Background and Objectives A main challenge in disease-management of rheumatoid arthritis (RA) is to establish objective criteria relevant for diagnosis and therapeutic stratification of patients. Clinical signs and symptoms, radiographic changes and routine laboratory tests have indispensable roles in diagnosis of RA. Nevertheless, a high degree of heterogeneity between RA patients and an increasing numbers of treatment opportunities, demands identification of personalised profiles that might indicate on and/or predict response to particular treatment.

In this study we applied global gene-expression profiling of cells from various body locations, including synovial tissue as a principal place of destruction and inflammation, and blood and bone marrow monocytes, as a cell type that acts as biosensor for revealing on-going alterations in the body. Based on transcriptome data candidate markers were selected and validated at the protein level.

Materials and Methods Affymetrix arrays were used for gene-expression profiles of synovial tissues, blood and bone marrow monocytes from RA and osteoarthritis (OA) patients. ELISA and multiplex immunoassays were used for validation of 28 markers in synovial fluid (SF) and matched serum samples from RA and OA patients.

Results Transcriptome analyses of synovial tissues from RA and OA indicated on infiltration and activation of various cell types in synovial tissue including monocytes/macrophages, T-, B-, NK-cells, fibroblast. Monocyte transcriptomes from blood and bone marrow showed a minor overlap, indicating that pathophysiological changes in RA were disease- and tissue-specific.

Candidate markers from these 3 transcripomtes were selected for measurement at the protein level in SF and serum. Out of 28 markers, 23 reached statistical significance when measured in SF, while 16 were significant when measured in serum. ROC analyses of individual markers from SF and serum enabled ranking for the best candidates, whose combinations allowed almost perfect classification of RA.

Conclusions RA transcriptome was the most robust in synovial tissue. Nevertheless, systemic nature of RA was also demonstrated by transcriptomes from blood and bone marrow monocytes. All transcriptomes demonstrated a great potential in identifying biomarkers whose concentrations were strongly diluted and neutralised in serum. Interestingly, none of the measured markers in serum perform well enough to stratify all RA patients. However, combination of markers as a disease-pattern allowed correct stratification of RA patients and demonstrated their heterogeneity. Further prediction analyses should indicate on a minimal number of top markers relevant for disease and therapeutic stratification, and this would be an indispensable step for establishment of personalised medicine in RA.