Evaluation of metformin in early breast cancer: a modification of the traditional paradigm for clinical testing of anti-cancer agents
Metformin, an inexpensive oral agent commonly used to treat type 2 diabetes, has been garnering increasing attention as a potential anti-cancer agent. Preclinical, epidemiologic, and clinical evidences suggest that metformin may reduce overall cancer risk and mortality, with specific effects in breast cancer. The extensive clinical experience with metformin, coupled with its known (and modest) toxicity, has allowed the traditional process of drug evaluation to be shortened. We review the rationale for a modified approach to evaluation and outline the key steps that will optimize development of this agent in breast cancer, including discussion of a Phase III adjuvant trial (NCIC MA.32) that has recently been initiated.
KeywordsBreast cancer Metformin Adjuvant Clinical testing
Traditionally, oncology drugs have emerged from preclinical research that identifies compounds having biologic effects that target attributes that are unique to, or more common in, cancer cells. In the past, these drugs have often targeted the machinery involved in cell division in a very blunt fashion (e.g., chemotherapeutic agents targeting DNA or mitotic spindles). Because these effects were not specific to cancer cells, toxicity to normal tissues (notably rapidly proliferating tissues such as white blood cells) was often high, limiting the clinical use of the agent. More recently, reflecting advances in understanding the molecular characteristics that distinguish cancer from non-cancer cells, this process has become more finely tuned, with a focus on identification of agents that target key cell surface receptors, oncogenes, and signaling pathways that are preferentially activated or suppressed in cancer cells. This approach has led to an explosion of potential biologic agents that may have more specific anti-cancer effects than traditional chemotherapeutic agents.
In the usual course of events, these novel agents are initially evaluated pre-clinically using in vitro and in vivo approaches; this preclinical study may be supplemented by observational clinical research that evaluates the prevalence and prognostic significance of the drug target in specific cancers. Promising drugs are then moved through a step-wise clinical evaluation, involving Phase 1, 2, and 3 trials that are conducted in progressively more targeted cancer populations to obtain information on pharmacokinetics, optimal dose, toxicity, anti-cancer activity, and, ultimately, the clinical utility of the new agent in relation to current standard treatment. Recent refinements, designed to speed up this process, have involved the use of combined Phase 1/2 designs as well as randomized Phase 2 trials.
One successful example of this molecularly based, step-wise approach in breast cancer involved the identification of the HER2/neu receptor as a key driver of growth in a subset of breast cancers . This was followed by the development and demonstration of efficacy of agents targeting the HER2/neu receptor (first in the metastatic setting, and then in the adjuvant setting) [2, 3]. The initial agent to be developed was the monoclonal antibody trastuzumab; other agents, such as the tyrosine kinase inhibitor lapatinib, were subsequently identified, and others are under active investigation . This approach, which has dramatically improved outcomes in HER2 positive breast cancer, is a powerful one. However, progress for other agents has not been as straightforward as initially anticipated because of difficulties in identifying targets that are key drivers of cell growth in specific cancer types (the HER2/trastuzumab success may prove to be the exception rather than the rule), the development of compensatory molecular mechanisms within cancer cells that lead to drug resistance and the emergence of clinically significant toxicities that limit drug use.
In the case of metformin, an oral drug that is widely used in the treatment of type 2 diabetes, two key factors have modified the approach to its evaluation as an anti-cancer agent. First, the initial interest in this agent has come from clinical and epidemiologic research which has provided an empiric basis for its evaluation in the clinical setting, even though the precise mechanism(s) by which metformin may influence cancer have not yet been defined. Ongoing pre-clinical research has informed and intensified interest in the development of metformin. The second factor modifying the approach is the extensive clinical experience that exists with metformin; its well-defined (and modest) toxicity profile has allowed at least parts of the traditional process of evaluation to be avoided.
The evidence suggesting that metformin may influence breast cancer outcomes comes from several sources. A series of observational studies published in the last 5 years have reported reduced cancer incidence and/or mortality among diabetics who receive metformin (vs. other drugs) to treat type 2 diabetes [5, 6, 7, 8, 9, 10]. These studies also provide evidence that those receiving the highest doses or the longest durations of metformin therapy may have the lowest cancer rates. Although a causal role of metformin in cancer incidence and mortality cannot be demonstrated using epidemiologic data alone, and there is potential for confounding of these associations by factors, such as obesity, which may influence both the risk of cancer and the selection of metformin to treat diabetes, the consistency and magnitude of the cancer-lowering effect seen across these studies that have used different designs and have been conducted in different settings strengthen the justification for clinical evaluation.
