INTRODUCTION — Patients with inflammatory bowel disease (IBD) are at increased risk for colorectal cancer [1-8]. The risk is related to the duration and anatomic extent of the disease. The mortality in patients diagnosed with colorectal cancer in the setting of IBD is higher than for sporadic colorectal cancer [7].

Although no large controlled trials have proven that surveillance reduces mortality, surveillance is widely practiced and recommended by the American Gastroenterological Association, the American College of Gastroenterology [9], and the American Society for Gastrointestinal Endoscopy (the ASGE recommendations are available online at: www.asge.org). The AGA guideline for colorectal cancer screening and surveillance [10], as well as other AGA guidelines, can be accessed through the AGA web site at http://www.gastro.org/wmspage.cfm?parm1=4453.

The epidemiology and pathology of colon cancer in IBD and the evidence supporting a role for cancer surveillance will be reviewed here. Newer techniques for cancer surveillance will also be discussed, although there is insufficient evidence to support their widespread use.

EPIDEMIOLOGY — The risk of colorectal cancer (CRC) is increased in ulcerative colitis (UC) and Crohn's disease (CD). However, much more is known about the risk in UC. A population-based study from Sweden estimated that the overall risk of colorectal cancer in IBD was 95 cases per 100,000 population [11]. The incidence of colorectal cancer showed a trend toward a decline from the 1960s to 2004, while colorectal cancer mortality decreased significantly. (See "Colorectal cancer: Epidemiology, risk factors, and protective factors".)

Ulcerative colitis — The risk of CRC in UC depends upon the duration and extent of disease [1-3,12-14]. A population-based study in the United States estimated the risk was significantly increased in those with extensive or pancolitis (standardized incidence ratio (SIR) 2.4, 95% CI 0.6-6.0). In addition, patients with UC complicating primary sclerosing cholangitis (PSC) may be at increased risk for CRC compared to those without PSC. A case control study (in which cases and controls were matched for the extent and duration of disease) found that the risk of CRC was reduced with use of anti-inflammatory agents (including aspirin, nonsteroidal anti-inflammatory drugs and 5-aminosalicylic acid agents) and by surveillance colonoscopy while it was increased in patients with a history of postinflammatory pseudopolyps [15].

Pancolitis — Patients with disease extending to the hepatic flexure or more proximally have the greatest risk of CRC. Compared to an age-matched population, the risk begins to increase 8 to 10 years following the onset of symptoms [16-18]. The approximate cumulative incidence of CRC is 5 to 10 percent after 20 years and 12 to 20 percent after 30 years of disease [4,16,19-21]. Lower rates of CRC have also been found [22,23]. In a population-based study from Copenhagen, the risk of CRC was not different from the general population, a finding that the authors hypothesize may have been due to an active surgical approach in medical treatment failures and long-term use of 5-ASA drugs [23]. The highest cancer risks have been reported by medical centers that predominantly receive referral patients who may have more severe or longstanding disease [12].

In one series, the absolute risk of CRC in patients with pancolitis was 30 percent after 35 years of disease [4]. The risk was increased in those with the onset of symptoms prior to age 15. However, in other reports, the age of onset of colitis did not increase the risk of CRC after adjusting for the longer period of time that young patients were at risk and the extent of the disease [8].

One study suggested that the presence of "backwash ileitis" (in which mucosal inflammation involves the terminal ileum) may be an independent risk factor for CRC [24]. However, other studies have not confirmed this association [25,26]. (See "Endoscopic diagnosis of inflammatory bowel disease", section on 'Backwash ileitis',The severity of inflammation may also be an important marker of risk. A case-control study found a significant correlation between the severity of inflammation as assessed by histology and the risk of colorectal neoplasia (OR 4.7) [25]. A similar degree of risk was found in a later cohort study [27].

Left-sided colitis — Most studies have found that the risk of CRC increases after 15 to 20 years (approximately one decade later than in pancolitis) in patients with colitis confined to the left colon (ie, distal to the splenic flexure) [28]. However, rates of CRC and dysplasia similar to those seen in patients with pancolitis have been described [29].

