Cytoplasmic islet cell autoantibodies (ICAs) of 13 coded sera were determined by 26 laboratories. ICAs were determined by 24 laboratories according to a standard protocol based on the conventional indirectimmunofluorescence technique on cryostat sections of human pancreas. In addition, these 24 laboratories had performed any of 8 modifications of the assay. Test samples were titrated to end point, and the results obtained with the different methods were compared with those obtained by the standard protocol. The remaining 2 laboratories used either exclusively pancreatic sections of monkey instead of human as substrate (lab 23) or applied immunohistochemical staining (lab 22).
By following the standard protocol, interlaboratory concordance was >90% for the presence or absence of islet cell antibodies in 7 of the 13 samples circulated. However, a wide variability of titers was recorded, ranging from negative to 128.
Of the modifications, prolonged incubation in the presence of aprotinin was performed by 10 laboratories and was found to improve the sensitivity in 57 of 70 (81%) determinations with samples that had been ICA positive by the standard protocol. Improved sensitivity was also noted by 2 laboratories with sections of monkey pancreas. Acetone-fixed sections, used by 6 laboratories, or a two-color immunofluorescence method, applied by 3 laboratories, did not change the titers in 27 of 35 (77%) and 14 of 25 (56%) determinations with samples that had been ICA positive by the standard protocol. In contrast, heat inactivation of the samples before testing, performed by 5 laboratories, resulted in a decrease in titers in 25 of 39 (64%) determinations. Four laboratories used fluorescein-labeled protein A as reagent and observed a decrease in ICA titers in 22 of 27 (81%) determinations. Complement-fixing antibodies were determined by 20 laboratories and were present in 66 of 136 (48%) determinations with ICA-positive samples, and in general, they were associated with high titers of the latter. Three laboratories reported results obtained by immunohistochemical staining with glucose oxidase, and at this stage, the data indicated the requirement of both method improvement and exchange of reference reagents for interlaboratory comparison.