Background: In early rheumatoid arthritis (RA), patients with similar clinical characteristics may display heterogeneous disease evolution and therapeutic responses. Synovial pathotypes reflect distinct biological states of synovial tissue; however, it remains unclear to what extent these states can be identified using accessible biomarkers in peripheral blood. The identification of serum molecular patterns reflecting synovial biology, integrated with clinical and histological data, may facilitate patient stratification and treatment personalization in the early phases of the disease.

Objectives: To classify and characterize early RA patients by synovial pathotypes (MD, LM, PI) using baseline clinical and histological data, and assess their associations with post-biopsy b/tsDMARD initiation and 6-month clinical response. To identify pathotype-specific serum proteomic signatures and characterize their underlying biological processes. To evaluate whether pathotype-associated serum proteomic modules are linked to 6-month treatment response to csDMARDs versus b/tsDMARDs, exploring their potential as biomarkers for therapeutic stratification.

Methods: Histomorphological features of synovial explants obtained through ultrasound-guided biopsies from 25 patients with early RA were assessed, including stromal cellularity, synovial inflammation, and immune cell infiltrates (CD3+ T cells, CD68+ macrophages in lining/sublining, CD20+ B cells, CD138+ plasma cells). Accordingly, patients were stratified into three synovial pathotypes based on prior classifications: myeloid (MD), lymphoid-myeloid (LM), and pauci-immune (PI). Clinical evaluations encompassed disease activity (DAS28-PCR, SDAI, CDAI), inflammatory markers (CRP, ESR), autoantibodies (RF, ACPA titers), time to biologic therapy initiation, and response to 6 months therapy with conventional synthetic DMARDs (csDMARDs) or biological or targeted synthetic DMARDs (b/tsDMARDs) (EULAR response criteria). Serum proteomic profiling was performed in these patients using proximity extension assay technology, analyzing a 92-protein inflammation-related panel. Proteomic data underwent pairwise comparisons and module analysis to define pathotype-specific signatures, with z-score associations to treatment responses. Functional enrichment analyses (using the software String) were performed to characterize the biological processes underlying each signature. Associations between synovial pathotypes, serum molecular signatures, and 6-month clinical response were explored using descriptive and statistical analyses.

Results: After classification of patients according to synovial pathotype, baseline clinical characteristics were assessed. The three groups (MD, LM and PI) were comparable in age and sex. The LM pathotype showed the greatest number of tender and swollen joints, as well as the highest disease activity indices (DAS28-CRP/ESR, SDAI, CDAI). Positivity for RF and ACPAs was mainly observed in both, the LM and MD pathotypes, with higher titers in the LM group. At the histological level, LM and MD pathotypes showed higher stromal cellularity and synovial inflammation compared with the PI pathotype. The LM pathotype exhibited the most pronounced lymphoid infiltrate, with increased T cells, B cells and plasma cells, whereas no significant differences were observed in macrophage infiltration. During follow-up, initiation of b/tsDMARDs differed across synovial pathotypes. LM patients initiated biologics more frequently and earlier post-biopsy compared with MD and PI. csDMARD response at 6 months was superior in MD (highest responders), intermediate in LM, and lowest in PI; conversely, biologic response was best in PI, intermediate in LM, and lowest in MD. Serum proteomics revealed distinct signatures: MD (immune chemotaxis/inflammation: upregulated IL7, CXCL9, CXCL11, CXCL6, OSM, TNFSF14, EN-RAGE, MMP-1); LM (endothelial remodeling/proliferation control: FGF-21, IL-24, TGF-β, CCL19, IL-10, MMP-1); PI (tissue remodeling/chemotaxis: uPA, CXCL10).

Individual proteins did not predict response, but low z-scores of pathotype-specific protein modules correlated with better outcomes (low LM module with csDMARD response; low MD module with b/tsDMARD response).

Conclusions: Synovial pathotypes and associated serum proteomic modules identify molecular heterogeneity in early RA that is not apparent from baseline clinical or histological assessments alone. These pathotypes correlate with distinct clinical trajectories, including time to biologic initiation (shorter in LM) and 6-month responses to csDMARDs (superior in MD) or b/tsDMARDs (superior in PI). Low z-scores of pathotype-specific protein modules (LM for csDMARD response; MD for b/tsDMARDs response) further refine outcome prediction. Combined clinicopathological and molecular profiling provides complementary prognostic information to standard evaluations, supporting further investigation into its utility for therapeutic stratification in early RA.

REFERENCES: NIL.

Acknowledgments: NIL.

Disclosure of Interests: None declared.