Plasma Bile Acids More Closely Align With Insulin Resistance, Visceral and Hepatic Adiposity Than Total Adiposity

Abstract

Context

The etiological mechanism of bile acid (BA) effects on insulin resistance and obesity is unknown.

Objective

This work aimed to determine whether plasma BAs are elevated in human obesity and/or insulin resistance.

Methods

This observational study was conducted at an academic research center. Seventy-one adult volunteers formed 4 groups: lean insulin-sensitive (body mass index [BMI] ≤ 25 kg/m², Homeostatic Model Assessment of Insulin Resistance [HOMA-IR] < 2.0, n = 19), overweight/obese nondiabetic who were either insulin sensitive (Ob_sensitive, BMI > 25 kg/m², HOMA-IR < 1.5, n = 11) or insulin resistant (Ob_resistant, BMI > 25 kg/m², HOMA-IR > 3.0, n = 20), and type 2 diabetes (T2D, n = 21). Main outcome measures included insulin sensitivity by hyperinsulinemic-euglycemic clamp, body composition by dual energy x-ray absorptiometry, abdominal fat distribution, and liver density by computed tomography and plasma BA.

Results

In the Ob_resistant group, glucose infusion rate/fat-free mass (GIR/FFM, an inverse measure of insulin resistance) was significantly lower, and visceral and liver fat higher, compared to lean and Ob_sensitive individuals, despite similar total adiposity in Ob_resistant and Ob_sensitive. Total BA concentrations were higher in Ob_resistant (2.62 ± 0.333 mmol/L, P = .03) and T2D (3.36 ± 0.582 mmol/L, P < .001) vs Ob_sensitive (1.16 ± 0.143 mmol/L), but were similar between Ob_sensitive and lean (2.31 ± 0.329 mmol/L) individuals. Total BAs were positively associated with waist circumference (R = 0.245, P = .041), visceral fat (R = 0.360, P = .002), and fibroblast growth factor 21 (R = 0.341, P = .004) and negatively associated with insulin sensitivity (R = –0.395, P = .001), abdominal subcutaneous fat (R = –0.352, P = .003), adiponectin (R = –0.375, P = .001), and liver fat (Hounsfield units, an inverse marker of liver fat, R = –0.245, P = .04). Conjugated BAs were additionally elevated in T2D individuals (P < .001).

Conclusions

BA concentrations correlated with abdominal, visceral, and liver fat in humans, though an etiological role in insulin resistance remains to be verified.

bile acid, insulin resistance, obesity, type 2 diabetes, visceral fat mass, metabolic associated fatty liver disease

It has been known for more than 30 years that bile acids (BAs) play important roles in modulating glucose and lipid metabolism beyond their well-known actions as facilitators of hepatobiliary detoxification, absorption of fat-soluble vitamins, and biliary secretion of lipids (1, 2). It is also now acknowledged that abnormalities of BA signaling are associated with multiple disease states, including metabolic-associated fatty liver disease (MAFLD), cholestatic liver disease, dyslipidemia, type 2 diabetes mellitus (T2D), and cardiovascular disease (3-5). In line with this, BA sequestration or targeting BA receptor binding improves glucose metabolism (6-8). Proposed mechanistic hypotheses include amplification of incretin-induced insulin production, enhanced splanchnic glucose use, improved hepatic and/or peripheral insulin sensitivity, and inhibition of food intake (8).

BAs are produced from cholesterol in the liver via either the classical (neutral acid) or acidic pathways; the former produces the primary BA species, cholic acid (CA) and chenodeoxycholic acid (CDCA) (4), both constituting 80% of the total BA pool. They are conjugated to glycine or taurine to improve their solubility and secreted into bile and concentrated in the gallbladder during fasting; in response to nutrient lipid ingestion, they are expelled into the intestinal lumen and undergo enzymatic biotransformation to secondary species, deoxycholic acid (DCA) and lithocholic acid by gut microbiota. They are then reabsorbed in the ileum and colon and circulate back to the liver (9).

Recently, BA signaling via Takeda G protein–coupled receptor (TGR-5) (especially via DCA), (10, 11), has been proposed to be the principal site by which incretin-dependent glucose modulation via the actions of glucagon like peptide-1 (GLP-1) result in improved glucose homeostasis, whereas Farnesoid X-receptor (FXR)-dependent BA signaling (primarily via CA and CDCA) is responsible for peripheral tissue insulin sensitivity and glucose uptake (12, 13). High plasma BA levels, in contrast, are associated with human obesity according to a recent study (14).

