Colorectal cancer (CRC) is the second leading cause of cancer death in the United States, with projections of 55,000 deaths and 149,000 new cases diagnosed in 2006.¹ About three-quarters of CRC cases are sporadic, apparently resulting from environmental factors, diet, and aging (Table 1). The remaining 25% are familial or inherited. Familial CRC is not well understood and is characterized by increased risk of CRC and an unclear pattern of inheritance. Inherited CRC, on the other hand, exhibits a clear pattern of inheritance and includes hereditary nonpolyposis colorectal cancer (HNPCC) and familial adenomatous polyposis (FAP) as well as the rarer MYH-associated neoplasia, Peutz-Jeghers, and juvenile polyposis syndromes. Determining the type of CRC has important implications for screening and follow-up of patients and family members.

Molecular Characteristics of CRC

The natural progression of CRC from benign adenoma to carcinoma to metastatic disease is associated with 1 of 2 distinct molecular pathways. The most common pathway, chromosomal instability, occurs in 80% to 85% of cases and is characterized by allelic loss (loss of heterozygosity), chromosomal rearrangements, or loss of whole chromosomes. Such characteristics are associated with most sporadic CRCs and FAP. The remaining 15% to 20% are associated with impairment of nucleotide mismatch repair (MMR) that normally occurs during DNA replication or recombination.

In tumor tissue, MMR defects are characterized by microsatellite instability (MSI), defined as insertions or deletions of nucleotides within repeated DNA nucleotide sequences known as microsatellites. The MMR defect pathway is exemplified by HNPCC, an autosomal dominant syndrome also known as Lynch syndrome. Individuals with HNPCC typically inherit 1 copy of a defective MMR allele; subsequent somatic mutations may cause loss of the normal allele, leading to defective DNA repair. Approximately 51% of the mutations occur in MLH1, 38% in MSH2, 10% in MSH6, and 2% in PMS2.⁴

Germline mutation of the adenomatous polyposis coli (APC) gene, a tumor suppressor gene, causes FAP, an autosomal dominant disorder with a prevalence of approximately 1 in 8000.

Screening for CRC

The American Cancer Society (ACS) recommends screening average-risk individuals with one of several options beginning at age 50 years (Table 2).⁵ Screening options should be chosen based on individual risk, personal preference, and access. Of these screening tests, the fecal occult blood test (FOBT) or fecal immunochemical test (FIT) are the only non-invasive tests; FOBT has led to the detection and surgical excision of precancerous polyps thereby significantly reducing the incidence of CRC and related mortality.⁶ Flexible sigmoidoscopy is also associated with reduced mortality for CRC,⁷ and combining FOBT with flexible sigmoidoscopy is more effective than either test alone.⁵ Although less commonly used, double contrast barium enema has the advantage of examining the entire colon, but is less sensitive than colonoscopy in detecting polyps.⁷

The American Gastroenterology Association recommends that high-risk individuals be screened at a younger age and that screening be performed more frequently (Table 2).⁷

Differential Diagnosis of CRC and Risk Assessment

Because HNPCC is the most common of the hereditary CRCs, criteria to identify families with the disorder have been developed by the National Cancer Institute (Bethesda guidelines)⁸ and the International Collaborative Group on HNPCC (Amsterdam criteria).⁹

HNPCC is characterized by a high lifetime cancer risk and early age at onset (mean ~45 years). Because HNPCC progresses rapidly, early detection is critical for affected individuals and their first-degree relatives; close surveillance of affected family members can reduce overall mortality by ~65%.¹⁰ Additionally, women with HNPCC are at high risk for endometrial cancer (40% to 70%) and ovarian cancer (10% to 12%),⁴ and early detection of HNPCC or familial MMR mutations allows close monitoring for these cancers.

FAP is characterized clinically by the presence of hundreds to thousands of colorectal polyps identified at an early age (<30 years). FAP will progress to colon cancer unless colectomy is performed. It is also associated with an increased lifetime risk of other cancers, including duodenal cancer (5% to 11% risk).¹¹

Selection of Therapy for Patients with CRC

Pharmacogenomics, the study of genetic influences on drug response, is becoming increasingly important in the selection of therapeutic agents. Genetic polymorphisms are partially responsible for inter-patient variability in drug efficacy and/or toxicity; detecting such polymorphisms prior to treatment may help optimize drug selection and dosage.

