Identification of broadly discriminatory tissue biomarkers of synovitis with binary and multicategory receiver operating characteristic analysis.

Immunohistochemical synovial tissue biomarkers are used increasingly to classify arthropathies, study their pathogenesis, and to measure disease activity in clinical trials. We have used receiver operating characteristic (ROC) analysis to quantify the discriminatory abilities of markers for common inflammatory cells (subintimal CD15, CD68, CD3, CD20, CD38, and lining CD68), proliferating cells (Ki-67) and blood vessels (von Willebrand factor, vWF) among inflammatory (chronic septic arthritis, early arthritis and rheumatoid arthritis (RA)) and degenerative arthropathies (osteoarthritis (OA) and orthopedic arthropathies) and normal synovium. Six of the eight markers distinguished accurately between RA and the degenerative arthropathies (area under the curve (AUC) 0.91-0.97), whereas subintimal CD68 (AUC 0.92) and Ki-67 (AUC 0.87) distinguished best between OA and normal synovium. Fold differences in mean expression correlated only modestly with AUCs (r(2) = 0.44). Multicategory ROC analysis ranked Ki-67, subintimal CD68, and CD15 as discriminating best among all six sample groups, and thus identified them as the most broadly applicable markers.