Metabolomic alterations associated with Behçet's disease

Abstract

Background: The diagnosis of Behçet's disease (BD) remains challenging due to the lack of diagnostic biomarkers. This study aims to identify potential serum metabolites associated with BD and its disease activity.

Methods: Medical records and serum samples of 24 pretreated BD patients, 12 post-treated BD patients, and age-matched healthy controls (HC) were collected for metabolomics and lipidomics profiling using UPLC-QTOF-MS and UPLC-QTOF-MS^E approaches. Additionally, serum samples from an independent cohort of BD patients, disease controls including rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), Takayasu's arteritis (TA), Crohn's disease (CD) patients, and HC were collected for further validation of two potential biomarkers using UPLC-QTOFMS analysis.

Results: Unsupervised principal component analysis (PCA) showed a clear separation of metabolomics profiles of BD patients from HC. Statistical analysis of the data revealed differential metabolites between BD patients and HC. The serum levels of some phosphatidylcholines (PCs) were found to be significantly lower in BD patients, while the levels of several polyunsaturated fatty acids (PUFAs) were increased markedly in the BD group compared with HC. Furthermore, the serum level of two omega-6 PUFAs, linoleic acid (LA) and arachidonic acid (AA), were dramatically decreased in patients with remission. A validation cohort confirmed that the serum LA and AA levels in BD patients were significantly higher than those in HC and patients with RA, SLE, TA, and CD. In addition, receiver operating characteristic (ROC) analysis indicated good sensitivity and specificity.

Conclusions: The serum metabolomics profiles in BD patients are altered. Serum LA and AA are promising diagnostic biomarkers for BD.

Keywords: Autoinflammatory disease; Behçet’s disease; Biomarker; Lipidomics; Metabolomics.

Fig. 1

Multivariate data analyses of serum metabolomic profiles segregate Behçet’s disease (BD) patients from the healthy cohort (HC; Nor). a Principle components analysis (PCA) unsupervised clustering plots for healthy (green triangle) and BD (purple triangle) are shown. b The volcano plot displays t test results of samples from BD and HC. Ions marked in red show a significant difference in intensity between BD and HC. c KEGG pathway enrichment analysis results of the differential ions identified in the multivariate analysis of serum metabolomic profiles from HC and BD patients. FDR false discovery rate

Fig. 2

Serum lipidomic profile phenotype of treated Behçet’s disease (BD) patients moves towards that of the healthy cohort (HC). a PCA clustering plot of samples from these three groups with differential metabolites identified between pretreatment BD patients and HC. b Heat map of the metabolites with significantly different abundance between BD and HC

Fig. 3

Levels of phosphatidylcholines (PCs) and polyunsaturated fatty acids (PUFAs) are affected by treatments in a different way. Comparison of the level of three PCs (a) and two PUFAs (b) in healthy controls (HC), pretreatment Behçet’s disease (BD) (Pre-) patients, and post-treatment BD (Post-) patients. The significance of metabolite levels was determined using a two-tailed student t test. *p < 0.05, ***p < 0.001. ns not significant

Fig. 4

Serum level of arachidonic acid (AA) and linoleic acid (LA) in the validation cohort. Levels of AA (a) and LA (b) in patients with Behçet’s disease (BD) (n = 25), systemic lupus erythematosus (SLE) (n = 12), rheumatoid arthritis (RA) (n = 12), Takayasu’s arteritis (TA) (n = 15), and Crohn’s disease (CD) (n = 15), and healthy controls (HC) (n = 19). Statistical significance between disease groups and HC was determined using a two-tailed student t test. *p < 0.05, **p < 0.01, ***p < 0.001. n.s. not significant