There is strong clinical evidence that fasting insulin levels are associated with poor BC outcomes across the normal range of insulin levels that is typically seen in women with early stage breast cancer, without evidence of a floor or ceiling effect. At the dose we plan to use in MA.32, metformin safely lowers insulin levels by about 20% in non-diabetic BC and non-cancer subjects who have baseline fasting insulin levels in the normal range. Observational evidence suggests that metformin use in diabetics may reduce cancer risk and mortality and may be associated with enhanced responses to neoadjuvant systemic therapy in diabetic women with BC. We hypothesize that metformin will reduce BC growth through its effect on insulin levels and we believe there is no justification for selecting subjects by insulin levels based on an insulin dependent action of metformin. Growing molecular evidence that metformin may also exert an insulin independent effect on breast cancer via activation of an insulin independent, AMPK-dependent energy stress response and/or inhibition of PI3K/Akt/mTOR signaling strengthens the rationale for studying metformin in BC. The existence of this direct (non-insulin dependent) metformin effect on AMPK/mTOR provides additional justification for inclusion of women with a broad range of insulin levels in MA.32 rather than limiting enrolment to those with insulin levels above an arbitrarily chosen cutpoint. Because metformin is a readily available, inexpensive oral agent, with known (and easily manageable) toxicities, we believe evaluation of its therapeutic effects in early stage BC can proceed at an accelerated pace.