared with placebo. In one trial, which was terminated early because of significant results, 635 patients with a history of curative resection of colorectal cancer and randomized to 325 mg/day of aspirin (or placebo) were found to have a significant reduction in incident adenomas after a mean of 12.8 months.⁹¹ Another trial in patients with prior colorectal adenomas compared 81 mg/day or 325 mg/day of aspirin with placebo and found that the 81-mg dose reduced the relative risk of developing adenomas and advanced adenomas by 19% and 41%, respectively, with a similar, but not significant, trend found with the higher dose.⁹² The one-year results of a trial comparing two doses of lysine acetylsalicylate with placebo found a significant reduction in recurrence of adenomas larger than 5 mm in both treatment groups but a greater effect with the higher dose.⁹³ In contrast, in the Physicians’ Health Study, administration of aspirin 325 mg every other day over five years showed no reduction compared with placebo in advanced adenomas or colon cancers.⁹⁴ A meta-analysis pooled data from four randomized trials of aspirin at doses from 81 mg to 325 mg daily compared with placebo. More than 2900 patients with a history of an adenoma were included. At a median follow-up of 33 months, the rate of recurrent adenoma was 37% in the placebo group and 33% in the aspirin group. The absolute risk reduction with aspirin use was 6.7%.⁹⁵

NSAIDs act by inhibiting cyclooxygenase (COX)-1 and COX-2 enzymes, which thereby reduces cellular proliferation, enhances apoptosis, and reduces angiogenesis; other COX-independent effects also are operative. Based on these observations as well as studies showing that selective COX-2 inhibition reduces the number of adenomas in patients with FAP (see later), three large-scale prospective studies of COX-2 inhibitors (celecoxib and rofecoxib) have been undertaken.⁹⁶ The APPROVe, APC, and PreSap trials randomized more than 5000 patients with a history of adenomas to treatment with celecoxib or rofecoxib or placebo. In all three studies, celecoxib or rofecoxib was superior to placebo in preventing recurrent adenomas and, specifically, advanced adenomas; enthusiasm for these agents was tempered by the finding of increased cardiovascular adverse events in the COX-2 arms. The APPROVe study was terminated early and rofecoxib was withdrawn from the market, but the studies illustrated proof of principle that COX-2 inhibition leads to decreased polyp burden.⁹⁷

Calcium supplements have been shown in two randomized, placebo-controlled phase III studies to reduce adenoma recurrence by approximately 19% to 34%, with effects noticed even after one year of supplementation.^98,99 It has even been suggested that calcium supplements might have a more pronounced effect on advanced adenomas.¹⁰⁰ The mechanism for this protective effect likely is multifactorial, because calcium has been shown to decrease proliferation of colonic epithelial cells and to inhibit mucosal injury induced by bile acids and carcinogens in the fecal stream. Calcium can act by neutralizing the mutagenic effects of bile acids on the colonic mucosa or by directly inhibiting epithelial cell proliferation. Data suggest that the benefit of calcium in reducing the adenoma burden is maintained for at least five years after stopping calcium therapy.¹⁰¹

In a study investigating chemoprevention of skin cancer, selenium supplements were associated with a 58% reduction in colon cancer incidence as a secondary end-point.⁸⁹ This result has prompted additional trials to evaluate the effects of selenium on recurrent formation of adenoma. HRT has been associated in many studies with an approximately 20% reduction in colon cancer risk and a protective effect against colonic adenomas.⁸⁹ As with NSAIDs, the adverse side effects of HRT often outweigh the beneficial chemopreventive effects of these agents. A randomized, placebo-controlled trial in patients with a history of adenomas found that use of ursodeoxycholic acid at a dose of 8 to 10 mg/kg over three years did not lead to a statistically significant decrease in recurrent adenomas; however, there was a statistically significant 39% decrease in adenomas with high-grade dysplasia.¹⁰² Initial evidence from multiple cohort studies, including the Nurses’ Health Study and Canadian National Breast Cancer Screening study, suggested a 40% lower risk of colorectal adenomas in women with the highest levels of folate intake. A randomized, controlled trial of folate alone or in combination with aspirin, however, did not show a decreased risk of polyps or advanced neoplasia in the folate arms.⁹⁷ Currently, it is unclear if folate has a true benefit in reducing polyp burden.

Difluoromethylornithine (DFMO) suppresses polyamine levels in rectal mucosa likely by modulating ornithine decarboxylase, which can be abnormally expressed in the presence of an APC mutation. DFMO in combination with sulindac has been compared with placebo in patients with a history of an adenoma. At the three-year colonoscopic follow-up, the rate of adenomas was 41% in the placebo group and 12% in the treatment group. The rate of advanced adenoma was 8% in the placebo group and 0.7% in the treatment group.¹⁰³ DFMO is relatively well tolerated, although ototoxicity is a concern at higher doses. Further work is needed to confirm the efficacy of this agent.

