Guidelines NCCN Guidelines for Colon Cancer NCCN Guidelines for Rectal Cancer Guidelines for Colonoscopy Surveillance after Cancer Resection

June 2005

Section 5 Gastrointestinal Tract and Abdomen

Colorectal cancer (CRC) remains a major public health problem throughout the world. In the United States, CRC is the third most frequently diagnosed cancer in both men and women and the second most common fatal cancer (behind lung cancer).¹ During 2004, there were an estimated 106,000 cases of colon cancer and 41,000 cases of rectal cancer in the United States, resulting in 57,000 total deaths.¹ The cost of treating colorectal cancer in the United States is believed to be between 5.5 and 6.5 billion dollars a year.² Worldwide, the risk of death from CRC is highest in developed countries and especially low in Asia and Africa.³

Figure 1. Frequencies of CRC

Data from the Surveillance, Epidemiology, and End Results (SEER) program indicate that the overall incidence of and mortality from CRC have been decreasing in the United States among both men and women,⁴ though they remain generally higher among men than among women. Overall, the incidence of and mortality from CRC are highest among African Americans, somewhat lower among European Americans, and lowest among Native, Asian, and Hispanic Americans [see Table 1].⁵ Most CRCs still occur in the distal colon (beyond the splenic flexure), but the incidence of proximal adenocarcinomas relative to that of distal adenocarcinomas has been increasing over the past 25 years [see Figure 1].⁶ The cause of this shift is not known.

Genetics

---

Figure 2. Genetic model of colorectal tumorigenesis

The development of CRC involves a progression from normal mucosa through adenoma to carcinoma.⁷ A genetic model of colorectal carcinogenesis has been proposed that describes a sequence of key mutations driving the process of colorectal carcinogenesis [see Figure 2].⁸ This process may involve the accumulation of mutations in both tumor suppressor genes and proto-oncogenes, as well as epigenetic phenomena such as DNA hypermethylation or hypomethylation.⁹ The onset of genomic instability increases the mutation rate and accelerates this progression. Inactivation of the adenomatous polyposis coli (APC) gene on chromosome 5q is thought to be one of the earliest mutations in sporadic cancers and is seen as a germline mutation in patients with familial polyposis [see 5:14 Hereditary Colorectal Cancer and Polyposis Syndromes]. Mutations in other tumor suppressor genes play an important role in this pathway as well, including mutations in DCC, SMAD2, and SMAD4 on chromosome 18q and p53 on chromosome 17p; these events are thought to occur at a later stage of tumor progression. Mutations in the K-ras oncogene occur at an intermediate stage. The accumulation of additional mutations (as yet poorly defined) allows metastases to develop.

Microsatellite instability (MSI) is an alternative pathway to genomic instability and subsequent colorectal carcinogenesis. This phenomenon arises from defects in mismatch repair genes, which cause significantly increased mutation rates in comparison with those in normal cells. MSI in hereditary nonpolyposis colorectal cancer (HNPCC) [see Risk Factors, below] is most commonly attributable to germline mutations in the hMLH1 and hMSH2 genes.¹⁰ MSI in sporadic CRC is most frequently associated with hypermethylation of the promoter region of the hMLH1 gene,¹¹ which leads to inactivation of the gene and loss of expression of the hMLH1 protein.

A number of risk factors for CRC have been described, including a family history of cancer or adenomatous polyps, familial CRC syndromes, inflammatory bowel disease (both ulcerative colitis and Crohn disease), and dietary and lifestyle factors.^12,13 The vast majority of CRCs worldwide are sporadic—that is, they are not associated with known genetic syndromes. In the United States, no more than 5% of CRCs are associated with known genetic syndromes.

In a meta-analysis of studies addressing CRC risk and family history, the relative risk of CRC in those with an affected first-degree relative was 2.25; this figure rose to 4.25 if more than one relative was involved and to 3.87 if CRC was diagnosed before the age of 45.¹⁴ The National Polyp Study found that the relative risk of CRC was 1.78 in first-degree relatives of patients with adenomatous polyps.¹⁵ In another study, the relative risk of CRC was 1.74 in first-degree relatives of patients with adenomatous polyps and was especially high (4.36) in those diagnosed with polyps at or before the age of 50.¹⁶

The most common of the genetic syndromes known to be associated with CRC is HNPCC, which accounts for the majority of patients with familial CRC. MSI is the characteristic finding of HNPCC, though it is also present in approximately 15% of all sporadic CRCs. HNPCC can be diagnosed clinically on the basis of what are known as the Amsterdam Criteria.¹⁷ Polyposis syndromes (e.g., familial polyposis and juvenile polyposis) account for the remainder of patients with familial CRC syndromes. HNPCC and polyposis syndromes are discussed further elsewhere [see 5:14 Hereditary Colorectal Cancer and Polyposis Syndromes].

As determined by a 2001 meta-analysis, the lifetime risk of CRC for patients with ulcerative colitis is 3.7%, which increases to 5.4% for patients with pancolitis and rises further with greater duration of disease.¹⁸ Despite the common misconception, Crohn disease may be associated with a similarly increased risk of CRC.¹⁹

Numerous lifestyle and dietary factors have been put forward as potential causes of increased CRC risk. Lower levels of physical activity and increased body mass are associated with an increased risk of CRC in both men and women.²⁰ The Western-style diet, which is high in calories and fat and low in fiber, is associated with high rates of CRC. There is evidence that increased dietary intake of calcium may confer some protection against the development of CRC and adenomatous polyps. The Calcium Polyp Prevention Study, a large randomized trial done in the United States, reported a small but statistically significant reduction in the incidence of recurrent colorectal adenomas with dietary calcium supplementation.²¹ To date, the evidence from randomized trials has not shown dietary fiber supplementation to have a similar effect. In Japan, where the incidence of CRC has traditionally been low, CRC has become considerably more common in the past few decades.²² This increased incidence is believed to be the result of post—World War II lifestyle changes (e.g., increased consumption of animal fat and decreased expenditure of energy) that mirror Western habits.

