Metformin: Review of Epidemiology and Mechanisms of Action in Pancreatic Cancer

Abstract

Pancreatic ductal adenocarcinoma continues to be a lethal disease, for which efficient treatment options are very limited. Increasing efforts have been taken to understand how to prevent or intercept this disease at an early stage. There is convincing evidence from epidemiologic and preclinical studies that the anti-diabetic drug metformin possesses beneficial effects in pancreatic cancer, including reducing the risk of developing the disease and improving survival in patients with early stage disease. This review will summarize the current literature about the epidemiological data on metformin and pancreatic cancer as well as describe the preclinical evidence illustrating the anti-cancer effects of metformin in pancreatic cancer. Underlying mechanisms and targets of metformin will also be discussed. These include direct effects on transformed pancreatic epithelial cells and indirect, systemic effects on extra-pancreatic tissues.

Pancreatic cancer is a highly significant public health problem

Pancreatic ductal adenocarcinoma (PDAC), the most common histologic subtype of pancreatic malignancies, is still one of the most lethal types of cancer. The latest report of the American Cancer Society estimates 60,430 new cases (28,480 females and 31,950 males) of pancreatic cancer in the year 2021 [1]. An estimated 22,950 female and 25,270 male patients will succumb to this disease in 2021, putting pancreatic cancer as the fourth leading cause of cancer mortality in women and men [1]. Alarmingly, the incidence of pancreatic cancer has increased over the last years. In 2016, the American Cancer Society’s estimates for pancreatic cancer were 53,070 (25,400 females and 27,670 males) [2], marking an almost 13.9% increase (12.1% in females and 15.5% in males) in total estimated pancreatic cancer cases in the last 5 years. In fact, it has been projected that the mortality of pancreatic cancers will surpass the deaths from colorectal cancer by 2030, making pancreatic cancers the second leading cause of cancer-related deaths [3]. Only a minority of patients with PDAC are eligible for surgical resection due to the early local and distant spread. Although advances have been made in neoadjuvant and adjuvant chemotherapeutic regimens, the benefit in prolonging survival is still disappointingly small and recurrences occur often. In addition, despite significant progress in the discovery and understanding of molecular pathways driving PDAC growth, molecular targeted therapy has not yet resulted in significant breakthroughs. A different approach to reduce the burden of pancreatic cancer on patients and to lower the number of deaths from this disease is to identify strategies that either prevent or intercept PDAC [4, 5]. In this context, a detailed understanding of modifiable risk factors and repurposing currently used drugs will most likely guide the rapid implementation of new preventive strategies.

Preclinical animal models are invaluable to study the mechanisms of risk factors and to develop interventional strategies. It is well established now that most PDACs arise through a stepwise progression from precursor lesions [6], e.g. pancreatic intraepithelial neoplasias (PanINs). There is general consensus about the importance of mutated KRAS in the initiation of PDAC [7]. This notion is strongly supported by genetically engineered animal models of PDAC development [8, 9]. Accordingly, the model that faithfully recapitulates the progression of human PDAC in mice involves the expression of a mutant Kras (Kras^G12D) from the endogenous Kras locus through a Cre recombinase that is under control of the pancreas-specific promoter p48 (LSL-Kras^G12D; p48-Cre mice referred to as the KC model [10]). In this model the expression of mutated Kras in pancreatic progenitor cells during embryogenesis leads to PanIN formation and progression to PDAC [11]. However, in the conditional Kras mouse model (without additional genetic alterations) the development of invasive PDAC occurs very late (usually after 9 months) and only in ~ 10% of the animals [8]. Mutated Kras is necessary for the initiation but is not sufficient for driving the progression of PanINs to invasive PDAC. However, the presence of another mutation, e.g. in Trp53 (transformation-related protein 53), and/or certain environmental conditions greatly accelerate PDAC development [9].

Metabolic disorders, including obesity and type 2 diabetes mellitus (T2DM) are associated with an increased risk and worse clinical outcomes for developing PDAC and other cancers [12–17]. These connections have been recapitulated in preclinical mouse models of PDAC. In particular, diet-induced and genetic obesity strikingly accelerate the progression of PanINs to invasive cancer in mice [18–22]. Diet-related metabolic disturbances, including obesity and early stages of T2DM, which are prominent in the western world [23–26], are multifaceted but characterized by peripheral insulin resistance, compensatory overproduction of insulin, and increased bioavailability of IGF-1 (insulin-like growth factor-1). Elevated levels of insulin and IGF-1, which are potent growth factors for PDAC cells, are generally thought to be of great importance for the development and progression of cancers, including PDAC [27–30].