Other agents that continue to be examined as chemoprevention agents include 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors, vitamin E, and mesalamine compounds. Overall, given the need for long-term therapy and the potential adverse events associated with many of these agents, they are likely best studied in patients at high risk for colorectal neoplasia.

Conditions Associated with Adenomatous Polyps

A variety of clinical circumstances have been associated with adenomatous polyps. Of the conditions discussed here, the predisposition to have or to develop adenomas is strongest for ureterosigmoidostomy, acromegaly, and Streptococcus bovis bacteremia. Patients with any of these three conditions should undergo a thorough colorectal examination and, in the former two conditions, periodic surveillance should be considered (although the frequency of such examinations is not well defined). As for the other conditions, either data are conflicting or the risk is not strong enough to recommend a policy of surveillance.

Fecal Occult Blood Testing

The actual frequency of bleeding from adenomas is difficult to determine. A significant adenoma (i.e., >1 cm or carcinoma in situ) is the cause in less than 10% of people who report frank rectal bleeding.¹²¹ In general, polyps smaller than 1 cm do not bleed. This dictum is supported by quantitative measurements of fecal blood loss in people with known adenomas; these measurements indicate that only patients with adenomas larger than 1.5 to 2.0 cm lose more than the usual amounts of blood, regardless of the location of the polyp within the colon.¹²² Thus, less than 40% of patients with known adenomas show positive fecal occult blood test (FOBT) results, and the higher rates occur primarily in patients with larger and distal polyps.¹²²

When asymptomatic persons older than 40 years of age undergo colon cancer screening with guaiac-based FOBTs, about 1% to 3% will have a positive result.¹²³ Upon colonoscopic evaluation, less than half of these people will have a colorectal neoplasm, and among the lesions found, adenomas outnumber carcinomas by 3 : 1. Thus the proportions of all positive guaiac tests attributable to colonic neoplasms (i.e., positive predictive values) are 30% to 35% for adenomas and 8% to 12% for cancer.¹²⁴ Despite the predominance of adenomas among lesions detected, 75% of adenomas still may be missed by guaiac testing (i.e., false-negative values) unless they are large or located in the distal portion of the colon. Positive test results for occult blood one to two years after a negative search will detect some of these missed polyps.¹²⁵ Because small polyps seldom bleed and their rate of detection by occult blood testing is low, sigmoidoscopy or colonoscopy has been recommended to complement FOBTs.

Fecal Immunochemical Testing

Guaiac-based testing relies on a peroxidase reaction; false positives can occur if the patient recently had ingested vegetable peroxidases or rare red meat, and false negatives can occur if they had ingested high doses of antioxidants such as vitamin C. To avoid some of these drawbacks, fecal immunochemical testing (FIT) was developed using an immunoassay to detect human globin in stool. The advantages of FIT are that it has increased specificity for human blood over guaiac-based testing, it does not require dietary restrictions, and it can be quantitative.

Two types of FIT studies have been performed: comparison studies to FOBT, in which all patients received a colonoscopy, and in vitro studies to determine at what level FIT can detect blood in various samples. Like the guaiac-based FOBT, FIT detects more cancers than adenomas. Most studies report sensitivities of 30% to 60% for detecting either cancer or advanced neoplasia.^126,127 Few studies have looked at the sensitivity for only adenomas. One large study analyzed seven different FIT tests in a screening population of more than 1300 patients. FITs were better than guaiac-based FOBTs.¹²⁸ The two best-performing assays had sensitivities of 25% to 27% for advanced adenomas. Specificity was high (97% for all adenomas and 93% for advanced adenomas). When quantitative FIT tests were set at detecting 50 ng Hb/mL in stool, the sensitivity for advanced adenomas was 55%, and higher rates of detection correlated with larger polyps and multiple adenomas.¹²⁹ At present, although FIT is an improvement over guaiac-based FOBT, it remains inadequate for adenoma screening and is best used as a screening test primarily for colorectal cancer.

Sigmoidoscopy

For several decades, sigmoidoscopy was the mainstay of endoscopic colon cancer screening. Rigid sigmoidoscopy would detect polyps (of all histologic types) in about 7% of asymptomatic persons older than 40 years,¹³⁰ whereas the flexible sigmoidoscope would find polyps in 10% to 15%, principally because a greater length of bowel could be examined.¹³¹ Screening sigmoidoscopy reduced mortality from distal rectosigmoid cancers by as much as 60% to 75%, in several retrospective case-control studies.⁶⁸ Conclusions regarding mortality reduction from three large-scale prospective studies of screening sigmoidoscopy—the PLCO Study in the United States,⁷⁰ the United Kingdom Flexible Sigmoidoscopy Screening Trial,¹³² and a “once-only” sigmoidoscopy trial in Italy¹³³—are awaited. Baseline findings of the U.K. study were that 12% and 0.3% of subjects between ages 55 and 64 years were found to have distal adenomas and distal cancers, respectively.¹³² The Italian study found an 11% prevalence of distal adenomas in the same age group. In the meantime, the increasing use of colonoscopy in the United States has resulted in a marked reduction in screening sigmoidoscopy examinations.

