Neurology 2019 Aug 29;93(9):e889-e894. Epub 2019 Jul 29.
From Ghent University Hospital (P.A.O., B.D.P., R.C., J.L.D.B.), Belgium; Academic Medical Centre (E.A., F.B., M.d.V.), University of Amsterdam, the Netherlands; Harvard Medical School (A.A.), Boston, MA; CHU Henri Mondor (D.D.), Créteil, France; Hospital Santa Creu i Sant Pau (E.G.), Barcelona, Spain; Université Paris Est-Créteil (R.G.), France; Charité-Universitätsmedizin Berlin (H.-H.G., W.S.), Germany; John Radcliffe Hospital (D.H.-J.); Oxford University Hospitals (M.H.); UCL Institute of Neurology (J.H.), London, UK; Syddansk Universitet (H.d.S.), Odense, Denmark; and Mayo Clinic (D.S.), Rochester, MN.
Objective: To determine interrater variability in diagnosing individual muscle biopsy abnormalities and diagnosis.
Methods: We developed a scoring tool to analyze consensus in muscle biopsy reading of an ad hoc workgroup of international experts. Twenty-four samples from patients with suspected idiopathic inflammatory myopathy (IIM) were randomly selected, providing sections that were stained with standard histologic and immunohistochemical methods. Sections were made available on an online platform, and experts were queried about myopathologic features within 4 pathologic domains: muscle fibers, inflammation, connective tissue, and vasculature. A short clinical presentation of cases was included, and experts were asked to give a tentative diagnosis of polymyositis, dermatomyositis, inclusion-body myositis, antisynthetase syndrome-related myositis, immune-mediated necrotizing myopathy, nonspecific myositis, or other disease. Fleiss κ values, scoring interrater variability, showed the highest agreement within the muscle fiber and connective tissue domains.
Results: Despite overall low κ values, moderate agreement was achieved for tentative diagnosis, supporting the idea of using holistic muscle biopsy interpretation rather than adding up individual features.
Conclusion: The assessment of individual pathologic features needs to be standardized and harmonized and should be measured for sensitivity and specificity for subgroup classification. Standardizing the process of diagnostic muscle biopsy reading would allow identification of more homogeneous patient cohorts for upcoming treatment trials.