Metformin as an intriguing agent in pancreatic cancer: epidemiological evidence

Metformin (1,1-dimethylbiguanide hydrochloride) is the most widely prescribed drug for prevention and treatment of T2DM worldwide [31]. The primary systemic effect of metformin is the lowering of blood glucose levels predominantly through reduced hepatic gluconeogenesis. The molecular pathways responsible for this reduction in glucose production remain a subject of debate [32]. A number of studies suggest that a considerable proportion of the antidiabetic action of metformin derives from effects on the gastrointestinal (GI) tract [33–35], and this will revisited in a later section.

Over the past 10–15 years there have been many epidemiologic studies that analyzed a potential association of metformin use with a reduced risk of PDAC (preventive effect) or improved survival of patients with PDAC (therapeutic effect). Overall, the available literature suggests a beneficial effect of preventive and therapeutic efficacy of metformin in patients with underlying T2DM, with stronger evidence for a preventive effect, although the published reports are not universally consistent. Most epidemiologic studies found an association of metformin use with a reduced incidence of PDAC in patients with T2DM. In a single-institution, case-control study diabetic patients who had taken metformin had a significantly lower risk of PDAC compared with those who had not taken metformin (Odds Ratio [OR]: 0.38; 95% confidence interval [CI]: 0.22–0.69) [36]. Interestingly, diabetic patients who had taken insulin or insulin secretagogues had a significantly higher risk of PDAC compared with diabetic patients who had not taken these drugs [36], supporting the hypothesis that insulin is a potent mitogenic and cancer promoting factor. In contrast, a large cohort study from the Netherlands did not detect an association of current and cumulative use of metformin with a decreased risk of GI cancer, including PDAC [37]. A meta-analysis published in 2010, analyzing eleven studies, reported a significant inverse association between PDAC risk and metformin use [38]. Another meta-analysis in 2013 analyzing 37 studies comprising over 1,500,000 participants found a risk reduction of 46% for developing PDAC in metformin users [39]. Finally, a recent umbrella review on the potential effect of metformin on cancer, which included 21 systematic reviews and meta-analyses covering 11 major anatomical sites, found strong evidence for the association between metformin use and decreased pancreatic cancer incidence [40].

