Intended for healthcare professionals


Prostate specific antigen for detecting early prostate cancer

BMJ 2009; 339 doi: (Published 24 September 2009) Cite this as: BMJ 2009;339:b3572
  1. Dragan Ilic, senior lecturer,
  2. Sally Green, professorial fellow
  1. 1Monash Institute of Health Services Research, School of Public Health and Preventive Medicine, Monash University, Locked Bag 29, Monash Medical Centre, Clayton, Vic 3168, Australia
  1. dragan.ilic{at}

    Evidence is inconclusive, so patient education and shared decision making are essential

    Clinicians currently rely on prostate specific antigen (PSA) measurement and digital rectal examination as frontline tests for screening and diagnosing prostate cancer. Much of the controversy surrounding the merits of the PSA test is based on uncertainty about its performance in population screening. In the linked study (doi:10.1136/bmj.b3537), Holmström and colleagues show the test’s limitations in this context.3

    In the early 1990s, the PSA test was at the forefront of prostate cancer screening and a threshold of 4 µg/l was set. This value was thought to allow detection of many curable cancers, while limiting the number of false positives and the need for men to undergo further invasive tests.1 Since then the value of this threshold has been much debated because high grade prostate cancer has been found in people with levels below 4 µg/l.2

    In their case-control study, Holmström and colleagues assessed the validity of PSA for predicting a subsequent prostate cancer diagnosis by record linkage to the regional cancer registry.3 They matched 540 cases and 1034 controls taken from the Northern Sweden Health and Disease Cohort by age and date of blood sampling. The positive likelihood ratios were 4.5, 5.5, and 6.4 for PSA cut-off values of 3 µg/l, 4 µg/l, and 5 µg/l, respectively. None of these likelihood ratios provides strong evidence to rule in the disease in the event of a positive test.3 The results echo previous findings, which indicated that a cut-off value of 4 µg/l may increase the specificity of the test but results in a low sensitivity.4 5 Lowering the cut-off value to 3 µg/l increases the sensitivity, but reduces the specificity. Holmström and colleagues conclude that using a PSA cut-off value below 1 µg/l increases the specificity to a degree where the PSA test may be useful in ruling out the disease in patients at low risk.

    Two large randomised control trials investigating the merit of screening for prostate cancer, with respect to prostate cancer specific mortality, illustrate the challenge in finding an appropriate threshold for the PSA test. The European Randomised study of Screening for Prostate Cancer reported a 20% reduction in the risk of prostate cancer mortality with screening.6 Most participating countries in the study used a threshold of 3 µg/l, although some study sites ordered additional diagnostic tests (such as biopsy) for men with a value of 2.5-3.9 µg/l.6 Conversely, the Prostate, Lung, Colorectal and Ovarian screening trial adopted a cut-off value of 4 µg/l and found no significant difference in prostate cancer mortality between screened and unscreened groups.7 Contamination between screening and control groups affected both studies, but this methodological problem may be difficult to overcome in view of the high level of public awareness about screening.

    Clinicians and patients are faced with conflicting evidence about the merits of screening and how to use the PSA test (alone or in combination with digital rectal examination) as a screening tool.6 7 8 9 The lack of a clear PSA cut-off level that differentiates men who have prostate cancer from men who do not is also problematic.2 Adopting a threshold of 1 µg/l may rule out the disease for a certain population of men at a lower risk of developing prostate cancer. A PSA value of 4 µg/l is commonly used as the threshold in practice, but what are the implications for results between 1 µg/l and 4 µg/l?1 Lowering the cut-off value to 1 µg/l may only expose more men who do not need them (that is, false positives) to further biopsies and increase their probably of experiencing an adverse event as a direct result of screening for prostate cancer.

    In the absence of conclusive evidence to inform PSA testing, clinicians should initiate an informed discussion with patients, with the aid of appropriate patient education materials to ensure that patients understand the implications. An ideal test for prostate cancer would be highly specific and highly sensitive.10 Converting sensitivity and specificity into likelihood ratios would help clinicians to evaluate how effective the test might be in ruling in or ruling out the disease. Clinicians should consider using likelihood ratios together with a patient’s individual risk factors for the disease to explain the potential benefits and harms of the PSA test, allowing patients to contribute to decisions.11

    The current limitations of the PSA test, as shown by Holmström and colleagues, have driven research into the potential of other biomarkers in screening for prostate cancer.12 Further research is needed to develop and evaluate a screening test that could discriminate between high and low grade prostate cancer and can be implemented in a wider population.


    Cite this as: BMJ 2009;339:b3572


    • Research, doi:10.1136/bmj.b3537
    • Competing interests: DI and SG are members of the Cochrane Prostatic Diseases and Urologic Cancers Group and have previously received funding from Andrology Australia, which is a community and professional education programme funded by the Australian government that provides health information on male reproductive health. SG is also a member of the Management Group for Andrology Australia.

    • Provenance and peer review: Commissioned, not externally peer reviewed.


    View Abstract