Background/Purpose: Classification of chronic vasculitis to clinically or etiologically meaningful groups has been challenging.  In 2008, the 1990 ACR classification criteria for GPA was adapted for children as the 2008 EULAR/PRINTO/PRES criteria. In 2017, ACR/EULAR proposed new criteria for classifying GPA using new knowledge, and based on data from 1500 adult patients in the Diagnosis and Classification Criteria in Vasculitis (DCVAS) initiative. We aimed to evaluate the 2017 criteria for GPA in a cohort of children and compare the 2008 criteria.

Methods: Time of diagnosis data was retrieved from the international Registry of Childhood Vasculitis (ARChiVe) in PedVAS. The 2017 Classification criteria and scoring system are shown in Table 1 with adaptations of criteria to align with the available ARChiVe dataset.  All patients except Takayasu  (TAK) were included. Classification of GPA versus not GPA patients by both the 2017 and 2008 criteria was computed from data, and sensitivity / specificity measured against Expert classification.  Expert Opinion was determined as follows: If physician submitted diagnosis matched both 1990 and 2008 criteria for GPA, or not GPA, the diagnosis was considered as verified; an expert (KM or DC) evaluated clinical vignettes of remaining patients and verified the diagnosis if it matched submitted diagnosis. Both experts evaluated remaining patients and when consensus could not be reached patients were considered unclassifiable.

Results: After applying the 1990 and 2008 criteria for GPA in all 376 included patients (Table 2), those fulfilling both criteria matched submitted diagnosis of GPA in 163; and for those not fulfilling either criteria there was a matched submitted diagnosis that was not-GPA in 89. Clinical vignettes of the remaining 124 patients underwent expert review for classification. Ultimately 245 patients had GPA and 131 did not.  The 2017 criteria for GPA and the 2008 criteria were each applied to all patients and sensitivity specificity respectively measured against ÒexpertÓ classification: 2017 criteria 93% and 72%, and 2009 criteria 95% and 70%. The 2017 criteria classified 6 additional patients.

Conclusion: Performance of the 2017 and 2008 classification criteria was similar but applicability of the 2017 criteria across the ages is valuable; low specificity may reflect relatively low non GPA numbers and exclusion of TAK. The value of criteria in both adults and children will be best assessed if classification to GPA is seen to be mutually exclusive of classification to other closely related diseases, when these criteria also become available from DCVAS.