Proctitis — Patients with ulcerative proctitis and proctosigmoiditis are probably not at increased risk for CRC [30].

Primary sclerosing cholangitis — An increased risk of CRC has been observed in patients with UC complicated by PSC. Cancer in patients with PSC was more likely to be in the right colon, suggesting a possible role of bile acids in oncogenesis (a hypothesis supported by studies showing a protective effect of ursodeoxycholic acid). (See "Colorectal cancer and primary sclerosing cholangitis".)

Crohn's disease — The risk of CRC in longstanding Crohn's disease involving the colon is probably comparable to ulcerative colitis [5-8,31-34]. In a population-based study from Sweden, for example, the relative risk of colon cancer was 2.5 overall in patients with CD and 5.6 in those with disease restricted to the colon [8]. The relative risk was even greater in patients who were less than 30 at the time of diagnosis (relative risk 21 compared to those diagnosed after age 30). Similar findings have been reported in other studies [5,35,36].

However, not all studies reached these conclusions and thus the magnitude of risk in patients with CD remains unsettled. Earlier reports found no increased risk of CRC in patients with CD, but most did not adjust for the duration and extent of the disease [37-40]. On the other hand, a population-based study from the United States (which did consider disease duration) found a significantly increased risk of small bowel but not colorectal cancer [14].

CRC in CD is observed in a similar time frame as in UC [41,42]. This was illustrated in one series that included 80 patients with CRC complicating UC or CD [41]. The median duration of disease prior to the diagnosis of CRC was comparable for CD and UC (15 and 18 years, respectively). The median age at diagnosis of CRC was 55 in CD and 43 in UC. Another report suggested the benefit from regular surveillance in patients with Crohn's colitis of at least seven years duration (and affecting at least one-third of the colon) was similar to that of ulcerative colitis [34].

Despite the increased risk, the absolute number of patients at risk is relatively small. Many patients with extensive colitis undergo colectomy early in the course of disease to relieve persistent symptoms [42]. Furthermore, sporadic CRC is far more common than CRC arising in the setting of IBD. As a result, CD accounts for a small proportion of CRC for the overall population (graph 1).

PATHOGENESIS — The pathogenesis of colon cancer in IBD is poorly understood. However, several lines of evidence suggest that the pathobiology is different than for sporadic CRC. (See "Molecular genetics of colorectal cancer".)

• The mean age of developing CRC in the setting of IBD is lower than for sporadic CRC (40 to 50 versus 60 years).
• Dysplasia in UC is preceded by a long history of chronic inflammation and can be found at distant sites from the cancer. In contrast, dysplasia in sporadic colon cancer is usually associated with a discrete polyp without inflammation.
• Mutations in the ras protooncogene are present in 40 to 60 percent of sporadic colon cancers and are probably an early event; in contrast, these mutations are less frequently observed in cancer associated with UC, and are probably a late event [43-45].
• Loss of heterozygosity for the p53 gene and src activation occur earlier in cancers associated with IBD than in sporadic CRC [43,45-47]. Src activity in UC correlates with the degree of dysplasia [45].
• Abnormalities of the p53 locus are absent in nondysplastic mucosa of patients with sporadic colon cancer [48]. In contrast, nondysplastic mucosa in UC frequently has aneuploid DNA content and may show clones of cells with loss of heterozygosity of the p53 gene [48].

PATHOLOGY — CRC complicating inflammatory bowel disease may appear polypoid, nodular, ulcerated, or plaque-like [49]. As in sporadic colorectal cancer, most lesions in the colon are adenocarcinomas [50]. However, poorly differentiated, anaplastic, and mucinous carcinomas are more common in colitis-associated CRC than in sporadic CRC [7,45,51,52]. (See "Pathology and prognostic determinants of colorectal cancer".)

