Conceived and designed the experiments: DMM DR KC. Performed the experiments: DMM JCF DR JLB CP KC. Analyzed the data: DMM JCF JCW. Contributed reagents/materials/analysis tools: DMM JCF DR JCW JLB CP KC. Wrote the paper: DMM JCF DR JCW KC.
Current address: Human Health and Nutritional Sciences, University of Guelph, Guelph, Ontario, Canada
Some of the authors are employed by a company (metanomics GmbH or metanomics Health GmbH); however, this work was performed purely in the spirit of a scientific collaboration and not as a service.
Roux-en-Y gastric bypass (RYGB) surgery is associated with weight loss, improved insulin sensitivity and glucose homeostasis, and a reduction in co-morbidities such as diabetes and coronary heart disease. To generate further insight into the numerous metabolic adaptations associated with RYGB surgery, we profiled serum metabolites before and after gastric bypass surgery and integrated metabolite changes with clinical data.
Serum metabolites were detected by gas and liquid chromatography-coupled mass spectrometry before, and 3 and 6 months after RYGB in morbidly obese female subjects (n = 14; BMI = 46.2±1.7). Subjects showed decreases in weight-related parameters and improvements in insulin sensitivity post surgery. The abundance of 48% (83 of 172) of the measured metabolites changed significantly within the first 3 months post RYGB (p<0.05), including sphingosines, unsaturated fatty acids, and branched chain amino acids. Dividing subjects into obese (n = 9) and obese/diabetic (n = 5) groups identified 8 metabolites that differed consistently at all time points and whose serum levels changed following RYGB: asparagine, lysophosphatidylcholine (C18:2), nervonic (C24:1) acid, p-Cresol sulfate, lactate, lycopene, glucose, and mannose. Changes in the aforementioned metabolites were integrated with clinical data for body mass index (BMI) and estimates for insulin resistance (HOMA-IR). Of these, nervonic acid was significantly and negatively correlated with HOMA-IR (p = 0.001, R = −0.55).
Global metabolite profiling in morbidly obese subjects after RYGB has provided new information regarding the considerable metabolic alterations associated with this surgical procedure. Integrating clinical measurements with metabolomics data is capable of identifying markers that reflect the metabolic adaptations following RYGB.
Obesity is characterized by the accumulation of excess body fat to an extent that health is adversely affected via the development of co-morbidities. Due to a scarcity of validated and safe therapies, Roux-en-Y gastric bypass (RYGB) has become an increasingly effective treatment for severely obese patients
Unraveling the immediate and long-term adaptations associated with gastric bypass surgery has proved challenging, predominantly because the consequences of this procedure include caloric restriction, diminished nutrient absorption, reduced adipose mass, modified gut hormone signaling, and changes in whole-body glucose metabolism that can each cause numerous physiological and metabolic adaptations. Because of a growing interest to treat type-II diabetes with gastric bypass surgery, the rapid and long-term improvement of insulin sensitivity and the reduction of diabetes in subjects post surgery has become a primary axis of interest. Several hypotheses have been postulated to explain the improved insulin sensitivity witnessed post surgery, and include the altered secretion of gut hormones
One informative approach that has not yet been used to study the metabolic adaptations following RYGB is metabolite profiling, despite the fact that this approach has successfully generated new knowledge regarding the metabolic modifications associated with obesity and diet-induced weight loss
The Ethics Committees of the Hôtel-Dieu Hospital approved the clinical investigations and all subjects gave written consent.