In parallel to this epidemiologic research, observational clinical studies have reported associations of circulating levels of insulin (or C-peptide, which is cleaved from pro-insulin when insulin is released) with outcomes in non-diabetic patients with breast cancer [11, 12, 13]—higher insulin levels (predominantly within the range considered normal) have been shown to be independently associated with poor outcomes; in one study, these effects persisted for 5 years post-diagnosis but not thereafter . An association of insulin with breast cancer outcome is biologically plausible given that insulin receptors, most commonly the fetal form of the receptor, are over-expressed on breast cancer cells . The fetal insulin receptor binds insulin-like growth factors (IGFs) 1 and 2 in addition to insulin, and it frequently hybridizes with the IGF 1 receptor (usually present on breast cancer cells at a lower concentration than the insulin receptor) to bind these same ligands. Receptor activation stimulates signaling through predominantly proliferative pathways (in contrast to the metabolic pathways that are stimulated when insulin binds to the adult insulin receptor) leading to enhanced growth. These insights have led to the hypothesis that the fetal insulin receptor may act as a molecular switch on breast cancer cells which allows circulating insulin to stimulate cell growth .
Additional support for a role of metformin in breast cancer outcomes comes from a retrospective clinical study that reported significantly increased pathologic complete response (pCR) rates to standard neoadjuvant chemotherapy in diabetic breast cancer patients who were receiving metformin (24% pCR) compared to diabetics not receiving metformin (8% pCR), with intermediate rates in non-diabetics who did not receive metformin (16% pCR) . In a separate study in non-diabetic breast cancer survivors without clinical evidence of cancer, results showed that metformin lowered circulating insulin levels by 22% . Emerging results of pre-operative window of opportunity studies in non-diabetic breast cancer patients suggest that metformin impacts expression of a range of genes in breast cancer cells, including those related to adenosine monophosphate kinase (AMPK), which is thought to mediate metformin effects on mammalian target of rapamycin (mTOR) (see below).
In evaluating this observational evidence, an important caveat is that much of the evidence regarding potential beneficial effects of metformin has been generated in diabetic subjects. Many of these individuals (and the cancers they develop) have been exposed to prolonged periods of hyperinsulinemia during a pre-diabetic phase that may last a decade or longer. Both type 2 diabetes and hyperinsulinemia have been associated with increased breast cancer risk in epidemiologic studies . It is possible that different associations of metformin with cancer may exist in non-diabetic subjects who have not had similar exposure. The emergence of clinical data in non-diabetic breast cancer patients (reviewed above), coupled with observations that low dose metformin (250 mg/day) may reduce aberrant crypt foci in rectal epithelium , reduces these concerns.
Concurrent preclinical study has magnified interest in metformin as an anti-cancer agent and has identified a range of potential mechanisms of anti-cancer action. Reports that metformin activates AMPK through an LKB1/STK11-mediated mechanism [20, 21] (believed to be the mechanism by which metformin reduces gluconeogenesis in the liver leading to lower circulating insulin levels) have raised the possibility that metformin may also influence cancer cells directly via this mechanism (i.e., independent of changes in circulating insulin). A series of studies has provided in vivo and in vitro evidence that direct LKB1/STK11 activation of AMPK by metformin in tumor cells results in downstream inhibition of mTOR signaling and leads to reduced protein synthesis and cell proliferation [22, 23]. Subsequent reports have identified other potential insulin-independent anti-cancer effects that may not be mediated by AMPK, including generalized effects on gene expression (notably cell cycle related genes), reduction of cyclins D1 and E [24, 25], effects on cancer stem cells that may involve synergy with chemotherapeutic agents, increased autophagy, and others . Additional studies have reported that metformin may lower aromatase activity and, in HER2-positive breast cancer cell lines, reduce HER2 protein expression and overcome resistance to HER2-targeted agents, suggesting it may have added effects when used in combination [27, 28]. Metformin may have unique effects in triple negative cell lines , including S-phase cell cycle arrest as well as reduced proliferation, colony formation, and apoptosis. There is no current consensus regarding the relative importance of these potential mechanisms of metformin activity in breast cancer. Notwithstanding these reports of breast cancer subtype-specific effects, metformin has also been reported to inhibit growth of a broad variety of breast cancer call lines regardless of ER, PgR, HER2, or p53 status providing support for evaluation of metformin in all breast cancer subtypes [25, 29].
