Standardization of assay methods reduces variability of total PSA measurements: an Irish study

Study Type – Diagnosis (quality control) Level of Evidence 2b What's known on the subject? and What does the study add? Today, numerous assays for PSA detection are available from various manufacturers. However, these various assays do not detect PSA equally and several studies have demonstrated variability between them. In order to harmonise PSA results and reduce the discrepancies, reference materials are available for assay calibration. We have demonstrated significantly variability between 6 different assay methods currently in use in 9 hospitals despite assay calibration. Variability in PSA values was reduced with the standardisation of the assay method in 4 hospitals. Our results highlight the dilemma of PSA assay variability and stress the need for nationwide standardisation of PSA testing. OBJECTIVE • To determine whether standardization of total prostate‐specific antigen (tPSA) assay methods reduces variability in tPSA measurements. PATIENTS AND METHODS • Blood samples from 84 patients attending a single urology department were distributed across nine hospitals selected throughout Ireland for the independent determination of tPSA under the same conditions. • The selected hospitals collectively used six different assay methods for tPSA detection: Beckman Hybritech WHO Calibrated (used as reference method), Tosoh AIA 1800, Roche E170 (used in three hospitals), Abbott AxSYM, Immulite 2500 2nd Generation (used in two hospitals) and Siemens ADVIA Centaur. • The method of tPSA detection was next standardized in a subset of four hospitals using the same assay method and the measurements were repeated. • The difference in mean tPSA in the cohort across the hospitals tested was determined and the Bland–Altman test was used to assess the agreement between each test. Analysis was performed over both the full (0.5–30 µg/L, N= 84) and a narrow (3–7 µg/L, n= 25) tPSA range. RESULTS • The range and the mean tPSA of the full cohort were inflated across the eight test hospitals, when compared with the reference hospital. • The poorest agreement between assay methods was associated with a bias of 2.2 ± 2.4 µg/L. The variability in tPSA measurements between assay methods was inconsistent across the range of tPSA values tested and increased with increasing mean tPSA. • Agreement in reported tPSA was excellent after standardization of tPSA assay methods (bias