Screening for Colorectal Cancer: Analysis of Evidence and Cost-Effectiveness

Colorectal cancer causes significant morbidity and mortality and remains the second most common cause of cancer-related death in the United States. A number of screening approaches have been shown to decrease colon cancer–related mortality and are cost-effective. In this review, we discuss available screening approaches and suggest strategies to implement effective screening using the current tools available. Colorectal cancer (CRC) is one of the most prevalent cancers in the United States, affecting 1 in 18 Americans [1]. Half of the affected patients will die from the disease because of tumor spread, resulting in more than 56,000 deaths per year. It is the second most common cause of cancer-related death [2]. The economic burden of CRC is large, with the annual direct costs for CRC estimated at $4.8 billion in 1998, making it the third most costly gastrointestinalrelated illness behind gastroesophageal reflux (GERD) and gallbladder disease [3]. Indirect costs, such as loss in productivity and the intangible costs of pain and suffering, are difficult to measure but are most likely substantial and probably higher in CRC than in nonmalignant conditions. The greatest predictor of prognosis is the stage at which the cancer is detected [4]. Screening can detect earlier-staged cancers and thus extend survival [5,6]. Although CRC screening tests are widely available to the U.S. population at risk, as few as 30% of eligible patients have undergone screening [7], a percentage much lower than that for other organ-based screening procedures. Acombination of patient resistance, the sluggish acceptance of screening by health care systems, and slow implementation by physicians is likely responsible for this disturbingly low adherence rate [8]. Indeed, Medicare did not begin covering colorectal screening modalities until 1998 and only recently began covering screening colonoscopies in July 2001. Another reason for slow implementation of CRC screening may be due to multiple modalities for screening. The adenoma-to-carcinoma transformation sequence in colorectal carcinogenesis is well established and accepted; detection and removal of adenomas reduces the incidence of colon cancer, and finding earlier-staged cancers improves survival [6]. Screening for CRC detects cancer or its precursors in asymptomatic individuals. The ideal screening test should be sensitive, specific, affordable, as well as acceptable to asymptomatic individuals. The overall goal of screening is to reduce morbidity and mortality. The tests that have been evaluated for CRC screening include fecal occult blood testing (FOBT) and flexible sigmoidoscopy. Trials looking into the validity of colonoscopy, doublecontrast barium enema (DCBE), fecal genetic markers, and virtual colonoscopy are lacking. Serum tumor markers have failed to show any benefit as a screening tool for CRC [9]. The American Cancer Society (ACS) guidelines for CRC screening in asymptomatic adults greater than 50 years of age give 5 options: annual FOBT; flexible sigmoidoscopy every 5 years; annual FOBT plus flexible sigmoidoscopy every 5 years; DCBE every 5 years; or colonoscopy every 10 years [10]. In this review, we attempt to summarize the existing data that led to the current recommendations for population-based screening. Our aim is to aid the reader in choosing the appropriate screening modality. Throughout this review, we will focus on recommendations for the average-risk individual (a man or woman 50 years of age or older with no personal history of CRC, with no first-degree relative affected with polyps or colorectal neoplasia, and without a history of inflammatory bowel disease). For individuals with a higher risk of developing colon cancer, recommendations for screening and surveillance intervals as well as age of initiation of screening are different and beyond the scope of this review. Fecal Occult Blood Test The most comprehensive literature evaluating colorectal screening tools stems from trials that test for occult blood in the stool. In normal feces, approximately 0.7 mL of blood per day on average is excreted. This increases to 1 to 2 mL/day with early colorectal neoplasms. Blood loss appears to correlate with tumor size, with larger lesions shedding more blood [11,12]. Detection is also related to the location of the neoplasm. Stool From the University of California, San Diego, and the San Diego VA Healthcare System, San Diego, CA. is more liquid in the right colon where blood and stool mix and may result in the degradation of hemoglobin. With left-sided lesions, blood is deposited on the surface of more solid stool, where it can be sampled more easily and detected by FOBT. The detection of occult blood in the stool is based on the oxidation capacity of the heme moiety of hemoglobin, which turns the guaiac on the stool card blue. A variety of agents can lead to false-positive FOBT results, the most important being red meat and NSAIDs. Vitamin C, on the other hand, may interfere with the sensitivity of FOBT, leading to a falsenegative result. In general, the patient being screened avoids the agents listed above, scrapes stool from 3 consecutive fecal passes, places the sample on the FOBT card, and