Prediction of the intolerance or non-responder to Janus kinase inhibitors in patients with rheumatoid arthritis: a preliminary retrospective study with integrative cluster analysis

Objective To identify the subpopulation of rheumatoid arthritis (RA) non-responders to Janus kinase inhibitors (JAKis) using cluster analysis. Methods This retrospective study enrolled RA patients who had been treated with JAKis (tofacitinib or baricitinib) between July 2013 and September 2019 in six centres. The endpoint was set as inadequate response to JAKis (JAKis-IR), defined as either non-response to JAKis or their intolerance. Non-response to JAKis was defined as achieving neither American College of Rheumatology 20% response nor Disease Activity Score (ΔDAS28-CRP) >1.2 at 12 weeks. Withdrawal time point included earlier than after 12 weeks from baseline. A hierarchical cluster analysis was performed with variables related with clinical and serological parameters at baseline. Results The 132 RA patients enrolled were classified into four groups (Group A-D). Groups consisted of three components defined at baseline, as seropositivity, advanced joint destruction, interstitial lung disease presumably associated with RA (RA-ILD). Group A (n=32): seronegative, presence of advanced joint destruction, absence of RA-ILD. Group B (n=35): seropositive, absence of advanced joint destruction and RA-ILD. Group C (n=20): seropositive, absence of advanced joint destruction, presence of RA-ILD. Group D (n=45): seropositive, presence of advanced joint destruction and RA-ILD. The rate of JAKis-IR in four groups was as follows: A, 34.3%; B, 17.1%; C, 20.0%; and D, 8.9%. The difference in JAKis-IR rate between group A and D was statistically significant. Conclusion A subpopulation of RA patients with a combination of the following three components, seronegativity, advanced joint destruction and absence of RA-ILD, was identified as being prone to JAKis-IR.