Re:Re: Metformin as an anti-cancer agent: actions and mechanisms targeting cancer stem cells

[문형철]

beyond reason

메트포민의 암줄기세포 항암효과 기전

Metformin as an anti-cancer agent: actions and mechanisms targeting cancer stem cells 

Nipun Saini,  Xiaohe Yang

Acta Biochimica et Biophysica Sinica, Volume 50, Issue 2, February 2018, Pages 133–143, https://doi.org/10.1093/abbs/gmx106

Published:

 

07 October 2017

 Article history

• PDF
•  Split View
• Cite
• Permissions Icon Permissions
•  
  Share

   

Abstract

Metformin, a first line medication for type II diabetes, initially entered the spotlight as a promising anti-cancer agent due to epidemiologic reports that found reduced cancer risk and improved clinical outcomes in diabetic patients taking metformin. To uncover the anti-cancer mechanisms of metformin, preclinical studies determined that metformin impairs cellular metabolism and suppresses oncogenic signaling pathways, including receptor tyrosine kinase, PI3K/Akt, and mTOR pathways. Recently, the anti-cancer potential of metformin has gained increasing interest due to its inhibitory effects on cancer stem cells (CSCs), which are associated with tumor metastasis, drug resistance, and relapse. Studies using various cancer models, including breast, pancreatic, prostate, and colon, have demonstrated the potency of metformin in attenuating CSCs through the targeting of specific pathways involved in cell differentiation, renewal, metastasis, and metabolism. In this review, we provide a comprehensive overview of the anti-cancer actions and mechanisms of metformin, including the regulation of CSCs and related pathways. We also discuss the potential anti-cancer applications of metformin as mono- or combination therapies.

metformin, cancer stem cells, AMPK/mTOR pathway, anti-cancer drugs, cellular metabolism

Topic:

 

• metformin
• antineoplastic agents
• cancer
• mtor serine-threonine kinases
• cancer stem cells

Issue Section:

 Review

Introduction: Metformin at a Glance

Metformin (1,1-dimethylbiguanide), a commonly prescribed anti-type II diabetes drug, belongs to the biguanide class of compounds, which also includes phenformin and buformin [1]. The glucose and the insulin lowering ability of metformin, along with reduced hepatic glucose output, are shown to lower blood glucose levels and improve several other diseases, including polycystic ovary syndrome and metabolic syndrome. In past decades, several epidemiologic studies have linked metformin use with a decreased risk of several types of cancers, including breast, prostate, pancreatic, and non-small cell lung (NSCLC) cancer. Numerous in vitro and in vivostudies, along with clinical trials, have further strengthened and supported the anti-cancer ability of metformin. In addition, the cost effectiveness of metformin, alongside its beneficial effects on weight loss and cardiovascular risk factors, including an improved lipid profile and reduced incidence of fatty liver, further adds to its superiority as a promising anti-cancer agent [2,3]. Importantly, metformin also has a well-established safety profile with the most common toxicity being mild-to-moderate gastrointestinal discomfort and metallic taste, which are diminished with continued metformin use [2]. Lactic acidosis, a potential side effect of other members of the biguanide family, is very rare in patients treated with metformin [2]. Together these economical and clinical benefits of metformin support its further development and potential clinical implementation as an anti-cancer therapy.

In this review, we provide a comprehensive overview of evidence supporting metformin as an anti-cancer agent and discuss the underlying mechanisms of metformin, including metformin-mediated regulation of cancer stem cells (CSCs). The search strategy used to retrieve previous studies involved search terms, such as ‘metformin and cancer’, ‘metformin and cancer stem cells’, ‘metformin and tumor stem cells’, ‘metformin and mammary stem cells’ and ‘metformin mechanism of action’, in PubMed and Google Scholar. Previous articles, specifically focusing on the in vitro and in vivo anti-cancer and anti-CSC mechanisms of action of metformin in different cancers, are included. Also, only relevant epidemiologic studies focusing on metformin and reduced cancer incidence are discussed. Information regarding clinical trials was retrieved from the ClinicalTrials.gov website (provided by the National Institutes of Health) using ‘cancer’ and ‘metformin’ in the search query.

Metformin and Cancer Prevention

Epidemiologic link

The association between metformin use and reduced cancer risk in patients with diabetes was suggested in a pioneering observational study published in 2005 which reported a 23% decrease in cancer risk with metformin use [4]. Since then, several epidemiologic studies have provided additional evidence linking lower cancer risk in diabetic patients treated with metformin than in non-metformin users [1]. In a cohort study of diabetic patients, survival analysis revealed reduced cancer risk (bowel, lung, and breast) in metformin users (n = 4085) versus non-metformin users (n = 4085) with a hazard ratio (HR) of 0.63 [5]. In another study of patients with diabetes and NSCLC, metformin use was associated with improved overall survival (OS) of 25.6 months as compared to 13.2 months in patients given other anti-diabetes treatments [6]. A population-based cohort study in Korea also reported a positive correlation between metformin use and reduced cancer-specific mortality and reduced occurrence of retreatment events in diabetic patients (n = 533 metformin users; n = 218 non-metformin users) with comorbid hepatocellular carcinoma (HCC) that were initially subject to hepatic resection [7]. Along with epidemiologic studies, several meta-analyses have also supported metformin use and reduced cancer risk in diabetic patients with cancer. As such, a significant association (31% reduction) between metformin use and cancer incidence (pancreatic and HCC) was reported in a meta-analysis of 11 studies consisting of 4042 patients with cancer and diabetes [8]. Moreover, the reduced incidence of liver, pancreatic, colorectal (CRC), and breast cancers in metformin users was reported to be 78%, 46%, 23%, and 6%, respectively, in a meta-analysis of 37 studies comprising 1,535,636 patients [9]. Another meta-analysis of 11 studies in breast cancer patients with diabetes (n = 2760 metformin users; n = 2704 non-metformin users) revealed a 65% improved OS and cancer-specific survival in metformin users as compared to non-users [10]. Similar results of improved OS and cancer-specific survival were reported with metformin use in a meta-analysis of eight studies with a total 254,329 kidney cancer patients with diabetes [11]. Metformin use is also associated with increased survival (HR = 0.59) and clinical beneficial effect (HR = 0.64) in diabetic liver cancer patients [12] and reduced cancer risk (n = 39,787 metformin users; n = 177,752 non-metformin users) in lung cancer patients [13].

