Untargeted Metabolomic Approach Shows No Differences in Subcutaneous Adipose Tissue of Diabetic and Non-Diabetic Subjects Undergoing Bariatric Surgery: An Exploratory Study

Abstract

Background:

Obesity plays a major role in the development of insulin resistance (IR) and diabetes (T2DM). Increased adipose tissue (AT) is particularly of interest because it activates a chronic inflammatory response in adipocytes and other tissues. AT plays key endocrine and metabolic functions, acting in the regulation of insulin sensitivity and energy homeostasis. Additionally, it can be easily collected during bariatric surgery. The purpose of this pilot study was to explore the potential differences in AT metabolism, through comparing the untargeted metabolomic profiles of diabetic and non-diabetic obese patients undergoing bariatric surgery.

Methods:

For this exploratory study, samples were collected from 17 subjects. Subcutaneous AT (SAT) samples from obese-diabetic (n = 8) and Obese-non-Diabetic (n = 9) subjects were obtained from the Human Metabolic Tissue Bank. Untargeted metabolomic profiling was performed by Metabolon® Inc. Statistical analysis was performed using the MetaboAnalyst 4.0 platform.

Results:

Among the 421 metabolites identified and analyzed there were no significant differences between the Obese-Diabetics and the Obese-non-Diabetics. Small changes were observed by fold change analysis mainly in lipid (n = 12; e.g. NEFAs) and amino acid (n = 8; e.g. BCAAs) metabolic pathways. Dysregulation of these metabolites has been associated with IR and other T2DM-related pathophysiological processes.

Conclusion:

Obesity may influence SAT metabolism masking T2DM-dependent dysregulation. Better understanding the metabolic differences within SAT in diabetic populations may help identify potential biomarkers for diagnosis and monitoring of T2DM in patients undergoing bariatric surgery.

Obesity is a worldwide epidemic. According to the World Health Organization, more than 650 million adults are obese (body mass index > 30 kg/m²), and approximately 51% of the world population is predicted to be obese by 2030 (Finkelstein et al., 2012; WHO, 2018). Obesity is influenced by the interaction of environmental, psychological, and genetic factors that contribute to excess caloric intake and a state of positive energy balance (Valencak et al., 2017). Obesity contributes to diabetogenic characteristics, such as dyslipidemia, insulin resistance (IR), and metabolic syndrome, which increase the risk of developing type 2 diabetes mellitus (T2DM; Burhans et al., 2018; Kusminski et al., 2016).

T2DM is also characterized by chronic hyperglycemia, dyslipidemia, and IR (Catalan et al., 2016; Engin, 2017a). Prior studies have demonstrated the important metabolic role of adipose tissue (AT) in the human body, including the regulation, storage, and release of energy (Valencak et al., 2017) and its involvement in the development of IR (Catalan et al., 2016; Engin, 2017a, 2017b; Greenberg & Obin, 2006; Guillet et al., 2012). Weight loss and medications that restore glycemic control and mediate insulin secretion are used in the management of T2DM in subjects whose BMI is 35kg/m² or higher (Kusminski et al., 2016; Nathan, 2002).

Bariatric surgery (gastric restrictive procedures or intestinal bypass procedures) is an effective intervention against severe obesity and T2DM (Buchwald et al., 2009; Kashyap et al., 2010; Sarosiek et al., 2016). Following bariatric surgery, 87% of diabetic patients achieve better glucose control with few medications and 78% achieve normal glycemia without any medication (Buchwald et al., 2009). The metabolic benefits of bariatric surgery include increased insulin sensitivity, modulation of adipokine secretion and decreased local adipose and systemic inflammation. In addition, it also remodels the interaction between adipose tissue and other organs (Mingrone & Cummings, 2016; Villarreal-Calderón et al., 2019). Thus, understanding the metabolites derived from AT may provide insights into the pathological mechanisms involved in T2DM in bariatric patients and aid in the treatment/management of T2DM in this population.

