Background: Systemic sclerosis (SSc) is a chronic and systemic autoimmune disease marked by the skin and visceral fibrosis. Metabolic alterations have been found in SSc patients; however, serum metabolomic profiling has not been thoroughly conducted. Our study aimed to identify alterations in the metabolic profile in both SSc patients before and during treatment, as well as in mouse models of fibrosis. Furthermore, the associations between metabolites and clinical parameters and disease progression were explored.
Methods: High-performance liquid chromatography quadrupole time-of-flight mass spectrometry (HPLC-Q-TOF-MS)/MS was performed in the serum of 326 human samples and 33 mouse samples. Human samples were collected from 142 healthy controls (HC), 127 newly diagnosed SSc patients without treatment (SSc baseline), and 57 treated SSc patients (SSc treatment). Mouse serum samples were collected from 11 control mice (NaCl), 11 mice with bleomycin (BLM)-induced fibrosis and 11 mice with hypochlorous acid (HOCl)-induced fibrosis. Both univariate analysis and multivariate analysis (orthogonal partial least-squares discriminate analysis (OPLS-DA)) were conducted to unravel differently expressed metabolites. KEGG pathway enrichment analysis was performed to characterize the dysregulated metabolic pathways in SSc. Associations between metabolites and clinical parameters of SSc patients were identified by Pearson's or Spearman's correlation analysis. Machine learning (ML) algorithms were applied to identify the important metabolites that have the potential to predict the progression of skin fibrosis.
Results: The newly diagnosed SSc patients without treatment showed a unique serum metabolic profile compared to HC. Treatment partially corrected the metabolic changes in SSc. Some metabolites (phloretin 2'-O-glucuronide, retinoyl b-glucuronide, all-trans-retinoic acid, and betaine) and metabolic pathways (starch and sucrose metabolism, proline metabolism, androgen and estrogen metabolism, and tryptophan metabolism) were dysregulated in new-onset SSc, but restored upon treatment. Some metabolic changes were associated with treatment response in SSc patients. Metabolic changes observed in SSc patients were mimicked in murine models of SSc, indicating that they may reflect general metabolic changes associated with fibrotic tissue remodeling. Several metabolic changes were associated with SSc clinical parameters. The levels of allysine and all-trans-retinoic acid were negatively correlated, while D-glucuronic acid and hexanoyl carnitine were positively correlated with modified Rodnan skin score (mRSS). In addition, a panel of metabolites including proline betaine, phloretin 2'-O-glucuronide, gamma-linolenic acid and L-cystathionine were associated with the presence of interstitial lung disease (ILD) in SSc. Specific metabolites identified by ML algorithms, such as medicagenic acid 3-O-b-D-glucuronide, 4'-O-methyl-(-)-epicatechin-3'-O-beta-glucuronide, valproic acid glucuronide, have the potential to predict the progression of skin fibrosis.
Conclusions: Serum of SSc patients demonstrates profound metabolic changes. Treatment partially restored the metabolic changes in SSc. Moreover, certain metabolic changes were associated with clinical manifestations such as skin fibrosis and ILD, and could predict the progression of skin fibrosis.
Keywords: Biomarkers; Machine learning algorithm; Metabolic reprogramming; Metabolomics; Systemic sclerosis.
Copyright © 2023 Elsevier Inc. All rights reserved.