To date, the contribution of insulin resistance in augmentation of circulating BA levels, and whether they more closely reflect excess adiposity more generally or at specific tissue sites (eg, hepatic, central, visceral) remains unclear, especially in well-characterized human studies.

Given that higher plasma BAs are associated with adverse metabolic disease states, our aim was to investigate whether circulating plasma BA levels correlated more closely with adiposity (total, visceral, or hepatic) or insulin resistance, measured by the gold-standard euglycemic-hyperinsulinemic clamp. Our hypothesis was that plasma BA concentrations aligned more closely with insulin resistance, a key driver of metabolic disease, rather than total adiposity in humans.

Materials and Methods

Participant recruitment

We have previously described our participant selection method (15). In brief, 1032 individuals responded to local advertisements, of whom 81 were included in the study after screening. Ten individuals (4 lean, 1 insulin-resistant overweight/obese nondiabetic [Ob_resistant] and 5 insulin-sensitive overweight/obese nondiabetic [Ob_sensitive]) were excluded for lack of available BA data, leaving 71 individuals to be included in the study analysis. Patients without a known diagnosis of T2D, impaired fasting glucose (IFG), or impaired glucose tolerance (IGT), underwent assessment with the Homeostatic Model of Assessment of Insulin Resistance (HOMA-IR) (16) and a 75-g oral glucose tolerance test and were subsequently classified using well-established international criteria as having T2D (ie, fasting blood glucose [FBGL] ≥ 7.0 mmol/L and/or a 2-hour postchallenge blood glucose level ≥ 11.1 mmol/L), IFG (ie, FBGL = 5.6-6.9mmol/L), or IGT (ie, 2-hour postglucose challenge BGL = 7.8-11.0 mmol/L). Participants were categorized accordingly into 1 of 4 groups based on screening investigations: lean insulin-sensitive controls (BMI ≤ 25 kg/m², HOMA-IR < 2.0, n = 19), Ob_resistant (BMI ≥ 25 kg/m², HOMA-IR ≥ 3.0, n = 20), those with T2D (BMI ≥ 25 kg/m², n = 21 [7 were new diagnoses]), Ob_sensitive (BMI ≥ 25 kg/m², HOMA-IR < 1.5, n = 11). The lean control group had normal glucose metabolism. Of the 20 in the Ob_resistant group, 2 had IFG, 6 IGT, and 1 participant had both. Two volunteers in the Ob_sensitive group had IGT.

Patients with T2D were included if their glycated hemoglobin A_1c was less than 9% (75 mmol/mol), they had no history or evidence of diabetic microvascular or macrovascular complications, and diagnosis occurred no more than 5 years earlier. The management of the T2D group was limited to lifestyle modification, metformin, and/or sulfonylurea only. Metformin was withheld 2 weeks prior to the study and a sulfonylurea added if FBGLs were consistently 7.0 mmol/L or greater on self-monitoring of BGLs (n = 2; ceasing the day before the studies).

Study volunteers were required to be nonsmokers or if reported smoking, smoke fewer than 10 cigarettes per day. They were also limited to less than 20 g (women) or 40 g (men) of alcohol per day, not taking medications affecting carbohydrate metabolism, did not participate in intensive physical activity, and did not report more than a 5% change in body weight in the preceding 3 months prior to study commencement. Menstruating women (n = 9) were studied in the follicular phase of their menstrual cycle. Ethical approval was obtained from the St Vincent’s Hospital Health Research Ethics Committee, Sydney, Australia. Informed written consent was obtained from all participants.

Hyperinsulinemic-euglycemic clamp

Volunteers were advised to avoid exercise and alcohol for 48 hours before study commencement. Participants underwent a 2.5-hour hyperinsulinemic (80 mUm⁻²min⁻¹)–euglycemic (5 mmol/L) clamp following an overnight fast, and the glucose infusion rate (GIR) was calculated during the last 30 minutes of the clamp. This was normalized to fat-free mass (FFM) as measured by dual energy x-ray absorptiometry (DEXA) scanning.

Body fat determination

Total body fat, FFM, and central fat were analyzed using DEXA. Liver attenuation was assessed using computed tomography at the T12/L1 level and expressed as Hounsfield units (HUs, where HU is inversely proportional to the amount of liver fat). The size of abdominal visceral and subcutaneous fat was determined using computed tomography at the L2/L3 and L4/L5 levels. The mean visceral fat areas at L2/L3 and L4/L5 and the mean subcutaneous fat areas at L2/L3 and L4/L5 were used in all the analyses.