Changes in AT plasticity (i.e., tissue expansion due to increased adipocyte size [hypertrophy] and/or number [hyperplasia], consequent remodeling of the vasculature system, and recruitment of inflammatory cells; Wang et al., 2013), together with the locally increased accumulation of lipids, are suggested to be early indicators of IR, as they are present long before the clinical expression of T2DM (Burhans et al., 2018; Cummins et al., 2014; Engin, 2017a, 2017c). These alterations in plasticity and lipid accumulation may greatly influence obesity-related dysfunction of subcutaneous adipose tissue (SAT): the largest AT storage depot (Abraham et al., 2015). Excessive energy intake may cause SAT to expand and exceed capacity, which facilitates lipid accumulation in non-SAT depots (visceral or ectopic), possibly leading to IR (Abraham et al., 2015; Longo et al., 2019; Neeland et al., 2012). Also, increased concentrations of non-esterified fatty acids (NEFAs), most prevalent in SAT, may potentially induce IR (Baron, 2002; Burhans et al., 2018; Ferrannini, 2014; Pallares-Mendez et al., 2016; Sobczak et al., 2019). Although the mechanisms by which SAT and NEFAs influence IR and T2DM remain unclear, previous studies have suggested that low-grade inflammation (Longo et al., 2019) and NEFA spillover (a process leading to aggregation of NEFAs in plasma instead of AT) may play a role (Karpe et al., 2011). Improved understanding of the complex relationship between abdominal SAT and T2DM may contribute to the development of innovative methods to diagnose and manage the obesity-related onset of T2DM.

Untargeted metabolomics, the unsupervised large scale study of small molecule substrates (metabolites), intermediates, and products of metabolisms, is a unique method of providing a direct readout of phenotype perturbations (Cummins et al., 2014; Filla & Edwards, 2016; Goodpaster & Sparks, 2017; Guasch-Ferré et al., 2016; Kucera et al., 2018; Riekeberg & Powers, 2017). Moreover, rapidly fluctuating metabolites produced by AT in obese subjects are emerging as key mediators of metabolic disease: they may be used as biomarkers of that obesity-dependent pathological state and act as endocrine regulatory signals (Greenberg & Obin, 2006; Hanzu et al., 2014; Saltiel & Olefsky, 2017; Sun et al., 2011). As such, metabolomics can aid in the detection and management of disease processes, including obesity-related T2DM (Filla & Edwards, 2016; Guasch-Ferré et al., 2016; Sarosiek et al., 2016; Wu et al., 2018).

Few studies have focused on the metabolomic changes within SAT (Cummins et al., 2014; Hanzu et al., 2014) and to our knowledge, no previous studies of SAT metabolites have compared the metabolic profile of Obese-Diabetic vs Obese-non-Diabetic populations. Conducting an untargeted metabolomic analysis in the AT of Obese-Diabetic patients may provide insight into the intermediate products of metabolism in this population. The purpose of the present study was to use an untargeted metabolomic approach to explore modifications in the metabolic pathways of abdominal SAT in obese individuals with and without T2DM undergoing bariatric surgery. While there is no cure for T2DM, the identification of new biomarkers can contribute to the development of early treatment alternatives that prevent the progression of T2DM (Wu et al., 2018).

Methods

Design and Setting

This was an exploratory study of AT metabolomics in Obese-Diabetic and Obese-non-Diabetic patients undergoing bariatric surgery. SAT samples (n = 17) were obtained from the Human Metabolic Tissue Bank (HMTB) at Penn Presbyterian Medical Center and the Hospital of the University of Pennsylvania. Prior to surgery individuals had a diagnosis of T2DM, as documented in the medical records. In addition, laboratory tests (insulin levels, HgbA_1c) verified the diagnosis during routine assessments prior to bariatric surgery. Once collected, SAT samples were immediately stored at −80 °C. Written informed consent was obtained from all participants. Demographic and clinical data (gender, age, race, medications) were collected from medical records. Body weight and height were used to calculate body mass index (BMI), weight/height (kg/m²).

Sample Accessioning

Following receipt, samples were inventoried and immediately stored at −80 °C. Each sample received was accessioned into the Metabolon® Laboratory Information Management System (LIMS) and was assigned a unique identifier by the LIMS that was associated with the original source identifier. This identifier was used to track all sample handling, tasks, results, etc. All samples were maintained at −80 °C until processed.

