A SEER-Medicare analysis of the impact of metformin on overall survival in ovarian cancer

Objective. Determine whether metformin use is associated with improved survival in patients with ovarian, fallopian tube or primary peritoneal cancer. Methods. All patients with a diagnosis of first epithelial ovarian cancer from 2007 to 2011 in the combined SEER-Medicare database were identified from the SEER registry primary site codes. Comorbidities, procedures and cancer treatment ICD-9 and HCPCS codes were used to search the Medicare claims files. Medication use was determined with National Drug Codes using the Medicare Part D event files. The primary outcome, overall survival, was assessed between metformin users and non-users using a Cox Proportional Hazards survival model. To control for confounding, metformin users were matched to non-metformin users using propensity scores. Effect of dosage on survival was assessed using discrete time survival analysis with pooled logistic regression (PLR). Results. There were 2291 cases that met our inclusion criteria. Of these, 180 (7.9%) had been on metformin. The median age was 73 years, with the majority of the population being White (83.5%) and treated with primary surgery (74.1%). Metformin use was not associated with overall survival in the entire cohort (HR 0.96, 95% CI 0.75-1.23) or in the matched sample cohort (HR 0.88, 95% CI 0.66-1.17). However, exploratory regression with time-varying coefficients suggests a protective metformin effect for women alive after 30 months follow-up (HR = 0.37, 95% 0.16-0.87). Conclusion. No statistically significant association was observed between metformin use and overall survival in a matched cohort of 360 ovarian cancer patients. However, exploratory modeling suggests metformin use may be protective in a certain subgroup of patients. (C) 2017 Elsevier Inc. All rights reserved.