Location — CRC associated with UC is most common in the rectum and sigmoid colon (picture 1) [41]. In contrast, CRC associated with CD is evenly distributed between the right colon and rectosigmoid. Cancer always occurs in areas involved in chronic inflammation [41]. Synchronous tumors are much more common in IBD than in sporadic CRC (12 versus 3 to 5 percent) [45]. This can include the colon, rectum, anus, and internal or external fistulous tracts [50].

Dysplasia — It is generally accepted that CRC in IBD is preceded by dysplasia. Thus, dysplastic epithelium may be a marker for coexisting malignancy, and provides the rationale for surveillance.

A uniform terminology for dysplasia in IBD has been proposed [53]. The classification categorizes histology as:

• Negative
• Indefinite
• Positive (with subgroups of low grade and high grade dysplasia)

Dysplastic areas are often difficult to recognize on endoscopy. They may appear as flat or only slightly elevated above the level of the mucosa. Dysplasia may also occur within or near raised plaque-like lesions, nodules, polyps, or masses (referred to as dysplasia-associated lesion or mass [DALM], (see 'Dysplasia associated lesion or mass' below.

Dysplasia may be difficult to distinguish from epithelial regeneration due to inflammation on histologic sections [54]. As a result, the presence of dysplasia should be confirmed by an experienced GI pathologist. The criteria for dysplasia stress the uniform clonal nature of dysplastic changes, which affects all parts of the crypt and surface epithelium. In contrast, regenerative changes are usually most prominent at the bases of the crypts and show evidence of surface maturation [53]. One study found that immunostaining for alpha-methylocyl-Coa-racemase (AMACR), a mitochondrial and peroxisomal enzyme overexpressed in many types of cancers, was highly specific for detecting dysplasia and distinguishing regenerating epithelium from true dysplasia [55].

Other architectural and cytological abnormalities seen in dysplastic epithelium include [53,56-58]:

• Increased mitoses (typical and atypical)
• Back to back gland pattern
• Increased nuclear size
• Variation in the size and shape of nuclei (pleomorphism)
• Increased nuclear/cytoplasmic ratio
• Altered nuclear polarity
• Hyperchromaticity
• Lack of surface maturation
• Stratification of nuclei

Dysplasia associated lesion or mass — Patients with ulcerative colitis have the same risk for sporadic adenomas as the general population. However, some patients with UC who are found to have dysplasia associated with a non-adenoma-like lesion or mass (DALM) may have an underlying invasive carcinoma, which may not be detectable by endoscopic biopsy and warrants colectomy. The risk of cancer associated with DALM was illustrated in a series that included 12 patients with DALM who underwent resection [59]. The endoscopic appearance of the lesions were:

• Single polypoid mass (five patients)
• Plaque-like lesion (two patients)
• Multiple polyps (five patients)

Invasive carcinoma was found on resection specimens in seven patients (60 percent), including all five who had a single polypoid mass. Malignancy was not detected in any of the patients prior to resection despite multiple biopsies.

Whereas non-adenoma-like DALMs is an indication for colectomy, sporadic adenomas may be removed endoscopically even if they occur in an area histologically involved with colitis. As a result, non-adenoma-like DALMs and sporadic adenomas need to be distinguished to prevent unnecessary colectomy. A number of clinical, histologic, and molecular features have been studied to help make this distinction:

• Patients with non-adenoma-like DALM are more likely to be younger and have a longer duration of disease, more extensive disease, and larger lesions (1.8 versus 0.5 cm in one study [60]).
• Lesions that appear endoscopically as adenomas (pedunculated or sessile) rather than having other characteristics (such as flat, ulcerated, or plaque-like appearance) even if within an area of histologic colitis have a favorable prognosis with endoscopic removal and close follow-up [61-64].

In addition to the clinical and histologic features described above, several molecular markers have been proposed to distinguish DALM from sporadic adenomas. Two of these are beta catenin and P53 [65,66]. Beta catenin is a cell membrane protein that accumulates more frequently in nuclei of cells within sporadic colon cancer compared to DALM. In comparison, mutations with P53 (a tumor suppressor gene) occur more frequently in DALM than sporadic adenomas. In one study, for example, these markers were compared in 17 adenoma-like DALM specimens and 21 sporadic adenomas in patients with UC [67]. Positive staining for P53 was more common in DALM (29 versus 5 percent), whereas positive staining for beta catenin was more common in sporadic adenomas (40 versus 8 percent).