Barium Enema

Detecting adenomas by barium enema depends on their size. In the National Polyp Study, the detection rates of adenomas smaller than 6 mm, 6 to 10 mm, and larger than 10 mm were 32%, 53%, and 48%, respectively.¹³⁴ Common sources of error include inadequate cleansing of the colon, which contributes to the 5% to 10% false-positive rate, and diagnostic difficulty caused by the presence of diverticulosis, redundant bowel, or poor mucosal coating, which results in a 10% false-negative rate. Because of these issues, as well as the fact that barium enema never has been formally tested as a colon cancer screening tool, the use of barium enema for colon cancer screening has all but been abandoned in favor of colonoscopy or CT colonography (see later).

Colonoscopy

Colonoscopy is preferred to double-contrast barium enema examination for detecting adenomas because it has enhanced diagnostic accuracy as well as therapeutic capability. This diagnostic superiority has been demonstrated in studies of patients with known polyps^134,135 as well as in symptomatic patients who have negative findings on proctosigmoidoscopic and barium enema examinations.¹³⁶ Colonoscopy has become the preferred colon cancer screening test in many settings.¹³⁷ Despite its reputation as the gold standard for detecting adenomas, colonoscopy does have some limitations.¹³⁸ Colonoscopy fails to reach the cecum in up to 10% of cases, it usually requires sedating the patient, and it carries a higher cost than FOBT, FIT, or sigmoidoscopy. Colonoscopy also can miss neoplasms, especially those located at flexures or behind folds. In general, adenomas that are missed tend to be small. Studies using a tandem colonoscopy design demonstrate adenoma miss rates of 0% to 6%, 12% to 13%, and 15% to 27% for adenomas larger than 1 cm, between 6 and 9 mm, and less than 6 mm, respectively.¹³⁸ CT colonography reveals that colonoscopy can miss 12% to 17% of adenomas larger than 1 cm.¹³⁸

Given the concern about polyp miss rates, there has been increasing attention to quality measures for colonoscopy. Key measures of high-quality colonoscopy include adequacy of preparation, cecal intubation rate, withdrawal time, and adenoma detection rate. Inadequate preparation contributes to prolonged procedure times, decreased detection of lesions, and the need for repeat colonoscopy before recommended surveillance intervals.¹³⁹ Colonoscopy is not considered complete unless cecal intubation is accomplished. The majority of screening colonoscopy studies report a cecal intubation rate greater than 95%. Current guidelines suggest that cecal intubation rates should be greater than 90% for all colonoscopies and greater than 95% in screening colonoscopies.¹³⁹ Most screening colonoscopy studies report adenoma detection rates of 25% to 40%. Men have been consistently found to have a higher burden of adenomas than women. Current guidelines suggest that adenoma detection rates should be at least 15% in women and 25% in men.¹³⁹

A key factor in adenoma detection rate is colonoscopic withdrawal time. A large study examined the effect of withdrawal time in more than 7800 colonoscopies performed by 12 endoscopists. The adenoma detection rate was 28.3% among endoscopists with a withdrawal time of six minutes or more compared with 11.8% when the withdrawal time was shorter than six minutes. The respective detection rates for advanced adenomas were 6.4% and 2.6%¹⁴⁰; slower withdrawal time has been validated by the same investigators in a follow-up study of over 2300 colonoscopies.¹⁴¹ Current recommendations suggest that a withdrawal time of at least six minutes is necessary to maximize detection of adenomas. Continued emphasis on adhering to quality measures and detailed elucidation of the reasons lesions are missed can serve to improve colonoscopy further.

Computed Tomography Colonography

Also known as virtual colonoscopy, CT colonography involves scanning the colon with a helical or spiral CT scanner to produce both two-dimensional and three-dimensional images of the colon and rectum. Patients undergo a standard bowel preparation, and the colon is distended with air or carbon dioxide while images are taken with the patient in the supine and prone positions without sedation.

In this rapidly emerging field, a number of studies have compared the performance characteristics of CT colonography with standard optical colonoscopy for detecting polyps.¹⁴² Factors affecting detection rates include the polyp prevalence rate in the population being studied, the experience of the radiologist, and technical aspects including bowel preparation techniques, software, and the use of single-row or multi-row scanners. In high-prevalence populations that included symptomatic patients, the sensitivity for detecting polyps ranged from 29% to 59% for small polyps, 47% to 82% for medium polyps, and 63% to 92% for large polyps (Fig. 122-5). Studies of cohorts with low polyp prevalences fared less well, with sensitivities ranging from 32% to 58% and specificities of 90% for polyps larger than 6 mm in diameter.

