This comprehensive meta-analysis of randomized controlled trials (RCTs) assessed the efficacy and safety of aspirin and other NSAIDs for the primary prevention of colorectal cancer (CRC) and colorectal adenoma (CRA). The review pooled data from trials conducted in Europe, North America, Australia, and Japan, encompassing a total sample size of 124,837 participants. The population consisted of the general population, and the intervention involved aspirin or NSAIDs compared against no treatment or placebo. Follow-up durations varied across studies, categorized into intervals of 5 to less than 10 years, 10 to less than 15 years, and 15 years or more. The certainty of the evidence was rated from very low to high, with downgrades applied for risk of bias and imprecision in several outcomes. Notably, no RCTs were identified evaluating non-aspirin NSAIDs specifically for primary CRC prevention, limiting the scope of the analysis to aspirin-based regimens.
Regarding the primary outcome of CRC incidence, the analysis found little to no difference between the intervention and comparator groups during the initial 5 to less than 10 years of follow-up. The hazard ratio was 1.00 (95% CI 0.81 to 1.24) based on 26,702 participants. Similarly, in the 10 to less than 15 year interval, the hazard ratio remained neutral at 0.95 (95% CI 0.77 to 1.17) among 42,412 participants. Only in the longest follow-up period of 15 years or more did the data suggest a potential slight reduction in incidence, with a hazard ratio of 0.78 (95% CI 0.67 to 0.91) derived from 47,464 participants. However, the authors caution that these long-term benefits are derived from observational follow-up phases where intention-to-treat analyses are not robust to post-randomization confounding.
The analysis of CRC mortality revealed a concerning pattern in the short term. During the first 5 to less than 10 years, the data suggested a potential increase in mortality, with a hazard ratio of 1.77 (95% CI 1.02 to 3.07) among 19,114 participants. In the subsequent 10 to less than 15 year period, the effect was neutral with a Peto odds ratio of 1.14 (95% CI 0.73 to 1.78). By the 15-year mark, the Peto odds ratio indicated a potential reduction in mortality at 0.74 (95% CI 0.60 to 0.90) across 53,909 participants. These findings underscore the uncertainty regarding long-term survival benefits versus immediate risks.
Secondary outcomes regarding colorectal adenoma incidence showed little to no difference in the first 5 to less than 10 years, with a Peto odds ratio of 0.42 (95% CI 0.10 to 1.87) in 12,546 participants. The confidence interval for this result is wide, indicating imprecision. Safety analyses demonstrated a clear increase in serious adverse events overall, with a relative risk of 1.06 (95% CI 0.84 to 1.34) in 16,442 participants. More specifically, the risk of serious extracranial hemorrhage was increased with a relative risk of 1.59 (95% CI 1.30 to 1.95) among 97,567 participants. Additionally, there was probably an increased risk of hemorrhagic stroke, with a Peto odds ratio of 1.40 (95% CI 1.11 to 1.77) in 105,037 participants. The certainty of evidence for these safety outcomes was rated as high or moderate.
Methodological limitations significantly impact the interpretation of these results. The certainty of evidence was downgraded for risk of bias and imprecision. A critical limitation is that the evidence of very low to moderate certainty suggests little to no benefit in the first 15 years. Furthermore, low-certainty evidence suggests a potential increase in CRC mortality within the first 5 to 10 years. The reliance on observational data for long-term benefit estimates introduces potential confounding that cannot be fully addressed by standard intention-to-treat analyses. Discontinuations and tolerability were not reported in the source data. Funding was provided by the China Postdoctoral Science Foundation.
Clinical implications of this review emphasize that practice should continue to center on an individualized assessment and a shared decision-making process. Clinicians must carefully balance a patient's established cardiovascular risk profile against their risk of bleeding. The evidence does not support definitive conclusions or specific implications for the routine use of aspirin for CRC primary prevention in the general population. The potential long-term benefits after 15 years remain uncertain due to methodological constraints. Reduction in CRC incidence and mortality at earlier time points is not supported by the high-certainty evidence. Questions remain unanswered regarding the optimal duration of therapy to achieve benefit without incurring unacceptable bleeding risks, particularly given the lack of data on non-aspirin NSAIDs.
View Original Abstract ↓
RATIONALE: The role of nonsteroidal anti-inflammatory drugs, particularly aspirin, in the primary prevention of colorectal cancer remains controversial. The debate over aspirin use is driven by the challenge of balancing uncertain preventive benefits against the risks of adverse effects. Given the inconsistent findings from clinical trials and conflicting clinical guidelines, a rigorous and updated systematic review is necessary to clarify the evidence base.
OBJECTIVES: To assess the benefits and harms of nonsteroidal anti-inflammatory drugs (NSAIDs), including aspirin, for preventing colorectal cancer (CRC) and colorectal adenoma (CRA) in the general population.
SEARCH METHODS: We searched CENTRAL, MEDLINE, Embase, and two clinical trial registers (ClinicalTrials.gov and WHO ICTRP) on 3 March 2025.
ELIGIBILITY CRITERIA: We included parallel-group or factorial-design randomized controlled trials (RCTs) comparing aspirin and other NSAIDs with either no treatment or a different treatment for preventing CRC or CRA in the general population.
