Abstract
Objectives: To analyze the serum and urine metabolites present in type 2 diabetes mellitus (T2DM) patients and T2DM patients with diabetic peripheral neuropathy (DPN) and to select differentially expressed biomarkers for early diagnosis of DPN. Methods: Serum and urine metabolites from 74 T2DM patients with peripheral neuropathy and 41 without peripheral neuropathy were analyzed using gas chromatograph system with time-of-flight mass spectrometer metabolomics to detect biomarkers of peripheral neuropathy in T2DM. Results: There were increased serum triglycerides, alanine aminotransferase, and decreased C-peptide, and total cholesterol levels in T2DM patients with DPN compared to those without peripheral neuropathy. Metabolomic analysis revealed visible differences in metabolic characteristics between two groups, and overall 53 serum differential metabolites and 56 urine differential metabolites were identified with variable influence on projection (VIP) >1 and P<0.05. To further analyze the correlation between the identified metabolites and DPN, four serum metabolites and six urine metabolites were selected with VIP>2, and fold change (FC) >1, including serum β-alanine, caproic acid, β-alanine/L-aspartic acid, and L-arabinose/L-arabitol, and urine gluconic acid, erythritol, galactonic acid, guanidoacetic acid, cytidine, and aminoadipic acid. Furthermore, five serum biomarkers and six urine biomarkers were found to show significant changes (P<0.05, VIP>1, and FC>1) respectively in patients with mild, moderate, and severe DPN. In addition, we found that glyoxylate and dicarboxylate metabolism was a differential metabolic pathway not only between T2DM and DPN, but also among different degrees of DPN. The differential metabolites such as β-alanine and caproic acid are expected to be biomarkers for DPN patients, and the significant changes in glyoxylate and dicarboxylate metabolism may be related to the pathogenesis of DPN. Conclusion: There were serum and urine spectrum metabolomic differences in patients with DPN, which could serve as biomarkers for T2DM and DPN patients.
Keywords: Type 2 diabetes mellitus, diabetic peripheral neuropathy, gas chromatograph system with time-of-flight mass spectrometer, metabolomics
Introduction
Diabetic peripheral neuropathy (DPN) is a type of peripheral nerve dysfunction that is the most common complication of diabetes. Typical DPN is a chronic symmetric, disease-dependent, sensorimotor multiple neuropathy [1]. It features progressive degeneration of the peripheral nerve from the distal to the proximal end, leading to a range of sensory symptoms such as pain, weakness, and/or loss of sensation. The International Diabetes Federation stated that the total number of diabetic patients in the world reached 382 million in 2013 and may increase to 592 million by 2035 [2]. The prevalence of DPN at about 5, 10, and 20 years after being diagnosed with diabetes is about 30%, 60%, and 90%, respectively [3]. DPN is closely related to morbidity and mortality. Currently, there is no effective method to treat DPN except strict blood glucose control. It can be seen that the quality of life of type 2 diabetes mellitus (T2DM) patients is seriously affected by peripheral neuropathy [4]. Since the pathophysiology of DPN is not fully understood, it is important to identify underlying biomarkers that lead to nerve damage and prevent nerve regeneration for successful development of therapeutic intervention [5].
Metabolomics is the high-throughput characterization of small molecule metabolites (<1500 Daltons) in biological matrices using analytical chemistry techniques [6]. It is mainly divided into two analytical techniques. One is called untargeted metabolomics analysis, which uses screening and metabolite identification to detect unknown important and relevant endogenous metabolites. The other is targeted metabolomics, a quantitative analysis of known endogenous metabolites [7]. The metabolome mainly includes small molecule metabolites such as amino acids, organic acids, fatty acids, lipids and carbohydrates. Metabolomics measures metabolic changes to understand the progression of the disease and is a rapidly evolving field in systems biology [8,9]. It is widely used in assessing the metabolic status of diabetes-related diseases and identifying candidate biomarkers [10]. Badeau et al. [11] concluded that short-term rosiglitazone treatment resulted in a slight improvement in metabolism in patients with T2DM and coronary heart disease (CHD) using serum nuclear magnetic resonance (NMR) metabolomics, and changes in serum lactic acid and glutamine concentrations reflected an improved insulin sensitivity. Zhao et al. [12] performed metabolic profiling based on an untargeted high-resolution liquid chromatography-mass spectrometry to identify seven plasma candidate biomarkers in T2DM and non-T2DM in American Indians. Wang et al. [13] identified candidate metabolite markers in diabetic retinopathy by gas chromatography-time-of-flight mass spectrometry (GC-TOFMS), revealing for the first time some novel metabolites in DR, such as d-2,3-dihydroxypropanoic acid, isocitric acid, and threonic acid. Our former study also showed serum and urine differential metabolites of T2DM patients with or without nephropathy based on GC-TOFMS metabolomics [14].
In this study, serum and urine metabolism were analyzed using GC-TOFMS between T2DM patients with or without peripheral neuropathy. In addition, we evaluated serum and urine metabolites in patients with varying degrees of DPN, and explored the relationship between these biomarkers and DPN progression. Our aim was to find alternative indicators to predict the diagnosis and development of DPN.
Materials and methods
Chemicals and reagents
Standards of pyridine, anhydrous sodium sulfate, methoxyamine HCl, and fatty acid methyl ester (C7-C30) were obtained from Sigma-Aldrich (St. Louis, MO, USA). N-methyl-N-trimethylsilyl trifluoroacetamide with 1% (vol) trimethylchlorosilane, methanol, hexane, acetonitrile, chloroform, dichloromethane and acetone were from Thermo-Fisher Scientific (FairLawn, NJ, USA). Ultrapure water was produced by a Mill-Q Reference system with a LC-MS Pak filter (Millipore, Billerica, MA).
Study population
74 T2DM patients with peripheral neuropathy and 41 without peripheral neuropathy were admitted to Shuguang Hospital, Putuo Hospital and Shanghai Traditional Chinese Medicine-integrated Hospital of Shanghai University of Traditional Chinese Medicine from March 2015 to March 2016.
T2DM was diagnosed according to the diabetes diagnosis standards of American Diabetes Association (ADA) in 2010: 1) glycosylated hemoglobin (HbA1c) ≥6.5%; 2) fasting blood glucose (FPG) ≥7.0 mmol/L (no calories intake for at least eight hours was the definition of fasting); 3) 2 hours of blood glucose during oral glucose tolerance test ≥11.1 mmol/L; 4) random blood glucose ≥11.1 mmol/L, in patients with typical hyperglycemia or hyperglycemia crisis [15]. DPN was diagnosed according to the Toronto clinical scoring system (TCSS) proposed by Perkins et al. [16] in 2001. It takes neuroelectro-physiological examination as the gold standard for diagnozing DPN, which has high diagnostic value for DPN and clinical application value in the classification of neuropathy. The TCSS score is evaluated from neurological symptoms, neural reflexes, and sensory function, with a total score of 0-19, as shown in Table S1.
All patients included in this study had no history of hypertension or other complications, and did not use relevant treatment drugs. Patients with neuropathy caused by lumbar spine disease, cerebral infarction, and Guillain-Barré syndrome were also excluded. Patients with severe arteriovenous vascular disease, drug-induced neurotoxicity, and renal insufficiency were also excluded.
According to the grading criteria of TCSS (Table S1), DPN is absent with a score of 0 to 5, mild with a score of 6 to 8, moderate with a score of 9 to 11, and severe with a score of 12 to 19. Thus in the present study, DPN patients were further divided into three groups: mild group (n=37), moderate group (n=24), and severe group (n=13).
Ethical approval of the present study was provided by the Ethics Committee of Shanghai University of Traditional Chinese Medicine. Institutional Review Board (IRB) approval report is provided in Supporting Information. Informed consent from all participants was obtained.
GC-TOFMS analysis
All serum samples were centrifuged for 5 min at 4°C and 3,000 g. 10 μL of internal standard solution, i.e. 2-chloro-phenylalanine with the concentration 0.5 mg/mL was mixed with each 50 μL serum sample, and then with 175 μL of pre-chilled methanol chloroform. Then samples were briefly evaporated to eliminate chloroform using a CentriVap vacuum concentrator (Labconco, Kansas City, MO, USA). We obtained quality control (QC) samples by combining sera from different groups.
All urine samples were centrifuged for 5 min at 4°C and 3,000 g. 10 μL of internal standard solution was mixed with each aliquot of 75 μL urine sample, and then lyophilized with a FreeZone dryer with a stopping tray dryer (Labconco, Kansas City, MO, USA). We obtained QC samples by combining urine from different groups.
After the mixture was stored in a refrigerator at -20°C for 20 min and centrifuged at 14,000 g and 4°C for 20 min, the supernatant was then carefully transferred to an autosampler vial (Agilent Technologies, Foster City, CA, USA). Afterwards, the samples were analyzed by GC-TOFMS system (Pegasus HT, Leco Corp., St. Joseph, MO, USA) coupled to an Agilent 7890B gas chromatography (Gerstel, Muehlheim, Germany) with a Rxi-5 MS capillary column (30 m*250 μm i.d., 0.25-μm; Restek corporation, Bellefonte, PA, USA). Carrier gas flow was 1.0 ml/min using helium. The temperatures of injection interface, transfer interface, and the source temperatures were 270°C, 270°C, 220°C respectively. The electron impact ionization was used 70 eV and mass spectrometry scanning range was 50-500 m/z. The QC sample was injected at the beginning of the run to ensure system equilibrium and subsequently every 15 samples to further monitor the stability of the whole analysis.