Besides strong epidemiologic evidence of an association of metformin use and decreased risk of PDAC, several studies assessed a potential association of metformin and improved survival of PDAC patients. A retrospective study of patients with diabetes and PDAC treated at the University of Texas MD Anderson Cancer Center showed a better survival in patients with metformin use [41]. The 2-year survival rate was 30.1% for the metformin group and 15.4% for the non-metformin group. The median overall survival time was 15.2 months for the metformin group, and 11.1 months for the non-metformin group [41]. In that study metformin use was significantly associated with longer survival only in patients with non-metastatic disease [41]. A nationwide population-based cohort study in Korea suggested that metformin might decrease cancer-specific mortality rates in localized resectable PDAC patients with pre-existing diabetes [42]. The cancer-specific survival (5-year, 31.9% vs. 22.2%) was significantly higher in the metformin users than in the diabetic metformin non-users [42]. In contrast, a large cohort study, which retrospectively analyzed 980 patients with PDAC failed to detect a benefit of metformin [43]. Median survival of metformin users versus nonusers was 9.9 versus 8.9 months, respectively and metformin use was not significantly associated with improved survival (Hazard Ratio [HR]=0.93; P=0.28) [43]. A double-blind, randomized, placebo-controlled phase 2 trial in the Netherlands in patients with advanced PDAC did not find a benefit of metformin when added to gemcitabine and erlotinib [44]. Overall survival at 6 months was 63.9% in the placebo group and 56.7% in the metformin group (P=0.41). There was no difference in overall survival between groups (median 7.6 months vs 6.8 months in the metformin group; HR=1.056; log-rank P=0.78) [44]. Several meta-analyses, however, described a beneficial association between metformin use and survival in PDAC patients. A systematic review and meta-analysis of nine studies found a relative survival benefit in patients with PDAC and preexisting T2DM associated with metformin treatment compared with non-metformin treatment in overall survival [45]. A recent meta-analysis in 2020 of 21 studies that included 38,772 patients with PDAC and concurrent T2DM found a significant survival benefit in the metformin treatment group compared with non-metformin group (HR=0.83) [46]. Interestingly, the survival benefit was observed in patients at early stage (HR=0.75) and mixed stage (HR=0.81), but not for patients at an advanced stage (HR=0.99) [46]. Also, the survival benefit was seen in patients receiving surgery (HR=0.82) and comprehensive treatment (HR=0.85), but not in the chemotherapy group (HR=0.99) [46]. Some of the contrasting results of epidemiologic studies may be related to weaknesses in the study design and data analysis. In fact, a recent study that included patients who only received metformin after surgery and corrected for immortal bias, a major weakness in many previous publications, concluded that metformin use is associated with a higher overall survival following pancreaticoduodenectomy in patients with T2DM and PDAC [47]. In contrast, another study assessing the impact of immortal time bias in previous cohort studies concluded that the correlation between the use of metformin and PDAC survival was exaggerated in several studies due to the existence of immortal time bias [48]. This contrasting conclusions highlight the importance of more rigorous statistical designs in investigating the effects of metformin on PDAC survival. Another potentially confounding and complicating factor in interpreting the efficacy of metformin on PDAC risk and survival in epidemiologic studies is the often concomitant use of several drugs, e.g. metformin, aspirin, statins, in patients with T2DM or metabolic syndrome [49]. A new retrospective analysis of 33,784 patients from the Korean National Health Insurance Service showed a reduction of all-cause mortality in diabetic cancer survivors using metformin [50]. Taken together the available epidemiologic evidence suggest a positive correlation between metformin use and survival in patients with early stage PDAC and co-existing T2DM. Indeed, a meta-analysis from 2017 indicated that the effects of metformin depend on tumor stage, with marked improved survival in patients with locally advanced disease but not in patients with metastatic PDAC [51].

Metformin as an intriguing agent in pancreatic cancer: preclinical evidence

Early preclinical evidence of a potential preventive effect of metformin in pancreatic carcinogenesis came from hamster studies. Metformin administration in the drinking water prevented n-nitrosobis-(2-oxopropyl)amine-induced pancreatic carcinogenesis in hamster fed a high fat diet [52]. Using an allograft model, metformin significantly reduced tumor growth in prediabetic C57BL/6 mice with diet-induced obesity (DIO) [53]. The anti-cancer effect of metformin in allograft/xenograft models of pancreatic cancer has been corroborated by several other studies [54–58]. Metformin and phenformin (another related antidiabetic drug from the biguanide class) have also been shown to significantly reduce tumor growth in some patient-derived xenografts [59] but not all [60]. Our own earlier studies have shown that intraperitoneal administration of metformin (at 250 mg/kg) significantly reduced the growth of subcutaneous pancreatic cancer xenografts (using MIAPaCa-2 and PANC-1 human pancreatic cancer cell lines) in nu/nu mice [61]. Subsequently, we reported that metformin (administered intraperitoneally) dose-dependently inhibited subcutaneous pancreatic cancer xenograft growth with a significant growth inhibitory effect starting at 50 mg/kg [62]. Importantly, metformin also inhibited pancreatic cancer growth when xenografted orthotopically and administered orally (2.5 mg/ml in drinking water) [62].

