Metformin in Breast Cancer: Time for Action

A recent convergence of clinical and epidemiologic evidence has linked hyperinsulinemia, insulin resistance, and diabetes to poor breast cancer outcomes. This has been coupled with enhanced understanding of molecular effects of metformin and its potential role in malignancy. It has been revealed that metformin may influence cancer cells through indirect (insulin-mediated) effects, or it may directly affect cell proliferation and apoptosis of cancer cells.^1,2 Preclinical work has demonstrated a beneficial effect of metformin in breast cancer.³ The first evidence of a potential effect of metformin in human breast cancer is reported by Jiralerspong et al⁴ in this issue of Journal of Clinical Oncology.

In their article, Jiralerspong et al⁴ observe that diabetic patients with breast cancer treated with metformin experienced higher pathologic complete response (pCR) rates with neoadjuvant chemotherapy than did those treated with other diabetes medications. The authors looked retrospectively at chemotherapy response rates in a group of 2,592 patients, including 157 women with diabetes, treated with neoadjuvant chemotherapy for early stage or locally advanced breast cancer between 1990 and 2007. Diabetic patients treated with metformin experienced a pCR rate of 24%, which was significantly greater than the pCR rate in diabetic women not treated with metformin (8%; P < .001) and numerically (but not statistically) greater than the pCR rate in women without diabetes (16%; P = .10). In multivariate models adjusting for body mass index, stage, tumor grade, hormone receptor and human epidermal growth factor receptor 2 status, age, presence of diabetes, and use of neoadjuvant taxanes, metformin use remained an independent predictor of pCR with an odds ratio of 2.95 (95% CI, 1.07 to 8.07; P = .04). In this model, metformin use was a better predictor of pCR than were established features such as tumor grade, hormone receptor status, and human epidermal growth factor receptor 2/neu overexpression.

Given that insulin use was two-fold higher in the nonmetformin group compared with the metformin group (33% v 16%), Jiralerspong et al⁴ performed exploratory analyses of the relationship between insulin use and pCR rates and observed that insulin use was associated with a significantly lower pCR rate in the nonmetformin group (0% v 12%; P = .05) but not in the metformin group (27% v 23%; P = .75). The authors concluded that insulin use could have contributed to the differences in pCR rates in the two diabetic groups. Despite the higher pCR rate in patients treated with metformin, there were no differences in rates of disease recurrence among the three groups (P = .66), and both diabetic groups had worse overall survival compared with the nondiabetic group (P = .02). Given the large number of deaths not related to breast cancer reported in the study (208 recurrences and 500 total deaths) and the excess mortality seen in diabetic patients, it is not surprising that there was no significant difference in survival between the diabetic patients who received metformin and those who did not. However, given the tripling of the pCR rate in diabetic patients treated with metformin compared with those not receiving the drug in this study, additional work will be necessary to determine whether pCR is a relevant prognostic marker in this group of patients.

A few limitations of the study by Jiralerspong et al⁴ should be acknowledged. First, identification of diabetic patients came from patient self-reports, and some patients could have been misclassified. Furthermore, details regarding medication use were obtained from medical record review and pharmacy records from a tertiary care cancer institute, and it is possible that patients may have been taking other medications prescribed by primary care or other physicians. No information was available regarding the rationale underlying the choice of diabetic agents for a particular patient, potentially introducing confounding factors into this retrospective analysis. Finally, the number of diabetic patients was small, and almost 50% of the diabetic patients initially identified in the database were excluded from the analysis for a number of reasons. Despite these limitations, the results of this study are intriguing.

A beneficial effect of metformin in breast cancer appears to be biologically plausible. By inhibiting transcription of key gluconeogenesis genes in the liver and increasing glucose uptake in skeletal muscle, metformin reduces levels of circulating glucose, increases insulin sensitivity, and reduces the hyperinsulinemia associated with insulin resistance,⁵ all of which are factors associated with breast cancer prognosis. Although several mechanisms of metformin action have been proposed (Fig 1), activation of adenosine monophosphate-activated protein kinase (AMPK) has been found to play a prominent role in mediating the effects of this drug.⁶ AMPK is a central cellular energy sensor, activation of which leads to suppression of many of the processes highly dependent on ample cellular adenosine triphosphate supply, including gluconeogenesis, protein and fatty acid synthesis, and cholesterol biosynthesis, as well as promotion of catabolic processes such as fatty acid beta oxidation and glycolysis.⁷