Early diagnosis of colorectal neoplasms at a presymptomatic stage is important for improving survival. Polypectomy has consistently been shown to decrease the subsequent development of CRC: the National Polyp Study found that the incidence of CRC in patients who underwent colonoscopic polypectomy was as much as 90% less than would otherwise have been expected.²³ Identifying patients with early-stage disease that has not yet metastasized can prevent many CRC-related deaths. Early detection of and screening for CRC have become important components of routine care and public health programs both in the United States and abroad. The benefits of screening for CRC are especially substantial in patients who are at high risk for CRC (e.g., those with affected first-degree relatives), but even average-risk patients derive some benefit.

There is no ideal method of screening for CRC that is applicable to all patients. Physical examination is generally not helpful in making the diagnosis; various investigative tests are used instead. Modalities commonly employed for CRC screening and early detection include fecal occult blood testing (FOBT), double-contrast barium enema (DCBE), flexible sigmoidoscopy, and colonoscopy. Of these, only FOBT and sigmoidoscopy have been tested in randomized trials.²⁴ It is clear, however, that these tests are less sensitive and specific than colonoscopy. There is evidence that colonoscopy detects many CRCs in asymptomatic patients that would not be detected by sigmoidoscopy.^25,26 Colonoscopy has been shown to be a safe and effective method of CRC screening in asymptomatic, average-risk patients.²⁷

Newer screening modalities, such as virtual colonoscopy and stool DNA assays, are currently being developed and tested. Virtual colonoscopy, which uses high-resolution computed tomographic scanning to image the colon, has been evaluated in at least two multicenter trials in the United States, with varying results.^28,29 One of the studies reported a sensitivity and a specificity of 89% and 80%, respectively, for polyps larger than 6 mm and up to 94% and 96%, respectively, for polyps larger than 10 mm.²⁸ The sensitivities were equivalent to those of optical colonoscopy in this group of asymptomatic average-risk patients. The second study, however, found that virtual colonoscopy had a sensitivity of only 39% for lesions larger than 6 mm and 55% for lesions larger than 10 mm.²⁹ Given these divergent findings, it appears that there are issues related to equipment, software, and training that remain to be addressed before virtual colonoscopy can be recommended as a routine screening modality. Another consideration is that patients with lesions detected by means of virtual colonoscopy must still undergo optical colonoscopy for treatment or tissue diagnosis. Fecal DNA assays have been developed to test for mutations in multiple genes known to be involved in colorectal neoplasia and are currently being evaluated in clinical trials.³⁰ These assays are not as sensitive as colonoscopy but may be useful in patients who are unable or unwilling to comply with endoscopic screening.³¹

Many groups have advocated CRC screening, and published guidelines are available from several organizations, including the American Cancer Society,³² the American Gastroenterologic Association,³³ and the U.S. Preventive Services Task Force.³⁴ All of these guidelines recommend that screening begin at age 50 for average-risk patients. The recommended screening options are consistent among the various organizations and include (1) FOBT yearly, (2) flexible sigmoidoscopy every 5 years, (3) yearly FOBT and flexible sigmoidoscopy every 5 years, (4) DCBE every 5 years, and (5) colonoscopy every 10 years. In high-risk patients (e.g., those with a family history of CRC), screening may begin at an earlier age—generally, 10 years younger than the age of the affected first-degree relative. There are also specific intensive screening and follow-up regimens for patients with known or suspected familial cancer syndromes.

As a consequence of the use of screening modalities, patients with CRC are often asymptomatic at diagnosis. Some CRC patients present with occult GI bleeding and anemia. Many patients do not exhibit symptoms until relatively late in the course of the disease. The duration of symptoms, however, is not necessarily associated with the stage of the tumor.³⁵

The most common symptoms of CRC are bleeding per rectum, abdominal or back pain, and changes in bowel habits or stool caliber. Other symptoms are fatigue, anorexia, weight loss, nausea, and vomiting. Some patients present with acute bowel obstruction or perforation.

Staging and Prognosis

---

Figure 3. Classification of CRC

Accurate staging of CRC is extremely important for determining patient prognosis and assessing the need for adjuvant therapy. Traditionally, staging of CRC has been based on modifications of the Dukes classification, which was initially developed as a prognostic tool for rectal cancer in the 1930s.³⁶ Since this classification was first implemented, it has undergone multiple modifications, of which the most widely used is the modified Astler-Coller system, initially introduced in the 1950s.³⁷ Currently, the TNM classification, developed by the American Joint Committee on Cancer (AJCC) and the International Union against Cancer (UICC), is the preferred staging system [see Tables 2 and 3].³⁸ This system takes into account the depth of penetration into the bowel wall (T) [see Figure 3], the presence and number of involved mesenteric nodes (N), and the presence of distant metastases (M).

There is some debate regarding the utility of preoperative CT scans in the management of primary colon cancer. The rationale for obtaining these scans includes evaluation of potential metastatic disease and assessment of the local extent of disease. In a 2002 study of preoperative CT in patients with intraperitoneal colon cancer, however, the results of the imaging changed management in only 19% of patients, and CT had a sensitivity of only 78% for all metastatic disease.³⁹ Nonetheless, many surgeons routinely perform staging CT in patients with primary colon cancer. PET is a sensitive study, but its routine use for staging primary CRC is not generally recommended. PET may be considered for high-risk patients in whom the detection of metastases would change initial management.⁴⁰

In cases of rectal cancer, locoregional staging may significantly affect therapeutic decision making. Such staging includes determination of the depth of invasion of the rectal wall and the degree of regional node involvement. Modalities commonly used include CT, MRI, and endoscopic ultrasonography (EUS). In a 2004 meta-analysis that examined the relative utility of each of these studies in rectal cancer staging,⁴¹ EUS proved to be the most accurate technique for evaluating muscularis propria involvement and perirectal tissue invasion. The various techniques were equally accurate in assessing lymph node involvement, with none of them being highly sensitive.

