# Integrative Bioinformatics and Machine Learning Identify Novel Diagnostic Biomarkers and Molecular Mechanisms in Sjögren’s Syndrome

## TL;DR

This study uses bioinformatics and machine learning to find new biomarkers and understand the molecular basis of Sjögren’s Syndrome, a difficult-to-diagnose autoimmune disease.

## Contribution

The study introduces 12 novel diagnostic biomarkers for Sjögren’s Syndrome and reveals their molecular mechanisms and immune-related functions.

## Key findings

- Twelve hub genes (e.g., EPSTI1, IFIH1) were identified with strong diagnostic performance across multiple datasets.
- Immune infiltration analysis showed immune dysregulation in SS patients, including reduced CD8+ T cells and Tregs.
- Drug repurposing suggested FDA-approved drugs like nisoldipine and exemestane as potential treatments.

## Abstract

Sjögren’s syndrome (SS) is a chronic autoimmune disorder characterized by significant diagnostic challenges due to nonspecific symptoms and a lack of reliable biomarkers, often resulting in delayed diagnosis and suboptimal patient management.

This study is aimed at identifying novel diagnostic biomarkers and elucidating the molecular mechanisms underlying SS pathogenesis through integrative bioinformatics and machine learning approaches.

We analyzed three peripheral blood transcriptomic datasets (GSE51092, GSE66795, and GSE84844) comprising a total of 351 SS patients and 91 healthy controls. Differential expression analysis, weighted gene coexpression network analysis (WGCNA), and 12 machine learning algorithms were employed to identify robust diagnostic biomarkers. Immune cell infiltration was assessed using CIBERSORT, and single‐cell RNA sequencing data (GSE157278) were analyzed to validate cell‐type‐specific expression patterns. Drug repurposing analysis was conducted using the L1000FWD platform.

We identified 12 hub genes (EPSTI1, IFIH1, CXCL10, TNFSF10, GBP5, PARP9, IFI44, LAP3, IFIT2, IFI44L, PARP12, and OAS1) with exceptional diagnostic performance (AUC = 0.994 in training, 0.838 in internal validation, and 0.825 in external validation). These biomarkers showed significant correlations with clinical indicators including ANA, Ro/SSA, and La/SSB (p < 0.05). Immune‐infiltration analysis revealed pronounced immune dysregulation in SS patients, characterized by an imbalance between naive and memory B cells and reduced CD8+ T cells and regulatory T cells (Tregs). Single‐cell transcriptomics confirmed predominant expression in monocytes and dendritic cells, with additional significant expression in B cells and CD4+ T cells. Virtual knockdown analysis implicated these genes in antigen presentation, interferon signaling, and leukocyte trafficking. Drug repurposing identified FDA‐approved candidates such as nisoldipine and exemestane as potential therapeutics.

Our integrative approach identifies 12 robust diagnostic biomarkers for SS, offering new insights into disease mechanisms and highlighting potential therapeutic targets for this challenging autoimmune disorder.