Assays

BGL was analyzed using a glucose oxidase electrode (YSI Life Sciences). Screening insulin levels were analyzed by the Advia Centaur immunoassay. Clamp plasma insulin levels were measured using radioimmunoassay (Millipore; intra-assay and interassay coefficients of variation, 3.1%-4.4% and 2.9%-6.0%, respectively). Nonesterified fatty acids (NEFA) were assayed using an enzymatic calorimetric method (Wako). Serum adiponectin and fibroblast growth factor 21 (FGF21) were measured by sandwich enzyme-linked immunosorbent assay (Antibody and Immunoassay Service, University of Hong Kong). More detailed assay information has been previously reported (15).

Bile acids

Individual plasma BAs were quantified by high-performance liquid chromatography–mass spectrometry (LC/MS) using authentic BA standards and deuterated internal standards as described (17). The detection limit for individual BAs was 10 to 50 nmol/L. Plasma BAs were collected in fasting patients and at the end of the clamp protocol. For CA and CDCA values, 5 and 2 lean participants had no CA and CDCA detectable, respectively, whereas 3 Ob_resistant individuals had no CA detectable. In T2D individuals, 1 participant had no CA detectable and 1 had no data, with 2 having no data for CDCA. In the Ob_sensitive cohort, 1 and 3 participants had no detectable CA and CDCA, respectively. There were 2 individuals in the lean cohort who had undetectable DCA and 1 participant in the T2D cohort with missing data. Further information regarding the metabolic profiling and factors affecting detection of BA by LC/MS are reported in more details elsewhere (18, 19).

Statistical analysis

Data are reported as means ± SE. Two-way analysis of variance was used to determine differences in the mean plasma BA concentrations and subspecies preclamp and clamp steady state infusion studies using the Tukey multiple comparisons test. BA species were tested for normality by Kolmogorov-Smirnov test and log transformed in correlation analyses. Pearson coefficients were determined to assess correlations between BA and continuous variables. Analyses were performed using GraphPad Prism 7.0b and SPSS Statistics for Windows, (IBM Corp, released 2012, IBM, version 21.0). Two-tailed P less than .05 was considered statistically significant.

Results

Baseline characteristics of the study participants are summarized in Table 1. There was an approximately equal distribution of study participants in the 4 phenotypic subgroups with respect to age with the exception of Ob_sensitive being slightly older than lean controls (62.8 ± 4.3 vs 55.0 ± 8.4 years, P < .05). Nearly all indices of central adiposity and insulin resistance were greater in the Ob_resistant vs lean controls. By definition, key parameters of insulin resistance (GIR/FFM, HOMA-IR) were similar between Ob_sensitive and lean controls and additionally, hepatic fat and fasting levels of FGF21, NEFA, and adiponectin were also similar between the 2 cohorts, despite a statistically significant greater waist circumference, BMI, total body (kilogram [kg], percentage [%]), central (kg, %), visceral, and average subcutaneous fat in the former (see Table 1). Waist circumference, BMI, central fat (kg), hepatic and average visceral fat, but not average subcutaneous fat, were significantly greater in Ob_resistant vs Ob_sensitive. Interestingly, Ob_sensitive participants had similar hepatic fat compared with lean controls (59.3 ± 6.2 vs 61.9 ± 4.3 HU, P = .6).

Table 1.

Clinical and biochemical characteristics of study participants (means ± SD)

