CA 15-3 : a multicentre evaluation of automated and manual tests

The introduction of a new and automated CA 15-3 immunoassay (IMx Abbott) prompted us to compare the analytical performance of this new test with established tests from CIS ELSA, Sorin, and Boehringer Mannheim in a multicentre study. CA 15-3 measurements in blood samples of breast tumour patients, comparison of intra- and inter-assay variation, dilution linearity, and lower limit of detection are described. The study showed improved precision for the automated over the manual test systems (intra-assay variation : IMx 10% ; inter-assay variation : IMx ≤ 8%, CIS ELSA ≤ 19%, ES 300 ≤ 9% and ETI Sorin ≤ 27%). Results on patients' samples (n = 101 to 184) showed highly comparable results ; IMx vs CIS ELSA site 1 : r = 0.950 ; IMx vs CIS ELSA site 2 : r = 0.998 ; IMx vs ES 300 : r = 0.980 ; IMx vs ETI Sorin : r = 0.931. Slopes of regression lines varied from 0.666 for IMx vs ETI Sorin to 0.988 for IMx vs CIS ELSA (site 1, where heparin plasma was used instead of serum as recommended by the manufacturer). Slopes and correlation coefficients were found to be only slightly dependent on assay ranges analysed by statistical procedures applied. Despite good correlations between methods, it is recommended that samples collected in the follow-up of disease and at higher CA 15-3 concentrations are analysed by the same test ; a changeover to another test is not encouraged.