Gluco-Metabolic Effects of Pharmacotherapy-Induced Modulation of Bile Acid Physiology Free

Abstract

Context

The discovery and characterization of the bile acid specific receptors farnesoid X receptor (FXR) and Takeda G protein-coupled receptor 5 (TGR5) have facilitated a wealth of research focusing on the link between bile acid physiology and glucose metabolism. Modulation of FXR and TGR5 activation have been demonstrated to affect the secretion of glucagon-like peptide 1, insulin, and glucagon as well as energy expenditure and gut microbiota composition, with potential beneficial effects on glucose metabolism.

Evidence Acquisition

A search strategy based on literature searches in on PubMed with various combinations of the key words FXR, TGR5, agonist, apical sodium-dependent bile acid transporter (ASBT), bile acid sequestrant, metformin, and glucose metabolism has been applied to obtain material for the present review. Furthermore, manual searches including scanning of reference lists in relevant papers and conference proceedings have been performed.

Evidence Synthesis

This review provides an outline of the link between bile acid and glucose metabolism, with a special focus on the gluco-metabolic impact of treatment modalities with modulating effects on bile acid physiology; including FXR agonists, TGR5 agonists, ASBT inhibitors, bile acid sequestrants, and metformin.

Conclusions

Any potential beneficial gluco-metabolic effects of FXR agonists remain to be established, whereas the clinical relevance of TGR5-based treatment modalities seems limited because of substantial safety concerns of TGR5 agonists observed in animal models. The glucose-lowering effects of ASBT inhibitors, bile acid sequestrants, and metformin are at least partly mediated by modulation of bile acid circulation, which might allow an optimization of these bile acid–modulating treatment modalities. (J Clin Endocrinol Metab 106: 362–373, 2020)

The discovery and characterization of the bile acid specific receptors farnesoid X receptor (FXR) (NR1H4) in 1999 (1–3) and the Takeda G protein-coupled receptor 5 (TGR5) (GPBAR1) in 2002 (4) have facilitated a wealth of research focusing on bile acids as gluco-metabolic integrators. Thus, several studies published during the previous decade have demonstrated bile acid–mediated activation of FXR and TGR5 to elicit modulation of important gluco-metabolic features such as secretion of glucagon-like peptide 1 (GLP-1) (5, 6), insulin, and glucagon (7, 8) as well as energy expenditure (9, 10) and gut microbiota composition (11), with potential impact on glucose metabolism. In addition, bile acids have been shown to affect both plasma glucose concentrations and GLP-1 secretion following oral, jejunal, or rectal administration of bile acids in humans (12–14).

A range of pharmaceutical compounds with effects on bile acid metabolism has been investigated in terms of implications for glucose metabolism. Within recent years, a variety of agonists for the FXR and TGR5, respectively, have been tested in both animal models and human settings. In addition, the concept of drug-mediated inhibition of the apical sodium-dependent bile acid transporter (ASBT) (SLC10A2) has been investigated and the specific ASBT inhibitor GSK672 is currently undergoing phase 2 evaluation for the treatment of type 2 diabetes. Interestingly, metformin, constituting the well-established first-line pharmacotherapy in type 2 diabetes, has been demonstrated to reduce the ASBT-mediated reabsorption of bile acids in the terminal ileum, which could contribute to the glucose-lowering effect of this drug (15, 16). In addition, the bile acid sequestrant colesevelam has been approved for the treatment of type 2 diabetes in the United States since 2008. This drug prompts a diversion of bile acids from the enterohepatic circulation resulting from complex binding and subsequent entrapment of bile acids in the gastrointestinal tract (17).

Attainment of glycemic control in patients with type 2 diabetes continues to represent a considerable challenge for both the individual patient and health care systems worldwide (18). An estimated number of nearly 400 million people worldwide have type 2 diabetes and, each year, around 5 million deaths are caused by diabetes (19). Thus, patients with type 2 diabetes have an estimated 10-year reduced life span expectancy, which is mainly a result of a deteriorated cardiovascular risk profile (20). Consequently, a continuous focus on optimization of type 2 diabetes treatment is very much needed (21).

The aim of the present review is to provide an overview on the link between bile acid physiology and glucose metabolism with a special focus on bile acid–modulating treatment modalities with effects on plasma glucose concentrations in humans.

