A total of 47 patients with gout who met the classification criteria approved by The European League Against Rheumatism Executive Committee and from the outpatient ward of the Affiliated Hospital of Nanjing University of Traditional Chinese Medicine (Nanjing, China) from June 2020 to January 2023(17). A total of 49 healthy volunteers without rheumatic diseases, were recruited from the Health Counseling and Physical Examination Center of the Affiliated Hospital of Nanjing University of Traditional Chinese Medicine at the same time. The patients with gout and healthy subjects were randomly assigned to the discovery set (34 healthy individuals and 32 patients with gout) and the validation set (15 healthy individuals and 15 patients with gout), respectively, at a 7:3 ratio based on previously published literature (18,19). The present study was approved by the Ethics Committee of the Affiliated Hospital of Nanjing University of Traditional Chinese Medicine (approval no. 2019NL-129-02; Nanjing). Written informed consent was obtained from the participants, who abided by the principle of privacy protection. The patients with gout and the healthy individuals were aged 18-70 years, were male or female, did not suffer from other rheumatic diseases and had not used long-term drugs or treatments, such as anti-inflammatory drugs, for 3 months. LC-MS/MS technology was used to analyze the peripheral blood of the two groups of patients and potential differential markers related to gout were identified (Fig. 1). Logistic regression and receiver operating characteristic (ROC) analyses were used. Kyoto Encyclopedia of Genes and Genomes (KEGG) and MetaboAnalyst databases (https://www.metaboanalyst.ca/) were used to analyze the pathways associated with the differentially expressed metabolites and identify the metabolic pathways of patients with gout.

A Vanquish ultrahigh-performance liquid chromatograph (Thermo Fisher Scientific, Inc.) was used to separate the target compounds using a Waters ACQUITY UPLC HSS T3 (2.1x100 mm; 1.8 µm) liquid chromatography column. The Orbitrap Exploris 120 mass spectrometer, which can perform primary and secondary mass spectrometry data acquisition, was also used (Xcalibur; version 4.4; Thermo Fisher Scientific, Inc.). LC-MS/MS analyses were performed using an UHPLC system (Vanquish; Thermo Fisher Scientific, Inc.) with a UPLC HSS T3 column (2.1x100 mm; 1.8 µm) coupled to an Orbitrap Exploris 120 mass spectrometer (Orbitrap MS; Thermo Fisher Scientific, Inc.). The mobile phase consisted of 5 mmol/l acetate and 5 mmol/l acetic acid in water (A) and acetonitrile (B). The auto-sampler temperature was 4˚C, and the injection volume was 2 µl. The Orbitrap Exploris 120 mass spectrometer was used for its ability to acquire MS/MS spectra on information-dependent acquisition mode in the control of the acquisition software (Xcalibur; Thermo Fisher Scientific, Inc.). In this mode, the acquisition software continuously evaluates the full scan MS spectrum. The ESI source conditions were set as following: Sheath gas flow rate, 50 Arb; Aux gas flow rate, 15 Arb; capillary temperature, 320˚C; full MS resolution, 60,000; MS/MS resolution, 15,000; collision energy, 10/30/60 in NCE mode; and spray voltage, 3.8 kV (positive) or -3.4 kV (negative).

After the original data were converted into the mzXML format using ProteoWizard (http://www.proteowizard.org/index.html), the peak recognition, peak extraction, peak alignment and integrals were processed using the self-written R package (Kernel XCMS). To reduce the influence of detection system errors on the results and ensure that the results better highlighted the biological significance, a series of data management steps were performed using the original data. This approach included the following steps: i) Deviation value filtering where a single peak was filtered to remove the noise and the deviation values were filtered based on the relative standard deviation and coefficient of variation; ii) missing value filtering where a single peak was filtered and only peak area data with a single set of null values ≤50% or all group null values ≤50% were retained; iii) missing value filling where the missing values in the original data were simulated and the numerical simulation method filled half of the minimum value; and iv) data normalization where a mixed isotope internal standard [succinic acid-2,2,3,3-d4,L-leucine-5,5,5-d3, (RING-2H5)-L-phenylalanine,2-chloro-L-phenylalanine] was used for normalization.

The significance of each metabolite was analyzed using the Mann-Whitney Wilcoxon test and the false discovery rate (FDR) was used for data correction. P<0.05 was considered to indicate a statistically significant difference. Then, the standardized data were imported into SIMCA-P (version 14.1; Sweden Umetrics) for principal component analysis (PCA) to observe the aggregation of samples. Orthogonal partial least squares discriminant analysis (OPLS-DA) was further used to assess the difference between the patients with gout and the healthy control group. The processed data were subjected to MetaboAnalyst (version 5.0; http://www.metaboanalyst.ca/). All data in this article were analyzed by one-way ANOVA using GraphPad Prism (version 9.5.1; Dotmatics), followed by ROC and correlation analysis. Correlation analysis and mapping were performed using SPSS (version 27.0; IBM Corp.) and Origin (version 2022; OriginLab Corporation). Endogenous biomarkers were identified using the Human Metabolomics Database (http://www.hmdb.ca/). KEGG (http://www.genome.jp/) and metabolic analysis were used to identify related metabolic pathways.

