Rheumatoid arthritis (RA) is a chronic systemic inflammatory disease characterized by significant clinical heterogeneity. This complexity poses persistent challenges for early diagnosis, accurate prognosis, and personalized therapy. Although modern management emphasizes rigorous disease control, conventional biomarkers like rheumatoid factor and anti-citrullinated protein autoantibodies (ACPA) often lack sufficient sensitivity and specificity. Specifically, they frequently fail to predict disease transition in at-risk individuals or to stratify clinical severity accurately. Given the molecular complexity of RA, there is an urgent need for novel biomarkers that reflect dynamic metabolic shifts across different disease stages. Therefore, this study sought to explore plasma metabolomic markers to better understand RA onset, assess disease activity, and evaluate the efficacy of therapeutic interventions.

The study utilized a longitudinal design involving 209 established RA patients who were disease-modifying antirheumatic drugs (DMARDs)-free for at least six months prior to enrollment, with 197 of these participants followed for 3–6 months to evaluate treatment response. Additionally, the research incorporated a cohort of 56 individuals at risk for RA, 34 of whom completed a 5–7-year follow-up to monitor disease progression. Plasma metabolomic profiling was conducted to identify signatures associated with RA development and severity. Predictive modeling was then performed using ridge regression, integrating metabolite and clinical features to forecast the response to two primary combination therapies: methotrexate (MTX) plus leflunomide and MTX plus hydroxychloroquine.

Analysis indicated that metabolites associated with methylation and redox imbalance, specifically S-adenosylmethionine, sarcosine, nicotinamide adenine dinucleotide, and glutathione, showed significant associations with RA development and clinical severity. These metabolomic alterations are associated with the observed disease heterogeneity across age, sex, and ACPA status. Furthermore, the ridge regression models demonstrated favorable predictive performance for therapeutic outcomes, yielding an average receiver operating characteristic (ROC) score of 0.83 for the MTX plus leflunomide regimen and 0.92 for the MTX plus hydroxychloroquine regimen.

In conclusion, these findings highlight potential plasma metabolomic alterations associated with the progression of RA, from the initial at-risk phase to established disease and subsequent treatment response. The identification of markers linked to methylation and redox pathways provides preliminary insights into the molecular factors underlying RA heterogeneity. These results offer potential avenues for supporting early diagnosis and clinical monitoring. Ultimately, while further validation is required, these metabolomic markers may contribute to the development of personalized treatment assessments and help refine the clinical management of patients with RA.