Numerous immunoassays have been commercialized to determine pancreatic elastase (PE) in feces in screening for exocrine pancreatic insufficiency (EPI), but how the different assays compare to one another is controversial, especially in the context that all methods use the same cut-off values for interpreting the results obtained on the presence or absence of EPI or the degree of insufficiency if it is present. Our aim was to analytically verify a new method for determining PE, compare the results with a previous method, and verify the declared cut-off values for interpretation of the results. PE in the stool was assayed using a previous monoclonal enzyme-linked immunosorbent assay ("ScheBo ELISA") and a new polyclonal particle-enhanced turbidimetric immunoassay ("Bühlmann PETIA"). The direct method comparison of two immunoassays was performed in 40 samples. Clinical comparisons were conducted against each other for the binary determination of "abnormal/normal" elastase levels and the three-way determination of "severe/moderate/no" EPI in 56 samples. The indirect comparison method used external quality assessment (EQA) data to compare the monoclonal and polyclonal immunoassays for PE, and additionally compare the monoclonal ScheBo ELISA to a monoclonal chemiluminescence immunoassay ("DiaSorin CLIA"). Precision in the series and intra-laboratory precision for Bühlmann PETIA met the manufacturer's specifications for the concentration range of limit/lower values and the range of normal values. The Bühlmann PETIA immunoassay on different analytical platforms yielded comparable results and nearly perfect agreement in the case of three-way classification (kappa = 0.89 with 95%CI from 0.79 to 1.00. ScheBo ELISA tends to generate higher values of pancreatic elastase than the Bühlmann PETIA; agreement between the methods was moderate in the case of binary classification (kappa = 0.43; 95% CI 0.25 to 0.62), and substantial in the case of three-way classification (kappa = 0.62; 95% CI 0.50 to 0.75). EQA data analysis showed a statistically significant difference between ScheBo ELISA and Bühlmann PETIA peer groups (p = 0.031), as well as the DiaSorin CLIA and ScheBo ELISA peer groups (p = 0.010). The ScheBo ELISA and Bühlmann PETIA do not appear to be commutable in the analytical and clinical context. Our data address a discordance between different mono- and polyclonal immunoassays for pancreatic elastase and the potential of misclassification using its universal cut-off values in screening suspected patients for exocrine pancreatic insufficiency.
Read full abstract