Introduction
For decades, the biguanide, metformin, has been the first-line oral glucose-lowering drug of choice when treating type 2 diabetes. The antihyperglycaemic effects of metformin include suppression of hepatic gluconeogenesis and increased glucose uptake in skeletal muscle tissue mediated by activation of adenosine monophosphate-activated protein kinase (AMPK) [
1‐
3]. Some studies also suggest the involvement of non-AMPK-mediated suppression of hepatic gluconeogenesis [
4,
5]. However, evidence suggests that the gastrointestinal tract and the microorganisms that reside within it are partly involved in mediating both beneficial and adverse effects of metformin. Intraluminal metformin concentration in the gastrointestinal tract is up to 300 times higher than in plasma [
6,
7], and metformin treatment decreases intestinal glucose absorption [
8]. Furthermore, AMPK activation in duodenal epithelium lowers the plasma glucose concentration in rats [
9]. Individuals with type 2 diabetes have an altered composition of the gut microbiota [
10‐
14], and part of the aberration of the microbiota is linked to metformin treatment [
14‐
18]. The gut microbiota is involved in intestinal bile acid metabolism and short-chain fatty acid production, which could explain some of the glucose-lowering effect of metformin, through effects on incretin secretion, hepatic glucose homeostasis and beta cell function [
19,
20]. In support of this, data from 22 patients with type 2 diabetes, treated with metformin for 3 days, showed a
Bacteroides fragilis-linked increase in the intestinal concentration of glycoursodeoxycholic bile acid, which in rodent studies has been associated with improvement of hyperglycaemia [
15]. Furthermore, the faecal concentration of the short-chain fatty acid propionate was increased in 15 metformin users compared with nine non-users [
21]. Similarly, during a 4 month intervention, faecal butyrate and propionate concentrations were increased significantly in men treated with metformin compared with placebo [
18], demonstrating an effect of metformin on fermentative metabolites involved in regulating human metabolism. A hyperglycaemia-independent effect of metformin on the induced perturbation of the gut microbiota has recently been shown after 1 week of metformin intake in 18 healthy individuals [
16], and elucidation hereon could further our understanding of the link between type 2 diabetes and the gut microbiota.
The most common side effect of metformin treatment is gastrointestinal discomfort, including nausea, bloating, flatulence and diarrhoea [
22,
23]. The mechanisms responsible for these adverse effects are poorly understood but a role of the gut microbiota as a potential mediator has been proposed [
14]. Whether the side effects commonly associated with metformin treatment arise from changes in the gut microbiota needs further exploration.
The primary objective of the present intervention in young, healthy men was to investigate compositional changes of the gut microbiota following metformin intake, independent of the physiological changes induced by the diabetic state. In post hoc analyses, we aimed to explore whether the pre-intervention gut microbiota profile was related to self-reported gastrointestinal discomfort following metformin intervention.
Methods
Faecal samples and data processing
Participants were instructed to deliver nine faecal samples throughout the study period. Samples were collected on the day of examination if possible and immediately frozen either at the study site or at home, in which case samples were transferred to the laboratory on dry ice. Samples were stored at −80°C until DNA extraction.
A total of 206 faecal samples were collected but two samples did not contain sufficient material for DNA extraction. Genomic DNA was isolated, followed by PCR amplification of the V4 hypervariable region of the bacterial 16S rRNA gene and sequencing on an Illumina MiSeq platform generating a total of 9,466,021 (mean 43,622 [SD 18,700]) paired-end (250 bp) reads. See
ESM Methods for detailed information. Twelve samples that failed during the first run were resequenced. Processing of raw sequencing data was performed using the dada2 (v1.4.0) R package [
26]. Following inspection of quality profiles, denoising, merging of reads and removal of chimeric sequences, a total of 7,214,117 reads (mean 34,702 [SD 12,889]; minimum 10,736) in 1764 unique amplicon sequence variants (ASVs) were available for downstream analyses. Taxonomical assignment of ASVs from kingdom to species was performed against the Silva v128 database, using the dada2 implementation of the naive Bayesian RDP classifier [
27].
Statistical analyses
The primary outcome was compositional change in abundance of ASVs agglomerated at the taxonomic level of genus. Pre-specified secondary outcomes were as follows: compositional change in abundance of non-agglomerated ASVs; change in intra-individual diversity quantified by the number of observed ASVs (richness), Shannon’s entropy and Pielou’s evenness (Shannon’s entropy/loge[richness]); change in community structure and membership quantified by Canberra and Jaccard distances, respectively, and change in anthropometric and biochemical traits. Post hoc analyses included random forest (RF) classification of individuals developing gastrointestinal discomfort and testing for bacterial genera and ASVs responding differently to metformin in these participants, compared with those who did not develop gastrointestinal side effects. Data from 24 participants examined at least once during the intervention period were included in these subanalyses. Data from participants not following protocol (e.g. reduced metformin intake, n = 2) were omitted from analyses from the time they deviated from protocol. In total, 25 participants completed visit 1 and were included in the statistical analyses, but one participant dropped out before initialisation of the intervention.