Fourteen obese women (11 Caucasian, 2 Caribbean from French Antilles, and 1 African) involved in a gastric surgery program were prospectively recruited from the Ile-de-France region between 2005 and 2006 in the Department of Nutrition, Center of Reference for Medical and Surgical Care of Obesity, CREMO, Hôtel-Dieu (Paris, France). Patients met the criteria for obesity surgery, i.e. BMI≥40 kg/m2 or ≥35 kg/m2 with at least two co-morbidities (hypertension, type-II diabetes, dyslipidemia or obstructive sleep apnea syndrome). The preoperative evaluation included medical history, physical, nutritional, metabolic, cardiopulmonary, and psychological assessments. Subject weight was stable (i.e. variation of less than ±2 kg) for at least 3 months prior to operation. Subjects did not demonstrate evidence of acute or chronic inflammatory disease, infectious diseases, viral infection, cancer and/or known alcohol consumption (>20 g per day). According to the criteria of fasting glycemia over 7 mM or the use of an anti-diabetic drug, 5 subjects were also type 2 diabetics (4 Caucasians and 1 Caribbean from French Antilles). These five subjects (referred to hereon as the OB/D group) were treated with metformin and hypolipemic drugs (either fibrates or statins). Two subjects were additionally treated with insulin. The duration of diabetes in these 5 subjects was 7.4±1.0 years (range between 5–11 years). In the non-diabetic group (referred to hereon as the OB group), none were treated with hypolipemic drugs. Furthermore, an oral glucose tolerance test (OGTT) was systematically performed before RYGB and confirmed that all patients in the OB group had glucose levels less than 11 mmol/l (200 mg/dl) in the two hours following a 75 g oral glucose challenge.
All patients were assessed prior to Roux-en-Y surgery (i.e. baseline or T0) and at 3 months and 6 months post surgery (T3 and T6). Blood samples were obtained at each time point and stored at −20°C until used to assess lipid, insulin and glucose values (enabling the determination of insulin sensitivity parameters), leptin, and many other factors, outlined in
Caloric intake and macronutrient proportions were evaluated by a registered dietician in the hospital nutrition department. Multivitamins and iron supplements were provided to avoid deficiencies, which are a well-known secondary effect of this bariatric surgery
Three types of mass spectrometry analyses were applied to all samples. GC-MS (gas chromatography-mass spectrometry; Agilent 6890 GC coupled to an Agilent 5973 MS-System, Agilent, Waldbronn, Germany) and LC-MS/MS (liquid chromatography-MS/MS; Agilent 1100 HPLC-System (Agilent, Waldbronn, Germany) coupled to an Applied Biosystems API4000 MS/MS-System (Applied Biosystems, Darmstadt, Germany)) were used for broad profiling, as described in
Technical reference samples were measured in parallel with the study samples in order to allow the relative quantification of metabolites in the study samples. These technical reference samples were generated by randomly pooling serum from 15 healthy female controls (average age = 48.9±2.1; BMI range of 21-24). A relative quantification for each metabolite was obtained by normalizing peak intensity in the study samples to the median peak intensity of the corresponding metabolite in the technical reference samples measured in the same batch.
Proteins were removed from serum samples (60 ul) by precipitation. Subsequently polar and non-polar fractions were separated for both GC-MS and LC-MS/MS analysis by adding water and a mixture of ethanol and dichloromethane. For GC-MS analyses, the non-polar fraction was treated with methanol under acidic conditions to yield the fatty acid methyl esters derived from both free fatty acids and hydrolyzed complex lipids. The polar and non-polar fractions were further derivatized with O-methyl-hydroxyamine hydrochloride (20 mg/ml in pyridine, 50 ul) to convert oxo-groups to O-methyloximes and subsequently with a silylating agent (MSTFA, 50 ul) before GC-MS analysis
