Location of joint involvement differentiates rheumatoid arthritis into different clinical subsets

This is the authors' abstract. We don't add key points for this paper.

Objectives To aid research on etiology and treatment of the heterogeneous rheumatoid arthritis (RA) population, we aimed to identify phenotypically distinct RA subsets using baseline clinical data. Method We collected hematology, serology, joint location, age and sex of RA-patients from the Leiden Rheumatology clinic(n=1,387). We used deep learning and clustering to identify phenotypically distinct RA subsets. To ensure robustness, we tested a) cluster stability, b) physician impact, c) association with remission and methotrexate failure, d) replication in clinical trial data (n=307) and independent secondary care (9 clinics, n=515). Results We identified four subsets: Cluster-1) arthritis in feet, Cluster-2) seropositive oligo-articular disease, Cluster-3) seronegative hand arthritis, and Cluster-4) polyarthritis. We found high cluster stability, no physician influence, significant difference in methotrexate failure( P <0.001) and occurrence of remission (P =0.007). The MTX-failure rates were recurrent in both replication sets(both P <0.001). The hand-Cluster-3 showed best outcomes, especially compared to Cluster-4 ( P <0.001) and Cluster-1 ( P =0.003). The MTX-failure difference was largest in the ACPA-positive stratum (Cluster-3 versus Cluster-1 (HR:0.3 (0.15-0.60) P <0.001), Cluster-3 versus Cluster-4 HR:0.33 (0.15-0.72) P =0.005). We observed this for Cluster-4 in all sets, and for Cluster-1 in two out of three. This was independent of baseline disease activity and symptom duration. The clusters significantly improved the MTX failure model on top of traditional risk factors. Conclusions We discovered and replicated four phenotypic subgroups of RA at baseline characterized by hand and foot involvement that associate with treatment response. Such knowledge on disease subgroups could enhance studies to the treatment and the mechanisms underlying RA. KEY MESSAGES What is already known about the subject? - Rheumatoid arthritis is a heterogeneous disease and clinicians have not completely identified the disease differentiating patterns in clinical practice. - Data-driven unsupervised techniques are able to identify hidden structures in big data. What does this study add? - We identified four RA clusters at baseline: feet involvement, oligo-articular disease, hand involvement and polyarthritis (both feet and hand involvement). The hand cluster shows a good treatment response, especially when compared to the polyarthritis and feet group. - The difference in treatment success between hands vs foot clusters was largest in the ACPA-positive stratum. How might this impact on clinical practice or future developments? . - This research supports future endeavors in identifying etiological mechanisms and tailored treatment solutions. - Feet involvement at disease presentation can be considered a risk factor for less successful treatment. - Depending on the location of inflamed joints at first presentation, RA-patients might need different treatment.