	Lean	Obresistant	T2D	Obsensitive
No. (M:F)	19 (8:11)	20 (13:7)	21 (8:12a)	11 (2:9)
Age, y	55.0 ± 8.4	56.4 ± 8.2	60.7 ± 7.8	62.8 ± 4.3b
Waist circumference, cm	79.6 ± 8.7	111.1 ± 10.2d	102.8 ± 11.6d	100.0 ± 13.8d,e
BMI, kg/m2	21.8 ± 1.9	34.0 ± 6.4d	30.2 ± 3.4d,e	29.4 ± 3.7d,e
HOMA-IR	1.1 ± 0.4	4.4 ± 1.0d	–	1.2 ± 0.3g
Fasting insulin, mU/L	11.1 ± 2.4	23.4 ± 8.6d	23.7 ± 10.7d	11.2 ± 3.6g,i
Steady-state insulin, mU/L	264 ± 48	332 ± 97	358 ± 119b	333 ± 114
GIR/FFM, μmol min–1kg–1	92 ± 24	60 ± 25d	45 ± 13d	94 ± 37f,i
Nonoxidative glucose disposal, μmol min–1kg–1	125 ± 35.3	100 ± 31.2	66.2 ± 20.8d,f	126 ± 47.1i
Adiponectin, fasting, μm/mL	25.8 ± 12.9	13.6 ± 7.6c	14.7 ± 8.8c	24.0 ± 9.3e
FGF21, fasting, μm/mL	105 ± 88.7	163 ± 118	119 ± 157	114 ± 82.5
NEFA, fasting mmol/L	0.37 ± 0.16	0.32 ± 0.12	0.37 ± 0.12	0.44 ± 0.12
NEFA, steady-state mmol/L	0.02 ± 0.02	0.02 ± 0.03	0.01 ± 0.01	0.02 ± 0.01
DEXA fat
Total body fat, kg	16.0 ± 5.1	40.7 ± 12.0d	33.0 ± 7.0d,e	32.5 ± 11.6d
Total body fat, %	26.1 ± 7.7	40.9 ± 9.7d	40.0 ± 7.1d	41.1 ± 8.8d
Central fat, kg	1.2 ± 0.5	3.5 ± 0.9d	2.9 ± 0.7d	2.5 ± 1.3d,e
Central fat, %	24 ± 8	44 ± 6d	42 ± 6d	38 ± 10d
Fat content of legs, kg	5.8 ± 2.0	12.2 ± 5.3d	9.4 ± 2.8b	11.7 ± 5.8c
CT fat
Hepatic fat, HU	61.9 ± 4.3	39.7 ± 15.6d	43.9 ± 16.2d	59.3 ± 6.2g,h
Average visceral fat, cm2	51.8 ± 27.4	220 ± 67.7d	151 ± 61.2d,f	124 ± 116b,f
Average subcutaneous fat, cm2	111 ± 60.4	337 ± 158d	269 ± 98.2d	275 ± 112c

Data are reported as means ± SEM.

^aMissing = 1. ^bP less than .05. ^cP less than .01. ^dP less than .001 vs lean controls. ^eP less than .05. ^fP less than .01. ^gP less than .001 vs Ob_resistant individuals; ^hP less than .01. ⁱP less than .001 vs diabetic participants.

Abbreviations: BMI, body mass index; CT, computed tomography; DEXA, dual energy x-ray absorptiometry; F, female; FGF21, fibroblast growth factor 21; GIR/FFM, glucose infusion rate/fat-free mass; HOMA-IR, Homeostatic Model Assessment of Insulin Resistance; HU, Hounsfield unit; M, male; NEFA, nonesterified fatty acids; Ob_resistant, insulin-resistant overweight/obese nondiabetic; Ob_sensitive, insulin-sensitive overweight/obese nondiabetic; T2D, type 2 diabetic.

Results of plasma BA levels measured at baseline and after hyperinsulinemic euglycemic clamp are shown in Fig. 1. Ob_resistant (2.62 ± 0.333, P = .03) and T2D participants (3.36 ± 0.582, P = .0004) had higher baseline total BA (Fig. 1A) concentration than Ob_sensitive (1.16 ± 0.143). Total and all BA species (A-F) were suppressed with the hyperinsulinemic-euglycemic clamp but to a much lesser degree in Ob_sensitive individuals. There was a statistically significant difference in primary BAs (sum of CA + CDCA; Fig. 1B) in Ob_resistant (0.714 ± 0.180) vs Ob_sensitive participants (0.186 ± 0.059, P = .01), likely driven by higher CDCA (0.439 ± 0.093 vs 0.107 ± 0.040, P = .01; Fig. 1C) in the former. There was a trend for higher CA levels observed in Ob_resistant vs Ob_sensitive (0.275 ± 0.094 vs 0.079 ± 0.022, P = .06; Fig. 1D). Fasting plasma secondary BAs (DCA) were higher in T2D vs lean controls (0.755 ± 0.127 vs 0.443 ± 0.076, P = .02; Fig. 1E). Unconjugated BA species (sum of CA, CDCA, and DCA) similarly discriminated between Ob_resistant (1.349 ± 0.248) and Ob_sensitive participants (0.672 ± 0.084, P = .04), whereas total conjugated BA species (glyco- and tauro-conjugated forms of CA, CDCA, and DCA) were statistically significantly higher in T2D vs Ob_sensitive participants (2.101 ± 0.541 vs 0.493 ± 0.075, P = .001). The distribution of all BAs and their subspecies are shown in a box-and-whisker plot and accessible in the supplementary data repository (20).