Statistical analysis
Clinical data were expressed as medians (interquartile ranges, IQRs). Statistical analysis was performed using SPSS version 26.0 (SPSS Inc., Chicago, IL, USA). Independent sample T-test was used to evaluate the significance of differences between the two groups. One-way analysis of variance (ANOVA) was carried out for comparison of variance of the variables among the 3 groups. A difference with a P value <0.05 was considered significant.
The raw data generated by GC-TOFMS analysis were processed using XploreMET (Metabo-Profile, Shanghai, China) for automated baseline denoizing and smoothing, peak picking and deconvolution, creating a reference database from the pooled QC samples, metabolite signal alignment, missing value correction and imputation, and QC correction. Categorical variables were compared by Chi-square test. Principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) were performed using SIMCA-P12.0 (Umetrics AB). The metabolites with P (adjusted by false discovery rate (FDR) method proposed by Benjamini and Hochburg [17]) <0.05, variable influence on projection (VIP) >1 and fold change (FC) value >1 were considered as candidate biomarkers for classification. Then, the differentially metabolites were presented in volcano plot. Data in Supplementary Materials are shown as medians (IQRs).
The differential compounds were matched to Human Metabolome Database (HMDB, http://www.hmdb.ca/) to identify putative differential metabolites. Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.kegg.jp/) pathway enrichment analysis was performed using the differentially expressed metabolites, and P value <0.05 was considered significantly enriched. Biomarker analysis of differentially expressed metabolites was conducted at MetaboAnalyst 3.0 web server (http://www.metaboanalyst.ca).
The diagnostic performance of biomarkers in serum and urine samples was visualized by receiver operating characteristic (ROC) curves. For the sensitivity, specificity, and predictive values were calculated at 95% confidence intervals (CI) and the area under the curve (AUC).
Results
Clinical characteristics of T2DM patients with and without peripheral neuropathy
Table 1 summarizes the clinical characteristics of 120 subjects, including 74 T2DM patients with and 41 without peripheral neuropathy. As shown in Table 1, there were no significant differences in the age, body mass index (BMI), or gender ratio between the two groups. In the T2DM patients with peripheral neuropathy group, waist-to-hip ratio, triglyceride, alanine aminotransferase (ALT), and aspartate aminotransferase (AST) level were distinctly increased more than those without peripheral neuropathy (P<0.05), while C-peptide, insulin and total cholesterol (TC) were lower than those in T2DM patients without peripheral neuropathy (P<0.05). In addition, accompanied by the severity of DPN, serum TC level gradually decreased. The detailed clinical features of T2DM patients with or without peripheral neuropathy are shown in Table S2.
Table 1.
Clinical information of T2DM patients with and without peripheral neuropathy
| Characteristic | T2DM (n=41) | DPN (n=74) | |||
|---|---|---|---|---|---|
|
| |||||
| Overall | DPN-Mild (n=37) | DPN-Moderate (n=24) | DPN-Severe (n=13) | ||
| Gender (Male/Female) | 29/12 | 46/28 | 19/18 | 19/5 | 8/5 |
| Age (years) | 71 (55-79) | 64 (58.75-74) | 63 (55.5-71) | 66 (60-76) | 62 (60-79) |
| Body mass index (kg/m2) | 24.34 (22.86-26.12) | 24.49 (23.2-27.04) | 24.04 (22.49-26.42) | 24.57 (23.51-26.69) | 26.2 (24.49-28.13) |
| Waist-to-hip ratio | 0.93 (0.91-0.96) | 0.95 (0.93-0.98)* | 0.95 (0.9-0.97) | 0.94 (0.94-0.99) | 0.95 (0.95-0.99) |
| Glucose (mmol/L) | 7.5 (5-9.6) | 7.6 (5.57-9.55) | 7.42 (5.24-9.03) | 7.2 (6.02-8.86) | 8.97 (6.46-10.6) |
| Two hours post-load plasma glucose (mmol/L) | 11 (6.98-14.53) | 11.54 (8.27-14.27) | 13.46 (9.19-14.8) | 9.86 (8.02-13) | 8.41 (6.89-12.1) |
| C-peptide (nmol/L) | 1.41 (0.88-2.01) | 0.71 (0.43-1.01)*** | 0.9 (0.48-1.12) | 0.63 (0.43-0.8) | 0.46 (0.35-0.67) |
| Two hours C-peptide (nmol/L) | 1.96 (1.07-4.01) | 1.34 (0.8-2.75) | 1.55 (0.94-3.12) | 1.38 (0.94-2.45) | 0.8 (0.51-1.5) |
| Insulin (mmol/L) | 69.74 (44.59-107.73) | 12.36 (5.56-40.12)*** | 11.15 (5.56-337.2) | 11.8 (6.8-26.41) | 37.52 (23.1-52.93) |
| HbA1c (%) | 8.4 (6.98-9.58) | 8.5 (7.05-10) | 8.5 (7.2-10.25) | 8.1 (7-8.9) | 9.25 (7.53-10.25) |
| ALT (U/L) | 14 (111-22) | 19.5 (15-27.75)** | 20 (16.2-27) | 18.1 (15-25.45) | 20 (12-32) |
| AST (U/L) | 17 (13-21) | 20.5 (16-25)** | 20 (16-28) | 21.5 (16.75-24.5) | 20 (16-22) |
| Triglyceride (mmol/L) | 1.05 (0.85-1.37) | 3.86 (1.52-4.86)*** | 3.81 (1.49-5.06) | 3.57 (1.49-4.57) | 4.04 (3.3-4.75) |
| TC (mmol/L) | 3.92 (3.53-5.09) | 2.6 (1.55-4.21)*** | 3.48 (2.01-4.51) | 2.75 (1.4-4.03) | 1.36 (0.99-1.84)## |
| High-density lipoprotein (mmol/L) | 1.01 (0.84-1.17) | 1 (0.85-1.13) | 0.91 (0.83-1.06) | 1.04 (0.9-1.21) | 1.08 (0.88-1.16) |
| Low-density lipoprotein (mmol/L) | 2.52 (2.06-3.29) | 2.56 (2.05-3.12) | 2.65 (2.23-3.21) | 2.47 (1.76-3.11) | 2.64 (2.06-2.92) |
Note: Data were expressed as the medians (IQRs). The Wilcoxon test was used to compare these variables between T2DM patients with and without peripheral neuropathy, and P<0.05 was significant. ANOVA was used to compare these variables among the three degrees of DPN.
P<0.05, Comparison of T2DM patients and DPN patients;
P<0.01, Comparison of T2DM patients and DPN patients;
P<0.001, Comparison of T2DM patients and DPN patients;
P<0.01, Comparison of severe DPN patients with moderate DPN patients.
T2DM, Type 2 diabetes mellitus; ANOVA, Wilcoxon test and analysis of variance; DPN, Diabetic peripheral neuropathy.
GC-TOFMS data analysis of serum and urine in T2DM patients with or without peripheral neuropathy
A total of 183 serum metabolites including amino acids, nucleotide, carbohydrates, fatty acids, organic acid, and others were identified. PCA was originally used to summarize GC-TOFMS data and exclude any outliers. From the PCA score plots for serum, it can be seen that most samples were clustered in the 95% confidence interval (the oval area of the score chart). According to the PCA score plots of serum, the clustering trend of the T2DM and DPN groups was not obvious (Figure S1A). To further distinguish the differences between the two groups, OPLS-DA analysis was performed (Figure 1A). The results showed that the established model was effective with good pattern quality. As shown in Figure 1A, the T2DM and DPN groups can be distinguished by serum OPLS-DA scores, and the clustering trend of the two groups was obvious. The values of R2X, R2Y, and Q2 of the two groups of OPLS-DA models were 0.277, 0.895, and 0.778, respectively, which suggested that the established model was reliable. The results of 1000 permutation tests showed that the Q2 intercept of serum model was -0.43, which was a negative value, indicating that the model was effective (Figure 1B).
Figure 1.
OPLS-DA score plot of T2DM patients with or without peripheral neuropathy. A. OPLS-DA score plot of serum metabolites. B. 1000 permutation test plot of serum metabolites. C. OPLS-DA score plot of urine metabolites. D. 1000 permutation test plot of urine metabolites. OPLS-DA, Orthogonal partial least squares-discriminant analysis; T2DM, Type 2 diabetes mellitus; DPN, Diabetic peripheral neuropathy.
We identified 161 urine metabolites including amino acids, nucleotides, carbohydrates, organic acids, vitamin and others. From the PCA score plots for urine, it can be seen that most samples were clustered in the 95% confidence interval (Figure S1B). To further distinguish the differences between the two groups, OPLS-DA analysis was performed (Figure 1C). As shown in Figure 1C, the T2DM and DPN groups can be distinguished, and a clustering trend of the two groups was obvious. The values of R2X, R2Y, and Q2 of the two groups of OPLS-DA models were 0.170, 0.821, and 0.628, respectively, which suggested that the established model was reliable. The results of 1000 permutation tests showed that the Q2 intercept of urine model was -0.393, which was a negative value, indicating that the model was effective (Figure 1D).
