Introduction

Metformin is one of the most commonly prescribed anti-diabetic drugs worldwide. Its history can be traced back to 1918 when guanidine, found in traditional herbal medicine in Europe known as Galega officinalis, was shown to lower glycemia [1]. A series of guanidine derivatives, including metformin, was subsequently synthesized [2]. Over time, the benefits associated with repurposing metformin for several challenging diseases, including obesity [3], cardiovascular diseases [4, 5], liver diseases [6], renal diseases [7], aging-related diseases [8], and cancers [9] have been shown. Epidemiological studies have revealed that metformin exerts protective effects on people with diabetes suffering from cancer [10,11,12]. Intriguingly, several clinical studies have also reported encouraging outcomes in non-diabetic cancer patients [13,14,15]. Given that metformin is safe, well-tolerated, and cost-effective, it is extremely appealing as a focus of antitumor research. Subsequent evidence has shown that metformin inhibits tumor growth, invasion, and metastasis both in vitro and in mouse tumor models for hepatocellular carcinoma [16], ocular melanoma [17], head and neck squamous cell carcinoma [18], and breast cancer [19], among others. Moreover, metformin has been used as a synergistic therapy for cancer, as it enhances sensitivity to radiotherapy [15, 20], chemotherapy [14, 21], and immunotherapy [22] and decreases side effects at lower therapeutic dosages of anticancer treatments.

Great interest has been attached to the basic and clinical study of metformin in cancer. The central mechanism by which metformin attenuates tumorigenesis and progression is through the regulation of energy metabolism. The master pathway of metformin anticancer activity is the activation of the adenosine monophosphate-activated protein kinase (AMPK)/mammalian target of rapamycin (mTOR) pathway triggered by inhibition of complex I in the mitochondrial respiratory chain [23,24,25]. However, the vague performance in a clinical study was in contrast with the excellent performance in a preclinical study. Metformin did not show any benefit in cancer treatment in some clinical trials. Therefore, there are great challenges in the clinical translation of metformin.

Abundant reviews have elaborated on the topic of metformin and cancer from different perspectives, such as specific cancer types [26, 27], diabetes [28, 29], pharmacology [30, 31], and molecular mechanisms [32, 33]. However, insight into the therapeutic repurposing of metformin is still insufficient [34]. Based on the literature review, we recognize that metformin exerts protective effects against multiple tumor types and an increasing number of subtypes [35, 36]. Hence, the mechanisms of action of metformin must be closely related to the hallmarks of cancer [37], which have been proposed as a common set of functional capabilities crucial to the transformation from normalcy to malignancy. This review focused on the effects of metformin on cancer cells in terms of the hallmarks of cancer and updated the clinical translation of metformin in cancer treatment. In this review, we aim to (1) update the readers on the molecular mechanisms through which metformin exhibits antitumor activities, (2) map the effects of metformin on cancer cells in terms of the hallmarks of cancer (Fig. 1), and (3) summarize seminal clinical trials and therapeutic prospects of metformin for cancer treatment.

Fig. 1
figure 1

Main anticancer mechanisms of action of metformin based on hallmarks of cancer. Metformin can combat cancer by affecting metabolism, epigenetics, cell cycle, migration, metastasis, cell death, cell senescence, cancer stem cells, immunity, and gut microbes

Full size image

Update on metformin’s molecular antitumor mechanisms of action

The classic modes of metformin’s antitumor effects are the inhibition of respiratory complex I in the mitochondria and the activation of AMPK in succession. Recently, metformin was defined to inhibit complex I by binding in the quinone channel and exert an independent localized chaotropic effect by combining cryo-electron microscopy and enzyme kinetics [38]. Although the direct interaction between metformin and complex I is essential, metformin-induced complex I inhibition is not a consequence of the direct interaction but instead occurs through an indirect mechanism [39]. Ma et al. [40] conducted another novel study that focused on direct molecular targets of metformin and identified PEN2, a subunit of γ-secretase, as a direct molecular target of metformin. PEN2 binds to ATP6AP1, inhibits the activity of v-ATPase without increasing AMP or ADP, and then activates the lysosomal AMP-independent AMPK pathway.

Epidemiologic studies indicate that metformin decreased the risk of cancer incidence compared to other anti-diabetic medications. Hence, the anticarcinogenic effects of metformin were traditionally divided into direct (blood glucose- and insulin-independent) and indirect (blood glucose- and insulin-dependent) effects, being mindful that none of the effects are mutually exclusive.