Though most studies have supported the reduced cancer incidence in metformin users as compared to non-users, some recent retrospective cohort studies in diabetic patients with breast [14], renal [15], prostate [16], and endometrial [17] cancers indicated no clear association between metformin use and improved OS or disease-free survival, as reviewed by Coperchini et al. [18]. Certain limitations associated with these studies include a small sample size of enrolling patients or restriction to a single healthcare system or ethnic group. Second, the follow-up time was also shorter for these studies, along with missing data on patient characteristics such as obesity, diet, and physical activity. Some reports also did not have a clear indication of the number of patients actually taking metformin among the included patients that were prescribed metformin. In addition, time-related biases, such as immortal time, time-window, and time-lag biases, have also been reported as factors leading to the overestimation of the protective effects of metformin [19]. Together, these factors suggest that the effect of metformin could be tumor site- or tumor type-specific, thus leading to the inconsistencies observed in clinical studies. However, taking into account the available studies favoring metformin use and the studies reporting inconsistent clinical outcomes, the vast majority of the data supports the potential of metformin in decreasing the risk of multiple cancers.

Preclinical studies

To understand the potential anti-cancer mechanisms of metformin, a multitude of studies using cell and animal models of human cancer have reported cellular and systemic effects. Importantly, metformin inhibits the growth of tumor cells by targeting numerous pathways involved in cell proliferation in vitro. A range of metformin concentrations (2–50 mM) has been tested in various cancer cells to depict its anti-cancer efficacy [20]. Metformin inhibits cell proliferation by inducing cell cycle arrest in G0/G1 phase in various cell line models of breast [21,22], renal [23], pancreatic [24], and prostate [25] cancers. A few studies have even demonstrated that metformin can induce both G0/G1 and G2/M arrest to inhibit cell growth, particularly in endometrial cancer cells [26]. Cell cycle arrest was also found to be concomitant with decreases in key cell cycle regulators, such as cyclin D1, Cdk4, and phosphorylation of retinoblastoma (Rb) protein, as well as the induction of apoptosis in metformin-treated cells.

Several cancer models, such as xenografts of primary cell lines, orthotopic tumors, carcinogen-induced tumors, and transgenic animals with spontaneous tumors, have been used to eval‎uate the in vivo effects of metformin on tumor prevention, development, and growth. In established pancreatic cancer xenograft models, metformin (50–250 mg/kg/day) dose-dependently inhibited tumor growth when given via intraperitoneal (i.p.) injections. Tumor volume was reduced by 80% and 67% when metformin was administered via i.p. injection (200 mg/kg/day) and in the drinking water (2.5 mg/ml/day), respectively [27]. Notably, another report found that low-dose metformin (human equivalent dose = 20 mg/kg) administered in the drinking water for 18 or 24 days also resulted in significant growth inhibition of pancreatic cancer xenografts [28]. Along with reductions in tumor growth and volume, metformin effectively targets tumor angiogenesis and metastasis in different cancer models. Metformin (200 mg/kg/day) significantly suppressed Her2-induced tumor angiogenesis via targeting Her2/HIF-1α/VEGF secretion axis in a breast cancer xenograft model [29]. Likewise, in an ovarian cancer xenograft model, metformin (100–200 mg/kg/day) significantly inhibited pulmonary metastasis and angiogenesis as compared to untreated control mice, which exhibited visible liver, spleen and kidney tumors [30]. Combination studies of metformin with other chemotherapeutic drugs, such as gefitinib (1 mg/ml/day metformin + 250 mg/l/day gefitinib in drinking water for 4 weeks) [31] and cisplatin (40 mg/kg metformin + 5 mg/kg cisplatin daily via i.p. injection for 18 days) [32], have also demonstrated significant reductions in tumor burden and prolonged survival in mice with combination treatments versus either treatment alone in lung cancer xenograft models.

Orthotopic models of cancer, which simulate organ-specific microenvironments, have also shown that metformin significantly reduces tumor growth, tumor volume, and metastasis, specifically in pancreatic cancer [27], Her2+/ErbB2+ and triple-negative breast cancer models [33]. Similarly, the combined treatments of metformin, given orally or via tail vein injections, with gemcitabine [34,35] or sorafenib [36] have shown significant suppression of tumor growth and postoperative tumor recurrence and metastasis as compared to vehicle or either treatment alone in pancreatic and HCC orthotopic models, respectively. These studies further emphasize the potential therapeutic applications of metformin in regard to tumor recurrence and metastasis.

The impact of metformin treatment on the prevention of tumorigenesis has also been investigated. In a Her2/neu transgenic murine model of breast cancer, long-term metformin treatment (100 mg/kg/day from 8 weeks of age to 52 weeks of age) demonstrated increased survival and life expectancy along with increased tumor latency as compared to control mice [37]. Additionally, in a carcinogen-induced model of bladder cancer, metformin (2 g/l in drinking water for 14 weeks) blocked the progression of N-methyl-N-nitrosourea (MNU)-induced precancerous lesions to carcinoma in situ (CIS) or invasive tumors as compared to the untreated MNU group [38]. Similarly, metformin (50 mg/kg/day in drinking water for 18 weeks) increased tumor latency, but not tumor incidence, in an MNU-induced mammary tumor model in rats. Also, in a diethylnitrosamine-induced liver tumorigenesis model, metformin (250 mg/kg/day in the chow diet for 36 weeks) significantly reduced tumor multiplicity and size along with an almost 80% reduction in the number of visible liver surface tumors as compared to the control mice [39].