Screening, Differential Diagnosis, and Risk Assessment

• Individuals recommended for CRC screening (Table 2)

• Individuals with CRC who meet any of the Bethesda criteria (Table 3)

• Individuals who meet the Amsterdam II criteria (Table 4)

• First-degree relatives of individuals with a known MMR gene mutation

Determine Prognosis, Select and Monitor Therapy

• Individuals with diagnosed CRC

Table 5 lists tests used to assess risk, screen, diagnose, determine prognosis, and select and monitor therapy for CRC.

FOBT Screening

In 2003 the ACS guidelines for stool blood tests were expanded to include fecal immunochemical tests (FITs).⁵ The traditional guaiac-based FOBTs (eg, Hemoccult^®) suffer from false-positive results due to ingestion of red meat, some raw fruits and uncooked vegetables, non-steroidal anti-inflammatory drugs, and aspirin. Additionally, ingestion of vitamin C (>250 mg/day) from supplements or citrus fruits may lead to false-negative results. Thus, dietary and medication restrictions are required prior to sample collection. Conversely, FITs such as InSure^® are more specific and consequently do not require dietary and medication restrictions.¹³ Furthermore, InSure is more specific for occult bleeding in the colon and rectum thus making it less likely that bleeding is from the upper gastrointestinal tract.¹³ In a recent clinical study comparing InSure with a guaiac-based FOBT, InSure had a better true-positive rate for early stage CRC (92.3% vs 30.8%, n = 13 stage I patients), all stages of CRC (87.5% vs 54.2%, n = 24), and significant adenoma (42.6% vs 23.0%, n = 61).¹⁴  False-positive rates were ~3%.

Positive FOBT or FIT results generally reflect the presence of blood in the stool and may be associated with CRC. The ACS recommends colonoscopy as follow-up to a positive test; flexible sigmoidoscopy or repeat testing are not indicated.⁵ Negative results do not rule out CRC; false-negative results can occur because of uneven distribution of blood in the feces or intermittent bleeding.

Differential Diagnosis of CRC and Risk Assessment of Relatives

Since patients with CRC present with non-specific symptoms (eg, change in bowel habit, unexplained weight loss, abdominal pain, mucous discharge, or rectal bleeding) or with no symptoms, diagnosis is based on colonoscopy and pathologic examination of the suspicious tissue.

Once CRC is diagnosed, it is important to determine if the cancer is hereditary. If HNPCC is diagnosed, or if family members have an HNPCC-associated mutation, increased surveillance is required. The Figure depicts a suggested testing algorithm for the differential diagnosis of CRC and risk assessment of family members. Details regarding these tests follow.

Microsatellite Instability
The diagnosis of HNPCC begins with consideration of MSI testing. Colorectal tumors should be tested for MSI when any of the Bethesda criteria are met (Table 3). In a study of 1222 patients with CRC, combining Bethesda criteria with MSI testing was more effective in determining which patients should be tested for MMR mutations than the use of either alone.¹⁵ The combination resulted in a sensitivity of 81.8%, a specificity of 98.0%, and an overall accuracy of 97.9% for identifying patients with MLH1 or MSH2 mutations. Furthermore, combining Bethesda criteria with MSI testing was more cost effective.

Results are reported as MSI-high (MSI-H), MSI-low (MSI-L), or negative for MSI (microsatellite stable, MSS). A MSI-H result is reported if ≥2 of the 5 National Cancer Institute-recommended markers show instability and requires follow-up with MMR gene mutation testing (Figure). Although an MSI-H result is the hallmark of HNPCC, it is also found in 15% to 20% of sporadic CRC cases.¹¹ An MSI-L result, reflecting instability in 1 marker, is found in <10% of HNPCC cases and in most MSI-positive sporadic CRCs. Since MSI results do not rule out HNPCC, MMR gene mutation testing should be considered regardless of MSI status in families with a strong suspicion of HNPCC.⁸

MMR Gene Mutation Testing
For patients with suspected HNPCC and no known familial mutation, begin MMR gene testing with MLH1 and MSH2. MLH1/MSH2 mutation testing should include DNA sequence changes, which are more commonly identified, and also deletion/duplication mutations, which are found in ~27% of families with HNPCC.¹⁶ If no MLH1/MSH2 mutations are detected, MSH6 mutation testing should be performed (Figure). Detection of a deleterious MMR gene mutation in a patient with CRC is diagnostic of HNPCC. Failure to identify a relevant mutation makes HNPCC unlikely but does not rule it out.

