Method comparison of diabetes mellitus associated autoantibodies in serum specimens

Autoantibodies against structure elements of the pancreas are an essential part of laboratory diagnosis of diabetes mellitus. The heterogeneity of current methods and difficult inter-assay comparability of results poses challenges to both clinical interpretation as well as research integrity.

We conducted a method validation study comparing the measurement of autoantibodies against glutamic acid decarboxylase (GADA), islet antigen 2 (IA-2A), and zinc transporter 8 (ZnT8A) on three commercially available platforms (EUROLabWorkstation analyzer, enzyme-linked immunosorbent assay (ELISA); iFlash 1800 (chemiluminescence immunoassay (CLIA); Maglumi 800 (CLIA)).

Compared to the results from EUROLabWorkstation analyzer that was routinely used in our laboratory, the Maglumi demonstrated an acceptable agreement and deviation for GADA (75.4-35.8 %) and IA-2A (71.7-50.3 %). The iFlash displayed less favorable results with agreement and deviation for GADA (51.5-108.8 %) and IA-2A (68-32.5 %). While the iFlash showed excellent agreement for ZnT8A, the Maglumi produced almost exclusively negative results below a 200 U/mL threshold which results in severe underestimation of results on the Maglumi. Overall, no consistent agreement was observed across all three parameters among the CLIA devices tested.

Our study reveals significant and clinically relevant discrepancies in the detection of GADA and IA-2A when comparing CLIA devices to the ELISA method routinely used in our laboratory. For individual antibodies up to 65 % of patients who tested positive using ELISA showed negative or borderline results on at least one CLIA device. These deviations are unpredictable and cannot be detected through standard calibration and internal control procedures.