Funding agencies: This research was supported by grants from the Swiss National Science Foundation (Grant No. 320000-118330 and Grant No. 320000-120020), and by the Stiftung zur Förderung der gastroenterologischen Forschung, and an unrestricted grant of Hoffmann-La Roche.
Disclosure: The authors declared no conflict interest.
Increased delivery of bile acid salts (BA) to distal L-cells and altered TGR5 receptor activation may contribute to the early and substantial increases in gut peptide secretion seen after bariatric surgery. To further elucidate a potential role of BA in the secretion of GLP-1 and PYY, we analyzed plasma BA concentrations in 14 morbidly obese patients undergoing gastric bypass or sleeve gastrectomy in a prospective, randomized 1-year trial.
Design and Methods
Patients received a standard test meal and blood was collected before and after eating, prior to, and 1 week, 3 months, and 12 months after surgery.
Pre-surgery, basal BA concentrations were significantly lower in bariatric patients than in healthy controls. One year post-surgery, bariatric patients expressed variably increased BA concentrations (gastric bypass patients ∼2 fold increase, P ≤ 0.05). However, whereas in both patient groups, marked increases in GLP-1 and PYY and improved glycemic control was seen already 1 week and 3 months post-surgery, changes in plasma BA followed a different pattern: basal and postprandial plasma BA concentrations increased much slower, more progressively with significant increases only 1-year post-surgery.
Based on these findings, BA do not appear to be key mediators of the early increase in GLP-1 and PYY response in post-bariatric patients.
Bile acid salts (BA) have traditionally been considered mediators of lipid absorption and cholesterol metabolism. However, in recent years, this picture has changed dramatically and it now appears that BA are complex metabolic molecules with direct effects not only on lipid metabolism but also on gut peptide secretion and glycemic control .
Katsuma et al.  and Parker et al.  showed recently in enteroendocrine STC-1 and GLUTag cells that BA trigger the secretion of glucagon-like peptide-1 (GLP-1) through TGR5, a G-protein coupled receptor expressed on enteroendocrine L-cells, in a dose-dependent manner and that reduced expression of TGR5 results in reduced GLP-1 release. Under physiological conditions, GLP-1 is released in response to food ingestion, promotes insulin secretion and suppresses glucagon release from the endocrine pancreas . In addition, GLP-1 delays gastric emptying and suppresses appetite, and thus, plays a central role in glycemic control [4, 5]. The first functional evidence for a link between BA, TGR5, GLP-1, and improved glycemic control came from Thomas et al. . They found that glucose tolerance was reduced by TGR5 deficiency and improved by TGR5 over-expression in high-fat diet-fed mice via increased GLP-1 and insulin secretion. In 2012, Adrian et al.  reported in humans that intrarectal infusion of taurocholic acid (TCA) increases plasma GLP-1 and PYY, an effect that was associated with increased insulin secretion, decreased blood glucose, and a suppression in food intake.
It has also been speculated that increased delivery of BA to distal L-cells may contribute to the early and substantial increase in gut peptide secretion and the initial improvement in glycemic control seen after bariatric surgery [8-11]. For example, in a cross-sectional analysis, Patti et al.  reported that fasting plasma BA concentrations were higher in patients with gastric bypass 2-4 years post-surgery compared to both overweight and severely obese and that total plasma BA were inversely correlated with 2 h post-meal glucose concentrations and positively correlated with GLP-1.
So far no data are available investigating the physiological postprandial changes of BA in response to a test meal in obese patients before and after bariatric surgery throughout a 1-year period. In this pilot investigation, we have assayed available samples from our previous study  to address the hypothesis that peripheral fasting and postprandial plasma BA concentrations are associated with the marked and early increases in GLP-1 and PYY secretion and the associated improvement in glycemic control seen in these patients shortly after surgery. We have analyzed postprandial plasma BA concentrations in healthy control subjects and in patients undergoing two anatomically different bariatric procedures (laparoscopic Roux-en-Y gastric bypass [LRYGB] and laparoscopic sleeve gastrectomy [LSG]) in a prospective, randomized 1-year trial. The interpretation of the present data is based on the assumption that postprandial changes in peripheral plasma BA concentrations reflect, at least to some extent, the postprandial changes in luminal BA concentrations [13-15].
