LOW-DOSE ASPIRIN AND INCIDENCE OF COLORECTAL TUMORS IN A RANDOMIZED TRIAL

Background: Laboratory, clinical, and epidemiologic studies have recently suggested that regular use of aspirin can reduce colorectal cancer incidence or mortality. However, observational epidemiologic analyses have had limited opportunity to control for confounding bias or to specify aspirin doses used. Purpose: Our purpose was to examine the relationship between regular use of low-dose aspirin and incidence of invasive and noninvasive colorectal tumors by utilizing data from the Physicians' Health Study, a randomized, double-blinded, placebo-controlled trial of aspirin and beta carotene. We also attempted to determine whether invasive cancers among aspirin users were associated with rectal bleeding and early stage at diagnosis. Methods: The Physicians' Health Study includes 22071 U.S. male physicians. The aspirin arm was terminated in 1988 after a mean follow-up of 5 years. Stage at diagnosis and signs and/or symptoms during presentation were abstracted from medical records. Cox proportional hazards models were used to estimate relative risk (RR), 95% confidence intervals (CIs), and the association between aspirin and bleeding. Differences between aspirin and placebo groups in tumor risk over time were visualized with Kaplan-Meier curves. We assessed the association between aspirin and stage at diagnosis with a Mann-Whitney rank sum statistic for non-parametric comparison of two ordinal distributions. Results: The RR of developing colorectal cancer for aspirin compared with placebo was 1.15 (95% CI = 0.80-1.65). For in situ cancers and polyps, the RR was 0.86 (95% CI = 0.68-1.10). There was no significant trend for decreasing RR by year of follow-up for invasive cancers (P = .09) or noninvasive tumors (P = .96). Aspirin and placebo groups did not differ in stage or prevalence of rectal bleeding at diagnosis. Conclusions: Regular aspirin use, at a dose adequate for preventing myocardial infarction, was not associated with a substantial reduction in the incidence of colorectal cancer during 5 years of randomized treatment and follow-up. A small decrease in polyps in the aspirin group could not be reliably distinguished from a chance association. Our results suggest that among low-dose aspirin users, (a) colorectal cancer mortality is not likely to be reduced by earlier detection and (b) incidence is not likely to be increased due to aspirin-induced gastrointestinal bleeding. Implications: The potential for a benefit from higher doses of aspirin or longer duration of use should be addressed by more detailed observational epidemiologic studies and prevention trials with longer follow-up of randomized participants.