Metformin and Chemoprevention: Potential for Heart-Healthy Targeting of Biologically Aggressive Breast Cancer

Currently, tamoxifen is the only drug approved for reduction of breast cancer risk in premenopausal women. The significant cardiovascular side effects of tamoxifen, coupled with lack of a survival benefit, potential for genotoxicity, and failure to provide a significant risk-reduction for estrogen receptor-negative breast cancer, all contribute to the low acceptance of tamoxifen chemoprevention in premenopausal women at high-risk for breast cancer. While other prevention options exist for postmenopausal women, there is a search for well-tolerated prevention agents that can simultaneously reduce risk of breast cancers, cardiovascular disease, and type-2 diabetes. Metformin is a well-tolerated oral biguanide hypoglycemic agent that is prescribed worldwide to over 120 million individuals with type-2 diabetes. Metformin is inexpensive, safe during pregnancy, and the combination of metformin, healthy lifestyle, and exercise has been shown to be effective in preventing diabetes. There is a growing awareness that prevention drugs and interventions should make the “whole woman healthy.” To this end, current efforts have focused on finding low toxicity alternatives, particularly repurposed drugs for chemoprevention of breast cancer, including metformin. Metformin's mechanisms of actions are complex but clearly involve secondary lowering of circulating insulin. Signaling pathways activated by insulin also drive biologically aggressive breast cancer and predict poor survival in women with breast cancer. The mechanistic rationale for metformin chemoprevention is well-supported by the scientific literature. Metformin is cheap, safe during pregnancy, and has the potential to provide heart-healthy breast cancer prevention. On-going primary and secondary prevention trials will provide evidence whether metformin is effective in preventing breast cancer.

Drug Administration (FDA) in October 1998 to approve tamoxifen as a chemoprevention agent for premenopausal highrisk women.
The benefit of tamoxifen appears to be in risk-reduction of ER+ breast cancer; tamoxifen has failed to demonstrate in highrisk women (1) a significant risk reduction for ER-breast cancer and (2) a survival benefit. An extended analysis (median 16 years) of IBIS-I study participants, continues to shows in the tamoxifen vs. placebo arms "no difference in the number of breast cancer deaths (p = 0.8)" (12).
Despite initial recommendations by the FDA and American Society for Clinical Oncology, very few women take tamoxifen (11); it is estimated that only 5-12% of women offered tamoxifen chemoprevention elect to take tamoxifen (11).
Tamoxifen has been shown to increase risk for cardiovascular events, including venous thrombosis, pulmonary embolism, and stroke, and increases risk for endometrial cancer (12)(13)(14). Other side effects of tamoxifen include hot flashes, dyspareunia, depression, cataracts, weight gain, and bone loss in premenopausal women (12)(13)(14)(15). Consistent with the increased risk of endometrial cancer in humans, a 2013 study in rats showed that 13-week tamoxifen treatment increased DNA point mutations in the liver (16). Lastly, a concern was raised that tamoxifen may be less active in the 5-10% of individuals who carried homozygous variant of the CYP2D2 gene; this gene variant has low activity to convert tamoxifen to its more active metabolite, 4-hydroxytamoxifen. Lacking in the analysis was a consideration of the concentration of 4-hydroxytamoxifen required to saturate ER; consequently, prospective clinical studies did not demonstrate a reduction in tamoxifen efficacy in individuals with the CYP2D2 variant (17).
While tamoxifen is the only agent approved for breast cancer prevention in premenopausal women, other agents have been approved for postmenopausal women. In the NSABP Study of Tamoxifen and Raloxifene (STAR) trial (raloxifene 60 mg vs. tamoxifen 20 mg), raloxifene was shown to reduce the incidence of breast cancer in postmenopausal women (18). Raloxifene does not increase the risk of endometrial cancer, however, the incidence of ischemic heart disease and stroke was equivalent to the risk associated with tamoxifen (18). IBIS-II tested anastrozole (1.0 mg) vs. placebo in postmenopausal women; the study found a significant decrease in breast cancer in women who took anastrozole; there was no increased incidence of fractures or cardiovascular disease (19). In the Mammary Prevention.3 trial (MAP.3) exemestane (25 mg) vs. placebo in postmenopausal women was associated with a decreased incidence of both ductal carcinoma in situ and invasive breast cancer; with a median follow-up of 3 years, side effects and impact on quality of life were minimal (20).

