RESUMO
AIM: To evaluate the effect of metformin on cancer incidence in subjects with overweight/obesity and/or prediabetes/diabetes. MATERIALS AND METHODS: We searched MEDLINE, Embase and CENTRAL for randomized controlled trials (RCTs) in adults with overweight/obesity and/or prediabetes/diabetes that compared metformin to other interventions for ≥24 weeks. Independent reviewers selected and extracted data including population and intervention characteristics and new diagnoses of cancer. We used the RoB 2.0 risk-of-bias tool and the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) framework to assess risk of bias and certainty of evidence. RESULTS: From 14 895 records after removal of duplicates, 27 trials were included, providing a total of 10 717 subjects in the metformin group and 10 003 in the control group, with 170 and 208 new cases of cancer, respectively. Using a random-effects model, the relative risk was 1.07 (95% confidence interval 0.87-1.31), with similar results in subgroup analyses by study duration or effect of control intervention on weight. Risk of bias in most studies was low, and no evidence of publication bias was found. Trial sequential analysis provided evidence that the cumulative sample size was large enough to exclude a significant effect of metformin on cancer incidence. CONCLUSIONS: Metformin did not reduce cancer incidence in RCTs involving subjects with overweight/obesity and/or prediabetes/diabetes.