PSA and the Detection of Prostate Cancer After 2005. Part II: Ways Out of the PSA Dilemma?

This part II of a paper on the “dilemma of PSA” is a continuation of part I which appeared in the first issue of the EAU-EBU update series. The paper deals with the two dilemma's of PSA: (1) the decreased prediction of PSA as a diagnostic tool in pre-screened populations. (2) the fact that even at in very low PSA ranges large proportions of biopsy detectable cancers are present and that Gleason 7 or higher cancers have a low but possibly clinically relevant prevalence in these populations.In the first place the issue of test performance of PSA is addressed. It is pointed out that we cannot really calculate test sensitivity and specificity as long as we do not know the prevalence of those cancers we wish to put into the denominator of the equation. Any calculation of sensitivity which uses a “gold standard” leads to erroneous numbers. The discussion of this issue concludes that clinically the most useful denominator in calculating sensitivity would be the rate of aggressive cancers which do not include those indolent cases which are most likely to contribute to overdiagnosis.