Objective The results of cohort studies are controversial regarding the association between preexisting diabetes mellitus (DM) and the risk of endometrial cancer (EC) incidence and mortality. The aim of this meta-analysis was to assess whether an association exists between them in prospective studies.

Methods The PubMed and EMBASE database were searched for prospective cohort studies of preexisting DM on EC occurrence, mortality outcomes, updated up to June 2012. Data were independently extracted by 2 investigators using a standardized protocol. Summary relative risks (SRRs) with 95% confidence intervals (CIs) were estimated by the use of fixed-effect or random-effect models. Between-study heterogeneity was investigated using χ²-based Cochran Q statistic test and I² statistics.

Results A total of 21 cohort studies (20 articles) involving 12,195 incident cases and 575 deaths caused by EC were included in the analysis. Among these, 15 studies reported the risk of EC incidence, and 6 studies reported risk of EC mortality. The meta-analysis showed that DM was associated with an increased incidence of EC (SRR, 1.81; 95% CI, 1.38–2.37; test for heterogeneity: Q = 280.96, P < 0.001, I² = 95.4%), compared with individuals without diabetes or the general population. Subgroup analyses were conducted by geographical region, study design, year of publication, diabetes ascertainment, type of DM, and adjustments for some variables, and the positive association between DM and risk of EC was consistently observed in each subgroup. Sensitivity analyses also showed that no single study significantly influenced the pooled relative risk. No significant publication bias (P = 0.173) was found. Diabetes mellitus was not positively associated with EC mortality (SRR, 1.23; 95% CI, 0.80–1.90; test for heterogeneity: Q = 11.95, P = 0.035, I² = 58.2%).

Conclusions This meta-analysis of cohort studies suggests that there is a significant association between DM and increased risk of EC incidence but no increased risk of EC mortality.