Sample Preparation

Samples were sent to Metabolon® for processing, including metabolite identification and quantification. Samples were prepared using the automated MicroLab STAR® system from Hamilton Company. Recovery standards were added prior to the first step in the extraction process for Quality Control (QC) purposes. To remove protein, to dissociate small molecules bound to protein or trapped in the precipitated protein matrix, and to recover chemically diverse metabolites, proteins were precipitated with methanol under vigorous shaking for two minutes (Glen Mills GenoGrinder 2000, Germany) followed by centrifugation. QC procedures are described in the Supplementary Materials. Samples were divided into five sections: two for analysis by two separate Reversed Phase Ultrahigh Performance Liquid Chromatography-Tandem Mass Spectroscopy ([RP]/UPLC-MS/MS) methods with positive ion mode electrospray ionization (ESI), one for analysis by RP/UPLC-MS/MS with negative ion mode ESI, one for analysis by HILIC/UPLC-MS/MS with negative ion mode ESI, and one sample was stored. Samples were placed on a TurboVap® (Zymark) to remove the organic solvent and were stored for analysis.

Metabolomic Profiling

Untargeted metabolomic profiling was performed by Ultrahigh Performance Liquid Chromatography-Tandem Mass Spectroscopy (UPLC-MS/MS). Waters ACQUITY ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-Exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer (operated at 35,000 mass resolution) were used. The samples were dried and then reconstituted in solvents compatible with each of the four methods. Each reconstitution solvent contained a series of standards at fixed concentrations to ensure injection and chromatographic consistency. Briefly, the first method used acidic, positive ionization conditions chromatographically optimized for hydrophilic compounds. The second method used the same acidic positive ionization conditions, but was chromatographically optimized for hydrophobic compounds. The third method used negative ionization optimized conditions, and the last method utilized negative ionization optimized conditions with hydrophilic interaction liquid chromatography.

Data Extraction and Compound Identification

Raw data was extracted, peaks were-identified, and QCs were processed using Metabolon’s hardware and software. Compounds were identified by comparison to library entries of purified standards or recurrent unknown entities. Metabolon® maintains a library based on authenticated standards that contains the retention time/index (RI), mass to charge ratio (m/z), and chromatographic data (including MS/MS spectral data) on all molecules present in the library.

Data Processing: Univariate and Multivariate Analysis

Datasets derived from Metabolon® were normalized by sample mass available/utilized for extraction. Then, each biochemical in the original scale was rescaled to set the median equal to 1. Lastly, missing values were replaced with the minimum values registered in the samples. Subsequently the datasets were loaded into MetaboAnalyst 4.0 (Chong et al., 2019; https://www.metaboanalyst.ca/home.xhtml) for metabolite analysis. The metabolic diversity between groups was quantified using univariate (t-test and Fold Change Analysis [FCA]) and multivariate Principal Component Analysis (PCA) analyses. The false discovery rate (FDR) approach described by Storey and Tibshirani (2003) was used to account for multiple testing. All p-values were two-tailed and a corrected p-value < 0.05 was considered significant for the statistical analyses in this study. Fold change was considered significant above 2. For untargeted metabolomics, the fold change serves as a measure for the relative change in a given metabolite’s concentration in the different conditions under investigation.

Results

Population Characteristics

This was an exploratory study of adult Obese-Diabetic vs non-Diabetic patients with obesity undergoing bariatric surgery. Eight samples from diabetic patients and nine from non-diabetics were acquired from the HMTB. Participants were matched by demographic (gender, age and race) and clinical characteristics (Table 1). Most of the patients (94.1%) were women, with an average age of 45.3 years (SD = 11) and a mean BMI of 46.2 kg/m² (SD = 7.7).

Table 1.

Demographic and Clinical Features of the Study Sample.