Pathologic Staging

Figure 4. Five-year survival rates for colon cancer

Figure 5. Five-year survival rates for rectal cancer

Figure 6. Five-year survival rates: stage II colon cancer

Definitive pathologic staging is carried out after surgical exploration and examination of the resected specimen. The final stage of the cancer is then determined on the basis of the TNM system [see Tables 2 and 3] . Survival is correlated with the stage of the tumor [see Figures 4 and 5] . In the current (sixth) edition of the AJCC staging system,³⁸ stage II is subdivided into stages IIA and IIB, and stage III is subdivided into stages IIIA, IIIB, and IIIC on the basis of both the extent of wall penetration and the number of nodes involved. These changes were implemented as a result of studies demonstrating differences in survival among these subgroups [see Figure 6].⁴²

Numerous other criteria have been evaluated as additional prognostic factors in CRC. The degree of lymphatic invasion and the extent of vascular invasion are important adjuncts to the TNM staging system and are incorporated in the current schema.³⁸ Certain histologic types, including signet-ring and mucinous carcinomas, are associated with poor outcomes. The preoperative serum carcinoembryonic antigen (CEA) level may be an independent prognostic factor that is predictive of resectability and the presence of distant metastases.⁴³

Various molecular markers have been investigated with respect to prognosis and response to therapy in CRC patients. Unfortunately, there are conflicting data on the prognostic impact and clinical utility of most of these markers. As noted [see Risk Factors, above], MSI is seen in as many as 15% of patients with sporadic CRC. Patients with MSI typically have proximal, poorly differentiated tumors with mucinous or signet-ring components, but they usually exhibit improved overall survival.⁴⁴ These patients may be less sensitive to 5-fluorouracil (5-FU)-based chemotherapy.⁴⁵

The long arm of chromosome 18 (18q) harbors at least three candidate tumor suppressor genes, including DCC, SMAD2, and SMAD4. Deletions of chromosome 18q in CRC patients are associated with decreased survival. One study found that patients with stage II cancers and 18q allelic loss had a prognosis similar to that of patients with stage III disease.⁴⁶ In addition, p53 mutations and overexpression are associated with poor outcomes in CRC.⁴⁷ Thymidylate synthase is an enzyme active in DNA synthesis that is targeted by 5-FU and similar chemotherapeutic agents. Overexpression of this enzyme is associated with a poor prognosis but also with improved sensitivity to 5-FU—based chemotherapy.⁴⁸

All of these molecular alterations, as well as others (e.g., K-ras mutations and 5q deletions), are commonly observed in CRC patients, but further study is required to establish their real prognostic significance.

Surgical Therapy

Figure 7. Treatment of colon cancer

Surgery with curative intent remains the mainstay of therapy for colon cancer [see Figure 7]. Complete R0 resection (leaving no gross or microscopic disease) with wide margins along the bowel wall, coupled with regional lymphadenectomy, is the standard of care. The major arterial vessels supplying the segment of the colon containing the tumor should be excised at their origins. A minimum margin of 5 cm of normal bowel on each side of the tumor is considered adequate.

Extent of Resection

The standard extent of resection for various colon cancers has been defined. For tumors of the cecum and the ascending colon, a right hemicolectomy that includes the right branch of the middle colic artery at its origin should be performed. For tumors of the hepatic flexure, an extended right colectomy that includes the entire middle colic artery is indicated. For tumors of the transverse colon, an extended right or left colectomy or a transverse colectomy may be performed. For tumors of the splenic flexure region, a left hemicolectomy is performed, and for sigmoid tumors, a sigmoid colectomy is performed.

In patients who have small or flat tumors or who are undergoing resection after a polypectomy, intraoperative identification of the tumor may be difficult. This is especially true with laparoscopic procedures, in which the bowel often cannot be palpated. If the lesion is in the cecum, the ileocecal valve and the appendiceal orifice are visualized endoscopically, and localization of the tumor is simple. If the lesion is at another location, endoscopic measurements of the distance from the anus or estimates of the location of the tumor may be inaccurate. Endoscopic tattooing, a process in which an agent is injected into the bowel wall submucosally at or near the site of the lesion, has been employed to facilitate intraoperative identification of the tumor site. India ink is the agent most commonly used for this purpose and generally yields excellent results.⁴⁹ As an alternative, many institutions use a commercially available sterile suspension of carbon particles, which is also very safe and effective.⁵⁰ Intraoperative endoscopy is another option for locating these lesions.

Surgical Staging

The selection of patients for adjuvant therapy relies heavily on accurate staging. A significant percentage of patients with early-stage node-negative disease present with recurrences or metastases; such a presentation implies that the patients had occult metastatic disease at the time of operation. Surgical resection of CRC should include division of the appropriate mesenteric vessels at their origins, along with resection of the regional nodes. Optimal staging of CRC patients, especially with regard to nodal status, remains controversial. One area of debate is the number of nodes that must be examined to confirm node-negativity. This number depends both on the surgeon's technique (i.e., how many nodes were resected) and on the pathologist's efforts to harvest nodes from the specimen. Most groups recommend analysis of at least 12 nodes to confirm node-negativity.⁵¹

Because of the importance of nodal status, ultrastaging of harvested nodes with techniques such as serial sectioning, immunohistochemistry (IHC), and reverse transcriptase polymerase chain reaction (RT-PCR) has been proposed as a means of detecting micrometastases. All of these techniques may result in upstaging of patients who are node negative on standard pathologic analysis, which involves only bivalving the nodes and examining a limited number of sections. The prognostic impact of micrometastases that are detected only by IHC or RT-PCR and are not verified by hematoxylin-eosin staining remains unclear. It is impractical to perform these assays on all nodes harvested; accordingly, the use of lymphatic mapping to identify sentinel lymph nodes (SLNs) has been proposed as a means of selecting a small number of nodes for further analysis.

