In search of biomarkers for prediction of drug treatment responses in rheumatoid arthritis: Lessons learned and future perspectives

Abstract

Prompt initiation of effective drug treatment is crucial for controlling inflammation and preventing disease progression in rheumatoid arthritis, the most prevalent systemic rheumatic disease. The growing range of drug therapies over the past three decades and the fact that only a minority of patients achieve sustained long-term remission with any given therapy, make imperative the need for biomarkers predicting responses to specific drugs. Moreover, promising therapeutic approaches under development, namely cellular therapies, could be promptly applicable at earlier disease stages in about 10-15 % of RA patients who will be refractory to all approved drugs. In this scoping review of original articles published until 25th of July 2025, we present a critical overview of the literature pertaining to the prognostic value of blood immunophenotyping, circulating proteins and blood proteomics, transcriptomics, metabolomics and lipidomics, as well as of endogenous cortisol production and synovial histopathology. We also discuss the emerging use of artificial intelligence-based approaches for developing response prediction models that integrate clinical features with molecular profiling. We conclude that current knowledge does not allow to discern future responders to methotrexate and/or to different biologic agents from non-responders because established biomarkers to identify those patients who will benefit the most from each therapeutic option are lacking. We also emphasize the lack of standardized research approaches to discover biomarkers predicting drug treatment responses and try to identify the relevant pitfalls and describe the lessons learned over the years. Finally, we propose a roadmap and the application of advanced analytical and machine learning techniques for future research in this area.