Application of Metabolomics to Study Effects of Bariatric Surgery

Bariatric surgery was born in the 1950s at the University of Minnesota. From this time, it continues to evolve and, by the same token, gives new or better possibilities to treat not only obesity but also associated comorbidities. Metabolomics is also a relatively young science discipline, and similarly, it shows great potential for the comprehensive study of the dynamic alterations of the metabolome. It has been widely used in medicine, biology studies, biomarker discovery, and prognostic evaluations. Currently, several dozen metabolomics studies were performed to study the effects of bariatric surgery. LC-MS and NMR are the most frequently used techniques to study main effects of RYGB or SG. Research has yield many interesting results involving not only clinical parameters but also molecular modulations. Detected changes pertain to amino acid, lipids, carbohydrates, or gut microbiota alterations. It proves that including bariatric surgery to metabolic surgery is warranted. However, many molecular modulations after those procedures remain unexplained. Therefore, application of metabolomics to study this field seems to be a proper solution. New findings can suggest new directions of surgery technics modifications, contribute to broadening knowledge about obesity and diseases related to it, and perhaps develop nonsurgical methods of treatment in the future.


Introduction
Based on historical reports, origins of bariatric surgery date back to the 10th century, when King of León, Sancho I (called the Fat), was treated by famous Jewish doctor, Hasdai ibn Shaprut [1]. The homeland of bariatric surgery in modern meaning is the University of Minnesota. The first metabolic surgery was the jejunoileal bypass performed there by Arnold J. Kremen in 1954 [1,2]. Nevertheless, success has many parents; therefore, many great minds were involved in the bariatric surgery development. As the pioneers in this medicine field, Henry Buchwald, Richard L. Varco, Edward E. Manson (long recognized as the father of bariatric surgery), Allan C. Wittgrove, Nicola Scopinaro, Walter J. Pories, Picard Marceau, and Douglas S. Hess have to be mentioned [1,3]. Of course, they are a small group from all people developing bariatric surgery discipline, but their contributions to this area are incontrovertible.
Bariatric surgery has come a long way since the first procedures. Its evolution can be seen from several perspectives. Henry Buchwald spotlights the fact that bariatric surgery is metabolic surgery. In 1978, he and Richard L. Varco published a book entitled Metabolic Surgery in which they defined it as "the operative manipulation of a normal organ system to achieve a biological result for a potential health gain." Therefore, in the early beginnings of bariatric surgery, not only weight loss but also general health improvement was already expected. Other years carried out other studies, and new reports about benefits are evoked by bariatric procedures. Toward the end of the previous century, it was communicated that bariatric surgery can resolve type 2 diabetes mellitus. What is really interesting is that the remission of T2DM can occur very fast, even before the patients can reduce their weight [4][5][6]. It has also been pointed out that bariatric procedures improve many other clinical parameters (BMI, HbA1c, glucose and cholesterol levels, insulin resistance, and modulations of gut hormones) [7]. However, those modulations are not the core of this paper. Evidently, like every surgical intervention, it is related to some risks and disadvantages. Depending on the type of surgery, patients can be exposed to, for example, dumping syndrome or be forced to lifetime vitamin supplementation [6]. However, the growth of laparoscopic surgery with its reduced complications such as wound infection, incisional hernias, and lower early postoperative morbidity and mortality resulted in shorter hospital stay, faster recovery, lower morbidity, and improved effects, which has led to an ever-increasing patient demand [1,8,9]. Therefore, bariatric surgery has become the most effective treatment of morbid obesity and associated comorbidities, such as sleep apnea, hypertension, dyslipidaemia, and type 2 diabetes [8,10,11]. Thus, these kinds of surgical procedures not only reduce overall mortality but also improve patients' quality of life [9,12].