Taken together, preclinical studies have implicated the anti-cancer efficacy of metformin at a range of doses administered via various routes in several cancer models. Though the doses used in these studies are often higher than what is typically used in the clinics, the potential of metformin in preventing tumorigenesis and inhibiting tumor growth is recognized in vivo. Additionally, studies have demonstrated that the efficacy of metformin is affected by the change in the expression‎ levels of membrane transporters (OCT1-4, PMAT, and MATE1-2) involved in the uptake and secretion of metformin [3]. For instance, the bioavailability, tissue distribution, and clearance of metformin, along with its ability to phosphorylate AMP-activated protein kinase (AMPK), are reduced significantly in the adipose tissue of OCT3-knockout mice as compared to wild-type controls [40]. Similarly, in OCT3-overexpressing breast cancer cell line and xenograft models, metformin treatment increased AMPK activation, reduced pS6K phosphorylation and enhanced anti-tumor activities as compared to the wild-type cells and tumors that expressed low endogenous levels of OCT3 [41]. In epithelial ovarian cancer cells, siRNA knockdown of OCT1 attenuated the efficiency of metformin to activate the AMPK pathway and inhibited the anti-proliferative capacity of metformin in vitro[42]. Also, in a rat model of high fat diet-induced overweight and carcinogen-induced mammary tumorigenesis, the reduction in tumor volume associated with metformin treatment was positively correlated with the intratumoral accumulation of metformin and increased OCT2 protein expression‎, suggesting a link between the cellular uptake of metformin by transport proteins and the anti-cancer efficacy of metformin [43]. Thus, concerns regarding the usage of superphysiological concentrations of metformin in preclinical studies could be somewhat resolved by altering the expression‎ of membrane transport proteins through the use of drugs, such as antibiotics and proton pump inhibitors [44], in combination with metformin to increase cellular uptake and accumulation in tumor cells. Future studies to better understand the role of membrane transport proteins in enhancing metformin’s potency as an anti-cancer agent are imperative.

Clinical studies

Numerous clinical trials are underway to eval‎uate metformin as a monotherapy or a combination therapy in breast, pancreatic, endometrial, lung, and prostate cancers. Therapeutic strategies being tested include metformin in combination with other chemo-drugs and/or radiation therapy. The chemotherapeutic drugs being eval‎uated for enhanced anti-cancer effects in combination with metformin include: cyclophosphamide, doxorubicin, docetaxel, epirubicin, everolimus, exemestane, trastuzumab, atorvastatin, letrozole, megestrol acetate, carboplatin, and fluorouracil (5-FU). The primary objective of these trials is to determine the maximum tolerable dose, progression-free survival (PFS), overall response rate (ORR), and recurrence-free survival (RFS) in metformin-treated patients. A completed Phase II trial of metformin and medroxyprogesterone acetate combination treatment in atypical endometrial hyperplasia and endometrial cancer reported complete and partial response rates of 81% and 14%, respectively, and an RFS rate of 89% with no severe toxicities [45]. Moreover, metformin in combination with 5-FU demonstrated ‘overall modest activity’ in metastatic CRC patients in a Phase II trial, [46], while metformin as a chemopreventive monotherapy reduced metachronous colorectal adenomas or polyps in a Phase III trial [47]. Current clinical trials are also investigating secondary outcomes, such as proliferation markers (Ki67) and pathway biomarkers (phosphorylation status of pS6K, 4EBP-1, AMPK, Akt, and Erk). However, results are not yet available for most of these studies. Details of inactive and active clinical trials testing the safety and efficacy of metformin in different cancers can be viewed at: https://clinicaltrials.gov/ct2/results?term=+cancer+AND+metformin. Several concerns need attention regarding these clinical trials. First, most of the clinical studies target patients with diabetes and insulin resistance, which may modulate the anti-cancer benefits of metformin. Therefore, more clinical studies targeting non-diabetic cancer patients are needed. Second, the efficacy of metformin as a cancer preventive and/or therapeutic agent still needs investigation. Finally, the endpoint goals of future clinical trials need to shift toward long-term, RFS with minimal side effects in monotherapy or adjuvant applications in order to better understand the potential of metformin in clinical settings.

Anti-cancer mechanisms of metformin at the molecular level

At the molecular level, the major effects of metformin are predominantly exerted through the inhibition of oxidative phosphorylation in mitochondria and activation of AMPK (Fig. 1) [48,49]. The inhibition of mitochondrial complex I by metformin treatment induces metabolic stress, which increases endogenous levels of reactive oxygen species (ROS). In turn, oxidative stress mediates the death of cancer cells that rely on oxidative phosphorylation for energy production [50–52]. Metformin-induced inhibition of mitochondrial complex I is also accompanied by an increase in glycolysis to compensate for reduced ATP production. To maintain cellular homeostasis in response to metformin-induced changes in AMP/ATP ratio, AMPK is activated by the phosphorylation of LKB1, a tumor suppressor, at Thr172, and anabolic and catabolic pathways are subsequently inhibited and activated, respectively [53]. In particular, AMPK activation inhibits the mTOR pathway via the phosphorylation of TSC1/2, tumor suppressors that negatively regulate mTOR. Metformin-mediated activation of AMPK also leads to activation of p53, a tumor suppressor that promotes apoptosis, autophagy and inhibition of the Akt and mTOR pathways [49,53]. In addition, AMPK activation can inhibit receptor tyrosine kinase pathways, including EGFR and ErbB2 signaling, which further target the downstream effectors Akt, mTOR, and Erk [54]. Metformin also inhibits the mTOR pathway in an AMPK-independent manner by inactivating Rag GTPases [55] or by upregulating the expression‎ of REDD1 (regulated in development and DNA damage responses 1), a negative regulator of mTOR [56]. mTOR inhibition further suppresses downstream targets, including 4EBPs, pS6Ks, and initiation factor eIF4G [20,53]. mTOR is also a critical mediator of the PI3K signaling pathway, which is involved in cellular growth and survival [53]. Thus, metformin restricts cancer cell proliferation by inhibiting protein translation via PI3K/Akt/mTOR pathways. 

Figure 1.

Open in new tabDownload slide

Molecular mechanisms associated with classical anti-cancer and anti-CSC effects of metformin Classical anti-cancer and anti-CSC pathways activated by metformin are indicated by solid line arrows and those pathways inhibited by metformin are shown in dotted lines. Abbreviations: ACC (acetyl-coA carboxylase); Akt (protein kinase B); AMP (adenosine monophosphate); ATP (adenosine triphosphate); AMPK (AMP-activated protein kinase); EMT (epithelial to mesenchymal transition); FASN (fatty acid synthase); IGF1R (insulin growth factor-1 receptor); IL (interleukin); IRS1 (insulin receptor substrate-1); LKB1 (liver kinase B1); mTOR (mammalian target of rapamycin); NF-κB (nuclear factor kappa-light-chain-enhancer of activated B-cells); NTPs (nucleotide triphosphates); PI3K (phosphatidylinositol-4,5-bisphosphate 3-kinase); REDD1 (regulated in development and DNA damage response 1); ROS (reactive oxygen species); Shh (sonic hedgehog); TAM (tumor-associated macrophage); TGFβ (transforming growth factor beta); TNFα (tumor necrosis factor alpha); TSC1/2 (tuberous sclerosis 1 and 2).