Figure 122-5. Findings on computed tomographic colonography (CTC) in a man at average risk for colorectal cancer. A, An endoluminal three-dimensional CTC and colonoscopy image shows a 33-mm lobulated rectal polyp (arrow) and a 13-mm polyp (arrowhead) near the rectosigmoid junction. B and C, Two-dimensional coronal (B) and sagittal (C) CTC images confirm the presence and soft-tissue composition of the larger polyp (arrows). D, A digital photograph from same-day optical colonoscopy shows the endoscopic capture of the polyp immediately before resection. Pathologic evaluation revealed a large benign tubulovillous adenoma with high-grade dysplasia. The second lesion also had benign tubulovillous histologic characteristics but without high-grade dysplasia.

(From Kim DH, Pickhardt PJ, Taylor AJ, et al. CT colonography versus colonoscopy for the detection of advanced neoplasia. N Engl J Med 2007; 357:1403-12, with permission).

In the first large study to involve a pure asymptomatic screening population, CT colonography had a sensitivity of 86% for polyps 5 to 9 mm and 92% for polyps larger than 10 mm in the hands of highly skilled radiologists and with patients ingesting oral contrast prior to the study (fecal tagging) to facilitate distinguishing stool from mucosal abnormalities⁶⁰; a subsequent meta-analysis found very similar results.¹⁴³ A large multicenter study described a sensitivity of 90% for polyps larger than 10 mm and 78% for polyps between 6 mm and 9 mm.¹⁴⁴ In all studies, the detection of polyps smaller than 5 mm by CT colonography has been consistently low.

Several important questions remain unanswered regarding CT colonography. The management of polyps smaller than 10 mm is debated. The sensitivity of CT colonography for polyps smaller than 5 mm is quite low, and, even if polyps of this size are detected, they are often not reported. Various strategies have been proposed for polyps from 6 to 9 mm including referral for polypectomy and CT colonography surveillance. There is a small but definite risk of advanced neoplasia in these polyps. Given the paucity of natural history data on polyps of this size, the safety of CT colonography surveillance of these polyps is unclear.

Concerns have been raised about radiation exposure in healthy screening populations. It has been estimated that the lifetime risk of cancer in any organ associated with a screening CT at age 50 years is 0.14%.¹³⁷ Given that CT colonography images the entire abdomen, incidental extracolonic findings are often encountered; the rate is as high as 70%, and up to 11% of these incidental findings are clinically significant.¹³⁷ These incidental findings can prompt additional testing, cost, and anxiety. Lastly, questions remain about the frequency of CT colonography surveillance intervals.

Newer Methods for Detecting Adenomas

Based on our knowledge of molecular genetic alterations in colon carcinogenesis, a noninvasive method for detecting altered human DNA in stool has been developed.¹⁴⁵ In initial studies of patients with colon cancer, the sensitivity of the test was approximately 65%. In a large multicenter study of average-risk, asymptomatic persons older than 50 years, the sensitivity for colon cancer was 52% and specificity was 95%.⁶¹ This study demonstrated that stool DNA analysis was three to four times more sensitive than guaiac-based FOBT for detecting invasive cancers and adenomas with high-grade dysplasia. A second-generation stool DNA assay was found to have a sensitivity of 83% and a specificity of 82% for cancer.¹⁴⁶ A similar stool DNA test detected 46% of adenomas larger than 1 cm compared to a detection rate of 10% to 17% with guaiac-based FOBT.¹⁴⁷ Currently, stool DNA testing, like all noninvasive stool-based screening tests, has a low rate of adenoma detection. New technology, however, is already detecting 59% of advanced adenomas¹⁴⁸ and is expected to improve in the years ahead.

Natural History

Little is known about adenoma growth rate, primarily because polyps now are removed readily at endoscopy, thereby interrupting the natural history of their growth. Thus, our limited knowledge about polyp growth rate has been pieced together from two main types of studies: longitudinal follow-up studies on patients with nonresected (i.e., untreated) polyps and studies that compare the age distribution of people with adenomas with that of people with carcinomas.

Initial Treatment

If a polyp is detected by barium enema or CT colonography, a colonoscopy is recommended to afford the simultaneous opportunities to remove the polyp and search for synchronous neoplasms. There is some debate as to whether rectosigmoid adenomas found at sigmoidoscopy are markers for proximal colonic neoplasia and thereby prompt the need for a full colonoscopy. This controversy applies particularly to patients who have only a single small tubular adenoma with low-grade dysplasia. Some studies indicate that in this subset of patients, colonoscopy does not discover a substantial number of proximal AAPs or cancers,^165–167 nor is there a subsequent risk of developing proximal cancer among those who do not undergo colonoscopy.^168,169 In contrast, two screening colonoscopy studies suggest that the odds of finding advanced proximal neoplasia, even with a single small tubular adenoma, are 2.6-fold to 4-fold that of finding no distal pathology (Table 122-9).^58,75