Bile Acid Physiology

The enterohepatic circulation of bile acids

Bile acids are synthesized from cholesterol in the liver by 2 distinct pathways. The primary (classical or neutral) pathway constitutes more than 90% of the de novo synthesis of bile acids and takes place in the endoplasmic reticulum of hepatocytes (22, 23), whereas the secondary (alternative or acidic) pathway is initiated in the mitochondria (24). The cholesterol 7 alpha-hydroxylase–mediated 7α-hydroxylation of cholesterol represents the rate-limiting step in the classical pathway toward the synthesis of the primary bile acids cholic acid and chenodeoxycholic acid (25). The primary bile acids are conjugated with either taurine or glycine within the hepatocyte before being actively secreted to the bile canaliculi (23, 26), with continuous drainage into larger and larger ducts before exiting the liver through the common hepatic duct (27). It has been reported that 50% of the bile acids secreted from the liver bypasses storage in the gallbladder during fasting conditions. However, repeated enterohepatic cycling of these released bile acids terminally secures storage and up-concentration of bile in the gallbladder during overnight fasting (28, 29). Cholecystokinin (CCK) is secreted by enteroendocrine I cells in the epithelium of the proximal small intestine following ingestion of protein and particularly fat (30), thereby eliciting release of bile acids to the duodenum as a result of CCK-induced contraction of gallbladder smooth muscle cells (31, 32). The intestinal reabsorption of bile acids is mainly carried out by active ASBT-facilitated transport in the terminal ileum with only a minor contribution from passive uptake along the more proximal part of the small intestine (33–35). Binding of bile acids to the ileal lipid-binding protein (FABP6) within enterocytes facilitates transfer through the cytosol and subsequent export to the portal vein mediated by the organic solute transporter OSTα-OSTβ (SLC51A/SLC51B) (36). The portal flow returns the bile acids to the liver with possible reconjugation in hepatocytes before secretion to the bile canaliculi and completion of the enterohepatic cycle (26). Under normal conditions, no more than 5% of the circulating bile acids escape reabsorption in the ileum and spill over to the colon (23). This transfer to the colon allows processing of the primary bile acids by the intestinal bacterial flora. These microbiota-derived modifications involve deconjugation of taurine and glycine-conjugated cholic and chenodeoxycholic acid with potential subsequent 7α-dehydroxylation to the secondary bile acids deoxycholic and lithocholic acid (26). The microbiota-mediated processing increases the hydrophobicity of the bile acids, which supports the passive absorption across the colonic epithelium and return to the liver for reuse in the enterohepatic circulation (26, 28). This enterohepatic cycle of bile acids repeats several times each day and only 500–600 mg/day are lost via fecal excretion, which is compensated by de novo synthesis in the liver to preserve a stable bile acid pool (23, 26).

Bile acid receptors and target tissues

A range of bile acid receptors and various physiological functions of bile acid–mediated receptor activation have been described in the literature. In particular, the discovery of the receptors FXR (1–3) and TGR5 (4) has instigated investigations of the potential link between bile acids as glucose metabolism. Additional bile acid receptors include the G protein-coupled sphingosine 1-phosphate receptor 2 (S1PR2) that has been reported to contribute to the regulation of hepatic lipid metabolism (37, 38), and the nuclear vitamin D receptor (NR1I1) (39) and pregnane X receptor (NR1I2) with potential beneficial bile acid–detoxifying properties (40). However, none of these receptors are specifically activated by bile acids, and, furthermore, the link between these receptors and glucose metabolism at this point remains speculative (41).

FXR

The nuclear FXR is mainly located in the liver and in enterocytes of the gastrointestinal tract, with the highest density in the distal part of the small intestine (42, 43). The structure of FXR is characterized by a ligand-binding domain, an activation domain for coregulation, and a DNA-binding domain for interaction with promoter sequences of target genes (44). FXR is a major regulator of human bile acid metabolism via diverse effects on transport and synthesis of bile acids as a result of intestinal and hepatic FXR activation, thereby playing an important role in avoiding potential bile acid toxicity within hepatocytes and enterocytes (41). Thus, intestinal FXR activation causes an inhibitory effect on the apical ASBT, whereas expression of the basolateral OSTα-OSTβ is increased (45, 46). In addition, an induced secretion of the hormone fibroblast growth-factor 19 (FGF-19) is evident following FXR activation in enterocytes (47). This hormone suppresses cholesterol 7 alpha-hydroxylase activity and the expression of basolateral bile acid transporters in the liver, which induce reduced de novo synthesis (48, 49) and hepatic uptake of bile acids (50). Hepatic FXR activation instigates induction of the excretory apical bile acid transporters and suppression of the basolateral transporters that facilitates bile acid uptake to the hepatocytes from the portal circulation (51).