Patient age, C-reactive protein, erythrocyte sedimentation rate, uric acid, alanine transaminase, aspartate aminotransferase, creatinine, triglycerides, total cholesterol and low-density lipoprotein (LDL) cholesterol were significantly increased in the patients with gout compared with those in the healthy control group, while high-density lipoprotein cholesterol was decreased. The changes in total protein, albumin and globulin were not statistically significant (Table I). Although subjects with other metabolic diseases were excluded to minimize confounding factors, the aforementioned differences still existed, indicating that these indicators may change during the onset of gout.

Nontargeted metabolomics analysis was performed using an LC-MS/MS system quadrupole-electrostatic field orbital trap Orbitrap mass spectrometer, and 26,854 peaks were detected in positive and negative ionization modes. After data collation and standardization, 991 metabolites were identified. Through the Mann-Whitney Wilcoxon test combined with FDR correlation analysis, 295/991 compounds showed significant differences in metabolic characteristics. Of these compounds, 207 were significantly more abundant in patients with gout compared with healthy subjects, while 88 of these compounds were significantly less abundant (Fig. 2). Unsupervised PCA was used to compare the metabolic profiles between patients with gout and healthy individuals to assess the difference between the two groups of patients, and then OPLS-DA was used to demonstrate that there was a significant overall separation between patients with gout and healthy individuals. The PCA plot indicated that there was a significant metabolic difference between patients with gout and healthy subjects at the molecular level (Fig. 3A). The cumulative R2Y and Q2 of the discovery set were 0.962 and 0.801, respectively (Fig. 3B), which also indicated that there was a significant metabolic difference, consistent with the results shown in the PCA plot. The OPLS model was verified using a permutation test. These results demonstrated that the Y intercept of R2 was 0.803 and the Y intercept of Q2 was 0.391 (Fig. 3C), which indicated that the model significantly differed between patients with gout and healthy subjects, and was reliable. The difference between the group of patients with gout and the healthy group was also obvious and the model was more reliable in the validation set than the discovery set (Fig. 3D-F). These observations were used to further explore metabolites that could potentially be used to identify the development of gout.

According to the parameters of VIP ≥1.0, P<0.05 and adjusted P≤0.05, 186 differentially expressed metabolites were identified (Table SI). A visual cluster analysis of the 186 differentially expressed metabolites was performed using a heatmap and the top 20 differentially expressed metabolites were selected for further analysis. The metabolites which were significantly different between the group of patients with gout compared with the healthy control group were separated according to the heatmap, which provided a visual display of the overall distribution of metabolic differences between the groups (Fig. 4A).

The group of biomarkers for the diagnosis of patients with gout were screened according to the area under the ROC curve (AUC). The top 10 endogenous metabolites with the highest diagnostic rates were UA, L-glutamic acid, benzaldehyde, indolelactic acid (ILA), 2-methylbutyroylcarnitine and 2-hydroxy-3-methylbutyric acid, D-glutamic acid, isobutyrylglycine, eugenol and 3-(3,4,5-trimethoxyphenyl)propanoic acid, which had AUC values of 0.9511, 0.9467, 0.9422, 0.9378, 0.9289, 0.9244, 0.9156, 0.9156, 0.9067 and 0.9022, respectively (Fig. S1). Among these biomarkers, ILA could be used to successfully distinguish between the patients with gout and the healthy group when used as a diagnostic marker. The AUC value in the discovery set reached 0.9378, the sensitivity was 86.67%, the specificity was 86.67% and the 95% CI was 0.8580-1.000 (Fig. 4B).

Spearman correlation analysis was performed between the top 10 differentially abundant metabolites and clinical indicators revealed that triglycerides, creatinine, cholesterol, LDL-cholesterol and UA were significantly correlated with differentially abundant metabolites. Among the metabolites, L-glutamic acid and D-glutamic acid were the most related to clinical indicators. The correlations of UA, ILA, 2-methylbutyroylcarnitine and 2-hydroxy-3-methylbutyric acid were also positive, followed by benzaldehyde, isobutyrylglycine, eugenol and 3-(3,4,5-trimethoxyphenyl)propanoic acid (Fig. 4C). A correlation heatmap was produced to represent the different markers between patients with gout and healthy individuals (Fig. 4D). Each small square represented the correlation coefficient between the metabolites and the red color represented a positive correlation, whereas blue represented a negative correlation. The darker the color, the more related the compound is to gout. The correlations between the 10 endogenous compounds demonstrated that the levels of ILA and the other nine metabolic compounds were closely related and significant.

Pathway enrichment analysis demonstrated that the aforementioned metabolites were involved in 29 metabolic pathways. The identified metabolites were significantly enriched in pathways such as the malate-aspartate shuttle, beta-alanine metabolism and aspartate metabolism (Fig. S2). Pathway bubble diagram analysis demonstrated that six pathways, namely, histidine metabolism, D-glutamine and D-glutamate metabolism, alanine, aspartate and glutamate metabolism, steroid hormone biosynthesis, glycerophospholipid metabolism and arginine biosynthesis, were significantly differentially expressed between the patients with gout and the healthy group. Among these pathways, D-glutamine and D-glutamate metabolism were identified as the key nodes (Fig. 5).