Chemoprevention of colorectal cancer in individuals with previous colorectal neoplasia: systematic review and network meta-analysis

OBJECTIVE To assess the comparative efficacy and safety of candidate agents (low and high dose aspirin, nonaspirin non-steroidal anti-inflammatory drugs (NSAIDs), calcium, vitamin D, folic acid, alone or in combination) for prevention of advanced metachronous neoplasia (that is, occurring at different times after resection of initial neoplasia) in individuals with previous colorectal neoplasia, through a systematic review and network meta-analysis. DATA SOURCES Medline, Embase, Web of Science, from inception to 15 October 2015; clinical trial registries. STUDY SELECTION Randomized controlled trials in adults with previous colorectal neoplasia, treated with candidate chemoprevention agents, and compared with placebo or another candidate agent. Primary efficacy outcome was risk of advanced metachronous neoplasia; safety outcome was serious adverse events. DATA EXTRACTION Two investigators identified studies and abstracted data. A Bayesian network meta-analysis was performed and relative ranking of agents was assessed with surface under the cumulative ranking (SUCRA) probabilities (ranging from 1, indicating that the treatment has a high likelihood to be best, to 0, indicating the treatment has a high likelihood to be worst). Quality of evidence was appraised with GRADE criteria. RESULTS 15 randomized controlled trials (12 234 patients) comparing 10 different strategies were included. Compared with placebo, non-aspirin NSAIDs were ranked best for preventing advanced metachronous neoplasia (odds ratio 0.37, 95% credible interval 0.24 to 0.53; SUCRA=0.98; high quality evidence), followed by low-dose aspirin (0.71, 0.41 to 1.23; SUCRA=0.67; low quality evidence). Low dose aspirin, however, was ranked the safest among chemoprevention agents (0.78, 0.43 to 1.38; SUCRA=0.84), whereas non-aspirin NSAIDs (1.23, 0.95 to 1.64; SUCRA=0.26) were ranked low for safety. High dose aspirin was comparable with low dose aspirin in efficacy (1.12, 0.59 to 2.10; SUCRA=0.58) but had an inferior safety profile (SUCRA=0.51). Efficacy of agents for reducing metachronous colorectal cancer could not be estimated. CONCLUSIONS Among individuals with previous colorectal neoplasia, non-aspirin NSAIDs are the most effective agents for the prevention of advanced metachronous neoplasia, whereas low dose aspirin has the most favorable risk: benefit profile.