Steroids and their related metabolites were measured by online SPE-LC-MS/MS. Catecholamines and their related metabolites were measured by online SPE-LC-MS/MS, as described by Yamada et al
An ANOVA model was developed to determine differences in metabolite abundance over time (T0, T3, and T6) and between groups (OB vs. OB/D). Metabolite data was log10-transformed and a mixed-effects ANOVA was conducted for univariate data analysis. Two modeling approaches were applied, where both approaches began with fixed effects: sample_age (numeric, to account for serum storage time), group (categorical: OB or OB/D) and time (categorical: T0, T3, T6), and with random patient effect (categorical, patient 1–14). The second more general approach comprised an additional group:time interaction to account for possible subgroup-specific changes over time (i.e. RYGB having different effects on OB and OB/D subjects). The Akaike information criterion (AIC) was applied for metabolite-specific stepwise model evaluation and reduction of model complexity with regards to fixed effects (all fixed effects except time were considered optional). Experimental groups were compared and statistically evaluated by t-statistics of model contrasts. The estimated models were also used for data correction with regards to sample_age by metabolite-specific subtraction of linear sample_age estimations from log10-transformed data when required (i.e. when sample_age was not removed by AIC model complexity reduction). Analysis was conducted with statistical software R version 2.4.1 (Copyright (C) 2006 “The R Foundation for Statistical Computing”,
A profile is characterized by 3 dots, which represent T0 (prior to RYGB), T3 and T6 (post surgery). An angled slope between two time points indicates a significant change (p<0.05) and a flat slope between two time points indicates non-significant changes. Based on data derived from the mixed-effects ANOVA using all 14 subjects together. (*1): Structure annotation is based on strong analytical evidence (combinations of chromatography, mass spectrometry, chemical reactions, deuterium-labeling, database and literature search, as well as comparisons to similar/homologue/isomeric reference compounds). (*2): Metabolite exhibits identical qualitative analytical characteristics (chromatography and mass spectrometry) compared to status (*1). Further structural and analytical investigations of this metabolite - also in comparison to structurally identified or status (*1) metabolites - are still pending.
Partial least squares discriminant analysis (PLS-DA) of sample_age-corrected data for the 172 known metabolites was performed for classification analysis according to time (T0 vs. T3 vs. T6) and group (OB vs. OB/D, time-specific) using Umetrics SIMCA-P software (Version 11.0, Umetrics AB, Umea, Sweden). Model performance was quantified by Q2 values from leave-one-subject-out cross validation.
The distribution of clinical data prior to surgery was tested using the Shapiro-Wilk W Test. Non-parametric statistical tests were used to assess clinical data over time (Mann-Whitney U-Test) and between groups (Friedman test); performed using GraphPad Prism version 4.0 software (GraphPad Software, Inc., California, USA). All data is presented as mean±SEM.
Correlations between the abundance of 8 metabolites and two clinical parameters (BMI or HOMA-IR) were assessed with a Spearman correlation using JMP statistical software v5.1.2 (SAS Institute Inc., Cary, NC, USA).
Our study population consisted of 9 obese (OB) subjects and 5 obese/diabetic (OB/D) female subjects, with an average BMI of 46.2±1.7. The OB/D subjects were all treated with metformin and 2 of these subjects were additionally treated with insulin. RYGB resulted in the expected and significant decreases in body mass index (BMI), body weight (kg), fat mass (kg), and fat free mass (kg) from T0 → T3 → T6 (