Using a genetically engineered mouse model (LSL-Kras;Trp53), administration of metformin (intraperitoneally at 125 mg/kg) 1 or 3 weeks before tumor initiation (via direct orthotopic injection of adenoviral Cre constructs) and continued for another 6 weeks significantly reduced pancreatic tumor volume [63]. Other studies reported that daily oral gavage of metformin (at 200 mg/kg) inhibited tumor growth and pancreatitis and prolonged survival in LSL-Kras^G12D/+;Pdx1-Cre (KC) and LSL-Kras^G12D/+;Trp53^fl/+;Pdx1-Cre (KPC) mice [64] as well as inhibited angiogenesis and enhanced chemosensitivity of gemcitabine in KPC mice [65]. Our own studies demonstrated that metformin decreased the incidence of PDAC in Kras^G12D/+;p48-Cre (KC) mice with DIO [66, 67]. In that study male and female KC mice were randomly fed with a control diet (CD), high fat, high calorie diet (HFCD), or HFCD plus metformin (administered in the drinking water at 5 mg/ml) for 3 or 9 months. After 3 months, metformin prevented HFCD-induced weight gain, liver steatosis, formation of acinar-to-ductal metaplasia (ADM), and advanced PanIN lesions [66]. In addition to reversing hepatic and pancreatic histopathology, metformin normalized HFCD-induced hyperinsulinemia and hyperleptinemia in the 9-month cohort. Importantly, the HFCD-increased PDAC incidence was completely abrogated by metformin [66]. The chemopreventive effects of metformin in this model were associated with significant microbial changes in both the mucosal and luminal microbiome of the duodenum [67]. In contrast to our study, a recent report using KC mice with genetic obesity (Pdx1-Cre;LSL-Kras^G12D/+ mice crossed with leptin-deficient [ob/ob] mice; KCO) did not detect a beneficial effect of metformin [22]. In that study, KCO mice were given metformin in the drinking water (2 mg/ml) starting at 6 weeks of age and continued for another 6 weeks. There are important differences, including rapid vs. gradual weight gain, lack of vs. elevated leptin levels, between our model with DIO and the genetic obesity model, which may explain the different response to metformin [22, 68]. Importantly, metformin was administered at a lower concentration (2 mg/ml) for only 6 weeks in the KCO model, which was not associated with weight loss and a decrease in insulin levels [22], while our study used metformin at 5 mg/ml for 3 and 9 months, which was associated with weight loss (also seen in diabetic patients on metformin) and a significant reduction in insulin levels [66]. Overall, there is overwhelming preclinical evidence from allograft/xenograft and genetically engineered mouse models of PDAC that metformin potently inhibits tumor growth and reduces the incidence of PDAC formation in obese and non-obese conditions.

Metformin as an intriguing agent in pancreatic cancer: underlying mechanisms

The mechanism of action of metformin remains incompletely understood. The beneficial effects of metformin on PDAC development and growth may thereby be mediated by direct effects on transformed pancreatic cells, by indirect, systemic extra-pancreatic effects or by a combination of both mechanisms. The systemic effects may include metformin action in the liver, skeletal muscle, adipose tissue, brain, and/or gut [7, 34, 69–75].

Direct cellular effects:

At the cellular level metformin indirectly activates AMPK (adenosine monophosphate-activated protein kinase) [76], although AMPK-independent mechanisms have also been described [69, 77–79]. Most of these studies were carried out in liver cells and attempted to identify the mechanism, by which metformin decreases glucose output. A number of studies were performed using PDAC cells, but an important limitation of most studies is that metformin was used at very high concentrations. While circulating levels of metformin in humans and animals range from 10 μM to 40 μM (approximately 40–70 μM in the portal vein), effects of metformin on signaling pathways and proliferation in PDAC cells were examined using drug concentrations in the culture medium as high as 5–30 mM [57, 80–82]. Therefore, the clinical and mechanistic significance of the effects of metformin on PDAC cells in vitro has been questioned.

Within cells, AMPK exists as heterotrimeric complexes, composed of a catalytic kinase α subunit (isoforms α1 and α2), a scaffolding β subunit (isoforms β1 and β2) and a regulatory γ (isoforms γ1, γ2 and γ3), giving rise to 12 possible α/β/γ combinations in different cell types [83, 84]. It is plausible that the different AMPK complexes have different regulation (e.g. expression, turnover, subcellular localization), regulatory properties (e.g. dependency on 5’-AMP/ADP [adenosine monophosphate/adenosine diphosphate] concentration), and functions in different cell types [85]. Activation of AMPK is achieved by phosphorylation in the activation loop by LKB-1/STK11 (liver kinase B-1/serine-threonine kinase 11) [86] in conditions where cellular levels of ATP (adenosine triphosphate) decrease and 5’-AMP and ADP concentrations rise [31]. Metformin does not act directly on either AMPK or LKB1, but is generally thought to inhibit the complex I of the mitochondrial respiratory chain [87] leading to a decline in ATP production accompanied by a concomitant increase in ADP and AMP levels [88]. The binding of AMP or ADP to the γ1/2 subunits induces a conformational change that facilitates the phosphorylation of the activation loop of the α subunit by LKB-1/STK11 at the highly conserved Thr172 [85]. Interestingly, AMPK with mutant subunits that do not bind AMP are not sensitive to metformin. However, changes in adenine nucleotides can affect the activity of other enzymes in the cell and thus, metformin can exert effects through AMPK independent pathways [32] especially at high concentrations [89–91] that cause pronounced changes in adenine nucleotide levels [90]. A different mechanism of AMPK activation involving a lysosomal pathway has also been proposed [92].