Some authors have estimated that by not doing colonoscopy for a distal nonadvanced adenoma, 36% of advanced proximal neoplasms would be missed.¹⁷⁰ Furthermore, 52% of patients with advanced proximal neoplasia have no distal adenoma,⁵⁸ and 70% of proximal colon cancers lack a distal marker lesion.¹⁷¹ These observations support the use of colonoscopy as a primary screening modality. Because risk factors for finding advanced proximal neoplasia are increasing age,^58,59,75 a positive family history of colorectal cancer,^58,75 and male gender,⁵⁹ performing colonoscopy in these persons on the basis of finding a distal small tubular adenoma would seem prudent. A risk index that stratified patients into low, intermediate, or high risk based on the type and size of distal colonic polyps as well as patient age and male sex was able to predict the risk of proximal advanced colonic neoplasia.¹⁷²

There is little debate that if sigmoidoscopy reveals more than one adenoma or a distal AAP, a full colonoscopy is warranted because of the higher likelihood of finding synchronous proximal advanced neoplasms (see Table 122-9). Because negative biopsy results from a fractional sample of a polyp cannot possibly rule out cancer, total excision of a polyp is the only method of providing a thorough and accurate histologic diagnosis. For larger polyps, this can require piecemeal excision; for sessile growths, injection of saline into the polyp base can assist with complete endoscopic resection. After apparent complete removal of a large adenoma, it is advisable to repeat the colonoscopy in three to six months to document the completeness of the excision. If a polyp cannot be completely excised after two or three endoscopic sessions, surgical therapy is recommended.

Based on reasons listed in the previous discussion, screening colonoscopy is gaining acceptance in the United States as a primary screening tool, although many other countries with more limited resources do not yet subscribe to this philosophy. Two large-scale European trials are under way to evaluate the use of selective colonoscopy in patients based on the findings of a single sigmoidoscopy performed at about age 60 years.^132,133 After screening sigmoidoscopy, 5% to 8% of patients were referred for colonoscopy based on the finding of high-risk lesions or multiple polyps. Of these patients, 15% to 18% had proximal adenomas, 5% had advanced proximal adenomas, and fewer than 1% had cancer. Follow-up data regarding the incidence and mortality of colorectal cancer in these patients, as well as the results of the PLCO Cancer Screening Trial,⁷⁰ will provide additional data from prospective trials on the utility of screening sigmoidoscopy.

Management of the Malignant Polyp

The term malignant polyp refers to an adenoma in which a focus of carcinoma has invaded beyond the muscularis mucosae into the submucosa (see earlier). This term is not applied to adenomas containing either carcinoma in situ or intramucosal carcinoma because these lesions are not invasive and carry no metastatic potential. Rarely, polyps consist entirely of carcinoma. These polypoid carcinomas usually are considered a subset of malignant polyps, and they most likely represent a previous adenoma that has been completely replaced by carcinoma. Sometimes islands of benign adenomatous epithelium are found beneath the muscularis mucosae, and care must be taken not to mistake such pseudocarcinomatous invasion for true invasive carcinoma; this important distinction may be particularly difficult in the rare instance when the ectopic benign epithelium exhibits features of high-grade dysplasia. Ectopic benign epithelium is seen more often in larger pedunculated polyps, especially in the sigmoid colon. Because the distinction between carcinoma in situ and invasive carcinoma influence both management and prognosis, it is crucial that tissue be properly oriented for pathologic examination and that close communication takes place among endoscopist, surgeon, and pathologist.

Endoscopic removal of the vast majority of colorectal polyps raises two central and related questions: Is endoscopic polypectomy alone adequate therapy for the malignant polyp? If not, what features of the polyp can predict the presence of residual disease or subsequent recurrence? The answers are vital because they determine the decision for subsequent surgical resection of bowel.

Complete endoscopic removal of an adenoma with noninvasive carcinoma is curative. The decision regarding therapy becomes much more difficult when polyps contain invasive carcinoma. Although most of these lesions are treated adequately by endoscopic polypectomy, approximately 10% of patients experience an adverse outcome,¹⁷³ defined as residual cancer in the bowel wall or in lymph nodes either at the time of polypectomy or on follow-up examination. This rate of failure is comparable to that of the precolonoscopy era when adenomas were larger and patients underwent more complete surgical resections.¹⁷³ Because malignant polyps account for only 5% of all adenomas, and because not all patients who have had colonoscopic removal of malignant polyps have had surgical resection or are available for follow-up, conclusions often are based on small numbers.

Notwithstanding these limitations, combined experience has identified certain favorable and unfavorable histopathologic features of a colonoscopically resected malignant polyp that can be used to classify a patient as being at low or high risk for an adverse outcome (see later) (Table 122-10). If none of these unfavorable risk factors is found, the patient is considered to have been cured by the endoscopic polypectomy. This principle applies even to endoscopically removed polypoid carcinomas, which have been associated with a surprisingly good outcome when not complicated by any unfavorable histopathologic features. If one or more unfavorable features is found in a malignant polyp, the chance of an adverse outcome rises to about 10% to 25%.¹⁷⁴ In such cases, surgical resection usually should be performed, taking into account the risk of operative mortality in elderly patients with comorbid illnesses.

