To date, epidemiological evidence suggests that T2DM is one of the risk factors for CRC. However, the latest meta-analysis of observational studies displayed insulin therapy significantly increased the risk of CRC [risk ratio (95% confidence interval (CI)): 1.69 (1.25, 2.27)][6]. Compared to this, the latest meta-analysis from observational studies demonstrated that metformin therapy was associated with a significantly lower overall survival [pooled risk ratio = 0.75, 95%CI: 0.66-0.86][7], lower CRC-specific survival [hazard ratio (95%CI): 0.66 (0.50-0.87)][8], and a lower risk of colorectal neoplasm [risk ratio (95%CI): 0.63 (0.50-0.79)] in patients with T2DM[9]. In vivo, Tomimoto et al[10] suggested administration of metformin significantly reduced the number of tumors larger than 2 mm in diameter in Apc (Min/+) mice. Subsequent studies have shown that metformin suppressed colorectal aberrant crypt foci in humans[11,12] and an azoxymethane-induced animal mode of CRC[13]. In patients with T2DM, it has been reported that metformin has a positive prophylactic effect on colorectal adenomas[14] and CRC[15]. In a multicenter double-blind, placebo-controlled, randomized phase 3 trial, Higurashi et al[16] proved the administration of low-dose metformin for 1 year to patients was safe and reduced the prevalence and number of metachronous adenomas or polyps after polypectomy.

At the cellular level, the effect of metformin is mainly manifested in inhibiting the proliferation of CRC cells as a result of AMPK activation and increased reactive oxygen species production[17,18] on the one hand and accelerating the apoptosis of CRC cells induced by immune cells and cytokines on the other hand[19]. Complementarily, metformin may go through a variety of pathways or mechanisms, such as by alteration of cellular responses to oxidative stress, protection of mitochondrial structures via suppressing reactive oxygen species production and NF-kB activity[20-22], and repression of epithelial-mesenchymal transition induced by IL-6 or TGF-β[19,23]. At the genetic level, it has been shown that the upregulation of adenosine A1 receptor[24], blocking of encoding DNA replication proteins and protooncogene protein synthesis[25,26], and regulation of the SNAIL/miR-34:ZEB/miR-200 system[27] are the targets of metformin. In addition, key sites for metformin action have been shown in multiple signaling pathways, such as attenuation of cell stemness via inhibiting the Wnt3a/β-catenin[28] and AMPK/PI3K/Akt pathways[29].

To sum up, in both humans and animals or at both the individual and cellular level, metformin has shown an effect of inhibiting the development of CRC. However, as a treatment for T2DM, the underlying mechanism has not been fully elucidated.

The changes of gut microbiota have been investigated to be associated with metformin administration, contributing to the beneficial therapeutic effects[30,31]. Decreased profiles of Intestinibacter spp. and Clostridium spp. were observed upon administration of metformin to health young men with a corresponding increase of Escherichia/Shigella spp. and Bilophila wadsworthia[32]. To date, there are many studies focusing on the role of the gut microbiota that mediate the effect of metformin in regard to improving the metabolic phenotype.

In diet-induced obese mice, an increase in the population of Akkermansia, a mucin degradation bacteria, induced the improved glucose homeostasis in the cases of metformin administration[33]. Subsequent work focused on the effects of improving metabolism and delaying T2DM progression in animal models[34,35]. The study addressed by Forslund et al[36] supported the opinion from the perspective of the human gut genome that metformin could reduce the consumption of short-chain fatty acid-producing groups in T2DM and improve the transfer of functional microbial groups. Through the implementation of 16S rRNA gene sequencing, it was observed that the participating microbiota were mainly Akkermansia muciniphila and short-chain fatty acid-producing bacteria, including Butyrivibrio, Bifidobacterium bifidum, Megasphaera, and Prevotella after the metformin intervention[37,38].

Overall, at the phylum level, the main shifts by metformin in patients with T2DM are reflected in the increase of Bacteroidetes, Actinobacteria, and Proteobacteria with a corresponding decrease of Firmicutes and Verrucomicrobia. At the genus level, the increased bacteria were mainly Bacteroides, Streptococcus, Collinsella, Escherichia, Clostridium, and Subdoligranulum, while the decreased bacteria were Faecalibacterium and Ruminococcus[39,40]. Finally, through a comparative analysis of currently available data, we found that the alteration of gut microbiota involved in the inhibition of CRC by metformin in T2DM may be mainly the increase of Firmicutes and decrease of Bacteroidetes, Fusobacteria, and Bacteroidetes at the phyla levels. At the genus level, it is mainly manifested as the increase of Bifidobacterium and decrease of Fusobacterium (F.) nucleatum.