Once a diagnosis of HNPCC is established and an MMR gene mutation is identified, first-degree relatives should be tested for the mutation using a single-exon assay (Table 5; Figure). Relatives who carry the family mutation are at high risk for HNPCC and should be monitored closely (Table 2). For first-degree relatives of HNPCC patients who meet the Amsterdam criteria but do not know the family mutation, MLH1 and MSH2 mutation testing should be performed first; MSH6 mutation testing should then be considered for patients with negative results (Figure). Because not all HNPCC families meet the Amsterdam criteria, MMR mutation testing should also be considered when there is a strong suspicion of HNPCC.^3,16 Detection of an MMR gene mutation is associated with high risk for HNPCC, whereas negative results are consistent with average risk but do not rule out HNPCC.

Determining Prognosis

Tumor Staging

Tumor staging using either the American Joint Committee on Cancer (AJCC) system¹⁷ or the Dukes system has historically been a powerful tool for determining the prognosis of patients with CRC. Tumor staging reflects the extent of CRC and provides prognostic information, as demonstrated by the data in Table 6. The data were derived by correlating AJCC stage with survival in more than 119,000 colon cancer patients.¹⁸ Interestingly, stage IIIa CRC had a better prognosis than stage IIb (P <0.001); this may be due to the current practice of initiating chemotherapy for patients with stage III but not stage II disease.¹⁸ The Dukes stage that corresponds to the AJCC stage is provided to give the reader estimated survival times for the various Dukes stages.

Microsatellite Instability
Tissue testing for MSI may be used to assess prognosis independent of CRC tumor stage. In a meta-analysis combining the results from 32 studies including 7,642 cases, overall survival was more favorable with MSI than with MSS tumors (hazard ratio [HR] = 0.65, 95% confidence interval [CI] 0.59–0.71).¹⁹ The survival advantage of patients with MSI tumors was maintained when HRs were pooled for stage II and stage III CRC patients (n = 2,935). However, the use of microsatellite status to assess prognosis has not been investigated in a prospective clinical trial.

DNA Cell Cycle Analysis
DNA cell cycle analysis is used to determine ploidy, specifically aneuploidy (an abnormal complement of chromosomes), in CRC tissue. Because of conflicting data on the association of aneuploidy with a poor outcome, the American Society of Clinical Oncology (ASCO) does not recommend routine testing for ploidy.²⁰

p53 Oncoprotein
Overproduction of p53 oncoprotein serves as a surrogate marker of p53 oncogene mutation, which is a common event in tumorigenesis. Although some studies have found an association between p53-positive tumor tissue and poor prognosis in patients with CRC, results have been inconsistent.²⁰

Carcinoembryonic Antigen (CEA)
Elevated pretreatment CEA levels (>5 ng/mL) are associated with poor prognosis and predict eventual tumor recurrence.²⁰

Selecting and Monitoring Therapy

Carcinoembryonic Antigen (CEA)

A CEA level, if elevated preoperatively, can be used to detect residual disease following curative intent surgery in patients with primary CRC. If tumor removal was complete, the CEA level should return to normal within ~6 weeks following surgery; persistently elevated levels suggest residual or metastatic disease.²¹

Approximately 50% of patients who undergo surgery with curative intent develop recurrent or metastatic disease.²¹ Serial CEA monitoring postsurgery is useful for detecting these conditions: the sensitivity and specificity are ~80% and ~70%, respectively.²¹ Sensitivity is higher (~100%) for detecting liver metastasis, which accounts for about 80% of CRC recurrences, than for detecting locoregional recurrence (sensitivity ~60%).²¹ ASCO recommends testing every 3 months for at least 3 years following diagnosis of stage II or III disease, providing the patient is a candidate for further surgery or systemic therapy.²⁰ While stable or falling CEA levels suggest no disease progression, elevated levels, if confirmed by retesting, are associated with disease progression and warrant reevaluation for recurrent and metastatic disease.

Stage IV CRC (distant metastases) and locoregional recurrence may be treated with surgical resection, chemotherapy, or radiation, depending on the site and extent of the metastases or recurrence. In this setting, CEA levels are useful for evaluating the success of surgical removal of the metastasis and for monitoring chemotherapy. ASCO considers CEA to be the marker of choice for monitoring chemotherapy and recommends measurement before treatment and every 1 to 3 months during treatment.²⁰ While decreases in CEA levels during chemotherapy suggest a favorable treatment response, persistently rising values above pretreatment levels suggest disease progression. Rising values should prompt reevaluation and consideration of alternative treatment.^20,21 However, transient increases in CEA can occur with chemotherapy that are not associated with disease progression (eg, within 2 weeks following 5-FU-based treatment and within 4 to 6 weeks after oxaliplatin therapy).^20,22 Thus, timing of sample collection for CEA determination should be considered in context with the therapy prescribed.