The effects of FXR modulation in terms of glycemic control remain controversial. A study has demonstrated hepatic FXR activation to induce expression of the small heterodimer partner transcription factor (NR0B2), which instigates a decreased expression of phosphoenolpyruvate carboxykinase and glucose-6-phosphatase that are both key enzymes in gluconeogenesis (52). Along these lines, various animal models have reported beneficial gluco-metabolic effects of FXR activation in terms of repressed gluconeogenesis and increased hepatic glycogen synthesis (53–56). In addition, studies by Pathak et al. have demonstrated both systemic and intestine-restricted FXR activation to cause increased TGR5 signaling via induction of TGR5 expression and modulation of the gut microbiota, thereby eliciting improvements in GLP-1 secretion and hepatic insulin sensitivity (57, 58). Furthermore, the FXR-mediated increase in FGF-19 has been reported to elicit increased glycogen synthesis and reduced gluconeogenesis (49). On the other hand, a range of studies have demonstrated induction of GLP-1 secretion following suppressed FXR activation in L cells (6), decreased gluconeogenesis as a result of reduced circulating ceramide concentrations resulting from diminished intestinal FXR activation (59), and delayed intestinal glucose absorption in FXR^-/- mice (60). In addition, improved glucose homeostasis has been reported in FXR-deficient obese mice as well as after application of the FXR antagonist HS218 in a mouse model of type 2 diabetes (61, 62).

Likewise, the cardiovascular effects of FXR activation seem somewhat unresolved, with rodent models reporting FXR knockout to induce increased atherosclerosis in apolipoprotein E–deficient mice (63), whereas an atheroprotective effect of FXR deficiency was observed in low-density lipoprotein (LDL) receptor knockout mice (64).

TGR5

TGR5 is a bile acid–specific 7-transmembrane G protein–coupled receptor that has been localized to various tissues throughout the body, including the gastrointestinal tract, pancreas, liver, gallbladder, and adipose tissue (4). A range of animal models has reported bile acid–mediated TGR5 activation to suppress hepatic macrophage activation (65), induce gallbladder relaxation and refilling (66), and to promote intestinal motility (67). Also, administration of the TGR5 agonist INT-777 has been shown to diminish the formation of atherosclerotic lesions in LDL receptor knockout mice via TGR5 activation in macrophages and a subsequent reduction of proinflammatory cytokine production (68).

Interestingly, TGR5 activation has also been demonstrated to elicit several beneficial gluco-metabolic effects. Thomas et al. have demonstrated increased intracellular adenosine 5′-triphosphate/adenosine 5′-diphosphate ratio and cyclic adenosine 5′-monophosphate levels following TGR5 activation in L cells, thereby inducing GLP-1 secretion from these cells (5). The concept of TGR5-mediated GLP-1 secretion from intestinal L cells is substantiated by studies reporting increased GLP-1 concentrations in plasma following duodenal, colonic, or rectal administration of bile acids in human subjects (12, 14, 69, 70). Studies have demonstrated a basolateral localization of TGR5 in L cells, which necessitates intestinal bile acid absorption to obtain TGR5 activation in these cells (71, 72). In vitro studies performed by Kumar et al. have demonstrated induced insulin and suppressed glucagon secretion following TGR5 activation in pancreatic beta and alpha cells, respectively (7, 8). Thus, direct TGR5-mediated pancreatic effects with beneficial implications for glucose metabolism seem to be evident. Finally, TGR5 activation has been reported to increase energy expenditure via induction of the enzyme type 2 iodothyronine deiodinase in skeletal muscle and brown adipose tissue (9, 10).

Gluco-metabolic effects of pharmacotherapies modulating bile acid physiology

Pharmaceutical compounds with potential bile acid–mediated glucose-lowering effects include both bile acid receptor agonists and indirect modulators of bile acid circulation. These 2 quite different approaches for modulation of bile acid metabolism as well as the relevant specific treatment modalities will be outlined in the following section.