Long-term follow-up suggests that polypectomy with complete excision and continued surveillance provides adequate treatment of patients with an adenoma-like DALM. Of 34 patients with ulcerative colitis, 24 were diagnosed with an adenoma-like DALM and 10 with a coincidental sporadic adenoma, 28 of whom were treated by polypectomy and continued surveillance and 6 by colonic resection [62,68]. Outcomes were compared with 49 non-UC patients who were treated similarly for sporadic adenomas. Mean follow-up ranged from 72 to 82 months. Overall, 59 percent of the UC patients developed at least one further adenoma-like DALM. However, there were no significant differences in the incidence of polyp formation on follow-up between UC patients with an adenoma-like DALM and UC patients with a sporadic adenoma or non-UC sporadic adenoma patients.

It is important to acknowledge that the study was small (and therefore potentially underpowered to detect differences) and that these patients had been carefully selected based upon their endoscopic findings. Whether a similar degree of certainty in distinguishing adenoma-like DALMs from non-adenoma-like DALMs can be generalized to other clinical settings remains to be tested in a prospective fashion. Our approach to patients with ulcerative colitis who are found to have a polyp is discussed below (see 'Summary and recommendations' below.

Inflammatory pseudopolyps — Inflammatory pseudopolyps are irregularly shaped islands of residual intact colonic mucosa that are the result of the mucosal ulceration and regeneration that occurs in inflammatory bowel disease (IBD). These polyps are typically multiple and scattered throughout the colitic region of the colon (picture 2). They are not dysplastic and are not a risk factor for colon cancer. However, their presence can complicate the recognition of true adenomas and DALM. Inflammatory pseudopolyps can be recognized by their histologic features; thus a biopsy can help make the distinction in unclear cases.

PREDICTIVE VALUE OF DYSPLASIA — The predictive value of dysplasia has been best studied in UC. There is general agreement that the finding of high-grade dysplasia warrants prompt colectomy because of the high rate of synchronous colorectal cancer [69]. More controversial is the proper course of action in patients found to have low-grade dysplasia, although we currently recommend colectomy for patients with either high- or low-grade dysplasia [69-73].

Although dysplasia is a marker for future or concurrent malignancy, it can also regress or remain stable for long periods [56,74]. There is currently no reliable way to predict which of these outcomes is most likely. Furthermore, most patients with dysplasia do not have cancer [21,75,76], and dysplasia is absent in the colonic regions distant from the malignancy in 25 to 30 percent of patients [77-79].

To further complicate matters, the diagnosis of dysplasia is not straightforward, even when using specific criteria for diagnosis. Interpretation of biopsy samples may be confounded by interobserver variation in the recognition and grading of dysplasia [77,80,81]. As a result, it is important that difficult cases be reviewed by a pathologist who is experienced in this area.

The following studies illustrate the range of findings in some of the largest series:

• In one report based upon 21 years of experience with 332 patients in a surveillance program, the five-year predictive value of low-grade dysplasia for either CRC or high grade dysplasia was 54 percent [77].
• Another series included 590 patients who underwent total proctocolectomy or restorative proctocolectomy for chronic ulcerative colitis [82]. Colonoscopy had been performed in 160 patients within one year before surgery. Colorectal cancer was found significantly more often in resection specimens that contained dysplastic changes (43 versus 1 percent). Dysplasia diagnosed during preoperative colonoscopy had a sensitivity of 81 percent and a specificity of 79 percent for the diagnosis of synchronous cancer. The positive predictive value for cancer of a finding of preoperative dysplasia of any grade was 50 percent.
• A third report focused on 160 patients (seen at one institution between 1978 and 1990) with longstanding, extensive ulcerative colitis of whom 40 developed low-grade dysplasia during surveillance [71]. Of 128 patients who were available for follow-up, 29 had low-grade dysplasia. After 10 years, high grade dysplasia or colorectal cancer developed in three of these patients (10 percent) compared with four "controls" (4 percent) without low-grade dysplasia. The authors concluded that the diagnosis of low-grade dysplasia was not sufficiently reliable to justify prophylactic colectomy.
• Strikingly different conclusions were reached in another study that included 46 patients with low-grade dysplasia [83]. Among patients who elected not to undergo immediate colectomy, five (20 percent) were ultimately diagnosed with colorectal cancer. The rate of progression to high-grade dysplasia or cancer was estimated to be 53 percent at five years [83]. No clinical features predicted progression to advanced neoplasia. An important additional finding was that the cancers in those who underwent immediate resection were earlier stage than those in whom resection had been delayed. Furthermore, of five patients who ultimately developed colorectal cancer during surveillance, none had high-grade dysplasia detected prior to the diagnosis. The authors recommended that early colectomy should be recommended to patients with low-grade dysplasia.

The association between dysplasia and CRC in CD appears to be similar to that in UC. Dysplasia is present in 83 to 100 percent of patients diagnosed with CRC [7,51,52], and dysplasia distant from the cancer has been found in 23 to 70 percent [51,52,84,85]. In contrast, dysplasia is much less common in colectomy specimens of patients with CD without CRC, occurring in only 2 percent of specimens in one series [86].

Cost-effectiveness colectomy versus surveillance in dysplasia — A decision-analysis estimated the cost-effectiveness of enhanced surveillance (colonoscopy at 3, 6, and 12 months then annually) versus immediate colectomy in a simulated cohort of patients found to have unifocal, flat, low-grade dysplasia on initial surveillance colonoscopy [87]. Immediate colectomy was associated with higher quality-adjusted life years and lower costs. Enhanced surveillance was cost-effective only when considering health utilities of living with a proctocolectomy that were generally lower than what was considered to be likely.

SURVEILLANCE AND PROGNOSIS OF CANCER — The relatively young age at which CRC develops in patients with IBD have led to attempts to stratify patients at greatest risk. Because of the association of CRC with dysplasia in IBD, the goal of most surveillance programs has been the detection of dysplasia [29,41,88,89]. There are conflicting data as to whether patients with colon cancer associated with IBD have a worse prognosis compared with patients with sporadic colon cancer, although more recent data suggest that it is similar (or even better) when matched for stage [90-99]. However, patients diagnosed with CRC after they have developed symptoms (such as abdominal pain or bleeding) have a worse prognosis than those detected during surveillance [100].

Dysplasia surveillance — The optimal surveillance strategy remains controversial [101]. The debate revolves around the sensitivity of the detection system, the predictive value of dysplasia for assessing the risk of colorectal cancer (see 'Predictive value of dysplasia' above, and the costs [17,18,70,74,102].

A systematic review found no clear evidence that surveillance colonoscopy prolongs survival in patients with extensive colitis [103]. However, there was evidence that cancers tended to be detected at earlier stages as well as indirect evidence that surveillance was likely to be effective at reducing the risk of death from IBD-associated colorectal cancer and was cost-effective. In an earlier review of 10 studies of dysplasia surveillance that included 1225 patients with UC, the likelihood of finding cancer in patients found to have high or low grade dysplasia was 42 and 19 percent, respectively [69].

Another study compared 19 patients with CRC associated with IBD who had undergone surveillance to 22 patients with CRC who had not undergone surveillance [100]. Patients who had undergone surveillance had earlier stage carcinomas and significantly longer five-year survival (77 versus 36 percent). Similar results have been found in other studies [104]. However, advanced CRC can occur despite surveillance [13,74,105], and the effectiveness of surveillance programs has been limited by patient compliance [17,18] and unwillingness to undergo prophylactic colectomy based upon the finding of dysplasia [100].