AMPK activation by metformin also leads to the inactivation of insulin receptor substrate-1 (IRS1). IRS1 is an activator of IGF1R and PI3K/Akt signaling pathways. In turn, the suppression of IRS1 activity inhibits the IGF1/insulin signaling axis and subsequently PI3K/Akt/mTOR signaling [1,57]. Via the reduction of circulating insulin levels and targeting of the insulin/IGF1/PI3K signaling axis, metformin inhibits hyperinsulinemia-associated neoplastic activity [58]. Metformin-induced AMPK activation also inhibits acetyl-CoA carboxylase (ACC) and fatty acid synthase (FASN) activation, thereby preventing lipogenesis, a process required by tumor cells to accommodate increasing demands of continuous cellular growth, and subsequent cellular proliferation [2,48]. Increased cell proliferation also results from the induction and infiltration of pro-inflammatory cytokines. Metformin elicits anti-inflammatory and anti-angiogenic effects by decreasing the production of inflammatory cytokines, including tumor necrosis factor alpha (TNFα), interleukin-6 (IL-6), and IL-1β, and inhibiting nuclear factor kappa-light-chain-enhancer of activated B-cells (NF-κB) and hypoxia-inducible factor-1-alpha (HIF-1α), which in turn diminishes the production of vascular endothelial growth factor (VEGF) [48,59]. As such, metformin also inhibits TNFα-induced CXCL8 secretion, which is a downstream mediator of NF-κB signaling and is associated with tumor progression, in primary human normal thyroid cells and differentiated thyroid cancer cells [60].

Overall, the anti-cancer effects of metformin as a mono- or combination therapy in various cancers are innumerable. Epidemiologic, preclinical and clinical studies support the anti-neoplastic activity of metformin, further emphasizing its potential as a therapeutic agent. Although some studies report inconsistent or conflicting data, which warrant further investigation, the promising anti-cancer effects of metformin in preclinical settings cannot be negated.

CSC Theory and Characterization

According to the CSC theory, CSCs/tumor-initiating cells (TICs) are a population of cells that are capable of triggering tumorigenesis. CSCs possess stem cell properties, including self-renewal, proliferation, and differentiation potential, which give rise to heterogeneous populations consisting of both CSCs and non-stem cancer cells (NSCCs). NSCCs have limited proliferation and survival potential; therefore, self-renewal, clonal tumor initiation, and expansion into heterogeneous populations are important features specific to CSCs [61–63].

The earliest reports of CSCs in solid tumors came from the pioneering work of Al-Hajj et al. (2003), which identified a distinct tumor cell population derived from breast carcinomas that was capable of inducing tumors in NOD/SCID mice [64]. Very low cell numbers (as low as 100–200 cells) were needed to form tumors in the inoculated mice. This particular cell population, as characterized by CD44+, CD24low/−, and ESA+ (CD44+CD24−ESA+) expression‎, produced tumors with phenotypic heterogeneity comparable to the parent tumor and could be passaged serially. In comparison, other tumor cell populations (CD44+CD24+) were not tumorigenic despite xenograft inoculations with up to 2000 cells in vivo [64]. In the years following the identification of the tumorigenic CD44+CD24− population, an ALDH1+cell subpopulation was isolated from breast carcinomas using an ALDEFLUOR assay and also demonstrated the ability to stimulate xenograft tumor formation with inoculations of as low as 500 ALDH1+ cells in vivo [65]. To note, ALDH1− cells were not tumorigenic. To date, several additional markers have been identified and are routinely used to differentiate CSCs from NSCCs in different cancers as detailed in Table 1. In addition, pluripotent embryonic stem cell markers, like c-Myc, Nanog, Sox2, Klf-4, OCT4, and Lin28, have also been used to differentiate CSCs from NSCCs [66]. In vitro, CSCs are characterized by their ability to form microtumors/mammospheres under non-adherent and non-differentiating conditions with continual passages. In vivo, CSCs are a subset of cells capable of self-renewal and inducing new tumors when inoculated (at low cell numbers) into immunodeficient animal models. Additionally, a strong correlation between CSCs and tumor aggressiveness, metastasis, histological grade, and poor OS in different cancers further highlights the critical roles of CSCs in cancer initiation and progression [65,67–69]. Thus, their established association in tumor resistance and relapse makes CSCs important candidates for novel targeted-therapeutic approaches. 

Table 1.

Potential CSC markers in different cancer types

Type of cancer	CSC markers
Breast	CD44highCD24low/−, CD61highCD49fhigh, ALDH1+
Pancreatic	CD133+, EpCAM+, ALDH1+
Ovarian	CD133+, CD44+CD117+, ALDH1+
Lung	CD166, ALDH+, CD90
Prostate	CD44highCD24low/−, ALDH+, CD133+
Hepatocellular carcinoma	CD90/Thy-1 and EpCAM+AFP+
Melanoma	CD166/ALCAM

AFP, α-fetoprotein; ALCAM, activated leukocyte cell adhesion molecule.

Metformin and CSCs

Metformin as a monotherapy to target CSCs

Recent studies demonstrate that metformin-mediated anti-cancer activities involve specific targeting of CSCs/TICs. Metformin significantly inhibits the sphere-forming ability of CD44+CD24−, CD61highCD49fhigh, CD133+, ALDH1+, EpCAM+, CD133+CD44+, and CD44+CD117+ subpopulations in breast, pancreatic, glioblastoma, CRC, and ovarian cancer models [70–72]. The CD61highCD49fhigh population, which is enriched with CSC/TIC precursors in premalignant mammary tissues of MMTV-ErbB2 transgenic mice, and the ALDH1+ population, which is detected in ErbB2-overexpressing breast cancer cell lines and xenograft models, are significantly inhibited by metformin treatment via targeted inactivation of EGFR/ErbB2 signaling [70]. In pancreatic [24,73], colorectal [74], and glioblastoma [75,76] cancer cell and xenograft models, metformin-induced inhibition of the CSC subpopulation is associated with the downregulation of Akt/mTOR pathways, decrease in FASN levels and increase in the expression‎ of phosphatase and tensin homolog (PTEN), a tumor suppressor. Notably, in ovarian cancer cell and patient-derived tumor xenograft models, low doses of metformin (0.1 and 0.3 mM in vitro and 20 mg/kg/day in vivo) are associated with significant inhibition of the CD44+CD117+ subpopulation without affecting ALDH+ cells [71]. Higher doses of metformin (1 mM and 150 mg/kg/day, respectively) were needed to reduce the ALDH+ population in SKOV3 and A2780 cells in vitro and SKOV3 xenografts in vivo [77]. In addition to monotherapy strategies, several studies have also demonstrated the potency of metformin in targeting CSCs in combination with chemo- and radiation therapies, as detailed below.