Metabolic characters in serum and urine of T2DM patients with or without peripheral neuropathy
Combined with PCA and OPLS-DA analysis, differential serum metabolites were also selected using univariate statistical analysis (Figure S2A). Differential metabolites are shown as an enhanced volcano plot (V-plot) (Figure S2B). The P value and log 1.5 FC were calculated, with cut-off values of 0.05, P values of 0.01, and 1.5 for log 1.5 FC. The metabolites and metabolic pathways were further identified on the basis of accurate molecular weight and mass spectra, and verified using HMDB and KEGG databases. The information of 53 candidate metabolite markers with VIP>1 and P value <0.05 is provided in Table 2. Additional information is provided in Table S3. To further analyze the correlation between the identified metabolites and DPN, 4 metabolites were selected with P<0.05, VIP>2, and FC>1. Among them, Figure 2A shows the 4 most significant changes in serum metabolites. Serum β-alanine, ratio of β-alanine/L-aspartic acid, caproic acid and ratio of L-arabinose/L-arabitol in DPN patients was significantly increased compared to T2DM patients. To determine which serum metabolic pathways are primarily relevant for DPN development, we performed pathway analysis according to their P values in their enrichment and impact values. As shown in Figure S3A, the “metabolomics view” demonstrates all metabolic pathways, arranged based on the scores of enrichment analysis (Y-axis) and topological analysis (X-axis), with the most distinct P values shown in red and the least distinct P values shown in yellow and white. MetaboAnalyst 3.0 was used to perform functional enrichment analysis to observe the biological significance of metabolic changes [18], and metabolite set enrichment analysis (MSEA) on serum metabolites. Finally, there were 6 serum metabolic pathways (P value <0.05) with significant differences respectively (Figure 3A, Table 4). The occurrence of DPN serum metabolism was closely related to 6 metabolic pathways, including arginine biosynthesis, alanine, aspartate and glutamate metabolism, galactose metabolism, starch and sucrose metabolism, valine, leucine and isoleucine biosynthesis, glyoxylate and dicarboxylate metabolism.
Table 2.
Differential serum metabolites between T2DM patients with and without peripheral neuropathy
| Class | Name | P | VIP | FC |
|---|---|---|---|---|
| Amino Acid | β-Alanine | 2.78E-13 | 3.4 | 10.3 |
| Ratio of β-Alanine/L-Aspartic acid | 6.69E-12 | 3.3 | 8.3 | |
| Ratio of Urea/L-Arginine | 2.55E-03 | 1.5 | 2.2 | |
| Ketoleucine | 2.30E-03 | 1.3 | 1.3 | |
| L-Alpha-aminobutyric acid | 3.91E-03 | 1.3 | 1.3 | |
| Ratio of Ketoleucine/L-Leucine | 6.23E-03 | 1.2 | 1.3 | |
| Ratio of L-Serine/Glycine | 8.55E-03 | 1.2 | 1.2 | |
| L-Glutamine | 2.51E-04 | 1.6 | 0.9 | |
| L-Threonine | 1.34E-02 | 1.3 | 0.9 | |
| Ratio of Glycine/L-Serine | 6.48E-03 | 1.2 | 0.9 | |
| Ornithine | 1.91E-02 | 1.2 | 0.9 | |
| Dimethylglycine | 3.21E-02 | 1.2 | 0.9 | |
| Glycine | 3.52E-03 | 1.7 | 0.8 | |
| Ratio of L-Glutamic acid/Pyroglutamic acid | 5.98E-03 | 1.2 | 0.8 | |
| D-2-Hydroxyglutaric acid | 6.17E-03 | 1.2 | 0.8 | |
| Ratio of L-Glutamic acid/Oxoglutaric acid | 8.44E-03 | 1.1 | 0.8 | |
| 4-Hydroxyproline | 1.47E-02 | 1.1 | 0.8 | |
| L-Asparagine | 2.42E-08 | 2.5 | 0.7 | |
| Citrulline | 9.73E-04 | 1.5 | 0.7 | |
| L-Kynurenine | 1.95E-03 | 1.4 | 0.7 | |
| Ratio of L-Asparagine/L-Aspartic acid | 9.87E-04 | 1.3 | 0.7 | |
| 1-Methylhistidine | 9.02E-03 | 1.2 | 0.7 | |
| L-Arginine | 2.53E-06 | 2 | 0.6 | |
| N-Acetyl-L-aspartic acid | 4.40E-03 | 1.1 | 0.6 | |
| Homocysteine | 4.12E-09 | 2.4 | 0.5 | |
| Carbohydrates | Ratio of L-Arabinose/L-Arabitol | 8.12E-07 | 2.3 | 1.5 |
| Allose | 3.67E-03 | 1.3 | 0.9 | |
| D-Galactose | 2.16E-03 | 1.7 | 0.8 | |
| D-Mannose | 1.05E-02 | 1.2 | 0.8 | |
| Erythritol | 1.09E-02 | 1.1 | 0.8 | |
| D-Threitol | 1.10E-02 | 1.1 | 0.8 | |
| Threonic acid | 1.33E-02 | 1.1 | 0.8 | |
| L-Arabitol | 6.11E-06 | 2.1 | 0.7 | |
| D-Xylose | 3.26E-06 | 2 | 0.7 | |
| Rhamnose | 2.96E-04 | 1.6 | 0.7 | |
| Isomaltose | 1.82E-03 | 1.4 | 0.6 | |
| D-Maltose | 4.04E-03 | 1.3 | 0.6 | |
| Ribitol | 9.74E-03 | 1.4 | 0.4 | |
| Fatty Acids | Caproic acid | 3.46E-10 | 3 | 1.9 |
| Arachidic acid | 1.49E-04 | 1.6 | 1.5 | |
| Pelargonic acid | 9.61E-05 | 1.8 | 1.3 | |
| Docosahexaenoic acid | 7.34E-04 | 1.4 | 1.2 | |
| Organic Acids | L-Pipecolic acid | 3.15E-02 | 1.1 | 0.8 |
| Succinic acid | 5.54E-10 | 2.9 | 0.7 | |
| Glycolic acid | 7.71E-04 | 1.6 | 0.7 | |
| Nucleotide | Ratio of Uridine/Cytidine | 3.70E-03 | 1.1 | 1.6 |
| Cytidine | 2.84E-03 | 1.3 | 0.6 | |
| Alcohols | 2-Hydroxypyridine | 7.59E-04 | 1.6 | 0.9 |
| Myoinositol | 2.25E-03 | 1.5 | 0.9 | |
| Glycerol | 1.38E-03 | 1.5 | 0.8 | |
| Lipids | Cholesterol | 2.94E-02 | 1.1 | 1 |
| O-Phosphoethanolamine | 5.94E-05 | 1.8 | 0.7 | |
| Phosphate | Phosphate | 8.27E-05 | 1.9 | 0.7 |
Note: Differential metabolites were selected according to VIP>1 and P<0.05. P values were calculated by t test for continuous variables and adjusted by the FDR method. T2DM, Type 2 diabetes mellitus; VIP, Variable influence on projection; FC, Fold change; FDR, False discovery rate.
Figure 2.
Representative differential metabolites between T2DM patients with or without peripheral neuropathy. A. The peak height comparison of representative serum metabolites between two groups, T2DM (green), DPN (blue), P<0.05, VIP value >2, and FC>1. The statistical analysis of the significance between T2DM patients with or without peripheral neuropathy. β-Alanine, P=2.78E-13; ratio of β-Alanine/L-Aspartic acid, P=6.69E-12; caproic acid, P=3.46E-10; ratio of L-Arabinose/L-Arabitol, P=8.12E-07. B. The peak height comparison of representative urinary metabolites between two groups, T2DM (green), DPN (blue), P<0.05, VIP value >2 and FC>1. The statistical analysis of the significance between T2DM patients with and without peripheral neuropathy group. Gluconic acid, P=2.72E-06; Erythritol, P=3.65E-06; Galactonic acid, P=1.22E-05; Guanidoacetic acid, P=1.74E-05; Cytidine, P=3.53E-05. T2DM, Type 2 diabetes mellitus; DPN, Diabetic peripheral neuropathy; VIP, Variable influence on projection; FC, Fold change.
Figure 3.
Representative differential metabolic pathways between T2DM patients with or without peripheral neuropathy. A. The representative metabolic pathways of serum metabolic pathways. B. The representative metabolic pathways of urine metabolic pathways. The up-regulated or down-regulated metabolites are highlighted with red and yellow respectively. T2DM, type 2 diabetes mellitus; L-Asp, L-aspartic acid; L-Arg, L-arginine; Gly, Glyceric acid.
Table 4.