SLN biopsy in the setting of CRC remains investigational. Lymphatic mapping may be done with either in vivo or ex vivo injection of tracer dye. The dye rapidly diffuses through the lymphatic vessels, and SLNs can be identified and marked in the mesocolon within minutes. This procedure has been shown to be feasible in a number of studies⁵²; however, its sensitivity and false negative rates have been variable. In a 2004 multicenter trial, SLN biopsy with serial sectioning had a false negative rate of 54% in patients with node-positive colon cancer.⁵³ In a large single-institution trial, both SLNs and non-SLNs were studied with serial sectioning and IHC in patients who were node negative on routine pathologic analysis⁵⁴; 19.5% of patients were upstaged by the combination of serial sectioning and IHC of SLNs. These results imply that the main role of this technique may be in upstaging patients who are node negative on routine pathologic analysis. Further study is required before the use of SLN techniques in the context of CRC becomes standard clinical practice.

Occult metastatic disease may also be present in the peritoneal cavity or systemically in the blood or bone marrow at the time of operation. The presence of tumor cells in the peritoneum may be detected by performing cytologic analysis of washings done at the time of operation. In one study, disseminated tumor cells were identified in peritoneal washings or blood in 25% of patients, and their presence was found to be an independent prognostic factor for survival.⁵⁵ In another study, patients with positive peritoneal washings had significantly higher rates of local recurrence and peritoneal carcinomatosis but manifested no differences in survival.⁵⁶ Again, further study is required before these assays can be routinely used for staging CRC.

Laparoscopic versus Open Colectomy

At present, open colectomy is the most widely accepted treatment of resectable colon cancer. Laparoscopic techniques have been developed and are being tested in prospective, randomized multicenter trials [see 5:34 Segmental Colon Resection]. Initially, there were concerns about port-site recurrences,⁵⁷ but current data suggest that these concerns are unfounded.⁵⁸ With respect to comparative cost, data from a subset of patients in the European COlon cancer Laparoscopic or Open Resection (COLOR) trial demonstrated that although the total cost to society from laparoscopic colectomy is similar to that from open colectomy, the costs to the health care system are significantly higher with the former.⁵⁹ The Clinical Outcomes of Surgical Therapy (COST) study group, a large, randomized, multicenter trial conducted in the United States, found that laparoscopic-assisted colectomy conferred only minimal (though statistically significant) short-term quality-of-life benefits when compared with open colectomy.^60,61 Cancer-specific outcomes (e.g., recurrence rates, wound recurrences, and overall survival rates) were similar with the two approaches.⁶¹ The COST investigators concluded that laparoscopic colectomy is an acceptable alternative to open colectomy. Recurrence and survival data from other large multicenter trials (e.g., the COLOR trial⁶²) are not yet available.

Special Situations

Obstructing and perforated cancers Obstructing and perforated colon cancers are associated with a poor prognosis and with increased surgical morbidity (as a consequence of the need for emergency surgery). Perforation can occur either via direct erosion of the tumor through the wall of the colon or secondary to obstruction with resultant bowel distention proximal to the tumor. Patients with perforated colon cancer are managed with emergency laparotomy, washout, and resection of the primary lesion to prevent further soilage. A diverting stoma is usually indicated, with either a Hartmann pouch or a mucous fistula constructed distally. Select patients may be managed by means of primary anastomosis, with or without a proximal diverting colostomy or ileostomy.

Obstructing right-side cancers (up to the splenic flexure) can usually be treated with resection and primary anastomosis. The traditional emergency treatment of obstructing left-side colon cancers is a diverting colostomy, with or without resection of the lesion. In many such cases, the stoma is never taken down. Some surgeons advocate emergency treatment of these lesions with total abdominal colectomy and ileorectal anastomosis as a means of improving outcomes.⁶³ Another treatment option is primary resection and anastomosis, with or without on-table intestinal lavage. Yet another option for managing obstructing left-side colon and rectal cancers is the use of colorectal stents with the aim of avoiding emergency surgery. Stents can serve as a bridge to definitive resection by decompressing the colon and thereby allowing subsequent bowel preparation. In patients with advanced disease, stents may also be employed for palliation as an alternative to surgical resection or a diverting stoma.

Synchronous primary colorectal cancers The incidence of synchronous CRCs is reported to range from 3% to 5%^64,65 but may be as high as 11%.⁶⁶ Stage for stage, there appear to be no differences in survival between synchronous cancers and single primary cancers.^67,68 Synchronous adenomatous polyps are present in as many as 35% of patients undergoing surgical treatment of CRC.^65,68 In one study, the presence of synchronous lesions made the surgical procedure more extensive than was initially planned for resection of the primary tumor in 11% of patients.⁶⁵

Most synchronous polyps are identified on preoperative colonoscopy, and the colon can often be cleared of these lesions before operation. Management of adenomas not amenable to endoscopic resection and management of synchronous cancers are more challenging. Each primary cancer must be managed surgically according to sound oncologic principles. One option is to perform multiple segmental resections with multiple anastomoses. Another is to perform an extended resection that encompasses all of the lesions or even total abdominal colectomy if needed. The presence of a rectal cancer and a second synchronous lesion makes surgical treatment even more challenging, especially if sphincter preservation and a low rectal or coloanal anastomosis are contemplated.

The 5-year survival rate after resection of colon cancer is inversely correlated with the pathologic stage [see Figure 4]. The diminishing 5-year survival rates for stage II and III colon cancer became the basis of several phase III randomized trials designed to test the hypothesis that postoperative systemic adjuvant chemotherapy would significantly improve survival in patients with resected but high-risk cancers. Multi-institutional, cooperative cancer group trials were necessary to obtain populations large enough to test this hypothesis. The North Central Cancer Treatment Group (NCCTG) initiated a randomized trial of postoperative systemic adjuvant 5-FU plus levamisole for Dukes stage B and C (AJCC stage II and III) colon carcinomas.⁶⁹ Patients were randomly assigned to receive either levamisole alone or 5-FU plus levamisole. Overall survival was significantly improved in stage C patients treated with 5-FU plus levamisole. This was the first randomized trial to demonstrate the efficacy of systemic adjuvant therapy.

