Background: The diabetes drug metformin may improve BC outcomes through enhanced obesity-related physiology or direct anti-tumor effects. We studied the effect of metformin on CA15-3 (the soluble moiety of the MUC1 protein), a marker associated with BC prognosis that also has mitogenic and metabolic effects that favor tumorigenesis. Methods: 3,256 women with T1-3, N0-3, M0 BC who had completed standard therapy (ongoing hormone therapy permitted) provided fasting blood (stored at -80C) at baseline and 6 months. CA15-3 and insulin were measured by Roche ECLIA; leptin, and hs-CRP by Luminex Milliplex MAP and Roche ITA. Spearman coefficients were calculated and comparisons analyzed using Wilcoxon signed rank test and multivariable linear regression models. Tests were two-sided. Results: Mean age was 52.3 and BMI 28.6 kg/m². Tumor and treatment characteristics were balanced between arms (overall: T2/3 in 59.8%, N +ve in 52.9%, grade 1/2/3 in 9.1%/35.3%/54.4%, ER+ in 69.6%, HER2+ in 17.1%; 50.4% underwent mastectomy, 74.0% received radiation, 89.2% chemotherapy, 17% trastuzumab, and 64.4% hormones). Baseline values and 6 month changes are shown below. In multi-variable analyses (including age, BMI, tumor characteristics, treatment), metformin (vs placebo) led to a greater relative reduction in CA15-3 (-5.81%; 95% CI: -3.94% to -7.64%, p < 0.0001). CA15-3 change at 6 months significantly correlated with change in BMI (r=0.10, p<0.0001), glucose (r=0.05, p=0.011) and hsCRP (r=0.05, p=0.022). Conclusions: Metformin significantly lowered CA15-3; change in CA15-3 was associated with improved obesity-associated physiology and BMI, consistent with hypothesized beneficial actions of metformin. Clinical trial information: NCT01101438

Mean values. *significant difference between metformin and placebo, p < 0.05