Abstract
Systemic lupus erythematosus (SLE) patients are at greater risk of developing osteoporosis (OP) than the general population. This study aimed to identify crosstalk genes between SLE and OP and to validate their diagnostic accuracy as biomarkers. Data analysis based on Gene Expression Omnibus (GEO) datasets was conducted. We utilized Weighted Gene Co-Expression Network Analysis (WGCNA) and differential expression analysis to identify crosstalk genes (CGs). Machine learning algorithms and consensus clustering were applied to screen shared diagnostic biomarkers and construct two predictive models featuring key genes. We also investigated potential subgroups, immune infiltration across different subtypes, and validated hub mRNAs using quantitative real-time PCR (qPCR). Molecular docking was performed to simulate the interaction of a small molecule compound with its target. We identified 19 CGs and developed two predictive models: the IL1R2-GADD45B and CHI3L1-IL1R2-SPTLC2 diagnostic score thresholds. The CHI3L1-IL1R2-SPTLC2 model showed improved predictive accuracy for lupus-associated osteoporosis. The C2 subtype was found to potentially regulate bone metabolism in SLE patients. Immune infiltration analysis indicated a strong association between CGs and multiple immunocytes, with IL1R2 being a common element in both models. Molecular docking suggests that Anakinra's therapeutic effect may involve IL1R2. Our study introduces novel diagnostic biomarkers and predictive models for lupus-associated osteoporosis, with a particular focus on IL1R2 as an innovative biomarker and therapeutic target. These are anticipated to aid early screening and risk assessment in SLE patients, pending large-scale clinical validation.