George P. Kim*
Chris H. Takimoto†
Carmen J. Allegra‡
*Gastrointestinal Cancer Section, Mayo Clinic Cancer Center, Jacksonville, Florida
†Institute for Drug Development, Cancer, Therapy & Research Center, San Antonio, Texas
‡Network for Medical Communication and Research, Atlanta, Georgia
Although certain conditions predispose patients to develop colon cancer, up to 70% of patients have no identifiable risk factors:
Familial Adenomatous Polyposis Syndrome
FAP is an autosomal-dominant inherited syndrome with more than 90% penetrance, manifested by hundreds of polyps developing by late adolescence. The risk of developing invasive cancer over time is virtually 100%. Germline mutations in the adenomatous polyposis coli (APC) gene on chromosome 5q21 have been identified. The loss of the APC gene results in altered signal transduction with increased transcriptional activity of ß-catenin.
Attenuated FAP: These FAPs are flat adenomas that arise at an older age than FAPs do; mutations tend to occur in the proximal and distal portions of the APC gene.
Gardner syndrome: This syndrome is associated with desmoid tumors, lipomas, and fibromas of the mesentery or abdominal wall.
Turcot syndrome: This syndrome involves tumors of the central nervous system (CNS).
Peutz–Jeghers syndrome: This syndrome shows nonneoplastic hamartomatous polyps throughout the gastrointestinal tract and perioral melanin pigmentation.
Juvenile polyposis: These are hamartomas in colon, small bowel, and stomach.
Hereditary Nonpolyposis Colorectal Cancer
The Lynch syndromes, named after Henry T. Lynch, include Lynch I or the colonic syndrome, which is an autosomal-dominant trait characterized by distinct clinical features including proximal colon involvement, mucinous or poorly differentiated histology, pseudodiploidy, and the presence of synchronous or metachronous tumors. Increased survival has been observed in patients despite colon cancer developing before 50 years, with a lifetime risk of cancer approximating 75%. The Lynch II or extracolonic individuals are susceptible to malignancies in the endometrium, ovary, stomach, hepatobiliary tract, small intestine, and genitourinary tract.
The “Amsterdam Criteria” were established to identify potential kindreds and include:
Inclusion of extracolonic tumors and clinicopathological and age modifications were introduced by the “Bethesda Criteria” in 1997. Germline defects in DNA mismatch-repair genes (hMSH2, hMLH1, hPMS1, and hPMS2) have been detected, and resultant microsatellite instability (MSI)
can be identified in virtually all hereditary nonpolyposis colorectal cancer (HNPCC) kindreds and in 15% to 20% of sporadic colon cancers.
The American Cancer Society has developed screening guidelines for the early detection of colon cancer. There are a variety of available early detection tests for colon cancer. Starting at age 50, both men and women should discuss the full range of testing options with their physicians and choose one of the following:
It should be noted that all positive tests should be followed up with colonoscopy. Individuals with a family or personal history of colon cancer or polyps or a history of chronic inflammatory bowel disease should be tested earlier and may need to undergo testing more often.
A virtual colonoscopy or computerized tomographic colonography is an emerging technology in which a spiral computerized tomography (CT) scan of the colon is obtained and three-dimensional images are created and reviewed by a radiologist. A recent study demonstrated comparable sensitivity to conventional colonoscopy (88.7% versus 92.3% for polyps at least 6 mm in dimension). Earlier studies using two-dimensional technology and inexperienced radiologists observed equivocal results. Patients still require bowel preparation and colonic distension as well as ingestion of oral contrast. Additional studies are required before this technique can be used routinely.
Carcinoembryonic antigen (CEA) is not useful for general CRC screening purposes. CEA has a low positive predictive value whereby approximately 60% of cancers are missed.
The K-ras gene is mutated in 50% of CRCs, and its detection in stool represents a potential powerful screening strategy. This is currently an active area of clinical investigation.
