SCIENTIFIC REVIEW AND CLINICIAN’S CORNER CLINICAL APPLICATIONS Colorectal Cancer Screening Scientific Review Judith M. E. Walsh, MD, MPH Context Screening for colorectal cancer clearly reduces colorectal cancer mortality, Jonathan P. Terdiman, MD yet many eligible adults remain unscreened. Several screening tests are available, and OLORECTAL CANCER IS THE various professional organizations have differing recommendations on which screen- second leading cause of can- ing test to use. Clinicians are challenged to ensure that eligible patients undergo co- cer death in the United States. lorectal cancer screening and to guide patients in choosing what tests to receive. In women, it ranks third af- Objective To critically assess the evidence for use of the available colorectal cancer Cter lung and breast cancer, and in men, screening tests, including fecal occult blood tests, sigmoidoscopy, colonoscopy, double- it ranks third after lung and prostate contrast barium enema, and newer tests, such as virtual colonoscopy and stool-based cancer. Incidence and mortality from molecular screening. colorectal cancer are similar in both Data Sources All relevant English-language articles were identified using PubMed men and women. In 2001, an esti- (January 1966-August 2002), published meta-analyses, reference lists of key articles, mated 135400 cases were diagnosed and expert consultation. and an estimated 56700 deaths oc- Data Extraction Studies that evaluated colorectal cancer screening in healthy in- curred in the United States.1 dividuals and assessed clinical outcomes were included. Evidence from randomized con- Death from colorectal cancer is pre- trolled trials was considered to be of highest quality, followed by observational evi- dence. Diagnostic accuracy studies were evaluated when randomized controlled trials ventable. Effective, safe, and relatively and observational studies were not available or did not provide adequate evidence. inexpensive methods for screening for Studies were excluded if they did not evaluate colorectal screening tests and if they the disease have been available for de- did not evaluate average-risk individuals. cades, and screening is championed by Data Synthesis Randomized controlled trials have shown that fecal occult blood a large number of public, private, and testing can reduce colorectal cancer incidence and mortality. Case-control studies have professional organizations (Multidis- shown that sigmoidoscopy is associated with a reduction in mortality, and observa- 2 ciplinary Expert Panel, US Preven- tional studies suggest colonoscopy is effective as well. Combining fecal occult blood tive Services Task Force,3 the Ameri- testing and sigmoidoscopy may decrease mortality and can increase diagnostic yield. 4 can College of Gastroenterology, and Conclusion The recommendation that all men and women aged 50 years or older 5 the American Cancer Society ). This ar- undergo screening for colorectal cancer is supported by a large body of direct and in- ticle provides practicing physicians with direct evidence. At present, the available evidence does not currently support choos- an evidence-based review of the litera- ing one test over another. ture of the current status of colorectal JAMA. 2003;289:1288-1296 www.jama.com cancer screening, including the meth- ods of screening that are likely to be barium enema or colonoscopy with the Author Affiliations: Division of General Internal Medi- widely available within the next sev- term screening. To supplement the cine, Departments of Medicine and Epidemiology and eral years. Biostatistics (Dr Walsh) and Division of Gastroenter- search, we reviewed published meta- ology, Department of Medicine (Dr Terdiman), Uni- analyses and personal files and bibliog- versity of California, San Francisco. METHODS Corresponding Author and Reprints: Judith M. E. raphies from published articles, con- In addition to reviewing published sys- Walsh, MD, MPH, Women’s Health Clinical Re- ducted hand searches, including the search Center, University of California San Francisco, tematic reviews, we conducted a litera- Campus Box 1793, 1635 Divisadero Suite 600, San proceedings of recent national profes- ture search using PubMed for English- Francisco, CA 94115 (e-mail: [email protected] sional organization meetings, and con- .edu). language articles from January 1966 Financial Disclosure: Dr Terdiman is on the scientific sulted with experts on colorectal can- through August 2002. For the search, we advisory board at Exact Laboratories, Maynard, Mass. cer screening. Eligibility criteria for Scientific Review and Clinical Applications Section combined search terms colorectal neo- Editor: Wendy Levinson, MD, Contributing Editor. articles that evaluated colorectal screen- plasm or occult blood or sigmoidoscopy or We encourage authors to submit papers to “Scien- ing tests, such as fecal occult blood tests tific Review and Clinical Applications.” Please con- (FOBTs), sigmoidoscopy, colonos- tact Wendy Levinson, MD, Contributing Editor, JAMA; See also p 1297 and Patient Page. phone: 312-464-5204; fax: 312-464-5824; e-mail: copy, and double-contrast barium en- [email protected]. 