Similarly to those of bariatric surgery, the origins of metabolomics can be found far back in the past-in ancient Greece. Likewise, the beginning of metabolomics in modern meaning is estimated to be in the 1960s, when during metabolic-control analysis, the mathematical method for cell metabolism modelling was developed. The second starting point was the development of nuclear magnetic resonance (NMR) spectroscopy [17]. Currently, Oliver Fiehn and Jeremy K. Nicholson are considered as pioneers in the metabolomics (metabonomics) field [18][19][20]. Nowadays, both terms are used interchangeably. Jeremy K. Nicholson described differences between these terms as philosophical rather than technical. Metabolomics looks for an analytical description of complex biological samples. Its aim is to characterize and quantify all the small molecules in the studied sample. In the meantime, metabonomics is described as global measurement of dynamic metabolic response of living systems to biological stimuli or genetic manipulation. It is focused on understanding systemic modulations of complex multicellular systems through the time. Actually, modelling procedures for both of them are the same [17]. Metabolomics analyses are based on stand-alone hydrogen nuclear magnetic resonance ( 1 H NMR) technique or mass spectrometry technique combined with different metabolite chromatographic separation methods, that is, liquid chromatography (LC), gas chromatography (GC), or capillary electrophoresis (CE). This range of analytical platforms enable detection, characterization, and quantification of low-molecular-weight metabolites from different classes, for example, lipids, amino acids, peptides, nucleic acids, organic acids, vitamins, thiols, carbohydrates, and many other metabolites in which mentioned species can be metabolised. NMR can uniquely identify and simultaneously quantify a wide range of organic compounds in the micromolar range. It has been used for analysis of amino acids, nucleotides and nucleosides, vitamins, thiols, carbohydrates, and peptides. The LC-MS method has become a useful tool for the analysis of hundreds of polar metabolites in a complex sample. It is an important tool used for targeted or nontargeted metabolomics. Liquid chromatography separation is better suited for the analysis of labile and nonvolatile polar (hydrophilic interaction liquid chromatography (HILIC)) and nonpolar (reversed-phase chromatography) compounds in their native forms. Additionally, MS and LC are commonly used for compound characterization and to obtain structural information. GC-MS has been used as a platform especially for hydrophilic metabolites. Using this approach, one can directly separate and quantify the volatile metabolites. It allows to profile several hundreds of compounds including organic acids, most amino acids, sugars, sugar alcohols, aromatic amines, and fatty acids. CE-MS has been used for both targeted and nontargeted analyses of polar and ionic metabolites, including analysis of inorganic ions, organic acids, amino acids, nucleotides and nucleosides, vitamins, thiols, carbohydrates, and peptides [21,22].
Therefore, metabolomics is a powerful tool for the comprehensive study of the dynamic alterations of the metabolome. It has been widely used in the areas of medicine, biology, and physiology for biomarker discovery or for prognostic evaluations [23][24][25][26][27].
The aim of this study was to find and classify all studies in which the metabolomics approach was used to study the metabolic effects of bariatric surgery published up till now. Additionally, detected metabolites were investigated together to obtain conclusions about the impact of bariatric surgery on particular biochemical pathways. All studies, regardless of species (humans, rats, and mice) and biological samples used (blood, urine, and tissues), were included in this review. Intervention trials in which metabolomics techniques were used to study changes after bariatric surgery were investigated. In one study, influence of diet and surgery was compared, but only blood taken during bariatric procedure was analyzed [28]. This study (and animal studies) is reviewed, but its results were not considered during the MetaboAnalyst analysis.

Metabolomics
Data Analysis with MetaboAnalyst 3.0. MetaboAnalyst 3.0 was used to perform biochemical interpretation of all altered metabolites. This online tool (http://www.metaboanalyst.ca/) allows to analyze impact of particular compounds on biochemical pathways. In Meta-boAnalyst 3.0, there are currently 15 pathway libraries supported, with a total of 1173 pathways (80 for Homo sapiens). The pathway analysis module combines results from the powerful pathway enrichment analysis with those from the pathway topology analysis. Pathway analysis accepts a list of compound labels (common names, HMDB IDs, or KEGG IDs). Next, Fisher's exact test or hypergeometric test is used. The results from the pathway analysis are presented graphically as well as in a detailed table [29].

