The effect of diabetes and metformin on clinical outcomes is negligible in risk-adjusted endometrial cancer cohorts

Objective To examine the influence of diabetes and metformin therapy on overall survival (OS) and progression-free survival (PFS) in patients with endometrial cancer (EC) by using propensity score (PS) matching to account for confounding factors. Methods We retrospectively identified consecutive patients with stage I-IV EC managed surgically from 1999 through 2008 and stratified patients by diabetes status. PS matching was used to adjust for confounding covariates. OS and PFS were compared between diabetic and nondiabetic matched pairs and between matched pairs of diabetic patients with or without metformin therapy. Cox proportional hazards models were fit to estimate the effects on outcomes. Results Among 1303 eligible patients (79% stage I, 28% grade 3), 277 (21.3%) had a history of diabetes. Among diabetic patients, treatment consisted of metformin in 116 (41.9%); 57 (20.6%) had other oral agents, 51 (18.4%) insulin with or without other oral agents, and 53 (19.1%) diet modification only. For PS-matched diabetic and nondiabetic patients with EC, OS (hazard ratio [HR], 1.01; 95% CI, 0.72-1.42) and PFS (HR, 1.01; 95% CI, 0.60-1.69) were similar between matched subsets. No differences in OS and PFS were observed when comparing PS-matched metformin users with nondiabetic patients (OS HR, 1.03; 95% CI, 0.57-1.85; PFS HR, 1.14; 95% CI, 0.49-2.62) or with other diabetic patients (OS HR, 0.61; 95% CI, 0.30-1.23; PFS HR, 1.06; 95% CI, 0.34-3.30). Conclusions When adjusted for confounding covariates, OS and PFS are similar between diabetic and nondiabetic patients with EC and between metformin users and nonusers or nondiabetic patients.