Background/Purpose: Although not part of the formal ACR criteria for RA, PsA, or OA, synovial pathology can be a helpful tool in clinical practice. Histopathologic evaluation in RA has led to the development of a synovitis score and a joint pathology algorithm. Three synovial pathotypes—lympho-myeloid, diffuse-myeloid, and pauci-immune—have been described, but are based on a limited number of markers and currently provide limited clinical information. Recent advances allow for minimally invasive synovial biopsies and multiplex microscopy to provide mechanistic insights into tissue pathology. This is especially relevant as targeted therapies for OA are still lacking. Artificial intelligence (AI) enables detailed image analysis, preserving spatial context, allowing for single-cell quantification and identification of cellular interactions that may inform therapy and predict response. Our objective was to apply a novel high-plex IF platform coupled with a semi-automated image analysis approach using machine learning algorithms to characterize synovial inflammatory infiltrates in patients with PsA, RA, and OA

Methods: FFPE synovial tissue biopsies from nine patients from patients with RA (n=5), OA (n=2), and PsA (n=2) were stained with a 17-marker IF panel targeting nuclei, CD31, CD68, CD163, CD20, CD4, CD8a, CD45RO, PD-L1, CD3e, Ki-67, PD-1, FOXP3, CD45, Pan-CK, vimentin, and SMA. Some RA and PsA samples had prior classification (lymphoid or lympho-myeloid) based on conventional IF/IHC staining. Whole sections were scanned at 20x using the Orion platform.A semi-automated AI pipeline originally developed for pancreatic tissue was adapted to synovial tissue. Regions of interest (ROIs, lining and sublining) were selected. Cell classification was based on fluorescence intensity and morphology. Spatial context and cellular neighborhoods were also analyzed using unsupervised clustering algorithms

Results: Nine synovial biopsies from five male and five female (mean age 59.1 years, SD10.1) with a total of 105,364 cells across 130 ROIs were analyzed over 23.7 mm² of tissue. Cellular density was higher in the lining (mean 10,461 cells/mm²) than in the sublining (mean 5,017 cells/mm²). As expected, OA showed fewer immune cells, fibroblasts, and vessels than RA and PsA. Macrophage and fibroblast predominance was observed across all samples, especially in immune-mediated arthritis. In RA, T cell infiltration was more abundant in the sublining, and proportions of T cell subsets were similar across knees and wrists. Distinct cellular clusters were identified in the unsupervised analysis, differentiating between diseases

Conclusion: These preliminary findings support the suitability of this new high-plex microscopy technology and AI-driven image analysis in assessing synovial inflammation. Despite limited tissue, we achieved robust, high-resolution profiling of thousands of cells. We envision that this type of data may help create algorithms in joint pathology to further aid in patient management. Novel platforms currently restricted to research can be incorporated into the clinical arena to serve as a diagnostic tool as well as a personalized treatment tool based on a patient’s specific histological characteristics

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/spatial-and-quantitative-semiautomated-image-analysis-of-synovial-biopsies-studied-using-a-novel-high-plex-immunofluorescence-platform/