Xu WS, Xing H, Wang QQ, Qi H, He JT, Jin T, Kan YP, Sun SY, Wang JY, Lin FQ. Identification and validation of serum amino acids as diagnostic biomarkers for diabetic peripheral neuropathy. World J Diabetes 2025; 16(6): 105592 [DOI: 10.4239/wjd.v16.i6.105592]
Corresponding Author of This Article
Fu-Qing Lin, Chief Physician, Department of Pain Medicine, Shanghai Tenth People's Hospital, Tongji University School of Medicine, No. 301 Yanchang Middle Road, Jingan District, Shanghai 200072, China. fuqinglin@tongji.edu.cn
Research Domain of This Article
Endocrinology & Metabolism
Article-Type of This Article
Prospective Study
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Wei-Sheng Xu, Hui Qi, Jian-Tao He, Tong Jin, Yan-Peng Kan, Shi-Yu Sun, Ji-Ying Wang, Fu-Qing Lin, Department of Pain Medicine, Shanghai Tenth People's Hospital, Tongji University School of Medicine, Shanghai 200072, China
Huan Xing, Qing-Qing Wang, Department of Anesthesiology, Shanghai Tenth People's Hospital, Tongji University School of Medicine, Shanghai 200072, China
Co-corresponding authors: Ji-Ying Wang and Fu-Qing Lin.
Author contributions: Xu WS, Wang JY and Lin FQ conceptualized and designed the research; Wang QQ, Xing H, Qi H, He JT, Jin T and Kan YP screened patients and acquired clinical data; Xu WS, Xing H and Wang QQ collected blood specimen and performed laboratory analysis; Xu WS, Wang JY, Xing H and Lin FQ performed Data analysis; Xu WS, Wang JY and Xing H wrote the paper. Sun SY applied for and obtained the funds for this research project. All the authors have read and approved the final manuscript. Xu WS proposed, designed and conducted serum amino acids analysis, performed data analysis and prepared the first draft of the manuscript. Xing H was responsible for patient screening, enrollment, collection of clinical data and blood specimens. The two authors have made crucial and indispensable contributions towards the completion of the project and thus qualified as the co-first authors of the paper. Wang JY and Lin FQ have played important and indispensable roles in the experimental design, data interpretation and manuscript preparation as the co-corresponding authors.
Supported by the Youth Fund for Specialized Clinical Research of the Health Commission, No. 20214Y0149.
Institutional review board statement: The study was reviewed and approved by the Shanghai Tenth People’s Hospital Institutional Review Board, Approval No. SHYS-IEC-5.0/22K151/PO1.
Clinical trial registration statement: This study is registered at https://www.chictr.org.cn. The registration identification number is ChiCTR2400088659.
Informed consent statement: All study participants, or their legal guardian, provided informed written consent prior to study enrollment.
Conflict-of-interest statement: The authors report no conflicts of interest in this work.
CONSORT 2010 statement: The authors have read the CONSORT 2010 statement, and the manuscript was prepared and revised according to the CONSORT 2010 statement.
Data sharing statement: The datasets included in this study are available from the first author at Xuweisheng203@163.com upon request.
Open Access: This article is an open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: https://creativecommons.org/Licenses/by-nc/4.0/
Corresponding author: Fu-Qing Lin, Chief Physician, Department of Pain Medicine, Shanghai Tenth People's Hospital, Tongji University School of Medicine, No. 301 Yanchang Middle Road, Jingan District, Shanghai 200072, China. fuqinglin@tongji.edu.cn
Received: January 30, 2025 Revised: March 27, 2025 Accepted: May 14, 2025 Published online: June 15, 2025 Processing time: 136 Days and 2.9 Hours
Abstract
BACKGROUND
Diabetic peripheral neuropathy (DPN) is the most prevalent complication of type 2 diabetes mellitus (T2DM). Due to a lack of specific biomarkers, the early diagnosis of this disorder is limited.
AIM
To identify and validate serum amino acids that could discriminate T2DM patients with DPN from those without DPN.
