2020 Volume 69 Issue 2 Pages 184-192
The purpose of this study was to compare the diagnostic capability of different immunoassay reagents for detecting serum carbohydrate antigen 19-9 (CA19-9). The five reagents used were as follows: Lumipulse Presto CA19-9, CA19-9XR, Elecsys CA19-9 II, HISCLTM CA19-9 Assay Kit, and VITROS Immunodiagnostic Products CA 19-9 Reagent Pack. In terms of repeatability, coefficients of variation (CVs) of 0.5 to 4.3% were confirmed for the reagents, except for the ARCHITECT reagent, which showed a CV of 3.7–21.8%. In the evaluation of CA19-9 correlation with each reagent, the correlation coefficient for comparison between the Cobas reagent and the Vitros reagent was the best, but the measured value of the Vitros reagent tended to be about twice as large as that of the Cobas reagent. Furthermore, from the correlation evaluation near the cutoff value of CA19-9, it was confirmed that the correlation coefficient for the comparison of CA19-9 with the architect reagent was larger than those for the comparison with the other four reagents. It was also confirmed that the correlation coefficient for the comparison of CA19-9 with the HISCL reagent in the immunoassay using a non-NS19-9 antibody was dispersed compared with those for the other four reagents. On the other hand, interestingly, the CA19-9 pancreatic cancer diagnostic capability of all the reagents evaluated using the receiver operating characteristics (ROC) curve was relatively good with an area under the ROC curve of 0.739 or more. Consequently, there was a discrepancy in the correlation coefficient in the comparison between CA19-9 and other reagents depending on the type of reagent, but there was almost no effect of the pancreatic cancer diagnostic capability in terms of CA19-9 values.