ASV abundances were agglomerated at genus level using the phyloseq R package, normalised using total sum scaling, and log transformed following addition of a pseudocount corresponding to the lowest non-zero normalised abundance of each taxon. Intervention effect was modelled by a one-way repeated measures ANOVA using mixed linear regression as implemented in the lme4 R package. Models were fitted using restricted maximum likelihood using samples from all time points. The abundance of each genus at each time point during the metformin intervention was compared by a post hoc t test with the mean abundance averaged over the three samples collected during the pre-intervention period. Genera never present in >80% of participants were excluded from analyses. Intervention effect on non-agglomerated ASV abundances, bacterial richness, Shannon’s entropy and Pielou’s evenness was assessed using the same approach considering a two-tailed p value <0.05 as significant. Correction for false discovery rate was done for all genera/ASVs across all time points by the Benjamini–Hochberg method, applying a false discovery rate of 10% for significance.
Intervention effect on community structure and community membership was assessed by permutational multivariate ANOVA (PERMANOVA; vegan R Package v2.4.6) of Canberra and Jaccard distances, contrasting each intervention and post-intervention time point to the average across the three pre-intervention time points. Principal coordinate ordination was performed using the capscale function of the vegan package, specifying an unconstrained model.
RF classification was used to identify bacterial genera that at baseline were discriminative for the development of gastrointestinal adverse effects during the intervention. RF models were trained on total sum scaled abundances of 124 bacterial genera using the caret (v6.0.79) and randomForest (v4.6.12) R packages with 25 iterations of bootstrap resampling. Down-sampling during resampling was applied to address class imbalance. Performance across resamples was evaluated by the area under the receiver operating characteristic (ROC) curve and discriminant genera were ranked by the mean decrease in accuracy. Using the optimal variable settings from the naive RF classifier (mtry = 2), we applied the Boruta (v5.2.0) R package for feature selection of all-relevant discriminant genera identified across 200 repetitions using different random seeds. Selected genera were subsequently used to build an optimal RF classifier, as described above.
We used mixed linear regression to identify bacterial genera and ASVs responding differently to the metformin intervention in participants who developed gastrointestinal side effects compared with those who did not. A response profile model was specified with a main effect of time (categorical) and a time × group interaction as fixed effects and a random intercept for each participant, thereby testing the difference between groups at all time points during the intervention. Data were corrected for multiplicity as described above.
Change in clinical and biochemical traits was assessed using linear mixed model regression ANOVA as outlined above. Logarithmic transformation of the dependent variable was applied if deemed appropriate upon inspection of residual plots and normal probability plots. For VAS data, a constant of 1 was added prior to transformation. Difference in plasma metformin between visit 3 and 4 were tested with a Wilcoxon signed-rank test.
Discussion
We show that metformin intervention has an impact on the composition of the human gut microbiota at genus and ASV levels in healthy young men, and these changes are reversed after discontinuation of metformin. In post hoc, secondary aim analyses, we demonstrate that the pre-treatment composition of a subset of bacterial genera may predict risk of development of gastrointestinal adverse effects to metformin.
Until now, several studies have recognised that metformin treatment associates with a structural change of the gut microbial community. A double-blinded randomised study [
18] of 40 treatment-naive patients with type 2 diabetes given placebo or metformin for 4 months showed an increase in abundance of
Escherichia spp. and
Bilophila wadsworthia along with a decrease in
Intestinibacter spp. and
Clostridium spp. Similarly, using shotgun sequencing-based metagenomic analyses we previously reported an increase of
Escherichia spp. and a reduced abundance of
Intestinibacter spp. in metformin-treated type 2 diabetes patients [
14]. In a recent study of 18 healthy participants given 850 mg metformin twice daily for 1 week an increase of
Escherichia/Shigella was also reported [
16]. Our analyses of healthy normoglycaemic individuals similarly showed a reduced abundance of
Intestinibacter spp. and
Clostridium spp., as well as an increased abundance of the genus
Escherichia/Shigella and
B. wadsworthia in response to metformin treatment. Collectively, the findings substantiate that the effect of metformin is independent of the dysbiosis induced by diabetes. Additionally, we identified seven genera changing in abundance during the metformin intervention, further demonstrating that metformin treatment has a profound impact on the gut microbiota.