Subjects and patients
All studies were performed in accordance with the Declaration of Helsinki. The Local Research and Ethics Committee in Basel approved the study that included 14 morbidly obese patients. Subjects were part of a larger prospective randomized trial with effectiveness and safety of LRYGB and LSG as primary endpoints . All patients were informed in detail about risks and benefits of each procedure, and each provided written, informed consent. Computer-generated random numbers were used to assign the type of surgery. All operations were performed laparoscopically and by the same surgeon. The LRYGB technique included a small gastric pouch with a 25-mm circular pouch-jejunostomy to a 150 cm antecolic Roux limb and an exclusion of 50 cm of biliopancreatic limb. The LSG was done along a 35-F bougie from the angle of His to approximately 3-4 cm orally to the pylorus. As a control group, the study included six healthy, non-smoking volunteers. The body weight of all control subjects was in the normal range for age, sex and height, and stable for at least 3 months. Table 1 provides baseline demographics.
Table 1. Baseline demographics
Healthy subjects (n = 6)
LRYGB (n = 7)
LSG (n = 7)
P ≤ 0.05 versus healthy subjects. No significant differences between laparoscopic Roux-en-Y gastric bypass (LRYGB) and laparoscopic sleeve gastrectomy (LSG) patients were detected.
ND = not determined, BMI = body mass index, Data are means ± SEM.
The study was conducted as a randomized, prospective, parallel-group trial (Clinical Trial Registration: ClinicalTrials.gov NCT00356213). All bariatric patients underwent complete evaluation before the operation and during follow-up, including medications, nutritional behavior, anthropometric and clinical parameters, and blood sampling for laboratory tests. For meal studies, bariatric patients were admitted to the Clinical Research Centre before the operation, and 1 week, and 3 and 12 months after the operation. Healthy subjects reported to the Clinical Research Centre on a separate day. After fasting overnight for at least 10 h, an antecubital vein catheter was inserted for phlebotomy. After taking the fasting sample, a 424 kcal (1775 kJ) liquid test meal containing 15 g carbohydrates, 25 g proteins, and 28 g fat was served to stimulate hormone release. Blood was drawn at the following times: pre-dose, 15 min, and 180 min. Samples (10 ml/withdrawal) were collected into EDTA tubes containing aprotinin at a final concentration of 500 KIU/ml and a DPP-IV inhibitor (50 µmol/l). Samples were immediately processed and kept on ice to retard peptide breakdown. After centrifugation at 4°C, plasma samples were kept frozen at −20°C until analysis.
The hormones investigated were GLP-1, PYY, and insulin. GLP-1 was measured by a commercially available ELISA kit (Linco Research Inc., St. Charles, MO). This kit is highly specific for measuring active GLP-1 but does not detect other GLP-1 forms (e.g., 1-36 amide, 1-37, 9-36 amide, or 9-37). The intra- and inter-assay variabilities were below 9.0 and 13.0%, respectively. The lowest level of GLP-1 detectable was 0.25 pmol/l. PYY was measured with a commercially available kit (Linco Research Inc., St. Charles, MO). The guinea pig antibody displays 100% cross-reactivity with human PYY [1-36] and human PYY [3-36], but not with human pancreatic polypeptide, neuropeptide Y, leptin, or ghrelin. The lowest level of PYY detectable was 10 pg/ml. The intra- and inter-assay variabilities were below 9.0 and 9.0%, respectively. Insulin was measured with a commercial radioimmunoassay (CisBio International, Bagnols-sur-Cèze, France). The lowest level of insulin detectable was 4.6 μU/ml. The intra- and inter-assay coefficients of variation were, respectively, 12.2 and 9.0%. Blood glucose concentrations were measured by a commercial hexokinase/glucose-6-phosphate-dihydrogenase method (Roche, Basel, Switzerland). More details for each assay have been described previously [16, 17].