NEED FOR HEART-HEALTHY BREAST CANCER CHEMOPREVENTION
Women are not just at risk for breast cancer but also face the risk of developing heart disease, obesity, and type-2 diabetes. Furthermore, with the risk of currently available chemoprevention agents potentiating cardiovascular disease, there is a need to identify agents that can effectively target both conditions: breast cancer and cardiovascular disease. To this end, current efforts have focused on finding alternative prevention strategies that have the potential to reduce not just breast cancer but also reduce the risk for cardiometabolic diseases. Potential strategies have included exercise, aspirin, and metformin.

Metformin
Metformin (1,1-dimethylbiguanide hydrochloride) is a welltolerated oral agent that is prescribed for first-line treatment of type-2 diabetes (21,22) and is approved for treatment of polycystic ovary and gestational diabetes (23). Metformin is well-tolerated by the majority of patients; common metformin side effects include lack of appetite, epigastric pain, nausea, and diarrhea (24). The most significant potential side effect is lactic acidosis; consequently, metformin is not prescribed in individuals with kidney and/or liver disease (23,25). The mechanism of action of metformin remains a topic of current investigations. It is accepted that metformin inhibits hepatic gluconeogenesis and decreases intestinal absorption of glucose, secondarily decreasing circulating insulin (21,26). Metformin is also thought to indirectly increase insulin sensitivity by increasing peripheral glucose utilization (21).
Until recently, most clinical care has focused on treatment of type-2 diabetes rather than its prevention. However, several wellcontrolled studies have shown that it is possible to prevent type-2 diabetes through a combination of diet, exercise, and metformin. The Diabetes Prevention Program/Diabetes Prevention Program Outcomes Study (DPP/DPPOS) is the largest and longest clinical trial of metformin for the prevention of type-2 diabetes (27,28). Study participants in the DPP/DPPOS cohort have over 15 years prospective assessment of the impact of metformin and lifestyle modification on type-2 diabetes, cardiovascular events, safety, and fiscal outcomes (27). Metformin and intensive lifestyle modification resulted in a 50% type-2 diabetes risk-reduction in women with a history of type-2 diabetes (29). Based on findings from the DPP/DPPOS study, in 2014, the American Diabetes Association (ADA) published formal recommendations for prevention of type-2 diabetes (30). Recommendations included: (1) individuals with impaired glucose tolerance or a HgbA1c 5.7-6.4 should be referred to a life-style modification (7% weight loss target) and moderate physical activity (e.g., walking) for 150 min/week (30). These recommendations may also prove beneficial in modifying breast cancer risk; as outlined below, metformin is undergoing testing for primary and secondary breast cancer prevention.

Metformin and Breast Cancer: Epidemiology Studies
Population-based studies provide evidence that cancer incidence and mortality decreased in individuals with cancer who took metformin (31)(32)(33). In a retrospective study of women with breast cancer who received neoadjuvant chemotherapy individuals who took metformin had a higher rate of pathologic complete remission vs. those did not [24 vs. 8%, p = 0.007; (34)]. In a 2014 meta-analysis, individuals who took metformin had a lower incidence of breast cancer (SRR = 0.94; 95% CI, 0.90-0.99) (35). These epidemiologic studies represent a starting point for recent prospective clinical trials testing the impact of metformin on primary and secondary breast cancer prevention.
Epidemiology studies investigating the impact of metformin on breast cancer incidence are limited by several factors. These factors include: (1) racial and ethnic differences in body mass index (BMI), (2) inability of BMI to precisely identify individuals who are metabolically unhealthy, and (3) the heterogeneity of breast cancer as a disease. A BMI ≥30 kg/m 2 is the most frequently used measure of adiposity (36). BMI is an inexact measure of risk, particularly when comparing individuals of different race and ethnicity. Muscle tissue weighs significantly more per unit volume than adipose tissue; consequently fit, muscular individuals can be mistakenly identified as overweight (BMI 25-30 kg/m 2 ) or obese.
BMI is not a precise measure of metabolic health. Over the past 20 years, the observation has been made that some individuals with a BMI > 30 kg/m 2 are metabolically healthy, "metabolically healthy obese" (37). In contrast to individuals who are obese but metabolically healthy, there are also individuals with a normal BMI (BMI <25 kg/m 2 ) who have abnormal metabolic profiles and are at increased risk for cardiovascular disease and type-2 diabetes. Current definition of metabolically unhealthy individuals with a normal BMI includes (1) BMI <25 kg/m 2 , (2) insulin-resistance, hypertriglyceridemia, (3) abdominal fat distribution, and (4) elevated blood pressure (37).