Serum and urine metabolic pathways that significantly changed between T2DM patients with and without peripheral neuropathy
| Sample | Pathway name | P | Impact | Up | Down |
|---|---|---|---|---|---|
| Serum | Arginine biosynthesis | 2.26E-04 | 0.38 | Urea | Ornithine; L-Arginine; L-Glutamine; Citrulline |
| Galactose metabolism | 9.00E-04 | 0.03 | D-Glucose | Glycerol; D-Galactose; D-Mannose; Myoinositol; Sorbitol | |
| Alanine, aspartate and glutamate metabolism | 3.33E-02 | 0.23 | L-Asparagine; Succinic acid; L-Glutamine; N-Acetyl-L-aspartic acid | ||
| Starch and sucrose metabolism | 4.31E-02 | 0.12 | D-Glucose | D-Maltose; Isomaltose | |
| Glyoxylate and dicarboxylate metabolism | 4.63E-02 | 0.12 | Glycolic acid; Glycine; Isocitric acid; L-Glutamine | ||
| Valine, leucine and isoleucine biosynthesis | 4.64E-02 | 0.25 | Ketoleucine | L-Threonine | |
| Urine | Arginine biosynthesis | 1.32E-05 | 0.56 | L-Arginine | L-Glutamic acid; L-Aspartic acid; Ornithine; L-Glutamine; Citrulline |
| D-Glutamine and D-Glutamate metabolism | 1.75E-02 | 0.67 | L-Glutamic acid; L-Glutamine | ||
| Arginine and proline metabolism | 2.31E-02 | 0.22 | Guanidoacetic acid; L-Arginine | Creatine; L-Glutamic acid; Ornithine | |
| Nitrogen metabolism | 2.56E-02 | 0.25 | L-Glutamic acid; L-Glutamine | ||
| Galactose metabolism | 2.81E-02 | 0.09 | D-Galactose; Sucrose | Myoinositol; Sorbitol | |
| Histidine metabolism | 3.03E-02 | 0.13 | 1-Methylhistidine | L-Glutamic acid; L-Aspartic acid | |
| Glyoxylate and dicarboxylate metabolism | 4.41E-02 | 0.16 | cis-Aconitic acid; Glycolic acid | L-Glutamic acid; L-Glutamine | |
| Ascorbate and aldarate metabolism | 4.51E-02 | 0.2 | D-Glucuronic acid; Myoinositol | ||
| Pantothenate and CoA biosynthesis | 4.78E-02 | 0.11 | Pantothenic acid | β-Alanine; L-Aspartic acid |
Note: Differential metabolic pathways were selected according to P<0.05. T2DM, Type 2 diabetes mellitus.
At the same time, we also used univariate statistical analysis to screen for differential urine metabolites (Figure S2C). Differential metabolites are shown in enhanced V-plot (Figure S2D). The information of 56 candidate metabolite marker with VIP>1 and P value <0.05 is provided in Table 3. Additional information is provided in Table S4. The 6 most significant urinary metabolites are shown in Figure 2B (P<0.05, VIP>2, and FC>1). Urine levels of gluconic acid, erythritol, galactonic acid, guanidoacetic acid, cytidine, aminoadipic acid were observably increased in the DPN group compared to T2DM patients. To determine which urine metabolic pathways are primarily relevant for DPN development, we performed pathway analysis according to their P values in their enrichment and impact values. Figure S3B presented a “metabolomics view” of all metabolic pathways. As shown in Table 4 and Figure 3B, 9 urinary metabolic pathways were associated with the occurrence of DPN, including arginine biosynthesis, arginine and proline metabolism, D-glutamine and D-glutamate metabolism, nitrogen metabolism, histidine metabolism, galactose metabolism, glyoxylate and dicarboxylate metabolism, ascorbate and aldarate metabolism, and pantothenate and CoA biosynthesis.
Table 3.
Differential urinary metabolites in T2DM patients with and without peripheral neuropathy
| Class | Name | P | VIP | FC |
|---|---|---|---|---|
| Amino Acid | Guanidoacetic acid | 1.74E-05 | 2.4 | 1.7 |
| L-Kynurenine | 2.26E-04 | 2.1 | 0.6 | |
| Aminoadipic acid | 6.27E-05 | 2 | 2.1 | |
| L-Glutamine | 2.10E-03 | 2 | 0.7 | |
| Ratio of Citrulline/L-Arginine | 1.69E-03 | 2 | 0.4 | |
| L-Methionine | 4.75E-04 | 1.9 | 0.7 | |
| 3-Nitrotyrosine | 9.05E-04 | 1.8 | 1.9 | |
| Ratio of Guanidoacetic acid/Glycine | 6.62E-03 | 1.8 | 1.8 | |
| Citrulline | 4.68E-03 | 1.7 | 0.8 | |
| 1-Methylhistidine | 9.72E-05 | 1.5 | 1.7 | |
| Ratio of Creatine/Guanidoacetic acid | 2.84E-03 | 1.5 | 0.4 | |
| Ratio of L-Aspartic acid/N-Acetyl-L-aspartic acid | 2.68E-03 | 1.4 | 0.7 | |
| L-Aspartic acid | 9.86E-03 | 1.4 | 0.7 | |
| 4-Hydroxy-L-proline | 4.43E-03 | 1.3 | 0.8 | |
| Ketoleucine | 5.88E-03 | 1.3 | 0.6 | |
| β-Alanine | 4.02E-02 | 1.2 | 0.9 | |
| Creatine | 1.00E-02 | 1.2 | 0.7 | |
| L-Glutamic acid | 4.39E-02 | 1.1 | 0.9 | |
| Ratio of L-Glutamine/L-Glutamic acid | 4.41E-02 | 1.1 | 0.8 | |
| Ratio of L-Serine/Glycine | 4.57E-02 | 1.1 | 0.8 | |
| Carbohydrates | Gluconic acid | 2.72E-06 | 2.8 | 1.7 |
| Galactonic acid | 1.22E-05 | 2.6 | 1.9 | |
| Erythritol | 3.65E-06 | 2.5 | 2.5 | |
| D-Glucuronic acid | 9.72E-05 | 2.3 | 0.7 | |
| Sorbitol | 1.30E-04 | 1.9 | 0.7 | |
| D-Ribose | 1.25E-04 | 1.8 | 1.3 | |
| Ribitol | 3.40E-03 | 1.7 | 1.5 | |
| Rhamnose | 3.34E-04 | 1.6 | 1.8 | |
| D-Galactose | 1.39E-02 | 1.6 | 1.5 | |
| Ratio of Gluconic acid/Gluconolactone | 1.49E-02 | 1.5 | 2.5 | |
| Ribonolactone | 1.06E-02 | 1.4 | 1.3 | |
| Mannitol | 4.66E-02 | 1.3 | 0.7 | |
| D-Xylose | 1.64E-03 | 1.2 | 1.4 | |
| Ratio of L-Arabinose/L-Arabitol | 3.64E-02 | 1.2 | 1.1 | |
| Sucrose | 4.18E-02 | 1.1 | 1.7 | |
| D-Threitol | 4.37E-02 | 1.1 | 1.2 | |
| Organic Acids | Oxalic acid | 2.25E-02 | 1.6 | 1.6 |
| p-Hydroxyphenylacetic acid | 1.49E-02 | 1.6 | 0.9 | |
| Malic acid | 2.15E-03 | 1.6 | 0.5 | |
| Tartaric acid | 7.27E-03 | 1.4 | 0.3 | |
| Suberic acid | 2.32E-02 | 1.3 | 0.7 | |
| Guanidinosuccinic acid | 3.89E-03 | 1.3 | 0.6 | |
| Glycolic acid | 3.66E-02 | 1.2 | 1.4 | |
| Uric acid | 6.12E-03 | 1.2 | 1 | |
| Hydroxyphenyllactic acid | 3.97E-02 | 1.2 | 0.8 | |
| Pimelic acid | 4.64E-02 | 1.1 | 0.8 | |
| Vanillic acid | 1.47E-02 | 1.1 | 0.7 | |
| Nucleotide | Cytidine | 3.53E-05 | 2.3 | 3.1 |
| Ratio of Uridine/Cytidine | 6.34E-04 | 1.7 | 0.3 | |
| Ratio of Hypoxanthine/Inosine | 1.14E-02 | 1.2 | 1.6 | |
| Ratio of Xanthine/Xanthosine | 4.15E-02 | 1.1 | 1 | |
| Alcohols | Myoinositol | 1.65E-02 | 1.1 | 0.7 |
| Lipids | O-Phosphoethanolamine | 3.43E-02 | 1.2 | 1 |
| Vitamin | Pantothenic acid | 1.94E-04 | 1.9 | 1.7 |
| Hormone | Normetanephrine | 4.56E-02 | 1.2 | 0.7 |
| Indoles | Indoxyl sulfate | 6.65E-03 | 1.2 | 0.6 |
Note: Differential metabolites were selected according to VIP>1 and P<0.05. P values were calculated from t test for continuous variables and adjusted by the FDR method. T2DM, Type 2 diabetes mellitus; VIP, Variable influence on projection; FC, Fold change; FDR, False discovery rate.