Figure 1.

Plasma bile acid (BA) concentrations at baseline and following hyperinsulinemic euglycemic clamp. A, Baseline total BA levels. Total BA and all BA species, A to F, were suppressed with hyperinsulinemic euglycemic clamp but to a much lesser degree in Ob_sensitive. B, Primary BA (sum of CA + CDCA). C, (Primary BA) CA levels. D, (Primary BA) CDCA. E, (secondary BA) DCA. F, Unconjugated and E, conjugated BA. Data are reported as means ± SEM. P less than .05 was considered to be statistically significant. CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; Ob_sensitive, insulin-sensitive overweight/obese nondiabetic.

A Pearson correlations heat map is shown in Fig. 2. Positive associations were found between baseline fasting total BA and waist circumference (R = 0.245, P = .041), mean visceral fat (% at L2/L3, R = 0.360, P = .002), central abdominal fat (kg, R = 0.263, P = .028), and FGF21 (R = 0.341, P = .004). Negative associations with insulin sensitivity (GIR/FFM/insulin, R = –0.395, P = .001), nonoxidative glucose disposal (R = –0.320, P = .007), subcutaneous fat (L2/L3 %, R = –0.352, P = .003), adiponectin (R = –0.375, P = .001), and liver fat (or HU, an inverse marker of liver fat content, R = –0.245, P = .041) were demonstrated. Similar correlations were observed for primary BA species—CA and CDCA—with the latter additionally and positively associated with weight (R = 0.313, P = .015), visceral (R = 0.345, P = .005), and central (R = 0.247, P = .049) fat but negatively associated with insulin sensitivity (R = –0.284, P = .026), subcutaneous fat (R = –0.333, P = .007), and adiponectin (R = –0.254, P = .043). The secondary BA DCA similarly correlated with waist circumference (R = 0.283, P = .019), visceral fat (R = 0.294, P = .015), and FGF21 (R = 0.321, P = .008) but negatively with insulin sensitivity (GIR/FFM/insulin, R = –0.382, P = .002), (R = 0.294, P = .015), subcutaneous fat (R = –0.298, P = .014), adiponectin (R = -0.407, P = .001), and NEFA (R = –0.283, P = .019). Tertiary species (guanosine [G]-CDCA) were similarly positively associated with visceral fat and FGF21 and negatively with subcutaneous fat and adiponectin, whereas G-DCA correlated negatively with insulin resistance and adiponectin but positively with FGF21 (all P < .05).

Figure 2.

Correlation heat map of bile acid (BA) species with metabolic parameters of insulin resistance and obesity. General trends for a positive association between total, primary (CA + CDCA), secondary (DCA), and conjugated BA with several metabolically adverse indices such as waist circumference, mean visceral fat, central abdominal fat, liver adiposity and FGF21 but negatively with insulin sensitivity, nonoxidative glucose disposal, subcutaneous fat, and adiponectin. Of the conjugated species, (guanosine [G]-CDCA) was positively associated with visceral fat and FGF21 and negatively with subcutaneous fat and adiponectin whereas G-DCA correlated negatively with insulin resistance and adiponectin but positively with FGF21 (all P < .05). P less than .05 was considered to be statistically significant. CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; FGF21, fibroblast growth factor 21; GIR/FFM/insulin, glucose infusion rate/fat-free mass/insulin; NEFA, nonesterified fatty acids; NOGD, nonoxidative glucose disposal; WC, waist circumference.

Discussion

This is the first study to investigate BA composition in Ob_sensitive vs Ob_resistant individuals stratified based on the gold-standard measure of insulin resistance, the hyperinsulinemic-euglycemic clamp. The study design allowed us to accurately study BA signatures in overweight and obese individuals without the limitations of preclinical models.