Using human PDAC cells exposed to medium containing a physiological concentration of glucose (5 mM), our studies showed that metformin inhibited cell proliferation at relevant concentrations (50–100 μM). Knock down of AMPKs prevented the inhibitory effect of low (but not high) concentrations of metformin on DNA synthesis in PDAC cells [90, 91]. These studies emphasized that the mechanism of the inhibitory effects of metformin on PDAC cell signaling and proliferation is sharply dose-dependent. At low concentrations, metformin inhibited cell proliferation via AMPK, whereas at high concentrations (>1 mM), metformin elicited cellular responses through AMPK-independent pathways.

A key pathway in KRAS signaling is phosphatidylinositol 3-kinase (PI3K)/Akt leading to mTOR (mechanistic target of rapamycin) activation [93–95]. This signaling module plays a pivotal role in stimulating proliferation of PDAC cells [96, 97], is activated in PDAC tissues and limits catabolic processes, including autophagy [98]. mTOR functions as a catalytic subunit in two distinct multi-protein complexes, mTORC1 (mechanistic target of rapamycin complex 1) and mTORC2 [99]. mTORC1, characterized by Raptor, phosphorylates and controls at least two regulators of protein synthesis, the 40S ribosomal protein subunit S6 kinase (S6K) and the inhibitor of protein synthesis 4E-binding protein 1, referred as 4EBP1 [100–103]. As depicted in Figure 1, AMPK has been shown to inhibit mTORC1 through several mechanisms. First, AMPK can phosphorylate TSC2 (tuberous sclerosis complex 2) on Ser1345 [104–106], thereby activating the complex, which in turn prevents the accumulation of Rheb-GTP (ras homolog enriched in brain-guanosine tri-phosphate), thus inhibiting mTORC1 activation, a process that occurs at the lysosomal membrane. Second, AMPK can also phosphorylate Raptor, which disrupts its binding to mTOR, thereby inhibiting mTORC1 [107]. Finally, AMPK has been shown to phosphorylate IRS-1 (insulin receptor substrate-1) at Ser794, which impedes activation of PI3K [108, 109], thereby leading to inhibition of mTORC1 activation induced by insulin and/or IGF-1. Our own studies have demonstrated that metformin at low concentrations activates AMPK in PDAC cells [90, 91, 96], as judged by the increase in the phosphorylation of both Raptor and ACC (acetyl-CoA carboxylase) Ser79 phosphorylation, thereby inhibiting mTORC1, ERK (extracellular signal-regulated kinase), and DNA synthesis. Accordingly, oral administration of metformin prevented the increase in PDAC incidence in KC mice with DIO [66], which was associated with an increase in AMPK activity in the pancreas (as measured by ACC Ser79 phosphorylation).

Fig. 1.

Schematic of direct molecular, AMPK-dependent effects of metformin

Metformin indirectly activates AMPK, which in turn phosphorylates multiple targets leading to inhibition of the Raf/MEK/ERK/RSK, mTORC1, and YAP/TAZ pathways, all critical oncogenic drivers in pancreatic cancer. Activating phosphorylations are depicted in black, inhibiting phosphorylations in red. Partly created with Servier Medical Art.

