The Use of Aspirin for Primary Prevention of Colorectal Cancer: A Systematic Review Prepared for the U.S. Preventive Services Task Force

Cancer accounts for 23% of all deaths in the United States. It is the second leading cause of death after heart disease, and the leading cause of death in those younger than age 65 years. Colorectal cancer is the third most common type of cancer in both men and women and is the second and third leading cause of cancer-related deaths in men and women, respectively. In 2006, an estimated 148610 new cases of colorectal cancer occurred and 51170 patients died of this disease (1, 2). It is widely accepted that colorectal adenomatous polyps are the precursors of the vast majority of colorectal cancer cases, so the early detection and removal of these lesions are presumed to reduce the incidence and mortality of colorectal cancer. In addition, cases of cancer detected by screening may be in the early stage and therefore curable. Colorectal cancer has many characteristics of a disorder that would be amenable to screening, as recently reviewed by the U.S. Preventive Services Task Force (USPSTF) (3). Several screening methods are available, but despite the evidence of effectiveness, widespread routine screening of eligible individuals by any method continues to be low in the United States (47). An alternative and possibly complementary strategy to screening is prevention. This can include a variety of lifestyle and dietary changes or, as is the focus of this review, aspirin chemoprevention. Several basic science, population-based, and clinical trials have suggested a protective effect of aspirin as well as nonaspirin nonsteroidal anti-inflammatory drugs (NSAIDs), including cyclooxygenase-2 (COX-2) inhibitors, against colorectal adenomas and colorectal cancer. Since age is a major risk factor for colorectal cancer, with approximately 90% of cases occurring after age 50 (1), aspirin may be a particularly attractive intervention; it has documented efficacy in both the primary and the secondary prevention of cardiovascular disease (3). However, aspirin is not risk free; it can increase the risk for hemorrhagic stroke and gastrointestinal bleeding (3). Potential harms must be considered in light of the possibly long period of aspirin exposure used for colorectal cancer prevention. Furthermore, reductions in colorectal cancer mortality with chemoprevention would have to be great enough to compete with the 21% mortality reduction achieved with simple biannual fecal occult blood testing, or with the 60% mortality reduction seen with flexible sigmoidoscopy for lesions within reach of the sigmoidoscope. Furthermore, data suggest that sigmoidoscopy followed by colonoscopy when polyps are found could decrease colorectal cancer incidence by up to 80% (8). The USPSTF strongly recommends screening of men and women older than age 50 years (grade A recommendation) (9). A preventive strategy using aspirin may still have a role as an adjunct treatment, but the benefits would have to balance increased risks; in addition, the cost-effectiveness of this strategy would need to be favorable. Finally, although adherence to colorectal cancer screening is poor, long-term adherence to therapy with a chemopreventive agent in otherwise healthy individuals will probably have a similar limitation. We conducted this systematic review to examine the evidence on the effectiveness of aspirin for chemoprevention of colorectal adenomas, colorectal cancer, and colorectal cancer mortality, as well as the harms of aspirin use in this setting. Methods Data Sources The search strategy was developed in MEDLINE and modified for the other databases. The search was limited to English-language reports of human studies. Databases searched were MEDLINE, 1966 to December (week 4) 2006; preMEDLINE, through 5 April 2005; EMBASE, 1980 to week 14 of 2005 (publication years 2003 to 2005); and CENTRAL and the Cochrane Library, Issue 4, 2004. Beyond these dates, we surveyed several sources to ascertain additional potentially eligible studies. The PubMed Cancer subset was searched for non-MEDLINE material. Terms were derived from the National Cancer Institute cancer topic searches for colorectal cancer and adenomatous polyps. A comprehensive retrieval strategy was derived from the indexing in both MEDLINE and EMBASE, investigator-nominated terms, and previous reviews (1012). A search strategy to find recent systematic reviews of aspirin that appeared to address harm was developed and run in MEDLINE (2003 to December [week 4] 2006). The Cochrane Database of Systematic Reviews and Database of Abstracts of Reviews of Effects (DARE) (Cochrane Library, third quarter 2004) were searched for all systematic reviews related to aspirin, without date restrictions. Study Selection At each screening level, 2 members of the review team independently selected articles for inclusion, after an initial calibration exercise. After identifying potentially relevant articles in the initial screening level, team members assessed