Predictive abilities of DPN-related biomarkers
It is well-known that untargeted metabolomics data have a wide variety of compound classes, massive data files, complex compound structures, and a lack of databases for searching possible compounds [19]. These factors make it difficult to screen for useful biomarkers and analyze metabolic pathways. In this study, to select the metabolites as candidates that may participate in DPN pathophysiology, compounds with significant changes between groups (P value <0.05, VIP>2 and FC>1) were selected as biomarkers. Four and six metabolites with the most significant changes were found in serum and urine samples from patients with DPN, respectively. Serum metabolites were β-alanine, ratio of β-alanine/L-aspartic acid, caproic acid, and ratio of L-arabinose/L-arabitol (Figure 2A). Urine metabolites were gluconic acid, erythritol, galactonic acid, guanidoacetic acid, cytidine, and aminoadipic acid (Figure 2B). We then performed ROC analysis of serum metabolites and urine metabolites separately to evaluate the clinical potential of these metabolites to predict DPN in T2DM patients. As shown in Figure S4 and Table S5, all four metabolites in serum samples showed good accuracy with high AUC values (AUC=0.946) for the prediction of DPN in T2DM patients. In urine samples, AUC value for these six metabolites were also high, at 0.946 for predicting DPN in T2DM patients.
Serum and urinary metabolic profiles in T2DM patients with peripheral neuropathy of different degrees
We further analyzed the serum metabolites of the three groups with different degrees of DPN. PCA was utilized to show similarities and differences among the mild, moderate, and severe groups of DPN patients in serum metabolites (Figure S5A). For the OPLS-DA model, the score of serum in mild, moderate, and severe groups could easily distinguish the three groups (Figure 4A-F). The values of R2X, R2Y, and Q2 of the three groups of OPLS-DA models were 0.252, 0.444, and -0.0965, respectively. 5 candidate serum biomarkers were finally identified (P<0.05, VIP>1, and FC>1), including β-alanine, hydroxylamine, putrescine, L-kynurenine, and ratio of β-alanine/L-aspartic acid (Figure 5A). Details are shown in Table 5, and additional information is provided in Table S6. So as to ascertain which serum metabolic pathways were most affected in different degrees of DPN, we analyzed these pathways based on the enrichment and impact values of P values. As shown in Figure S6A-C, the “metabolome view”, which demonstrates all metabolic pathways, was arranged with the most significant P values. As shown in Table 6 and Figure 6A, the occurrence of DPN serum metabolism was closely related to 10 metabolic pathways, including arginine biosynthesis, D-glutamine and D-glutamate metabolism, alanine, aspartate and glutamate metabolism, nitrogen metabolism, histidine metabolism, glyoxylate and dicarboxylate metabolism, galactose metabolism, starch and sucrose metabolism, valine, leucine and isoleucine biosynthesis, and cysteine and methionine metabolism.
Figure 4.
OPLS-DA score plots for metabolic profiling in different degrees of DPN. A. OPLS-DA score plot of serum metabolites between mild group and moderate group. B. 1000 permutation test plot. C. OPLS-DA score plot of serum metabolites between mild group and severe group. D. 1000 permutation test plot. E. OPLS-DA score plot of serum metabolites between moderate group and severe group. F. 1000 permutation test plot. G. OPLS-DA score plot of urinary metabolites between mild group and moderate group. H. 1000 permutation test plot. I. OPLS-DA score plot of urinary metabolites between mild group and severe group. J. 1000 permutation test plot. K. OPLS-DA score plot of urinary metabolites between moderate group and severe group. L. 1000 permutation test plot. OPLS-DA, Orthogonal partial least squares-discriminant analysis; DPN, Diabetic peripheral neuropathy.
Figure 5.
Representative differential metabolites in different degrees of DPN. A. Pairwise comparison of peak heights of three representative serum metabolites, Mild (green), Moderate (blue), Severe (red), P<0.05, VIP>1, and FC>1. B. Pairwise comparison of peak heights of three representative urinary metabolites, Mild (green), Moderate (blue), Severe (red), P<0.05, VIP>1, and FC>1. DPN, Diabetic peripheral neuropathy; VIP, Variable influence on projection; FC, Fold change.
Table 5.
Serum and urine metabolites that significantly changed among DPN mild, moderate, and severe groups
| Sample | Class | Name | Mod-Mild | Sev-Mild | Sev-Mod | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
|
|
|||||||||
| P | VIP | FC | P | VIP | FC | P | VIP | FC | |||
| Serum | Amino Acids | β-Alanine | 0.221 | 0.7 | 1.7 | 1.46E-02 | 1.8 | 1.8 | 0.17 | 1.2 | 1 |
| Ratio of β-Alanine/L-Aspartic acid | 0.534 | 0.1 | 1.9 | 4.82E-02 | 1.4 | 2.4 | 0.179 | 1.2 | 1.3 | ||
| L-Kynurenine | 4.12E-02 | 1.5 | 0.9 | 2.00E-02 | 1.6 | 1.1 | 0.399 | 0.8 | 1.2 | ||
| Alkylamines | Hydroxylamine | 0.498 | 0.2 | 0.9 | 4.12E-02 | 1.5 | 1.2 | 1.64E-02 | 2.1 | 1.3 | |
| Putrescine | 4.47E-02 | 2.7 | 1.3 | 0.663 | 0.4 | 0.9 | 0.345 | 1.1 | 0.7 | ||
| Urine | Amino Acids | D-2-Hydroxyglutaric acid | 0.505 | 0.2 | 1.1 | 0.112 | 2 | 1.4 | 3.57E-02 | 2.9 | 1.4 |
| Glycine | 1.02E-02 | 1.9 | 1.6 | 0.386 | 1.4 | 1.3 | 0.319 | 0.8 | 0.8 | ||
| Carbohydrates | L-Arabinose | 0.228 | 0.9 | 0.8 | 0.052 | 2.2 | 1.5 | 1.79E-02 | 2.8 | 1.9 | |
| D-Xylose | 0.099 | 1.3 | 1.2 | 1.71E-02 | 2 | 1.4 | 0.23 | 0.9 | 1.1 | ||
| Organic Acids | Adipic acid | 0.054 | 1.4 | 0.9 | 0.426 | 0.6 | 1.1 | 4.50E-02 | 2.6 | 1.2 | |
| cis-Aconitic acid | 2.19E-02 | 2.5 | 1.5 | 4.57E-02 | 2.3 | 1.6 | 0.583 | 0.4 | 1.1 | ||
Note: Differential metabolites were selected according to P<0.05, VIP>1, and FC>1. ANOVA was conducted in the comparison among DPN different degrees. T-test was performed in the comparison between two groups. DPN, Diabetic peripheral neuropathy; VIP, Variable influence on projection; FC, Fold change; ANOVA, Wilcoxon test and analysis of variance.
Table 6.
Serum and urine metabolic pathways that significantly changed among DPN mild, moderate, and severe groups
| Sample | Differential Metabolic Pathway | Pathway name | P | Impact | Up | Down |
|---|---|---|---|---|---|---|
| Serum | Mod_Mild | Arginine biosynthesis | 7.76E-05 | 0.25 | L-Glutamic acid; Ornithine; L-Glutamine | Fumaric acid; Urea |
| D-Glutamine and D-Glutamate metabolism | 1.18E-02 | 0.67 | L-Glutamic acid; L-Glutamine | |||
| Alanine, aspartate and glutamate metabolism | 1.60E-02 | 0.32 | L-Glutamic acid; L-Glutamine | Fumaric acid; L-Alanine | ||
| Nitrogen metabolism | 1.73E-02 | 0.25 | L-Glutamic acid; L-Glutamine | |||
| Histidine metabolism | 1.76e-02 | 0.33 | L-Glutamic acid | 1-Methylhistidine; L-Histidine | ||
| Sev_Mild | Glyoxylate and dicarboxylate metabolism | 1.06E-02 | 0.2 | Glycolic acid; Glyceric acid; L-Glutamic acid; L-Serine; Isocitric acid | ||
| Galactose metabolism | 2.95E-02 | 0.09 | D-Glucose; D-Galactose; Myoinositol; Sucrose | |||
| Starch and sucrose metabolism | 4.31E-02 | 0.12 | Fructose 6-phosphate | D-Glucose; Sucrose | ||
| Valine, leucine and isoleucine biosynthesis | 4.64E-02 | 0.25 | L-Leucine; Ketoleucine | |||
| Sev_Mod | Glyoxylate and dicarboxylate metabolism | 9.31E-03 | 0.16 | Pyruvic acid | Glycolic acid; L-Glutamic acid; L-Serine; Isocitric acid | |
| Alanine, aspartate and glutamate metabolism | 3.01E-02 | 0.19 | Pyruvic acid | L-Glutamic acid; L-Asparagine; N-Acetyl-L-aspartic acid | ||
| Cysteine and methionine metabolism | 4.65E-02 | 0.15 | L-Cystine; Pyruvic acid | L-Serine; Homocysteine | ||
| Urine | Mod_Mild | Phenylalanine, tyrosine and tryptophan biosynthesis | 1.08E-02 | 0.75 | L-Phenylalanine | Phenylpyruvic acid |
| Glyoxylate and dicarboxylate metabolism | 4.41E-02 | 0.24 | cis-Aconitic acid; Citric acid; Glycolic acid; Glycine | |||
| Valine, leucine and isoleucine biosynthesis | 4.51E-02 | 0 | L-Valine | Ketoleucine | ||
| Sev_Mild | Glyoxylate and dicarboxylate metabolism | 1.52E-03 | 0.32 | cis-Aconitic acid; Citric acid; Glycolic acid; Glycine | Glyceric acid; L-Glutamic acid | |
| Pentose and glucuronate interconversions | 3.83E-02 | 0.14 | D-Xylose; L-Arabinose; L-Arabitol | |||
| Ascorbate and aldarate metabolism | 4.26E-02 | 0.2 | Myoinositol; D-Glucaric acid | |||
| Sev_Mod | β-Alanine metabolism | 1.58E-02 | 0.29 | Uracil | β-Alanine; L-Histidine; L-Aspartic acid | |
| Arginine biosynthesis | 2.65E-02 | 0.12 | Fumaric acid; Oxoglutaric acid | L-Aspartic acid | ||
Note: Differential metabolic pathways were selected according to P<0.05. DPN, Diabetic peripheral neuropathy.
