Thompson IM et al. (2005) Operating characteristics of prostate-specific antigen in men with an initial PSA level of 3.0 ng/ml or lower. JAMA 294: 66–70

An important study has shown that, contrary to popular belief, there is no such thing as a 'normal' or 'abnormal' prostate-specific antigen (PSA) level. PSA screening in men aged over 50 has been widespread in the US for almost 20 years, yet no studies have ever shown it to reduce prostate cancer mortality.

Thompson and colleagues set out to measure the diagnostic accuracy of PSA for prostate cancer detection by estimating the 'receiver operating characteristic' curve. This is a plot of 1 – specificity versus sensitivity for each cut-off value of PSA. The team analyzed data from 8,575 men in the placebo group of the Prostate Cancer Prevention Trial, a prospective, multicenter, randomized, controlled trial that ran from 1993 to 2003. The value of this study population was the end-of-study prostate biopsy performed on all cancer-free men, irrespective of PSA level. This avoided verification bias and made it possible to evaluate the operating characteristics of PSA screening.

The researchers found that no cut-off value simultaneously yields both high sensitivity and high specificity, making it difficult for clinicians to know when to recommend a prostate biopsy. The commonly used cut-off value of 4.1 ng/ml detected only 20.5% of prostate cancers. However, lowering the cut-off point so that 83.4% of cases were detected would result in a false-positive rate of 61.1%.

Clinicians and patients must now learn to think of PSA levels in terms of a continuum of prostate cancer risk, rather than definitive cut-off points.