A Review of the Evidence for the U.S. Preventive Services Task Force

Release Date: October 2011

By Roger Chou, MD; Jennifer M. Croswell, MD, MPH; Tracy Dana, MLS; Christina Bougatsos, BS; Ian Blazina, MPH; Rongwei Fu, PhD; Ken Gleitsmann, MD, MPH; Helen C. Koenig, MD, MPH; Clarence Lam, MD, MPH; Ashley Maltz, MD, MPH; J. Bruin Rugge, MD, MPH; and Kenneth Lin, MD

The information in this article is intended to help clinicians, employers, policymakers, and others make informed decisions about the provision of health care services. This article is intended as a reference and not as a substitute for clinical judgment.

This article may be used, in whole or in part, as the basis for the development of clinical practice guidelines and other quality enhancement tools, or as a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.

This article was first published in Annals of Internal Medicine on October 7, 2011 (www.annals.org).

Background: Screening can detect prostate cancer at earlier, asymptomatic stages, when treatments might be more effective.

Purpose: To update the 2002 and 2008 U.S. Preventive Services Task Force evidence reviews on screening and treatments for prostate cancer.

Data Sources: MEDLINE (2002 to July 2011) and the Cochrane Library Database (through second quarter of 2011).

Study Selection: Randomized trials of prostate-specific antigen–based screening, randomized trials and cohort studies of prostatectomy or radiation therapy versus watchful waiting, and large observational studies of perioperative harms.

Data Extraction: Investigators abstracted and checked study details and quality using predefined criteria.

Data Synthesis: Of 5 screening trials, the 2 largest and highest-quality studies reported conflicting results. One found that screening was associated with reduced prostate cancer–specific mortality compared with no screening in a subgroup of men aged 55 to 69 years after 9 years (relative risk, 0.80 [95% CI, 0.65 to 0.98]; absolute risk reduction, 0.07 percentage point). The other found no statistically significant effect after 10 years (relative risk, 1.1 [CI, 0.80 to 1.5]). After 3 or 4 screening rounds, 12% to 13% of screened men had false-positive results. Serious infections or urinary retention occurred after 0.5% to 1.0% of prostate biopsies. There were 3 randomized trials and 23 cohort studies of treatments. One good-quality trial found that prostatectomy for localized prostate cancer decreased risk for prostate cancer–specific mortality compared with watchful waiting through 13 years of follow-up (relative risk, 0.62 [CI, 0.44 to 0.87]; absolute risk reduction, 6.1%). Benefits seemed to be limited to men younger than 65 years. Treating approximately 3 men with prostatectomy or 7 men with radiation therapy instead of watchful waiting would each result in 1 additional case of erectile dysfunction. Treating approximately 5 men with prostatectomy would result in 1 additional case of urinary incontinence. Prostatectomy was associated with perioperative death (about 0.5%) and cardiovascular events (0.6% to 3%), and radiation therapy was associated with bowel dysfunction.

Limitation: Only English-language articles were included. Few studies evaluated newer therapies.

Conclusion: Prostate-specific antigen–based screening results in small or no reduction in prostate cancer–specific mortality and is associated with harms related to subsequent evaluation and treatments, some of which may be unnecessary.

Primary Funding Source: Agency for Healthcare Research and Quality

Prostate cancer is the most commonly diagnosed cancer in U.S. men ^1-3. Prostate-specific antigen (PSA)–based screening can detect prostate cancers at earlier, asymptomatic stages, when treatments might be more effective.

The U.S. Preventive Services Task Force (USPSTF) last reviewed the evidence on prostate cancer screening ⁴ and issued recommendations in 2008 ⁵. Since then, large trials of prostate cancer screening have been published ^{6, 7}. Benefits and harms of treatments for prostate cancer were last reviewed by the USPSTF in 2002 ⁸. This article summarizes 2 recent reviews commissioned by the USPSTF to synthesize the current evidence on screening ⁹ and treatments ¹⁰ for localized prostate cancer.

