Alternate-day, low-dose aspirin and cancer risk.

Most apparent medical discoveries are not subsequently confirmed, and reports of unexpected benefits of established treatments are particularly prone to being disproved. It is noteworthy, therefore, that the aspirin and cancer prevention story continues to stand up to testing, with the latest report from the Women’s Health Study (WHS) in this issue (1) confirming previous observations and offering important new insights. During the 25 years since studies first suggested that regular aspirin use might prevent colorectal cancer, more than 150 case–control studies and about 50 cohort studies have reported consistent associations between regular aspirin use and substantially reduced risks for colorectal, esophageal, and stomach cancer (2). However, data from observational studies can be misleading, and evidence from randomized trials is required. About 10 years ago, randomized trials of the use of aspirin (81 to 325 mg daily) for the secondary prevention of colorectal adenomas reported a 15% to 30% reduction in recurrent adenomas (3). However, follow-up was too short to determine any effect on risk for colorectal cancer, and 10-year follow-up of the 2 largest randomized trials of aspirin in primary prevention of vascular events (the WHS and the Physicians’ Health Study) showed no such effect (4, 5). Nevertheless, given an estimated average delay of 10 years between early development of an adenoma—the point at which aspirin is believed to act—and clinical presentation with colorectal cancer, it was argued that follow-up of 15 to 20 years would be needed to determine any effect on incidence of colorectal cancer (6). By collating centralized cancer registration and death certificate data, 20-year follow-up was first obtained for patients who had been randomly assigned to high-dose (300 to 1500 mg) aspirin or control in 2 U.K. trials on the prevention of vascular events (6). Analysis confirmed the lack of effect of aspirin on 10-year incidence of colorectal cancer but showed a substantial reduction in risk thereafter (hazard ratio, 0.60 [95% CI, 0.42 to 0.87]; P 0.007) (6). Similar posttrial follow-up of 2 trials of daily low-dose (75 to 100 mg) aspirin subsequently confirmed this finding (7), and a pooled analysis of individual patient data from the 4 trials (14 033 patients with 391 cases of colorectal cancer) showed that treatment with aspirin for about 5 years reduced the 20-year risk for cancer of the proximal colon but not the distal colon or rectum (Table) (7). Similar delayed reductions were also observed in the long-term risks for esophageal, stomach, and lung cancer (8), and aspirin was also found to reduce distant metastasis (9). The report of 18-year follow-up of the WHS, a randomized trial of about 10 years’ treatment with 100 mg of alternate-day aspirin versus placebo in middle-aged women, summarizes the first test of these observations in a trial in which cancer was a prespecified outcome and provides the first reliable data in women. In line with previous trials involving mainly men (6, 7), a reduction in colorectal cancer risk appeared in the aspirin group after 10 years of follow-up, again because of a substantial delayed reduction in risk for cancer of the proximal colon (Table). Of note, the WHS also prospectively determined rates of colonoscopy and adenomas and showed that both were balanced across the treatment groups during and after the trial, thus helping to rule out any potential bias due to earlier diagnosis and removal of adenomas resulting from increased bleeding in the aspirin group. There was also no evidence of earlier diagnosis of colorectal cancer in the aspirin groups in previous trials (6, 7). The reduction in risk for proximal colon cancer without a reduction in overall number of adenomas might reflect the particular morphology of adenomas in the proximal colon and their relatively rapid growth, which at least partly explains the relative ineffectiveness of colonoscopic screening in preventing proximal colon cancer. The proximal and distal colon also differ in embryologic origins, risk factors for cancer, mechanisms of carcinogenesis, and the molecular and genetic characteristics of cancer. Although the numbers of esophageal and stomach cancer cases in the WHS were small, the trends in effect were consistent with the significant reductions in long-term risk observed on follow-up of trials of daily aspirin (7, 8), as was the effect on risk for gastrointestinal cancer as a whole (Table). The WHS also showed an early reduction in risk for lung cancer in the aspirin group (5), which reversed during posttrial follow-up. This could be a chance finding, but there was a significant reduction in lung cancer deaths during follow-up of trials of daily aspirin (8), particularly for adenocarcinomas. Further analyses of the WHS might determine whether the reversal of the apparent in-trial effect was related to posttrial aspirin withdrawal. However, the WHS showed no reduction in incidence of other types of cancer, particularly breast and gynecologic cancer, which were underrepresented in the predominantly male populations of the trials of daily aspirin (6–10) but accounted for more than half of the cancer cases in the WHS. Although the WHS did not show the reduction in overall cancer deaths that was seen after 5 years of follow-up in the trials of daily aspirin (8, 10), there was evidence of the reduction in distant cancer metastasis. The posttrial reductions in risk for metastatic adenocarcinoma (hazard ratio, 0.73 [CI, 0.56 to 0.96]; P 0.025) and Duke stage C or D colorectal cancer (hazard ratio, 0.44 [CI, 0.25 to 0.77]; P 0.004) are consistent with observations in trials of daily aspirin (9) but were more delayed. The early effects on cancer death that were seen with daily aspirin, particularly in trials with higher doses (9, 10), suggested reduced growth or metastasis from occult cancer that was present at randomization, whereas the delayed Annals of Internal Medicine Editorial