Group		Diabetics			Non-Diabetics
Characteristic
Gender (n)		7F/1M			9F/0M
		Mean	SD	Range	Mean	SD	Range
Age (years)		50	7	42–62	42	13	26–58
BMI (kg/m2)		45.6	7	38–56	46.7	9	38–63
Race (n)	Non-Hispanic white		3			5
	African American		3			4
	Hispanic		1			0
Medications (n)	Acetaminophen		—			3
	Albuterol		1			2
	Aspirin		1			2
	Atorvastatin		2			—
	Cholecalciferol		1			3
	Citalopram		2			—
	Fluticasone		2			2
	Furosemide		2			—
	Metformin		4			—
	Hydrochlorothiazide		2			—
	Omeprazole		1			4
Hypertension (n)			3			5

Note. Selected clinical features of population study. F/M = females/males. Medications shown represents consumption by more than one patient.

Metabolomic Analysis

Based on liquid chromatography/mass spectrometry, 421 known compounds were identified and quantified by Metabolon®. These compounds were analyzed on the MetaboAnalyst platform and statistical analysis was performed. The t-test and FCA were used to look at differences between the individual metabolites while PCA was used to examine the data as a whole, according to the differences resulting from the metabolic profiling. After correcting for multiple comparisons, no significant differences were found between Obese-Diabetics and Obese-non-Diabetics (Figure 1A). These results are consistent with the unsupervised multivariate analysis (PCA), which showed no segregation between the two groups (Figure 1B). The FCA (Figure 2, Table 2) showed that for thresholds above two, significant changes were identified in lipids (n = 12), amino acids (n = 8), carbohydrates (n = 2), nucleotides (n = 2), and peptide (n = 1) metabolism.

Figure 1.

(a) t-Test Results. After FDR, no significant features were identified. (b) Principal component analysis shows no discrimination between Obese-Diabetics vs Obese-non-Diabetics (red/green, respectively).

Figure 2.

Features identified by the FCA. Red/green (Obese-Diabetics/Obese-non-Diabetics) bars represent features above 2 [Log2(FC = 1)] represented by the discontinuous line.

Table 2.

Relevant Biochemicals and Corresponding Pathways Identified by the FCA.

Super-pathway	Sub-pathway	Biochemical	FC	log2(FC)	p-val	Mean ob-dia ± SD	Mean ob-non-dia ± SD
Amino Acid	Creatine Metabolism	creatine phosphate	2.10	1.07	0.47	1.83 ± 3.53	0.87 ± 1.44
	Histidine Metabolism	cis-urocanate	0.42	−1.24	0.28	2.28 ± 1.12	5.37 ± 7.79
	Leucine, Isoleucine and Valine Metabolism	1-carboxyethylisoleucine	2.10	1.07	0.34	0.91 ± 1.2	0.43 ± 0.77
		3-methyl-2-oxovalerate	2.46	1.30	0.07	1.58 ± 1.45	0.64 ± 0.24
		4-methyl-2-oxopentanoate	3.50	1.81	0.02	3.24 ± 2.67	0.93 ± 0.58
	Lysine Metabolism	fructosyllysine	2.68	1.42	0.21	10.59 ± 14.7	3.95 ± 3.89
	Urea cycle; Arginine and Proline Metabolism	3-amino-2-piperidone	2.23	1.15	0.04	3.91 ± 2.67	1.75 ± 0.91
		homoarginine	2.50	1.32	0.21	2.73 ± 3.59	1.09 ± 1.15
Carbohydrate	Aminosugar Metabolism	N-acetylglucosamine 6-phosphate	0.37	−1.42	0.22	0.27 ± 0.28	0.72 ± 0.97
	Glycolysis, Gluconeogenesis, and Pyruvate Metabolism	1,5-anhydroglucitol (1,5-AG)	0.46	−1.13	0.02	1.72 ± 0.83	3.76 ± 2.02
Lipid	Fatty Acid Metabolism	3-hydroxyoleoylcarnitine	0.49	−1.02	0.14	0.54 ± 0.18	1.09 ± 0.99
		sebacate (C10-DC)	2.27	1.18	0.17	2.25 ± 2.58	0.99 ± 0.37
		3-hydroxyhexanoate	2.43	1.28	0.17	1.45 ± 1.73	0.6 ± 0.36
		alpha-hydroxycaproate	2.50	1.32	0.18	0.96 ± 1.25	0.38 ± 0.06
		8-hydroxyoctanoate	3.23	1.69	0.24	2.79 ± 4.72	0.87 ± 0.39
		2-hydroxydecanoate	3.24	1.70	0.22	2.45 ± 3.99	0.75 ± 0.21
		maleate	3.88	1.95	0.24	3.93 ± 7.18	1.01 ± 0.4
		azelate (nonanedioate; C9)	3.97	1.99	0.28	5.29 ± 10.7	1.33 ± 0.77
		3-hydroxyoctanoate	4.86	2.28	0.23	5.51 ± 10.53	1.13 ± 0.43
		2-hydroxyheptanoate*	18.03	4.17	0.21	14 ± 30.86	0.78 ± 0.3
	Phosphatidylglycerol (PG)	1,2-dioleoyl-GPG (18:1/18:1)	5.94	2.57	0.31	6.86 ± 16.3	1.15 ± 1.19
	Sphingolipid Synthesis	sphingadienine	0.38	−1.38	0.31	1.1 ± 0.46	2.87 ± 4.72
Nucleotide	Purine Metabolism, Adenine containing	adenine	2.18	1.13	0.36	5.54 ± 8.71	2.54 ± 3.67
	Pyrimidine Metabolism, Cytidine containing	cytosine	2.14	1.10	0.07	3.56 ± 2.66	1.66 ± 1.11
Peptide	Dipeptide	prolylglycine	4.48	2.16	0.31	2.4 ± 5.3	0.53 ± 0.72

