Three candidate genes may help distinguish rheumatoid arthritis from osteoarthritis, according to a study published in Frontiers in Medicine, although the findings are based on transcriptomic analyses and early experimental validation.
Distinguishing rheumatoid arthritis from osteoarthritis can be challenging in early disease because of overlapping clinical features, and misclassification may affect treatment decisions.
Researchers analyzed publicly available Gene Expression Omnibus synovial-tissue data sets from patients with rheumatoid arthritis, osteoarthritis, and healthy controls, applying machine learning methods to identify potential diagnostic markers. The final gene set reflected the overlap between both machine learning approaches, strengthening confidence that the findings were not specific to a single algorithm.
Across both methods, three genes—EPYC, MAGED1, and LAP3—were consistently identified. In receiver operating characteristic analyses, each demonstrated good diagnostic performance, with area under the curve values above 0.85 in both validation data sets. In model testing on the study data sets, the support vector machine approach achieved about 93% accuracy, with an error rate near 7%.
In a tumor necrosis factor alpha–stimulated fibroblast-like synoviocyte model derived from patients with rheumatoid arthritis, EPYC and LAP3 expression increased, while MAGED1 expression decreased, consistent with the computational findings.
Functional analyses suggested that differentially expressed genes in rheumatoid arthritis were enriched in immune-related pathways, including T-cell activation and cytokine signaling.
Secondary analyses showed that EPYC and LAP3 expression correlated with multiple immune cell populations in rheumatoid arthritis, while in osteoarthritis this pattern was observed mainly for LAP3. The researchers noted that these findings may reflect general immune activity within joint tissue rather than disease-specific mechanisms.
The study had several limitations. Sample sizes were small, and the analysis relied on cross-sectional transcriptomic data, which does not establish causation. Experimental validation was limited to messenger RNA expression in a single cell model, without protein-level or clinical validation. Drug sensitivity analyses were based on cancer cell data and may not reflect responses in rheumatoid arthritis.
Overall, the findings identify EPYC, MAGED1, and LAP3 as candidate biomarkers for distinguishing rheumatoid arthritis from osteoarthritis, but further validation is required before clinical use. “Further validation in independent clinical cohorts and mechanistic studies are needed before clinical application,” wrote Zhibin Zhang, of Chengde Medical University, and colleagues.
The researchers reported no conflicts of interest. The study received institutional and grant funding support.
Source: Frontiers in Medicine