Scope of the Review

We followed a standardized protocol and developed an analytic framework that focused on the following key questions:

1. Does PSA-based screening decrease prostate cancer–specific or all-cause mortality?
2. What are the harms of PSA-based screening for prostate cancer?
3. What are the benefits of treatment of early-stage or screening-detected prostate cancer?
4. What are the harms of treatment of early-stage or screening-detected prostate cancer?

Detailed methods and data for the review, including search strategies, multiple evidence tables with quality ratings of individual studies, and pooled analyses of some harms data, are available in the full report ¹⁰. Also of note, androgen deprivation therapy, cryotherapy, and high-intensity focused ultrasonography are reviewed in the full report ¹⁰ but are not presented in this article.

Data Sources and Searches

We searched OVID MEDLINE from 2002 to July 2011, PubMed from 2007 to July 2011, and the Cochrane Library Database through the second quarter of 2011 and reviewed reference lists to identify relevant articles published in English.

Study Selection

At least 2 reviewers independently evaluated each study to determine inclusion eligibility. We restricted inclusion to published studies. We included randomized trials of screening for prostate cancer in asymptomatic men (including those with chronic, mild lower urinary tract symptoms) that incorporated 1 or more PSA measurements, with or without additional methods, such as digital rectal examination, and reported all-cause or prostate cancer–specific mortality or harms associated with screening. We also included randomized trials and cohort studies of men with screening-detected prostate cancer that compared radical prostatectomy or radiation therapy (the most common primary treatments for localized prostate cancer ^{11, 12} with watchful waiting and reported all-cause mortality, prostate cancer–specific mortality, or prespecified harms (quality of life or functional status, urinary incontinence, bowel dysfunction, erectile dysfunction, psychological effects, and surgical complications). We included studies of clinically localized (T1 or T2) prostate cancer because more than 90% of screening-detected prostate cancers are localized ^{6, 7, 13}. We included only studies that reported risk estimates for mortality adjusted at a minimum for age at diagnosis and tumor grade (no study reported adjusted risk estimates for treatment harms). We also included large (>1000 participants) uncontrolled observational studies of perioperative mortality and surgical complications.

We classified “no treatment,” “observation,” or “deferred treatment” as watchful waiting because patients probably received at least watchful waiting. We also grouped watchful waiting with active surveillance because studies of active surveillance provided insufficient information to determine whether more active follow-up actually occurred ¹⁴, and older studies used these terms interchangeably.

Data Extraction and Quality Assessment

One investigator abstracted details about the patient population, study design, analysis, duration of follow-up, and results. A second investigator reviewed data abstraction for accuracy. Two investigators independently applied criteria developed by the USPSTF ¹⁵ to rate the quality of each study as good, fair, or poor. Discrepancies were resolved through a consensus process.

Data Synthesis and Analysis

We assessed the aggregate internal validity (quality) of the body of evidence for each key question (good, fair, and poor) by using methods developed by the USPSTF on the basis of the number, quality, and size of studies; consistency of results between studies; and directness of evidence ¹⁵. We synthesized results of treatment studies descriptively, using medians and ranges, because few randomized, controlled trials (RCTs) were available and studies varied in the populations and interventions evaluated, methodologic quality, duration of follow-up, and other factors. We stratified results according to study type and qualitatively assessed the effects of study quality, duration of follow-up, year of publication, and mean age on results.

Role of the Funding Source

This study was funded by the Agency for Healthcare Research and Quality (AHRQ) under a contract to support the work of the USPSTF. Staff at AHRQ and USPSTF members helped develop the scope of this work and reviewed draft manuscripts. The draft systematic reviews were reviewed by external peer reviewers not affiliated with the USPSTF, then revised for the final version. Approval from AHRQ was required before this manuscript could be submitted for publication, but the authors are solely responsible for the content and the decision to submit.

Appendix Figure 1 and Appendix Figure 2 show the results of the search and study selection process.

We identified 2 fair-quality ^{6, 7} and 3 poor-quality ^16-20 randomized trials of PSA-based screening (Appendix Table 1). We also included a report describing results from a single center ²¹ participating in a fair-quality trial ⁷. Sample sizes ranged from 9026 to 182,160 and maximum follow-up from 11 to 20 years (median, 6 to 14 years).

