M. K. Breitenstein,1 L. Wang,1 R. M. Weinshilboum,1 G. J. Simon,2 J. Pathak1; 1Mayo Clinic, Rochester, MN, 2University of Minnesota, Minneapolis, MN

BACKGROUND: Metformin use in Type 2 diabetes mellitus (T2DM) patients has been shown to reduce the incidence of breast cancer. However, the impact of metformin and insulin on breast cancer-specific survival outcomes remains controversial. Further, the impact of metformin and insulin on breast cancer-specific mortality outcomes beyond those attributable to mortality outcomes due to T2DM severity remains unclear.
METHODS: 1,180 female patients with unilateral, non-recurrent breast cancer diagnosis between 1998 and 2011 and T2DM at breast cancer diagnosis were identified using Mayo Clinic electronic health record (EHR) data. Median age at breast cancer diagnosis was 67 (34-95) and median follow-up time was 83 (6-191) months. Stratified Cox proportional hazard regression was utilized to separate the effects of metformin (n=299) and insulin (n=197) ≥6 months on T2DM severity (breast cancer-specific mortality events removed) and breast cancer treatment outcomes. A linear offset of T2DM severity was utilized to separately study the effect of metformin and insulin in multivariate Cox models of breast cancer treatment and biology.
RESULTS: Significant protective effects for metformin (HR=0.544, p=0.0017) and detrimental effects for insulin (HR=2.144, p<0.0001) were observed due to their implication for T2DM severity. For breast cancer-specific disease impact, significant (p<0.0500) effects in models of breast cancer treatment and biology adjusted for T2DM severity were observed for metformin (HRs=0.388, 0.339) and insulin (HRs=1.956, 2.201).
CONCLUSION: In this EHR-driven study, we have demonstrated a significant protective effect for metformin and a significant detrimental effect for insulin on breast cancer survival outcomes that is beyond their implication for T2DM severity.