FXR agonists

As described previously, FXR is a key regulator of bile acid metabolism and transport. The specific FXR agonist obeticholic acid, a semisynthetic derivative of the primary bile acid chenodeoxycholic acid, was approved in 2016 for the treatment of primary biliary cholangitis (73, 74). Interestingly, obeticholic acid has been demonstrated to reduce steatosis in both rodent models (75, 76) and in patients with nonalcoholic fatty liver disease (76, 77). Clinical phase 2 and 3 studies are under way to further investigate the safety and efficacy of obeticholic acid in patients with nonalcoholic steatohepatitis (Randomized Global Phase 3 Study to Evaluate the Impact on NASH With Fibrosis of Obeticholic Acid Treatment [NCT02548351] and Combination Obeticholic Acid (OCA) and Statins for Monitoring of Lipids [NCT02633956]).

As described, the effects of FXR modulation in terms of glycemic control seem less clear with animal studies reporting conflicting results on this matter (53–56, 59–62). Furthermore, the 2 available human studies examining the glycemic effects of treatment with obeticholic acid have reported conflicting results in terms of gluco-metabolic effects. A randomized, controlled trial in patients with type 2 diabetes and nonalcoholic fatty liver disease examined the effects of treatment with 25 mg (n = 20) or 50 mg (n = 21) of obeticholic acid administered once daily compared with placebo (n = 23) (77). Insulin sensitivity was assessed by the hyperinsulinemic-euglycemic clamp method before and after 6 weeks of treatment. Only patients with successfully performed baseline and end-of-treatment clamps were included in the analysis (placebo [n = 17], 25 mg obeticholic acid [n = 15], 50 mg obeticholic acid [n = 12]) and no additional glycemic outcomes (e.g., basal plasma glucose concentration or HbA1c) were reported in this study. A statistically significant increase in glucose infusion rates of approximately 25% were observed for the 2 obeticholic acid treatment groups combined compared with placebo, which point to improved insulin sensitivity (77). However, a subsequent randomized, placebo-controlled trial including 283 patients with nonalcoholic steatohepatitis (50% had type 2 diabetes) demonstrated no placebo-corrected improvements in HbA1c or basal plasma glucose concentrations following 72 weeks of treatment with obeticholic acid 25 mg once daily. Treatment with obeticholic acid was shown to cause improved liver histology (decrease in nonalcoholic fatty liver disease activity score by at least 2 points without worsening of fibrosis) in 45% of patients compared with 21% of placebo-treated patients (relative risk, 1.9; 95% confidence interval [CI], 1.3-2.8). Nevertheless, a statistically significant placebo-corrected deterioration of insulin sensitivity as measured by the homeostatic model assessment of insulin resistance Repetition (HOMA-IR) was evident in patients treated with obeticholic acid (78). A post hoc analysis from this study examined the combined effects of weight loss and obeticholic acid on metabolic features, including basal plasma glucose, HbA1c, and HOMA-IR (79). The analysis included 102 participants treated with obeticholic acid and 98 treated with placebo that all had completed the study and undergone an end-of-treatment liver biopsy. These cohorts had similar baseline characteristics compared with the complete study population (n = 283). Interestingly, the post hoc analysis demonstrated treatment with obeticholic acid to reverse beneficial effects of weight reduction on basal plasma glucose, HbA1c, and HOMA-IR observed in placebo-treated group (79). In terms of adverse events, treatment with obeticholic acid seemed to be well tolerated with increased frequencies of only mild constipation and pruritus in the 2 trials, respectively (77, 78). However, it should be emphasized that obeticholic acid has been linked to cases of liver injury or death following treatment in patients with decompensated cirrhosis of the liver, which in 2017 gave rise to a US Food and Drug Administration black box warning on the use of obeticholic acid in this group of patients (80).

TGR5 agonists

The potential gluco-metabolic effects of TGR5 activation outlined here have prompted the development and pharmacodynamic testing of various specific TGR5 agonists. A range of these compounds has demonstrated glycemic improvements in rodent models (81, 82), but at the cost of significant adverse events including gallbladder dilation, pancreatitis, and hepatic necrosis (82–84).