mails the card to their physician or laboratory. The cost of a 3-sample test based on Medicare reimbursement rates for 2001 is about $5 for the Hemoccult II test [13]. Multiple large population-based screening studies reveal that overall, 1% to 2.4% of patients older than age 50 years who underwent the screening were FOBT-positive [14–16]. Of the FOBT-positives, 22% to 58% had an adenoma or carcinoma identified with colonoscopy or barium enema and sigmoidoscopy. Of importance was the finding that earlystaged cancers (Dukes A and B) were increased significantly following FOBT screening in each trial as compared with the unscreened controls. In an attempt to improve on the sensitivity of FOBT, Allison et al applied 2 FOBT methods of detection of occult blood consecutively to one stool sample. In their protocol, the Hemoccult Sensa II, a more sensitive guaiac test than the Hemoccult II, was followed by the HemeSelect test, an immunochemical test specific to human hemoglobin. This study’s combined testing did improve the sensitivity (to above 60%) without reducing the specificity in testing for fecal occult blood [17]. This approach could be improved upon further in the future if FOBT remains a component of CRC screening recommendations. There are a number of studies that show that FOBT reduces CRC mortality. A retrospective case-controlled study conducted at the Kaiser Permanente Medical Care Program showed a 25% reduction in CRC mortality after FOBT screening compared with sex-matched controls [18]. A clear benefit was shown when screening every 1 to 2 years, but no decrease in mortality could be found at longer intervals. The “Minnesota” trial of 46,551 patients found a decreased mortality from colon cancer by 33%, but only if screened yearly [16]. Mortality in individuals screened with FOBT every 2 years was similar to controls, emphasizing the need for yearly testing with FOBT [16]. The authors rehydrated the FOBT cards, which increased the number of positive FOBT tests from 1–2% to 10%. The higher rate of (rehydrated) positive FOBTs resulted in 38% of their enrolled patients undergoing colonoscopy over the 14-year course of the study, a number that may be too high for general population screening. In a randomized prospective trial from the Memorial Sloan-Kettering Cancer Center, 5806 patients underwent rigid sigmoidoscopy, half of whom were additionally tested with nonrehydrated FOBT. The FOBT testing resulted in a 43% reduction in CRC [19]. In 2 large randomized trials published in 1996, an every-2-year screening interval with nonhydrated Hemoccult test resulted in a 15% [15] and 18% reduction in CRC mortality [14]. These numbers are about half of those from the studies with annual FOBT screening. A meta-analysis of the 4 randomized trials concluded that mortality from CRC was reduced by 16% in those randomized to screening [5]. In summary, FOBT can diagnose asymptomatic CRC. It detects early-stage cancer, as 65% to 90% of cancers that are picked up by this screening method are Dukes A or B. With annual FOBT screening alone, CRC mortality can be reduced up to 33%. It is the only screening tool that has been shown by randomized clinical trials to reduce death from CRC. Because of the poor sensitivity of FOBT, as many as 30% to 50% of cancers may be missed if this method is used alone. Flexible Sigmoidoscopy The advantage of flexible sigmoidoscopy over FOBT is the direct visualization of mucosa with the potential to remove polyps. It is performed without sedation and requires a preparation of the colon with either a laxative or an enema. It is considered a safe test when performed by an experienced health care provider [20]. Its cost based on Medicare reimbursement rates is about $99 per procedure [13]. A gastroenterologist, a trained primary care physician or surgeon, or a trained nurse practitioner or physician’s assistant may perform the procedure. The rationale for screening with sigmoidoscopy stems from previous observations that more than 50% of CRCs are left-sided [21], but this may not be the case since the widespread use of colonoscopy and the observation of more right-sided lesions [22–26]. Gilbertsen and Nelms from the University of Minnesota reported on the mortality of 18,158 patients who underwent rigid sigmoidoscopy with removal of all rectal lesions. They noted an 85% reduction in late-stage rectal cancer compared with historical controls [27]. In a study from Kaiser Permanente, a 70% reduction in mortality from rectosigmoid cancers was observed in the group screened within 10 years prior to diagnosis [6]. The lesson put forth is twofold: (1) mortality can be greatly reduced by directly observing the area of the colon and removing precancerous lesions, and (2) screening intervals could be stretched significantly using this technique. Similar findings were reported from the Marshfield Community Health Plan, where a casecontrol study of 66 patients revealed a 79% reduction in death related to CRC [28]. In these studies, the mortality reduction only pertained to cancers in the screened portion of the colon. Findings on flexible sigmoidoscopy might be used to predict advanced right-sided polyps and the need for a colon88 JCOM February 2003 Vol. 10, No. 2 www.turner-white.com COLORECTAL CANCER SCREENING