The NCCTG trial led to second-generation trials of adjuvant therapy for patients with resected colon cancers. In one such study, patients with high-risk stage II or stage III colon cancer were randomly assigned to receive either 5-FU plus leucovorin and levamisole or 5-FU plus levamisole.⁷⁰ Survival rates after 12 months of adjuvant chemotherapy were no better than those after 6 months of chemotherapy; however, 5-FU plus levamisole proved to be inferior to 5-FU plus leucovorin and levamisole with respect to survival.

National Surgical Adjuvant Breast and Bowel Project (NSABP) protocol C-04 randomly assigned Dukes stage B and C colon cancer patients to receive (1) postoperative 5-FU plus leucovorin, (2) 5-FU plus levamisole, or (3) 5-FU plus leucovorin and levamisole.⁷¹ A slight improvement in 5-year disease-free survival was noted with 5-FU plus leucovorin, but overall 5-year survival did not differ significantly among the three treatment arms. Accordingly, 5-FU plus leucovorin became the standard adjuvant regimen.

Intergroup Trial 0089 randomly assigned patients with high-risk stage II and III disease to receive either 5-FU plus high-dose leucovorin or 5-FU plus low-dose leucovorin. The investigators concluded that (1) the high-dose and low-dose regimens were equivalent, (2) a regimen consisting of four cycles of 5-FU with high-dose weekly leucovorin was equivalent to the low-dose leucovorin Mayo Clinic regimen, and (3) the addition of levamisole to the 5-FU plus leucovorin regimen did not improve survival.

These clinical trials have established 5-FU plus leucovorin as standard therapy for patients with high-risk stage II and stage III colon cancer. The next generation of clinical investigations should provide data on the potential benefits of augmenting this regimen with irinotecan or oxaliplatin in an adjuvant setting.⁷²

Routine use of systemic adjuvant therapy for stage II colon cancer remains controversial. Patients with stage II colon cancers, including those at high risk (e.g., those who present with large bowel obstruction or perforation), are typically included in adjuvant chemotherapy trials. A meta-analysis of stage II patients included in NSABP colon cancer trials demonstrated that adjuvant chemotherapy did confer a survival benefit at this disease stage.⁷³ This study was criticized, however, for having included patients from trials that lacked a surgery-only arm, as well as from trials that employed outmoded chemotherapeutic regimens.⁷⁴ Another meta-analysis, which included only trials that compared 5-FU plus leucovorin with observation after curative resection in stage II patients, found no statistically significant survival benefit with chemotherapy.⁷⁵ A 2004 meta-analysis formulated recommendations on this controversial topic and provided a Web-based tool for calculating risk.⁷⁶ This report included data from seven randomized trials that compared surgery alone with surgery plus chemotherapy. Patients with node-negative disease derived a much lower reduction in risk and no statistically significant improvement in overall survival. The authors concluded that the use of postoperative adjuvant chemotherapy for stage II colon cancer patients should be individualized on the basis of the estimated prognosis and the potential treatment benefit.

In summary, postoperative systemic adjuvant therapy is the standard of care in patients with stage III disease. In stage II colon cancer patients who have undergone complete surgical resection, the relative risk of recurrence is small enough that adjuvant chemotherapy yields relatively little benefit in terms of survival. There is, however, a subgroup of patients who have recognized prognostic factors that significantly reduce survival and in whom adjuvant therapy is therefore more likely to be beneficial. These risk factors include (1) bowel obstruction, (2) colonic perforation, (3) high-grade or lymphovascular invasion, and (4) the presence of fewer than 12 lymph nodes in the resected specimen.

Rectal cancer presents special management issues with respect to local recurrence after surgical resection. With cancer of the intraperitoneal colon, local recurrence is rare. With rectal cancer, however, surgical treatment alone results in recurrence rates of 16.2% after low anterior resection (LAR) and 19.3% after abdominoperineal resection (APR).⁷⁷ Higher stages are associated with higher recurrence rates: 8.5% for Dukes stage A, 16.3% for stage B, and 26% for stage C.⁷⁷ Multimodality management, including adjuvant radiation therapy or chemotherapy (or both) in combination with appropriate operative therapy, can reduce local recurrence rates significantly.

Radical resection

Figure 8. Treatment of rectal cancer

Traditionally, tumors of the rectum have been treated with either LAR or APR [see Figure 8]; in numerous series, APR rates of 60% or higher have been reported. Surgical techniques such as stapled or handsewn coloanal anastomoses [see 5:29 Intestinal Anastomosis].

The morbidity associated with radical rectal resection can be substantial. Anastomotic leakage rates vary widely, ranging from less than 10% to more than 30% after resection with anastomosis. Leaks can lead to substantial morbidity and mortality and can necessitate reoperation. Such concerns have prompted the use of temporary diverting ileostomies or colostomies in patients with low rectal anastomoses. Defunctioning stomas may be overused, however, thereby increasing the cost of care in low-risk patients.⁷⁹ Preoperative chemoradiation therapy has not been shown to increase anastomotic leakage rates. Urinary and sexual dysfunction are also fairly common after radical resection of rectal cancer. Autonomic nerve preservation in conjunction with TME may improve the functional results of these procedures.⁸⁰ The use of local resection techniques (see below) is another means of reducing surgical morbidity and mortality in rectal cancer patients.

Local excision Local excision-including transanal, transsphincteric, and transcoccygeal techniques, as well as transanal endoscopic microsurgery (TEM) [see 5:35 Procedures for Rectal Cancer]—is another option for curative resection of low rectal cancers with preservation of sphincter function. These procedures were initially implemented for local control in patients who were medically unfit for or unwilling to undergo major resections. Transsphincteric and transcoccygeal resections have been associated with an increased incidence of complications, including fecal fistulas and incontinence, and have largely been abandoned now that other, better techniques are available.

Local excision with curative intent is generally reserved for the treatment of early-stage (T1-2N0) lesions. Selection of patients for these procedures is critical and is based on preoperative staging and on the probability of harboring nodal metastases, which increases with the T stage. EUS has become an important staging procedure in these patients, both for assessing the depth of tissue invasion and for detecting the presence of nodal disease. CT is generally performed to rule out distant metastases. Palliative procedures (e.g., fulguration and endocavitary irradiation) may also be considered in patients who are unfit for major surgery.