Table 122-10 Favorable and Unfavorable Features for Adverse Outcomes in Patients with Malignant Polyps

Several aspects of defining these risk factors demand close collaboration between endoscopists and pathologists. First, some malignant polyps contain only a small focus of poorly differentiated carcinoma, and therefore meticulous pathologic analysis is essential. Second, identification of vessel invasion by cancer cells in malignant polyps may be difficult, in which case special stains of vascular endothelium can be used for clarification. Third, although the polypectomy margin may be found on microscopy to contain cancer cells, some studies suggest that if the endoscopist believes that a complete excision was achieved, surgical resection might not be necessary because the electrocautery might have effectively destroyed residual tumor in the bowel wall.¹⁷⁵ Finally, when judging the adequacy of endoscopic polypectomy, the issue of submucosal invasion is important.

A pedunculated polyp differs anatomically from a sessile polyp in that the submucosa of the former projects up into the stalk, whereas the submucosa of the latter is in direct continuity with the bowel wall proper (Fig. 122-6). If cancer in a pedunculated polyp is confined to the submucosa of the stalk and all other histologic features are favorable, surgery is not indicated because the chance of an adverse outcome from endoscopic polypectomy is less than the operative mortality. Once the submucosa of the bowel wall is involved with cancer (a situation that occurs more readily in sessile polyps), the chance of an adverse outcome often outweighs the operative mortality, thereby justifying surgical resection. Furthermore, because endoscopic technique purposely avoids cutting deep into the bowel wall submucosa, there are few published examples of sessile lesions that have been completely excised endoscopically with clear margins and no other unfavorable features.

Figure 122-6. Carcinoma in situ versus invasive cancer. Carcinoma is considered intramucosal or carcinoma in situ, as indicated by the area labeled “1,” either in a pedunculated adenoma (left) or in a sessile adenoma (right). This lesion, as a rule, does not metastasize. Carcinoma in an adenomatous polyp is considered invasive when it breaches the muscularis mucosae, as indicated by the areas labeled “2.” Invasive cancer in a pedunculated polyp (left) is unlikely to metastasize, and it is managed differently from invasive cancer in a sessile polyp (right), which often requires surgical resection.

Deciding on the optimal plan of management after polypectomy involves weighing the risks of morbidity and mortality from potential residual or recurrent cancer against the risks of morbidity and mortality from a surgical attempt to cure any such residual disease or lymph node metastasis. A few general recommendations, however, can be offered. If excision of an adenoma is complete, endoscopic polypectomy alone is adequate therapy for adenomas that contain carcinoma in situ, pedunculated adenomas that harbor well-differentiated or moderately differentiated invasive carcinoma, and polypoid carcinomas. Resectional surgery is indicated for malignant polyps in which the invasive carcinoma is poorly differentiated, involves endothelium-lined channels (lymphatics, blood vessels), extends to or within 2 mm of the polypectomy margin, or involves the submucosa of the colonic wall (including all sessile adenomas). Studies have examined the depth of submucosal invasion as a predictor of lymph node metastasis. Two studies found the rate of lymph node metastasis was zero when depth of submucosal invasion was less than 1 mm in one study and 2 mm in the other.¹⁷⁶

Clearly, the ultimate plan of therapy must be individualized according to each patient’s medical condition. For most patients with malignant polyps, polypectomy without surgical resection seems adequate, with the caveat that the polyp was deemed to be completely resected endoscopically, and that postpolypectomy endoscopic surveillance will be incorporated into the patient’s health care regimen.

Postpolypectomy Management

Polyp Recurrence Rates

Although patients in whom a colorectal adenoma has been excised completely are likely to develop subsequent (metachronous) neoplasms, the frequency and time course of these occurrences are not well understood. In long-term retrospective studies, the cumulative risk of adenoma recurrence is nearly linear, being 20% at five years after polypectomy and rising to 50% after 15 years (Fig. 122-7).¹⁷⁷ Information on this subject is inexact because earlier studies were primarily retrospective, involved short follow-up periods, and differed in endoscopic indications. Moreover, recurrence rates tend to be somewhat inflated because lesions missed during the index colonoscopy might be considered recurrences. With these caveats in mind, it is estimated that one third of people who have undergone polypectomy develop recurrent adenomas.^178–180 The recurrence rate at one year is as low as 5% to 15%^177,180 but, more realistically, it ranges from 30% to 45% based on prospective colonoscopy studies.^{178,181–184} In the National Polyp Study, the overall adenoma recurrence rate was 42% for patients who underwent surveillance colonoscopy at one and three years after index polypectomy compared with 32% in the group that was examined only at three years.⁴⁶ There is general consensus that recurrent adenomas typically are smaller and less likely to harbor advanced pathology than are index adenomas.

