Development and validation of a serum multi-omics based diagnostic model for systemic lupus erythematosus

Abstract Objective. The study aimed to identify potential biomarkers (metabonomics and proteomics) of systemic lupus erythematosus (SLE), and develop serum multi-omics based diagnostic models for SLE. Methods. Serum samples were collected from 111 SLE patients and 38 healthy controls (HCs). Gas chromatography-mass spectrometry (GC-MS) and Tandem Mass Tag (TMT) quantitative detection technology were respectively used to evaluate the subjects' serum metabolic spectrum and protein. Least absolute shrinkage and selection operator (LASSO) regression and stepwise logistic regression were used to select clinical characteristics, metabonomics and proteomics parameters associated with SLE and the severity of SLE. Three diagnosis models were developed and validated to diagnosis SLE and the severity of SLE. The receiver operating characteristic curve (ROC), sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were utilized to evaluated three diagnostic models. Three nomogram were plotted to visualize the three models. Results. 12 variables of all variables (2 demographic characteristics, 9 laboratory indexes, 8 proteomic indexes and 339 metabolomic indexes) were selected, which can be used for the diagnosis of patients with different severity of SLE. Area under ROC (AUC) of three models were 0.932, 0.840 and 0.867, respectively. Meanwhile, the sensitivity and specificity of three models were high (Mode l: 0.956 and0.857, Model 2: 0.700 and 0.900, Model 3: 1.000 and 0.667). Conclusion. The changes of serum metabolites and proteins in patients with SLE were studied by multi-omics analysis, and three multi-omics based diagnosis models and nomograms for were developed and validated to diagnosis SLE and the severity of SLE.