Direct effects of metformin

Metformin can exert direct effects on cancer cells independent of blood glucose and insulin levels, partly through AMPK activation. It is generally acknowledged that metformin inhibits complex 1 (NADH-coenzyme Q oxidoreductase) of the mitochondrial respiratory chain, which leads to membrane depolarization, reactive oxygen species (ROS) release, and a decrease in the ATP/ADP ratio [41, 42]. Metformin requires a robust inner mitochondrial membrane potential to accumulate within the mitochondrial matrix and reversibly inhibits complex 1 [23]. This inhibition of complex I limits the electron flow to complex III, where ROS are generated. Mitochondrial complex III ROS are hypoxic activators of HIF-1 [43]. Therefore, metformin reduces the hypoxic stabilization of HIF-1α protein and HIF-dependent target genes. Additionally, metformin reduces DNA damage and the production of oxidative stress through mitochondrial respiratory chain inhibition [44]. Metformin depleted the tricarboxylic acid (TCA) cycle and blocked the production of biosynthetic precursors. Nearly all TCA cycle metabolites decrease considerably with metformin treatment [45]. Metformin can also inhibit cancer cell growth by decreasing the cellular energy status, and the effects can be reversed by the expression of the metformin-resistant yeast-derived complex I NADH dehydrogenase NDI1 [46].

A series of complicated signal pathways are activated by metformin (Fig. 2). First, metformin is a well-known AMPK activator and a key enzyme in glucose homeostasis, gluconeogenesis, and lipid metabolism. AMPK is directly activated by an increase in either the AMP/ATP or ADP/ATP ratio [47] and is indirectly activated by upstream kinases, including LKB1 [48], Ca(2+)/calmodulin-dependent protein kinase kinase (CaMKK) beta [49] and TGFβ-activated kinase-1 (TAK1) [50]. Wu et al. [51] recently demonstrated that metformin protects AMPK-mediated phosphorylation of serine 99, thus increasing TET2 stability and 5-hydroxymethylcytosine (5hmC) levels. A pathway linking diabetes to cancer was revealed through the definition of a novel ‘phospho-switch’ that regulates TET2 stability and a regulatory pathway that links glucose and AMPK to TET2 and 5hmC.

Fig. 2
figure 2

Main molecular anticancer mechanism of action of metformin. The pathways associated with anticancer action may be dependent on AMPK or independent of AMPK

Full size image

AMPK-dependent mTOR complex 1 (mTORC1) inhibition occurs via multiple downstream effectors that switch on ATP-producing processes and switch off ATP-consuming pathways [52]. These effects can be mediated by the activation of TSC1/TSC2 tumor suppressor genes [53, 54]. TSC1/2 can inhibit mTORC1 and the phosphorylation of its downstream effectors 4EBP1 and S6K [55]. mTORC1 inhibition also occurs as a result of the direct phosphorylation of S722 and S792 on Raptor, a vital mTORC1-binding partner [56].

Metformin can inhibit mTOR through Rag GTPase inactivation or REDD1 activation independent of AMPK activation. Metformin can also inhibit mTORC1 signaling independent of AMPK or TSC1/2, although it is dependent on Rag GTPases. Metformin inhibits growth by inhibiting the mitochondrial respiratory capacity, which inhibits the transit of the RagA-RagC GTPase heterodimer through the nuclear pore complex (NPC). A key transcriptional target, acyl-CoA dehydrogenase family member-10 (ACAD10), is activated when metformin induces the nuclear exclusion of the GTPase RagC, thereby inhibiting mTORC1 [57]. REDD1 (REgulated in Development and DNA damage responses 1), also known as RTP801, Dig2, or DDIT4, has been deemed a hypoxia-inducible factor-1 (HIF-1) target gene and plays a significant role in inhibiting mTORC1 signaling during hypoxic stress. Several other pathways are involved in the anticancer action of metformin, including PI3K/AKT/mTOR [58,59,60], K-Ras [61], nemo-like kinase (NLK) [62], c-Jun-N-terminal kinase (JNK) [63], and Stat3-Bcl-2 [64].

Recently, some studies revealed prognostic and predictive biomarkers, as well as a promising therapeutic target of metformin. ** them evade senescence. Metformin also effectively blocks senescence induced by E6/E7 inhibition or chemotherapy in HPV-positive cancer cells.