Colon carcinogenesis involves progression from hyperproliferative mucosa to polyp formation, with dysplasia, and transformation to noninvasive lesions and subsequent tumor cells with invasive and metastatic capabilities. CRC is a unique model of multistep carcinogenesis resulting from the accumulation of multiple genetic alterations. Stage-by-stage molecular analysis has revealed that this progression involves several types of genetic instability, including loss of heterozygosity, with chromosomes 8p, 17p, and 18q representing the most common chromosomal losses. The 17p deletion accounts for loss of p53 function, and 18q contains the tumor-suppressor genes deleted in colon cancer (i.e., DCC) and the gene deleted in pancreatic 4 (i.e., DPC4). The loss of heterozygosity of chromosome 18q has prognostic significance.
Colon carcinogenesis also occurs as a consequence of defects in the DNA mismatch repair system. The loss of hMLH1 and hMSH2, predominantly, in sporadic cancers leads to accelerated accumulation of additions or deletions in repeating DNA nucleotide units. This MSI contributes to the loss of growth inhibition mediated by transforming growth factor-ß (TGF-ß) due to a mutation in the type II receptor. Mutations in the APC gene on chromosome 5q21 are responsible for FAP and are involved in cell signaling and in cellular adhesion, with binding of ß-catenin. Alterations in the APC gene occur early in tumor progression. Mutations in the protooncogene ras family, including K-ras and N-ras, are important for transformation and also are common in early tumor development.
More than 90% of CRCs are adenocarcinomas, with proximal tumors becoming increasingly more common. Left-sided cancers tend to be annular, leading to obstruction, whereas right-sided cancers are more commonly polypoid and clinically silent. One-third of patients will initially be seen with metastatic disease, whereas 50% will eventually develop metastases.
Signs and Symptoms
The American Joint Committee on Cancer (AJCC) (1) staging of colon cancer using the TNM classification was updated in 2003 (see Fig. 8.1). Patients with stage II and III disease have been further stratified, and vascular or lymphatic invasion has been included (see Table 8.1). The tumor designation, or T stage, defines the extent of bowel wall penetration, as opposed to tumor size. The AJCC staging system accounts for the number of lymph nodes involved as a significant predictor of survival. Four or more positive lymph nodes or gross versus microscopic bowel wall penetration lead to diminished survival. Some patients with stage
II disease exhibit a heterogeneous outcome and are at high risk for relapse, with outcomes similar to those of node-positive patients (seeTable 8.2).
FIG. 8.1. Staging classification of colorectal cancer. Classification is based on modifications of Dukes' system. Stages B3 and C3 (not shown) signify invasion of contiguous organs or structures (T4). Prognosis is also determined by the number of positive lymph nodes: more than four (N2) lymph nodes predicts a worse outcome than one to three (N1) lymph nodes, and a poor histopathological differentiation, vascular or lymphatic invasion, and a positive preoperative CEA value of >5 ng per mL implies a worse outcome. According to the revised TNM classification system, stage I equals T1 or T2 N0 (Dukes' stage A and B1); stage II equals T3 or T4 N0 (Dukes' stage B2 and B3); stage III equals any T plus N1, N2, or N3 (Dukes' stage C1, C2 and C3); and stage IV equals any T any N plus M1 (Dukes' stage D).
TABLE 8.1. American Joint Committee on Cancer (AJCC) Staging Classification (2003)
TABLE 8.2. Five-year Survival Prognosis by Stage
Adverse Prognostic Factors
The adverse prognostic factors include:
Adjuvant Chemotherapy for Colon Cancer
This large Intergroup trial of 5-FU and levamisole (Lev) is of historic importance because it reported a 41% reduction in the relapse rate and a 33% decrease in overall cancer mortality (4). This study resulted in the National Institutes of Health consensus panel recommending that 5-FU–based adjuvant therapy be administered to all patients with resected stage III colon cancer.
Intergroup 0089 is a landmark study that randomized 3,759 patients with stage II or III disease to one of four arms (5). The results demonstrated that the 5-FU and leucovorin (LV)–containing schedules (Mayo Clinic and Roswell Park) were equivalent and that the three-drug combination
only increased toxicity. The 5-FU and Lev arm was effective but required 12 months of treatment versus the 6-month schedules of the 5-FU and LV arm.
