THU0418 Automated interpretation of ANCA patterns - a new approach in the serology of ANCA associated vasculitis

Background Indirect immunofluorescence (IIF), utilizing ethanol-fixed neutrophils as substrate, is still the method of choice for detecting antineutrophil cytoplasmic antibodies (ANCA) in vasculitis. The conventional immunofluorescence microscopy for detection of ANCA is subjective and interlaboratory tests showed occasionally large variances in ANCA determination. Objectives Aim of this study was to develop a novel ANCA module for the automated immunofluorescence reading system AKLIDES® for detection of ANCA and to assess its diagnostic value. Methods Samples were analyzed by IIF with ethanol- and formalin-fixed neutrophils (ethN and formN). Seventy ANCA positive samples with defined antigen specificities (myeloperoxidase, n=20 and proteinase 3, n=50) were used as a “training set” for the development of the new ANCA module of AKLIDES®. Sera from 291 patients (“test set”) with ANCA associated vasculitis and other systemic rheumatic diseases were tested for ANCA using the novel ANCA module and conventional fluorescence microscopy. Results For the analysis of IIF patterns of the nucleus and cytoplasm new algorithms of image processing were developed by using the training set. The module is divided into two profiles for the recognition of ANCA patterns of ethN and formN. Furthermore, a differentiation between positive and negative samples and a detailed separation into the different basic patterns C-ANCA (cytoplasmic pattern) and P-ANCA (perinuclear pattern) were implemented. The ANCA module was evaluated with the test set of 291 sera with ethN and formN slides regarding to the differentiation between positive and negative samples and compared with the manual interpretation of the slides. The Cohen’s kappa values for this comparison were 0.871 for ethN and 0.866 for formN. Furthermore, the analysis of the “test set” with regard to the differentiation of the C-ANCA and P-ANCA pattern showed a high agreement for ethN (kappa =0.739), formN (kappa =0.742), and for the combined result (kappa =0.901). The resulting sensitivities/specificities and negative/positive likelihood ratios for the comparison between manual and automatic reading (ethN, formN and combined result) showed medium to high agreement. Conclusions In summary, a novel ANCA fluorescence pattern recognition algorithm could be implemented successfully into AKLIDES®. The preliminary investigation of the automatic reading of ethanol- and formalin-fixed neutrophils demonstrated a high diagnostic performance for the detection of ANCA. Further studies are needed to evaluate this novel fully automated method for initial screening of suspected vasculitis patients. Disclosure of Interest None Declared