TYPE-2 DIABETES, METFORMIN, AND BREAST CANCER SUBTYPES
Type-2 diabetes is well-established to increase a woman's risk of developing breast cancer. The association between Type-2 diabetes and breast cancer subtypes, however, remains a work in progress, particularly since the majority of studies are underpowered. A case-control study of 916 postmenopausal women with breast cancer cases and 1,094 population-based controls conducted by Garcia-Esquinas et al. found that type-2 diabetes was associated with a 2.25-fold increased risk for triple negative breast cancer (TNBC) (38); this study was limited by a low number of TNBC and the study of only postmenopausal women. The Carolina Breast Cancer Study included 225 women with TNBC; no statistical association was found between type-2 diabetes and TNBC; unfortunately, this study did not test for the association between insulinresistance and TNBC (39). A case-case study by Lara-Medina et al. of Latinas with breast cancer (469 women with TNBC) found no statistical association between type-2 diabetes and TNBC (40).
The most complete and well-designed epidemiologic study was a retrospective multi-center population-based case-case study of 4,557 women with breast cancer ages 20-69 years old performed by Chen et al.; 1,446 women had TNBC (41). The investigators identified that women with type-2 diabetes had a 38% (95% CI: 1.01-1.89) increased odds of having TNBC (vs. women without type-2 diabetes) (41).
Interestingly, Chen et al. also found that current and extended-time metformin use (13-24 months metformin) within 2 years of diagnosis, increased the odds of a woman having TNBC (OR = 1.54; 95% CI: 1.07-2.22 and OR = 1.80; 95% CI:1.13-2.85, respectively) (41). These latter results are puzzling, given the ability of insulin to activate signaling pathways that drive the aggressive biology of TNBC and the known ability of metformin to lower circulating insulin.
Epidemiologic studies are powerful tools for generating associations but do not test mechanisms. First off, as pointed out by Chen et al., it may be that the women who had the most poorly controlled diabetes (41), were the individuals who had the longest use of metformin; HgbA1c values for these individuals were not reported. While the number of women Frontiers in Public Health | www.frontiersin.org using metformin were carefully determined, it is not clear that the investigators incorporated insulin-use (insulin-dependent type-2 diabetes) in their risk models. Furthermore, these risk models do not account for individuals with insulin-resistance ( Figure 1). Ultimately, the studies by Chen et al. are extremely important because they highlight how complex the associations between metformin-use, insulin-use, and TNBC are likely to be and underscore the importance of window-of-opportunity trials and ongoing prospective metformin prevention trials (such as MA-32, described below).