The 5-year DFS and overall survival (OS) for each of the four arms in the study were as follows:
A European study of 2,219 patients with stage II (40%) and III (60%) disease who were treated with infusional 5-FU with LV modulation versus the same combination with oxaliplatin (FOLFOX4) every 2 weeks for 6 months (6) demonstrated a 3-year DFS benefit favoring the FOLFOX4 combination [78.2% for 5-FU and LV versus 72.9% for FOLFOX4, hazard ratio (HR) 0.77; 95% CI, 0.65 to 0.92, p = 0.002], although the OS between the two arms was not statistically different when the study was reported. A 3-year disease-free endpoint was chosen because a recent retrospective analysis of more than 17,400 patients demonstrated that the 3-year disease-free endpoint is equivalent to the conventional 5-year OS benchmark. Treatment with FOLFOX4 was well tolerated, with 41% patients having grade 3 and 4 neutropenia, with only 0.7% being associated with fever. Anticipated peripheral neuropathy or paresthesias were observed (grade 2–32% and grade 3–12%) but was almost entirely resolved 1 year later (grade 2–5% and grade 3–1%).
A 1987 study of patients with stage III disease compared capecitabine (1,250 mg per m2 b.i.d. for 14 days, every 3 weeks) with the Mayo Clinic bolus of 5-FU and LV (7). The study was designed to demonstrate equivalency, with a primary endpoint of 3-year DFS. The capecitabine arm demonstrated a trend toward superiority in this endpoint (64.2% versus 60.6%, HR 0.87; 95% CI, 0.75 to 1.00, Log-rankp = 0.0526).
CALGB 89803 was a study of irinotecan with bolus 5-FU and LV (IFL) versus weekly 5-FU in patients with stage III disease (8). Increased grade 3 and 4 neutropenia and early deaths were observed in the experimental arm, and a higher number of patients withdrew from the study. Overall, IFL was not better than the 5-FU and LV arm. The use of the IFL regimen in the adjuvant setting cannot be recommended at present. An important study of irinotecan in combination with infusional 5-FU or FOLFIRI (continuous infusion 5-FU with biweekly irinotecan) has completed accrual of 1,800 patients, and preliminary analyses report no significant increases in treatment-related toxicity.
Adjuvant Chemotherapy Regimens for Colon Cancer
On the basis of these adjuvant chemotherapy studies, the use of one of the following regimens is recommended for patients with stage III colon cancer. The results with capecitabine are promising in the adjuvant setting.
5-fluorouracil and Calcium Leucovorin:
The toxicity profile of these two regimens differs. Myelosuppression and oral mucositis are more common with the daily Mayo Clinic regimen, whereas diarrhea may be more severe with the weekly schedule. Cryotherapy with ice held in the mouth during the 5-FU infusion may help lessen the mucositis associated with the therapy.
Oxaliplatin, 5-fluorouracil (Infusional and Bolus), and Leucovorin (FOLFOX4):
FOLFOX6, which omits the day 2 bolus 5-FU and LV, uses 2,400 mg per m2 of continuous 5-FU over 46 hours and appears to have activity equivalent to that of FOLFOX4 in the advanced disease setting. The oxaliplatin dose is increased to 100 mg per m2, but ongoing studies are evaluating the 85 mg per m2 dose.
Adjuvant Chemotherapy for Stage II Colon Cancer
Despite the 75% 5-year survival with surgery alone, some patients with stage II disease have a higher risk of relapse, with outcomes being similar to those of node-positive patients. Adjuvant chemotherapy provides up to 33% OS advantage, resulting in an absolute treatment benefit of approximately 5%.
Several analyses have reported varying outcomes in patients with stage II disease who received adjuvant treatment:
Immunotherapy (Edrecolomab, 17-1A)
Riethmuller et al. (13) treated 189 patients postoperatively with the monoclonal antibody edrecolomab, resulting in a 27% decrease in recurrence rate and a 30% reduction in mortality
rate. In a comparison of patients with stage III disease treated with 5-FU–based chemotherapy without or with edrecolomab, a survival advantage was suggested for the latter (HR 0.785; 95% CI, 0.638 to 0.967). The results from a study of edrecolomab in patients with stage II disease are pending.
Adjuvant Treatment for Rectal Cancer
In contrast to colon cancer, treatment failures after potentially curative resections tend to occur more locally in 10% to 18% of patients. Combined-modality adjuvant chemotherapy and radiation therapy is now the standard therapy for patients with stages II and III rectal cancer (T3, T4, and nodal disease N+).
