The diagnostic value of synovial histopathology in undifferentiated inflammatory arthritis: a prospective study in 154 consecutive patients
© The Author(s) 2003
Received: 14 January 2003
Published: 24 February 2003
To explore prospectively the diagnostic value of synovial histopathology in undifferentiated inflammatory arthritis by analysing the positive predictive value (PPV) of single histological markers and the additive value of multiparameter models.
Synovial biopsies were obtained in 154 consecutive patients presenting for diagnostic work-out: 67 were clinically classified as rheumatoid arthritis (RA), spondyloarthropathy (SpA), or other disease at time of arthroscopy (cohort 1). Patients with undifferentiated arthritis (n = 87) were re-evaluated after 6 months, yielding a diagnosis in 53 (cohort 2). Macroscopic, histological, and immunohistochemical synovial parameters with diagnostic value were identified in cohort 1. Subsequently, the PPV of single parameters and multiparameter models was tested in cohort 2.
Four single parameters yielded PPVs of more than 95% in cohort 1: anticitrulline staining, mAb 12A staining (recognising a specific MHC/peptide complex), and lining layer hyperplasia of more than five cell layers for RA and crystal depositions for other diseases. In undifferentiated arthritis, only anticitrulline staining, mAb 12A staining, and crystal depositions had a PPV of more than 90%. Using these parameters, 39.6% of the undifferentiated patients were classified, with a PPV of 90.5%.
Using cohort 1 as a learning file, two decision trees were developed which made a prediction in 74.6% and 67.2% of the patients, with PPVs of 88.0% and 86.7%, respectively. When this was applied to cohort 2, a diagnostic prediction was made in 79.2% and 71.7% of the patients, with a PPV of 81.0% and 86.8%, respectively.
This study supports the idea that synovial histopathology can be used for the early diagnosis of undifferentiated arthritis and that multiparameter models have an additive value compared with single parameters.
The authors thank Annemieke Boots and Peter Steenbakkers, Organon NV, for providing mAb 12A.