Screening for prostate cancer using PSA is a careful balance of benefits and harms. But current US practice involves testing older men who have little to gain and aggressively treating low-risk cancers. Debates about whether to test need to be replaced by debates on how to test better.
The US Preventive Services Task Force (USPSTF) recently issued a recommendation against the use of prostate-specific antigen (PSA) testing for prostate cancer screening. They concluded that “there is moderate or high certainty that [prostate cancer screening] has no net benefit or that the harms outweigh the benefits.” In this article, we review the USPSTF report and make three simple points. First, the USPSTF report is riddled with errors, so much so that we would be sympathetic to accusations that the task force was biased. Second, if the USPSTF were indeed biased against PSA screening, this would be entirely understandable: urologists, radiation oncologists and others have made such a mess out of PSA screening that it is easy to see why a group of family practitioners, obstetricians and paediatricians would like to write the whole thing off. Third, PSA screening can be done in different ways, and the ratio of benefit to harm will depend on choices regarding how PSA tests are used. As mid-life levels of PSA are strongly predictive of long-term risk of prostate cancer morbidity,[2,3] we would argue for risk-stratified approaches, to minimise harms for men unlikely to benefit from screening and ensure careful follow up of those at the highest risk of unfavourable outcome.
Regarding our first point, the USPSTF report is riddled with errors of fact, interpretation and statistics. Some of these errors might be considered understandable. Take, for example, the claim that in the interim report from the European randomised screening trial (ERSPC) after a median follow up of 9 years,4 “48 men received treatment for every prostate cancer-specific death prevented.” The number of 48 patients was obtained by dividing the between-group difference in prostate cancer diagnoses with the between-group difference in cancer deaths. As not all men diagnosed with prostate cancer in this study were treated — some were placed on active surveillance — the USPSTF statement is incorrect. It is also highly misleading, as the ratio of diagnoses to deaths that are avoided is time dependent; consider that this ratio is infinity at early follow up because screening does not prevent death in a man diagnosed with advanced-stage cancer at his first PSA test. The empirical estimate from the Göteborg arm of ERSPC, which has longer follow up (14 years) is that 12 men need to be diagnosed to prevent one death from prostate cancer. Still, the ERSPC report4 used the phrase “number-needed-to-treat” and cited the number 48, so perhaps the USPSTF error is understandable.
The principal flaw of the USPSTF might also be seen as an understandable mistake. Specifically, the USPSTF draws definitive conclusions of “moderate or high certainty of no benefit” on the basis of interim data: the largest randomised trial of prostate cancer screening – the European ERSPC trial – has not yet reported on the main endpoint of cancer-related mortality at its prespecified primary timepoint, data were only reported because the difference between groups crossed a prespecified significance boundary at interim analysis. It seems bizarre to be certain of “no benefit” when a major trial is yet to report in full.
What is less understandable is that the USPSTF make unsupportable claims that seem designed to emphasise that screening is harmful and that there should be less of it. For example, the USPSTF cites a perioperative mortality rate from radical prostatectomy of 0.5%, far higher than most contemporary estimates, such as 0.1%. This is because they used a study of Medicare patients to draw their conclusions, that is, the oldest patients at highest risk for perioperative death. In addition, it is hard to understand the biological mechanism behind the claim that because “the [negative US] trial evaluated a shorter screening interval [than the positive European trial] … more conservative screening and treatment strategies might be more effective than more aggressive ones.” Less regular screening may well decrease the harms of screening, but there is simply no mechanism by which it could be more effective.