Several criteria have been established to identify patients who may be candidates for transanal excision (TAE).⁸¹ Generally, the lesion must be no more than 4 cm in diameter, must encompass no more than one third of the circumference of the rectum, and must be less than 8 cm from the anal verge. With the advent of TEM, these criteria have been expanded to include patients with higher lesions. Poorly differentiated tumors and the presence of lymphovascular invasion may also be associated with increased nodal involvement and higher recurrence rates. At least two prospective trials have reported their results with TAE.^82,83 Local recurrence rates ranged from 5% to 7% for T1 lesions treated with surgery alone. Results were not as promising for T2 lesions: local recurrence rates ranged from 14% to 16%, even when adjuvant radiation or chemoradiation therapy was provided.

The use of local excision in patients with more locally advanced disease is even more controversial. Such patients are at considerably greater risk for nodal metastases and thus for local recurrence even after adequate resection of the primary lesion. Traditionally, local excision in patients with locally advanced disease has been associated with unacceptably high recurrence rates. Some authors advocate combining chemoradiation therapy with local excision to manage these patients. In a 2004 retrospective series, the results of local excision were comparable to those of radical resection in T3 patients who had a good response to preoperative chemoradiation therapy and who refused or were medically unfit for major surgery.⁸⁴ The role of local excision in these patients remains poorly defined.

Patients undergoing local resection must receive careful follow-up, including digital examination, measurement of CEA levels, proctoscopy, and, possibly, transanal ultrasonography. A subset of these patients with local-only recurrences who are medically fit for surgery may be candidates for resection. At present, few good data are available on the results of salvage surgery for local recurrence after local excision of rectal cancer, but it is unlikely that outcomes are equivalent to those of initial radical resection.⁸⁵

Importance of Radial and Distal Resection Margins

There has been a great deal of debate about what constitutes an adequate margin of resection in surgical treatment of rectal cancer. With respect to distal margins, 2 to 5 cm has traditionally been considered to be the minimum necessary for curative resection. Growing interest in sphincter preservation has led investigators to consider smaller distal margins (i.e., < 2 cm). Studies have shown that clear margins smaller than 2 cm are not associated with higher local recurrence rates or reduced survival.⁸⁶ Subsequent reports have suggested that even smaller histologically negative margins (i.e., < 1 cm) may be adequate in patients receiving adjuvant chemoradiation therapy.^87,88

The importance of radial margin involvement after rectal cancer resection was not recognized until comparatively recently.⁸⁹ Radial margins are assessed by means of serial slicing and evaluation of multiple coronal sections of the tumor and the mesorectum.⁸⁹ Involvement of radial margins is a predictor of both local recurrence and survival after potentially curative rectal cancer surgery⁹⁰ and may be associated with an increased risk of distant metastases.⁹¹ Radial margins smaller than 2 mm are associated with increased local recurrence rates.⁹¹ Adjuvant radiation therapy does not compensate for the adverse impact of positive margins on local recurrence rates.⁹²

Adjuvant therapy for rectal cancer has focused both on locoregional control of disease and on treatment of systemic disease. Several large studies have evaluated local recurrence of disease after surgical resection alone. Local failure rates of 30% to 40% for T2N0 disease and 50% to 70% for node-positive disease strongly suggested that postoperative adjuvant therapy was needed.^93–95 In distinct contrast to these data, however, other series in which TME was performed reported extremely low local recurrence rates with surgery alone.^96,97

A series of randomized trials were conducted to assess adjuvant therapy for rectal cancer. Initial studies reported a decrease in local recurrence rates with postoperative radiation therapy.^98,99 In a multi-institutional trial conducted by the NCCTG, the combination of 5-FU with radiation therapy led to improvements in local control rates and in survival.¹⁰⁰ These results were confirmed in large intergroup trials, the results of which indicated that continuous infusion of 5-FU during radiation therapy resulted in significantly better disease-free survival and overall survival than bolus infusion of 5-FU.

Simultaneously with the ongoing development of postoperative locoregional adjuvant therapy for rectal cancer, interest in preoperative therapy has been growing. Preoperative radiation therapy has been associated with excellent local control of disease, sphincter preservation, and acceptable postoperative recovery. There is evidence that rectal adenocarcinoma is sensitive to preoperative radiation therapy, with or without 5-FU. Pathologic complete response rates of 10% to 20% have been noted in resected rectal specimens¹⁰¹; pathologic complete response is associated with improved outcomes.¹⁰²

Perhaps the strongest reason to consider preoperative therapy for rectal cancer is its potential for inducing significant tumor regression before surgical resection. Such regression makes clear radial and distal margins easier to obtain. Moreover, tumor regression with preoperative therapy may result in higher sphincter preservation rates. In many published series, the APR rate in rectal cancer patients is between 40% and 60%; more aggressive preoperative efforts to induce regression may give surgeons a better chance to achieve sphincter preservation without compromising local control of disease.¹⁰¹

There remains significant controversy regarding the choice between preoperative and postoperative radiation therapy for rectal cancer. An advantage of the postoperative approach is that the disease is more accurately staged before adjuvant therapy begins, and thus, patients with early-stage disease are less likely to be overtreated. Two trials attempted to compare preoperative and postoperative radiation therapy for rectal cancer in an effort to determine their relative effects on local control, overall survival, and sphincter preservation. Both studies were unsuccessful, however—probably because of bias on the part of the treating physicians in favor of either preoperative or postoperative radiation therapy—and were closed because of slow accrual. At present, there is greater enthusiasm for preoperative therapy and pretreatment staging with transrectal ultrasonography, which is 90% accurate for determining the T stage. Patients with T3 or T4 rectal cancer on ultrasonography would be eligible for preoperative treatment.

