Metformin Use Is Associated With Longer Progression-Free Survival of Patients With Diabetes and Pancreatic Neuroendocrine Tumors Receiving Everolimus and/or Somatostatin Analogues

BACKGROUND & AIMS: Metformin seems to have anticancer effects. However, it is not clear whether use of glycemia and metformin affect outcomes of patients with advanced pancreatic neuroendocrine tumors (pNETs). We investigated the association between glycemia and progression-free survival (PFS) of patients with pNETs treated with everolimus and/or somatostatin analogues, as well as the association between metformin use and PFS time. METHODS: We performed a retrospective analysis of 445 patients with advanced pNET treated at 24 medical centers in Italy from 1999 through 2015. Data on levels of glycemia were collected at time of diagnosis of pNET, before treatment initiation, and during treatment with everolimus (with or without somatostatin analogues), octreotide, or lanreotide. Diabetes was defined as prior or current use of glycemia control medication and/or fasting plasma glucose level 126 mg/dL, hemoglobin A1c 6.5% (48 mmol/L), or a random sample of plasma glucose 200 mg/dL (11.1 mmol/L), with reported classic symptoms of hyperglycemia or hyperglycemic crisis. Patients were assigned to groups based on diagnosis of diabetes before or during antitumor therapy. PFS was compared between patients with vs without diabetes. Among patients with diabetes, the association between metformin use and PFS was assessed. We performed sensitivity and landmark analyses to exclude patients who developed diabetes while receiving cancer treatment and to exclude a potential immortal time bias related to metformin intake. RESULTS: PFS was significantly longer in patients with diabetes (median, 32.0 months) than without diabetes (median, 15.1 months) (hazard ratio for patients with vs without diabetes, 0.63; 95% confidence interval, 0.50–0.80; P ¼ .0002). PFS of patients treated with metformin was significantly longer (median PFS, 44.2 months) than for patients without diabetes (hazard ratio for survival of patients with diabetes receiving metformin vs without diabetes, 0.45; 95% confidence interval, 0.32–0.62; P < .00001) and longer than for patients with diabetes receiving other treatments (median PFS, 20.8 months; hazard ratio, 0.49; 95% confidence interval, 0.34–0.69; P < .0001). In multivariable analysis, adjusted for other factors associated with outcomes, metformin was associated with longer PFS but level of glycemia was not. Metformin was associated with increased PFS of patients receiving somatostatin analogues and in those receiving everolimus, with or without somatostatin analogues. Sensitivity and landmark analyses produced similar results. CONCLUSIONS: In a retrospective study of patients with pNETs, we found a significant association between metformin use and longer PFS.