BACKGROUND: The UK Prostate Cancer Risk Management Programme has recommended that all assays for prostate-specific antigen (PSA) should be both equimolar in their response to free and complexed PSA and calibrated to the World Health Organization (WHO) First International Standard for PSA (90:10). To determine which assays currently being used by diagnostic laboratories in England fulfil these criteria, a PSA recovery experiment was performed. METHODS: In all, 15 samples containing varying mixtures and concentrations of the WHO International Standards for PSA and free PSA were sent to 223 laboratories in England who participate in the UK National External Quality Assessment Service (UK NEQAS) scheme for analysis. Analytical platforms were assigned to one of 11 methods, and the results were converted into recovery percentages and analysed by means of a linear random intercept model. RESULTS: No method was both unbiased and equimolar; estimates of bias ranged from -5% to 22% and estimates of non-equimolarity (change in bias per percent increase in free PSA) ranged from -0.18% to 0.28%. CONCLUSIONS: Analytical methods currently in use for PSA in England have inter-method differences; some methods have estimated biases over 10% which can result in unacceptable clinical performance characteristics. Methods for PSA should be equimolar and calibrated to the international standard to minimize the likelihood of clinical errors.
BACKGROUND: The UK Prostate Cancer Risk Management Programme has recommended that all assays for prostate-specific antigen (PSA) should be both equimolar in their response to free and complexed PSA and calibrated to the World Health Organization (WHO) First International Standard for PSA (90:10). To determine which assays currently being used by diagnostic laboratories in England fulfil these criteria, a PSA recovery experiment was performed. METHODS: In all, 15 samples containing varying mixtures and concentrations of the WHO International Standards for PSA and free PSA were sent to 223 laboratories in England who participate in the UK National External Quality Assessment Service (UK NEQAS) scheme for analysis. Analytical platforms were assigned to one of 11 methods, and the results were converted into recovery percentages and analysed by means of a linear random intercept model. RESULTS: No method was both unbiased and equimolar; estimates of bias ranged from -5% to 22% and estimates of non-equimolarity (change in bias per percent increase in free PSA) ranged from -0.18% to 0.28%. CONCLUSIONS: Analytical methods currently in use for PSA in England have inter-method differences; some methods have estimated biases over 10% which can result in unacceptable clinical performance characteristics. Methods for PSA should be equimolar and calibrated to the international standard to minimize the likelihood of clinical errors.