BA analysis was performed using an established method modified for application to human plasma . Briefly, after removal of neutral sterols from plasma, the samples were hydrolyzed, acidified, and BA were extracted. A gas chromatography procedure was realized by injection of BA as their trimethyl silyl ether methyl ester into a gas chromatography-mass spectrometry instrument (QP 5000, Shimadzu, Kyoto, Japan). Hyodeoxycholic acid (HDCA) was used as internal standard. The standard substances of deoxycholic acid (DCA), cholic acid (CA), chenodeoxycholic acid (CDCA), and HDCA were purchased from Sigma-Aldrich (Munich, Germany). The mass spectrometric detection was realized in multi-ion current (DCA: m/z = 255.3 U, CA: m/z = 253.2 U, CDCA: m/z = 73.10 U, HDCA: m/z = 81.15 U) and quantification was carried out using data-handling software GCMSsolution® (Shimadzu, Kyoto, Japan). Due to hydrolysis of conjugated BA, measured CDCA, CA, and DCA concentrations represent unconjugated and conjugated BA. The sum of CDCA, CA, and DCA (here defined as total plasma BAs) reflect about 90% of plasma BA [19-22]. Lithocholic acid (LCA) concentrations were below the limit of detection.
Data analysis was performed using the statistical software package SPSS for Windows V. 19.0 (SPSS Inc., Chicago, IL). Values are reported as means ± SEM. All parameters were analyzed by calculating baseline values, time courses, and/or area under the plasma concentration time curves (AUC). The General Linear Model procedure of repeated measures ANOVA, using simple contrast, was used to test for significant differences in longitudinal changes from baseline. To test for differences between the two treatment groups and healthy subjects, one-way ANOVA followed by Bonferroni–Holm post-hoc comparisons were used. All data were log transformed when appropriate and tested for normality using the Shapiro–Wilk or Kolmogorov–Smirnov test methods. All tests were two-tailed, with P < 0.05 considered statistically significant. Correlations among parameters were assessed using Pearson's correlation coefficient. The coefficient of determination (r2) was obtained by squaring the correlation coefficient (r). Within-subject correlations between basal and postprandial BA concentrations and GLP-1, PYY, and body mass index (BMI) over the repeated visits were calculated using the methods described by Bland and Altman [23, 24].
Twelve months after surgery, both procedures, LRYGB and LSG, were followed by a marked reduction in BMI (P ≤ 0.05, Figures 1 and 5). The change in BMI between LRYGB and LSG subjects was not significantly different.
Pre-surgery, parameters of glycemic control were similar in LRYGB and LSG patients with no significant differences between the groups. LRYGB and LSG patients demonstrated elevated fasting insulin and glucose concentrations, an increased homeostasis model assessment of insulin resistance index (HOMA-IR) and slightly increased hemoglobin A1c (Table 1, Figure 2A–C). Post-surgery, in both LRYGB and LSG patients, improvement in glycemic control was seen already 1 week after surgery, with continued improvement during the next 12 months. The HOMA-IR was significantly decreased 12 months after surgery (P ≤ 0.05, Figure 2C). A comparison between LRYGB and LSG subjects showed no significant difference.
Plasma GLP-1 and PYY
Pre-surgery, time courses for GLP-1 and PYY documented an impaired postprandial secretory response to the test meal (Figure 3A–F). As early as 1 week post-surgery, GLP-1 and PYY secretion was markedly increased in both patient groups (P ≤ 0.05, Figures 3A–F and 5). The GLP-1 response was further increased 3 months post-surgery, whereas, PYY secretion slightly decreased but remained markedly elevated. A comparison between LRYGB and LSG subjects showed no significant difference.