Another putative target of AMPK of high significance is the highly conserved Hippo pathway, which is attracting intense attention as a key regulator of development, organ-size, tissue regeneration and tumorigenesis [110, 111]. Hippo signals are transduced through a serine/threonine kinase cascade wherein Mst1/2 kinases (mammalian STE20-like protein kinases 1/2) phosphorylate and activate Lats1/2 (large tumor suppressor 1/2). In turn, Lats1/2 phosphorylates the transcriptional co-activators YAP (Yes-associated protein) and its paralog TAZ (WW-domain-containing transcriptional co-activator with PDZ-binding motif), two central effectors of the Hippo pathway [112–114]. The phosphorylation of YAP/TAZ by Lats1/2 restricts their activity, cellular localization, and stability. In the absence of phosphorylation, YAP/TAZ localizes to the nucleus where it binds and activates predominantly the TEA-domain DNA-binding transcription factors (TEAD 1–4) thereby stimulating the expression of genes. Besides its recognized role in the regulation of growth and development, recent studies show that Hippo kinases and YAP/TAZ transcriptional coactivators, are regulated by metabolism and conversely that the Hippo/YAP/TAZ pathway controls metabolic processes in physiological and pathologic conditions, including obesity and T2DM [115, 116]. Although YAP and TAZ have similar structural topologies, share nearly half of the overall amino acid sequence, and are thought to be largely redundant, they may differ in their regulation and downstream functions [117, 118]. The YAP/TAZ pathway assumes an added importance in PDAC given that YAP is a major downstream target of Ras required for PanIN progression into PDAC [119, 120]. YAP is overexpressed in human pancreatic cancer, particularly in the squamous subtype, and correlates to poor survival [121–126]. YAP/TAZ are upregulated in cerulein-induced acute pancreatitis and strongly expressed in ADM and PanIN lesions in the KC mouse model [120]. YAP has been shown to drive KrasG12D-independent tumor maintenance [123] and collaborates with oncogenic Kras to drive tumorigenesis [127].

Metformin has been shown to inhibit the activity or expression of YAP/TAZ in several (cancer) models [128–133]. The inhibition of YAP/TAZ by metformin is mediated by AMPK, which opposes YAP activity via multiple mechanisms, including direct YAP phosphorylation on Ser94 [134, 135], a site that is critical for the interaction of YAP with TEAD transcription factors. In addition, AMPK has been shown to phosphorylate HMG-CoA (hydroxy-methyl-glutaryl-coenzyme A) reductase on Ser872, which inhibits its activity and decreases mevalonic acid synthesis [136], a pathway critical for YAP regulation through the formation of GG-PP (geranyl-geranyl pyrophosphate) and prenylation of Rho GTPases [137]. In addition, AMPK phosphorylates and activates upstream regulators of the Hippo pathway [138], including LATS, leading to inhibition of YAP activity. In our own studies, we found that KC mice with DIO had markedly elevated YAP and TAZ expression in the pancreas, which was significantly prevented by oral administration of metformin [66]. A schematic overview of AMPK-dependent effects of metformin is shown in Figure 1.

Indirect systemic, extra-pancreatic effects:

As mentioned earlier, the anti-diabetic, glucose-lowering effects of metformin are mediated systemically by decreasing the production of glucose in the liver and by improving the insulin sensitivity by virtue of increasing the uptake of glucose into skeletal muscle cells and adipocytes [139]. In addition, metformin is also able to lower the levels of insulin and IGF-1 in diabetic and non-diabetic patients [140, 141]. There is substantial evidence that suggests a prominent role of insulin and IGF in PDAC development and progression [142–149]. The beneficial effects of metformin in PDAC patients, especially with co-existing obesity and/or T2DM, might thereby be mediated, at least in part, by reducing or normalizing the systemically elevated insulin and IGF levels. This was also seen in our animal study, in which oral administration of metformin normalized the hyperinsulinemia in KC mice with DIO, which was associated with a marked reduction in PDAC incidence [66]. Interestingly, in that study metformin treatment also correlated with a substantial improvement of liver steatosis seen in obese KC mice [66]. In this context it is important to note that non-alcoholic fatty liver disease (NAFLD), often seen in obesity, has been found to be associated with an increased risk of PDAC and with a poorer survival of patients with PDAC [150–152]. However, further research is needed to determine whether the link between NAFLD and PDAC merely reflects the general promotional effect of obesity (and visceral adiposity) on PDAC development or whether specific mechanisms exists that are unique to hepatic steatosis and PDAC formation.

