Chemoprevention of Gastrointestinal Cancers: An Umbrella Review of Meta-Analyses of Randomized Controlled Trials and Cohort Studies.

Several meta-analyses have investigated the association between chemopreventive agents (CPAs) and the risk of gastrointestinal cancers, but syntheses of the quality of evidence in aggregate are lacking. This umbrella review aimed to assess the quality of evidence from meta-analyses of randomized controlled trials (RCTs) and cohort studies that examine inverse associations between CPAs and the risk of gastrointestinal cancers or any premalignant conditions. Summary effect sizes from random-effects models, between-study heterogeneity, 95% prediction interval, small-study effect, excess significance, and credibility ceilings were devised to classify the credibility of evidence from meta-analyses of cohort studies, whereas the GRADE approach was used for meta-analyses of RCTs. From 20,296 publications, 577 full-text articles were evaluated for eligibility, and 69 articles that provided 194 unique meta-analyses were included. Among meta-analyses of RCTs (N = 93), 26 reached statistical significance (p < 0.05). Seven inverse associations were graded as either high quality (celecoxib and colorectal adenomas, (N = 4)) or moderate (aspirin and colorectal adenomas, (N = 2) and H-pylori eradication and gastric cancer (N = 1)). Among meta-analyses of cohort studies (N = 101), 60 reached statistical significance. Four inverse associations were graded as either convincing (antivirals with hepatocellular carcinoma (HCC); N = 1) or highly suggestive (aspirin with HCC (N = 2) and colorectal cancer (N = 1)). This review suggests that the associations with the most consistent empirical evidence were confined to those targeting the well-established risk factors of gastrointestinal cancer progression. Despite the limited established evidence, the inverse associations observed between metformin and colorectal, esophageal, and gastric cancers, as well as between statins and HCC and gastric cancer, merit further research.