METHODS
T2DM patients with DPN, T2DM patients without DPN, and healthy controls were recruited for this study. The participants comprised two nonoverlapping cohorts: A training cohort (DPN = 84 participants, T2DM = 82 participants, normal = 50 participants) and a validation cohort (DPN = 112 participants, T2DM = 93 participants, normal = 58 participants). A prediction model of the ability of serum amino acids to distinguish DPN from T2DM was established using a logistic regression model, and area under the curve (AUC) analysis was used to evaluate the diagnostic ability of the model. In addition, the serum amino acid levels of 13 DPN patients were also detected before treatment and after 3 months of treatment.
RESULTS
A clinical detection method for the diagnosis of DPN based on a biomarker panel of three serum amino acids and diabetes duration was developed. The diagnostic model demonstrated AUC values of 0.805 (95%CI: 0.739-0.871) and 0.810 (95%CI: 0.750-0.870) in the training and verification cohorts, respectively. In the identification of T2DM patients and normal controls, the AUC values were 0.891 (95%CI: 0.836-0.945) and 0.883 (95%CI: 0.832-0.934) in the training and validation cohorts, respectively. Arginine and tyrosine levels were increased after treatment, whereas aspartic acid levels were decreased after treatment.
CONCLUSION
This study successfully identified and validated the metabolomic significance of arginine, tyrosine, and glutamic acid as potential biomarkers for diagnosing DPN. These findings are particularly valuable, as they establish a foundational step toward developing the first routine laboratory test for DPN. Moreover, the diagnostic model that was constructed in this study effectively distinguishes DPN patients from those with T2DM without neuropathy, thereby potentially facilitating early diagnosis and intervention.
Core Tip: This study involves the most extensive identification of serum amino acids as biomarkers for the detection of diabetic peripheral neuropathy (DPN), including the use of both training and validation cohorts. This is the first study demonstrating that a signature consisting of 3 serum amino acids (arginine, tyrosine, and glutamic acid) and diabetes duration can successfully distinguish DPN from type 2 diabetes mellitus with acceptable accuracy. Thirteen DPN patients received treatment in our department, and their clinical symptoms were relieved; moreover, their serum arginine and tyrosine levels were significantly increased, which may reflect the reliability of our model. These findings may provide potential therapeutic targets for the diagnosis and treatment of DPN.
Citation: Xu WS, Xing H, Wang QQ, Qi H, He JT, Jin T, Kan YP, Sun SY, Wang JY, Lin FQ. Identification and validation of serum amino acids as diagnostic biomarkers for diabetic peripheral neuropathy. World J Diabetes 2025; 16(6): 105592
Diabetic peripheral neuropathy (DPN) is among the most rapidly increasing diseases worldwide and represents the most prevalent complication of type 2 diabetes mellitus (T2DM), affecting more than 50% of patients[1,2]. The global prevalence of diabetes is estimated to increase to 12.2% (783.2 million individuals) by 2045, with T2DM accounting for 95% of cases[3]. DPN is characterized by neuropathic pain, numbness, and sensory abnormalities in symmetrical, bilateral distal limbs resulting from injury to peripheral motor and sensory nerves[4]. Although substantial progress has been made in exploring the potential pathogenesis of DPN in recent years[5], few advances have been made regarding clinical diagnosis and treatment.
Prediabetic patients may exhibit neuropathy, and the onset of DPN is sometimes insidious, with many patients experiencing asymptomatic onset of the condition for a long period of time[6]. Additionally, current methodological tools do not effectively screen all DPN patients, thus leading to potential missed diagnoses. Therefore, it is imperative to investigate a serum biomarker for the identification of DPN, especially at the early stage when preventive measures can be most effective. Adequate prevention and desirable diagnostic screening are highly valuable for DPN patients and can contribute to reducing the severe impact of DPN on patients’ health, as well as the associated considerable financial burden.
Given that metabolic factors play an important role in the progression of DPN[7] and with the development of metabolomics, certain serum metabolites have been demonstrated to be associated with DPN[8,9], mainly involving the metabolism of three major nutrients. As one of the most fundamental substances in the human body, serum amino acids are the basic units of protein, and the metabolism of serum amino acids, glucose, and lipids is interconnected and interactive[10]. The relationship between serum amino acid metabolism and DPN has received considerable attention, and previous studies have confirmed that some serum amino acids are associated with DPN; however, the results have not been consistent[11-14]. Moreover, these studies have demonstrated several disadvantages, such as a lack of hypoglycemic medications, a small sample size, and a lack of a prediction model and external validation.