Over the last decade there has been growing evidence indicating the importance of the gut microbiome in obesity and other metabolic conditions [153, 154]. Accordingly, a number of studies have found that the anti-diabetic and anti-obesity actions of metformin may be mediated by its effects on the gut, including the gut microbiome. In fact, the concentration of metformin in the intestine is 30–300 times higher than the levels of the drug in the circulation, implying striking accumulation of metformin in enterocytes [155]. A large observational study and a randomized controlled trial have identified a microbiome signature of metformin [156, 157]. The beneficial anti-diabetic effects of metformin can be transferred to germ free mice through fecal transplantation, providing strong evidence that changes of the gut microbiome underlie at least some of glucose lowering effects of metformin [157]. Findings from human studies have been corroborated by animal studies. Oral administration of metformin to high fat diet-fed mice for 6 weeks improved the glycemic profile, which was associated with a higher abundance of the mucin-degrading bacterium Akkermansia [158]. Oral administration of Akkermansia muciniphila to mice fed the high fat diet but not treated with metformin significantly improved glucose tolerance [158] providing a causal link between microbiome effects of metformin to its anti-diabetic actions. There has been increasing evidence of the importance of the gut microbiome in pancreatic cancer [159–161]. In our own studies we found that KC mice fed a high fat high calorie diet for 3 months displayed a significant shift in the microbial composition and α-diversity in the duodenum, a sign of dysbiosis or microbial imbalance [67]. Oral administration of metformin, which was associated with an attenuation of PDAC development, prevented the changes in the duodenal microbiome. Specifically, the high fat, high calorie diet increased substantially the genus Clostridium sensu stricto within the duodenum, which was prevented by metformin [67]. Importantly, Clostridium sensu stricto negatively correlated with the extent of intact pancreatic acini in KC mice [67]. Overall, there is increasing evidence that at least some of the beneficial effects of metformin in reducing PDAC development may be mediated by its actions on the gut microbiome. However, there is a paucity of causal links and further experimental work is needed to elucidate the role of the intestinal microbiota and their metabolites in PDAC de and in the inhibitory effect of metformin on PDAC development.

In addition to its actions on the gut microbiome, the glucose-lowering effects of metformin in diabetic patients have also been shown to be mediated by increasing the plasma levels of the gut incretin hormone, glucagon-like peptide 1 (GLP1) [162, 163], which is secreted by intestinal enteroendocrine L cells. In turn, GLP1 released locally in the gut can activate a gut-brain-liver neuronal network to inhibit hepatic glucose production [164]. Moreover, a recent elegant study has demonstrated that oral metformin increases circulating growth differentiation factor 15 (GDF15), expressed mainly in the intestine and kidney, which is necessary for metformin’s beneficial effects on energy balance, food intake, and body weight [73]. The effects of metformin on GDF15 were also produced in intestinal organoids growing in three-dimensional cultures [165]. The reduction of food intake and body weight by metformin are also thought to contribute to the anti-cancer activities of metformin, especially in obese and diabetic patients. Figure 2 illustrates some of the indirect, systemic extra-pancreatic effects of metformin.

Fig. 2.

Overview of indirect, systemic extra-pancreatic effects of metformin

Besides its direct effects on transformed pancreatic cells, metformin acts on several other organs/tissues, including liver, skeletal muscle, fat, gut, and brain, leading to a reduction of blood glucose and insulin levels, reduced food intake, weight loss, and gut microbiome changes that all together may also have profound opposing/preventive effects on pancreatic cancer development and growth. Partly created with Servier Medical Art.

Conclusion

There is overwhelming evidence from epidemiologic and preclinical studies that the anti-diabetic drug metformin reduces the risk of developing pancreatic cancer and improves outcome in patients with early stage pancreatic cancer. These anti-cancer effects of metformin may thereby be mediated by its direct actions on transformed pancreatic cells or indirectly by its systemic actions on extra-pancreatic tissues/organs. Among the direct mechanisms on transformed pancreatic cells, inhibition of the YAP/TAZ transcriptional co-activators by metformin may be of paramount importance. In addition, the glucose-lowering properties of metformin together with a decrease of the levels of insulin and IGF as well as its normalizing effects on the gut dysbiosis in diabetic and/or obese patients may contribute to attenuating PDAC development and progression. Further mechanistic studies are needed to dissect the individual contributions of the direct and indirect effects of metformin. However, we postulate that metformin is an exciting and promising drug for PDAC prevention and therapy exactly because it counteracts PDAC development and growth on several levels (direct and indirect).