Note. Increases in biochemicals are represented by dark gray (Obese-Diabetic group) and light gray (Obese-non-Diabetic group). FCA = fold change analysis, FC = fold change, raw p-val = p-value, ob-dia = Obese-Diabetic group, ob-non-dia = Obese-non-Diabetic group.

Lipids

In the Obese-Diabetic subjects, most of the lipids (n = 10/12) were higher compared to the Obese-non-Diabetics Most NEFAs exhibited a fold change above 3 (i.e., 8-hydroxyoctanoate, 2-hydroxydecanoate, maleate, azelate, and 3-hydroxyoctanoate) and the 2-hydroxyheptanoate* had a FC of 18, with higher values found in the Obese-Diabetic subjects. In the same group, 3-hydroxyoleoylcarnitine (fatty acid) and sphingadienine (sphingolipid) were decreased (FC ≈ 2; Table 2).

Amino Acids

Our results also revealed changes in amino acid metabolism. Particularly, we noted an increase in several metabolites (n = 7/8) in the Obese-Diabetics subjects, except for cis-urocanate (histidine metabolism), which was higher in the Obese-non-Diabetic patients. All branched chain amino acids (leucine, isoleucine, and valine) metabolites, such as 4-methyl-2-oxopentanoate, 1-carboxyethylisoleucine, and 3-methyl-2-oxovalerate (FC = 2.2, FC = 3.5, and FC = 2.5 respectively), were higher in the samples from Obese-Diabetic patients as well as amino acids (n = 4) involved in the metabolism of creatine, lysine, and urea cycle metabolisms.

Carbohydrates

We observed differences in two metabolites between the Obese-Diabetic cohort and their Obese-non-Diabetic counterparts. Using FCA, we identified increases in the Obese-non-Diabetic samples in metabolites involved in amino-sugar metabolism processes including glycolysis, gluconeogenesis, and pyruvate metabolism, such as 1,5-anhydroglucitol (1,5-AG) and N-acetylglucosamine 6-phosphate (FC = −4.5 and FC = −6.1; respectively).

Nucleotides

Adenine (FC = 3.1) and cytosine (FC = 3.3), two metabolites involved in the nucleotide pathway, were higher in the Obese-Diabetic subjects.

Discussion

The aim of this exploratory study was to compare the metabolite profiles in the SAT of diabetic and non-diabetic patients with obesity undergoing bariatric surgery. A total of 421 metabolites were identified and analyzed. Although we did not find significant differences in the metabolic pathways between the groups using multiple testing corrected statistical tests, we did observe fold change differences between Obese-Diabetics and Obese-non-Diabetics. These differences were captured predominantly in lipids (NEFAs) and amino acids pathways, where 20 out of 25 metabolites were higher in the Obese-Diabetics subjects; few metabolites were involved in carbohydrates (higher in the non-diabetic-obese subjects) and nucleotide metabolic routes. The observed fold changes suggest that these lipid and amino acid pathways differences may be induced by T2DM. Confirming and better understanding the role of these metabolite changes in larger populations could aid in our understanding of the pathophysiology of T2DM in obese populations.