A single randomized and placebo-controlled study including 51 patients with type 2 diabetes has examined the treatment effects of the selective TGR5 agonist, SB-756050, in humans (85). Treatment with this agonist in various dosing regimens throughout 6 days elicited no improvements in GLP-1 secretion or glycemic control compared with placebo. In fact, statistically significant increases in both basal and postprandial plasma glucose concentrations were observed following treatment with SB-756050 in the 2 lowest examined doses (15 and 50 mg once daily) compared with placebo (85). Intestinally targeted TGR5 agonists with no or low systemic exposure have been developed to overcome the considerable safety issues (82–84) associated with TGR5 activation in extraintestinal tissues observed in animal studies. However, no such compounds with beneficial gluco-metabolic effects and a favorable safety profile have been presented so far (86, 87), which might be due to the basolateral localization of TGR5 on L cells (71, 72).

ASBT inhibitors

The metabolic treatment potential of ASBT inhibition has been investigated during the previous 2 decades. In 2002, initial animal studies within this field reported treatment with ASBT inhibitors to cause reductions in LDL cholesterol (88, 89), which seem in line with the well-established compensatory increase in hepatic LDL receptor expression and cholesterol uptake because of increased fecal loss of bile acids (90). Along these lines, studies in guinea pigs and mice have suggested atheroprotective effects of ASBT inhibition (91, 92). In addition, studies applying rodent models of diabetes have demonstrated ASBT inhibition to elicit increased plasma GLP-1 and reduced plasma glucose concentrations (93, 94). One of these studies observed reduced hepatic and intestinal FXR activation following treatment with the potent ASBT inhibitor 264W94, but no deterioration of the glucose-lowering effect was evident following concomitant treatment with the FXR agonist GW4064 (94). Thus, the beneficial gluco-metabolic effects might be mediated by induced TGR5 activation and subsequent increased GLP-1 secretion resulting from increased amounts of free bile acids in the colon (94, 95). Various ASBT inhibitors have been developed and tested in animal models within recent years to optimize the pharmacokinetic and pharmacodynamic profiles of these compounds (95).

The highly potent, nonabsorbable, and specific ASBT inhibitor GSK672 is undergoing phase 2 evaluation for the treatment of type 2 diabetes and, until now, 2 studies have reported results on the glucose-lowering impact of this compound in humans (96). The first was a randomized, placebo-controlled crossover study examining the effects of GSK672 as an add-on to metformin in patients with type 2 diabetes (n = 15) and a baseline HbA1c (mean [standard deviation]) of 73.1 (9.5) mmol/mol. Seven days of treatment with GSK672 (titrated to 90 mg twice daily) caused a baseline-corrected reduction in weighted mean 24-hour plasma glucose of 1.93 mmol/L (95% CI, 0.82-3.04) compared with placebo (96). The predominant adverse event was diarrhea, with 11 participants reporting this side effect during treatment with GSK672 compared with 7 during treatment with placebo. Two subjects withdrew from the study because of diarrhea and cholecystitis, respectively (96). The second study was a randomized, controlled trial investigating the effect of 14 days of treatment with GSK672 in various dosing regimens as an add-on to metformin. Baseline HbA1c was around 65 mmol/mol in groups treated with GSK672 90 mg twice daily (n = 10) or the comparator placebo (n = 13) and sitagliptin 50 mg once daily (n = 13) (96). Treatment with GSK672 elicited placebo-corrected reductions from baseline in fasting plasma glucose of 1.21 mmol/L (95% CI, 0.28-2.14) and weighted mean 24-hour plasma glucose of 1.33 mmol/L (95% CI, 0.35-2.30). The observed glucose-lowering impact of GSK672 was similar to the effects seen after treatment with sitagliptin. Once again, diarrhea was the most frequent adverse event and reported by 6 participants in the GSK672 group compared with 4 and 2 reports in the groups treated with placebo and sitagliptin, respectively (96). Interestingly, both studies found reduced plasma insulin concentrations following treatment with GSK672, which point to an insulin-independent glucose-lowering mechanism of this drug. This observation seems contradictory to the previously outlined potential GLP-1–mediated effect of ASBT inhibition.