Extension of preclinical in vitro and in vivo research to investigate the contributions of both insulin-mediated and insulin-independent mechanisms of action across the full range of molecularly defined breast cancer subtypes. This research should also examine effects and mechanisms of metformin action across the spectrum of mammary neoplasia ranging from carcinogenesis to local growth and metastasis, and it should explore potential interactions (synergistic and antagonistic) of metformin with existing agents used to treat breast cancer. Many of these studies are ongoing.
Conduct of small clinical studies (mainly Phase 2) in the neoadjuvant or metastatic settings (including window of opportunity studies) that incorporate serial collection of clinical samples (blood, tumor) to explore both insulin-dependent and insulin-independent metformin effects in a broad range of breast cancer patients. Short term window of opportunity pre-operative studies will typically focus on mechanistic effects, while longer-term, neoadjuvant and metastatic studies will also provide evidence regarding anti-tumor efficacy and potential synergistic effects with other agents, including emerging biologic agents that target elements of pathways potentially affected by metformin, such as phosphatidylinositol-3 kinase (PI3K), mTOR and IGF-1 receptors. Studies addressing at least some of these issues in a variety of types of breast cancer are ongoing or planned.
Initiation of large scale Phase 3 trials in the adjuvant setting. Traditionally, Phase 3 adjuvant trial development would await further preclinical and clinical research (including Phase 3 trials in the metastatic setting) that would be conducted over a number of years. Such research could potentially define the relative importance of insulin-dependent versus insulin-independent mechanisms of action in the clinical setting and lead to targeting of metformin use to individuals with specific host factors such as high insulin levels or obesity, or to those with specific tumor factors (e.g., triple negative, LKB1 positive, etc.). It might also identify the magnitude of anti-tumor efficacy of metformin, if it exists, using a smaller sample size than is required in the adjuvant setting. However, given the epidemiologic and clinical evidence reviewed above, coupled with a favorable toxicity profile, we believe sufficient evidence currently exists to embark on a Phase 3 adjuvant trial evaluating the effect of adding metformin (versus placebo) to standard therapy on cancer outcomes in a non-targeted early-stage breast cancer population. The adjuvant setting offers advantages in terms of tumor biology; cancer cells are less likely to have unfavorable genetic and molecular profiles that may arise spontaneously or secondary to prior drug exposure in the metastatic setting. This setting also offers the opportunity for maximal impact on patient outcome, given the aim of extending the period of survival without invasive disease in otherwise healthy women. The Canadian NCIC Clinical Trials Group (CTG), with funding from the Cancer Therapy Evaluation Program (CTEP) of the National Cancer Institute, has recently initiated such a trial in the North American Breast Cancer Group. The trial, NCIC CTG MA.32, will enroll 3582 patients with node-positive or high-risk node-negative breast cancer, who are receiving standard therapy for their breast cancer (surgery, radiation, chemotherapy, biologics apart from those targeting the IGF1, or PI3K pathways). Subjects will be randomized to receive metformin 850 mg (or placebo) twice daily for 5 years, including a 1-month ramp up in dose. Women with diabetes or elevated glucose are excluded (because of the use of placebo) as are those at increased risk of metformin associated lactic acidosis (age > 75 years; heart, liver, or kidney organ dysfunction). The primary outcome is invasive disease-free survival (IDFS); secondary outcomes include overall and distant disease-free survival, breast cancer-specific survival, as well as adverse events (including cardiovascular- and diabetes-related hospitalizations), and quality of life. The trial is powered to identify a hazard ratio 0.76 in the metformin arm, with 80% power and a type 1 error of 0.05 (2-tail), allowing for two interim analyses. The primary efficacy analysis is anticipated 6 years after the first subject is enrolled. In designing the trial, evidence obtained from preclinical research that metformin may have insulin-independent mechanisms of actions and that in vitro activity has been seen across a range of breast cancer subtypes has led to inclusion of an unselected population of early-stage patients. Because of uncertainty regarding these issues, correlative research that will examine whether baseline fasting insulin levels and tumor insulin receptor expression predict effects of metformin has been embedded, allowing investigation of the role of insulin-dependent mechanisms of action (most strongly supported by observational studies in humans). Formal subset analyses are planned in hormone receptor-negative and triple-negative cancers to examine key hypotheses relating to differential effects across breast cancer subtypes. A broad range of additional correlative research will examine other predictive markers of metformin benefit.