Locking phenotypic plasticity

Cancer stem cells (CSCs), a cluster of tumor cells possessing clonogenicity and self-renewal abilities, may play a role in tumor recurrence and metastasis. Metformin has been shown to be preferentially cytotoxic to CSCs compared to non-CSCs [132]. Clonal cell growth and cancer sphere formation are hallmarks of CSCs that can be inhibited by metformin. Metformin has been reported to suppress the expression of CSC markers, including CD44, EpCAM, EZH2, Notch-1, Nanog, and Oct4 in pancreatic cells [133]; CD44 and Sox2 in gastric cancer [110]; Nanog, c-Myc, and TLF4 in NSCLC [62]; and upregulate the expression of differentiation markers, such as Kruppel-like factor 4 (KLF4) and MUC5AC in gastric cancer [110]. Metformin suppresses the self-renewal ability and tumorigenicity of osteosarcoma stem cells via ROS-mediated apoptosis and autophagy [104].

Inspiring inflammation and immunity in cancer (Fig. 4)

Fig. 4
figure 4

Metformin regulates diverse factors to modulate immune cells in the tumor microenvironment to inspire immunity in cancer. Metformin can modulate tumor infiltrating lymphocytes (TIL), tumor-associated macrophages (TAMs), Treg, myeloid-derived suppressor cells (MDSCs), and PDL1 to increase the number and function of T cells and decrease T cell surveillance escape. Metformin can also downregulate PDL1 to increase cytotoxic T cells

Full size image

Breakthroughs in cancer immunotherapy have expanded the possibilities for cancer therapy over the last decade. Although cancer immunity has continued to be underappreciated, an increasing number of studies have focused on the relationship between cancer, immunity, and potential immunotherapy. Metformin has been found to interact with immune regulators, such as inhibitory immune checkpoints, M2-like tumor-associated macrophages (TAMs), regulatory T cells (T-regs), and myeloid-derived suppressor cells (MDSCs), to inhibit immune destruction.

CD8+ tumor-infiltrating lymphocytes (TILs) inevitably undergo immune exhaustion, which is characterized by decreased production of multiple cytokines, such as IL-2, TNFα, and IFNγ, followed by a reduction in apoptosis. Metformin increases CD8+ TILs and protects them from exhaustion and apoptosis in the TME. Furthermore, the adoptive transfer of metformin-treated antigen-specific CD8+ T cells efficiently migrates into tumors and maintains multifunctionality in a manner that is sensitive to the AMPK inhibitor compound C [134]. Metformin also inhibits TAM infiltration during prostate cancer initiation and progression by inhibiting the COX2/PGE2 axis [135].

A window of opportunity trial for HNSCC has demonstrated that metformin modulates metabolism in the HNSCC microenvironment [136]. Metformin decreases infiltration of FOXP3+ T regulatory cells in intratumor regions, increases CD8+ cytotoxic T cell infiltration in the peritumoral leading edge stroma, and increases the CD8/FOXP3 ratio both in the tumor and leading-edge stroma of primary HNSCC tumors [137]. Metformin may positively interact with the immune TME in HNSCC, regardless of HPV status. Metformin inhibits CCR1 surface expression in HNSCC cells and the expression of CCL15 in M2-type TAMs, which promote HNSCC cell resistance to gefitinib under hypoxic conditions through the CCL15-CCR1-NF-κB pathway [138].

In a zebrafish model of nonalcoholic fatty liver disease (NAFLD)-associated hepatocellular carcinoma, metformin was found to alter macrophage polarization and exacerbate the liver inflammatory microenvironment and cancer progression [139]. In addition, metformin rescued the effects of a high-fat diet (HFD) on liver tumorigenesis (angiogenesis, steatosis, lipotoxicity), inflammation, and T cell recruitment to the liver.

