Article révisé par les pairs
Résumé : We evaluated new immunochemiluminometric assays (ICMAs) for insulin and C-peptide (ADVIA Centaur Insulin & C-Peptide-Serum assays). Both ADVIA Centaur assays are two-site sandwich immunoassays using direct chemiluminescent technology. Precision was investigated using serum pools at three levels of the two analytes, measured in duplicate for 10 days. Total Coefficient of Variations (CVs) were 5, 7 and 4% for insulin and 9, 6 and 10% for C-peptide, with intra-assay precisions of 5, 4 and 5% and 5, 3 and 3%, respectively. The minimum detectable concentrations were 0.5 mU/L and < 0.1 microg/L for insulin and for C-peptide, respectively. Day-to-day reproducibility of single measurements was 5.4, 7.1 and 4.3% for pools with an insulin concentration of 0.6 mU/L, 2.0 mU/L and 4.0 mU/L; it was 4.4, 6.6 and 5.3% for pools with a C-peptide concentration of 0.2, 0.3 and 1.0 microg/L. The functional sensitivity did not differ from 3 SD Minimal Detectable Concentration (MDC) (0.5 mU/L for insulin and < 0.1 microg/L for C-peptide). The linearity was good in the range of 0.6-20 mU/L for insulin and 0.3-9 microg/L for C-peptide. The comparison with the RIA used in our laboratory was analyzed by Passing-Bablok and Bland-Altman plots and revealed a proportional bias of approximately 20% (slope: 1.20; CI: 1.14 to 1.26) for C-peptide and a systematic bias of -1.6 mU/L (slope: 0.94; CI: 2.7 to -0.5) for insulin which should not have any clinical consequence in the interpretation of results. Finally, we tested the influence of hemolysis on insulin in serum and plasma and found the same negative effect for both samples when more than 2% of red cells were hemolyzed, and this effect increased with the lag time before freezing. In conclusion, both assays were satisfactorily correlated with the routine RIA test used in our laboratory. The major problem was the sensitivity to hemolysis which is common to all insulin immunometric assays.