Figure 122-7. Risk of recurrent adenomas and of colon cancers after removal of adenomas. The risk of developing recurrent adenomas (orange, yellow) or cancer (blue, green) is higher in patients who initially underwent removal of multiple (>1) adenomas (orange, blue) than removal of a single adenoma (yellow, green).

(Modified from Morson BC, Bussey HJR. Magnitude of risk for cancer in patients with colorectal adenomas. Br J Surg 1985; 72[Suppl]:S23.)

Can histopathologic features of adenomas at the time of the index polypectomy be used to help predict recurrence of adenomas? Virtually all studies agree that the presence of multiple index adenomas is an important predictor of subsequent recurrence of adenoma (and carcinoma) (see Fig. 122-7).^{177,178,180,185,186} This dictum applies despite negative findings on colonoscopy one year after polypectomy.¹⁸¹ Some studies suggest that polyp size greater than 1 cm,^179,180,186 severe dysplasia,¹⁸⁷ villous histology,^180,185 and older age^178,180 also are risk factors for adenoma recurrence, but the independent relative importance of each of these factors is uncertain.

It can be argued that the most clinically important recurrence is an AAP. Some studies have reported a 6.3% to 7.0% recurrence rate for AAP over a four-year follow-up period.¹⁸³ In the National Polyp Study, the recurrence rate of advanced adenomas was 3.3%, regardless of whether patients underwent colonoscopy at one and three years, or only three years, after polypectomy. The cumulative incidence of AAP was 4% at three years and 8% at six years of follow-up. Independent predictors for AAP at follow-up included more than three adenomas at the index colonoscopy, and age older than 60 years at the time of adenoma diagnosis, with a parent who had colorectal cancer.¹⁸⁸ In the presence of these two risk factors, cumulative AAP recurrence rates rose to 10% at three years and 20% at six years of follow-up. The persons at lowest risk for developing AAP were those with a single adenoma diagnosed when they were younger than 60 years. Another colonoscopic follow-up study reported that multiple adenomas at the index colonoscopy increased the risk of AAP, but it also found that adenoma size greater than 1 cm and proximal location were additional risk factors.¹⁸⁶ A pooled analysis of eight prospective trials with more than 9000 patients found that more than five adenomas, size greater than 20 mm, proximal location, and the presence of villous features were the strongest predictors of recurrent advanced adenomas.¹⁸⁹ Interestingly, the presence of high-grade dysplasia was not associated with recurrent adenomas.

Histologic Features

Hyperplastic polyps typically are small sessile lesions that are grossly indistinguishable from small adenomatous polyps. Microscopically, the colonic crypts are elongated and the epithelial cells assume a characteristic papillary configuration (Fig. 122-8). The epithelium is made up of well-differentiated goblet and absorptive cells. The cytologic atypia that is characteristic of adenomatous polyps is not seen. Mitoses and DNA synthesis are limited to the base of the crypts, and orderly cell maturation is preserved. The epithelial cell and attendant pericryptal sheath fibroblast make up an epithelial-mesenchymal unit that migrates up the colonic crypt. In contrast to adenomatous polyps, in which the epithelium and fibroblast appear to be immature, this tissue is more differentiated, and abundant collagen is synthesized in the basement membrane.¹⁹⁸ It is thought that the migration of epithelial cells up the colonic crypt is slow and that hyperplastic polyps develop from the failure of mature cells to detach normally.

Figure 122-8. Hyperplastic polyp. This high-power photomicrograph demonstrates the crypts of a hyperplastic polyp, consisting of elongated epithelial cells with nuclei that retain their basal orientation and demonstrate no atypia. The surface of the crypts assumes a serrated appearance.

(Courtesy Noam Harpaz, MD, PhD, New York.)

For the most part, true sporadic hyperplastic polyps are considered to have little if any intrinsic malignant potential; however, it is important to consider that hyperplastic polyps and neoplastic lesions can appear in the same colon, suggesting that the two may be pathogenetically related. Indeed, a germline mutation of the APC gene has been associated with an unusually large number of colorectal hyperplastic polyps in association with adenomas.¹⁹⁹ Coexisting hyperplastic and adenomatous polyps also are common in the setting of a strong family history of colorectal cancer, including HNPCC.²⁰⁰ It appears, however, that in patients with sporadic adenomas, the coexistence of hyperplastic polyps does not confer an increased risk of recurrent adenomas during a three-year follow-up period.²⁰¹