A four-arm study of 1,695 patients compared 5-FU alone, 5-FU and LV combination, 5-FU and Lev combination, and 5-FU and LV and Lev combination (14). Two cycles of chemotherapy was administered before and after chemotherapy in combination with external beam radiation (50.4 Gy to 45 Gy with 5.4 Gy boost). The chemotherapy during the radiation was given as a bolus with or without LV. The DFS and OS was similar in all treatment arms, leading to the conclusion that 5-FU alone was as effective as other combinations.
Both DFS and OS advantages were observed in patients receiving continuous infusion of 5-FU during radiation when compared with those receiving bolus 5-FU (15). This survival benefit has led to continuous infusion 5-FU during radiation being considered as the standard.
Adjuvant Combined-modality Regimens for Rectal Cancer
FOLLOW-UP AFTER ADJUVANT TREATMENT
Eighty percent of recurrences are seen within 2 years of initial therapy. The American Cancer Society recommends total colonic evaluation with either colonoscopy or double-contrast barium
enema within 1 year of resection, followed every 3 to 5 years if findings remain normal. Synchronous cancers must be excluded during initial surgical extirpation, and metachronous malignancies in the form of polyps must be detected and excised before more malignant behavior develops.
TREATMENT FOR ADVANCED DISEASE
Single-arm phase II studies of 5-FU and LV chemotherapy regimens in advanced CRC have reported response rates ranging from 0% to 70%, but most larger studies have observed objective response rates of 15% to 20%, with median survival of 8 to 12 months.
Continuous Infusion of 5-fluorouracil
The efficiency of continuous infusion of 5-FU may be equivalent to or slightly better than that of bolus 5-FU and LV and is generally well tolerated despite the inconvenience of a prolonged intravenous infusion apparatus (16,17). 5-FU at 300 mg/m2/day is infused continuously by an ambulatory infusion pump. Toxicities include mucositis; however, myelosuppression is less common. Palmar–plantar erythrodysesthesia (hand–foot syndrome) is common and may respond to pyridoxine, 50 to 150 mg/m2/day. Continuous infusions of 5-FU may have modest activity in patients who have progressed on a bolus 5-FU regimen.
Oxaliplatin is an agent that differs structurally from other platinums in its 1,2-diaminocyclohexane (DACH) moiety. At doses resulting in equivalent cytotoxicity, oxaliplatin forms fewer DNA adducts than does cisplatin, suggesting that oxaliplatin lesions are more lethal than cis-platinum adducts. Oxaliplatin exhibits synergy with 5-FU because increased response rates as high as 66% have been observed even in patients who are refractory to 5-FU. Despite its unique toxicities (i.e., reversible peripheral neuropathy, laryngopharyngeal dysesthesias, and cold hypersensitivities), oxaliplatin lacks the emetogenic and nephrogenic toxicities of cisplatin.
Oxaliplatin was approved for second-line therapy in metastatic patients on the basis of a study comparing FOLFOX4 with oxaliplatin alone and with infusional or bolus 5-FU and LV. In this study, response rate, time to progression, and relief of tumor-related symptoms were improved with FOLFOX4 when compared to the other treatment arms. Despite the improved time to progression, the OS difference was not statistically significant (9.8 versus 8.7 and 8.1 months, respectively).
The North Central Cancer Treatment Group (N9741) conducted a trial comparing first-line FOLFOX4 versus IFL versus IROX (irinotecan in combination with oxaliplatin). The study, designed by Richard Goldberg et al. (18), originally consisted of six arms, but three were eliminated on the basis of changes in the standard of care or toxicity. In addition, a higher 60-day mortality was detected in
the IFL arm, resulting in a dose reduction to 100 mg per m2. The response rate, time to progression, and OS were significantly better in the FOLFOX4 arm than in the IFL arm. Interestingly, survival in the patients on IROX regimen was better than in the IFL-treated patients and was not significantly different from those on the FOLFOX combination. Imbalances in the second-line chemotherapy administered to patients in this study may confound the survival differences. Approximately 60% of the oxaliplatin failures were treated with irinotecan, whereas only 24% of patients who are refractory to irinotecan received oxaliplatin. In addition, the study was not designed to address the effect of infusional 5-FU. The observed toxicities in the study were reflective of the specific drug combinations and included grade 3 or higher paresthesias (18%) in the FOLFOX arm and a 28% incidence of diarrhea in the IFL arm. Despite a higher degree of neutropenia (50% in FOLFOX versus 40% in IFL) with FOLFOX, febrile neutropenia was significantly greater in the IFL arm (15% with FOLFOX versus 4% with IFL). IROX also exhibited significant toxicities. Oxaliplatin has been approved by the FDA for use in the first-line treatment of patients with metastatic CRC largely on the basis of this study.