HOMA-IR was estimated for 12 subjects, 9 OB and 3 OB/D subjects. The OB/D subjects were treated with metformin and not with insulin. A significant reduction (T0 → T3, p = 0.014; T0 → T6, p = 0.001; T3 → T6, p = 0.123) in HOMA-IR occurred following RYGB, as assessed using a Friedman test. Prior to RYGB, significant variability in HOMA-IR estimations was observed between subjects (because OB and OB/D subjects are combined); however, post RYGB, all subjects demonstrated a major improvement in insulin sensitivity (illustrated by smaller error bars). Box plots indicate no outlying data (i.e. above or below the whiskers), and the band in the middle of the box indicates the median. **
Parameter | T0 | T3 | T6 |
45.4±3.6 | |||
46.2±1.7 | 38.7±1.6** | 35.1±1.7 ** | |
125.4±4.2 | 104.6±3.9 ** | 95.1±4.0 ** | |
59.3±3.5 | 46.9±3.2 ** | 39.2±2.9 ** | |
60.2±1.5 | 54.1±1.7 ** | 51.9±1.8** | |
34.3±0.4 | 33.6±0.4** | 32.8±0.3** | |
4.7±0.3 | 4.3±0.3 | 4.4±0.3 | |
1.5±0.1 | 1.3±0.1 ** | 1.5±0.1 | |
1.3±0.2 | 1.1±0.1** | 1.0±0.1** | |
1.4±0.1 | n/a | 1.4±0.1 | |
69.1±6.5 | 37.9±5.2** | 24.2±4.0** | |
1861±149.2 | 991±101.6** | 1106±98.4** | |
19.2±1.1 | 19.6±0.8 | 19.1±1.1 | |
33.1±2.0 | 33.7±1.9 | 32.0±1.8 | |
47.6±2.2 | 46.7±1.9 | 47.5±2.8 | |
5.44±0.34 | 4.92±0.17* | 4.53±0.10** | |
15.6±3.5 | 7.8±1.0** | 5.8±0.7** | |
2.0±0.4 | 1.0±0.1** | 0.7±0.1** | |
70.3±10.0 | 129.8±29.7** | 207.8±66.2** | |
134.5±16.9 | 101.0±12.1** | 92.5±7.6** |
All 14 subjects are included in the analysis of parameters related to body weight and lipids. Decreases after RYGB were observed for BMI, weight, fat mass, fat free mass, resting energy expenditure and triglycerides. Leptin was also decreased significantly. While HDL-cholesterol decreased from T0 to T3, HDL-Cholesterol levels recovered by T6 and are confirmed by the lack of change in Apo-A1 levels. Total caloric intake decreased after RYGB; however, the relative proportion of lipid, carbohydrate, and protein consumed remained stable. When considering the 12 subjects not treated with insulin, glucose and insulin levels decreased post RYGB. Estimates for HOMA-IR and HOMA%B decreased while HOMA%S increased after surgery. Data presented as mean±standard error. * represents p<0.1 and ** represents p<0.05, assessed by a Friedman test.
Subjects in the OB/D group showed further improvements in relation to their diabetic status. Prior to RYGB, HbA1c levels were 8.1%±0.3 and at T3 the HbA1c levels dropped to 6.8%±0.2 (p = 0.06). Blood glucose levels in three of the five subjects normalized after RYGB and anti-diabetic treatments were stopped, indicating a resolution of diabetes. The two OB/D subjects treated with insulin prior to surgery showed improvements post surgery (HbA1c levels dropped from 9.4% and 9.0% at T0 to 7.1% and 7.8% at T3); however insulin therapy was maintained and diabetes was not resolved.
Partial least squares discriminant (PLS-DA) analysis of the 172 known metabolites measured in the serum of 14 subjects led to a clear separation between the metabolite profiles before (T0), 3 months after (T3), and 6 months after (T6) gastric bypass surgery (similarly observed with principal component analysis – data not shown). The model, consisting of four components, had a R2X(cumulative) of 0.430 and a Q2(cumulative) of 0.806 (selected by leave-one-subject-out cross validation according to maximal Q2(cum)).
The first component distinctly classifies T0 vs. after RYGB (Q2 = 0.423), the second component further distinguishes T3 and T6 (Q2(cum) = 0.640 for two-component PLS-DA model). Model performance was quantified by Q2 values from leave-one-subject-out cross validation. Q2 depicts the fraction of the total variation that is predicted by each PLS component, while Q2(cum) depicts the cumulated fraction of the total variation predicted by the model.