Bile acid sequestrants

The bile acid sequestrant colesevelam has been demonstrated to elicit improvements of glycemic control in patients with type 2 diabetes alongside reduction of plasma LDL cholesterol (97). As previously described, complex binding of bile acids to sequestrants causes a depletion of the bile acid pool accompanied by a compensatory increase in the hepatic LDL receptor expression and uptake of circulating cholesterol, which ultimately causes a reduction in plasma LDL cholesterol concentration (90). Contrastingly, the underlying glucose-lowering mechanisms of bile acid sequestration remain to be clarified. Various studies have addressed this subject by investigating the metabolic effects of bile acid sequestrant-mediated modulation of the enterohepatic circulation of bile acids. The most consistently proposed mechanisms involve increased GLP-1 secretion from L cells (98, 99) and increased splanchnic glucose utilization (100–102), but so far, no firm conclusions have been made on this matter. Previous studies have reported conflicting results with respect to bile acid sequestrant-mediated GLP-1 secretion (98, 99, 101–103). Sequestration of bile acids in the intestinal lumen should be expected to reduce the activation of both the nuclear FXR receptor and the G protein–coupled TGR5 in ileac L cells. Potthoff et al. have previously suggested TGR5 activation by bile acids complexed to colesevelam (104). However, studies by Ullmer et al. and Brighton et al. have demonstrated an exclusively basolateral localization of TGR5, which should exclude the possibility of in vivo interaction between complexed bile acids and this receptor (71, 72). Thus, bile acid sequestrants should, because of the well-established TGR5-mediated potentiation of GLP-1 secretion, instigate a negative or at best neutral effect on TGR5-mediated GLP-1 secretion. On the other hand, Trabelsi et al. have reported induced GLP-1 secretion as a result of decreased FXR activation in L cells, which could indicate a positive effect of bile acid sequestrants on GLP-1 secretion (6). Stable and permanent binding of bile acids throughout the intestinal tract should result in likewise reverse effects with respect to TGR5 and FXR activation in the L cells of the colon. However, potential dissociation of bile acids from the sequestrant complex in the distal colon could instigate bile acid–mediated activation of both TGR5 and FXR in distally located L cells (104, 105). The apparent reverse effects of TGR5 and FXR make it relevant to consider the impact of reduced bile acid reabsorption on receptor activation following treatment with sequestrants. Human studies do not enable any direct quantification of these matters, but a recent study from our group found single-dose administration of the bile acid–sequestering resin sevelamer to eliminate bile acid–induced GLP-1 secretion in patients with type 2 diabetes (106).

The potential sequestrant-mediated glucose-lowering effect resulting from increased splanchnic glucose utilization could involve modulation of intestinal and hepatic FXR activation (100). Thus, human studies have demonstrated treatment with bile acid sequestrants to reduce plasma FGF-19 concentration, which point to reduced intestinal FXR activation (102, 106). Furthermore, the reduced FGF-19 concentration and increased fecal loss of bile acids might instigate a sequestrant-mediated shift in bile acid composition, which is in line with findings from previous studies (102, 107). Changes in plasma profile alongside the overall composition of bile acids could instigate a modulation of hepatic FXR activation with potential implications for glucose metabolism. Nevertheless, and as presented previously, the most relevant approach for FXR modulation in terms of improving glycemic control has so far not been established.

Overall, the dual effects of bile acid sequestrants on glucose and lipid metabolism constitute an interesting approach to the treatment of patients with type 2 diabetes. The clinical relevance of this drug class is substantiated by a previous study demonstrating reduced risk of cardiovascular morbidity and mortality during treatment with cholestyramine compared with placebo in patients with dyslipidemia (108). A meta-analysis has reported treatment with bile acid sequestrants to elicit a reduction in LDL cholesterol of 0.61 mmol/L (95% CI, 0.26-0.95) compared with placebo, whereas the potential effect on cardiovascular outcomes in patients with type 2 diabetes remains to be clarified (109). Furthermore, the nonabsorbable nature of bile acid sequestrants warrants no need for precautions in patients with impaired renal or hepatic function, which are both common challenges in patients with type 2 diabetes (110, 111). A Cochrane review has reported the bile acid sequestrant colesevelam to reduce HbA1c by 0.5% (5 mmol/mol) (95% CI, 0.4-0.6) in patients with type 2 diabetes (97). This drug is included as a treatment option for the management of hyperglycemia in patients with type 2 diabetes in a position statement from the American Diabetes Association and the European Association for the Study of Diabetes. However, colesevelam should only be considered in selected patients because of modest efficacy and/or limiting gastrointestinal side effects (112).

