Clinical interpretation of prostate-specific antigen values: Type of applied cut-off value exceeds methods bias as the major source of variation.

Abstract

Prostate-specific antigen is the biochemical gold standard for the (early) detection and monitoring of prostate cancer. Interpretation of prostate-specific antigen is both dependent on the method and cut-off. The aim of this study was to examine the effect of method-specific differences and cut-off values in a national external quality assessment scheme (EQAS). The Dutch EQAS for prostate-specific antigen comprised an annual distribution of 12 control materials. The results of two distributions were combined with the corresponding cut-off value. Differences between methods were quantified by simple linear regression based on the all laboratory trimmed mean. To assess the clinical consequence of method-specific differences and cut-off values, a clinical data-set of 1040 patients with an initial prostate-specific antigen measurement and concomitant conclusive prostate biopsy was retrospectively collected. Sensitivity and specificity for prostate cancer were calculated for all EQAS participants individually. In the Netherlands, seven different prostate-specific antigen methods are used. Interestingly, 67% of these laboratories apply age-specific cut-off values. Methods showed a maximal relative difference of 26%, which were not reflected in the cut-off values. The largest differences were caused by the type of cut-off, for example in the Roche group the cut-off value differed maximal 217%. Clinically, a fixed prostate-specific antigen cut-off has a higher sensitivity than an age-specific cut-off (mean 89% range 86-93% versus 79% range 63-95%, respectively). This study shows that the differences in cut-off values exceed the method-specific differences. These results emphasize the need for (inter)national harmonization/standardization programmes including cut-off values to allow for laboratory-independent clinical decision-making.