As previously mentioned, the population consisted of 9 OB and 5 OB/D subjects; therefore, we subsequently compared the metabolite profiles of these two subgroups. PLS-DA displayed a separation between OB and OB/D groups at each time point before and after surgery (1 component PLS-DA model selected by cross validation for T0 and T3, 2 component model for T6; T0 → R2X = 0.156 and Q2 = 0.528; T3 → R2X = 0.146 and Q2 = 0.529; and T6 → R2X = 0.149 and Q2_comp1 = 0.585, Q2(cum) = 0.711). While many metabolites may contribute to this separation, we considered only those metabolites that differed between OB and OB/D subjects with p<0.05 at each time point according to the mixed-effects ANOVA, allowing group:time-interactions when statistically adequate. If the differences in metabolite abundance between OB and OB/D groups did not vary significantly with time (i.e. group:time interaction excluded by AIC), the same p-value appears at all three time points. We identified 33, 32, and 28 metabolites that differed significantly (p<0.05) between OB and OB/D subjects at T0, T3, and T6, respectively (
The top two metabolites distinguishing OB and OB/D subjects were identified with the mixed-effects ANOVA allowing for group:time-interactions, where * indicates
Prior to surgery (T0) | 3 months post surgery (T3) | 6 months post surgery (T6) | ||||||
Metabolite | p-value | Greater in | Metabolite | p-value | Greater in | Metabolite | p-value | Greater in |
0.00004 | OB | 0.0002 | OB | Valine | 0.0004 | OB/D | ||
0.0020 | OB/D | 0.0020 | OB/D | 0.0020 | OB/D | |||
Testosterone | 0.0024 | OB | 11-Deoxycortisol | 0.0032 | OB/D | Leucine | 0.0030 | OB/D |
Normetanephrine | 0.0040 | OB | Hypoxanthine (minor: Inosine) | 0.0041 | OB/D | 0.0032 | OB | |
0.0043 | OB | 0.0043 | OB | myo-Inositol-2-phosphate (minor: Fructose-6-phosphate, Glucose-6-phosphate, myo-Inositol-1-phosphate, myo-Inositol-4-phosphate) | 0.0033 | OB/D | ||
0.0046 | OB/D | 0.0046 | OB/D | 0.0043 | OB | |||
0.0059 | OB | 0.0059 | OB | 0.0046 | OB/D | |||
Androstenedione | 0.0062 | OB | Phosphatidylcholine (C18:0, C18:1) (*2) | 0.0064 | OB/D | 0.0059 | OB | |
Linoleic acid (C18:cis |
0.0070 | OB | 0.0088 | OB/D | Proline | 0.0061 | OB/D | |
Phosphatidylcholine No 02 (*2) | 0.0083 | OB/D | 0.0091 | OB/D | Xanthine | 0.0082 | OB/D | |
0.0091 | OB/D | 0.0091 | OB/D | 0.0091 | OB/D | |||
0.0091 | OB/D | Valine | 0.0115 | OB/D | 0.0091 | OB/D | ||
5-Hydroxy-3-indoleacetic acid (5-HIAA) | 0.0116 | OB/D | 0.0133 | OB/D | Phenylalanine | 0.0112 | OB/D | |
0.0133 | OB/D | 0.0140 | OB/D | 0.0133 | OB/D | |||
0.0140 | OB/D | Ribose | 0.0148 | OB/D | 0.0140 | OB/D | ||
Phosphatidylcholine (C18:2, C20:4) (*2) | 0.0146 | OB/D | 0.0167 | OB/D | 0.0167 | OB/D | ||
0.0167 | OB/D | 0.0176 | OB/D | 0.0176 | OB/D | |||
0.0176 | OB/D | Taurine | 0.0181 | OB/D | Isoleucine | 0.0209 | OB/D | |
3-O-Methylsphingosine (minor: Sphingolipids, erythro-Sphingosine, threo-Sphingosine) (*1) | 0.0176 | OB | myo-Inositol-2-phosphate (minor: Fructose-6-phosphate, Glucose-6-phosphate, myo-Inositol-1-phosphate, myo-Inositol-4-phosphate) | 0.0195 | OB/D | Phosphatidylcholine (C18:0, C18:1) (*2) | 0.0223 | OB/D |
threo-Sphingosine (minor: Sphingolipids) | 0.0232 | OB | Citrulline | 0.0241 | OB/D | Ornithine (minor: Arginine, Citrulline) | 0.0237 | OB/D |
Threitol | 0.0241 | OB/D | Xanthine | 0.0264 | OB/D | 0.0265 | OB | |
erythro-Sphingosine (minor: Sphingolipids) | 0.0243 | OB | 0.0265 | OB | 0.0296 | OB/D | ||
5-O-Methylsphingosine (minor: Sphingolipids, erythro-Sphingosine, threo-Sphingosine) (*1) | 0.0249 | OB | 0.0296 | OB/D | 0.0299 | OB | ||
N-Acetylneuraminic acid, lipid fraction | 0.0256 | OB | 0.0299 | OB | Taurine | 0.0309 | OB/D | |
0.0265 | OB | gamma-Linolenic acid (C18:cis |
0.0300 | OB/D | 0.0325 | OB/D | ||