Metformin

The biguanide metformin is the well-established first-line and most often applied glucose-lowering pharmacotherapy in type 2 diabetes (112). Metformin has a pronounced glycemic effect with a reduction in HbA1c of 1.1% (12 mmol/mol) (95% CI, 0.92-1.32) compared with placebo (113) and the historically accepted glucose-lowering mechanisms include suppressed hepatic glucose production accompanied by improved peripheral insulin sensitivity (114). However, despite being introduced as a drug more than 60 years ago, the glucose-lowering modes of action of metformin remain to be fully clarified. Along these lines, novel gut-derived modes of action including modulation of enterohepatic circulation of bile acids (15), increased GLP-1 secretion (115–118), and changes in gut microbiota composition (119–121) have surfaced within recent years. The combination of labeled ¹¹C-metformin and positron emission tomography imaging has proven to be a useful modality for assessment of metformin biodistribution in humans (122). By this technique, metformin has been demonstrated to accumulate within enterocytes following oral administration, whereas only a minor increase in these cells was observed after intravenous administration because of a low basolateral metformin transporting capacity in the enterocytes (122). These findings might partly explain the lack of acute gluco-metabolic effects after intravenous administration of metformin in both healthy subjects and patients with type 2 diabetes (123, 124), which supports the importance of gut-derived modes of action of metformin. Furthermore, the relevance of metformin-induced modes of action involving the gut is underlined by the noninferior glycemic effect of delayed release metformin compared with instant release and extended release formulations, in spite of a ∼50% lower plasma bioavailability of the former compound (125).

Metformin has been demonstrated to reduce the ASBT-mediated reabsorption of bile acids in the terminal ileum (15, 115, 126, 127) with potential modulation of TGR5 and FXR activation in L cells as well as implications for gut microbiota composition. Metformin-mediated inhibition of the ASBT with reduced reabsorption of bile acids should most likely convey reductions of both FXR and TGR5 activation in L cells in the terminal ileum. In contrast, the ensuing increased amount of bile acids reaching the colon is likely to cause increased receptor activation in colonic L cells because the reabsorption of bile acids in this part of the gastrointestinal tract is passive and independent of ASBT (128). The balance between the previously described reverse effects of TGR5 and FXR modulation with respect to GLP-1 secretion seem important in terms of the potential metabolic outcome. However, the substantial glucose-lowering impact in humans and increased GLP-1 concentration in rodents after treatment with ASBT inhibitors point to a beneficial effect of ASBT inhibition (94, 96). In addition, a study including healthy young men demonstrated increased plasma GLP-1 concentrations following isolated oral administration of metformin and intravenous infusion of CCK, respectively (129). Interestingly, an additive effect of the single-dose metformin and CCK-mediated gallbladder emptying on GLP-1 secretion was observed in this study, which point to a metformin-induced potentiation of bile-mediated GLP-1 secretion (129). Furthermore, a similar study including patients with type 2 diabetes demonstrated single-dose metformin to enhance bile-mediated GLP-1 secretion following CCK-mediated gallbladder emptying (130).

A randomized, placebo-controlled study by Wu et al. in treatment-naive patients with type 2 diabetes demonstrated 4 months of treatment with metformin to elicit a strong impact on the gut microbiome (121). This is in line with findings from previous studies within this field (119, 120). Interestingly, subsequent transfer of fecal samples from metformin-treated donors was shown to improve glucose tolerance in germ-free mice, which suggest modulated gut microbiota composition to be part of the glucose-lowering mechanism of metformin (121). A variety of direct effects of metformin on gut microbiota has been suggested (121, 131), whereas the findings from Wu et al. could impose an additional potential link between bile acids and metformin-induced glucose-lowering effects. Thus, alterations in bile acid composition have been reported following treatment with metformin (121, 132) and it is now evident that a significant interdependence exists between bile acids and the gut microbiota (133, 134). Bile acids contribute to the regulation of microbiota composition through mechanisms that have been suggested to involve production of antimicrobial peptides (as a result of bile acid–mediated FXR activation) alongside a direct disturbance of bacterial membrane integrity (11, 135–137). Along these lines, a study has demonstrated intestinal FXR activation to cause an increase in lithocholic acid-producing bacteria with subsequent induction of GLP-1 secretion and beneficial effects on hepatic glucose metabolism (58). The potency rank order for activation of TGR5 favors the hydrophobic secondary bile acids with lithocholic acid being the most potent ligand for the receptor (4). Nevertheless, a bile acid–mediated contribution to the described effects of metformin on gut microbiota and glucose metabolism remains speculative.