Our study aimed to distinguish DPN from T2DM by establishing a training cohort and a validation cohort. A total of 479 participants were recruited for the study, including normal controls and T2DM patients with and without peripheral neuropathy. To identify serum amino acids that can potentially improve the prediction of the future manifestation of DPN, a liquid chromatography-mass spectrometry (LC-MS) metabolomics approach based on multiple serum amino acid sample release agents was used for the identification of metabolic alterations associated with DPN onset, with 20 serum amino acids being detected. The DPN-associated serum amino acids were subsequently validated in an independent cohort. Thirteen DPN patients were also observed to determine whether their serum amino acid levels were altered after treatment, thus providing an early intervention therapeutic strategy.
MATERIALS AND METHODS
Participants and design
The study design and workflow are illustrated in Figure 1, and the reported research aligns with the STROCSS criteria[15]. Two normal control groups (a training cohort of 50 individuals and a validation cohort of 58 individuals), T2DM patients without DPN (a training cohort of 82 patients and a validation cohort of 93 patients), and T2DM patients with DPN (a training cohort of 84 patients and a validation cohort of 112 patients) were consecutively enrolled in this study at Shanghai Tenth People’s Hospital from October 2022 to March 2024. All of the T2DM patients were recruited according to the American Diabetes Association diagnostic criteria of 2019[16]. Patients were excluded if they had any of the following conditions: History of cerebral infarction; heart disease; liver cirrhosis; liver or renal dysfunction; malignant tumors; gastrointestinal diseases; acute complications of diabetes and other endocrine diseases; arteriovenous vascular diseases; cervical and lumbar spine lesions; other diseases affecting peripheral nerve function; tuberculosis; viral hepatitis; other infectious diseases; or histories of alcohol consumption and smoking.
Figure 1 Flow diagram of the cohorts used for the generation and validation of serum amino acid signatures.
DPN: Diabetic peripheral neuropathy; T2DM: Type 2 diabetes mellitus.
Assessment of peripheral neuropathy
The inclusion criteria for the DPN group[4,17] were as follows: (1) Main clinical symptoms such as numbness, pain, or other signs of paresthesia at the ends of the limbs; (2) A Toronto Clinical Scoring System (TCSS) score of no less than six[18]; and (3) Electromyography findings indicating multiple peripheral neuropathies[19]. All of the patients were divided into the following two groups: T2DM patients without DPN (the T2DM group) and T2DM patients with DPN (the DPN group).
Collection of clinical data and biochemical indices
The patients’ demographic characteristics and blood biochemical indices included the following parameters: Age, sex, diabetes duration, body mass index (BMI), TCSS score, visual analog scale (VAS) score, patient health questionnaire-9 (PHQ-9), the 7-item generalized anxiety disorder questionnaire (GAD-7), medication use, systolic blood pressure, diastolic blood pressure, fasting blood glucose, glycated hemoglobin A1c (HbA1c), albumin (ALB), brain natriuretic peptide (BNP), aspartate aminotransferase (AST), alanine aminotransferase, glomerular filtration rate, serum creatinine, uric acid (UA), triglyceride, high-density lipoprotein, low-density lipoprotein, parathyroid hormone, thyroid stimulating hormone (TSH), free triiodothyronine, and free thyroxine (FT4).
The following electromyography indicators were used: Common peroneal nerve F wave latency (CPNFL), common peroneal nerve motor amplitude (CPNMA), common peroneal nerve motor conduction velocity (CPNMCV), common peroneal nerve motor latency, median nerve F wave latency, median nerve motor amplitude, median nerve motor conduction velocity (MNMCV), median nerve motor latency, median nerve sensory amplitude (MNSA), median nerve sensory conduction velocity (MNSCV), median nerve sensory latency, superficial peroneal nerve sensory amplitude, superficial peroneal nerve sensory conduction velocity, superficial peroneal nerve sensory latency, ulnar nerve F wave latency (UNFL), ulnar nerve motor amplitude, ulnar nerve motor conduction velocity (UNMCV), ulnar nerve motor latency, ulnar nerve sensory amplitude (UNSA), ulnar nerve sensory conduction velocity (UNSCV), and ulnar nerve sensory latency.