In summary, the data supporting the effectiveness of surveillance of dysplasia in patients with IBD are not uniform but suggest a reduction in mortality from CRC in patients undergoing surveillance who are willing to undergo prophylactic colectomy. As a result, the balance of evidence supports surveillance for dysplasia. This recommendation may eventually change with the development of new biologic markers of cancer risk.

Biopsy protocol — Multiple biopsies are required to adequately sample the colon. In one report, it was estimated that 33 biopsy specimens from a single colonoscopy would be required to detect dysplasia with 90 percent confidence [106]. In addition, areas of mucosal irregularity should be biopsied [107]. Guidelines for biopsies have been proposed in the society guidelines summarized below. (See 'Recommendations of major societies' below.)

A study comparing eight paired biopsy specimens from the rectosigmoid obtained by either jumbo or standard large-capacity forceps concluded that the jumbo forceps were superior for obtaining diagnostically adequate surveillance biopsy specimens [108]. Thus, use of jumbo forceps has the potential to improve the dysplasia detection rate.

New markers — Because of the limitations of surveillance strategies based upon the detection of dysplasia, newer techniques involving the detection of alterations in mucosal antigens and genetic abnormalities are being investigated [48,109]. As examples:

• Sialosyl-Tn is a mucin-associated carbohydrate antigen that may presage the development of dysplasia and CRC by several years [45,110].
• Allelic deletions and point mutations of tumor suppressor genes, such as p53, Rb, mcc and APC have been found in dysplastic and cancerous UC [45,48,111].

It is likely that changes in nuclear DNA content are detectable earlier than histologic signs of premalignancy. As a result, detection of DNA aneuploidy by flow cytometry of mucosal specimens in UC patients may be more objective and less sensitive to assessment error than dysplasia [109]. DNA aneuploidy detected by these techniques has been correlated with dysplasia and carcinoma [80,112,113]. However, as with dysplasia, the predictive value of aneuploidy is uncertain. Neoplasms can arise without preexisting aneuploidy, and conversely, aneuploidy may exist for many years without progression to malignancy [45].

At the present time, these newer techniques have been insufficiently standardized and validated in the clinical setting to be recommended for general use. Until further data are available, evaluation for dysplasia remains the standard for cancer surveillance programs in IBD.

Chromoendoscopy — Chromoendoscopy involves the topical application of stains or pigments to improve tissue localization, characterization, or diagnosis during endoscopy. A least one controlled trial suggested that staining with methylene blue enhanced the ability to detect the extent of inflammatory changes and identify intraepithelial neoplasia in patients with ulcerative colitis undergoing surveillance [114]. Other studies have suggested that staining with indigo carmine permitted better detection of dysplasia [115,116]. The clinical implications and generalizability of these findings require further clarification. Other staining techniques designed to augment visualization of dysplastic changes have also been described. (See "Chromoendoscopy".)

CHEMOPROPHYLAXIS — There is presently no drug that is generally considered to be effective in the primary prevention of colorectal cancer in patients with inflammatory bowel disease. Thus, chemoprophylaxis is not recommended. However, several drugs have been studied in the non IBD population (particularly nonsteroidal antiinflammatory drugs and calcium) and some have also been evaluated in IBD. (See "Colorectal cancer: Epidemiology, risk factors, and protective factors" and "NSAIDs (including aspirin): Role in prevention of colorectal cancer".) The latter include folic acid, fatty acids, and ursodeoxycholic acid [117-121]. (See "Colorectal cancer and primary sclerosing cholangitis".) Several lines of evidence also suggest that the risk may be reduced with aminosalicylates, but studies have been mixed and thus magnitude of benefit is unclear. (See "Sulfasalazine and 5-aminosalicylates in the treatment of inflammatory bowel disease".) By contrast, 6-MP or azathioprine do not appear to influence the rate of neoplastic transformation [122].

RECOMMENDATIONS OF MAJOR SOCIETIES — Optimal surveillance strategies for colon cancer in patients with IBD have not been established. The following recommendations have been issued by the major national societies in the United States and the British Society of Gastroenterology:

Ulcerative colitis

American Gastroenterological Association — The American Gastroenterological Association (AGA) recommends that colonoscopic surveillance should begin after eight years in patients with pancolitis, and 15 years in patients with colitis involving the left colon. Colonoscopy should be repeated every one to two years.

