Abstract
Background: Rheumatoid arthritis (RA) and Spondyloarthritis (SpA) are chronic inflammatory diseases characterized by joint inflammation and systemic involvement. Current diagnostic methods lack sufficient specificity and sensitivity, often leading to delayed or inaccurate diagnoses.
Objective: This study aims to utilize spatial transcriptomics and machine learning to identify differentially expressed genes (DEGs) and potential biomarkers associated with RA and SpA, enhancing our understanding of their molecular mechanisms.
Methods: High-dimensional spatial transcriptomics data and high-resolution tissue images from six synovial biopsy samples (three RA and three SpA) were analyzed. DEGs were identified using statistical criteria, and machine learning models were applied to classify disease status based on gene expression patterns. Functional enrichment analyses were performed to explore the biological significance of the identified DEGs.
Results: A total of 49 DEGs in SpA and 30 DEGs in RA were identified, all of which were upregulated. Key DEGs were further refined using feature selection methods. Machine learning models demonstrated moderate performance in classifying disease status, with the Light Gradient Boosting Machine (LGBM) model achieving the highest accuracy. Functional analyses indicated that the DEGs are predominantly involved in immune-related processes and cellular stress responses.
Conclusion: The study provides preliminary insights into the molecular mechanisms of RA and SpA, identifying potential biomarkers for further investigation. However, the limited sample size and lack of experimental validation necessitate caution in interpreting the results. Future studies with larger cohorts and experimental validation are required to confirm these findings and explore their clinical applicability.
Keywords
Rheumatoid Arthritis, Spondyloarthritis, Machine Learning, Diagnosis, Biomarker