Metformin use is associated with better survival of diabetic patients with pancreatic cancer

Abstract

Purpose: Accumulating evidence suggests that metformin has antitumor activity. The aim of this study was to determine whether metformin use has a survival benefit in patients with pancreatic cancer.

Experimental design: We conducted a retrospective study of patients with diabetes and pancreatic cancer treated at The University of Texas MD Anderson Cancer Center (Houston, TX). Information on diabetes history, including treatment modalities and clinical outcome of pancreatic cancer, was collected using personal interviews and medical record review. Survival analysis was carried out using a Kaplan-Meier plot, log-rank test, and Cox proportional hazards regression models.

Results: Among the 302 patients identified, there were no significant differences in demographic or major clinical characteristics between the patients who had received metformin (n = 117) and those who had not (n = 185). The 2-year survival rate was 30.1% for the metformin group and 15.4% for the non-metformin group (P = 0.004; χ(2) test). The median overall survival time was 15.2 months for the metformin group, and 11.1 months for the non-metformin group (P = 0.004, log-rank test). Metformin users had a 32% lower risk of death; the HR (95% confidence interval) was 0.68 (0.52-0.89) in a univariate model (P = 0.004), 0.64 (0.48-0.86) after adjusting for other clinical predictors (P = 0.003), and 0.62 (0.44-0.87) after excluding insulin users (P = 0.006). Metformin use was significantly associated with longer survival in patients with nonmetastatic disease only.

Conclusions: Our finding that metformin use was associated with improved outcome of patients with diabetes and pancreatic cancer should be confirmed in independent studies. Future research should prospectively evaluate metformin as a supplemental therapy in this population.

©2012 AACR.

Figures

Fig. 1

Survival curves for the metformin group (solid line) and nonmetformin group (dotted line) in patients with resectable (panel A), unresectable (panel B), and metastatic (panel C) pancreatic cancer. P values (log rank test) were 0.293, 0.001, and 0.482 for panels A, B, and C, respectively.