Two Swedish studies studied the role of preoperative radiation therapy in treating rectal cancer. The first demonstrated that a short course of preoperative radiation therapy (2.5 Gy in 5 fractions) was comparable to high-dose postoperative radiation therapy (60 Gy over a period of 8 weeks). The local recurrence rate was significantly lower with the short-course preoperative regimen (12% versus 21%), and there was no overall survival difference between the two regimens.¹⁰³ In the second trial, patients received either a short course of preoperative radiation therapy or surgery alone.¹⁰⁴ The local recurrence rate for preoperative therapy plus surgery was 11%, compared with 27% for surgery alone. The combined regimen also resulted in significantly better 5-year survival (58% versus 48%).

The question of the relative merits of preoperative and postoperative radiation therapy may be resolved by the findings from a 2004 German trial that randomly assigned patients to receive either preoperative or postoperative 5-FU plus radiation, followed by systemic 5-FU therapy.⁷⁸ This study was limited to patients with locally advanced disease, including those who had T3 or T4 disease or were node positive on ultrasonography. TME was performed in all patients and was done 6 weeks after treatment in patients receiving preoperative chemoradiation therapy. The primary end point of this study was overall survival; secondary end points included disease-free survival, local and distant control of disease, sphincter preservation, toxicity of adjuvant therapy, surgical complications, and quality of life. There was no difference between the preoperative group and the postoperative group with respect to 5-year survival, but the local recurrence rate was significantly lower with the former (6% versus 13%), as were both the short-term and the long-term toxicity of adjuvant therapy. Although overall, the rates of complete (R0) resection and sphincter preservation were similar in the two groups, the APR rate was significantly lower in patients determined by the surgeon to require APR before randomization.

In a retrospective review of patients presenting with unresectable stage IV CRC, there was no difference in survival between those who were initially managed surgically and those who were initially managed nonoperatively.¹⁰⁵ In the surgical group, the morbidity rate was 30% and the mortality 5%. Only 9% of the nonoperative patients subsequently required surgical intervention for bowel obstruction. In another retrospective series, patients managed surgically had significantly better overall survival than those managed nonoperatively but had a lesser tumor burden¹⁰⁶; 29% of the nonoperative patients eventually required surgery for bowel obstruction. When prognostic factors were evaluated in the surgical arm of this series, the only factor associated with improved outcomes was a less than 25% extent of liver involvement. On the basis of these and other studies, asymptomatic patients with unresectable metastatic CRC should be managed selectively: those with limited tumor burdens may benefit from surgical treatment, whereas those with more extensive disease (especially extensive liver involvement) may initially be managed nonoperatively.

Management of patients with synchronous resectable isolated liver metastases continues to evolve. Many studies have documented improved survival after liver resection in patients with metastatic disease that is confined to the liver. Patients presenting with synchronous lesions have a worse prognosis than those presenting with metachronous lesions.¹⁰⁷ Many of these patients have been managed with staged resections of the primary cancers and the liver metastases. Several groups have reported that such combined procedures do not substantially increase surgical morbidity and mortality or compromise cancer survival.^108,109 These combined procedures should be done only in carefully selected patients at specialized centers with significant experience in resection of both CRC and liver tumors.

Peritoneal carcinomatosis develops in approximately 13% of all CRC patients.¹¹⁰ The survival rate of patients who present with peritoneal carcinomatosis from CRC is dismal. In patients with stage IV CRC, the presence of carcinomatosis is associated with a significant reduction in survival (from 18.1 months to 6.7 months).¹¹¹ Treatment has traditionally included systemic chemotherapy, with surgery reserved for palliation of symptoms such as bowel obstruction. Newer chemotherapy regimens that include agents such as oxaliplatin may improve survival, but they certainly are not curative.

Peritoneal carcinomatosis is often associated with hematogenous metastases, but in some 25% of patients, the peritoneal cavity is the only site of disease. Several groups have advocated the use of cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (HIPEC) as a means of improving survival in these patients.¹¹² This treatment, however, is associated with significant morbidity and mortality.¹¹³ A randomized trial from the Netherlands that compared cytoreduction surgery plus HIPEC with systemic chemotherapy plus palliative surgery found that patients in the former group exhibited a statistically significant improvement in median survival (22.3 months versus 12.6 months).¹¹⁴ Cytoreductive surgery plus HIPEC seems to be a viable option for the treatment of peritoneal carcinomatosis. Patient selection for these aggressive procedures remains a major issue, given the substantial morbidity and mortality associated with them.

The goal of any CRC follow-up regimen should be to detect any recurrences or metachronous lesions that are potentially curable. In a large, multicenter trial, the incidence of second primary CRCs in patients with resected stage II and III lesions was found to be 1.5% at 5 years.¹¹⁵ Between 40% and 50% of patients experience relapses after potentially curative resection of CRC. Detection and treatment of recurrent disease before symptom development may improve survival. The time to recurrence is critical, in that as many as 80% of recurrences occur within the first 2 years and as many as 90% within the first 4 years. Patterns of recurrence should also be taken into account—for example, the markedly increased risk of local recurrence in rectal cancer patients compared with that in colon cancer patients. Even when recurrent CRC is detected, only a small percentage of patients are candidates for reoperation, and resection in these patients may not improve overall survival. Systemic therapy may improve survival in some patients who have unresectable recurrent lesions.

Various modalities are available for follow-up after surgical treatment of CRC. The history and the physical examination continue to be useful, in that a significant percentage of patients present with symptomatic recurrences. Measurement of serum CEA levels has proved effective in detecting asymptomatic recurrences. Other studies, such as liver function tests (LFTs), complete blood count (CBC), chest x-ray, and imaging studies (e.g., CT and ultrasonography), have not been consistently shown to detect asymptomatic resectable recurrences. One study that evaluated routine CEA measurement and CT scanning of the chest, the abdomen, and the pelvis for follow-up of stage II and III CRC demonstrated that both modalities were able to identify asymptomatic patients with resectable disease.¹¹⁶ Colonoscopy is valuable for detecting metachronous cancers and polyps.