METFORMIN TRANSPORT AND MECHANISM OF ACTION
After oral administration, the oral bioavailability is 55 ± 16% (mean ± standard deviation); metformin is predominantly absorbed in the small intestine (43). Metformin is excreted unchanged in the urine and has a half-life between 4 and 8 h (44). Metformin's absorption and renal clearance is primarily mediated by OCT2/MATE1/MATE2-K (organic cation transporter 2/multidrug and toxin extrusion 1/ multidrug and toxin extrusion 2-K) (45). There are frequent polymorphisms in OCT2, MATE1, and MATE2-K that impact clearance metformin [ Table 1; (46,62)]. Up to 9% of non-Hispanic Whites exhibit an "OCT1 null phenotype" (46). To date, there have been variable findings in pharmacogenomic studies in humans. However, there is evidence that cancer cell lines with high MATE2 expression may be resistant metformin's growth inhibitor effects (63). Despite metformin being one of our oldest medications, the precise molecular mechanism(s) underlying metformin's insulin-lowering effects, as well as its potential anti-neoplastic potential, are not completely understood. It is well-accepted that metformin inhibits hepatic gluconeogenesis and secondarily lowers circulating insulin. However, the precise mechanism(s) of metformin-action remains a work in progress. Two major pathways are thought to account for the main actions of metformin and metformin's proposed anti-cancer effects ( Figure 2); both pathways converge on mammalian target of rapamycin (mTOR): (1) AMPK (adenosine monophosphateactivated protein kinase) independent, driven by metformin's FIGURE 2 | Impact of metformin on insulin-resistance, serum insulin, and signaling pathways important for breast cancer aggressive biology.
Frontiers in Public Health | www.frontiersin.org ability to secondarily lower serum insulin and (2) AMPKdependent, regulated by metformin-suppression inhibition of mitochondrial complex-I (complex-I).
Increasing attention has been paid to identifying molecular mechanisms that promote chemotherapy-resistance. Kevin Struhl's group first showed in 2009 that 0.1 mM metformin in vitro blocked transformation and killed cancer-like stem cells (74). The combination of metformin and doxorubicin in a mouse xenograft model (metformin 100 µg/ml) exhibited synergy. These results provided a potentially novel mechanism of action for metformin and an experimental rationale for using the combination of metformin and chemotherapy. The metformin doses in this study, however, were supratherapeutic and this very interesting mechanism of metformin-action remains an area of active investigation.
There is also evidence that metformin acts on the tumor microenvironment. Metformin increases intracellular oxygen; this increase is thought to reduce tumor hypoxia (75). Metformin's decrease in hypoxia has been shown to inhibit hypoxia-inducible factor 1 (HIF1) and vascular endothelial growth factor A (VEGFA) driven angiogenesis; there is also evidence for a direct anti-tumor effect on endothelial cells (76,77). Metformin's increase in tumor oxygenation and or activation of AMPK is thought to shift cancer associated macrophages from a M2 to an M1 phenotype (78). Metformin has been shown to reduce programmed death-ligand 1 (PD-L1) expression on cancer cells, increase lymphocyte anti-tumor cytotoxicity, and downregulate myeloid derived tumor cell activity (79)(80)(81)(82). Taken together, these findings highlight a potential role for metformin to be used in concert with immune-therapy.

Current Consensus
While the study of metformin's molecular mechanisms of actions remain an area of active research, there is a growing consensus of the key signaling targets of metformin. The following consensus statement for metformin's key mechanisms of actions is updated from Pernicova and Korbonits (83): • Metformin alters cellular energy metabolism and promotes metabolic reprogramming. • Metformin acts to lower glucose and increase insulinsensitivity: (1) primarily by inhibiting hepatic gluconeogenesis and glucagon-signaling and (2) to a lesser degree, in the skeletal muscle by increasing glucose uptake.

