Comparative Evaluation of Chemiluminescence Immunoassay and ELISA for the Detection of Islet Autoantibodies in Type 1 Diabetes

Background: Early detection of type 1 diabetes (T1D) through major islet antibodies screening is gaining momentum as public health strategy, with Italy recently implementing a pilot pediatric screening program. The transition from research-based screening to large-scale population initiatives needs automated and standardized assays capable of processing extensive sample volumes. This study was hence aimed at evaluating the analytical performance and comparability of a fully-automated chemiluminescence immunoassay (CLIA) compared to a conventional enzyme-linked immunosorbent assay (ELISA) for detection of three classes of major islet antibodies anti-GAD (GADA), anti-IA-2 (IA-2A) and anti-ZnT8 (ZnT8A). Methods: A total of 104 serum specimens were analyzed for each autoantibody using both ELISA (DYNES, DSX) and CLIA (MAGLUMI 800). Assay precision and linearity were assessed through intra-assay variability studies and dilution protocols. Methods agreement was evaluated with Passing-Bablok regression, Spearman’s correlation, Bland-Altman analysis and Cohen’s kappa statistics. Results: CLIA showed good precision and excellent linearity across clinically relevant concentration ranges of all islet antibodies. Correlation coefficients and categorical agreement between CLIA and ELISA were high (r>0.96 and Cohen’s kappa >0.8 for all), with ZnT8A exhibiting the highest concordance. However, proportional biases were found, as CLIA systematically underestimated GADA and ZnT8A levels, while overestimated IA-2A compared with ELISA. Conclusions: CLIA assays displayed satisfactory precision and agreement with ELISA for GADA, IA-2A and ZnT8A detection. Our findings support the use of these automated immunoassay in large-scale population initiatives for diagnosing TD1, but we also highlight the need for further efforts to achieve better inter-assay harmonization.