Measurement of serum amino acids
A specific kit (Hangzhou Baichen Medical Equipment Limited Company) was used to obtain serum amino acids, which was able to detect the following 20 amino acids simultaneously via LC-MS[20]: Glycine, serine, asparagine, arginine, alanine, homoproline, proline, valine, methionine, tyrosine, phenylalanine, isoleucine, leucine, hydroxyproline, threonine, tryptophane, glutamic acid, aspartic acid, histidine and lysine.
Preprocessing
Venous blood samples (3.0 mL) were collected from the study participants in the morning by professional nurses in strict accordance with standard aseptic procedures after an overnight fast of at least 8 hours. The samples were placed into EDTA-containing anticoagulant vacuum negative pressure blood collection tubes (purple) and subsequently centrifuged at 3000 revolutions per minute (rpm) for 10 minutes at 4 °C to obtain the serum. A variety of amino acid standards and isotope internal standards were used, and dissolution diluent (liquid A) was used to prepare the internal standard working fluid. Additionally, 50 μL of serum and quality control samples were pipetted into a 1.5 mL centrifuge tube; afterwards, 400 μL of amino acid internal standard working mixture was added, and the mixture was subsequently vortexed for 3 minutes.
After the mixture was homogenized, it was centrifuged at 11000 rpm for 10 minutes. Seventy microliters of the supernatant was then pipetted into a V-type 96-well plate, which was subsequently placed under a nitrogen-blowing instrument to remove the solvent.
Derivatization procedure
Sixty microliters of derivating fluid (liquid B) was added to the blow-dried sample, which was subsequently placed in a thermostatic shaking incubator to shake and incubate for 30 minutes (700 rpm, 60 °C) and blown dry with nitrogen. Afterwards, 100 μL of redissolved liquid (liquid C) was pipetted into a V-type 96-well plate, which was subsequently placed in a thermostatic shaking incubator to shake and incubate for 3 minutes (700 rpm, room temperature). Finally, upon completion of incubation, the microplate was placed in a sample manager system for analysis.
Sample detection
The test samples were placed in a high-performance liquid chromatography tandem mass spectrometer for detection with a positive electrospray ionization ion source in a multiple reaction ion monitoring mode. The capillary voltage was 3.0 kV, and the source temperature was 150 °C. The desolventization gas temperature was 400 °C, and the flow rate was 800 L/hour. Moreover, the flow rate of the cone hole blowback gas was 50 L/hour. The desolventization gas and cone hole blowback gas were nitrogen (purity of 99.9%), and the collision gas was argon (purity of 99.999%).
Statistical analysis
One-way ANOVA was used to compare the three groups, due to the fact that the continuous variables, which are expressed as the mean ± SD, were approximately normally distributed. Categorical variables are presented as proportions and were analyzed by using the χ2 test. A paired t test was performed to analyze the differences between pretreatment and posttreatment values. A backward stepwise logistic regression (LR) model was used to select a set of variables based on the training cohort. A receiver operating characteristic (ROC) curve was established, and the area under the curve (AUC) value was calculated to examine the diagnostic value of the prediction model.
Pearson correlation analysis was used to determine the correlations between amino acids and both electromyogram indicators and clinical scores.
All of the statistical analyses were performed using SPSS software (version 25.0), GraphPad Prism software (version 9, United States), R software (version 4.2.3), and the MetWare Cloud platform (https://cloud.metware.cn).
RESULTS
Clinical features and basic information
The 479 enrolled participants originated from three different groups, and the characteristics of the two cohorts are shown in Table 1. In both cohorts, the duration of diabetes in DPN patients was longer than that in T2DM patients (P < 0.05). There was no significant difference observed between the T2DM and DPN groups in terms of other clinical indicators (Table 1) or medication use (Supplementary Table 1).
Table 1 Clinical characteristics of the participants in the two cohorts, n (%).
The serum concentrations of 20 amino acids are shown in Table 2. Arginine and tyrosine levels were significantly lower, whereas glutamic acid levels were greater, in DPN patients than in T2DM patients in both cohorts (Figure 2). The other serum amino acids were not significantly different between the two groups (Supplementary Figure 1). There were significant increases in the serum valine, leucine, isoleucine, and homoproline levels, as well as a significant decrease in the arginine level, in T2DM patients compared with normal controls (Figure 2).