Results
Of all initially retrieved studies, 30 successfully fit the criteria for this review. The first study on the application of NMR to examine effects of biliopancreatic diversion and Roux-en-Y gastric bypass was published in 2010. Since then, the number of papers per year is still rising (Figure 1). Additionally, also other metabolomics platforms were used to investigate the influence of different types of surgeries on metabolome. This proves the growing interest on bariatric surgery and applicability of metabolomics to investigate this metabolic surgery.
Most of the studies were focused on human samples; only few were performed on rats [11,13,28,45] or mice [42]. In some of them, animal and human studies were combined [28,42].
Most of the studies were focused on obesity, including morbid obesity, and type 2 diabetes. Interestingly, in one research, another disease, that is, nonalcoholic fatty liver disease (NAFLD), was examined [40]. Some of the studies were focused only on effects of bariatric surgery. In one interesting study, measurement of biological age was performed on a group of obese patients after bariatric surgery [26].
Usually, researches were performed on medium-sized groups, that is, 10-20 patients. There were also few studies on really small (below 10), big , or really big (up to 100) [38] groups of patients. For animal study standards, a really large group of rats (27 animals) was used to study RYGB surgery effects [11].
Considering time intervals, very different time points were examined. Most of the studies contained baseline and then one or more follow-up points. The shortest difference between baseline and the first analyzed time point (below 7 days after surgery) was presented by Jüllig et al. [27], Nemati et al. [30], Arora et al. [36], Gralka et al. [38], and Friedrich et al. [39]. The longest follow-up was presented in the study performed by Heffron-1, 2, 3, 4, and 5 years after the surgery [46]. From examined studies, over 300 metabolites were selected. After removing duplicates, standardizing names (taking into account synonyms), and assigning HMDB IDs, we introduced 224 compounds into MetaboAnalyst 3.0 software. Finally, the software used 211 of them for pathway analysis. Obtained results showed that metabolites altered by bariatric surgery belong to 63 biochemical pathways. A statistically significant influence was exhibited for 23 of them. Based on the MetaboAnalyst results, the most impacted pathways after bariatric interventions are aminoacyl-tRNA biosynthesis; glycine, serine, and threonine metabolism; nitrogen metabolism; phenylalanine metabolism; cysteine and methionine metabolism; TCA cycle (citrate cycle); taurine and hypotaurine metabolism; valine, leucine, and isoleucine biosyntheses; propanoate metabolism; and nicotinate and nicotinamide metabolism ( Figure 2).

Discussion
Despite the substantial research activities in the last years, many molecular aspects of bariatric surgery consequences leading to observed surgery effects are still unknown. However, alterations of particular metabolite groups detected till now allow to deduce about general changes associated with this kind of surgical procedure.
Higher serum concentrations of phenylalanine, tyrosine, leucine, isoleucine, valine, and glutamate were noticed in obese individuals. Roux-en-Y gastric bypass caused a decrease of circulating aromatic amino acids (AAA): methionine, alanine, and lysine. Serum concentrations of serine and glycine were found to be increased after sleeve gastrectomy [42]. RYGB accelerates caseinate digestion and amino acid absorption, resulting in a faster and higher but also more transient postprandial elevation of plasma amino acids [34]. In the group of patients with diabetes remission, relatively to nonremission significant decrease in alanine after one year was observed [32]. Branched-chain amino acid levels were found to be correlated with decreased insulin resistance [7].
Higher baseline stearic acid/palmitic acid ratio was associated with greater probability of diabetes remission after RYGB and may serve as a diagnostic marker in preoperative patient assessment. Correlation analysis demonstrated that the stearic acid/palmitic acid ratio negatively correlated with HbA1c, TG, TC, LDL-c, and HOMA-IR and positively correlated with HDL-c in overweight and obese subjects [51]. Arora et al. reported early alterations in the metabolome and lipidome after gastric bypass in insulin-resistant morbidly obese subjects. The beneficial effects of surgery included a reduction in BCAA metabolites and short-chain TGs [36]. Data obtained by Oberbach et al. showed that LSG affects the amino acid and lipid metabolism. It leads to modification of amino acids and lipid metabolism as indicated by changes in glycerol-phosphatidylcholines and SM levels [41].