Prevalence

The prevalence of hyperplastic polyps is not known with precision, but these growths are common. In colonoscopic examinations of asymptomatic patients older than 50 years, hyperplastic polyps were found in 9% to 10%,⁶⁶ although this frequency was higher (30% to 31%) among male veterans.²⁰² Sigmoidoscopic screening of asymptomatic relatives of adenoma-prone kindreds revealed 26% with hyperplastic polyps, a prevalence that was essentially identical (28%) to that of asymptomatic spouse controls.²⁰³ Autopsy data report a prevalence rate of 20% to 35%.^8,12

The incidence of hyperplastic polyps depends largely on the site of the colon being examined and on the age of the patient. Autopsy studies repeatedly observe a distal predominance of hyperplastic polyps.^8,11,12 Of course, sigmoidoscopic studies, which focus on the distal colon and rectum, find many hyperplastic polyps, but even screening colonoscopy studies detect more hyperplastic polyps in the distal than in the proximal colon.⁶² Among all diminutive polyps (<5 mm) removed during colonoscopy, hyperplastic polyps outnumber adenomatous polyps in the rectum and sigmoid, and adenomas predominate in the remainder of the colon.^15,204 The prevalence of hyperplastic polyps increases with age.^10,53 There also is an association between prevalences of hyperplastic polyps and colon cancer (Table 122-12), although this correlation is not as firm as the association between adenomas and colon cancers, nor does it necessarily imply any premalignant potential of the hyperplastic polyp itself. Data suggest that current smoking is a risk factor for hyperplastic polyps.⁸⁸

Treatment

Hyperplastic polyps remain small, usually are sessile, and seldom, if ever, cause symptoms. Inasmuch as they are not likely to give rise to cancer, little is gained by removing them, but because they cannot be distinguished from neoplastic or serrated polyps simply by gross examination, they usually are removed. Given their usual predominance in the distal colorectum, finding hyperplastic polyps in this location is not an alarming finding, particularly in the elderly. Because sigmoidoscopic findings sometimes are used to decide the need for a full colonoscopy, the importance of the distal hyperplastic polyp as a marker of proximal neoplasia has been questioned. A systematic review of the subject found that in asymptomatic persons, a distal hyperplastic polyp is associated with a 21% to 25% risk for any proximal neoplasia and a 4% to 5% risk of advanced proximal neoplasia, thus justifying colonoscopy.²⁰⁵ Thus, although most patients with a distal hyperplastic polyp likely will undergo colonoscopy, current guidelines suggest that if no adenomas are found, such patients can subsequently be followed at intervals similar to those without any polyps.¹⁹⁷

Serrated Polyps and the Serrated Neoplasia Pathway

Evidence is mounting to support the concept of a distinct serrated neoplasia pathway.^206,207 Polyps that display features of both hyperplastic and adenomatous transformation have been described; these mixed hyperplastic-adenomatous serrated polyps account for about 13% of hyperplastic polyps.²⁰⁸ It has become clear that polyps with a serrated phenotype, instead of representing simple hyperplastic polyps, might represent more clinically significant lesions. Serrated polyps are now classified into hyperplastic polyps (with two subtypes: goblet cell and microvesicular), and two types of serrated adenomas: traditional serrated adenoma (TSA) and sessile serrated adenoma (SSA, sometimes referred to as sessile serrated polyp) (Fig. 122-9).²⁰⁹

Figure 122-9. A, Sessile serrated polyp (sessile serrated adenoma). The crypts at the base of the polyp are broad and flattened, with nuclear pleomorphism and prominent nucleoli. The mid and upper portion of the same crypts show maturation toward more normal appearing nuclei but with a serrated appearance on the luminal surface and dystrophic goblet cells. B, Traditional serrated adenoma. The long fronds of this adenoma are lined by dysplastic epithelial cells characterized by nuclear pleomorphism interspersed with cells that have dystrophic goblet cell vacuoles.

(Courtesy Noam Harpaz, MD, PhD, New York.)

SSAs are considered the precursor lesions for the large hyperplastic polyps found in the proximal colon of patients with hyperplastic polyposis.²⁰⁹ When sporadic, SSAs are found more commonly in the proximal colon. By contrast, TSAs look and behave like conventional adenomas: They are often pedunculated, have unequivocal adenomatous dysplasia (albeit with crypt serration), and are found more often in the distal colon. At the molecular level, SSAs often have extensive BRAF mutations and extensive DNA methylation, whereas TSAs usually do not have BRAF mutations and have only infrequent mutations of APC, TP53, and MSI, but they can have K-ras mutations.²⁰⁹

Serrated adenomas are much less common than hyperplastic polyps, accounting for less than 1% of all polyps and between 1% and 11% of adenomas.²⁰⁷ The prevalence of high-grade dysplasia and cancer in these lesions, however, may be as high as 5% to 16%.²⁰⁷ The risk of colorectal cancer may be as high as 1 in 17 in a patient with a proximal sessile serrated adenoma. There are no guidelines for the management of SSAs. It is generally recommended that surveillance intervals for serrated adenomas, although not yet defined, should follow that of other adenomas, bearing in mind their possibly more rapid progression.²⁰⁹