In addition, Bauer et al[41], Sun et al[42], and Pryor et al[43] confirmed that metformin improved metabolic dysfunction, such as hyperglycemia, via the glucose-SGLT1-sensing glucoregulatory pathway, B. fragilis–GUDCA–intestinal FXR axis, and the model of host-microbe-drug-nutrient interactions, respectively. These three new findings are profound to help understand how metformin and gut bacteria together affect disease progression. Table 3 lists their details and the involved types of gut microbiota. Not only that, with the participation of gut microbiota, other fields have gradually revealed the role of metformin, including anti-inflammatory effects, downregulation of interleukin expression, and ameliorating polycystic ovary syndrome in an animal model[44]. Moreover, metformin accelerated fatty acid oxidation by regulating host metabolism and longevity under the action of regulating a multipressure metabolic system[43].

Therefore, it can be confidently said that the gut microbiota, such as Akkermansia, B. fragilis and E. coli, mediate the pharmacological effects of metformin in different fields separately.

At present, many studies have confirmed that gut microbiota affect the occurrence and development of CRC. In humans, Yachida et al[2] used metagenomics analysis techniques to analyze the changes of gut microbiota in different stages of CRC on samples from a large cohort of 616 participants. They found the relative abundance of Fusobacterium nucleatum spp. was significantly elevated continuously from intramucosal carcinoma to more advanced stages. In addition, Atopobium parvulum and Actinomyces odontolyticus, which co-occurred in intramucosal carcinomas, were significantly increased only in multiple polypoid adenomas and/or intramucosal carcinomas. Moreover, Yu et al[45] has shown that F. nucleatum was abundant in CRC tissues in patients with recurrence post chemotherapy. F. nucleatum promoted CRC resistance to chemotherapy by targeting TLR4 and MYD88 innate immune signaling and specific microRNAs to activate the autophagy pathway. By 16S rRNA gene sequencing of stool samples in taxon-based analysis, stool of conventional adenoma patients was depleted in a network of Clostridia operational taxonomic units from families Ruminococcaceae, Clostridiaceae, and Lachnospiraceae and enriched in the classes Bacilli and Gammaproteobacteria, order Enterobacteriales, and genera Actinomyces and Streptococcus[46].

In animal models through fecal bacteria transplantation, the intestinal microbiota of CRC patients promoted the progression of intestinal adenomas in Apc^min/+ mice[47]. Donohoe et al[48] proved that dietary fiber protects against colorectal tumorigenesis in a microbiota- and butyrate-dependent manner by a gnotobiotic mouse model. Zhu et al[49] found that editing of the gut microbiota reduced carcinogenesis in colitis-associated CRC. Qin et al[50] suggested that the gut microbiome, under specific dietary exposures, stimulates a reprogramming of the enhancer landscape in the colon with downstream effects on transcription factors, which may be associated with CRC development. In addition, the immune system plays an important role in the effect of gut microbiota on CRC. Cremonesi et al[51] demonstrated that gut microbiota modulated T cell trafficking in human CRC. Wang et al[52] has shown that a purified membrane protein from Akkermansia muciniphila or pasteurized bacterium blunts colitis-associated tumorigenesis by modulation of CD8+ T cells. Meanwhile, Long et al[53] has also shown that Peptostreptococcus anaerobius promoted the occurrence of CRC and regulated tumor-related immunity. At the cellular level, Belcheva et al[54] demonstrated that gut microbial metabolism drives transformation of MSH2-deficient colon epithelial cells.

To sum up, in both humans and animals, gut microbiota is closely related to CRC, and this relationship refers to the immune system. In humans, studies have elaborated changes in the types of gut microbiota, and animal and cell experiments have further elaborated the possible mechanisms by which gut microbiota affects CRC. In Wong et al[55]’s review, the gut microbiota is thought to influence colorectal carcinogenesis via microbial-derived factors such as metabolites or genotoxins, promotion of cancer, or proliferating as opportunistic microorganisms in the tumor-associated microenvironment and the activation of procarcinogenic signaling pathways.

As we described before, the changes in the gut microbiota are inconsistent at different stages of development of CRC. According to the interaction between metformin and gut microbiota, does metformin mediate the changes of different gut microbiota at different stages of the development of CRC?