In addition to the effects of metformin and AMPK on metabolic processes, activation of AMPK results in rapid inhibition of cellular protein synthesis and growth. Mechanistically, AMPK achieves this by phosphorylation and stabilization of the protein product of the tuberous sclerosis complex tumor suppressor gene TSC2, which serves as an integrator of various regulatory inputs implicated in cell growth and transmits them to the master regulator of cellular protein synthesis, the mammalian target of rapamycin (mTOR).⁸ In addition to mediating the inhibitory effects of AMPK on protein synthesis, TSC2 integrates several regulatory inputs that affect cellular protein translation, notably signals emanating from the availability of oxygen and growth factor–dependent stimulation of the phosphoinositide 3-kinase/phosphatase and tensin homolog/v-akt murine thymoma viral oncogene homolog and the Ras/Raf/extracellular signal-regulated kinase pathways, two of the most frequently deregulated signaling cascades in human breast cancer.^9,10 Significantly, activation of mTOR-dependent protein translation is often found in breast cancer specimens and has been shown to correlate with malignant progression and adverse prognosis.¹¹ Moreover, the mTOR signaling throughput correlates with hyperplastic changes in breast epithelium—from normal breast epithelium to hyperplasia and atypia to invasive lesions¹²—and contributes to resistance of breast cancer cells to chemotherapy, trastuzumab, and tamoxifen.¹³ A note of caution regarding use of metformin in breast cancer was injected by a recent report suggesting that metformin promoted an angiogenic phenotype in estrogen receptor–negative, melanoma differentiation-associated MB-435 cells.¹⁴ However, this report has been criticized because these cells are actually derived from melanoma, not breast cancer,¹⁵ and the dose of metformin was a minimum 45-fold excess of the recommended human dose,¹⁶ raising concerns about the relevance of the observation to human breast cancer.

Thus, there is a strong preclinical rationale for a potential beneficial effect of metformin in breast cancer outcomes. The results presented by Jiralerspong et al⁴ provide clinical support for a beneficial effect of metformin, although as noted above, interpretation of these results is limited by the observational nature of the data. Additional support has arisen from epidemiologic studies. Evans et al¹⁷ reported that the risk of subsequent cancer diagnosis (all cancer types, including breast cancer) was reduced in patients with type II diabetes who received metformin (with an odds ratio of 0.85 for any metformin exposure v no metformin exposure). The protective effect increased with greater metformin exposure (measured as total dose prescribed or total duration of use). Additionally, Bowker et al¹⁸ reported that cancer mortality was lower in patients with diabetes receiving metformin versus sulfonylureas or insulin (hazard ratio, 0.55 to 0.77), but they did not study patients with diabetes who were not receiving any drug therapy. A recent study¹⁹ has provided evidence that reduction in cancer risk does not occur with all oral diabetic agents. Thiazolidenediones had no effect, whereas effects of sulfonylureas (insulin secretagogues) differed according to the specific drug studied; for example, glibenclamide was associated with increased cancer risk.

Taken together, current evidence suggests that there is a real possibility that metformin may have beneficial effects on breast cancer outcome. Metformin is an inexpensive and safe drug. Its most serious toxicity is lactic acidosis, occurring in three of 100,000 patient-years of use. There is debate about whether lactic acidosis in diabetics is a result of metformin use or underlying diabetes. Risk is significantly reduced when metformin use is avoided in those patients with hepatic, cardiac, or renal compromise and in those age 80 years or older. Minor gastrointestinal upset is the commonest toxicity, leading to cessation of therapy in approximately 10% of individuals.

We are currently involved in developing a large-scale phase III trial of metformin in the adjuvant breast cancer setting. This intergroup clinical trial, led by the National Cancer Institute of Canada Clinical Trials Group (MA.32), is being proposed to evaluate the effects of metformin on breast cancer outcomes, including recurrence and death. The trial will be powered to identify clinically plausible and important effects (hazard ratio, 0.76), and it will include key correlative studies that will explore whether any beneficial effect is seen across a broad group of patients with breast cancer (consistent with a direct effect of metformin on AMPK) or whether beneficial effects are restricted to women with hyperinsulinemia, to those whose tumors are insulin receptor positive, or to those whose insulin levels fall in response to metformin treatment (consistent with an indirect effect of metformin acting through an insulin-mediated mechanism). We believe the science underlying such a trial is strong, the novelty of the intervention is high, and the potential for benefit is large. Although no single piece of evidence is adequate to justify the initiation of such a trial, we believe the totality of currently available evidence, including the results reported by Jiralersprong et al,⁴ has reached a level sufficient to warrant action in initiating such a trial in the breast cancer adjuvant setting.