Combined with immune checkpoint blockade (ICB), metformin has been found to decrease T-reg and MDSC levels and increase CD8+ levels in murine models [140]. Notably, only long-term metformin treatment is sufficient to reduce cancer cell growth. Programmed cell death 1 (PD-1)/programmed death ligand 1 (PD-L1), a representative ICB, has initiated a new era in cancer treatment. Cha et al. [40] revealed that PEN2 is the direct molecular target of metformin using a photoactive metformin probe. Bridges et al. [38] recently defined the inhibitory drug-target interactions of metformin with mammalian respiratory complex I by combining cryo-electron microscopy and enzyme kinetics. The identification of the direct target of metformin’s anticancer effects may help to further investigation for drug development. Second, in in vitro research, the direct effects are emphasized and well-studied, while the indirect effects cannot be mimicked. Some in vivo and clinical studies suggest that indirect insulin-dependent effects may be of great significance in at least some cancers, such as breast cancer and lung cancer. The research directions varied for the two modes of anticancer effects (Table 2). In terms of “direct effects,” more effort should be put into markers such as LKB1, Rag GTPases and REDD1. In terms of “indirect effects,” more effort should be put into markers such as blood glucose and insulin levels, insulin resistance, expression of insulin receptors and insulin-like growth factor receptor 1, and targets in the liver. Understanding the markers helps to predict the therapeutic response of patients. For example, the synergistic effects between metformin and gefitinib were reported to rely on the presence of wild-type LKB1 in NSCLC cells [178]. Third, since indirect effects could not be simulated in vitro, more in vivo models are needed to reexamine the direct and indirect effects of metformin and the possible interactions. Some attempts have been made to develop related models. For instance, in research on the anticancer activity of metformin, hyperglycemic mice were reported to lose sensitivity to metformin compared with normoglycemic mice, probably through increased c-Myc expression, glycolytic enzymes hexokinase 2 and pyruvate dehydrogenase kinase 1 [179]. We expect more in vivo models with more complex designs that are currently used in most in vitro research.

Table 2 Markers with potential predictive value
Full size table

There is a discrepancy in the antitumor effect of metformin between clinical research and preclinical studies, although metformin has shown notable benefits for cancer prevention and treatment in preclinical research, and the related molecular mechanisms have been extensively studied. The challenges mainly include (1) simulating pharmacokinetics consistent with clinical settings, including appropriate metformin concentrations and dosing time; (2) exploring suitable synergetic therapies and patients who are more sensitive to metformin; and (3) utilizing other forms of biguanides, such as phenformin or modified biguanides, which have better performances. Therefore, further research is required regarding the critical aspects mentioned above. First, many preclinical studies have employed metformin at concentrations that are considerably higher than what would be deemed safe in clinical settings [180]. The plasma concentrations of metformin were reported to be 5–30 μmol/L in patients taking clinical doses of 1.5–2 g per day, which were the most common doses for diabetes and were used in most clinical trials for cancer treatment [181]. However, the concentrations in most preclinical studies in vitro (300 μmol/L–10 mmol/L) were dozens or even a thousand times the clinical concentrations. The dosages of in vivo studies (200–1000 mg/kg per day) were 6–30 times the clinical dosages (approximately 30 mg/kg per day) with metformin diluted in the drinking water or intraperitoneally injected. Recent studies have noted this issue. Metformin can suppress cancer at a clinically safe concentration in vitro [40] and in vivo [16, 35, 183]. In the clinical setting, improved outcomes were observed only among patients with early-stage NSCLC or those who took metformin before the NSCLC diagnosis [184]. Therefore, it enlightens us to apply metformin once cancer is diagnosed or even in people with a high risk of cancer if possible. Second, based on our understanding of metformin’s effects on cancer hallmarks, metformin could be a useful adjuvant agent in combination therapy to combat cancer synergistically in certain patients with certain cancers [36]. It can be administered along with chemotherapy, radiotherapy, immunotherapy or targeted therapy. Taking immunotherapy as an example, a high level of lactate can lead to tumor immune tolerance, while metformin was reported to increase the level of lactate in the intra- and extracellular environment [31, 185]. However, from the current evidence, acidification of the TME made tumors more susceptible to metformin due to the loss of NAD+ regeneration capacity [78, 79, 186]. Therefore, whether metformin can cause tumor immune tolerance by increasing the acidification of the TME remains an interesting issue to explore. If indeed, the combination of metformin and immunotherapy might be a possible direction in further research. Besides, metformin was reported to show no more benefits in some cancers with certain mutations [158] or advanced stages [164, 175, 176] or in patients without diabetes [157, 158]. We call for further high-quality clinical trials on metformin combined with other therapies in different types of physiological conditions and cancers. Third, it was reported that phenformin may outperform metformin owing to its unique pharmacokinetic characteristics, which include better absorption and inhibition of the mitochondria [187]. Although the incidence of lactic acidosis associated with phenformin is higher than that associated with metformin, phenformin is in any case safer than other cancer treatments. Moreover, given the pharmacokinetic differences between metformin and phenformin, we can obtain more insights regarding drug modification. Once there is more evidence, we expect metformin or other forms of biguanides to exert a greater influence on anticancer therapy, at the appropriate dosage, on patients of appropriate metabolic state, and in combination with other therapies.

Conclusion

The review details the possible molecular mechanisms of metformin in cancer prevention and treatment, elucidates its role in terms of cancer hallmarks, and more importantly, analyses current challenges and future directions in clinical translation.