Metformin as a combination therapy to target CSCs

The ability of metformin to target chemo- and radiation-resistant CSCs in combination with other drugs is demonstrated in various cancer cell and xenograft models. Chemotherapy and radiation therapy are conventional approaches used for the treatment of cancer [78]; however, resistance to these strategies still poses a major challenge. Metformin sensitizes cancer cells to radiotherapy by activation of AMPK and DNA repair pathways [79]. Metformin sensitizes esophageal cancer cells to irradiation and induces cell cycle arrest and apoptosis by targeting the ataxia-telangiectasia mutated (ATM) and AMPK/mTOR/HIF-1α pathways [80]. Importantly, the combination of metformin (1 mM) and radiation (3, 5, or 7 Gy) significantly attenuates radiation-induced increases in ALDH1+ and CD44+CD24− CSC populations in FSaII and MCF7 cells, respectively. Metformin (25 mg/kg) + radiation (20 Gy) also significantly reduces FSaII xenograft tumor size and prolongs tumor latency, which corresponded with metformin-induced AMPK activation and mTOR suppression, as compared to either treatment alone in C3H mice [81]. Similarly, combinations of metformin (30–100 μM) and radiation (2–8 Gy) significantly attenuated clonogenic and tumorsphere CSC survival in Panc1 and MiaPaCa-2 pancreatic cells as compared to either treatment alone [82]. Metformin in combination with radiation markedly induced G2/M arrest and DNA damage in MiaPaCa-2 cells as well. These responses were also AMPK-dependent. Although studies using metformin to target radiation-resistant CSCs are limited, these reports provide supportive evidence that indicates the potential of metformin to sensitize CSCs to ionizing radiation.

Similarly, metformin in combination with chemotherapeutic drugs have shown significant reductions in CSCs and prolonged tumor remission. In trastuzumab-resistant breast cancer cell and xenograft models, the combination treatment of metformin and trastuzumab significantly inhibited the CD44+CD24− CSC subpopulation along with significant reductions in tumor volumes, thereby demonstrating the translational potential of this combination treatment strategy [83–85]. Metformin in combination with doxorubicin, paclitaxel, or carboplatin demonstrated similar results with nearly complete tumor elimination alongside prolonged remission in breast xenograft tumor models [86]. Metformin and doxorubicin combination treatments also suppressed tumorigenesis in prostate and lung xenograft tumor models. Notably, metformin reduced the dose of doxorubicin that was needed to inhibit tumor growth by 4 folds [86]. Doxorubicin or cisplatin combined with metformin has also shown efficacy in eradicating OCT4+ CSCs in doxorubicin-resistant thyroid cancer models [87], and ALDH+ and CD44+CD117+ CSCs in doxorubicin-resistant ovarian cancer models [71,77]. Metformin + 5-FU also significantly reduced CD133+ CSCs in CRC cells in vitro [72] and esophageal xenograft tumor growth [88], as compared to 5-FU treatment alone. Moreover, significant reductions in glioblastoma stem cell proliferation and tumor growth, and prolonged OS of tumor-bearing mice were demonstrated upon treatments with metformin + temozolomide, as compared to either treatment alone [89,90]. Furthermore, the combination of metformin with chemotherapy and irradiation (30 μM metformin + 0.2 μM gemcitabine + 8 Gy irradiation) enhanced the reduction in clonogenic survival in MiaPaCa-2 cells [82].

To further enhance the targeted delivery of metformin alone or in combination with other chemotherapeutic drugs, strategies involving the encapsulation of metformin in liposomes and nanoparticles have been explored. In murine sarcoma S180 cell and xenograft models, treatment with coencapsulated epirubicin and metformin in polyethylene glycolated (PEGylated) liposomes selectively increased cytotoxicity in CD133+ cells, which include a subpopulation of cancer stem-like cells. The coencapsulated combination treatments also induced complete tumor elimination and increased survival by 58.5 days in vivo as compared to the control groups or either encapsulated drug alone [91]. Similarly, metformin-loaded BSA nanoparticles amplified ROS production and increased the inhibition of cell proliferation in MiaPaCa-2 pancreatic cancer cells as compared to metformin treatment without the nanoparticle carrier [92]. Metformin-loaded alginate nanocapsules also reduced the dosage needed to maintain blood glucose levels in diabetic rats [93]. Overall, the majority of reports demonstrate that combination therapies with metformin produce nearly total eradication of CSCs and further reduce the effective dosages of chemotherapeutic drugs, which will in turn help to minimize potential related toxicities. Furthermore, drug delivery systems involving encapsulation and/or nanoparticles of metformin in combination with other therapeutics can potentially further reduce metformin and chemotherapeutic drug dosages needed for anti-cancer responses, as well as enhance targeted drug delivery to the cancer cells.

Anti-CSC Mechanisms

Inhibition of self-renewal and metastatic pathways

Pathways involved in development, self-renewal, progression, and metastasis are often deregulated in cancer [94]. Metformin is reported to effectively inhibit pathways associated with self-renewal and metastasis in various cancers, including the hedgehog (Hh), Wnt, and transforming growth factor beta (TGFβ) pathways. The anti-CSC mechanisms of metformin are illustrated in Fig. 1.