RATIONALE FOR METFORMIN'S ABILITY TO PREVENT BIOLOGICALLY AGGRESSIVE BREAST CANCERS
In breast cancer, particularly TNBC and basal-type breast cancer, activation of PI3K/AKT/mTOR-signaling pathway is associated with poor prognosis (84,85). Activation of the PI3K/AKT/mTOR results in cell cycle progression, apoptosisresistance, and invasion (86,87). PI3K/AKT/mTOR is a regulator of glucose metabolism and aerobic glycolysis (Warburg effect) (88)(89)(90). The Warburg effect is directly linked to aggressive cancer biology due to its impact on glycolysis/glucose-uptake; increased glycolysis/glucose-uptake promotes increased growth, mitochondrial dysfunction, and apoptosis-resistance. Metformin targets the PI3K/AKT/mTOR pathway and promotes metabolic reprogramming. These actions support the use of metformin for prevention of biologically aggressive breast cancers (Figures 2-4).
Prevention options for premenopausal women who carry a deleterious germline BRCA mutation are limited. There is strong scientific rationale for testing metformin in chemoprevention of breast cancer in BRCA mutation carriers: (1) metformin activates AMPK and (2) signaling networks regulated by both AMPK and BRCA1, include PTEN, p53, and acetyl coenzyme A carboxylase alpha (ACCA) (83,91,92). AMPK regulates the phosphorylation/dephosphorylation cycles of ACCA (93,94). Given that AMPK and BRCA1 both inactivate ACCA, it is hypothesized that metformin might compensate for BRCA1-loss. Further rational for metformin prevention in BRCA1 mutation carriers has been provided by Cuyas et al. (95). Introduction of BRCA1 mutation185delAG in MCF10A cells resulted in metabolic reprograming including (1) mitochondrial activation, (2) increased glucose-and glutamine-dependent activation of the tricarboxylic acid cycle (TCA), and (3) increased production of acetyl-CoA and malonyl-CoA (95). Metformin was shown in vitro to inhibit (1) mitochondrial biosynthetic capacity, (2) the TCA cycle, and (3) generation of lipogeneic precursors. The authors hypothesize that the ability of metformin to block ("starve") mitochondrial-generated biosynthesis, might provide further rationale for using metformin for cancer prevention in women with germline BRCA1-mutation (95). As described below, to date, the epidemiologic and clinical trials using metformin have yielded conflicting results. The ability of metformin to prevent biologically aggressive breast cancers, particularly TNBC, requires the completion of the on-going prospective trials, such as MA-32.