1288 JAMA, March 12, 2003—Vol 289, No. 10 (Reprinted) ©2003 American Medical Association. All rights reserved. COLORECTAL CANCER SCREENING ema, were that they include asymptom- of cancers arise.6-9 The majority of these tion for colorectal cancer death rang- atic subjects (hence, screening) who adenomas are polypoid growths. But as ing from a low of 0.8 per 1000 person- were at average-risk and that informa- many as 20% to 30% of adenomas are flat years with biennial screening in the UK tion be provided on clinical outcomes. or depressed, which make them more study to a high of 4.6 per 1000 person- The clinical outcomes included mortal- difficult to detect and remove.10,11 years with annual screening in the Min- ity, cancer incidence, and identifica- The optimal means to prevent colo- nesota trial. Recent 18-year follow-up tion of adenomas. Studies to be of the rectal cancer remain uncertain. Evi- data from the Minnesota trial demon- highest quality were randomized con- dence for the efficacy of the com- strate that annual and biennial serial trolled trials (RCTs), which assessed monly practiced colorectal cancer FOBT screening reduces colorectal can- morbidity and/or mortality, followed by screening tests are reviewed. cer incidence by 17% to 20% as well.18 observational studies. Diagnostic accu- Fecal occult blood testing had 3 RCTs How the Test Is Performed. The racy studies were evaluated when RCTs and sigmoidoscopy had 1 randomized FOBT detects blood loss in the stool. and observational studies were not avail- trial and 3 case-control studies that as- It can detect blood loss caused by co- able or did not provide adequate evi- sessed mortality. No RCTs were found lorectal neoplasms, which tend to bleed dence. The highest quality evidence for colonoscopy, but 4 observational more than normal colonic mucosa. A available was included for each topic: for studies were identified in which asymp- variety of FOBTs are available,19,20 but FOBT, RCTs; for sigmoidoscopy, RCTs tomatic individuals underwent colonos- the Hemoccult II is most widely used and observational studies; for FOBT plus copy and in which adenoma and can- in the United States (Beckman- sigmoidoscopy, controlled trials, obser- cer incidences were assessed. Combined Coulter, Palo Alto, Calif). This test de- vational studies, and diagnostic accu- FOBT and sigmoidoscopy screening had tects the pseudoperoxidase activity racy studies; for double-contrast barium 5 trials that assessed clinical outcomes: found in hemoglobin when it inter- enema, diagnostic accuracy studies; and 1 trial assessed mortality, and the other acts with a guaiac-impregnated card in for colonoscopy, observational studies trials assessed incidence of polyps and/or the presence of a hydrogen peroxide de- and diagnostic accuracy studies. This re- cancer. Finally, for double-contrast veloper. A positive result is indicated view included additional literature to barium enema, no RCTs or observa- by the immediate appearance of a blue provide background material on the per- tional studies were identified that as- color on addition of the hydrogen per- formance, interpretation, and safety of sessed clinical effectiveness and 4 diag- oxide developer. The testing process re- colorectal cancer screening tests and to nostic accuracy studies were identified quires that the patient apply 2 distinct provide preliminary data on new, emerg- and evaluated. samples of 3 different stools to 6 test ing methods for colorectal cancer screen- card windows. Because the test de- ing, such as computed tomography (CT) Fecal Occult Blood Testing tects peroxidase or pseudoperoxidase colonography and stool-based molecu- Evidence of Clinical Efficacy. Results activity in stool, it is not specific for hu- lar testing. from 3 large RCTs of serial FOBTs con- man hemoglobin. Dietary substances ducted in Minnesota,12,13 United King- can result in false positive (eg, rare red EVIDENCE FOR COLORECTAL dom,14,15 and Denmark,16,17 involving meat, turnips, horseradish) or false CANCER SCREENING more that 250000 subjects followed for negative (eg, vitamin C) results. How- Death from colorectal cancer can be pre- up to 18 years, have consistently dem- ever, a recent systematic review found vented by the detection of early-stage dis- onstrated that serial FOBT reduces co- that a restricted diet does not reduce the ease