Based on the eighth Union for International Cancer Control Tumor Node Metastasis Classification of Malignant Tumors, at the phylum level in stage 0/pTis, the enriched bacteria are mainly Actinobacteria accompanied by a certain amount of Proteobacteria, Firmicutes, and Fusobacteria. In stage SI/II, the bacteria that are enriched are Proteobacteri, Firmicutes, Fusobacteria, and Bacteroidetes, and they show roughly the same abundance. In the stage SIII/IV, the enrichment of Firmicutes and Bacteroidetes is more obvious. Proteobacteri, Actinobacteria, and Fusobacteria are also present, but their abundance is not very high and is roughly equal[2].

In a nonblinded, one-armed intervention study, the relative abundance of 11 bacterial genera significantly changed during metformin intervention but returned to baseline levels after treatment cessation[32], which demonstrated that the effects of metformin on the gut microbiota are transient. However, for patients with T2DM, this condition is nothing to worry about because they require long-term medication. Moreover, there was a significant decrease of the phylum Firmicutes and of the ratio of Firmicutes to Bacteroidetes after taking metformin[39,56]. The other result of the microbiome analyses indicated a shift in the bacterial distribution in fish exposed to metformin, leading to an increase of Proteobacteria and a reduction of Firmicutes and Actinobacteria[57]. Compared with the previous data, metformin use from intramucosal cancer to advanced stage is likely to mediate its anticancer effect by provoking different gut microbiota. In stage 0/pTis, Actinobacteria may be the main target of the action of metformin, which is specifically manifested by attenuating its abundance. In stage SI/II and SIII/IV, the decrease of the phylum Firmicutes and of the ratio of Firmicutes to Bacteroidetes may be the main function of metformin. Of course, these processes may also be accompanied by the regulation of other flora. Table 3 details the changes. Therefore, the use of metformin to interfere with different types of gut microbiota at different stages of the development of CRC may be a means to improve the survival rate and prognosis of CRC in the future.

Patients with stage III colon cancer have a risk of recurrence ranging between 15% and 50%[58]. Interestingly, the administration of metformin is associated with decreased recurrence of carcinoma[59], covering the colorectal adenoma (hazard ratio 0.572, 95%CI 0.385-0.852) in diabetic patients with previous colorectal adenoma[60].

So, how does the gut microbiota influence CRC recurrence? A recent study showed that higher fear of cancer recurrence was associated with lower relative abundance of Firmicutes and higher relative abundance of Bacteroidetes at the phylum level and higher relative abundance of Bacteroides and lower relative abundance of Lachnospiraceae and Ruminococcus at the genus level. The chemotherapy-induced changes in gut microbiota influence the fear of cancer recurrence[61]. In addition, colorectal adenoma resection gave rise to a significant increase of Parabacteroides postoperatively. The microbiota signature of Parabacteroides, Streptococcus, and Ruminococcus showed an optimal discriminating performance of postoperative status[62]. These changes suggested that gut microbiota may be used in the future to prevent postoperative recurrence of CRC. Finally, it has been stressed that the intervention of metformin can directly affect the gut microbiota profile. In other words, metformin may inhibit the recurrence of CRC by changing the abundance structure of the corresponding gut microbiota.

The recurrence of CRC is resistant to routine chemotherapy and thought to be due to the enrichment of cancer stem cells. It has been reported that long term treatment of metformin impedes development of chemoresistance by regulating cancer stem cell differentiation in cancer cells[63]. Combination metformin with chemotherapeutics showed an overall modest but intriguing activity in patients with refractory CRC[64] and a synergistic inhibitory cytotoxicity in human CRC cells[65]. Moreover, metformin can act as an alternative radiosensitizing agent to 5-fluorouracil during neoadjuvant treatment for rectal cancer[66]. In addition, a large number of studies have confirmed that metformin, as a cancer stem cell-targeting agent in ovarian cancer, has a direct inhibitory effect on cancer stem cells[63,67]. The metformin intervention reduced expression of cancer stem cell markers and stemness-related genes in primary oral cancer cells[68].

Interestingly, microbiota also have a substantial role in the effectiveness of chemotherapy, chemoresistance, and related side effects[69]. For instance, the presence of specific bacterial species such as Slackia and Blautia obeum may act as microbial markers associated with drug resistance monitoring[70]. The inhibition of the growth of F. nucleatum significantly augments the efficiency of first-line chemotherapy treatments of CRC[71]. In addition, chemotherapy-induced changes in gut microbiota impact chemotherapy via influencing fear of cancer recurrence[61]. Moreover, in terms of cancer stem cells, gut microbiota regulate tumor metastasis via impacting circular RNA expression to regulate levels of corresponding miRNAs[72].