We identified 11 studies (2 RCTs ^22-29 and 9 cohort studies ^30-38) on benefits of prostate cancer treatments and 16 studies (2 RCTs ^39-42 and 14 cohort studies ^43-58) on harms (Appendix Table 2). Sample sizes ranged from 72 to 44,630 and duration of follow-up from 1 to 23 years. Four studies were rated good quality ^{23, 42, 52, 56, 58}, 1 poor quality ²⁹, and the remainder fair quality. Frequent methodologic shortcomings were failure to describe loss to follow-up (6 cohort studies and all 3 RCTs met this criterion) and inadequate blinding of outcome assessors (no cohort studies and 1 RCT met this criterion). Only 2 studies ^{33, 40} clearly described the control group intervention (Appendix Table 2). We also included 6 observational studies ^59-64 of surgical complications after prostatectomy.

Key Question 1: Does PSA-Based Screening Decrease Prostate Cancer–Specific or All-Cause Mortality?

The fair-quality U.S. Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial randomly assigned 76,693 men between 55 and 74 years of age to annual PSA screening in combination with digital rectal examination versus usual care ⁶. After 7 years' (complete) follow-up, screening was associated with increased prostate cancer incidence (relative risk [RR], 1.2 [95% CI, 1.2 to 1.3]) but no effect on prostate cancer–specific (RR, 1.1 [CI, 0.75 to 1.7]) or all-cause (RR, 0.98 [CI, 0.92 to 1.0]) mortality. Similar results were observed after 10 years (67% of sample; RR, 1.1 [CI, 0.80 to 1.5]). Up to 52% of men assigned to usual care underwent a PSA test at some point during the trial, and 44% of trial participants had undergone PSA screening before entry.

The fair-quality European Randomized Study of Screening for Prostate Cancer (ERSPC) randomly assigned 182,000 men aged 50 to 74 years from 7 countries to PSA testing every 2 to 7 years (depending on center and year) or to usual care ⁷. Data from 2 other study centers were excluded for reasons not specified in the study protocol. Levels of PSA for diagnostic evaluation ranged from 2.5 to 4.0 µg/L (1 center used 10 µg/L for several years). Recruitment and randomization procedures and age eligibility also varied. After a median of 9 years, prostate cancer incidence was higher in the screened group (net increase, 34 per 1000 men), but there was no statistically significant difference in prostate cancer–specific mortality (RR, 0.85 [CI, 0.73 to 1.0]). A prespecified subgroup analysis of 162,243 men aged 55 to 69 years found that screening was associated with reduced prostate cancer–specific mortality (RR, 0.80 [CI, 0.65 to 0.98]; absolute risk reduction, 0.07 percentage point), for an estimated 1410 men invited to screening and 48 treated to prevent 1 prostate cancer–specific death.

After publication of the main ERSPC results, 1 participating center (Göteborg, Sweden) reported results separately ²¹. It found PSA screening (threshold, 2.5 to 3.0 µg/L) every 2 years in 20,000 men aged 50 to 64 years to be associated with increased prostate cancer incidence (hazard ratio [HR], 1.6 [CI, 1.5 to 1.8]) and decreased risk for prostate cancer–specific mortality (RR, 0.56 [CI, 0.39 to 0.82]; absolute risk reduction, 0.34 percentage point) after a median of 14 years. Outcomes for 60% of participants were included in the main ERSPC report ⁷. Although no other center separately reported results, only exclusion of the Swedish center data from the overall ERSPC analysis resulted in loss of the statistically significant effect of screening on prostate cancer–specific mortality (RR, 0.84 [CI, 0.70 to 1.01]), suggesting better results than the other centers ⁷.

Three poor-quality trials (number of men invited to screening ranged from 1494 to 31,333) found no difference between screening-invited and control groups in prostate cancer–specific mortality risk ^{16, 17, 20}. Two of the trials ^{17, 19} were included in the 2008 USPSTF review ⁴; results after 5 years' additional follow-up are now available from 1 of the trials ²⁰. Methodological shortcomings in these trials included failure to describe adequate randomization or allocation concealment methods, poorly described loss to follow-up, and unclear masking of outcomes assessors. One trial used a high PSA cut point (10 µg/L) ¹⁶.