Figure 6.
Representative differential metabolic pathways in different degrees of DPN. A. Representative differential metabolic pathways of serum between mild group and moderate group, between mild group and severe group, between moderate group and severe group. B. Representative differential metabolic pathways of urine between mild group and moderate group, between mild group and severe group, between moderate group and severe group. The up-regulated or down-regulated metabolites are highlighted with red and yellow respectively. DPN, Diabetic peripheral neuropathy; L-Asp, L-aspartic acid; L-Arg, L-arginine; Gly, Glyceric acid.
We also analyzed the urine metabolites of the three groups with different degrees of DPN. PCA was utilized to show similarities and differences among the mild, moderate, and severe groups of DPN patients in urine metabolites (Figure S5B). For the OPLS-DA model, the score of urine in mild, moderate and severe groups could easily distinguish the three groups (Figure 4G-L). The values of R2X, R2Y, and Q2 of the three groups of OPLS-DA models were 0.199, 0.569, and -0.0839, respectively. 6 candidate urine biomarkers were finally identified (P<0.05, VIP>1, and FC>1), including L-arabinose, D-2-hydroxyglutaric acid, adipic acid, cis-aconitic acid, D-xylose, and glycine (Figure 5B, Table 5 and Table S6). So as to ascertain which urine metabolic pathways were most affected in different degrees of DPN, we analyzed these pathways based on the enrichment and impact values of P values (Figure S6D-F). As shown in Table 6 and Figure 6B, 7 urinary metabolic pathways were closely associated with the occurrence of DPN, including glyoxylate and dicarboxylate metabolism, phenylalanine, tyrosine and tryptophan biosynthesis, valine, leucine and isoleucine biosynthesis, pentose and glucuronate interconversions, ascorbate and aldarate metabolism, β-alanine metabolism, and arginine biosynthesis.
Discussion
T2DM is a disorder of glucose and lipid metabolism. Studies show that among adults aged 20-79, the prevalence of diabetes is estimated to be 8.8% by 2015, and will rise to 10.4% by 2040 [20]. In recent years, the incidence of T2DM and its complications has also increased accompanied by changes in people’s lifestyle in China [21]. DPN is a common complication of T2DM, and is found in 50% patients with T2DM [22]. About 11% of patients with DPN have chronic pain symptoms, which severely decreases the quality of life of T2DM patients and is one of the main reasons for the increased incidence and mortality of DPN [23]. According to epidemiologic studies, hypertension and obesity, high blood glucose and glycated hemoglobin levels, duration of diabetes, and increased albumin excretion rates are major risk factors for DPN [24]. However, the pathological progress of DPN is still unclear. Thus, it is important to identify biomarkers in T2DM for better diagnosis of DPN.
In our study, as shown in Table 1, the serum levels of C-peptide in DPN patients were significantly decreased compared to T2DM patients. C-peptide is a hormone-active peptide [25]. Qiao et al. [26] recruited 220 patients with T2DM, assessed DPN by clinical symptoms, signs, and electromyography, and found that C-peptide was negatively correlated with DPN in patients with T2DM. Wahren et al. [27] showed that C-peptide can improve vibration perception threshold compared with the placebo group. These results show that C-peptide replacement therapy can improve the neurological function of patients with early symptoms of diabetic neuropathy, and also provide a basis for the reduction of serum C-peptide in patients with DPN. In addition, we also noticed that serum TC levels in DPN patients were significantly reduced compared with T2DM patients. Along with the severity of DPN, serum TC level gradually decreased. Our finding was consistent with that of Xu et al. [28] which showed that TC was one significant risk determinant for DPN in patients with T2DM. However, the relationship between TC and DPN was unclear until now.
Differential serum and urine metabolites between T2DM patients with or without peripheral neuropathy
In the present study, GC-TOFMS based metabolomics was used to identify related metabolic changes in T2DM patients with and without peripheral neuropathy. We screened out 53 serum metabolites and 56 urine metabolites that had significant changes. We focused on the 4 serum metabolites and 6 urine metabolites that changed most significantly. Simultaneously, we also performed MSEA on metabolites in serum and urine, and screened out 6 serum and 9 urinary metabolic pathways in patients with DPN, respectively. It was shown that compared to T2DM patients, serum level of β-alanine and ratio of β-alanine/L-aspartic acid in DPN patients was significantly increased, and compared to mild DPN patients, severe DPN patients had significantly higher serum β-alanine levels and ratio of β-alanine/L-aspartic acid. When present in high levels, β-alanine is a neurotoxin, and damages the brain or nerve tissue. In addition to neurotoxicity, taurine levels in cells can be reduced by β-alanine, which is also necessary for normal respiratory chain function [29]. It is well-known that depletion of cellular taurine reduces respiratory function and increases mitochondrial superoxide production, thereby damaging mitochondria and increasing oxidative stress [30,31]. Increased oxidative stress leads to a decrease in enzyme activity, which weakens the phosphoryl transfer network and decreases the levels of creatine and pyruvate. Both creatine and pyruvate are important antioxidants and neuroprotective agents [32]. As a result, the weakening of the antioxidant defense capacity will increase the inhibitory effect of the kinase, thereby reducing the content of pyruvate and creatine, and eventually forming a vicious cycle. With the progression of DPN, the continuous accumulation of serum β-alanine and ratio of β-alanine/L-aspartic acid further aggravates the peripheral nerve injury. Therefore, β-alanine and ratio of β-alanine/L-aspartic acid are biomarkers to assess the progress of DPN.
In addition, serum levels of caproic acid in DPN patients were found to be significantly higher than those in T2DM patients without peripheral neuropathy. Caproic acid is a medium-chain fatty acid. Saresella et al. [33] confirmed that caproic acid promotes T cell differentiation, thereby aggravating the inflammatory response of the neurological disease multiple sclerosis. This result suggests that the metabolism of caproic acid in the blood of DPN patients may affect the progression of DPN, which is worthy of further exploration. L-arabinose is a five-carbon sugar like D-xylose [34]. L-arabitol is a downstream product of L-arabinose degradation pathway, and L-arabitol inhibited the growth on L-arabinose [35,36]. Onkenhout et al. [37] have pointed out that L-arabinose metabolism is disturbed at the level of L-arabitol degradation due to the lack of L-arabitol dehydrogenase, resulting in the accumulation of L-arabinose and L-arabitol in patients. Due to the toxicity of L-arabitol to the central nervous system, together with our results, the ratio of L-arabinose/L-arabitol was significantly higher in DPN patients compared with T2DM patients without peripheral neuropathy. The ratio of L-arabinose/L-arabitol may be an important factor affecting the progression of DPN. In addition, there are currently no relevant reports in the literature on the relationship between the 6 urine metabolites and DPN.
ROC curve analysis showed that using these four serum metabolites and six urine metabolites could distinguish patients with DPN from T2DM without it, with good accuracy in both serum and urine samples. Collectively, these results further extend our understanding of key regulatory metabolic pathways involved in the pathophysiology of DPN, as well as provide some promising biomarkers for effective DPN early diagnosis.
Differential serum and urine metabolites in T2DM patients with peripheral neuropathy of different degrees
Serum and urine metabolites of three groups of patients with different degrees of DPN were further analyzed. 5 candidate serum biomarkers and 6 candidate urine biomarkers were found to have the most significant changes. Among them, serum hydroxylamine attracted our attention, and it also shows significant differences in different degrees of DPN. Compared with mild and moderate DPN patients, serum hydroxylamine of severe DPN patients was increased significantly. Hydroxylamine is a derivative of ammonium formed during nitrification and anaerobic ammonium oxidation processes [38]. It is a known genotoxic impurity compound [39]. Toxic reactions occur when the body’s hydroxylamine concentration is significantly higher than normal cellular metabolism [40]. Therefore, elevated hydroxylamine concentrations may be a factor promoting the progression of DPN.