We confirmed the finding of higher fasting total and primary BA species observed in individuals with obesity and insulin resistance (14) (see Fig. 1A and 1B); however, total and primary BA species were interestingly significantly lower in Ob_sensitive individuals, suggesting that higher fasting levels of these classical pathway BA subspecies are associated with an insulin-resistant phenotype independent of total body adiposity. Indeed, previous reports correlated elevated CA (which requires 12α-hydroxylation) with insulin resistance, MAFLD, and diabetes (14, 21, 22) via intestinal absorption of dietary cholesterol and fats. Additionally, our study demonstrated that fasting plasma CDCA, a hydrophobic, proinflammatory BA subspecies, was significantly higher in Ob_resistant vs Ob_sensitive individuals, and found to correlate significantly with weight, insulin resistance, and visceral fat but negatively with subcutaneous fat and adiponectin levels. Of all BA subspecies, CDCA is the most potent natural BA ligand for FXR-signaling, which is present in gut and white adipose tissue and vital for regulating peripheral glucose metabolism (23, 24). This highlights the potential of CDCA (non-12α -hydroxylated branch of primary BA synthesis) as an important regulator of insulin sensitivity that is independent of body weight, BMI, or total body fat. In part, these findings demonstrate why cholestyramine, a potent BA sequestrant that preferentially binds the hydrophobic CDCA over CA, exerts its antiglycemic and hypocholesterolemic properties (25).

Admittedly, elucidation of BA signaling via FXR in preclinical models by BA metabolites have been blurred by discordant results (23, 26-28). Some of this discordance may be attributable to the additional signaling modulation by TGR-5, a G protein–coupled receptor and a major receptor target for BA signaling present mainly in the intestine, brown and white adipose tissue, and gallbladder, with lower levels present in the liver and skeletal muscle (3). Activation of TGR-5 generally induces metabolically favorable downstream effects by modulating energy expenditure, stimulating GLP1 release from intestinal L cells, suppressing hepatic glycogenolysis, reducing inflammation and macrophage activation, and improving fatty liver disease and pancreatic function (29). CA is conjugated by colonic bacteria into the secondary species DCA, a toxic hydrophobic BA that contributes 20% of the total BA pool and has one of the strongest affinities for TGR-5. Our study demonstrated a positive association of higher fasting DCA concentration with worsening metabolic phenotype, however, with interestingly low and similar concentrations between lean and Ob_sensitive cohorts.

The correlation of DCA with insulin resistance and visceral fat may also highlight the influence of the microbiome in influencing BA pool distribution in metabolic health (and vice versa) (30). Our study demonstrated a positive correlation between DCA with waist circumference (rather than weight or BMI), as well as visceral fat, but negatively with insulin sensitivity, subcutaneous fat, adiponectin levels, FGF21, and NEFA. This is corroborated by clinical and preclinical studies illustrating the role of DCA in disrupting whole-body glucose homeostasis and when sequestered pharmacologically (by its hydrophilic antagonist, ursodeoxycholic acid) or reduced via bariatric surgery, improvement in glycemic control, though GLP-1 potentiation and MAFLD is observed (7, 31-33). The more hydrophobic the BA pool is, the greater its detergent properties and propensity for proinflammatory effects of cell membrane disruption (34). These findings, taken together with an elevated fasting CDCA in Ob_resistant vs Ob_sensitive individuals, suggest that the hydrophobicity index of the BA pool is a key feature of insulin resistance, visceral fat mass, and T2D independent of total or subcutaneous body fat.

Toxic hydrophobic BA metabolites may be associated with insulin resistance by several mechanisms. The FXR receptor (via primary BAs CA and CDCA) regulates BA synthesis by modulating CYP7A1 but also directly influences glucose and lipid metabolism when activated by BA, leading to decreased gluconeogenesis, increasing glycolysis leading to improved glucose tolerance and insulin sensitivity (29). In humans, increased production of primary BAs and/or 12α-hydroxylated BAs is seen in obesity and T2D and is associated with increased BA synthesis, as well as changes in BA transport (14, 35).

Assessing the acute changes in BA concentrations following hyperinsulinemic-euglycemic clamp, especially in lean individuals, has been scarcely reported; our study confirmed previous reports of a reduction in circulating BA concentrations following insulin infusion, with some speculating that this is likely related to increased uptake from the portal vein into the liver (14). We did not specifically find differences in the degree of BA suppression following insulin infusion in the different metabolic phenotype cohorts as some have reported (14), though this is perhaps related to our small sample size. Additionally, we observed that conjugated BAs, which are produced from conjugation of primary BA to either taurine or glycine (and subsequently stored in the gallbladder), were elevated in individuals with T2D. This is supported by previous work showing significant elevations of conjugated BA species, specifically taurine-conjugated BA, in T2D patients as well as positive associations with fasting glucose, glycated hemoglobin A_1c, and HOMA-IR, though interestingly, treatment with insulin did not alter the levels of total levels or composition of BAs (36). Speculation as to the mechanisms include secondary effects of obesity or insulin-resistance on the development of aberrations in the expression and function of sodium/BA cotransporter (Na⁺-taurocholate cotransporting polypeptide, NTCP) as NTCP preferentially transports conjugated BAs (4, 14).

