MultiOmics landscape of pathotypes in rheumatoid arthritis – enroute to digital health using AI clinical diagnostics by nanoproteomics and AI based imaging diagnostics

Introduction & Aim: Rheumatoid arthritis (RA) is a chronic autoimmune disease that primarily affects synovial joints. Despite significant advances in our understanding of RA, the underlying molecular mechanisms within synovial tissue of RA at the protein level remain largely unexplored. Additionally, the factors contributing to synovial histological heterogeneity, which may impact treatment outcomes, also remain largely unknown at the protein level. Mapping the proteomic landscape of RA-affected synovial tissue, extracellular matrix (ECM), and identifying determinants of histological pathotypes, could provide valuable insights into the molecular mechanisms of RA and identify new therapeutic targets. This knowledge could also pave the way for personalized medicine options, ultimately leading to improved treatment outcomes for RA patients.

Methods: Synovial tissue by ultrasound-guided nano-biopsy and matched plasma samples were collected from patients with early untreated RA and longstanding RA as well as healthy controls. All samples were characterized using optimized clinical discovery proteomics profiling (EvosepOne-timsTOF PRO2), autoantibody profiling and clinical immunological profiling. Next, data integration - of the synovial tissue proteome, plasma proteome, autoantigen profiles (Sengenics), clinical scores including Krenn synovitis score, immune cell scores, synovial immune infiltration and number of circulating immune cells - was conducted to identify the molecular and cellular determinants of synovial histological pathotypes.

Results: In the SYPRORA study we performed a multimodal integration of data types (mixOMICs) including deep plasma protein (>400 proteins) and synovial tissue nano-proteome (>3400 proteins) were identified and quantified after data preprocessing. Data integration identified distinct molecular and cellular signatures underlying synovial pathotypes that suggests the difference between lymphoid, myeloid, and fibroid pathotypes is a continuous spectrum that is linearly modellable. The nanoproteomic findings correlated with ARTHUR - the world's first fully automated clinical robot that performs ultrasound scans on RA patients and assesses the disease with the help of artificial intelligence.

Conclusion: Proteomic characterization of nano-proteomics synovial tissue biopsies elucidates RA pathogenesis at different stages of disease and identifies molecular and cellular determinants of synovial heterogeneity and impact to ECM.