Background/Purpose: Early and targeted treatment strategies have led to increasing numbers of patients with RA achieving sustained clinical remission. If and when to taper Disease Modifying Anti-Rheumatic Drug (DMARD) therapy is an emerging question for both clinicians and patients. Unfortunately, data on clinical, biochemical or imaging modalities that can reliably predict relapse following therapy withdrawal are limited to date. The objectives of our study were two-fold; first to generate unsupervised serum proteomic profiles using baseline samples from patients with RA in remission. We also aimed to identify serum biomarkers that predict relapse following DMARD withdrawal.

Methods: The RETRO (Reducing therapy in Rheumatoid Arthritis Patients in ongoing Remission) study is a multicentre, randomized, prospective trial enrolling RA patients in sustained clinical remission (DAS28 2.6) to undergo DMARD withdrawal. 130 baseline serum samples from the RETRO study were analyzed using SOMAscan (SOMAlogic Inc.), allowing the generation of comparative quantitative levels of > 1300 proteins. Unsupervised clustering analysis was performed to identify subgroups of patients. Unique clusters were analyzed with Ingenuity Pathway Analysis (QIAGEN). Differentially expressed proteins between patients who relapsed, and those who remained in remission were combined to synthesize a predictive biomarker scoring algorithm. A multivariate logistic regression model that included seropositive status, current treatment, both disease and remission duration, and the biomarker score was used to analyze predictors of relapse.

Results: Unsupervised clustering analysis distinguished 4 unique subpopulations of patients. Compared to the remaining population, Cluster 3 (n = 14) had shorter disease duration (p = 0.012), higher seropositivity rates (p = 0.02) and no use of biologic DMARDs (p = 0.001). Cluster 3 had 145 proteins with significantly (p <0.05) reduced expression (log2 difference = 0.50 – 1.17) compared to the remaining population. Pathway analysis of these proteins suggested downregulation of Acute Phase Response signaling. Additionally, we created a predictive biomarker withdrawal score (SOMA score; range = 1.5 to 26.0) comprised of 8 unique biomarkers using our data. Patients failing DMARD withdrawal had higher mean SOMA scores compared to those who sustained remission (13.4 vs 8.0, p < 0.001). 83.7% of patients who failed withdrawal had scores greater than a cut off of 9.64, while only 25.6% of those remaining in remission had scores above this cut off. Multivariate regression analysis identified SOMA score as the only independent predictor for relapse after DMARD withdrawal (OR = 15.5, 4.5 – 53.4, p < 0.001).

Conclusion: Unsupervised proteomic analysis of RA patients in sustained clinical remission generates distinct clusters of subpopulations. We discovered a unique protein signature in a group of patients who, despite being seropositive, achieved sustained clinical remission without requiring biologic therapy. We constructed a serum protein biomarker score that successfully predicts failure of DMARD withdrawal. Validation of our predictive score will be undertaken to assess its use as a clinical decision-making tool.

« Back to 2017 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/serum-proteomic-signatures-predict-relapse-in-rheumatoid-arthritis-patients-undergoing-dmard-withdrawal/