Figure 2 Scatter plots of the serum relative concentrations of amino acids among diabetic peripheral neuropathy patients, type 2 diabetes mellitus patients and normal controls in the training cohort and validation cohort.aP < 0.05; bP < 0.01; cP < 0.001; dP < 0.0001; NS: Not significant; TC: Training cohort; VC: Validation cohort (one-way ANOVA).
Table 2 Serum levels of twenty amino acids among diabetic peripheral neuropathy patients, type 2 diabetes mellitus patients and normal individuals in the two cohorts.
In the training cohort, the serum samples of 84 DPN patients, 82 T2DM patients, and 50 normal controls were used to establish two LR models. Three serum amino acids and diabetes duration (which exhibited significant differences between DPN patients and T2DM patients) were selected as the independent variables, and the dependent variable was the presence or absence of DPN in the patients. The LR model (Table 3) was created as follows: Logit (P = DPN) = -0.062 × arginine - 0.028 × tyrosine + 0.010 × glutamic acid + 0.060 × duration + 1.896. Moreover, five serum amino acids and glucose (which exhibited significant differences between T2DM patients and normal controls) were selected as independent variables, and the dependent variable was the presence or absence of T2DM in the patients. The LR model (Table 4) was created as follows: Logit (P = T2DM) = -0.072 × arginine + 0.060 × homoproline + 0.010 × valine + 0.027 × leucine + 0.246 × glucose - 6.046. Finally, the predicted probabilities for the 84 DPN patients and 82 T2DM patients were calculated and utilized for ROC analysis to determine the optimal threshold (P = 0.502). The model achieved an AUC of 0.805 in distinguishing DPN patients from T2DM patients, with a sensitivity of 75.0%, a specificity of 73.2%, and an accuracy of 74.1% (Figure 3A). In the training cohort, the LR model discriminated T2DM patients from normal controls, with an AUC of 0.891, a sensitivity of 81.7%, a specificity of 86.0%, and an accuracy of 83.3% at a cutoff value of P = 0.590 (Figure 3B). The predictive value of fasting blood glucose for detecting T2DM exhibited an AUC of 0.620 (Supplementary Figure 2), which was lower than that of our model. Furthermore, the calibration curve was close to the ideal diagonal line (Figure 4A and B), and the Hosmer-Lemeshow test demonstrated that the model aligned with the observed data (P > 0.05).
Figure 3 Receiver operating characteristic area under the curve for the specificity and sensitivity of the predictive model in the training cohort and validation cohort.
A: Performance of the prediction model for discriminating diabetic peripheral neuropathy (DPN) patients from type 2 diabetes mellitus (T2DM) patients; B: T2DM patients from normal controls; C: Discriminating DPN patients from T2DM patients; D: T2DM patients from normal controls. AUC: Area under the curve; T2DM: Type 2 diabetes mellitus; DPN: Diabetic peripheral neuropathy; TC: Training cohort; VC: Validation cohort.
Figure 4 Calibration curve for predicting the probability of diabetic peripheral neuropathy and type 2 diabetes mellitus.
A: Training cohort (TC) of diabetic peripheral neuropathy (DPN) patients and type 2 diabetes mellitus (T2DM) patients; B: TC of T2DM patients and normal controls; C: Validation cohort (VC) of DPN patients and T2DM patients; D: VC of T2DM patients and normal controls. AUC: Area under the curve; T2DM: Type 2 diabetes mellitus; DPN: Diabetic peripheral neuropathy; TC: Training cohort; VC: Validation cohort.
Table 3 Backward stepwise logistic analysis of diabetic peripheral neuropathy and type 2 diabetes mellitus patients.
Variables
Β
SE
Wald
P value
Arginine
-0.062
0.014
18.786
< 0.001
Tyrosine
-0.028
0.009
9.782
0.002
Glutamic acid
0.01
0.004
4.728
0.03
Duration
0.06
0.027
4.765
0.029
Constant
1.896
1.189
2.54
0.111
Table 4 Backward stepwise logistic analysis of type 2 diabetes mellitus patients and normal controls.
