Abstract P3-06-01: Association between metformin use and improved survival in breast cancer patients: A population-based study

Abstract

Background: Long-term use of AMP-kinase agonist Metformin, the first-line oral hypoglycemic agent, has recently been shown to be associated with both a decreased incidence of and improved survival following diagnosis of breast and other solid tumors. However, the sample size of most reported studies has been limited and has not been able to clearly delineate if there are molecular subtype-dependent effects. Method: We implemented a population-based, retrospective case-control analysis of the effect of metformin on breast cancer outcomes. We comprehensively searched the electronic medical records of all female patients who were cared for within the University of Pennsylvania Health System between the years 1997 and 2013. To ensure the accuracy of patient selection, we focused our preliminary analysis on only patients (n = 20,684) who received an ICD9 code for breast cancer (BC) on at least 2 separate in-person visits. We assessed the use of metformin by identifying those patients who have at least a record of a metformin prescription both before the first breast cancer diagnosis ICD9 code and after the first breast cancer diagnosis visit. Overall survival was determined by assessing a combination of electronic medical record of patient expiration and the SSN Death Certificate Index. Statistical tests, including chi-sq and cox-hazard modeling were implemented in R, Graphpad Prism or EXCEL. Results: In the preliminary analysis, we identified a total of 29,251 female patients either diagnosed or reported a history of BC within the UPHS EMRs. Stratifying by per patient visits, we limited the subsequent analysis to the 20,684 BC patients with at least 2 in-person visits where BC was recorded as a diagnosis code. Of these, there were 534 patients who had a record of metformin prescription prior to and after the first BC diagnosis code and 20,150 BC patients who did not. We found that with a mean follow up of (6.40 years), there were a total of 8 deaths (1.28%) in metformin users and 526 deaths (2.62%) among the no metformin users. Using a two-tailed chi-sq test, we found that metformin use was significantly associated with improved overall survival rates (98.50 vs 96.94%, p<0.05). We are currently assessing the role of age, ethnicity, the presence of diabetes, health metrics related to insulin resistance, and the use of other metabolic regulators (insulin, sulfonylurea) using a cox-hazard proportional risk model. Conclusion: Our preliminary analyses suggest that metformin use before and following diagnosis may be protective against overall mortality among women diagnosed with breast cancer. A more refined Cox proportional hazard modelling is ongoing, which will shed light on how factors such as the use of other metabolic agents, age, and breast cancer subtype may interact with metformin's effect on overall and disease-free BC survival. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P3-06-01.