American College of Gastroenterology — The American College of Gastroenterology (ACG) recommends annual surveillance colonoscopy beginning after 8 to 10 years of disease in patients who are surgical candidates. Multiple biopsies should be obtained at regular intervals. The finding of definite dysplasia (of any grade) should be confirmed by an expert pathologist and is an indication for colectomy. Patients whose biopsies are indefinite for dysplasia after review by an expert pathologist should undergo repeat surveillance colonoscopy at a shorter interval. The ACG recommendations do not specify whether surveillance can begin later in patients with disease limited to the left colon.

American Society for Gastrointestinal Endoscopy — The American Society for Gastrointestinal Endoscopy (ASGE) recommends that patients with ulcerative colitis who have pancolitis should begin surveillance colonoscopy after eight years of disease. Four biopsies should be obtained every 10 cm from the cecum to the rectum. In addition, any suspicious lesions or masses should be biopsied. Colonoscopy should be repeated every one to three years.

The finding of carcinoma or high grade dysplasia is an indication for colectomy. Colectomy is also indicated for any degree of dysplasia associated with a lesion or mass. The finding of dysplasia should be confirmed by an expert pathologist.

For patients with low grade dysplasia confirmed by an expert pathologist, the ASGE acknowledges that most experts recommend colectomy. However, in patients in whom colectomy is not feasible or is unacceptable, frequent surveillance (eg, every three to six months) is considered an acceptable alternative.

For patients with left-sided colitis, the ASGE recommends that surveillance should begin after 15 years of disease. Surveillance is not indicated in ulcerative proctitis. The ASGE also recommends that prophylactic colectomy should be considered in patients with longstanding colitis. The definition of longstanding was not specified.

British Society of Gastroenterology — A guideline issued by the British Society of Gastroenterology recommends the following [123]:

• Surveillance colonoscopies should be performed when the disease is in remission.
• All patients should have a screening colonoscopy after 8 to 10 years that will also clarify the disease extent.
• Regular surveillance should begin after 8 to 10 years (from onset of symptoms) for pancolitis and after 15 to 20 years for left-sided colitis.
• As the risk of cancer increases exponentially with time, there should be a decrease in the screening interval with increasing disease duration. For patients with pancolitis, a colonoscopy should be conducted every three years in the second decade of disease, every two years in the third decade, and yearly by the fourth decade of disease.
• Two to four random biopsy specimens every 10 cm from the entire colon should be taken with additional samples of suspicious areas. Patients with primary sclerosing cholangitis (including those with an orthotopic liver transplant) represent a subgroup at higher risk of cancer and they should have annual colonoscopy.

Crohn's disease

American Gastroenterological Association — The AGA concluded that the risk of colorectal cancer associated with ulcerative colitis and Crohn's colitis is similar for comparable extent, duration, and age of onset of inflammatory disease. As a result, the surveillance strategy discussed above for UC also applies for Crohn's colitis.

American College of Gastroenterology — The ACG recognizes the expanding evidence of the carcinogenic potential of long-standing Crohn's disease but found insufficient evidence to provide guidelines for surveillance [124].

American Society for Gastrointestinal Endoscopy — The ASGE also recognizes that the risk of colorectal cancer is increased in Crohn's colitis. Surveillance colonoscopy and biopsy for dysplasia should be offered to patients with longstanding Crohn's colitis. However, the ASGE finds that the data guiding the selection of intervals and proper performance of biopsies are limited.

British Society of Gastroenterology — Recommendations for ulcerative colitis also apply to Crohn's colitis in a guideline issued by the British Society of Gastroenterology (see 'British Society of Gastroenterology' above [123].

INFORMATION FOR PATIENTS — Educational materials on this topic are available for patients. (See "Patient information: Colon cancer screening" and "Patient information: Ulcerative colitis" and "Patient information: Crohn's disease" and "Patient information: Colonoscopy".) We encourage you to print or e-mail these topic reviews, or to refer patients to our public web site, www.uptodate.com/patients, which includes these and other topics.