All in all, in the recurrence of CRC due to both the chemoresistance and the enrichment of cancer stem cells the gut microbiota plays a vital role, which may be one of the reasons why metformin can inhibit the recurrence of CRC in T2DM patients.

Metformin, known as a “magic drug”, reduces the incidence of CRC in people with T2DM. Moreover, the gut microbiota has been shown to mediate the pharmacological effects of metformin and the progression of CRC. Finally, the shift of gut microbiota has similarity in the occurrence and development of T2DM and CRC. Therefore, a network of metformin-gut microbiota-CRC (in type 2 diabetes) axis could be assumed. In the network, the gut microbiota plays a key role as a bridge that affects not only the pharmacological effects of metformin but also the occurrence of CRC in T2DM patients.

We have hypothesized the potential mechanism of action of this axis around the gut microbiota. On the one hand, these mechanisms will pave the way for us to understand how gut microbiota can participate in drug therapy. On the other hand, they will provide ideas for the prospective clinical application and transformation of gut microbiota. More importantly, this axis can be expanded into a profound network of gut microbiota-drug-disease to explain the crosstalk relationship between gut microbiota, drugs, and diseases.

Noticeably, before oral drugs are absorbed into our bloodstream, they must pass through our intestines for catabolism. In this process, the drugs are likely to first react with the intestinal microorganisms colonized in the intestinal epithelium, one of which is gut microbiota. Therefore, the drug may cause a cascade-like reaction related to gut microbiota by changing the composition and abundance of the gut microbiota. The drug may also be metabolized into other secondary products under the action of gut microbiota to exert subsequent efficacy. But the causation and deep interaction mechanism still remain exclusive. We have mentioned above that the gut microbiota involved in the inhibition of CRC by metformin in T2DM may be mainly Bacteroides, Ruminococcus, Clostridium, Firmicute, Lactobacillus, and E. coli. After these gut microbiota are changed, it may cause subsequent cancer suppression via regulation of inflammation and immunity and reduction of the production of genotoxic metabolites, etc.

Intestinal inflammation and related pathways are closely related to the initiation of CRC. In terms of regulating inflammation, after the mice were treated with metformin, the level of inflammatory markers TNF-α, IL-6, and IL-17α in plasma was reduced with a corresponding increase of the level of IL-10, which was accompanied by a reduction in Helicobacter pylori. Lee et al[44] suggested that fecal microbiota transplantation using metformin-treated mouse fecal material can upregulate the expression of GLP-1 and pattern recognition receptors TLR1 and TLR4. It has been shown that CRC-enriched genotoxic polyketide synthase (pks) + E. coli, E. faecalis, and A. finegoldii and TLR2 and/or TLR4 pathway-related bacteria, such as F. nucleatum and Peptostreptococcus anaerobius, are related closely to intestinal inflammation[55]. In our previous discussion, these bacteria are likely to be the targets of metformin, which provides a theoretical basis for our conjecture.

In terms of innate immune response, it has been proven that metformin specifically affects the accumulation of IFN-γ producing NK cells, IL-4, and IL-17 producing NKT cells and IL-12 producing macrophages/dendritic cells in the intestine in animal models[73]. In addition, the latest studies have shown that metformin could upregulate protein molecules involved in the classical immune pathway such as PMK-1/p38MAPK[74], RAGE ligands[75], and TLRs[76]. It should be mentioned that this process is likely to be mediated by gut microbiota. F. nucleatum has been shown to activate autophagy via the TLR4 and MYD88 pathways closely related to innate immunity[45].

In terms of adaptive immune response, it has been mentioned that metformin regulates adaptive immune cell infiltration[77,78], the expression of related immune factors, such as IFN-γ[79], IL-2[80], and IL-6[23,81,82], and T cell metabolic reprogramming in tumor tissues[83]. Nevertheless, it must be pointed out that these effects may also be manifested in the presence of gut microbiota. Cremonesi et al[51] proved that gut microbiota modulated T cell trafficking into human CRC including CCL5, CXCL9, and CXCL10 for cytotoxic T lymphocytes and T helper (Th) 1 cells, CCL17, CCL22, and CXCL12 for Th1 and regulatory T cells, CXCl3 for follicular Th cells, and CCL20 and CCL17 for IL-17-producing Th cells. Sethi et al[84] showed that gut microbiota depletion significantly reduced tumor burden in subcutaneous and liver metastasis models of pancreatic cancer, colon cancer, and melanoma. However, the effect was counteracted in Rag1-knockout mice that lacked mature T and B cells. In fact, gut microbiota is likely more complicated in the mechanisms of drug-mediated immunity affecting the development of CRC. A lot of work needs to be done to supplement this field.