Almost all the changes in lipid metabolism were associated with higher fold changes in lipids (especially NEFA) in the Obese-Diabetic group compared to the Obese-non-Diabetic group. Dysregulation of fatty acid metabolism, manifested as high NEFAs in circulation and excess lipid deposition in liver and skeletal muscle, is closely associated with IR (Karpe et al., 2011; Sobczak et al., 2019). Also, the increase in saturated FA (SFAs) would likely promote the generation of reactive oxygen species, which contributes to inflammation and the onset/progression of IR (Paniagua, 2016). Thus, the observed perturbations in lipid homeostasis may represent a marker of IR. Although we did not find significant differences, previous literature supports this hypothesis (Holland & Summers, 2008; Neeland et al., 2012). Moreover, another study from Hanzu et al. (2014), studying SAT within an obese cohort, shows differences obesity-related rather than driven by T2DM.

Amino acids play an important role in all living organisms, and regulate key metabolic and physiological pathways in human homeostasis (Wu, 2009). The branched chained amino acids (BCAAs) are important constituents of proteins, but are also readily degraded into carbon skeletons that may enter anabolic pathways (i.e., gluconeogenesis or fatty acid synthesis) and other pathways involved in energy balance (i.e., the TCA cycle). In the current study, we detected an accumulation of branched-chain ketoacids (e.g., 4-methyl-2-oxopentanoate, 3-methyl-2-oxovalerate) in the Obese-Diabetic group vs the Obese-non-Diabetic group. Perturbations in the metabolism of BCAAs have been linked to obesity and cardiometabolic syndrome; BCAAs and their metabolites show positive associations with IR in several metabolomic studies (Cummins et al., 2014; Guasch-Ferré et al., 2016; Vangipurapu et al., 2019; Zhou et al., 2019). Although these differences did not reach statistical significance in our study, we measured higher levels of metabolites involved in leucine, isoleucine, valine, alanine, creatine in the Obese-Diabetic vs. Obese-non-Diabetic cohort. These amino acids were recently linked to IR and decreased insulin secretion (Zhou et al., 2019).

Regarding the changes in the carbohydrates, both metabolites identified by FCA were higher in the Obese-non-Diabetic subjects, in agreement with the metabolic profile observed in studies using plasma (Kim & Park, 2013; Martins et al., 2019). For example, 1,5-AG is derived from diet and, during hyperglycemia, it competes with glucose for renal tubular reabsorption. The decrease in circulating levels of 1,5-AG are commonly observed in IR, and it has been used as a biomarker for both type I diabetes and T2DM in serum (Kim & Park, 2013; Martins et al., 2019; McGill et al., 2004).

Nucleotide levels are variable in subjects who are obese. In some patients with obesity, high level of nucleotides found in plasma seem to be associated with the initial stage of IR (Sparks et al., 2014). Our results revealed higher levels purine (adenine) and pyrimidine metabolites (cytosine) in the SAT of Obese-Diabetic individuals. Although research is limited, previous studies have demonstrated an association between adenine nucleotides (e.g., adenosine triphosphate [ATP]), IR, and T2DM; Koster et al., 2005; Schmid et al., 2011). For example, in the pancreatic β-cell, ATP-sensitive potassium channels play a key role in coupling membrane excitability with glucose-stimulated insulin secretion (Koster et al., 2005). Dysfunction of this process may lead to IR (Schmid et al., 2011).