0.0279 | OB | Phosphatidylcholine (C18:1, C18:2) (*2) | 0.0327 | OB/D | 0.0376 | OB/D | ||
0.0296 | OB/D | Threitol | 0.0366 | OB/D | Citrulline | 0.0439 | OB/D | |
0.0299 | OB | 0.0376 | OB/D | 0.0468 | OB/D | |||
Phytosphingosine | 0.0302 | OB | Sucrose | 0.0390 | OB/D | |||
0.0376 | OB/D | Ornithine (minor: Arginine, Citrulline) | 0.0409 | OB/D | ||||
0.0468 | OB/D | 0.0468 | OB/D | |||||
Arginine | 0.0469 | OB | Phosphatidylcholine (C16:0, C16:0) (*2) | 0.0498 | OB | |||
Sphingomyelin No 02 (*2) | 0.0484 | OB |
Metabolites were identified using a mixed-effects ANOVA: 33 metabolites at T0, 32 metabolites at T3, and 28 metabolites at T6 (p<0.05), with indication of whether serum levels are higher in obese (OB) or obese/diabetic (OB/D) subjects. Those metabolites present in all three lists are indicated in bold font. The profiles for metabolites whose abundance changed post surgery are found in
The 15 remaining metabolites that differed between OB and OB/D subjects at all time points were asparagine, cysteine, lysophosphatidylcholine (C18:2), phosphatidylcholine (C16:0, C18:2), nervonic acid (C24:cis
We then integrated clinical and metabolite data from all three time points in the 12 non-insulin treated subjects with the goal of determining whether any of the aforementioned metabolites could be used as ‘candidate’ markers that reflect the clinical adaptations seen following RYGB. Independent Spearman correlations were performed for each of the 8 aforementioned metabolites versus BMI and HOMA-IR. Of the 8 metabolites examined, we identified a significant negative correlation between nervonic acid and HOMA-IR (p = 0.001, R = −0.55), i.e. HOMA-IR decreased (
A significant increase in serum abundance of nervonic acid occurred following RYGB in both OB and OB/D subjects, as assessed using a mixed-effects ANOVA with group:time interaction (**
Gastric bypass surgery results in significant metabolic changes associated with weight loss, improved insulin sensitivity and glucose homeostasis, and a reduction in co-morbidities such as diabetes and coronary heart disease. A retrospective study in approximately 10,000 subjects who had undergone gastric bypass surgery revealed that mortality related to diabetes decreased by over 90%, leading to this medical procedure being proposed as a means to treat diabetes, even in subjects who are obese rather than morbidly obese
All 14 patients experienced an improvement in the clinical parameters measured. Decreases in BMI, body weight, fat mass, fat free mass, triglycerides, insulin resistance (HOMA-IR), and leptin were observed following RYGB. Subjects also experienced a decrease in resting energy expenditure. Furthermore, while total caloric intake decreased dramatically post surgery, the relative proportion of protein, lipid, and carbohydrate consumed in the diet remained stable. The OB/D subjects had additional physiological improvements post surgery related to their diabetes status. OB/D subjects experienced a drop in HbA1C levels at 3 months post RYGB. Three of the 5 OB/D subjects demonstrated a resolution of diabetes, as suggested by the normalization of glucose levels and the fact they no longer needed anti-diabetic and dyslipidemia treatments after 3 months. The remaining 2 OB/D subjects demonstrated improvements in their diabetic status, but insulin therapy was still required. Interestingly, these two subjects both had diabetes for 11 years prior to surgery and higher pre-surgery HbA1c levels in contrast with the other three subjects, who had diabetes for 5 years prior to surgery and lower pre-surgery HbA1c levels. This suggests the RYGB surgery may be more efficacious for resolving diabetes if performed soon after diagnosis in subjects with low HbA1C levels.