Serum putrescine and L-kynurenine have significant differences in different degrees of DPN. Compared to patients with mild DPN, patients with moderate DPN have significantly higher levels of putrescine in serum. Elevated putrescine is known to cause neurological impairment [41]. Stewart et al. [42] showed that patients with mild cognitive impairment who subsequently developed Alzheimer’s disease had 10% higher putrescine. In combination with the above research, we present that putrescine plays important roles in the development of DPN. Tryptophan is a major precursor in kynurenine pathway and it has been discussed in various in vitro studies that the metabolites quinolinic acid causes neurotoxicity and kynurenic acid acts as neuroprotectants respectively [43]. Studies have shown that L-kynurenine plays a crucial role in the modulation of pain processing both peripherally and centrally [44]. Our results showed that serum L-kynurenine was significantly elevated in patients with severe DPN compared to patients with mild DPN. We speculate that whether L-kynurenine is protective or toxic to the nervous system is concentration-related, and the role of L-kynurenine in the occurrence and development of DPN remains to be further studied.
Compared to mild DPN patients, urine D-xylose of severe DPN patients increased significantly. Kim et al. [45] found that D-xylose regulated blood glucose levels through suppressing phosphoenolpyruvate carboxylase in diabetic rats and enhancing glucose uptake in vitro. This further proves that D-xylose may be a factor affecting the progression of DPN. However, other urine metabolites related to the different development degrees of DPN found in this study have not been mentioned in the relevant literature.
Differential metabolic pathways in T2DM patients with or without peripheral neuropathy and DPN patients of different degrees
We found a number of serum and urine metabolic pathways closely related to DPN. By integrating the information of these metabolic pathways, we found that arginine biosynthesis, galactose metabolism, glyoxylate and dicarboxylate metabolism were the main pathways involved in serum metabolism as well as important pathways in urinary metabolism in T2DM patients with or without peripheral neuropathy. We also found that glyoxylate and dicarboxylate metabolism and valine, leucine, and isoleucine biosynthesis were the main pathways involved both in serum metabolism and urine metabolism in DPN of different degrees. Glyoxylate and dicarboxylate metabolism may be involved not only in the occurrence but also in the progression of DPN. In future research, we will focus on the above related metabolic pathways, and explore key factors regulating progression of DPN.
Limitations
There are some limitations in our current research. First, because different instruments for detecting metabolites have their own limitations, it is impossible to detect all metabolites in serum and urine. Second, there were many metabolites and metabolic pathways screened in this research. Although the data are comprehensive, the research depth is not enough. Third, in order to understand the precise biological processes that lead to the development of DPN by metabolites, it is necessary to further study the relationship between metabolites and related metabolic pathways in combination with other omics techniques or mechanism study.
Conclusion
This study used GC-TOFMS technology to conduct non-targeted metabolomic studies on serum and urine samples of patients with DPN and T2DM. Our key finding is that the occurrence and development of disease in DPN patients are associated with metabolism disorders of amino acids, carbohydrates, fatty acids, and organic acids. In summary, compared to T2DM patients, DPN patients have higher serum levels of β-alanine, ratio of β-alanine/L-aspartic acid, caproic acid, and ratio of L-arabinose/L-arabitol. Compared to mild DPN patients, severe DPN patients have lower serum levels of β-alanine, ratio of β-alanine/L-aspartic acid, L-kynurenine, and hydroxylamine. In addition, we found that glyoxylate and dicarboxylate metabolism was a differential metabolic pathway not only between T2DM and DPN, but also in different degrees of DPN. Further understanding of the role of β-alanine, D-xylose, and other differential metabolites as well as the role of glyoxylate and dicarboxylate metabolic pathway in the pathogenesis of DPN patients may be of great significance for the preliminary clinical screening and diagnosis of DPN.
Acknowledgements
The metabolites involved in this paper can be found in HMDB (http://www.hmdb.ca/), and the metabolic pathway can be found in KEGG (http://www.kegg.jp/). This work was supported by the National Natural Science Foundation of China (81620108030, 81873076) and the Hundred Talents Program from Shanghai University of Traditional Chinese Medicine.
Disclosure of conflict of interest
None.
Supporting Information
References
- 1.Selvarajah D, Kar D, Khunti K, Davies MJ, Scott AR, Walker J, Tesfaye S. Diabetic peripheral neuropathy: advances in diagnosis and strategies for screening and early intervention. Lancet Diabetes Endocrinol. 2019;7:938–948. doi: 10.1016/S2213-8587(19)30081-6. [DOI] [PubMed] [Google Scholar]
- 2.Cho NH, Shaw JE, Karuranga S, Huang Y, da Rocha Fernandes JD, Ohlrogge AW, Malanda B. IDF diabetes atlas: global estimates of diabetes prevalence for 2017 and projections for 2045. Diabetes Res Clin Pract. 2018;138:271–281. doi: 10.1016/j.diabres.2018.02.023. [DOI] [PubMed] [Google Scholar]
- 3.Zheng C, Ou W, Shen H, Zhou Z, Wang J. Combined therapy of diabetic peripheral neuropathy with breviscapine and mecobalamin: a systematic review and a meta-analysis of Chinese studies. Biomed Res Int. 2015;2015:680756. doi: 10.1155/2015/680756. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Naranjo C, Ortega-Jiménez P, Del Reguero L, Moratalla G, Failde I. Relationship between diabetic neuropathic pain and comorbidity. Their impact on pain intensity, diabetes complications and quality of life in patients with type-2 diabetes mellitus. Diabetes Res Clin Pract. 2020;165:108236. doi: 10.1016/j.diabres.2020.108236. [DOI] [PubMed] [Google Scholar]
- 5.Katona I, Weis J. Diseases of the peripheral nerves. Handb Clin Neurol. 2017;145:453–474. doi: 10.1016/B978-0-12-802395-2.00031-6. [DOI] [PubMed] [Google Scholar]
- 6.Wishart DS. Metabolomics for investigating physiological and pathophysiological processes. Physiol Rev. 2019;99:1819–1875. doi: 10.1152/physrev.00035.2018. [DOI] [PubMed] [Google Scholar]
- 7.Koal T, Deigner HP. Challenges in mass spectrometry based targeted metabolomics. Curr Mol Med. 2010;10:216–226. doi: 10.2174/156652410790963312. [DOI] [PubMed] [Google Scholar]
- 8.Liu J, Wang D, Chen Y, Sun H, He S, Wang C, Yang G, Shi M, Zhang J, Ren Y, Wang L, Lu Y, Cheng J. 1H NMR-based metabonomic analysis of serum and urine in a nonhuman primate model of diabetic nephropathy. Mol Biosyst. 2013;9:2645–52. doi: 10.1039/c3mb70212j. [DOI] [PubMed] [Google Scholar]
- 9.Fan Y, Li Y, Chen Y, Zhao YJ, Liu LW, Li J, Wang SL, Alolga RN, Yin Y, Wang XM, Zhao DS, Shen JH, Meng FQ, Zhou X, Xu H, He GP, Lai MD, Li P, Zhu W, Qi LW. Comprehensive metabolomic characterization of coronary artery diseases. J Am Coll Cardiol. 2016;68:1281–1293. doi: 10.1016/j.jacc.2016.06.044. [DOI] [PubMed] [Google Scholar]
- 10.Tam ZY, Ng SP, Tan LQ, Lin CH, Rothenbacher D, Klenk J, Boehm BO SPC Team; ActiFE Study Group. Metabolite profiling in identifying metabolic biomarkers in older people with late-onset type 2 diabetes mellitus. Sci Rep. 2017;7:4392. doi: 10.1038/s41598-017-01735-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Badeau RM, Honka MJ, Lautamaki R, Stewart M, Kangas AJ, Soininen P, Ala-Korpela M, Nuutila P. Systemic metabolic markers and myocardial glucose uptake in type 2 diabetic and coronary artery disease patients treated for 16 weeks with rosiglitazone, a PPARgamma agonist. Ann Med. 2014;46:18–23. doi: 10.3109/07853890.2013.853369. [DOI] [PubMed] [Google Scholar]
- 12.Zhao J, Zhu Y, Hyun N, Zeng D, Uppal K, Tran VT, Yu T, Jones D, He J, Lee ET, Howard BV. Novel metabolic markers for the risk of diabetes development in American Indians. Diabetes Care. 2015;38:220–227. doi: 10.2337/dc14-2033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wang H, Fang J, Chen F, Sun Q, Xu X, Lin SH, Liu K. Metabolomic profile of diabetic retinopathy: a GC-TOFMS-based approach using vitreous and aqueous humor. Acta Diabetol. 2020;57:41–51. doi: 10.1007/s00592-019-01363-0. [DOI] [PubMed] [Google Scholar]