Variables
β
SE
Wald
P value
Arginine
-0.072
0.019
14.218
< 0.001
Homoproline
0.06
0.027
4.808
0.028
Valine
0.01
0.004
6.879
0.009
Leucine
0.027
0.01
7.975
0.005
Glucose
0.246
0.109
5.047
0.025
Constant
-6.046
1.87
10.45
0.001
External validation of the diagnostic model
External validation was used to assess the classification performance of the LR model for DPN and T2DM patients. Discrimination between DPN and T2DM yielded an AUC of 0.810, with a sensitivity of 79.5%, a specificity of 69.9%, and a diagnostic accuracy of 75.1% (Figure 3C). Additionally, the model was able to discriminate T2DM patients from normal controls, with an AUC of 0.883, a sensitivity of 73.1%, a specificity of 91.4%, and a diagnostic accuracy of 80.1% (Figure 3D). The calibration curve of the validation cohort closely followed the ideal diagonal line (Figure 4C and D). Additionally, the Hosmer-Lemeshow test confirmed that the model aligned well with the observed data. The results of the decision curve analysis for the nomogram are presented in Figure 5. In both the training and validation cohorts, the decision curve demonstrated that if the threshold probability of a patient was in the range of 0-0.99, the use of the models achieved more net benefits compared to the “treat all” or “treat none” strategies, thus suggesting that the models were good assessment tools.
Figure 5 Decision curve analysis.
A and B: Decision curve analysis (DCA) for diabetic peripheral neuropathy prediction in the training cohort (A) and validation cohort (B); C and D: DCA for type 2 diabetes mellitus prediction in the training cohort (C) and validation cohort (D). T2DM: Type 2 diabetes mellitus; DPN: Diabetic peripheral neuropathy; TC: Training cohort; VC: Validation cohort.
Relationships between serum amino acids, electromyography parameters and clinical scores
Correlation heatmaps were generated to estimate the relationships between serum amino acid levels and electromyography parameters (Figure 6A) and between serum amino acid levels and clinical scores (Figure 6B). Arginine, tyrosine, and glutamic acid were most strongly associated with electromyographic parameters. Specifically, glutamic acid was negatively correlated with MNSCV, UNSA and UNSCV and positively correlated with CPNFL. Moreover, arginine and tyrosine were positively correlated with UNMCV, MNMCV, CPNMCV, UNSCV, MNSCV, CPNSCV, UNSA, MNSA and CPNMA but negatively correlated with UNFL and CPNFL. Arginine was significantly negatively correlated with the VAS, TCSS, GAD-7, and PHQ-9 scale scores. Furthermore, aspartic acid was positively correlated with the GAD-7 and PHQ-9 scale scores.
Figure 6 Heatmap correlation between the measured serum amino acid levels and both electromyography parameters and clinical scores.
A: Electromyography parameters; B: Clinical scores. aP < 0.05; bP < 0.01; cP < 0.001; PHQ: Patient health questionnaire; VAS: Visual analog scale; TCSS: Toronto Clinical Scoring System; GAD: Generalized anxiety disorder questionnaire.
Changes in serum amino acid levels before and after treatment
Some DPN patients were treated with mecobalamin, pregabalin, epalrestat, or stellate ganglion pulsed radiofrequency. A total of 13 patients were followed up after 3 months, and three serum amino acids were observed to be altered. Specifically, arginine and tyrosine levels were significantly increased, whereas aspartic acid levels were significantly decreased (Figure 7). The other serum amino acid levels did not differ before and after treatment (Supplementary Table 2).
Figure 7 Changes in the serum arginine, aspartic acid and tyrosine levels of 13 diabetic peripheral neuropathy patients after treatment for 3 months.aP < 0.05; dP < 0.0001.
DISCUSSION
Recent evidence suggests that multiple metabolic disorders are involved in the pathogenesis of DPN; however, the results of different studies have not been entirely consistent[21-23]. Our study is the first to demonstrate and validate biomarkers of 20 serum amino acids for DPN and highlight potential diagnostic applications in T2DM patients with DPN. A profile of the concentrations of multiple biochemical molecules is likely to be advantageous over the use of a single biomarker for detecting DPN, due to the fact that the dysfunction of many metabolic pathways may precede the development of T2DM. To date, no biomarker has been recommended for the diagnosis of DPN in medical guidelines. In recent years, several serum indicators, such as ALB, TSH, FT4, BNP, and UA, have been detected as promising DPN biomarkers[24-31]. Although these studies have offered promising results, the abovementioned candidate biomarkers have largely lacked reproducibility, and the results have been inconsistent. Moreover, the abovementioned markers were not observed to be significantly different between DPN patients and T2DM patients in our study, which was likely due to the fact that age, BMI and other basic indicators were not different between the two groups.
