Comparison of 6 automated assays for total and free prostate-specific antigen with special reference to their reactivity toward the WHO 96/670 reference preparation.

BACKGROUND Prostate-specific antigen (PSA) assays have historically produced different results. Our aim was to investigate the comparability of assay results of selected commercially available assay methods designed to measure total, free, or complexed PSA (tPSA, fPSA, and cPSA). METHODS We measured tPSA, fPSA, and cPSA in 70 samples and in the WHO PSA 96/670 reference preparation with 6 assays (Beckman-Coulter Access, Abbott ARCHITECT and AxSYM, Bayer Advia Centaur, DPC IMMULITE 2000, and Roche Modular Analytics E170). We also calculated the fPSA/tPSA ratio. RESULTS The mean deviations from the expected tPSA and fPSA values for the WHO 96/670 reference preparation were 0.37 (range, 0.01-1.32) and 0.19 (range, 0.05-0.49) microg/L, respectively. When plotted against the expected WHO 96/670 reference preparation value, regression slopes varied from 0.99 to 1.22 and r2 from 0.9996 to 1.000. When total PSA was measured in mixtures of sera with high and low tPSA concentrations, the mean (SD) slope of regression of different assays against an in-house method was 1.04 (0.09). In these specimens, the fPSA/tPSA ratio was 0.11-0.14 with different methods. The tPSA and fPSA values in patient samples measured in different assays and plotted against ARCHITECT gave regression slopes from 0.88 to 0.97. The results of the studied assays for tPSA in serum samples agreed within 15%, from each other, and all results for the WHO 96/670 reference preparation were within 6.8% (confidence interval, 1.7%-15.2%) of the expected value. The results for fPSA were more diverse. CONCLUSIONS Differences among PSA assays appear to have decreased since introduction of the WHO 96/670 reference preparation, but further efforts are needed to harmonize fPSA assays.