- 14.Shao M, Lu H, Yang M, Liu Y, Yin P, Li G, Wang Y, Chen L, Chen Q, Zhao C, Lu Q, Wu T, Ji G. Serum and urine metabolomics reveal potential biomarkers of T2DM patients with nephropathy. Ann Transl Med. 2020;8:199. doi: 10.21037/atm.2020.01.42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Xu Y, Wang L, He J, Bi Y, Li M, Wang T, Wang L, Jiang Y, Dai M, Lu J, Xu M, Li Y, Hu N, Li J, Mi S, Chen CS, Li G, Mu Y, Zhao J, Kong L, Chen J, Lai S, Wang W, Zhao W, Ning G. Prevalence and control of diabetes in Chinese adults. JAMA. 2013;310:948–959. doi: 10.1001/jama.2013.168118. [DOI] [PubMed] [Google Scholar]
- 16.Perkins BA, Olaleye D, Zinman B, Bril V. Simple screening tests for peripheral neuropathy in the diabetes clinic. Diabetes Care. 2001;24:250–256. doi: 10.2337/diacare.24.2.250. [DOI] [PubMed] [Google Scholar]
- 17.Benjamini Y, Hochberg Y. Controlling the false discovery rate - a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol. 1995;57:289–300. [Google Scholar]
- 18.Xia J, Wishart DS. Using metaboAnalyst 3.0 for comprehensive metabolomics data analysis. Curr Protoc Bioinformatics. 2016;55:14.10.1–14.10.91. doi: 10.1002/cpbi.11. [DOI] [PubMed] [Google Scholar]
- 19.Liu T, Li J, Xu F, Wang M, Ding S, Xu H, Dong F. Comprehensive analysis of serum metabolites in gestational diabetes mellitus by UPLC/Q-TOF-MS. Anal Bioanal Chem. 2016;408:1125–1135. doi: 10.1007/s00216-015-9211-3. [DOI] [PubMed] [Google Scholar]
- 20.Ogurtsova K, da Rocha Fernandes JD, Huang Y, Linnenkamp U, Guariguata L, Cho NH, Cavan D, Shaw JE, Makaroff LE. IDF diabetes atlas: global estimates for the prevalence of diabetes for 2015 and 2040. Diabetes Res Clin Pract. 2017;128:40–50. doi: 10.1016/j.diabres.2017.03.024. [DOI] [PubMed] [Google Scholar]
- 21.Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nat Rev Endocrinol. 2018;14:88–98. doi: 10.1038/nrendo.2017.151. [DOI] [PubMed] [Google Scholar]
- 22.Iqbal Z, Azmi S, Yadav R, Ferdousi M, Kumar M, Cuthbertson DJ, Lim J, Malik RA, Alam U. Diabetic peripheral neuropathy: epidemiology, diagnosis, and pharmacotherapy. Clin Ther. 2018;40:828–849. doi: 10.1016/j.clinthera.2018.04.001. [DOI] [PubMed] [Google Scholar]
- 23.Tesfaye S, Selvarajah D. Advances in the epidemiology, pathogenesis and management of diabetic peripheral neuropathy. Diabetes Metab Res Rev. 2012;28(Suppl 1):8–14. doi: 10.1002/dmrr.2239. [DOI] [PubMed] [Google Scholar]
- 24.Yagihashi S. Recent advances in clinical practice and in basic research on diabetic neuropathy. Brain Nerve. 2011;63:571–582. [PubMed] [Google Scholar]
- 25.Yosten GL, Kolar GR. The physiology of proinsulin c-peptide: unanswered questions and a proposed model. Physiology (Bethesda) 2015;30:327–332. doi: 10.1152/physiol.00008.2015. [DOI] [PubMed] [Google Scholar]
- 26.Qiao X, Zheng H, Zhang S, Liu S, Xiong Q, Mao F, Zhang Z, Wen J, Ye H, Li Y, Lu B. C-peptide is independent associated with diabetic peripheral neuropathy: a community-based study. Diabetol Metab Syndr. 2017;9:12. doi: 10.1186/s13098-017-0208-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wahren J, Foyt H, Daniels M, Arezzo JC. Long-acting C-peptide and neuropathy in type 1 diabetes: a 12-month clinical trial. Diabetes Care. 2016;39:596–602. doi: 10.2337/dc15-2068. [DOI] [PubMed] [Google Scholar]
- 28.Xu F, Zhao LH, Su JB, Chen T, Wang XQ, Chen JF, Wu G, Jin Y, Wang XH. The relationship between glycemic variability and diabetic peripheral neuropathy in type 2 diabetes with well-controlled HbA1c. Diabetol Metab Syndr. 2014;6:139. doi: 10.1186/1758-5996-6-139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Jong CJ, Ito T, Mozaffari M, Azuma J, Schaffer S. Effect of beta-alanine treatment on mitochondrial taurine level and 5-taurinomethyluridine content. J Biomed Sci. 2010;17(Suppl 1):S25. doi: 10.1186/1423-0127-17-S1-S25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Shetewy A, Shimada-Takaura K, Warner D, Jong CJ, Mehdi AB, Alexeyev M, Takahashi K, Schaffer SW. Mitochondrial defects associated with beta-alanine toxicity: relevance to hyper-beta-alaninemia. Mol Cell Biochem. 2016;416:11–22. doi: 10.1007/s11010-016-2688-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Schaffer SW, Shimada-Takaura K, Jong CJ, Ito T, Takahashi K. Impaired energy metabolism of the taurinedeficient heart. Amino Acids. 2016;48:549–558. doi: 10.1007/s00726-015-2110-2. [DOI] [PubMed] [Google Scholar]
- 32.Bortoluzzi VT, Brust L, Preissler T, de Franceschi ID, Wannmacher CMD. Creatine plus pyruvate supplementation prevents oxidative stress and phosphotransfer network disturbances in the brain of rats subjected to chemically-induced phenylketonuria. Metab Brain Dis. 2019;34:1649–1660. doi: 10.1007/s11011-019-00472-7. [DOI] [PubMed] [Google Scholar]
- 33.Saresella M, Marventano I, Barone M, La Rosa F, Piancone F, Mendozzi L, d’Arma A, Rossi V, Pugnetti L, Roda G, Casagni E, Cas MD, Paroni R, Brigidi P, Turroni S, Clerici M. Alterations in circulating fatty acid are associated with gut microbiota dysbiosis and infl ammation in multiple sclerosis. Front Immunol. 2020;11:1390. doi: 10.3389/fimmu.2020.01390. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ye S, Kim JW, Kim SR. Metabolic engineering for improved fermentation of L-arabinose. J Microbiol Biotechnol. 2019;29:339–346. doi: 10.4014/jmb.1812.12015. [DOI] [PubMed] [Google Scholar]
- 35.Seiboth B, Metz B. Fungal arabinan and L-arabinose metabolism. Appl Microbiol Biotechnol. 2011;89:1665–1673. doi: 10.1007/s00253-010-3071-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Izumori K, Yamanaka K. Selective inhibition of Klebsiella aerogenes growth on pentoses by pentitols. J Bacteriol. 1978;134:713–717. doi: 10.1128/jb.134.3.713-717.1978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Onkenhout W, Groener JE, Verhoeven NM, Yin C, Laan LA. L-Arabinosuria: a new defect in human pentose metabolism. Mol Genet Metab. 2002;77:80–85. doi: 10.1016/s1096-7192(02)00125-7. [DOI] [PubMed] [Google Scholar]
- 38.Kumar T, Xavier N, Ramya M. A high-performance liquid chromatography method for determination of genotoxic impurity hydroxylamine in drug substances. J Chromatogr Sci. 2019;57:63–70. doi: 10.1093/chromsci/bmy082. [DOI] [PubMed] [Google Scholar]
- 39.Kumar T, Ramya M, Srinivasan V, Xavier N. A simple and direct LC-MS method for determination of genotoxic impurity hydroxylamine in pharmaceutical compounds. J Chromatogr Sci. 2017;55:683–689. doi: 10.1093/chromsci/bmx019. [DOI] [PubMed] [Google Scholar]
- 40.Gross P. Biologic activity of hydroxylamine: a review. Crit Rev Toxicol. 1985;14:87–99. doi: 10.3109/10408448509023765. [DOI] [PubMed] [Google Scholar]
- 41.Fernandes J, Chandler JD, Liu KH, Uppal K, Go YM, Jones DP. Putrescine as indicator of manganese neurotoxicity: dose-response study in human SH-SY5Y cells. Food Chem Toxicol. 2018;116:272–280. doi: 10.1016/j.fct.2018.04.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Graham SF, Chevallier OP, Elliott CT, Hölscher C, Johnston J, McGuinness B, Kehoe PG, Passmore AP, Green BD. Untargeted metabolomic analysis of human plasma indicates differentially affected polyamine and L-arginine metabolism in mild cognitive impairment subjects converting to Alzheimer’s disease. PLoS One. 2015;10:e0119452. doi: 10.1371/journal.pone.0119452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Venkatesan D, Iyer M, Narayanasamy A, Siva K, Vellingiri B. Kynurenine pathway in Parkinson’s disease-an update. eNeurologicalSci. 2020;21:100270. doi: 10.1016/j.ensci.2020.100270. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Guo S, Vecsei L, Ashina M. The L-kynurenine signalling pathway in trigeminal pain processing: a potential therapeutic target in migraine? Cephalalgia. 2011;31:1029–1038. doi: 10.1177/0333102411404717. [DOI] [PubMed] [Google Scholar]
- 45.Kim E, Kim YS, Kim KM, Jung S, Yoo SH, Kim Y. D-Xylose as a sugar complement regulates blood glucose levels by suppressing phosphoenolpyruvate carboxylase (PEPCK) in streptozotocin-nicotinamide-induced diabetic rats and by enhancing glucose uptake in vitro. Nutr Res Pract. 2016;10:11–18. doi: 10.4162/nrp.2016.10.1.11. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.