Platinum Priority – Editorial and Reply from Authors Referring to the article published on pp. 455–466 of this issue PSA is Dead, Long Live PSA Andrew J. Vickers a, *, Hans Lilja b,c,d a Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY, USA; b Departments of Laboratory Medicine, Surgery (Urology), Medicine (GU-Oncology), Memorial Sloan-Kettering Cancer Center, New York, NY, USA; c Department of Laboratory Medicine, Division of Clinical Chemistry, Lund University, Ska ˚ ne University Hospital, Malmo ¨, Sweden; d Institute for Biosciences and Medical Technology, University of Tampere, Tampere, Finland Prostate-specific antigen (PSA) has been called many things in recent years, few of them nice. Some of the words used to describe PSA include over, discredited, and lost. The chair of the US Preventive Services Task Force stated that PSA ‘‘cannot tell the difference between cancers that will and will not affect a man during his natural lifetime.’’ Some have gone further, calling PSA ‘‘an unconscionable crime against humanity’’ (references available on request). All this might be seen as somewhat odd, seeing as PSA is a very strong predictor of the long-term risk of aggressive prostate cancer. PSA in blood measured at age 60 in unscreened men has an area under the curve (AUC) of 0.90 for predicting cancer-specific death within 25 yr [1]. This level of predictive accuracy is pretty much unheard of in cancer care. So why the bad press? The first problem is that of end points. It is true that PSA is not very good at predicting prostate cancer on biopsy [2], but, then again, biopsy-detectable cancer may not be a particularly important end point. It well known that a very high incidence of cancer is found in the prostates of men who die of causes other than cancer and that many older men harbor indolent disease. A study finding that PSA cannot distinguish well between what is likely to be an indolent cancer versus no cancer at all does not imply that PSA has no clinical utility. The second problem is that of cohorts. Many papers reporting that PSA has little or no correlation with biopsy outcome included men who were subject to repeated PSA testing and who had an extremely narrow range of PSA levels, just above common biopsy thresholds. By way of analogy, take a group of 12-yr-old boys and have them play basketball; height will be a major predictor of ability. Then select out the tallest boys and have them play again; height will no longer discriminate, but that does not imply that being tall is not an advantage for basketball players. To illustrate this point, we reanalyzed data from a study [3] of men undergoing repeat biopsy for persistently raised PSA after an initial negative biopsy—by definition, a group with rather homogenous PSA levels (50% were between 3 and 5 ng/ml). The AUC for PSA to predict positive biopsy was 0.55, which is statistically better than a coin flip but not by much. One could stop there and declare PSA ‘‘over,’’ ‘‘lost,’’ or ‘‘discredited,’’ or one could note that PSA was a good predictor of high-grade (Gleason 7) cancer (AUC: 0.68) and even for predicting Gleason 8 disease (AUC: 0.79). There is another reason why PSA is more useful than it is so often made out to be. When we currently say that a man has a serum PSA of 2 ng/ml, for example, this is often taken to mean that there are 2 ng of the PSA molecule in every milliliter of that man’s blood. But what the conventional PSA test measures is actually several different molecules, with the final concentration being their sum. Urologists generally recognize that PSA occurs in blood either bound to protein (‘‘complexed’’ PSA) or ‘‘free.’’ But due to its nature as a proteolytic enzyme, PSA can be bound to different proteins, primarily a measurable complex with a1-antichymotrypsin or disguised when it is bound to a2-macroglobulin. Free (ie, unbound) PSA can also occur in multiple different forms. Just as we now look at diets in terms of fat, protein, and carbohydrate rather than just in terms of total calories, it seems reasonable to suppose that examining different PSA fractions separately, rather than lumping them together, might be of value. EUROPEAN UROLOGY 61 (2012) 467–470 available at www.sciencedirect.com journal homepage: www.europeanurology.com DOI of original article: 10.1016/j.eururo.2011.10.038 * Corresponding author. Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, 307 E 63 rd St., New York, NY 10065, USA. Tel. +1 646 735 8142; Fax: +1 646 735 0011. E-mail address: vickersa@mskcc.org (A.J. Vickers). 0302-2838/$ – see back matter # 2011 European Association of Urology. Published by Elsevier B.V. All rights reserved.