SUMMARY AND RECOMMENDATIONS — There is consensus that patients with UC should undergo surveillance colonoscopy, although the interval in which this should be done and the optimal number and location that biopsies should be obtained are uncertain. Furthermore, a reduction in mortality due to surveillance has not been clearly established. Patients with Crohn's colitis probably have the same risk of colorectal cancer as those with UC, and should also undergo surveillance.

Our current approach is as follows:

• In patients with ulcerative colitis extending proximal to the splenic flexure or Crohn's colitis, we suggest initiating surveillance colonoscopy after eight years of disease followed by annual examinations (Grade 2B). Those with Crohn's colitis may require use of a thin caliber colonoscope to adequately sample the colon including strictured areas [32]. We offer surgical resection to patients with Crohn's colitis who have an impassable stricture since they cannot be surveyed adequately; we suggest a barium enema in those who decline resection.
• In patients with left-sided UC, we suggest beginning with colonoscopy after 12 years of disease; examinations are then performed every year thereafter (Grade 2C).
• In patients who have undergone a subtotal colectomy with an ileostomy and have rectum left in place and tied off (ie, a Hartmann's pouch), we suggest surveillance of the remaining rectum every year (Grade 2B).
• We suggest not performing surveillance for patients with ulcerative colitis limited to the rectum (Grade 2B).
• In patients with a pouch, we suggest flexible sigmoidoscopy every other year; however, if they have evidence of severe atrophy we increase the surveillance interval to yearly (Grade 2C).

Multiple biopsies are required to adequately sample the colon. In one report, it was estimated that 33 biopsy specimens from a single colonoscopy would be required to detect dysplasia with 90 percent confidence [106]. In addition, areas of mucosal irregularity should be biopsied [107].

We recommend colectomy for patients who are found to have high-grade dysplasia (Grade 1B). Although opinions continue to vary, we also suggest colectomy for those with low-grade dysplasia (Grade 2B).

As discussed above, patients with UC also develop sporadic colorectal adenomas at the same rate as the general population. However, in some instances, a spontaneous adenoma can be distinguished from a non-adenoma-like dysplasia associated lesion or mass (DALM), but usually this is not possible by either endoscopy or pathology. Lesions can definitely be considered sporadic adenomas if they occur proximal to known areas of colitis (eg, right-sided polyp in a patient with left-sided colitis). We suggest (Grade 2B) that rather than performing a colectomy, an adenomatous polyp can be removed endoscopically with close follow-up for polyps which:

• Appear endoscopically and histologically to be adenomas (ie, adenoma-like DALM) regardless of whether they occurred in an area of the colon involved with colitis
• Are not associated with dysplastic changes in flat mucosa elsewhere in the colon

For lesions that appear endoscopically and histologically similar to a sporadic adenoma, we suggest performing a polypectomy and extensive biopsies in the surrounding area, and in the rest of the colon (Grade 2C). In the absence of dysplasia in the flat mucosa in the surrounding area or elsewhere in the colon, or invasive carcinoma in the polyp, the clinical aspects of the tumor and the patient's overall clinical status should be considered for recommending further surveillance or colectomy (table 1 and algorithm 1). However, most, if not all, adenoma-like lesions can be treated safely by polypectomy and continuous surveillance. Standard surveillance intervals can be continued in patients with low-risk lesions (although shortening the interval between examinations is also reasonable). Patients considered to have a sporadic adenoma who did not have adequate surveillance biopsies obtained throughout the colon at the time the polyp was removed should undergo a repeat colonoscopy with surveillance biopsies obtained throughout the colon (and extra attention to the area of the polypectomy); standard surveillance can be resumed if there is no evidence of flat dysplasia.

Non-adenoma-like polyps or other masses occurring in an area of colon that has been histologically involved with colitis, with or without surrounding dysplasia, should be considered as potentially containing invasive carcinoma [29,77].