Association of type 2 diabetes with cancer: A meta-analysis with bias analysis for unmeasured confounding in 151 cohorts comprising 32 million people

Background: Whether the association between type 2 diabetes (T2D) and cancer is causal remains controversial. Purpose: To assess how robust are the observational associations between (T2D) and cancer to unmeasured confounding. Data sources: PubMed, Web of Science, and the Cochrane library were systematically searched on January 10th, 2019. Study selection: Observational studies investigating associations between T2D and cancer incidence or mortality. Data extraction: Cohort-level relative risk (RR). Data Synthesis: RRs were combined in random-effects meta-analyses and pooled estimates used in bias analyses. A total of 151 cohorts (over 32 million people, 1.1 million cancer cases and 150,000 cancer deaths) were included. In meta-analyses, T2D was associated with incidence of several cancers, from prostate (RR 0.83; 95% CI: 0.79, 0.88) to liver (2.23; 1.99, 2.49); and mortality of pancreatic cancer (1.67; 1.30, 2.14). In bias analyses, assuming an unmeasured confounding associated with both T2D and cancer with a RR of 1.5, the proportion of studies with a true effect size larger than a RR of 1.1 (i.e., 10% increased risk in individuals with T2D) was nearly 100% for liver, pancreatic, and endometrial; 86% for gallbladder; 67% for kidney; 64% for colon; 62% for colorectal; and less than 50% for other cancer incidence, and 92% for pancreatic cancer mortality. Limitations: Biases other than unmeasured confounding were not analytically assessed. Conclusions: Our findings strongly suggest a causal association between T2D and liver, pancreatic, and endometrial cancer incidence, and pancreatic cancer mortality; conversely, associations with other cancers were less robust to unmeasured confounding.