Scoring Sacroiliac Joints by Magnetic Resonance Imaging. A Multiple-Reader Reliability Experiment
ROBERT B.M. LANDEWÉ, KAY-GEERT A. HERMANN, DÉSIRÉE M.F.M. VAN DER HEIJDE,
XENOPHON BARALIAKOS, ANNE-GRETHE JURIK, ROBERT G. LAMBERT, MIKKEL ØSTERGAARD,
MARTIN RUDWALEIT, DAVID C. SALONEN, JÜRGEN BRAUN, and the ASAS/OMERACT MRI in AS Working Group
ABSTRACT.
Magnetic resonance imaging (MRI) of the sacroiliac (SI) joints and the spine is increasingly important in the assessment of inflammatory activity and structural damage in clinical trials with patients with ankylosing spondylitis (AS). We investigated inter-reader reliability and sensitivity to change of several scoring systems to assess disease activity and change in disease activity in patients with AS. Twenty sets of consecutive MRI, derived from a randomized clinical trial comparing an active drug with placebo and selected on the basis of the presence of activity at baseline, were presented electronically to 7 experienced readers from different countries (Europe, Canada). Readers scored the MRI by 3 different methods including: a global score (grading activity per SI joint); a more comprehensive global score (grading activity per SI joint per quadrant); and a detailed scoring system [Spondyloarthritis Research Consortium of Canada (SPARCC) scoring system], which scores 6 images, divided into quadrants, with additional scores for "depth" and "intensity." A fourth and a fifth scoring system were constructed afterwards. The fourth method included the SPARCC score minus the additional scores for "depth" and "intensity," and the fifth method included the SPARCC slice with the maximum score. Inter-reader reliability was investigated by calculating intraclass correlation coefficients (ICC) for all readers together and for all possible reader pairs. Sensitivity to change was investigated by calculating standardized response means (SRM) on change scores that were made positive. Overall inter-reader ICC per method were between 0.47 and 0.58 for scoring status, and between 0.40 and 0.53 for scoring change. ICC per possible reader pairs showed much more fluctuation per method, with lowest observed values close to zero (no agreement) and highest observed values over 0.80 (excellent agreement). In general, agreement of status scores was somewhat better than agreement of change scores, and agreement of the comprehensive SPARCC scoring system was somewhat better than agreement of the more condensed systems. Sensitivity to change differed per reader, but in general was somewhat better for the comprehensive SPARCC system. This experiment under "real life," far from optimal conditions demonstrates the feasibility of scoring exercises for method comparison, provides evidence for the reliability and sensitivity to change of scoring systems to be used in assessing activity of SI joints in clinical trials, and sets the conditions for further validation research in this field. (J Rheumatol 2005;32:2050-5)
Key Indexing Terms:
ANKYLOSING SPONDYLITIS
MAGNETIC RESONANCE IMAGING
SACROILIAC JOINT
VALIDITY
RELIABILITY
From the Department of Internal Medicine, Division of Rheumatology, University Hospital Maastricht, and CAPHRI Research Institute, University Maastricht, Maastricht, The Netherlands; Departments of Radiology and Rheumatology, Charité Medical School, Berlin, Germany; Rheumazentrum Ruhrgebiet, Herne, Germany; Department of Radiology, Aarhus University Hospital, Aarhus, Denmark; Department of Radiology, University of Alberta, Edmonton, Canada; University of Copenhagen Hvidovre Hospital, Hvidovre, Denmark; and Department of Rheumatology, University of Toronto, Toronto, Canada.
R. Landewé, MD, PhD, Department of Internal Medicine, Division of Rheumatology, University Hospital Maastricht, and CAPHRI Research Institute, University Maastricht; K-G. Hermann, MD, Department of Radiology, Charité Medical School; D.M.F.M. van der Heijde, MD, PhD, Department of Internal Medicine, Division of Rheumatology, University Hospital Maastricht, and CAPHRI Research Institute, University Maastricht; X. Baraliakos, Rheumazentrum Ruhrgebiet; A-G. Jurik, Department of Radiology, Aarhus University Hospital; R.G. Lambert, Department of Radiology, University of Alberta; M. Østergaard, MD, PhD, DMSc, University of Copenhagen, Hvidovre Hospital; M. Rudwaleit, MD, Department of Rheumatology Charité -- Campus Benjamin Franklin; D.C. Salonen, Department of Rheumatology, University of Toronto; J Braun, Rheumazentrum Ruhrgebiet.
The ASAS/OMERACT MRI in AS Working Group: Jürgen Braun, Désirée van der Heijde (chairs), Xenofon Baraliakos, Matthias Bollow, Paul Emery, Kay-Geert Hermann, Robert Inman, Anne-Grethe Jurik, Mart van de Laar, Robert G. Lambert, Robert Landewé, Walter Maksymowych, Helena Marzo-Ortega, Phil O'Connor, Mikkel Østergaard, Ans Oostveen, Martin Rudwaleit, David Salonen, Jochen Sieper, Millicent Stone, and Kurt de Vlam.
Address reprint requests to Dr. R. Landewé, Department of Internal Medicine/Rheumatology, University Hospital Maastricht, PO Box 5800, 6202 AZ Maastricht, The Netherlands. E-mail: rlan@sint.azm.nl