Our second point is that contemporary PSA screening and treatment is a farrago and so if the members of the USPSTF were indeed prejudiced against PSA screening, it is not hard to see why. There is a lot to dislike about how prostate cancer is detected and managed in the US. For example, PSA screening is routinely used in men who have nothing to gain from it, with testing applied to one-third of men aged over 70 years who have a greater than 50% risk of death within five years.7 In addition, digital rectal examination is widely used even though it is not informative in a screening setting.8 Urologists are then extremely quick to biopsy, with current guidelines recommending biopsy for almost any indication: a raised PSA, a lowered ratio of free-to-total PSA, a high PSA velocity or a positive digital rectal examination. Worst of all, radiotherapy or surgical treatment is almost universally recommended: empirical studies show that fewer than 10% of men with low-risk disease are offered active surveillance.9 Couple this with apparent conflicts of interest, such as groups of urologists purchasing radiation equipment and then self-referring patients, and it is not hard to see why those outside of the prostate cancer field see PSA testing as nothing more than a scam. Prostate cancer screening is not a single intervention, such as a certain dose of a specific drug; it can be implemented in numerous different ways. Starting PSA screening at, say, 70 years, using a very low PSA threshold for biopsy and then aggressively treating all cancers will lead to enormous amounts of overdiagnosis and overtreatment and will have little effect on mortality. Conversely, focusing on younger men, only biopsying those meeting stringent criteria, and managing low-risk cancers by active surveillance will lead to a better balance of harms and benefits. Indeed, given the diversity of approaches to PSA screening and subsequent management of PSA-detected tumours, it is hard to know whether it is even coherent to make statements such as “PSA screening is associated with a 42% rate of overdiagnosis” or “48 men need to be diagnosed after a PSA test to save one life”.1 We would argue that the interim analysis of ERSPC and prespecified analysis from the Göteborg randomised trial in Europe demonstrates that PSA-based screening can reduce cancer-specific mortality and, as such, our question should really be how to make it work better. A key method will clearly be risk stratification: focusing PSA screening on the men at highest risk of prostate cancer morbidity and mortality will improve the ratio of benefit to harms. As it turns out, the most powerful risk factor is PSA itself.2, 3 Indeed, re-analyses of the European ERSPC trial suggest that if men with a low baseline PSA level were exempted from further screening, there would be a dramatic reduction in the number of men screened, biopsied, diagnosed and treated per prostate cancer death avoided.
In summary, the question is should we abandon PSA testing? One answer might be that yes, we should: current PSA testing as it is commonly practised in the US is indefensible. However, we should avoid throwing out the baby with the bathwater and instead grasp the opportunity to implement a more-rational, risk-stratified approach to PSA screening, which avoids testing of men with little to benefit and uses active surveillance to manage low-risk prostate cancer. Such a strategy has the best chance to reduce prostate cancer mortality while minimising overdiagnosis and overtreatment.
1. R Chou et al. (2011) Screening for prostate cancer: a review of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 155:762–771
2. H Lilja et al. (2011) Prediction of significant prostate cancer diagnosed 20 to 30 years later with a single measure of prostate-specific antigen at or before age 50. Cancer 117:1210–19
3. AJ Vickers et al. (2010) Prostate specific antigen concentration at age 60 and death or metastasis from prostate cancer: case-control study. BMJ 341:c4521
4. FH Schröder et al. (2009) Screening and prostate-cancer mortality in a randomized European study. NEJM 360:1320–28
5. J Hugosson et al. (2010) Mortality results from the Göteborg randomised population-based prostate-cancer screening trial. Lancet Oncol 11:725–732
6. S Carlsson et al. (2009) Nationwide population-based study on 30-day mortality after radical prostatectomy in Sweden. Scand J Urol Nephrol 43:350–356
7. MW Drazer, D Huo, MA Schonberg et al. (2011) Population-based patterns and predictors of prostate-specific antigen screening among older men in the United States. JCO 29:1736–43
8. FH Schroder, M Roobol-Bouts, AN Vis et al. (2001) Prostate-specific antigen-based early detection of prostate cancer – validation of screening without rectal examination. Urology 57:83–90
9. MR Cooperberg, JM Broering and PR Carroll. (2010) Time trends and local variation in primary treatment of localized prostate cancer. JCO 28:1117–23
10. PJ van Leeuwen et al. (2010) Balancing the harms and benefits of early detection of prostate cancer. Cancer 116:4857–65
Acknowledgements: Supported in part by funds from the Goldstein Foundation, a P50-CA92629 SPORE grant and a R33 CA127768-03 grant from the National Cancer Institute, Swedish Cancer Society (08-0345), Swedish Research Council (Medicine no. 20095), David H. Koch provided through the Prostate Cancer Foundation and by the Sidney Kimmel Center for Prostate and Urologic Cancers
Author affiliations: Andrew J Vickers: Department of Epidemiology and Biostatistics, Hans Lilja: Departments of Laboratory Medicine and Surgery (Urology) and Department of Medicine (GU-Oncology), both at the Memorial Sloan-Kettering Cancer Center, New York, New York, USA
Competing interests statement: Hans Lilja is an inventor and owner of patents WO 0227323, US 2002123616, WO 0193861, WO 9201936, WO 9626442, EP 0635575, and DE 9117047. Andrew J Vickers declares no competing interests
Practice points: The outcomes of PSA screening could be dramatically improved by:
- Avoiding screening in older men (age ≥70 years)
- Use of active surveillance to manage low-risk disease
This article was first published in Nature Reviews Clinical Oncology vol. 9 no.1, and is published with permission. © 2012 Nature Publishing Group. doi:10.1038/nrclinonc.2011.181