Biomarker combination predicting imminent relapse after discontinuation of biological drugs in patients with rheumatoid arthritis in remission

Abstract

Objectives Compared to conventional disease-modifying antirheumatic drugs (DMARDs), biological DMARDs demonstrate superior efficacy but come with higher costs and increased infection risks. The ability to stop and resume biological DMARD treatment while maintaining remission would significantly alleviate these barriers and anxieties. The objective of this study was to identify biomarkers that can predict an imminent relapse, hopefully enabling the timely resumption of biological DMARDs before relapse occurs. Methods Forty patients with rheumatoid arthritis who had been in remission for more than 12 months were included in the study. The patients discontinued their biological DMARD treatment and were monitored monthly for the next 24 months. Out of the 40 patients, 14 (35%) remained in remission at the end of the 24-month period, while 26 (65%) experienced relapses at different time points. Among the relapse cases, 13 patients experienced early relapse within 6 months, and another 13 patients had late relapse between 6 months and 24 months. Seventy-three cytokines in the sera collected longitudinally from the 13 patients with late relapse were measured by multiplex immunoassay. Using cytokines at two time points, immediately after withdrawal and just before relapse, volcano plot and area under the receiver operating characteristic curves (AUC) were drawn to select cytokines that distinguished imminent relapse. Univariate and multivariate logistic regression analyses were used for the imminent relapse prediction model. Results IL-6, IL-29, MMP-3, and thymic stromal lymphopoietin (TSLP) were selected as potential biomarkers for imminent relapse prediction. All four cytokines were upregulated at imminent relapse time point. Univariate and multivariate logistic regression showed that a combination model with IL-6, MMP-3, and TSLP yielded an AUC of 0.828 as top predictors of imminent relapse. Conclusions This methodology allows for the prediction of imminent relapse while patients are in remission, potentially enabling the implementation of on- and off-treatments while maintaining remission. It also helps alleviate patient anxiety regarding the high cost and infection risks associated with biological DMARDs, which are the main obstacles to benefiting from their superb efficacy.