When considering DPN as a metabolic disease, a combination of markers (rather than a single serum indicator) may provide adequate discrimination. In this study, 2 independent cohorts from 2 different time periods with a total of 479 individuals were evaluated to identify and validate biomarkers for the prediction of DPN. This is the first study demonstrating the use of a signature consisting of 3 serum amino acids (arginine, tyrosine, and glutamic acid) and diabetes duration that could successfully distinguish between DPN and T2DM patients with acceptable accuracy (AUC > 0.8) in serum samples. These biomarkers could support the diagnosis of DPN and decision-making regarding a treatment strategy that would lead to earlier treatment initiation and better outcomes for patients. In conclusion, our study identified novel potential diagnostic blood-based amino acid signatures that may help to discriminate T2DM patients with DPN from those without DPN. Therefore, our findings contribute to knowledge in the field in terms of which amino acids may contribute to the ability to predict DPN. The early and accurate diagnosis of DPN would ensure more effective clinical management and reduce financial and life burdens.
Recently, metabolomics has been rapidly and widely developed into a useful technique for quantitatively analyzing multicomponent mixtures of biological samples, which provides important information for understanding the composition and function of biochemical networks[32]. As one of the most basic substances in the human body, amino acids are the basic units that make up proteins and constitute the material basis for metabolism[33]. Some amino acids generate neurotransmitters and antioxidants, as well as promote angiogenesis, among other functions; moreover, deficiencies in various amino acids can lead to the development of metabolic disorders in the body[10]. Patients with decompensated cirrhosis exhibit amino acid imbalances, and alcohol intake has been observed to exacerbate polyneuropathy[34,35]. Therefore, these types of patients were excluded from our study design. We identified and validated a metabolomic signature of serum amino acids that can successfully discriminate between DPN patients and T2DM patients. This signature demonstrates a potentially significant role in the future prediction and diagnosis of DPN because of the lack of effective biomarkers for early diagnosis of this condition. Although the amino acid signatures used in the current study require further independent validation, they demonstrated promising performance.
To the best of our knowledge, this study represents the most extensive identification of amino acids as biomarkers for the detection of DPN, including both training and validation cohorts. Previous studies exploring some amino acids associated with DPN, such as arginine[36] and serine[37,38], have demonstrated promising results but remain largely inconclusive, which is possibly due to small cohort sizes and a lack of validation technology, thus limiting further applications. In a previous study, low arginine levels were observed to contribute to the prediction of DPN, which was consistent with our findings; however, the effects of drug use were not considered in this previous study[36]. Moreover, clinical trials of oral arginine supplementation in DM or DPN patients have yielded inconsistent results[39]. Therefore, the possibility that the restoration of low arginine levels toward the normal range may diminish the onset of DPN in patients with T2DM and alleviate its progression to DPN seems worthy of consideration and prospective research. Moreover, there was no significant difference observed in serine levels between the T2DM group and the DPN group in our study, whereas other studies have indicated that a decrease in serine promotes the occurrence and development of DPN[37,38], in which obesity and dyslipidemia may play important roles.
The correlation analyses provided a predictive effect for the relationships between amino acids and both neural parameters and clinical scores. There was a significant positive correlation observed between both arginine and tyrosine and nerve conduction velocity, whereas glutamic acid was negatively correlated with nerve conduction velocity. In addition, 13 DPN patients received treatment in our department, and their clinical symptoms were relieved; moreover, their serum arginine and tyrosine levels significantly increased, which may reflect the reliability of our model. Arginine is a substrate for nitric oxide (NO) synthesis, which maintains glucose homeostasis in the body, improves insulin sensitivity, inhibits oxidative stress caused by diabetes, and improves the vasodilation response via NO[40]. A previous study revealed that low arginine could predict DPN[36], and other studies have demonstrated that L-arginine supplementation can prevent occurrences of mechanical hyperalgesia and abnormal heat pain in DPN rats[41]. As an excitatory amino acid in the dorsal horn of the spinal cord, glutamic acid plays an important role in transmitting nociceptive information. Mitochondrial dysfunction caused by hyperglycemia can be further enhanced by glutamic acid, and excessive glutamic acid release is harmful to neurons[11]. Pulsed radiofrequency neuromodulation reduces hyperalgesia and ectopic pain in DPN mice, which may be related to the inhibition of glutamate release from the spinal cord, thereby attenuating pain sensitivity in the dorsal horn of the spinal cord[42]. Our research results are consistent with these studies, and a diagnostic prediction model utilizing a combination of three amino acids would have very broad application prospects.