Conduct of a large Phase III adjuvant trial at this juncture is not without risk. False negative results may be obtained because of failure to target specific patient or tumor characteristics (such as patients with high baseline insulin levels or triple negative or HER2 positive breast cancer subtypes), because the administration of metformin after completion of systemic chemotherapy (if given) and for 5 years only is suboptimal or because compliance is lower than expected in a population of healthy breast cancer survivors. Unexpected adverse effects may occur—given the widespread use of metformin to treat diabetes, it is unlikely that new toxicities will be identified but it is possible that adverse anti-cancer effects may be identified (for example, one group of investigators using metformin in concentrations several hundred times maximum clinical concentrations has reported stimulatory angiogenic effects) . An independent data safety monitoring Board at NCIC CTG will oversee trial conduct by monitoring adverse events in real time, and reviewing protocol-specified interim analysis results (for benefit and futility). Inclusion of planned interim futility analyses allows early termination of the study if it appears that metformin will not favorably impact breast cancer outcomes.
Balancing these risks, it must be recognized that there are also potential risks associated with not proceeding with a Phase III adjuvant trial at this time. Metformin is readily available off study (by prescription)—as evidence of potential benefits on breast cancer outcomes grows, it is possible that metformin will be prescribed to breast cancer patients off label in the hopes of improving their breast cancer outcomes, given its low cost and toxicity. Outside of a clinical trial, this practice could expose patients to harm that is greater than would occur in the closely monitored research setting, it would not advance knowledge about the impact of metformin on breast cancer outcomes and, if it became widespread, it could preclude the future conduct of properly designed and powered randomized trials.
Potential rewards of evaluation of metformin in the adjuvant setting are numerous, and the demonstration of an important anti-cancer effect of a well-tolerated agent that costs mere pennies per day would be a major clinical advance, even if it is ultimately shown to have differential effects in subgroups defined by insulin levels or tumor type. Similar study is anticipated in other tumor types, including prostate, colorectal, pancreatic, and endometrial cancers—these activities will face similar challenges. In addition to the scientific considerations discussed here, the pace of development of metformin has been impacted by unique funding challenges which have resulted in delays because, as a generic drug, metformin does not receive the pharmaceutical support usually afforded to promising anti-cancer agents. This has required novel approaches to funding, with partnerships of government, charitable organizations, a generic drug company as well as the commitment of international academic research groups. Evaluation of metformin in the prevention setting may face additional challenges if known beneficial effects on diabetes, cardiovascular health, and other health outcomes are seen before anti-cancer effects, leading to early closure of studies before effects on cancer outcomes can be assessed.
The approach advocated here is supported by available epidemiologic, clinical, and preclinical evidence; the ratio of potential risks and benefits appears to be favorable. It is an approach that is most likely to lead to timely information about potential metformin benefit in breast cancer. Although NCIC CTG MA.32 has been designed as a pragmatic trial, its embedded correlative research will also provide important opportunities to advance understanding of the effect of metformin on breast cancer biology. It is anticipated other studies will be needed to refine understanding of metformin effects and to examine effects of other agents that target insulin, IGF, mTOR, and related signaling pathways (both separately and in conjunction with metformin) if beneficial effects are identified in NCIC CTG MA.32. NCIC CTG MA.32 should be viewed as an important step in the evaluation of metformin in breast cancer, but not the final step.
- 12.Pollak MN, Chapman JW, Shepherd L, Meng D, Richardson P, Wilson C et al (2006) Insulin resistance, estimated by serum C-peptide level, is associated with reduced event-free survival for postmenopausal women in NCIC CTG MA.14 adjuvant breast cancer trial. J Clin Oncol 24(Suppl 18S):(Abstr 524)Google Scholar
- 14.Goodwin PJ, Pritchard KI, Ennis M, Koo J, Hood N (2009) Long-term effects of fasting insulin in early stage breast cancer patients. Eur J Cancer Suppl 7:308 (abstr P-5160)Google Scholar
- 28.Vazquez-Martin A, Oliveras-Ferraros C, del Barco S, Martin-Castillo B, Menendez JA (2009) The antidiabetic drug metformin: a pharmaceutical AMPK activator to overcome breast cancer resistance to HER2 inhibitors while decreasing risk of cardiomyopathy. Ann Oncol 20:592–595CrossRefPubMedGoogle Scholar