Metabolic profiling of patients with different idiopathic inflammatory myopathy subtypes reveals potential biomarkers in plasma

Abstract

Idiopathic inflammatory myopathy (IIM) are heterogeneous autoimmune diseases that primarily affect the proximal muscles. IIM subtypes include dermatomyositis (DM), polymyositis (PM), and anti-synthetase syndrome (ASS). Metabolic disturbances may cause irreversible structural damage to muscle fibers in patients with IIM. However, the metabolite profile of patients with different IIM subtypes remains elusive. To investigate metabolic alterations and identify patients with different IIM subtypes, we comprehensively profiled plasma metabolomics of 46 DM, 13 PM, 12 ASS patients, and 30 healthy controls (HCs) using UHPLC-Q Exactive HF mass spectrometer. Multiple statistical analyses and random forest were used to discover differential metabolites and potential biomarkers. We found that tryptophan metabolism, phenylalanine and tyrosine metabolism, fatty acid biosynthesis, beta-oxidation of very long chain fatty acids, alpha-linolenic acid and linoleic acid metabolism, steroidogenesis, bile acid biosynthesis, purine metabolism, and caffeine metabolism are all enriched in the DM, PM, and ASS groups. We also found that different subtypes of IIM have their unique metabolic pathways. We constructed three models (five metabolites) to identify DM, PM, ASS from HC in the discovery and validation sets. Five to seven metabolites can distinguish DM from PM, DM from ASS, and PM from ASS. A panel of seven metabolites can identify anti-melanoma differentiation-associated gene 5 positive (MDA5 +) DM with high accuracy in the discovery and validation sets. Our results provide potential biomarkers for diagnosing different subtypes of IIM and a better understanding of the underlying mechanisms of IIM.

Keywords: Anti-MDA5 positive dermatomyositis; Biomarker; Idiopathic inflammatory myopathy; Metabolomics.

Fig. 1

Significant differences in plasma metabolites between patients with dermatomyositis, polymyositis, anti-synthetase syndrome, and healthy controls. a. Idiopathic inflammatory myopathy patient cohorts. b. Principal component analysis (PCA) score plots from DM, PM, ASS, and HC groups. c. PCA score plots from DM and HC groups. d. Partial least square discriminant analysis (PLS-DA) score plots from DM and HC groups. e. PLS-DA models were evaluated by 200 permutation tests. f. Volcano plots from DM and HC groups. g. Summary of differential features and metabolites between patients with different IIM subtypes and HCs. POS, positive; NEG, negative

Fig. 2

Five metabolites selected by random forest can identify dermatomyositis, polymyositis, and anti-synthetase syndrome from healthy controls. a. Enrichment analysis of differential metabolites with MSI level 1/2 between DM and HC groups. b. Summary of models of identifying DM, PM, and ASS from HC using five metabolites in plasma in the discovery and validation cohorts. The contribution rank of each metabolite in every model is listed as number 1, 2, …, and 5. The receiver operating characteristic (ROC) curve based on five metabolites can accurately identify DM (c), PM (d), and ASS (e) from HC in the discovery and validation sets. f. Contribution of five metabolites to the identification model of DM. g. The concentration trends of individual metabolites in the DM and HC groups in the discovery and validation sets. h. Detailed MS/MS spectra of five potential metabolite biomarkers for identification of DM. The measured MS/MS spectral fragment profile (top, black) matching the commercial standard/theoretical fragment (bottom, red); HC-D, DM-D, HC and DM groups in the discovery set, respectively; HC-V, DM-V, HC and DM groups in the validation set, respectively

Fig. 3

Five to seven metabolites selected by random forest can distinguish different idiopathic inflammatory myopathy subtypes. a. Enrichment analysis of differential metabolites with MSI level 1/2 between DM and PM groups. b. Summary of models of differentiating DM, PM, and ASS using five to seven metabolites in plasma. The contribution rank of each metabolite in every model is listed as number 1, 2, …, and 7. ROC curves based on five to seven metabolites can accurately distinguish DM from PM (c), DM from ASS (d), and PM from ASS (e). f. Contribution of five metabolites to the identification model of DM from PM. g. The concentration trends of individual metabolites in the DM and PM groups. h. Detailed MS/MS spectra of 3-hydroxydecanoic acid, Palmitoylcarnitine, and Nicotinamide

Fig. 4

Seven metabolites selected by random forest can predict MDA5 + dermatomyositis. a. Enrichment analysis of differential metabolites with MSI level 1/2 between MDA5 + and MDA5-DM. b. ROC curve based on seven metabolites can accurately predict MDA5 + DM. c. Contribution of seven metabolites to the prediction model of MDA5 + DM. d. The concentration trends of individual metabolites in the MDA5 + and MDA5-DM groups in the discovery and validation sets. MDA5 + _D, MDA5-_D, MDA5 + and MDA5-DM groups in the discovery set, respectively; MDA5 + _V, MDA5-_V, MDA5 + and MDA5-DM groups in the validation set, respectively