Clinical Studies
Dr. Pamela Goodwin has been a pioneer in the use of metformin for lowering insulin and breast cancer chemoprevention; she has developed some of the first trials testing metformin. In a trial of 32 women (4 dropout) with early stage breast cancer and fasting insulin of ≥45 pmol/L and glucose <7.0 mmol/L, administration of metformin 1500 mg per day for 6 months was associated with a 22.4% decrease in serum insulin [p = 0.024; (34)]. This study provided the rational for subsequent randomized clinical trials using metformin vs. placebo.
Window-of-opportunity trials provide important insight into metformin's mechanisms of action but have had conflicting results. In a Scottish trial, Hadad et al. tested the impact of metformin 500 mg ramp up and then 1,000 mg twice a day on Ki-67 and gene expression on 8 pilot women and a further 47 women with primary breast cancer; 7/32 women receiving metformin withdrew due to gastrointestinal upset (96). In women receiving metformin, Ki-67 fell significantly following metformin in both the pilot study (p = 0.041) and in the metformin arm (p = 0.027) but was unchanged in women who did not take metformin (96). Gene expression studies showed a decrease in mRNA expression in genes regulating AMPK; further analysis demonstrated that tumor necrosis factor receptor signaling, and mTOR-and AMPK-signaling were impacted by metformin (96).
The  (97). However, there was a differential impact on Ki-67 based on insulin-resistance (measured by homeostatic model assessment-HOMA). In women with HOMA >2.8 there was a 10.5% decrease in mean Ki-67 vs. an 11% increase in women with HOMA <2.8 (pinteraction = 0.045); women with Luminal B breast cancer had the greatest benefit [p = 0.005; (97)]. Further, biomarker analysis showed that this trial represented a significant accomplishment, given the difficulty of coordinating window-of-opportunity trials; importantly, this trial provided a key piece of evidence that non-diabetic metabolically unhealthy women may benefit from metformin chemoprevention (97). A third window-ofopportunity trial reported by Kalinsky et al. in women with early stage breast cancer and a BMI ≥30 reported that in women taking 1,500 mg metformin there were no significant differences in Ki-67 for either DCIS or invasive breast cancer (98). There has been significant discussion about the differences observed in these: trials; one potential difference is that women in the Scottish trial had larger breast cancers and therefore, had larger tumors for analysis [see Kalinsky and Hershman for a more in-depth analysis (99)]. Still, given the short duration of window-of-opportunity trials, longer duration trials with a cancer endpoint are required. See Table 2A for additional clinical   and window-of-opportunity metformin trials in women with breast cancer. Currently many ongoing prospective clinical studies are testing the metformin for primary and secondary prevention of breast cancer (Tables 2A,B). Together, these clinical studies represent an important investment by the National Institute of Health, United States (NIH), European Cancer trials groups, and the National Cancer Institute, Canada (NCIC) ( Table 2). The largest adjuvant (secondary prevention) trial is NCIC MA-32, comparing metformin 850 mg p.o. twice a day vs. placebo (NCT01101438) in women with breast cancer; the endpoint of this trial is breast cancer recurrence. After 2,382 women were enrolled, in 2012, the eligibility criteria were amended to mandate TNBC status for patients with T1cN0 disease and at least one adverse tumor characteristic for patients with T2N0 tumors. Interim analysis of the first 500 women taking metformin entered in MA-32, showed at 6 months there was a significant decrease in weight (−3.0%), serum glucose (−3.8%), and serum insulin (−11.1%) (102); further results from this trial are pending. ACTRN12610000219088 is currently testing the impact of metformin (1,000 mg) on LKB1 and AMPK signaling; NCT0430079 tests the impact of metformin in preventing grade 3-4 in (1) men and (2) post-menopausal women receiving treatment for ER/PR+, HER2-not amplified advanced breast cancer, with a PI3Kmutation [METALLICA trial]. Primary prevention studies include (1) NCT01793948: randomized testing the impact of metformin on postmenopausal women with high breast density, (2) NCT01905046: metformin vs. placebo in highrisk premenopausal women (including BRCA mutation carriers) with cytologic atypia, and (3) NCT01905046: randomized testing of whether metformin alters breast density, serum IGF-1/IGFBP-e ratios, IGF-2, and leptin/adiponectin ratios, body weight/body composition (109). See Table 2B for additional trials. Given the wealth of primary and secondary metformin chemoprevention trials, it is anticipated that over the next 5 years, these trials will provide important insights into whether metformin is a viable chemoprevention agent for breast cancer.

METFORMIN AND HEART-HEALTHY PREVENTION OF BIOLOGICALLY AGGRESSIVE BREAST CANCERS
Metformin is cheap, safe during pregnancy, and has shown to prevent type-2 diabetes. There is a need for prevention drugs that target both ER+ and ER-breast cancer as well as providing prevention for cardiometabolic disease. Metformin clearly lowers insulin-signaling; signaling pathways activated by insulin are known to drive biologically aggressive breast cancer and predict poor survival in women with breast cancer. Despite the fact that metformin targets many key breast cancer pathways, there is much to be learned about whether metformin can prevent breast cancer and/or breast cancer recurrence. Window-of-opportunity trials provide important clues to metformin's impact on normal and malignant breast tissue, but results have not been entirely consistent. Currently, it is unclear which breast cancer subtypes may benefit the most from metformin. It is likely that MA-32 will provide answers to many of these questions. There is also much to be learned about metformin, insulin resistance, and BMI; specifically, whether metformin's impact is only in women who are metabolically unhealthy and/or have high BMI, or whether metformin can benefit all women. Biomarker studies that define key signaling pathways impacted by metformin will be critical to design and inform future clinical trials. Over the next 5 years on-going primary and secondary prevention trials will show (or not show) the ability of metformin to prevent breast cancer. Hopefully, these studies will not just provide a yes/no answer also provide the biomarkers to determine which women will maximally benefit from metformin. In the words of several of my patients "Please do not quote statistics at me; these statistics are about other women. If I take a prevention agent, I want to know if the prevention agent is working in my breasts."

AUTHOR CONTRIBUTIONS
All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.