Sonic hedgehog signaling

In pancreatic cancer, overexpression‎ of sonic hedgehog (Shh), a ligand of Hh signaling, activates the Hh pathway, which is associated with stem cell populations, epithelial-mesenchymal transition (EMT) and promotes neo-vascularization during tumorigenesis [95]. Metformin (1 mM) inhibits Shh protein and mRNA levels in BxPC3 human pancreatic cancer cells, although the mechanism is not fully elucidated [95]. In multiple breast cancer cell lines, metformin treatment (3 mM) downregulates the gene and protein expression‎ of Shh, Smo, Ptch1, and Gli1, components of Shh signaling pathway, as compared to untreated controls [96]. Moreover, in recombinant human Shh (rhShh)-activated MDA-MB-231 human breast cancer cell and xenograft models, metformin effectively inhibited cell proliferation, migration, invasion, and tumor growth in an AMPK-dependent manner. Importantly, metformin significantly decreased the rhShh-induced CD44+CD24− mammary CSC population [96].

Wnt/β-catenin signaling

Wnt signaling is another important pathway involved in self-renewal and metastasis targeted by metformin. Metformin has been reported to inhibit the activation of Wnt/β-catenin signaling in cervical and breast cancer cells by targeting DVL3, a positive regulator in Wnt/β-catenin signaling [97,98]. It has also been reported to increase the expression‎ of Bambi, a TGFβ decoy receptor, and induce pro-survival Wnt/β-catenin signaling in hepatic stellate cells [99]. In combination with FuOx, a drug combination composed of 5-FU and oxaliplatin, metformin effectively inhibited proliferation, migration, stemness/colonosphere formation, and tumor growth in chemo-resistant colon cancer cell and xenograft models via downregulation of β-catenin and c-Myc expression‎ [100]. The role of metformin in the inhibition of Wnt-induced CSCs has not been fully investigated to date; however, a recent study using embryonic stem cell and zebrafish models of neural development reported that metformin can impede EMT, which is required for neural crest formation, via the disruption of Wnt signaling and microRNA expression‎ [101].

TGFβ signaling

TGFβ is often labeled a ‘double-edged sword’ in regard to its tumor suppressor actions, as well as its tumor-promoting properties that involve processes such as cell proliferation, invasion and metastasis [102,103]. Recently, it was demonstrated that TGFβ-treated human mammary epithelial cells undergo EMT and acquire stem cell properties, including high mammosphere formation efficiency (MSFE) and a CD44+CD24− antigen phenotype [104]. Studies have also shown the expression‎ of TGFβ1 and TGFβRII specifically in CD44+CD24− cells isolated from human breast cancer tissues and subsequent EMT reversal upon TGFβRI/II inhibitor administration, further supporting a link between TGFβ signaling, EMT and CSCs [105]. In particular, metformin reduces the CD44+CD24− population and reverses EMT in MDA-MB-231 breast cancer cells by inhibiting the mRNA levels of EMT-specific markers, including ZEB1, TWIST1, and SNAI2 transcription factors and TGFβ1-3 cytokines [106]. Metformin also reversed EMT (upregulated E-cadherin and downregulated vimentin protein expression‎) and reduced cell migration in TGFβ-stimulated human NSCLC cells, as compared to untreated TGFβ-stimulated cells[107]. Importantly, a recent study using a surface plasmon resonance-based assay reported that metformin directly binds to TGFβ1 to prevent its heterodimerization with TGFβRII and subsequent downstream signaling [108]. Moreover, metformin is unable to attenuate TGFβ signaling in TGFβRI-deficient MCF7 cells, which provides further evidence of the TGFβ-mediated effects of metformin [109].

Inhibition of inflammatory pathways

NF-κB promotes tumorigenesis by activating an inflammatory response mediated by pro-inflammatory cytokines, such as TNFα, IL-1, IL-6, and IL-8, and promoting cell proliferation, anti-apoptotic genes, EMT and metastasis [110]. NF-κB is also involved in the phenotypic change of MCF10A cells, upon ER-Src activation, into transformed cells that exhibit colony-forming ability, CD44 expression‎/CSC phenotype and mammary tumor formation in xenograft models [111]. Metformin (0.1 mM) significantly inhibits the MSFE of ER-Src MCF10A transformed cells and human breast cancer cells in vitro and prevents ER-Src MCF10A-derived tumor growth in vivo [112]. Additional work by Hirsch et al. [113] demonstrated that metformin delays the malignant transformation of ER-Src-activated MCF10A cells. Metformin also inhibits Lin28B and VEGF mRNA expression‎ and NF-κB nuclear localization in CSCs, as compared to NSCCs, isolated from transformed cells [113]. Notably, metformin not only decreases CSC populations in vitro and in xenograft models of transformed cells, but also displays enhanced tumor growth inhibition in inflammation-associated xenograft models of human liver, prostate, and skin (melanoma) cancers [113].

Recently, metformin was shown to suppress the M2 phenotype of human THP-1 macrophages that were cultured in conditioned medium from metformin-treated breast cancer cells, indicating that metformin can alter the profile of cytokines secreted by cancer cells [114]. In particular, metformin promoted the M1 phenotype by activating AMPK/NF-κB signaling in the treated breast cancer cells. Metformin also similarly induced the polarization of tumor-associated macrophages (TAMs) to the M1 phenotype in vivo. Indeed, NF-κB is involved in the polarization of TAMs from the classically activated M1 phenotype, which promotes pro-inflammatory activity and tumor lysis, to the alternatively activated M2 phenotype that promotes tumor growth [115,116]. Also, the interleukins secreted by TAMs promote CSC-like properties via the induction of EMT in HCC cells [117,118]. Thus, a strong association between NF-κB, TAMs, and CSCs has been suggested in multiple reports. The ability of metformin to convert TAMs to the M1 phenotype further indicates an indirect anti-cancer mechanism of metformin. However, further investigation is required to fully understand the link between inflammation-induced cancers, NF-κB, TAMs, CSCs, and metformin.

Inhibition of metabolic pathways

Although the role of metformin in different metabolic pathways was introduced earlier in this review, the effects of metformin on cellular metabolism as they relate to CSC regulation will be discussed in this section. In order to investigate the metabolic effect of metformin on neoplastic transformation and CSCs, Janzer et al.[119] utilized the ER-Src-inducible MCF10A system. In ER-Src-activated cells, metformin or phenformin significantly increased glycerol 3-phosphate levels, while also decreasing glycolytic intermediates and de novo lipogenesis. TCA cycle intermediates were also decreased after metformin treatments with a concurrent increase in glutamine uptake and ammonium production. This suggests that metformin increases glutamine utilization to feed TCA cycle intermediates via anaplerosis. Interestingly, metformin (300 μM) demonstrated marginal changes in glycolytic intermediates in CSC-enriched mammospheres from CAMA-1 transformed breast cancer cells as compared to parental CAMA-1 cells [119]. However, metformin significantly decreased nucleotide triphosphate levels with a concomitant increase in monophosphate levels and no change in diphosphate levels in the CAMA-1 CSC-enriched cells. These effects were specific to the CSCs since metformin did not induce an observable trend in the parental CAMA-1 cells [119]. Furthermore, metformin also induced the accumulation of folate and homocysteine in both CSCs and parental CAMA-1 cells, indicating abnormalities in nucleotide synthesis associated with defects in the tetrahydrofolate pathway [119]. Thus, CSCs and other transformed NSCCs appear to exert different metabolic responses to metformin treatment, suggesting complicated tumor metabolism.