In individuals with nonalcoholic steatophepatitis (NASH), BA concentrations in the blood and in liver specimens (37) were found to be higher and harbor altered composition compared to healthy controls. Of note, in individuals with NASH matched for BMI and insulin resistance, the association of increased BA was related to the presence of insulin resistance rather than liver necroinflammation (38), providing support that dysfunctional hepatic insulin signaling may be a key to the production of the altered and increased BA pool. As observed in our study, elevated FGF21 was elevated in Ob_resistant individuals and aligned significantly with total and primary BA, which corroborates previous work (39).

Our cross-sectional study design did not allow us to determine mechanistic pathways and modifications in BA receptor signaling, though this is a clear area of further research. Our cohort was well characterized but admittedly small, so inferences from this study should be taken with caution and should provide support for further inquiry.

In summary, our study suggests that insulin resistance is inextricably related to maladaptive alterations to the bile acid pool composition. Our study adds to previous works and provides evidence that elevated bile acids in insulin resistance is possibly contributed to by defects in insulin signaling, rather than obesity. BA physiology is complex, and studies in preclinical models have been clouded by inconsistent findings. The renewed interest in the ability of BAs to regulate glucose and metabolic signals outside their classical vocation as detergent and detoxification chaperones will likely require further clinical investigation to determine whether modulating BA pool composition promises to rectify some of the most prevalent chronic disease states affecting the globe.

Abbreviations

Abbreviations
 
• BA

  bile acid

 
• BGL

  blood glucose

 
• BMI

  body mass index

 
• CA

  cholic acid

 
• CDCA

  chenodeoxycholic acid

 
• DCA

  deoxycholic acid

 
• DEXA

  dual energy x-ray absorptiometry

 
• FBGL

  fasting blood glucose

 
• FFM

  fat-free mass

 
• FGF21

  fibroblast growth factor 21

 
• FXR

  Farnesoid X-receptor

 
• GIR

  glucose infusion rate

 
• GLP-1

  glucagon like peptide-1

 
• HOMA-IR

  Homeostatic Model Assessment of Insulin Resistance

 
• HU

  Hounsfield unit

 
• IFG

  impaired fasting glucose

 
• IGT

  impaired glucose tolerance

 
• LC/MS

  liquid chromatography–mass spectrometry

 
• MAFLD

  metabolic-associated fatty liver disease

 
• NASH

  nonalcoholic steatophepatitis

 
• NEFA

  nonesterified fatty acid

 
• NTCP

  Na⁺-taurocholate cotransporting polypeptide

 
• Ob_resistant

  insulin-resistant overweight/obese nondiabetic

 
• Ob_sensitive

  insulin-sensitive overweight/obese nondiabetic;

 
• T2D

  type 2 diabetes mellitus

 
• TGR-5

  Takeda G protein–coupled receptor

Acknowledgments

We thank Sally Coulter for performing the BA assays and Aimun Xu for analysis of adipokines.

Financial Support: The study was funded by a Diabetes Australia Research Trust Millennium Grant and, in part, by a National Health and Medical Research Council (NHMRC) Program Grant.

Author Contributions: R.H.B. analyzed the data, prepared the figures, and drafted the manuscript; K.T.T. recruited participants, collected and analyzed the data, and provided statistical and editorial input into the manuscript; J.G. provided assistance with bile acid assays and editorial input into the manuscript; D.S.B. provided assistance with data collection and editorial input into the manuscript; G.M.K. provided statistical assistance with data analysis; D.J.C. and D.E.J. provided editorial input into the manuscript; J.R.G. oversaw the project’s execution, data collection and analysis, and along with all the authors, endorsed the final manuscript draft for publication.

Additional Information

Disclosures: The authors have nothing to disclose.

Data Availability

Some or all data sets generated during and/or analyzed during the present study are not publicly available but are available from the corresponding author on reasonable request.