In addition, it should be mentioned that after metformin changes the abundance of gut microbiota, the metabolites of the gut microbiota such as carcinogenic-related metabolites and genotoxicity will also change accordingly. Some of these metabolites are cancer-promoting, such as products of protein fermentation, hydrogen sulfide, bile acid metabolism, and ethanol, but some show protective effects. For example, short-chain fatty acids (butyrate and propionate) are anti-inflammatory molecules. Butyrate can inhibit histone deacetylase in colon epithelial cells and immune cells to downregulate proinflammatory cytokines and induce apoptosis in CRC cell lines. Therefore, the reduction of gut microbiota that produces procarcinogenic metabolites and the increase of gut microbiota that produces anti-inflammatory metabolites are likely to be the reason of metformin’s action. Similarly, metformin can directly slow down the abundance of E. coli, restraining to the intestinal bacteria’s carcinogenic effects. On the other hand, metformin may also promote some floras that are beneficial for anticancer effects to become a dominant flora. This is consistent with the changes in gut microbiota caused by the metformin intervention mentioned in Table 3.

Polyamines are a class of compounds containing two or more amino groups. The raw materials for their synthesis are ornithine and arginine. The key enzymes are ornithine decarboxylase and arginine decarboxylase. It has been proven that they can regulate cell proliferation and apoptosis, and their biosynthesis is closely related to the formation and metastasis of cancer[85]. There is a correlation between the expression of ornithine decarboxylase and the prediction of cancer risk and treatment response in certain epithelial cancers. However, it must be pointed out that the function of the polyamine system is affected by metformin in the presence of several pathogens, including Shigella flexneri, Streptococcus pneumoniae, Salmonella enterica subsp., enterica serovar Typhimurium, and Helicobacter pylori[86]. Therefore, metformin may inhibit the development of CRC by regulating the polyamine system under the action of certain gut microbiota.

Agmatine, a biogenic amine, is a polyamine that is converted from arginine by the action of arginine decarboxylase on the mitochondrial membrane of cells, and it has vasomotor implication and regulation of anti-inflammation[87]. Interestingly, the ability of bacteria to produce agmatine was enhanced under the administration of metformin in T2DM in a nutrient-dependent manner[43]. In addition, it must be emphasized that agmatine has a directly inhibitory effect on the proliferation of tumor cells with the corresponding acceleration of apoptosis[88]. As for CRC, agmatine has been confirmed to inhibit the proliferation of six types of human intestinal tumor cell lines in a concentration-dependent manner[89]. It is able to arrest proliferation in cell lines by depleting intracellular polyamine levels and enter mammalian cells via the polyamine transport system.

To date, there are few reports on the use of agmatine metabolism as a direct therapeutic target in clinical research, but there are studies affirming that the agmatine metabolism system has the potential to treat tumors. The role of metformin and gut microbiota is like a fuse in the antitumor effect mediated by the agmatine metabolic system. Moreover, Pryor et al[43] suggested that the Escherichia, Bacteroides, Enterobacter, and Citrobacter genera produce the most agmatine in the intestine, and these bacteria were abundant in the gut of patients after metformin intervention. Furthermore, this inhibitory effect of agmatine on tumors was probably attributable to an interaction between agmatine and the intracellular polyamine metabolism system[90,91]. For example, agmatine regulates the functional gene delivery system of spermidine/spermine acetyltransferase in cancer cell[90]. Therefore, the exploration of a metformin-gut microbiota-cell polyamine metabolism system-cancer chain helps us to understand how metformin inhibits CRC by regulating the agmatine metabolism system in the presence of gut microbiota. From the perspective of drugs-gut microbiota-pharmacological effects-diseases, this will be an interesting and profound finding.

To sum up, metformin may indirectly restrain clinical colorectal tumors by regulating the agmatine metabolism system in the presence of gut microbiota. In future studies, more attention should be attached to the drug’s interaction with gut microbiota to the intracellular polyamine system because the intervention of the polyamine system may be one of the important methods for clinical cancerous treatment.