Increased Sensitivity of the European Medicines Agency Algorithm for Classification of Childhood Granulomatosis with Polyangiitis

Abstract

Granulomatosis with polyangiitis (Wegener's; GPA) and other antineutrophil cytoplasmic antibody (ANCA)-associated vasculitides (AAV) are rare in childhood and are sometimes difficult to discriminate. We compared use of adult-derived classification schemes for GPA against validated pediatric criteria in the ARChiVe (A Registry for Childhood Vasculitis e-entry) cohort, a Childhood Arthritis and Rheumatology Research Alliance initiative. Time-of-diagnosis data for children with physician (MD) diagnosis of AAV and unclassified vasculitis (UCV) from 33 US/Canadian centers were analyzed. The European Medicines Agency (EMA) classification algorithm and European League Against Rheumatism/Paediatric Rheumatology International Trials Organisation/Paediatric Rheumatology European Society (EULAR/PRINTO/PRES) and American College of Rheumatology (ACR) criteria for GPA were applied to all patients. Sensitivity and specificity were calculated (MD-diagnosis as reference). MD-diagnoses for 155 children were 100 GPA, 25 microscopic polyangiitis (MPA), 6 ANCA-positive pauciimmune glomerulonephritis, 3 Churg-Strauss syndrome, and 21 UCV. Of these, 114 had GPA as defined by EMA, 98 by EULAR/PRINTO/PRES, and 87 by ACR. Fourteen patients were identified as GPA by EULAR/PRINTO/PRES but not by ACR; 3 were identified as GPA by ACR but not EULAR/PRINTO/PRES. Using the EMA algorithm, 135 (87%) children were classifiable. The sensitivity of the EMA algorithm, the EULAR/PRINTO/PRES, and ACR criteria for classifying GPA was 90%, 77%, and 69%, respectively, with specificities of 56%, 62%, and 67%. The relatively poor sensitivity of the 2 criteria related to their inability to discriminate patients with MPA. EULAR/PRINTO/PRES was more sensitive than ACR criteria in classifying pediatric GPA. Neither classification system has criteria for MPA; therefore usefulness in discriminating patients in ARChiVe was limited. Even when using the most sensitive EMA algorithm, many children remained unclassified.