Arthritis & Rheumatism, Volume 60,
October 2009 Abstract Supplement

The 2009 ACR/ARHP Annual Scientific Meeting
Philadelphia October 16-21, 2009.


Assessing the Diagnostic Accuracy of a 3-Component Synovitis Score and Its Components with Receiver Operating Curve (ROC) Analysis

Slansky1,  Elisabeth, Li2,  Jialiang, Haupl3,  Thomas, Morawietz3,  Lars, Krenn4,  Veit, Pessler1,  Frank

Technical University Dresden Children's Hospital, 01307 Dresden, Germany
National University of Singapore, Singapore
Charite Berlin, Berlin, Germany
Univ. of Trier, Germany

Purpose:

To assess the diagnostic accuracy of a 3-component synovitis score and its components (intimal hyperplasia, stromal cellularity, inflammatory infiltration) with receiver operating characteristic (ROC) curve analysis and multi-category ROC analysis.

Method:

The histological grade of synovitis was scored in 666 synovial specimens carrying the following clinical diagnoses: normal synovium, n=33; post-traumatic arthropathy (PtA), n=29; osteoarthritis (OA), n=221; psoriatic arthritis (PsA), n=42; and rheumatoid arthritis (RA), n=341. The discriminatory abilities of the complete score and each of its components were quantified with the area under the ROC curve (AUC) in each of the 10 possible paired comparisons. Multi-category ROC analysis was used to rank the complete score and each of its components according to their overall discriminatory abilities.

Results:

The score differentiated all arthropathies accurately from normal tissue (AUCs: 0.87–0.98), and RA from OA or PtA (AUCs: 0.85 for both), but could not distinguish well between pairs of inflammatory (RA vs PsA, AUC 0.63) or degenerative (PtA vs OA, AUC 0.59) arthropathies. AUCs of the intimal hyperplasia and stromal cellularity components (r=0.94 and 0.91, respectively) correlated with the AUCs of the complete score markedly more strongly than the inflammatory infiltration component (r=0.60). Multi-category ROC analysis ranked the score and its components in the following order of overall diagnostic accuracy: complete score>stromal cellularity>intimal hyperplasia>inflammatory infiltration, with the complete score being clearly superior to any of its components (Table 1).

Table. Diagnostic ranks of the synovitis score and its components according to multi-category ROC analysis

RankTestHUM1
1Synovitis score0.0620
2Stromal cellularity0.0140
3Intimal hyperplasia0.0075
4Infiltration0.0005
5Nondiscriminatory marker20.000076
1 Values represent hyper volumes under the ROC manifolds (HUMs), calculated according to ref. 1.
2 Hypothetical marker corresponding to the null hypothesis.

Conclusion:

This 3-component synovitis score is an accurate tool for differentiating diseased from normal synovium and inflammatory from degenerative arthropathies. Its high discriminatory ability stems more from measuring proliferative than infiltrative aspects of synovitis. The data also suggest that multi-component synovitis scores may be superior to single component scores.

Reference

1. Li, J. & Fine, J.ROC analysis with multiple classes and multiple tests: methodology and its application in microarray studies. Biostatistics. 2008;9:566–76

To cite this abstract, please use the following information:
Slansky, Elisabeth, Li, Jialiang, Haupl, Thomas, Morawietz, Lars, Krenn, Veit, Pessler, Frank; Assessing the Diagnostic Accuracy of a 3-Component Synovitis Score and Its Components with Receiver Operating Curve (ROC) Analysis [abstract]. Arthritis Rheum 2009;60 Suppl 10 :959
DOI: 10.1002/art.26038

Abstract Supplement

Meeting Menu

2009 ACR/ARHP