Although FOLFOX is clearly a superior regimen compared to IFL, the use of infusional 5-FU with irinotecan may produce results similar to those seen using FOLFOX. Tournigand et al. reported an equivalent median survival of 21.5 months with FOLFIRI followed by FOLFOX and a median survival of 20.6 months with the opposite sequence (p = 0.99) (19). The conclusion is that similar survival is observed in patients receiving either sequence. Other investigators have suggested that the use of all active agents results in the best survival in patients with advanced CRC.
Irinotecan is a topoisomerase I–targeting agent, with activity in patients with advanced CRC and in patients deemed refractory to 5-FU. Response rates as high as 20% are observed, and an additional 45% of patients achieve disease stabilization. Significant survival advantages have been shown for irinotecan as second-line therapy after 5-FU compared with supportive care or with continuous-infusion 5-FU regimens. Several schedules are typically administered with and without 5-FU:
Delayed-onset diarrhea is common and requires close monitoring and aggressive management (high-dose loperamide, 4 mg initially and then 2 mg every 2 hours until diarrhea stops for at least 12 hours). Neutropenia and mild nausea and vomiting also are common. This combination of toxicities can be severe and life threatening, which was evident in a large phase III study, NCCTG 9741, comparing irinotecan and oxaliplatin combination regimens (see subsequent text). A higher 60-day mortality was observed (4.5% versus 1.8%), and the dose of the irinotecan weekly regimen reduced to 100 mg per m2.
Capecitabine, the oral fluoropyrimidine prodrug, undergoes a series of three enzymatic steps in its conversion to 5-FU. The final enzymatic step is catalyzed by thymidine phosphorylase, which is overexpressed in tumor than in normal tissues. Subsequently, the tumor tissue achieves higher concentrations of 5-FU, with thymidine phosphorylase preferentially activating the tumor tissue. Two phase III studies have compared single-agent capecitabine to the Mayo Clinic 5-FU
and LV regimen and demonstrated higher response rates for the former but equivalent time to progression and median survival (24). The toxicity profile favored the capecitabine arm because decreased gastrointestinal and hematologic toxicities and fewer hospitalizations were observed in this arm. An increased frequency of hand–foot syndrome and hyperbilirubinemia were noted with capecitabine. Early phase II studies with capecitabine in combination with either oxaliplatin or irinotecan demonstrated promising response rates as high as 50% to 65%, and several large phase III studies are ongoing.
Bevacizumab (BV) is a recombinant humanized anti–vascular endothelial cell growth factor (anti-VEGF) monoclonal antibody with amino acid sequence similarity of 97% to that of human IgG1. BV blocks VEGF-induced angiogenesis with an exceptionally high affinity for VEGF. One of the initial trials with BV in untreated CRC patients combined BV (doses of either 5 or 10 mg per kg every 2 weeks) with weekly 500–mg per m2 dose of 5-FU and LV. Interestingly, a 40% response rate and 21.5-month median survival was observed in the 5–mg per kg cohort. The major toxicities included thrombosis (13 patients with three treatment discontinuations and one patient death), proteinuria, and hypertension. Updated toxicity data reveals that full-dose anticoagulation can be administered with BV and that there is no increased risk of deep venous thrombus formation. The intriguing results presented by Hurwitz et al. (25) also reports a higher response rate (45% versus 35%, p<0.0029) and a longer median survival (20.3 versus 15.6 months) when BV is combined with IFL. A large phase III trial combining BV with FOLFOX as second-line treatment is awaited. BV has been approved by the FDA, largely on the basis of the results of these trials, for the treatment of patients with advanced CRC in combination with any intravenous 5-FU–based regimen.