OUTCOMES: Our critical outcomes were CRC incidence and serious adverse events (SAE). Our important outcomes included CRC mortality, CRA incidence, serious extracranial hemorrhage, and hemorrhagic stroke. When data were available, we categorized findings into prespecified follow-up intervals: 5 to < 10 years, 10 to < 15 years, and ≥ 15 years.
RISK OF BIAS: We assessed the risk of bias for study outcomes as high risk, some concerns, or low risk, using the Cochrane RoB 2 tool.
SYNTHESIS METHODS: We synthesized data for each outcome using random-effects meta-analysis. For time-dependent outcomes, we prioritized time-to-event data, calculating hazard ratios (HRs) with 95% confidence intervals (CIs). When these were unavailable, we used dichotomous data to calculate risk ratios (RRs). For rare outcomes, we calculated Peto odds ratios (ORs) using a fixed-effect model. We used GRADE to assess the certainty of the evidence.
INCLUDED STUDIES: We included 10 RCTs involving a total of 124,837 participants. These studies compared aspirin with either placebo or no treatment for the primary prevention of CRC. Low-dose aspirin (75 to 100 mg per day) was typically used, though three studies evaluated higher doses. The studies were mostly conducted in Europe and North America. One study had sites in Australia, and there were two large studies conducted in Japan. Seven studies reported long-term results with extended observational follow-up where blinding had ceased. We did not identify any RCTs that evaluated the use of non-aspirin NSAIDs by the general population for primary CRC prevention.
SYNTHESIS OF RESULTS: For the comparison of aspirin versus inactive control in the general population, we rated the certainty of the evidence as very low to high. We downgraded the certainty level for several outcomes due primarily to risk of bias and imprecision. Regarding CRC incidence, aspirin probably results in little to no difference at follow-up ≥ 5 to < 10 years (HR 1.00, 95% CI 0.81 to 1.24; 3 studies, 26,702 participants; moderate-certainty evidence) and at ≥ 10 to < 15 years (HR 0.95, 95% CI 0.77 to 1.17; 2 studies, 42,412 participants; moderate-certainty evidence). Aspirin may reduce CRC incidence slightly at follow-up ≥ 15 years (HR 0.78, 95% CI 0.67 to 0.91; 3 studies, 47,464 participants; very low-certainty evidence), but the evidence is very uncertain. Regarding CRC mortality, aspirin may increase mortality at follow-up ≥ 5 to < 10 years (HR 1.77, 95% CI 1.02 to 3.07; 1 study, 19,114 participants; low-certainty evidence), may result in little to no difference in mortality at follow-up ≥ 10 years and < 15 years (Peto OR 1.14, 95% CI 0.73 to 1.78; 1 study, 39,876 participants; low-certainty evidence), but may reduce mortality at follow-up ≥ 15 years (Peto OR 0.74, 95% CI 0.60 to 0.90; 5 studies, 53,909 participants; very low certainty evidence), but the evidence is very uncertain. Regarding CRA incidence, aspirin may result in little to no difference in CRA incidence at ≥ 5 to < 10 years of follow-up (Peto OR 0.42, 95% CI 0.10 to 1.87; 1 study, 12,546 participants; very low certainty evidence), but the evidence is very uncertain. Regarding safety, although aspirin probably results in little to no difference in overall SAE (RR 1.06, 95% CI 0.84 to 1.34; 3 studies, 16,442 participants; moderate-certainty evidence), aspirin does increase the risk of serious extracranial hemorrhage (RR 1.59, 95% CI 1.30 to 1.95; 8 studies, 97,567 participants; high-certainty evidence) and probably increases the risk of hemorrhagic stroke (Peto OR 1.40, 95% CI 1.11 to 1.77; 8 studies, 105,037 participants; moderate-certainty evidence).
AUTHORS' CONCLUSIONS: It is not possible to draw definitive conclusions or outline specific implications for the routine use of aspirin for CRC primary prevention based on the current evidence. Our findings reveal complex, time-dependent preventive effects and concerns about potential harms for clinicians and patients to consider. Evidence of very low to moderate certainty shows little to no benefit for CRC or CRA incidence in the first 15 years, and low-certainty evidence suggests a potential increase in CRC mortality in the first 5 to 10 years. Very low-certainty evidence suggests potential benefits for CRC incidence and mortality after long-term follow-up (≥ 15 years), but these potential long-term benefits are derived from findings in the observational follow-up phases of RCTs, where standard intention-to-treat analyses are not robust to post-randomization confounding from factors such as treatment contamination. The uncertain and delayed potential for benefit must be weighed against a definite harm. While aspirin probably has little to no effect on overall serious adverse events (moderate-certainty evidence), it increases the risk of serious extracranial hemorrhage (high-certainty evidence) and probably increases the risk of hemorrhagic stroke (moderate-certainty evidence). In light of the mixed evidence, clinical practice should continue to center on an individualized assessment and a shared decision-making process, carefully balancing a patient's established cardiovascular risk profile against their risk of bleeding.
FUNDING: This Cochrane review was funded (in part) by the China Postdoctoral Science Foundation (2024M752248) and the Postdoctoral Fellowship Program (Grade A) of China Postdoctoral Science Foundation (BX20230244).
REGISTRATION: Protocol available via doi.org/10.1002/14651858.CD015266.