Plasma bile acids
Pre-surgery, basal plasma BA concentrations were significantly lower in the obese (undergoing either LRYGB or LSG surgery) compared to healthy control subjects (P ≤ 0.01, Figure 4A). In contrast, total postprandial plasma BA (calculated as AUC) were not significantly different from healthy subjects (Figure 4B–D). One week post-surgery, total basal and postprandial plasma BA were decreased in LRYGB and in LSG patients. Three months post-surgery, a small but non-significant increase in total basal and postprandial plasma BA was seen in both groups (Figures 4A–D and 5). A marked and significant elevation in total basal and postprandial BA concentrations compared to pre-operative values and healthy control subjects was observed only 1-year post-surgery in LRYGB patients (P ≤ 0.05, Figures 4A–C and 5). In LSG patients, a significant elevation was observed one year post-surgery in total basal plasma BA concentrations only (Figure 4A). Individual plasma BA (CA, CDCA, and DCA) concentrations followed the same patterns (Table 2).
Table 2. Individual postprandial plasma bile acid salts (BA) (calculated as area under the curve, AUC) in healthy subjects and in laparoscopic Roux-en-Y gastric bypass (LRYGB) and laparoscopic sleeve gastrectomy (LSG) patients before, 1 week, 3 months, and 1 year after surgery
P ≤ 0.05 versus healthy subjects
P ≤ 0.05 versus pre-operative within group
P ≤ 0.05 LSG versus LRYGB at the same time point.
CA = cholic acid, CDCA = chenodeoxycholic acid, DCA = deoxycholic acid, NA = not applicable, AUC = area under the curve. Data are means ± SEM.
Correlations between bile acids, gut peptide secretion, and BMI
One year post-surgery, LRYGB and LSG patients expressed variably increased basal and postprandial plasma BA concentrations that were paralleled by a decrease in BMI, improvement in glycemic control and increased secretion of GLP-1 and PYY. However, whereas gut peptide secretion markedly increased already 1 week and 3 months post-surgery (i.e., peak effects were observed early after surgery within the first 3 months), the changes in total postprandial plasma BA followed a different more progressive pattern with small, non-significant increases in the first 3 months and a significant increase only 1 year after surgery (Figure 5A,B). No correlation between basal and postprandial plasma BA concentrations and the secretion of GLP-1 and PYY were found over the repeated visits (Supplementary Information Table 1). In contrast, a negative correlation between total basal plasma BA concentrations and BMI were found over time, as the progressive decrease in BMI was paralleled by a progressive increase in total basal plasma BA concentrations in both LRYGB and LSG patients (P ≤ 0.001, Figure 5C,D). Some correlations were also found between basal and postprandial plasma BA and GLP-1 and PYY at single time points, 3 and 12 month after surgery within each group (Supplementary Information Table 1), however, this analysis has to be considered preliminary because of the limited number of subjects included.
Many of the changes in body weight and glycemic control following gastric bypass have been attributed to markedly increased levels of postprandial GLP-1 and PYY secretion, and recent findings suggest that increased luminal delivery of BA may, at least in part, contribute to this effect. Here we report the physiological changes of basal and postprandial plasma BA concentrations in comparison to the changes in the secretion of GLP-1 and PYY and the associated alterations in BMI and glycemic control in obese patients before and after bariatric surgery. Two groups of patients, LRYGB and LSG, were randomized and followed for 1 year. We hypothesized that in LRYGB patients, the early and marked increase in GLP-1 and PYY secretion  is paralleled by a substantial rise in postprandial plasma BA concentrations. Based on previous reports [8, 11, 25, 26], we assumed that due to the surgical alterations in LRYGB patients, postprandially secreted BA progress down the biliopancreatic limb (duodenum and proximal jejunum) to distal L cells in a more undiluted state (without being mixed with ingested food) leading to increased availability of BA. This procedure should lead (i) to an increased potential of BA to engage TGR5 on L cells to stimulate the secretion of GLP-1 and PYY, and (ii) at the same time, to increased intestinal BA uptake with a consecutive increased spillover into the systemic circulation, and thus, a rise in peripheral plasma BA levels. In LSG patients, we did not expect major changes in postprandial BA concentrations because the integrity of the intestine is preserved in this type of surgery and BA recirculation not altered.