An important metabolic effect of metformin on cancer cells is the inhibition of mitochondrial complex I leading to an aberrant increase in the flow of electrons towards oxygen and generation of ROS (e.g. superoxide) [120]. In NSCLC [121], ovarian [120,122] and breast [123,124] cancer cells, metformin treatment significantly increased ROS levels and reduced mitochondrial membrane potential, leading to cell death via DNA damage-induced apoptosis. However, the pretreatment of ovarian cancer cells with ROS scavengers, such as N-acetyl-L-cysteine, did not reverse the cell death effects of metformin [120], suggesting that ROS-induced cell death is not the only mechanism of metformin action. Specifically, in CD133+ cells derived from pancreatic tumors, metformin treatment creates an energy crisis in stem-like cells, resulting in significant AMPK-independent ROS production and reduced membrane potential. These metformin-induced cellular responses ultimately led to CSC-specific cell death via apoptosis [24]. In a follow-up study, the authors showed that metformin-induced cell death via ROS generation may not be a major mechanism of metformin since metformin-treated animals exhibited patient-derived xenograft tumor relapse and developed metformin-resistant CSCs. Furthermore, animals treated with menadione, a ROS inducer whose mechanism of action to induce cell death relies on the inhibition of mitochondrial complex I and the generation of ROS, did not develop resistant CSCs [125]. Similar increased ROS production and lipid peroxidation leading to apoptotic cell death were reported in metformin-treated or sorafenib + metformin-treated glioblastoma stem-like cells [126]. In contrast, metformin pretreatment in AMPKα+/+ and AMPKα−/− mouse embryonic fibroblasts AMPK-independently attenuated paraquat-induced ROS production, but not H₂O₂-induced ROS, suggesting effects of metformin particularly on endogenous ROS levels [127]. These studies indicate an indirect anti-cancer mechanism of metformin that acts via ROS production with a potential role in cell death. Yet, the major mechanism of metformin remains the AMPK-dependent pathway to induce cell death, even when ROS production is not increased by the inhibition of mitochondrial complex I. Nevertheless, further eval‎uation of the metabolic effects of metformin on CSCs is required to better understand its complex inhibitory mechanisms.

Regulation of microRNA-mediated pathways

Metformin has been reported to target various microRNAs (miRNAs), proteins associated in the miRNA biogenesis pathway and target genes in CSCs and NSCCs. As such, metformin inhibits the proliferative capability of breast cancer cells by downregulating miR-27a [128] and upregulating miR-193 (miR-193a-3p and miR-193b) [129], which in turn increased AMPKα and decreased FASN levels, respectively. Notably, miR-193b inhibition blocks the ability of metformin to decrease FASN expression‎ and inhibit the MSFE of CD44+CD24− and ALDH+BT549 mammospheres [129]. In MCF7 human breast cancer cells, metformin also upregulates let-7a (a tumor suppressor miRNA) expression‎ and downregulates TGFβ-induced miR-181a (an oncogenic miRNA [oncomiR]) expression‎, which results in decreased MSFE in vitro [130]. In renal [23] and breast cancer cells [131], the anti-cancer effects of metformin have been reported to be associated with the upregulation of miR-34a, which suppresses cell proliferation and the Sirt1/Pgc1α/Nrf2 pathway, respectively. Notably, the combined treatment of metformin and FuOx is associated with marked reduction of miR-21 (an oncomiR) and induction of miR-145 (a tumor suppressor miRNA), which were consistent with the suppression of β-catenin and c-Myc expression‎, cell growth and colonosphere formation in chemo-resistant colon cancer cells [100]. In pancreatospheres derived from gemcitabine-sensitive and -resistant pancreatic cancer cells, metformin was found to upregulate let-7a, let-7b, miR-26a, miR-101, miR-200b, and miR-200c, which are typically suppressed in pancreatic cancer [132]. Importantly, the re-expression‎ of miR-26a is associated with a decrease in pancreatosphere formation and reduced mRNA levels of CSC markers, including EZH2, OCT4, Notch1, and EpCAM [132]. Let-7b re-expression‎ similarly blocks pancreatosphere formation as well, indicating that miR-26a and let-7b may be involved in metformin-mediated regulation of pancreatic CSCs. Metformin also activates the stress-induced senescence (SIS) response in human diploid fibroblasts and upregulates the expression‎ of miR-141, miR-200a, miR-205, and miR-429, which are miRNAs that promote the inhibition/reversal of EMT [133]. Additionally, the proliferation and colony-forming ability of SIS-resistant induced pluripotent stem cells (iPSCs) is significantly reduced after metformin treatment, suggesting metformin’s ability to also bypass SIS resistance [133]. Together, these studies present the regulatory capacity of metformin that is involved with miRNA-associated growth, self-renewal, migration, and differentiation of CSCs.

Overall, metformin, alone or in combination with other cancer therapies, effectively targets CSCs derived from various cancer cell and xenograft models. Promising results from recent reports demonstrate metformin’s ability to selectively target CSCs through the inhibition of various signaling pathways and/or regulatory molecules that inhibit the self-renewal, proliferation and metastatic ability of CSCs in vitro and in vivo. However, with the growing incidence of cancer resistance and relapse, more clinical studies testing the anti-cancer potential of metformin in humans are warranted. Nevertheless, the broad effects of metformin as anti-cancer and anti-CSC agent make it a suitable candidate for therapeutic interventions to improve clinical outcomes.