Compared with glucose alone, a panel of four serum amino acids (arginine, homoproline, valine and leucine) in combination with glucose demonstrated a slightly improved ability to discriminate T2DM patients from normal controls in our study. This panel was validated in an independent group and could exhibit some useful applications in the diagnosis of T2DM. However, HbA1c was excluded from our model due to the hospitalized T2DM patients with extremely poor glycemic control who exhibited significantly different HbA1c levels from the normal range. Further validation of these serum amino acids is needed to confirm whether such metabolites promote the development of diabetes or whether these metabolite alterations result from hyperglycemia.
There were several limitations in this study. First, serum amino acids were not measured at the same time for all of the participants, which could introduce a potential bias. However, the overall conclusions may not be substantially affected, due to the fact that all of the samples were detected using the same equipment. Second, the participants in the training and validation cohorts originated from the same hospital and had similar climatic characteristics and dietary patterns. Therefore, the results of this study may only reflect the situation of DPN in the region and are not representative. Third, patients with clinical syndromes but normal electromyogram data were excluded. Such patients may be in the early stages of DPN because electromyography generally reveals large-fiber neuropathy, whereas abnormal sensation is usually related to small nerve fibers[43], which can be reversed with early intervention. Fourth, the 13 patients who received treatment were part of descriptive research (rather than being involved in a before-after study) and may have changed their eating behaviors and withheld information during the course of the study. Therefore, further studies evaluating serum amino acid predictors in larger, multicenter cohorts at different stages of DPN are needed to translate our findings into beneficial clinical applications.
CONCLUSION
In conclusion, this study suggests that a prediction model combining three serum amino acids (arginine, tyrosine, and glutamic acid) and the duration of diabetes could distinguish T2DM patients with DPN from those without DPN. This measurement represents a useful auxiliary detection tool for the diagnosis of DPN and has the potential to be a cost-effective screening tool for DPN. Whether this model has clinical value for diagnosing the early stages of DPN or for monitoring disease progression requires further prospective research.
ACKNOWLEDGEMENTS
All of the authors are very grateful to the nurses in our department for their valuable assistance and to all of the patients involved in our study.
Footnotes
Provenance and peer review: Unsolicited article; Externally peer reviewed.
Peer-review model: Single blind
Specialty type: Endocrinology and metabolism
Country of origin: China
Peer-review report’s classification
Scientific Quality: Grade A, Grade B, Grade B, Grade B, Grade B, Grade C, Grade C
Novelty: Grade B, Grade B, Grade B, Grade C
Creativity or Innovation: Grade B, Grade B, Grade C, Grade C
Scientific Significance: Grade B, Grade C, Grade C, Grade C
P-Reviewer: Ahmed Salıh Gezh SAS; Dabla PK; Hwu CM; Kanda T; Papazafiropoulou A; Yang XY S-Editor: Li L L-Editor: A P-Editor: Xu ZH
Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan BB, Stein C, Basit A, Chan JCN, Mbanya JC, Pavkov ME, Ramachandaran A, Wild SH, James S, Herman WH, Zhang P, Bommer C, Kuo S, Boyko EJ, Magliano DJ. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045.Diabetes Res Clin Pract. 2022;183:109119.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 3033][Cited by in RCA: 4496][Article Influence: 1498.7][Reference Citation Analysis (36)]
Handzlik MK, Gengatharan JM, Frizzi KE, McGregor GH, Martino C, Rahman G, Gonzalez A, Moreno AM, Green CR, Guernsey LS, Lin T, Tseng P, Ideguchi Y, Fallon RJ, Chaix A, Panda S, Mali P, Wallace M, Knight R, Gantner ML, Calcutt NA, Metallo CM. Insulin-regulated serine and lipid metabolism drive peripheral neuropathy.Nature. 2023;614:118-124.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 46][Cited by in RCA: 93][Article Influence: 46.5][Reference Citation Analysis (0)]
