THU0052 A circulating protein signature correlates with synovial pathotypes in rheumatoid arthritis patients. (12th June 2018)

• Record Type: Journal Article
• Authors: Ruiz-Romero, C.; Picchi, F.; Fernández-Puente, P.; González-Rodríguez, L.; Hands, R.; Calamia, V.; Camacho-Encina, M.; Bessant, C.; Pitzalis, C.; Blanco, F.J.
• Is Part Of: Annals of the rheumatic diseases. Volume 77(2018)Supplement 2
• Page Start: 251
• Page End: 251
• Publication Date: 2018-06-12
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2018-eular.3447
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background: Rheumatoid arthritis (RA) is a disease characterised by high clinical variability and an underlying cellular and molecular heterogeneity. Efforts to find tools for the classification of the different disease phenotypes and patient stratification are essential to develop tailored therapies and improve its management. According to this, specific pathological phenotypes of synovial tissue (pathotypes) have emerged as associated with diverse clinical evolution and response to therapy. 1 Objectives: To identify signatures of circulating proteins associated with specific synovial pathotypes in RA patients. Methods: A proteomic analysis was carried out on samples from patients enrolled in the Pathobiology of Early Arthritis Cohort. Ultrasound-guided synovial biopsies from these patients allowed their classification into three groups: lymphoid (L), myeloid (M) or fibroid (F), according to the pathotype. The study was performed using 54 serum samples at baseline. Sera were analysed by nanoliquid chromatography coupled to mass spectrometry using a SWATH strategy on a tripleTOF (Sciex). The proteomic data were processed using ProteinPilot and PeakView. A two-stage support vector machine (TSSVM) with RBF kernel and 10 cross-fold validation for each meta-model was applied using the Classyfire, e1071 and caret R packages. Results: The proteomic analysis led to the identification and quantification of 229 proteins in all samples. A screening was performed on a group of 30 samples (Train set: 10 L, 10 M and 10 F). Data were pre-processed by PCA for dimension reduction. Then, application of machine learning tools led to the identification of a panel of 11 proteins whose different abundance is associated with a specific synovial phenotype (either L, M or F) in RA patients. As shown in the table 1, a very high accuracy and Kappa coefficient were achieved with this classification tool. The results were confirmed on an independent validation set of 24 samples (12 L, 8 M and 4 F) with also good performance. This protein signature allowed the correct classification of the samples into the three pathotypes with very high sensitivity and specificity (see table 1). Conclusions: A signature of 11 circulating proteins has been identified as associated with synovial pathotypes in RA patients. The putative correlation of this signature with the clinical evolution and/or response to therapy of the patients remains to be elucidated. Reference: [1] New learnings on the pathophysiology of RA from synovial biopsies. Pitzalis C, Kelly S, Humby F. Curr Opin Rheumatol. 2013May;25(3):334–44. Disclosure of Interest: None declared