Background: We compared two recently developed research assays for the measurement of human kallikrein 2 (hK2) in serum: one fully automated assay (Beckman Coulter Access immunoanalyzer) and one manual assay based on the DELFIA technology.
Methods: We used two subsets of clinical specimens consisting of 48 samples from prostate cancer patients and 210 samples from participants in an ongoing screening study (ERSPC). Both subsets were measured in the Rotterdam laboratory, and the prostate cancer samples were used for analytical comparison with the originating sites for the assays: Beckman Coulter Research Department (San Diego, CA) and Turku University (Turku, Finland).
Results: Both the Beckman Coulter and the Turku assays performed very similarly between the Rotterdam laboratory and the originating sites: the R(2) value for both comparisons was 0.99, and the slope difference between sites was <20%. Deming regression analysis of the DELFIA (y) and Access (x) assays yielded the following: for the prostate cancer group, y = 1.17x - 0.01 (R(2) = 0.88; n = 48); and for the ERSPC group, y = 0.62x - 0.01 (R(2) = 0.77). Breakdown of the latter group into subgroups (nondiseased, benign prostatic hyperplasia, and prostate cancer samples) gave only minor differences. The Access calibrators were underrecovered by 13% in the DELFIA assay, whereas the DELFIA calibrators were overrecovered by 45% in the Access assay.
Conclusion: The DELFIA and Access assays for hK2, which have similar analytical features, show differences that cannot be explained by calibration.