Although there is no universally accepted definition of what constitutes a clinically significant change in CEA levels, guidelines have been proposed: 1) ≥30% increase over the previous value, confirmed by a second sample collected within 1 month; or 2) >15% increase maintained over ≥3 successive samples.²¹

Since ~25% of patients with CRC do not have elevated levels of CEA, monitoring treatment with alternative tumor markers such as CA 19-9 may be of benefit.²¹ ASCO does not recommend the routine use of this marker, however.²⁰

Epidermal Growth Factor Receptor (EGFR)

Cetuximab (Erbitux^®) is a recombinant, human/mouse chimeric monoclonal antibody that blocks signal transduction and cell growth when bound to the extracellular domain of EGFR. Thus, EGFR expression was assumed to be a prerequisite for response to cetuximab, and the clinical trials conducted to demonstrate drug efficacy required a positive EGFR immunohistochemical (IHC) stain for patient inclusion. Later studies, however, showed that patients with EGFR-negative tumors might also respond.^23,24 Nevertheless, the Erbitux package insert states that a positive EGFR IHC stain is a prerequisite for use of the drug.²⁵ Clinical trials have shown no correlation between efficacy of treatment and the intensity (ie, 1+ to 4+) of EGFR IHC staining.²⁶

Determination of EGFR gene copy number by FISH or EGFR protein expression by ELISA (using serum rather than tissue) have been proposed as alternatives to IHC testing. Although preliminary results suggest FISH results predict response to cetuximab-based therapy,²⁷ confirmatory studies are required.

Pharmacogenomic Testing
The goal of pharmacogenomic testing is to improve patient outcome by enabling optimal patient-specific drug selection and dosing. Theoretically, genetic markers can identify patients who will 1) respond to a specific medication using standard doses, 2) respond only with increased doses, or 3) have a toxic reaction that either requires a reduced dose or selection of an alternative therapy. To date, most studies correlating genetic markers with CRC treatment outcomes are retrospective, derived from small sample sizes, and often originate from a single research group. The information presented below and in Table 7 summarizes the available literature and classifies the clinical utility of each test.

5-Fluorouracil (5-FU)
DPYD Polymorphism
Deficiency of dihydropyrimidine dehydrogenase (DPD), the rate-limiting enzyme in the catabolism of pyrimidine-based chemotherapeutic agents (eg, 5-FU and capecitabine), has been linked to severe myelosuppression and death in patients treated with standard doses.²⁸ Approximately 50% of DPD-deficient patients have 1 or 2 copies of a DPYD allele containing the IVS14+1G→A mutation; roughly one-third of patients with grade 3 or 4 toxicity due to 5-FU treatment have this mutation.^28-30

Consequently, presence of this mutation suggests an increased risk of severe myelosuppression in patients treated with 5-FU; however, a negative test result (lack of mutation) does not rule out this risk.

Irinotecan
UGT1A1 Polymorphism
Irinotecan (Camptosar^®) therapy can result in dose-limiting toxicity manifesting as neutropenia, diarrhea, and asthenia. The risk of toxicity can be assessed by testing for an additional TA repeat in the promoter region of the gene encoding uridine diphosphate glucuronosyltransferase 1A1 (UGT1A1).^31,32 This hepatic enzyme metabolizes SN-38, the active form of irinotecan and the cause of drug toxicity.

The presence of an additional TA repeat (ie, positive for TA repeat) is consistent with reduced UGT1A1 enzyme activity and SN-38 metabolism, leading to increased likelihood of irinotecan toxicity. Consequently, the irinotecan product insert suggests a reduced initial dose for patients homozygous for the TA repeat.³³ Heterozygous patients have intermediate enzyme activity and may be at increased risk for neutropenia; however, such patients have been shown to tolerate normal initial doses.³² Patients negative for the TA repeat are the least likely to suffer from dose-limiting toxicity. The UGT1A1 TA repeat assay does not detect other mutations in the UGT1A1 gene that may affect UGT1A1 enzyme activity.