Prostate Cancer Screening—The Evidence, the Recommendations, and the Clinical Implications

On October 11, 2011, the US Preventive Services Task Force (USPSTF) released its draft recommendation on prostate cancer screening. In it, the USPSTF recommended for the first time against prostate cancer screening for men of all ages (a grade “D” recommendation). Predictably, this was met with considerable controversy.

The mission of the USPSTF is to improve the health of all persons in the United States by making evidence-based recommendations about clinical preventive services. Each USPSTF recommendation is based on a systematic review of the evidence.¹ Before making a recommendation about a preventive service, the task force requires that the evidence be sufficient to estimate with at least moderate certainty the balance of benefits relative to harms.² In this case, across the population of screened men, the USPSTF determined that benefits of prostate-specific antigen (PSA)–based screening for prostate cancer were outweighed by harms.

As of 2010, the USPSTF process requires that the draft recommendation be posted on its Web site for public comment. The extent to which the final recommendation will differ from the draft is not known, as the public comment period has only recently ended, and thousands of comments were submitted and must be processed and reviewed. Nonetheless, the closing of the public comment period provides an opportunity to examine several key issues regarding the evidence supporting the USPSTF recommendation.

The primary benefit assessed in the evidence review was prostate cancer–specific mortality. The 2 highest-quality and largest trials of screening—the European Randomized Study of Screening for Prostate Cancer (ERSPC)³ and the US Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial⁴ —appeared to report conflicting results, although the confidence intervals of the estimates from the 2 trials overlapped. In ERSPC, screening was associated with a decreased absolute risk of prostate cancer mortality of 0.6 per 1000 men (median follow-up, 8.8 years), but in PLCO, screening was associated with an increased absolute risk of 0.2 per 1000 men (median follow-up, 6.3 years). Neither difference was statistically significant, although in a prespecified subgroup of men aged 55 to 69 years, ERSPC reported a statistically significant absolute reduction in risk of prostate cancer mortality of 0.7 per 1000 men, for a number needed to invite to screening of 1410 to prevent 1 prostate cancer death.

The PLCO trial reported a high contamination rate among men randomized to the no-screening group (ie, a high proportion had PSA screening performed outside of the trial) and a high rate of prior PSA testing in all enrollees. However, this does not mean that these results should be dismissed. Even though estimates of benefit would have been attenuated, PLCO should have been able to detect a signal suggesting benefit if it existed, because 20% more cancers—with lower Gleason scores—were identified among men in the screened group. In addition, longer 10-year results (based on two-thirds of the sample) also showed no benefit. Furthermore, PLCO is more applicable to the United States because the study included performance of annual screening (vs screening every 2 to 7 years in ERSPC) and use of potentially curative treatments was more consistent with US practice (about 90% compared with 80% in ERSPC).

ERSPC also had methodological issues, including postrandomization consent only in screened men in some centers and differential treatment between screened and nonscreened groups that may have created a bias toward benefit. ERSPC also excluded results from 2 participating centers.

It is not possible to determine with certainty the degree to which the inconsistency between the 2 trials is explained by these factors. Therefore, the evidence review presented the results separately and reported a range of potential benefits (small to none),¹ with ERSPC representing the upper bound. The USPSTF considered the results of both trials in its deliberations and determined that benefits would not exceed harms even if the PLCO results were excluded.

Will longer follow-up of the trials show greater benefits? Effects of screening on prostate cancer mortality might increase with more prolonged follow-up, given the prolonged natural history of most prostate cancers. On the other hand, competing mortality may attenuate the possibility of additional benefits, as men will be in their 70s and 80s by the time they reach 15 years of follow-up.

What about the Göteborg report, which showed greater effects of screening on prostate cancer mortality (absolute risk reduction, 0.34 percentage points after 14 years) than in ERSPC?⁵ The Göteborg results do not constitute an independent confirmatory study, as about 60% of the sample was included in the main analysis of ERSPC. Rather, these results primarily represent extended follow-up of a selected and previously reported cohort from ERSPC that appeared to report exceptionally good results. Selectively using ERSPC centers reporting the worst results to estimate benefits would be similarly misleading.

What about epidemiologic data showing substantial declines in prostate cancer mortality since the introduction of screening? Indeed, prostate cancer mortality peaked in the United States in 1992 and has declined since then.⁶ However, PSA testing had just been introduced at the end of the 1980s and did not become widespread until the mid- to late 1990s. The decline in prostate cancer mortality was similar in geographic areas that adopted screening early and those that adopted screening later.⁷ Furthermore, ERSPC clearly showed that if PSA screening reduces prostate cancer mortality, it does not occur for 7 to 10 years.³ Therefore, the decline observed from the 1990s to about 2000 could not be from screening but was probably due to other factors such as more effective treatments. Because epidemiologic data are poorly suited for understanding causality, they are weighted lower than randomized trials in USPSTF deliberations.

There is no debate that surgery and radiation therapy, the most commonly used therapies for localized prostate cancer, are associated with important harms (including perioperative death, cardiovascular complications, urinary incontinence, and erectile dysfunction) or that harms occur frequently.¹ Rather, it has been suggested that harms may be mitigated by more judicious use of such treatments for low-risk prostate cancers identified by screening or by refinements in technologies. However, estimates in the evidence review of harms and the proportion of men who undergo curative treatments were based on actual US data, which demonstrate that about 90% of men with localized prostate cancer (including about 75% of those with lower-risk cancers) undergo surgery or radiation therapy.⁸ Whether these rates can be substantially reduced in practice, and the effects such reductions might have on estimates of benefits relative to harms with screening, are yet to be determined.

Regardless of what the final USPSTF recommendation may be, some men will continue to request prostate cancer screening and some physicians will continue to offer it. An individual who is informed about the probability of benefit and harm may choose to be screened because he places a higher value on the possibility of benefit, however small, than the known harms that accompany screening and subsequent treatment—particularly harms related to overdiagnosis and overtreatment. No man should be screened without his explicit consent.

In addition, clinicians and health systems performing screening should demonstrate that they can in fact reduce avoidable harms by being more selective about use of potentially curative treatments for patients with low-risk prostate cancers, ideally informed by results from completed⁹ and ongoing trials¹⁰ of treatment for screen-detected cancers.

The controversies behind prostate cancer screening may never be fully resolved, but a solid grounding in what the evidence demonstrates—and using it to thoughtfully inform health policy and individual screening decisions—can help provide a rational basis for moving forward.