Discussion

The characterization of the bile acid receptors FXR and TGR5, alongside the extensive range of studies addressing the potential link between glucose metabolism and bile acid receptor activity, have substantiated the concept of bile acids as important gluco-metabolic integrators (138). Numerous metabolic effects resulting from modulated bile acid metabolism and receptor activation have been demonstrated. However, these findings derive mostly from in vitro studies and animal models, and thus, the potential clinical impact of these apparent beneficial effects remains somewhat unresolved. It is important to point out that considerable interspecies differences exist with respect to bile acid physiology. In rodents, the bile acid pool composition includes relatively large amounts of muricholic acids with FXR antagonistic effects, whereas the FXR antagonistic effects of ursodeoxycholic acids in the human pool most likely are negligible (100). These substantial differences with respect to bile acid composition and metabolism seem to constitute a challenge in terms of translating findings in rodents to human physiology (100, 133).

A range of compounds with potential bile acid–mediated glucose-lowering effects has been tested in human settings. These include agonists for the bile acid receptors FXR and TGR5 alongside ASBT inhibitors, bile acid sequestrants, and metformin that are indirect modulators of the enterohepatic circulation of bile acids. Metformin and the bile acid sequestrant colesevelam are both approved for the treatment of hyperglycemia in patients with type 2 diabetes, whereas the specific ASBT inhibitor GSK672 is currently undergoing phase 2 evaluation for the treatment of type 2 diabetes. Modulation of bile acid recirculation and metabolism most likely contribute to the glucose-lowering effects of these 3 pharmaceutical compounds. However, the potential bile acid–mediated glucose-lowering mechanisms of the indirect modulators of bile acid circulation remain to be clarified, and a range of other mechanisms also play a part in the glucose-lowering effect of metformin (16). In spite of their LDL cholesterol-lowering effect and their similar effect on glycemic control compared with the widespread dipeptidyl peptidase 4 inhibitors, the bile acid sequestrants are marginally positioned in treatment guidelines because of gastrointestinal adverse events combined with limited clinical effects (112). However, the development of humanized rodent models with respect to bile acid composition could enable valuable mechanistic and hypothesis-generating studies, which might allow for an optimization of bile acid–modulating treatment modalities (133). Studies with the FXR agonist obeticholic acid have reported conflicting results on gluco-metabolic effects (77, 78). However, the largest and most extensive of these trials demonstrated no beneficial glycemic effects following treatment with obeticholic acid for 72 weeks (78). In fact, a post hoc analysis from this trial found an apparent deteriorating effect of obeticholic on basal plasma glucose, HbA1c, and HOMA-IR (79). FXR constitutes an attractive target for the treatment of fatty liver disease, whereas the potential of FXR agonism in terms of improvements of glycemic control remains to be established (139). Only 1 human study has examined the effects of treatment with a specific TGR5 agonist and found no beneficial effects on GLP-1 secretion or plasma glucose concentrations. In addition, animal studies have demonstrated substantial safety concerns in relation to treatment with TGR5 agonists (82–84), which, for now, altogether makes the development of a clinical relevant TGR5-based treatment modality unlikely (83).

Additional Information

Disclosure Summary: The authors received no honorarium for the preparation of the present manuscript. FKK has received lecture fees from, participated in advisory boards of, consulted for and received research grants from Sanofi that produces and markets the bile acid sequestrant colesevelam included in this review.

Data Availability: Data sharing is not applicable to this article because no datasets were generated or analyzed during the current study.

Abbreviations

Abbreviations
 
• ASBT

  apical sodium-dependent bile acid transporter

 
• CI

  confidence interval

 
• CCK

  cholecystokinin

 
• FGF

  fibroblast growth-factor

 
• FXR

  farnesoid X receptor

 
• GLP-1

  glucagon-like peptide 1

 
• HOMA-IR

  homeostatic model assessment for insulin resistance

 
• LDL

  low-density lipoprotein

 
• TGR5

  Takeda G protein-coupled receptor 5.

References