More Screening Scrutiny: An Update on Screening for Prostate Cancer

Posted by Rena Xu • March 14th, 2012

Last year, the U.S. Preventive Services Task Force (USPSTF) issued a controversial draft recommendation against routine screening for prostate cancer.  The rationale was that for asymptomatic men, prostate-specific antigen (PSA)-based screening offered unclear survival benefit while posing at least a small risk.

In this week’s NEJM, an article by Schröder et al. report updated data from a European study of routine prostate cancer screening that challenges this premise.  The European Randomized Study of Screening for Prostate Cancer (ERSPC), which started in 1991, enrolled more than 180,000 men aged 50 to 74 years across eight European countries.  Men randomized to the study’s intervention arm were offered PSA-based screening; men in the control arm were not.  The primary outcome measured was death from prostate cancer.  Here, the authors updated their previously published findings to include two additional years of follow-up data, for a median follow-up of 11 years.

The updated findings were largely consistent with those previously reported.  In the screening group, there were 299 deaths from prostate cancer (0.39 per 1,000 person-years), as compared with 462 prostate cancer deaths in the control group (0.50 per 1,000 person-years).  This translated to a 21% relative reduction in prostate cancer mortality (29% after adjusting for non-compliance), similar to the degree of risk reduction that had been reported previously.  The absolute risk reduction continued to be modest at 1.07 deaths per 1,000 men.  This meant that more than 930 men would need to be invited for screening, and more than 30 cancers would need to be detected, in order to prevent one death from prostate cancer.  There continued to be no observed difference in all-cause mortality between the intervention and control groups.  Routine screening, in other words, was not associated with an overall lower rate of death for men in the studied age range.

Should these findings change how we think about prostate cancer screening in the US?  The answer depends on several things.  First, it depends on the validity of the study design.  Some experts have voiced concern that the results of the ERSPC may have been confounded by treatment inequities between the screening and control groups.  As Dr. Anthony B. Miller of the University of Toronto notes in an accompanying editorial, men diagnosed with prostate cancer in the screening group were more likely to be treated at an academic center than those in the control group.  If this is true, then the observed difference in prostate cancer mortality might be attributable to differences in the overall quality of care received, as opposed to differences in screening practices.

Second, the clinical relevance of the study findings depends on the degree to which the study population reflects the population at large.  As Dr. Miller points out, a similar trial conducted in the US – the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial – found no added benefit from routine screening, a result at odds with those of the ERSPC.  One of several factors potentially contributing to this discrepancy is the difference in rates of “opportunistic screening” – screening that happens outside of a routine schedule – in the general population of Europe versus the US.  In other words, even if routine screening were to yield a reduction in prostate cancer mortality in Europe, it might not do so in places like the US where opportunistic screening is more prevalent.

Finally, deciding whether the ERSPC results should influence clinical practice requires making a judgment about what costs –physical, psychological, and financial – should be considered acceptable as a trade-off for the anticipated benefits of routine screening.  The additional follow-up data only reinforce the finding that screening does not lower the rate of all-cause mortality.  Furthermore, it is clear that the costs of screening are not trivial: from false positives to over-diagnosis, screening can lead to unnecessary anxiety, infections from biopsy, surgical complications such as impotence and urinary incontinence, and diversion of resources from other health care needs.  Additional objective evidence about the effectiveness of prostate cancer screening will certainly help to guide this decision-making process by answering the question of what screening can or cannot do.  Ultimately, however, the question of what screening is worth – to an individual, and to society– is at least partially a subjective one.  Perhaps that is why an answer has proven so elusive and why the subject of screening promises to be a continuing source of controversy for the foreseeable future.

How do you approach the subject of prostate cancer screening with your patients?  How do you anticipate recommendations from the USPSTF influencing your clinical practice?

One Response to “More Screening Scrutiny: An Update on Screening for Prostate Cancer”

  1. Tim Bartik says:

    This blog entry, like much of the news coverage of the article, misinterprets the “all-cause mortality” results.

    The blog entry says “There continued to be no observed difference in all-cause mortality between the intervention and control groups. Routine screening, in other words, was not associated with an overall lower rate of death for men in the studied age range.”

    No, that is not what the study finds, or could be able to find. Actually, the POINT estimate was that all-cause mortality declined in the intervention group relative to the control group, by about twice the reduction in prostate cancer mortality between the two groups.

    What is true is that the confidence interval on the all-cause mortality result is wide enough that the all-cause mortality result is not significantly different from zero. But this result is to be expected given the sample size. The result is also not statistically significantly different from what would occur if the prostate cancer mortality reduction was reflected one-for-one in a reduction in all-cause mortality.

    Given that the all-cause mortality rate in the control group was almost 20%, we would not expect with a sample of this size to detect a decline in all-cause mortality of 0.1%, which is the observed decline in prostate cancer mortality. Given the study’s sample sizes, the “power” of a statistical test of a difference in all-cause mortality would only be 8%. To get a power of 0.8, that is an 80% chance of detecting a statistically significant effect on all-cause mortality from 20% to 19.9%, one would need an experiment with over 4 million participants.

    Therefore, this study’s statistically significant effects or lack thereof, on all-cause mortality are a red herring, a distraction, from the true issues here.

Leave a Reply

Please note: Comment moderation is enabled and may delay your comment. There is no need to resubmit your comment.

Submitting a comment indicates you have read and agreed to the Terms and Conditions.