In the present study we have identified 83 serum metabolites whose abundance changed significantly following RYGB; where the majority of these metabolites corresponded to lipids and amino acids. The use of a stringent Bonferroni adjustment to correct for multiple testing highlighted the large number of metabolites that decreased immediately post surgery and then did not change between T3 and T6 (
While many lipid species were found to increase in subjects post RYGB, several lipids were of particular interest, namely nervonic acid and various sphingosine species (3-O-methylsphingosine, threo-sphingosine, erythro-sphingosine, and 5-O-methylsphingosine). In light of recent evidence regarding the role of intramyocellular lipids (IMCL) and insulin resistance, it is interesting to speculate that plasma increases in the aforementioned lipids reflect their mobilization from non-adipose tissues such as the muscle. Elevated IMCL deposition has been observed in obese individuals, obese subjects with type 2 diabetes and insulin resistant subjects
Our work also revealed that branched chain amino acids (BCAA) serum levels decreased following RYGB. The decrease in valine was paralleled by an increase in serum β-aminoisobutyric acid, a product of valine catabolism. Our results are in line with a previous report indicating that plasma BCAA decrease by approximately 35% following gastric bypass surgery
Because this is the first study examining metabolic changes following RYGB, we have chosen to briefly discuss some metabolites that were significantly different at an unadjusted 0.05 p-level to highlight potentially promising metabolites for further study. These metabolites provide intriguing and novel information illustrating the many physiological adaptations arising post RYGB; although their relationship with RYGB, morbid obesity, and diabetes has not been extensively studied. For example, RYGB was previously shown to improve hypothyroidism
We further studied whether changes in metabolite abundance were associated with diabetes by separating our cohort into obese and obese/diabetic subjects. We felt this analysis was justified and valid for several reasons. Firstly, subgroup analysis between OB and OB/D subjects was only performed when statistical significance was achieved according to Akaike information criterion. Secondly, metabolite profiling revealed that glucose was greater in OB/D subjects compared to OB subjects at all three time points (p<0.05). Thirdly, the metabolite that differed the most between OB and OB/D subjects was 1,5-anhydrosorbitol (p<0.001), a biomarker currently used for assessing glycemic control
In summary, applying global metabolite profiling to the study of subjects before and after RYGB has provided considerable new information regarding the numerous alterations to serum metabolite levels arising with this surgical procedure. While we have identified metabolites not previously associated with the physiological and metabolic alterations arising with gastric bypass surgery, it is imperative that the next step taken in unraveling the physiological improvements following RYGB uses a systems biology approach to study metabolite fluxes between serum, muscle, liver, and adipose tissue. The additional knowledge regarding the communication between tissues and body fluids will permit a better understanding of whether metabolite changes are responsible for, or an indirect effect of, clinical improvements.
We thank Mme. Christine Baudouin, Dr. Florence Marchelli, and Mme. Patricia Ancel for their involvement in patient recruitment, data collection and sampling at the Center of Research on Human Nutrition, Paris Pitié-Salpêtrière Hospital. We thank Prof. Yves Boirie for helpful discussions regarding the interpretation of amino acid changes.