We found that 1-year post-surgery, both LRYGB and LSG patients expressed variably increased basal and postprandial plasma BA concentrations that were paralleled by increased secretion of GLP-1 and PYY, a decrease in body weight, and improvement in glycemic control. However, whereas the marked increase in GLP-1 and PYY secretion and the improvement in glycemic control was seen already 1 week after surgery with continued improvement over the next 12 months, the pattern was different for changes in BA concentrations, so that, basal and postprandial plasma BA concentrations increased much slower, more progressively with a significant increase only 1-year post-surgery. Based on the assumption that concentrations of BA measured in plasma, to some extent, reflect luminal BA concentrations [13-15], our findings do, thus, not support the hypothesis that luminal BA are key products of the proximal gut to mediate the initial increase in GLP-1 and PYY secretion. Similarly, our results do not suggest that the rapid improvement in glycemic control seen shortly after surgery is attributed to a BA activated GLP-1-dependent mechanism [8, 11, 25, 26]. The more progressive rise in plasma BA rather suggests that BA may contribute to metabolic improvements in a chronic stage later after surgery. In addition, BA may improve glycemic control via GLP-1-independent mechanisms: (i) Co-administration of TCA with glucose reduces postprandial glycemia also in GLP-1 receptor knockout mice ; (ii) Wu et al.  recently reported that in healthy humans, a 30 min intrajejunal infusion of TCA alone had no direct effects on GLP-1, blood glucose, insulin, C-peptide, or glucagon, moreover, a glucose-lowering effect of TCA that was present when TCA was co-infused with glucose occurred earlier than did stimulation of GLP-1; and (iii) In our own recent studies in healthy men, we also failed to observe major effects of intraduodenal infusions of physiological concentrations of CDCA on the secretion of GLP-1 and PYY . GLP-1-independent mechanisms may include activation of FXR and the release of FGF19. FXR knockout mice (FXR −/−) are glucose intolerant and insulin-resistant  and FGF19 has recently been shown to inhibit hepatic gluconeogenesis in mice . In contrast, studies arguing against a role of BA in the control of glucose metabolism come from Brufau et al. . They investigated changes in BA metabolism and explored whether they were associated with type 2 diabetes . Yet under conditions of profound perturbations in both glucose metabolism (type 2 diabetes patients) and BA metabolism (treatment with the BA sequestrant colesevelam), they were unable to find a correlation between markers of insulin resistance, glucose homeostasis and BA.
Strader et al.  and Kohli et al.  previously demonstrated in rats that surgical interposition of the distal ileum into the proximal jejunum was associated with higher BA levels in plasma as early as 6 weeks after surgery. Since the distal ileum has the highest expression of the apical sodium-dependent BA transporter and is the primary site for active BA reabsorption, they concluded that the higher absorptive capacity for BA in the “proximal ileal” section was responsible for the higher BA levels in plasma. In humans, Goldfine and colleagues reported that fasting plasma BA concentrations significantly increased in patients 1-year after gastric bypass  or were significantly higher 2-4 years post-surgery compared to obese individuals who have not undergone bariatric surgery . Nakatani et al.  and Pounaras et al.  found increased fasting plasma BA concentrations in gastric bypass patients 1 and 3 month after surgery.
Our data are consistent with the findings by Goldfine and colleagues , showing a significant increase in plasma BA concentrations 1-year post-surgery; however, we failed to observe a marked increase in basal or postprandial plasma BA within the first 3 month after surgery. We infer from these observations that altered anatomy after gastric bypass surgery may not be the only mechanism that is involved in elevated plasma BA concentrations post-surgery. The slow increase in plasma BA with a marked and significant elevation only 1-year post-surgery suggests a more progressive adaptive but not an initial mechanism. For example, BA metabolism may be linked to body adiposity—possibly adipokines, such as leptin or adiponectin. In line with this notion, we found that (i) basal plasma BA concentrations were higher in our healthy control subjects as compared to LRYGB or LSG patients before surgery, (ii) plasma BA concentrations were changed in LSG patients despite any anatomical alterations that would affect BA reabsorption, and (iii) in both LRYGB and LSG patients, the post-surgery increase in basal plasma BA concentration followed a pattern proportional to alterations in patients' BMI. Consistently, also Glicksman et al.  recently reported that obese subjects have a lower postprandial BA response relative to normal weight subjects.