Based on these findings, we hypothesize that the known obesity-driven changes in SAT metabolism may mask metabolomic changes induced by T2DM (Hanzu et al., 2014). Several studies (Cummins et al., 2014; Engin, 2017c; Greenberg & Obin, 2006; Kusminski et al., 2016) have demonstrated how SAT metabolism is disrupted in obese subjects leading to a cascade of events in SAT generally defined as “adipose tissue remodeling” (Catalan et al., 2016; Cummins et al., 2014; Longo et al., 2019; Spalding et al., 2008). These include: a) an increase and change in distribution of AT (ectopic fat accumulation on other tissues and accumulation of visceral AT; Zhang et al., 2015); b) lower number and changes in morphology and size of mitochondria (Cummins et al., 2014; Han, 2016); c) phenotypical changes and number in the adipocytes (Cummins et al., 2014; Han, 2016; McLaughlin et al., 2016); and d) uncontrolled inflammatory responses in AT, leading to chronic low-grade inflammation (Burhans et al., 2018; Catalan et al., 2016; Guillet et al., 2012). All these phenomena (Catalan et al., 2016) have been associated with the onset of IR (Burhans et al., 2018; Engin, 2017b, 2017c; Greenberg & Obin, 2006; Han, 2016; Longo et al., 2019; Neeland et al., 2012).

Although previous studies have found differences in metabolism comparing diabetic and non-diabetic subjects, we did not find statistically significant differences between these two groups in an obese cohort. We hypothesize that the obesity-dependent events described may be acting on SAT altering its metabolism and occulting the metabolic disruptions associated with T2DM.

Clinical Implications

Although we identified slight changes between the Obese-Diabetics and the Obese-non-Diabetics in SAT metabolites, these differences were not significant between groups. In order to establish the clinical applications of these results, further studies using a larger sample would be required (Cummins et al., 2014; Greenberg & Obin, 2006; Hanzu et al., 2014). A better understanding of the metabolites and metabolic pathways in SAT associated with T2DM could help to develop markers and therapeutic targets for nutritional or pharmacological interventions in obese patients undergoing bariatric surgery. For example, they may be used to study correlations between bariatric surgery outcome and pre-surgical metabolites. Moreover, metabolites present in the Obese-Diabetic group could be targeted for diagnostic purposes and those present in the Obese-non-Diabetic group could be studied as potentially protective against T2DM.

Importantly, as we hypothesize that some of the metabolite differences induced by T2DM may be masked by obesity, this should be considered when interpreting clinical data. For example, metabolites specific to diabetes (independent of obesity) would be needed to diagnose or manage the treatment of a non-obese patient with diabetes. Conversely, in a patient with multiple comorbidities the metabolite changes would have to be examined in the context of these diseases.

Future Direction and Limitations

In this preliminary study, small differences were found in lipid, amino acid, carbohydrate, and nucleotide metabolic pathways in a small cohort of Obese-Diabetic compared to non-diabetic obese patients. As this was a preliminary study, we had a limited sample size. Thus, future studies should expand this project in larger and more diverse obese populations. Although our sample size included mostly pre-menopausal women, we did not include participants’ menstrual status at the time of sample collection. The menstrual cycle may influence metabolite composition (Draper et al., 2018; Wallace et al., 2010) and SAT density/thickness (Perin et al., 2000); therefore, future metabolomic studies which take menstrual status into account are warranted. Additionally, specifically comparing a non-obese/normal weight control group, including longitudinal clinical data (i.e., bariatric surgery outcome), and comparing the metabolite differences across tissues would aid in establishing an association between metabolites changes, T2DM, and obesity. A normal weight group would serve as a negative control of obesity-derived changes. Longitudinal data could be collected to examine whether these metabolite profiles evolve over time and/or change with disease progression. Lastly, the studies of metabolites in SAT is innovative and particularly applicable to the study of T2DM in obese populations (characterized by excess SAT). However, future studies should examine whether the metabolite differences observed in SAT occur in blood or other tissues that could be more easily accessible for testing in a clinical setting.

Conclusions

Our untargeted metabolomic analyses revealed slight metabolite differences between Obese-Diabetics and Obese-non-Diabetics subjects. Although these differences were not significant, we identified fold changes mainly in lipids (NEFAs, phospholipids) and amino acids (BCAAs), increased in Obese-Diabetic subjects. We hypothesize that obesity may influences SAT metabolism masking T2DM-dependent dysregulation. Further research should be conducted including the use of a non-obese control group and longitudinal data to test the specificity of these metabolites to T2DM in larger populations. The successful identification and application of this knowledge could contribute to our understanding of the metabolic pathophysiology of T2DM in obese populations.