Mendelian Randomization and Machine Learning Reveal Immune Cell and Gene Drivers in Systemic Lupus Erythematosus

Background

Systemic lupus erythematosus (SLE) is a complex autoimmune disease with unclear pathogenesis. Recent studies suggest that immune cell phenotypes may play a causal role. This study aimed to uncover causal immune cell types, key genes, and potential biomarkers using Mendelian randomization (MR) and machine learning.

Methods

A two-sample MR analysis was performed on 731 immune cell traits to assess their causal relationship with SLE risk. Gene Expression Omnibus datasets were used to identify differentially expressed genes (DEGs), followed by immune infiltration analysis and machine learning-based gene selection. Key genes were validated using independent datasets and expression quantitative trait loci-MR analysis.

Results

Significant causal links with SLE were observed for 10 immune cell subtypes (p < 0.01). A total of 17 DEGs, including FCGR2A, TMEM181, and RASA3, were identified as being associated with immune infiltration. Single-sample gene set enrichment analysis revealed altered immune cell compositions in SLE. Five key genes (FCGR2A, TMEM181, RASA3, BCAR3, and MCTP2) with strong diagnostic potential (area under the curve = 0.948) were identified using a support vector machine model. Their causal relevance was confirmed by MR.

Conclusions

This integrative approach revealed 10 immune cell types and five genes with causal roles in SLE, offering novel insights into disease mechanisms and potential targets for precision medicine.