Cetuximab is a chimerized IgG1 antibody that prevents ligand binding to the epidermal growth factor receptor (EGFR) and its heterodimers. Cetuximab exhibits higher affinity (subnanomolar) or approximately 1-log greater binding than the natural ligands for EGFR. The agent blocks receptor dimerization, tyrosine kinase phosphorylation, and subsequent downstream signal transduction. Cetuximab is administered at a dose of 400 mg per m2 in the first week and then at a 250 mg per m2 dose each week thereafter.
In a study with patients refractory to irinotecan who were treated with the combination of cetuximab and irinotecan versus cetuximab alone, improvements in the response rate (22.9% versus 10.8%, 0.0074) and time to progression (4.1 versus 1.5 months, <0.0001) were reported (26). Despite manageable toxicity, no improvements in survival outcomes were observed. A correlation between the intensity of the skin rash and median survival was noted. Cetuximab has been approved by the FDA for the treatment of patients with EGFR-positive advanced CRC that is refractory to or intolerant of irinotecan.
CHEMOTHERAPY REGIMENS FOR METASTATIC COLORECTAL CANCER
CEA is an acid glycoprotein localized to the cell membrane, facilitating release into blood and surrounding body fluids. CEA is elevated in nonneoplastic processes such as smoking and inflammatory bowel disease and in cancers involving the breast, lungs, or pancreas. The degree of tumor differentiation correlates with CEA expression; up to 30% of colon cancers, in particular poorly differentiated tumors, exhibit no CEA elevation. Elevation is typically defined as a concentration >5 ng per mL and is associated with increased recurrence rate and decreased survival. A measurement of concentration >25 ng per mL is highly suggestive of metastatic disease. In patients preoperatively evaluated with CEA measurements, a sensitivity of 43% and a specificity of 90% were reported.
Persistent elevation of CEA postoperatively may suggest residual tumor or early metastasis. The routine use of CEA alone for evaluating treatment response is not recommended because up to 20% of patients exhibit conflicting declines in CEA levels despite disease progression. Patients with initially negative levels of CEA can subsequently exhibit positive levels. Serial CEA measurements after completion of postoperative chemotherapy may identify patients who are eligible for a curative resection, in particular, patients with a solitary liver or lung metastasis, but this is rare. Data from studies such as the Ohio State study report a 31% 5-year survival in select populations with aggressive CEA surveillance and second-look laparotomy to detect early recurrences. A joint National Institutes of Health (NIH) and U.K. trial failed to demonstrate survival differences with serial CEA measurements. The American Society of Clinical Oncology recommends CEA testing in patients with previous CRC diagnoses who would be eligible or be considered for surgical resection.
The liver is the most common site for metastasis, with one third of instances involving only the liver; liver is involved in two-thirds of patients dying from colon cancer. Approximately 25% of liver metastases are resectable, with certain patient subsets showing 30% to 40% 5-year survival after resection and 3% to 5% operative morbidity and mortality. Intraoperative ultrasound is the most sensitive test for initial detection, followed by CT scan or magnetic resonance imaging (MRI).
Nonresectable patients with disease limited to the liver can be treated with locoregional [by hepatic artery infusion (HAI)] or systemic chemotherapy. Postoperative chemotherapy trials have exhibited some benefit. Kemeny et al. (27) reported a 4-year DFS and hepatic disease-free benefit in patients with resected liver metastases who had received intraarterial floxuridine with systemic 5-FU compared to those who did not receive any postoperative therapy, although there was no statistically significant difference in OS (62% versus 53%, p = 0.06).
The feasibility of converting an initially unresectable disease to a potentially curative disease has been investigated by Bismuth and colleagues (28). Metastectomy was possible in 99 patients with either down-staged or stable disease, and the 3-year survival was encouraging (58% for responders, 45% for patients with stable disease). Similar observations have been reported by Alberts
using preoperative FOLFOX4 on 41% of patients undergoing resection with an observed median survival of 31.4 months (95% CI, 20.4 to 34.8) for the entire cohort (29).