Whereas, interspecies variations may account for the discrepancies between our data and the data in rats [it is well known that significant variations exist in BA homeostasis of humans, rodents, and other model organisms ], the reason for the discrepancies between our data and the data by Nakatani et al.  and Pounaras et al.  are unknown. Although both studies also report a more progressive rise in basal plasma BA, different peri- and post-bariatric anatomic and functional adaptive changes, genetics, altered dietary patterns, intestinal motility, mucosal hyperplasia, or changes of gut microflora may explain the differences. Moreover, different alterations in BA production or elimination may contribute to the observed BA patterns. Finally, the limited number of subject in our study may have prevented to detect small, but significant differences in the first 3 months post-surgery. The slight decrease in peripheral basal and postprandial BA concentrations 1 week post-surgery is most likely due to reduced food intake shortly after surgery. Thomas et al.  reported that fasting activates FXR, which in turn reduces BA biosynthesis and BA pool size. In support with a potential link between endogenous BA metabolism and body weight, we have recently also shown similar results for leptin, adiponectin, and basal insulin in these patients ; the secretion of these peptides was reduced 1-week post-surgery, and all of them are secreted proportionally to body fat mass .
There are some further considerations when interpreting the current data. First and most important, we measured only plasma concentrations of BA and, therefore, can provide only surrogate information about BA concentrations affecting the TGR5 receptor in the intestinal lumen. Nevertheless, the rise in plasma BA has been interpreted as direct 'spillover' from hepatic extraction and studies by Angelin and colleagues [13-15] suggest that the intestinal portal venous inflow can be monitored indirectly by analysis of peripheral BA plasma samples. Moreover, we recently observed that intraduodenally infused CDCA at different loads induces a dose-dependent rise in plasma BA concentrations . Second, although all our patients showed a disturbed glucose homeostasis pre-operatively (expressed as a pathological HOMA index), the use of non-diabetic patients may limit the interpretation of the role of BA on improved glucose homeostasis. Whether type 2 diabetics would show the same results remains to be determined. Third, the change of microbiota was not investigated here. Thus, with the current data, we cannot determine the role of this parameter on BA metabolism, gut peptide secretion and glycemic control. Fourth, although individual plasma BA (CA, CDCA, and DCA) followed the same secretory patterns, we did not characterize different BA conjugates that have different affinities for TGR5. Fifth, we only assessed associations between BA, gut peptides and parameters of glycemic control, however, BA may contribute to other metabolic improvements such as increased lipid oxidation and greater use of fat as energy substrate resulting in weight loss/maintenance in these patients [10, 37, 38]. Finally, as mentioned earlier the number of subjects included into the study was limited; however, these studies are difficult to perform as patients have very limited interest for repetitive examinations after surgery.
In conclusion, in this observational pilot study, we found that LRYGB and LSG patients expressed variably increased plasma BA concentrations that were paralleled by increased secretion of GLP-1 and PYY, a decrease in body weight, and improvement in glycemic control, 1-year post-surgery. However, whereas the marked increase in GLP-1 and PYY secretion and the improvement in glycemic control were seen already 1 week and 3 month after surgery, alterations in plasma BA followed a different pattern, so that basal and postprandial plasma BA concentrations increased much slower, more progressively with a significant increase only 1-year post-surgery. Based on the assumption that concentrations of BA measured in the plasma, to some extent, reflect luminal BA concentrations, the available data, thus, do not support the hypothesis that BA are key mediators of the early increase in postprandial GLP-1 and PYY secretion in post-bariatric patients. Future studies will be required to provide a clear conclusion with respect to the contribution of BA to gut peptide secretion.
We thank Kathleen Bucher for editorial assistance. We are indebted to Luisa Baselgia and the team at the Clinical Research Center for excellent assistance in performing the meal studies and Gerdien Gamboni for expert technical assistance in the laboratory.