Summary and Future Perspective

Metformin as a promising anti-cancer agent is supported by extensive epidemiologic, preclinical and clinical data. Inhibition of mitochondrial complex I and activation of AMPK are the major effects of metformin, though mechanisms targeting epigenetic regulation and other pathways have also been identified. Recently, metformin has entered the spotlight due to studies highlighting its ability to target CSCs, which is associated with drug resistance and tumor relapse. Various preclinical studies have suggested that metformin selectively inhibits CSCs via targeting of the AMPK/mTOR/PI3K, insulin/IGF1, Ras/Raf/Erk, Shh, Wnt, TGFβ, Notch, and NF-κB signaling pathways, which have diverse roles in cell proliferation, self-renewal, differentiation, metastasis and metabolism. Metformin-induced regulation of these key pathways has been outlined in Fig. 1, indicating the anti-cancer mechanisms of metformin. Despite promising preclinical data, several challenges lie ahead with regards to the potential clinical applications of metformin. As such, further studies are needed to identify immediate targets of metformin as well as the critical regulators/mediators of the anti-cancer responses that have been demonstrated in vitro and in vivo. By increasing our understanding of the anti-cancer mechanisms of metformin, this will help optimize treatment conditions of metformin as a monotherapy or in combination with other cancer therapeutic strategies, particularly in non-diabetic cancer patients. Moreover, clinical responsiveness to metformin in patients with aggressive subtypes or refractory cancers needs to be assessed. Overall, metformin exhibits potentially significant translational value due to its anti-cancer mechanisms and responses that may be capable of treating a broad spectrum of human cancers.

Acknowledgement

We thank Dr Erin Howard for her critical reading and editing of this manuscript.

References

1

Pierotti

 

MA

, 

Berrino

 

F

, 

Gariboldi

 

M

, 

Melani

 

C

, 

Mogavero

 

A

, 

Negri

 

T

, 

Pasanisi

 

P

, et al. 

. 

Targeting metabolism for cancer treatment and prevention: metformin, an old drug with multi-faceted effects

. 

Oncogene

 

2013

, 

32

: 

1475

–

1487

.

Google Scholar

Crossref

PubMed

 

2

Del Barco

 

S

, 

Vazquez-Martin

 

A

, 

Cufi

 

S

, 

Oliveras-Ferraros

 

C

, 

Bosch-Barrera

 

J

, 

Joven

 

J

, 

Martin-Castillo

 

B

, et al. 

. 

Metformin: multi-faceted protection against cancer

. 

Oncotarget

 

2011

, 

2

: 

896

–

917

.

Google Scholar

Crossref

PubMed

 

3

Pernicova

 

I

, 

Korbonits

 

M

. 

Metformin—mode of action and clinical implications for diabetes and cancer

. 

Nat Rev Endocrinol

 

2014

, 

10

: 

143

–

156

.

Google Scholar

Crossref

PubMed

 

4

Evans

 

JM

, 

Donnelly

 

LA

, 

Emslie-Smith

 

AM

, 

Alessi

 

DR

, 

Morris

 

AD

. 

Metformin and reduced risk of cancer in diabetic patients

. 

BMJ

 

2005

, 

330

: 

1304

–

1305

.

Google Scholar

Crossref

PubMed

 

5

Libby

 

G

, 

Donnelly

 

LA

, 

Donnan

 

PT

, 

Alessi

 

DR

, 

Morris

 

AD

, 

Evans

 

JM

. 

New users of metformin are at low risk of incident cancer: a cohort study among people with type 2 diabetes

. 

Diabetes Care

 

2009

, 

32

: 

1620

–

1625

.

Google Scholar

Crossref

PubMed

 

6

Arrieta

 

O

, 

Varela-Santoyo

 

E

, 

Soto-Perez-de-Celis

 

E

, 

Sanchez-Reyes

 

R

, 

De la Torre-Vallejo

M

, 

Muniz-Hernandez

 

S

, 

Cardona

 

AF

. 

Metformin use and its effect on survival in diabetic patients with advanced non-small cell lung cancer

. 

BMC Cancer

 

2016

, 

16

: 

633

.

Google Scholar

Crossref

PubMed

 

7

Seo

 

YS

, 

Kim

 

YJ

, 

Kim

 

MS

, 

Suh

 

KS

, 

Kim

 

SB

, 

Han

 

CJ

, 

Jang

 

WI

, et al. 

. 

Association of metformin use with cancer-specific mortality in hepatocellular carcinoma after curative resection: a nationwide population-based study

. 

Medicine (Baltimore)

 

2016

, 

95

: 

e3527

.

Google Scholar

Crossref

PubMed

 

8

Decensi

 

A

, 

Puntoni

 

M

, 

Goodwin

 

P

, 

Cazzaniga

 

M

, 

Gennari

 

A

, 

Bonanni

 

B

, 

Gandini

 

S

. 

Metformin and cancer risk in diabetic patients: a systematic review and meta-analysis

. 

Cancer Prev Res

 

2010

, 

3

: 

1451

–

1461

.

Google Scholar

Crossref

 

9

Zhang

 

P

, 

Li

 

H

, 

Tan

 

X

, 

Chen

 

L

, 

Wang

 

S

. 

Association of metformin use with cancer incidence and mortality: a meta-analysis

. 

Cancer Epidemiol

 

2013

, 

37

: 

207

–

218

.

Google Scholar

Crossref

PubMed

 

10

Xu

 

H

, 

Chen

 

K

, 

Jia

 

X

, 

Tian

 

Y

, 

Dai

 

Y

, 

Li

 

D

, 

Xie

 

J

, et al. 

. 

Metformin use is associated with better survival of breast cancer patients with diabetes: a meta-analysis

. 

Oncologist

2015

, 

20

: 

1236

–

1244

.

Google Scholar

Crossref

PubMed

 

11

Li

 

Y

, 

Hu

 

L

, 

Xia

 

Q

, 

Yuan

 

Y

, 

Mi

 

Y

. 

The impact of metformin use on survival in kidney cancer patients with diabetes: a meta-analysis

. 

Int Urol Nephrol

 

2017

, 

49

: 

975

–

981

.

Google Scholar

Crossref

PubMed

 

12

Ma

 

SJ

, 

Zheng

 

YX

, 

Zhou

 

PC

, 

Xiao

 

YN

, 

Tan

 

HZ

. 

Metformin use improves survival of diabetic liver cancer patients: systematic review and meta-analysis

. 

Oncotarget

 

2016

, 

7

: 

66202

–

66211

.

Google Scholar

PubMed