Some authorities advocate so-called intensive follow-up. However, this term lacks a standard definition, and such follow-up has not been conclusively shown to be beneficial. In a meta-analysis that compared an intensive follow-up regimen (including history, physical examination, and CEA measurement) with no follow-up, the former detected more candidates for curative re-resection and led to improvements in both overall survival and survival of patients with recurrences.¹¹⁷ Two other meta-analyses have been published that assessed the value of intensive follow-up of CRC patients.^118,119 Both of these meta-analyses included only randomized, controlled trials, and both documented a survival advantage with intensive follow-up. Some caution is required in interpreting these results, however, because the meta-analyses included trials with vastly different follow-up regimens in their baseline and intensive groups.

At present, the ideal follow-up regimen for CRC patients remains to be determined. Intensive follow-up regimens obviously are more costly. Patients with stage I disease are at very low risk for recurrence and therefore do not require intensive follow-up.¹²⁰ Patients with stage II and III disease are at significantly higher risk for recurrence and therefore need more specific cancer-related follow-up, but how intensive such a follow-up regimen should be is still a matter of debate.

Several organizations, including the American Society of Clinical Oncology,¹²¹ the National Comprehensive Cancer Network (NCCN),^122,123 and the American Society of Colon and Rectal Surgeons,¹²⁴ have developed algorithms for postoperative surveillance of CRC patients. Their recommendations generally apply to patients with stage II or III disease (and sometimes patients with T2 lesions) who are candidates for resection of recurrent disease. The recommendations vary somewhat among groups, but the following are generally agreed on:

1. Measurement of CEA levels every 2 to 3 months for 2 years, then every 3 to 6 months for 3 years, then annually.

2. Clinical examination every 3 to 6 months for 3 years, then annually.

3. Colonoscopy perioperatively, then every 3 to 5 years if the patient remains free of polyps and cancer (the NCCN also recommends colonoscopy 1 year after primary therapy).

Imaging studies (e.g., CT and chest x-ray) are not routinely recommended, nor are other blood tests (e.g., CBC and LFTs).

A complete review of the treatment of recurrent CRC is beyond the scope of this chapter. The primary aim of postoperative surveillance is the detection of treatable recurrences or metastatic disease. The most common sites of metastasis in CRC patients are the liver and the peritoneal cavity. Surgery is the only potentially curative therapy for recurrent CRC. Only a select group of patients with isolated peritoneal, liver, or lung metastases are candidates for surgical resection. As noted [see Special Considerations, Peritoneal Carcinomatosis, above], cytoreductive surgery and HIPEC improve survival in patients with peritoneal carcinomatosis and may lead to long-term survival in a very select group of patients.

Numerous studies have addressed the treatment of patients with isolated liver metastases from CRC. Resection of isolated hepatic metastases has been reported to yield 5-year survival rates higher than 30%, with acceptable surgical morbidity and mortality. Investigators from the Memorial Sloan-Kettering Cancer Center developed a staging system known as the clinical score in an attempt to predict which patients are likely to benefit from aggressive surgical resection.¹²⁵ This system used five factors that were found to be independent predictors of poor outcome: (1) node-positive primary disease, (2) a disease-free interval shorter than 12 months, (3) the presence of more than one hepatic tumor, (4) a maximum hepatic tumor size exceeding 5 cm, and (5) a CEA level higher than 200 ng/ml. Patients who met no more than two of these criteria generally had good outcomes, whereas those who met three or more were recommended for inclusion in adjuvant therapy trials.

PET scanning has also been used to detect occult metastatic disease and thus to aid in the selection of patients for surgical resection. In one series, a 5-year overall survival of 58% was reported after resection of CRC liver metastases in patients screened with PET.¹²⁶ When combined with the clinical risk score, PET was found to be helpful only in patients with a score of 1 or higher.¹²⁷

Modalities for treating unresectable disease confined to the liver include cryotherapy, radiofrequency (RF) ablation, hepatic artery infusion of chemotherapeutic agents, and hepatic perfusion. Of these, RF ablation is the one most commonly employed. It may be performed via an open approach, percutaneously, or laparoscopically; it may also be combined with resection and with local or systemic chemotherapy. The survival benefit (if any) associated with use of these modalities has not been well established.

Patients with isolated lung metastases from CRC may also benefit from surgical resection. Because there are relatively few of these patients, treatment of such metastases has not been studied as well as treatment of liver metastases. Some series have reported 5-year survival rates higher than 40% after complete resection. Patient selection remains a major issue. Several prognostic factors that may predict poor outcomes have been identified, including (1) a maximum tumor size greater than 3.75 cm, (2) a serum CEA level higher than 5 ng/ml, and (3) pulmonary or mediastinal lymph node involvement.^128,129 Patients with both pulmonary and hepatic metastases may also be considered for surgical resection.

Pelvic recurrences of rectal cancer present another difficult management issue. These tumors may cause significant pain and disability, and if they are not treated, survival is measured in months. Radiation and chemotherapy provide symptomatic relief and yield a modest increase in survival. Surgery may provide excellent palliation and is potentially curative in patients who do not have distant metastases.

Multimodality therapy has been advocated as a means of improving the chances of cure. In one study, a 37% 5-year survival rate was reported in patients who underwent multimodality therapy, including resection with negative margins.¹³⁰ A subgroup of patients in whom complete resection was impossible underwent intraoperative radiation therapy; the 5-year survival in this subgroup was 21%. Several predictors of poor outcomes were identified, including incomplete resection, multiple points of tumor fixation, and symptomatic pain. In another series, hydronephrosis was associated with the presence of unresectable disease.¹³¹ Selection of appropriate patients for curative surgery remains a major issue in the management of locally recurrent rectal cancer.

Chemotherapy is the mainstay of palliative treatment for patients with CRC and unresectable recurrent or metastatic disease. Combinations of 5-FU and leucovorin with newer agents such as irinotecan and oxaliplatin define the current standard. Patients in whom these regimens fail may be considered for treatment with other newer agents, including cetuximab, a monoclonal antibody against epidermal growth factor receptor, and bevacizumab, a monoclonal antibody against the vascular endothelial growth factor receptor.