Consistent correlations between alterations of gut microbiota composition and metformin intake, along with evidence that the intended effect of metformin is possibly co-mediated by the gut, suggest that alterations of the gut microbiota contribute to the therapeutic effect of metformin. However, long-term prospective studies of diabetic patients treated with metformin could determine causality. Identifying such a causality could be the first step in performing bacteria-based interventions in type 2 diabetes patients.
Adverse gastrointestinal effects are the primary cause of failing metformin compliance. In a recent study of 18 healthy young men exposed to metformin for 1 week, an association between increased abundance of
Escherichia/
Shigella spp. prior to the intervention and later development of gastrointestinal side effects was reported [
16]. However, no formal statistical evidence was provided to substantiate this finding. In the present study, we aimed to identify genera associated with development of gastrointestinal adverse effects. The composition of 12 bacterial genera of the pre-intervention gut microbiota was identified as a possible predictor of self-reported gastrointestinal adverse effects during metformin treatment. Among these, the highly abundant genera
Sutterella,
Akkermansia and
Allisonella had the greatest predictive value [
28]. Interestingly,
Sutterella spp. have been associated with infections of the gastrointestinal tract, inflammatory bowel disease [
29] and autism [
30].
Akkermansia muciniphila is one of the most abundant bacterial species of the human gut microbiota and specialises in mucin degradation. Using the glycosylated proteins of the epithelial mucus layer as its major source of carbon and nitrogen in fermentative processes producing the short-chain fatty acids acetate and propionate,
A. muciniphila strengthens the integrity of the intestinal epithelium and regulates the gut barrier function [
31]. In studies in animals [
32] and humans [
17,
33], metformin treatment was associated with increased abundance of
A. muciniphila and the bacterium has been linked to improved glycaemic control [
34]. We did not identify
A. muciniphila among the metformin-responsive bacteria in the present study, perhaps because metformin treatment in individuals with diabetes resets a perturbation in
A. muciniphila abundance caused by the disease; an imbalance that is not present in healthy individuals.
The genus
Allisonella, and the only known species
A. histaminiformans, produces histamine from histidine [
35]. Histamine is a potent vasoactive agent causing vasodilatation and increased vascular permeability, as well as a potent inducer of mucus secretion. Histamine is capable of inducing stomach ache, cramps, meteorism and diarrhoea, which are all known gastrointestinal side effects of metformin treatment [
36]. Interestingly, in vitro studies have shown that metformin inhibits degradation of histamine by diamine oxidase at concentrations achieved in the intestine after therapeutic doses [
37]. Whether metformin affects histamine production by the gut microbiota, how that may interact with effects on host capacity for histamine degradation to induce gastrointestinal intolerance and how the intolerance might be prevented or treated require testing in future studies. Yet, our findings of gastrointestinal side effects associated with pre-treatment bacterial composition must be interpreted with caution as our dataset is limited by its size and lack of validation in an independent cohort. Furthermore, there is a risk of overestimation due to small sample size and the large number of features in the model. Despite this major limitation, our data generate a hypothesis for future studies aiming to limit gastrointestinal side effects in patients introduced to metformin by modulating intestinal bacterial composition before medication is given.
A strength of the study is that multiple faecal samples were collected prior to the intervention (three samples over 6 weeks), enabling us to account for background fluctuations in gut microbiota composition, thereby improving the reliability of intra-individual modelling. Longitudinal sampling during and after the intervention also demonstrates the dynamics of the gut microbiota in response to the initiation and cessation of treatment. The design is, however, underpowered to test inter-individual effects, which is presumably why we were unable to identify bacterial genera or ASVs responding differently to the metformin intervention in participants who developed or did not develop gastrointestinal side effects. Much larger sample sizes are required to reliably identify interventional effects in between-group analyses [
38]. Other limitations include the lack of blinding and inclusion of a placebo control group. Technologically, the study was limited by usage of the 16S rRNA gene amplicon sequencing approach, which is well suited for detecting signals at the genus level but provides limited insight at the species level.
In conclusion, the blood glucose-lowering biguanide metformin changes the composition of the intestinal microbiota independent of the prevailing blood glucose level, showing that the effect is independent of the dysbiosis induced by diabetes. We propose that the pre-treatment composition of a subset of bacterial genera in the gut may predict risk of gastrointestinal side effects, hinting at the potential involvement of bacterial fermentation, gut barrier function and histamine in metformin intolerance.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.