gms | German Medical Science

GMDS 2012: 57. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

16. - 20.09.2012, Braunschweig

The relative composition of synovial inflammatory infiltrates as a tool for differential diagnosis of chronic joint diseases

Meeting Abstract

  • Cristina Della Beffa - Helmholtz Zentrum für Infektionsforschung, Braunschweig, Deutschland
  • Elisabeth Slansky - Helmholtz Zentrum für Infektionsforschung, Braunschweig, Deutschland
  • Claudia Pommerenke - Helmholtz Zentrum für Infektionsforschung, Braunschweig, Deutschland
  • Frank Klawonn - Helmholtz Zentrum für Infektionsforschung, Braunschweig, Deutschland
  • Jialiang Li - National University of Singapore, Singapore
  • Lie Dai - Sun Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou, Schweizina
  • H. Ralph Schumacher - hiladelphia VA Medical Center, Philadelphia, USA
  • Frank Pessler - Helmholtz-Zentrum für Infektionsforschung, Braunschweig, Deutschland; Medical Faculty “Carl Gustav Carus” and University Children’s Hospital, Technical University of Dresden, Deutschland

GMDS 2012. 57. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Braunschweig, 16.-20.09.2012. Düsseldorf: German Medical Science GMS Publishing House; 2012. Doc12gmds153

doi: 10.3205/12gmds153, urn:nbn:de:0183-12gmds1539

Published: September 13, 2012

© 2012 Della Beffa et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( You are free: to Share – to copy, distribute and transmit the work, provided the original author and source are credited.



Introduction: The synovial membrane is the key target of inflammation in chronic arthropathies, and the nature or intensity of the inflammatory infiltrates may vary according to the diagnosis. Traditionally, differences in absolute numbers of cells expressing a certain immunohistochemical marker (e.g., positive staining cells per mm2) have been used in histological synovial tissue classification. We have begun to evaluate the relative composition of the inflammatory infiltrates as an alternate classification tool.

Material: Synovial tissue specimens (normal synovium, n=15; orthopedic arthropathies, n=6; osteoarthritis, n=26; early undifferentiated arthritis, n=10; rheumatoid arthritis, n=24; and chronic septic arthritis, n=11) were stained immunohistochemically for the cell markers CD15 (neutrophilic granulocytes), CD68 (macrophages), CD3 (T cells), CD20 (B cells), and CD38 (plasma cells). Densities of cells expressing a given marker were determined in the superficial subintima.

Methods: The absolute densities of cells expressing a given marker (absolute method) were compared with the percentage of these cells in the total inflammatory cell population (relative method) to differentiate among the six sample groups. Statistical significance of expression differences was determined with the Mann-Whitney U (Wilcoxon rank-sum) test. Binary and multicategory receiver operating characteristic (ROC) analysis were used to assess the ability of each of the five markers (using either the absolute of relative values) to differentiate between any of the possible pairs of disease groups. Furthermore, naïve Bayes classifiers were used to predict the proportion of correctly classified disease states, using all markers or selected subsets of markers.

Results: The inflammatory infiltrates in normal synovium and the orthopedic arthropathies consisted almost exclusively of CD68+ and CD3+ cells. Notable fractions of CD20+ and CD38+ cells appeared in a subset of osteoarthritis samples, and increased further in early, rheumatoid and septic arthritis. The ROC analyses ranked the absolute method above the relative method in terms of overall discriminatory ability in that the mean values of the area under the ROC curve in binary ROC analysis (0.90 vs. 0.71, p=4.5x10-9) and the hypervolume under the manifold in multicategory ROC analysis (0.73 vs. 0.26, p=0.027) were significantly higher when the absolute method was applied. Similar results were obtained with the Naïve Bayes classifiers. However, the relative method did perform well in selected comparisons. Notably, the relative method became superior to the absolute method when the samples were also stratified according to the total number of inflammatory cells/mm2.

Conclusions: Measuring the relative proportions of inflammatory cell types aids in specific scenarios of synovial tissue classification. Concurrent stratification according to the intensity of inflammation broadens its applicability.