whether each article met the inclusion criteria. Conflicts were resolved by consensus. A third level of screening was included to discriminate the different study designs. Data were abstracted by one reviewer and checked by a second reviewer. Randomized, controlled trials (RCTs); controlled clinical trials; and observational studies (cohort and casecontrol studies) of the efficacy or effectiveness of aspirin were considered for inclusion if they fulfilled the population and outcome criteria: Participants were at average risk for colorectal cancer (that is, they had no known risk factors for colorectal adenoma or colorectal cancer other than age); could have a personal or family history of colorectal adenoma; and could have a family history of sporadic colorectal cancer (higher risk). Studies of familial adenomatous polyposis or hereditary nonpolyposis colon cancer syndromes (Lynch I or II) were excluded because these syndromes account for a small percentage of colorectal cancer cases. Secondary prevention studies of patients with a history of colorectal cancer were also excluded. Included studies addressed the incidence of colorectal adenoma or colorectal cancer and reductions in colorectal cancer mortality or overall mortality. We sought studies on gastrointestinal, cardiovascular, and renal harms associated with the aspirin use by identifying systematic reviews; we chose to identify reviews because of the large number of reviews on harms of aspirin already performed. Data Extraction and Quality Assessment Several members of the team extracted data independently by using a computerized Web-based system (SRS 4.0; Trialstat Corp., Ottawa, Ontario, Canada). The PICOS (participant, intervention/exposure, comparator, outcome and study design) approach was applied for data extraction. Predefined criteria from the USPSTF were used to assess the quality of included systematic reviews, trials, and observational studies, which were rated as good, fair, or poor (13). This scale relies on 4 criteria for systematic reviews, 6 criteria for casecontrol studies, 7 criteria for cohort studies, and 7 criteria for RCTs. Studies with a good rating met all criteria, fair studies met at least 80% of criteria and had no fatal flaw, and poor studies met fewer than 80% of criteria or had a fatal flaw. Data Synthesis and Analysis An analytic framework was used to facilitate study grouping and subsequent data analysis in an effort to produce logical groupings and to minimize clinical heterogeneity. Studies were initially grouped by the disorder (that is, colorectal adenoma vs. colorectal cancer), study design, study sample, and medication exposure and were subsequently subcategorized according to measures of dose effect, duration of exposure, and secondary outcomes when reported. Definition of such categories as regular use can be found elsewhere (13). Harms data from the included systematic reviews were summarized and presented as a qualitative systematic review. Results were combined numerically only if clinically and statistically appropriate. The effect measure chosen for synthesis was the relative risk (RR). In casecontrol studies, a direct estimate of the RR is not possible. The odds ratio (OR) may be estimated, however, and when event rates are low, as is the case here, the OR closely approximates the RR. In what follows, we simply refer to the RR. Heterogeneity was assessed by using the I2 statistic. Studies were combined when the I2 value was 50% or less (14). Point estimates of the adjusted RRs and their 95% CIs were directly abstracted from the reports of primary studies. One source of heterogeneity may be study-to-study variation in the method of selecting confounders to adjust for and the final set of confounders chosen. Appendix Tables 1 and 2 summarize these characteristics for each study. Furthermore, the USPSTF report discusses the methodologic considerations in detail (13). Standard errors were computed by dividing the CI width by (21.96). For 1 study that did not report CIs (15), the standard error was calculated by using a CI imputed from 2 different estimates in the report. Quantitative synthesis was conducted by using inverse-variance weighting and a random-effects model (16). Appendix Table 1. Confounders Controlled for in the Studies' Adjusted Relative Risk Estimates for the Incidence of Colorectal Cancer (Cohort and CaseControl Studies) Appendix Table 2. Confounders Controlled for in the Studies' Adjusted Relative Risk Estimates for the Incidence of Colorectal Adenoma (Cohort and CaseControl Studies) Role of the Funding Sources The evidence synthesis upon which this article was based was funded by the Centers for Disease Control and Prevention (CDC) for the Agency for Healthcare Research and Quality (AHRQ) and the USPSTF. Its design, conduct, and reporting were based on specific directives from these agencies. Data Synthesis Study Selection The literature search for the comprehensive USPSTF report (13) yielded 1790 potentially relevant bibliographic records addressing the use