Gut Microbiota-Related Metabolites.
Gut microbiota plays an important role in various processes including energy metabolism, lipid accumulation, homeostasis, regulation of brain function, and behavior. Its modulation is also one of the mechanisms by which bariatric surgery promotes    [53]. The human gastrointestinal tract (GIT) is dominated by two bacterial phyla, the Bacteroidetes and the Firmicutes, and the proportion of this phyla to each other has been already linked to obesity and type 2 diabetes mellitus [53][54][55][56]. Accordingly, alterations of compounds which can be linked to gut microbes and their modulations by surgery are important and interesting findings [38]. In several studies, changes in levels of SCFAs (i.e., butyric acid) [26, 27, 37-39, 44, 47], lactate [7,26,33,35,45], indole [36], and 3-indoxyl sulfate [7,26] after bariatric procedures were observed. Alterations in the level of sulfate-containing metabolites can be expected, as the largest group of sulfate-reducing bacteria is found among the Proteobacteria, present mainly in the duodenum, which is modified in some bariatric procedures [57]. Also levels of mentioned cholesterol [33,35,46], L-carnitine [7,11,26,43], and niacin or choline [7,25,26,32,45] can be linked to gut microbes [58]. Modesitt et al. linked perturbations in tryptophan, phenylalanine, and heme metabolism with decreased inflammation and alterations in the intestinal microbiome [7]. Moreover, tyrosine or phenylalanine fermentation by intestinal bacteria generates p-cresol. Bacteroides fragilis is one of the bacteria that have been shown to produce it [59]. Serum glutamate concentration was inversely correlated with the abundance of some Bacteroides species as well [42]. Changes in histidine and its metabolites following surgery might be an indication of altered gut microbiome ecology or liver function [44]. Another significant metabolite in which its modification can be linked to microbiota modification is beta-hydroxybutyrate (β-OHB). It is derived mainly from the oxidation of fatty acids and is the first ketone produced in the fasting state. Additionally, it is also produced in the form of poly-β-OHB by prokaryotes when carbon sources are freely available but other nutrients are limited [60].

Other Compounds.
Some of the metabolites reported in reviewed studies as significantly changing after bariatric surgery do not belong to the above-described groups of compounds. Therefore, in this section, metabolites from other biochemical pathways altered by metabolic surgery will be presented. Examples of such compounds are 8-oxoGua, 8-oxoGuo, and 8-oxodG, markers of DNA and RNA damage studied by Bankoglu et al. [13]. Alterations in the concentration of these compounds indicate the association of obesity with increased oxidative stress and DNA damage. Moreover, it was said that RYGB or caloric restriction can significantly reduce elevated oxidative or nitrative stress as well as genomic damage in obese subjects. Results obtained by Sarosiek et al. also suggested that bariatric surgery might promote antioxidant defence and insulin sensitivity through both increased heme synthesis and heme oxygenase (HO) activity or expression [44]. Modification of nucleotide metabolism after bariatric intervention was evaluated by adenosine, inosine, hypoxanthine, xanthine, urate, and allantoin profiling [7]. DJB surgery enhanced trans-sulfuration and its consecutive reactions such as detoxification and the scavenging activities of reactive oxygen species [45]. Metabolites detected by Narath et al. (trimethylamine N-oxide, alanine, phenylalanine, and indoxyl sulfate) are known as cardiovascular disease risk markers [32].
An important group of altered metabolites is compounds connected with energetic processes. Pyruvate [7,11,38,44,45], citric acid [11,26,27,38], carnitines [7,11,26,43], or the above-mentioned fatty acids belong to this group. Calvo et al. observed that the presence of moderate NAFLD is common in young patients with morbid obesity. Their data may be useful to explain the dissociation between excess lipid storage in adipose tissue, NAFLD, and insulin resistance [40]. After surgeries, energy metabolism, glucose homeostasis, and glycemic markers showed marked improvements, which manifested with reduced levels of glucose and the glycolytic end products of pyruvate and lactate. An increased level of chiroinositol may be associated with improved insulin signaling [7]. Narath et al. reported the decrease of lactate (Krebs' intermediate cycle) after RYGB. They also observed the higher levels of the high-density lipoprotein and phosphatidylcholine after bariatric surgery [32].
An interesting study about serum uric acid (sUA) was performed by Oberbach et al. An elevated level of sUA was observed in obese patients. However, twelve months after LSG and RYGB, a significant decrease in sUA and other parameters such as BMI, CVD risk factors, hepatic transaminases, and HOMA-IR was observed. Kwon et al. suggested new criteria-7-day metabolomics profile and 3-hydroxybutyrate to glucose ratio for the prediction of 3-month HbA1c. They suggested that this finding could augment current prognostic modalities and help clinicians decide if drug therapy is necessary [37]. Chouiali et al. performed also an interesting but more methodological study. The authors compared two methods (electro-chemiluminescence immunoassay and LC-MS/MS) for measurement of serum 25(OH)D by applying them to the bariatric population [48].
It has to be mentioned that changes presented above are only part of all detected modulations in all investigated studies. They were subjectively chosen by authors as the most interesting. There are still some intriguing modulations which are not described here but can be found in referred articles.

