Parallelism in immunoassays: two approaches applied to the determination of anti-deamidated gliadin peptide antibodies

Abstract

Introduction: Parallelism analysis examines whether the calibrators behave in the same manner as the endogenous analyte against the immunoreactants kit. Once confirmed, the kit can be applied for semiquantitative purposes in the diagnostic field.
Objectives: To analyze ELISAs parallelism as part of an operative verification program of two commercial immunoassays that quantitate anti-deamidated gliadin peptide antibodies of IgA and IgG class (DGP-A and DGP-G). Materials and methods: Four human samples termed M1 to M4 were processed. To analyze parallelism between samples and standards, a logistic fourparameter model (4PL) was adjusted to samples and standards, and then the data were linearized with the aid of Logit-Log transformation. ANOVA was used to compare the similarity between Hill coefficient (4PL) and slopes(linear adjustment). Then, to analyze parallelism among samples, we used an approach based on the variation coefficient (VC %). By processing serial dilutions of each sample, optical densities were interpolated in the calibration curve and the resulting concentrations were dilutioncorrected. Antibody levels obtained from this inverse procedure should remain consistent in the dilution range explored, without deviations from the tolerable VC% value. Results: Both approaches agreed in that M1 did not react in a predictable fashion with the DGP-A kit. The same conclusion was drawn from M3 analysis with the DGP-G kit. Slope analysis showed an exaggerated sensitivity when detecting parallelism deviations. Conclusions: It is recommended to prioritize graphical evaluation of 4PL curves and set aside the parameter similarity-based tests in favor of the VC%-based method.