Animal
Studies. The animal studies are an important part of the bariatric surgery research. The animal models allow not only to follow the general metabolic changes in blood and urine but also to focus on modulations, in particular organs, by analyses of tissue samples. Five [11,13,28,42,45] of all presented studies were performed on an animal model. In mentioned studies, not only blood or urine but also liver, heart, and adipose tissue samples were examined. Animal models in this kind of studies are also important because of the fact that they allows to compare particular bariatric procedures between themselves, as well as with sham operations, which cannot be performed in humans. Additionally, the animal models are characterized by high repeatability, which is meaningful when using the metabolomics approach [61].

From Metabolite to Metabolome-General Metabolic
Effects of Bariatric Intervention. All above-described alterations in combination with results from MetaboAnalyst showed that bariatric procedures have a huge impact on patients' metabolism. This influence can be observed by following clinical as well as molecular parameters. Investigation of clinical parameters in combination with multiple metabolites provides a broader picture than does evaluation of changes in selected metabolites. Thus, metabolomics is a perfect tool to study global effects evoked by bariatric surgery.
Of course, even the "image" of metabolic changes obtained here, based on all mentioned studies, is still incomplete. There are still some blank areas on this biochemical map. But step by step, using metabolomics techniques (especially combined together) to examine different procedures can bring us more interesting and useful knowledge. Although providing a wide spectrum of information, the metabolomics approach has some limitations. In case of LC-MS, the identification of metabolites could be improved-a relatively large percent of detected metabolites have remained unidentified. In case of GC-MS, identification is based on libraries, so the number of detected metabolites is always limited. NMR in comparison to MS-based platforms is less sensitive. Drawing conclusions is also limited by use of software for pathways analysis. In each software, the number of available metabolites and pathways is limited. MetaboAnalyst library contains 80 biochemical pathways for humans and similar or even less for other species. Therefore, many metabolites and pathways which can be affected by bariatric intervention are not included in such analysis. It is also worthy to collect more data for a particular procedure and to try to analyze them separately. It is very possible that different procedures will evoke distinct impact on biochemical pathways.

Conclusion
Bariatric procedure strongly influences the metabolism. Detected changes are tied with many compounds and biochemical pathways such as amino acids, lipids, carbohydrates, or gut microbiota alterations. It proves that classification of bariatric intervention as metabolic surgery is appropriate. However, many molecular modulations after those procedures are still unexplained. Therefore, the application of metabolomics in this field of medicine seems to be a right choice. New findings can suggest new directions for surgery technique modifications, contribute to broaden knowledge about obesity and related diseases, and perhaps develop nonsurgical methods of treatment in the future.

Conflicts of Interest
The authors declare no conflicts of interest.