Key Question 2: What Are the Harms of PSA-Based Screening for Prostate Cancer?

Direct harms of PSA-based screening were reported in the ERSPC and PLCO trials ^{6, 7}. The Finnish center of the ERSPC trial found that 12% of men received at least 1 false-positive result after 3 rounds of PSA testing (cutoff, 4.0 µg/L) ⁶⁵. For the entire ERPSC trial, 76% of prostate biopsies for an elevated PSA level identified no cancer ⁷. In the PLCO trial, the cumulative risk for at least 1 false-positive result was 13% after 4 PSA tests (cutoff, 4.0 µg/L), with a 5.5% risk for undergoing at least 1 biopsy due to a false-positive test result ⁶⁶.

Physical harms of screening in the PLCO trial included bleeding or pain from digital rectal examination (0.3 event per 10,000 screened); bruising or fainting due to venipuncture (26 events per 10,000 screened); and biopsy complications, such as infection, bleeding, and urinary difficulties (68 events per 10,000 evaluations) ⁶. The Rotterdam, Netherlands, center of the ERSPC trial reported that among 5802 biopsies performed, 200 men (3.5%) developed fever, 20 (0.4%) experienced urine retention, and 27 (0.5%) required hospitalization for signs of prostatitis or urosepsis ⁶⁷.

None of the RCTs of PSA-based screening provided information on potential psychological harms, such as anxiety, or adverse effects on health-related quality of life. The 2008 USPSTF review found evidence that false-positive PSA test results are associated with adverse psychological effects but could not estimate their magnitude ⁴.

Key Question 3: What Are the Benefits of Treatment of Early-Stage or Screening-Detected Prostate Cancer?

Prostatectomy

Prostatectomy was compared with watchful waiting in 1 good-quality RCT (n=695) of men with localized (stage T1b, T1c, or T2) prostate cancer (Appendix Table 3) ^{22-24, 28}. It did not specifically enroll men with screening-detected prostate cancer, and about 75% of cancers were palpable (stage T2). By comparison, 36% of localized cancers in the ERSPC screening trial were stage T2 ⁷. The 2002 USPSTF review included results through 6 years of follow-up ²⁸. Data now available through 15 years showed a sustained decrease in risk for prostate cancer–specific mortality (15% vs. 21%; RR, 0.62 [CI, 0.44 to 0.87]; absolute difference, 6.1 percentage points [CI, 0.2 to 12 percentage points]) and all-cause mortality (RR, 0.75 [CI, 0.61 to 0.92]; absolute difference, 6.6 percentage points [CI, −1.3 to 14 percentage points]) ²³. In subgroup analyses, benefits were restricted to men younger than 65 years of age (RR, 0.49 [CI, 0.31 to 0.79] for prostate cancer–specific mortality; RR, 0.52 [CI, 0.37 to 0.73] for all-cause mortality). A small (n=142), poor-quality RCT found no difference between prostatectomy and no prostatectomy for localized prostate cancer on overall survival through 23 years ²⁹. It did not report prostate cancer–specific mortality.

Eight cohort studies (median n=2264 [range, 316 to 25,900]) with a duration of follow-up ranging from 4 to 13 years consistently found prostatectomy for localized prostate cancer to be associated with decreased risk for all-cause mortality (6 studies; median adjusted HR, 0.46 [range, 0.32 to 0.67] ^{31, 33-37}) and prostate cancer–specific mortality (5 studies; median adjusted HR, 0.32 [range, 0.25 to 0.50] ^{30, 33, 35, 36, 38}) compared with watchful waiting (Appendix Table 3). The largest was a fair-quality, propensity-adjusted analysis of data from the U.S. Surveillance, Epidemiology, and End Results (SEER) program (n=25,900) of men 65 to 80 years of age that found decreased risk for all-cause mortality after 12 years (adjusted HR, 0.50 [CI, 0.66 to 0.72]) ³⁷. A large (n=22,385), fair-quality Swedish cohort study also found prostatectomy to be associated with decreased risk for all-cause mortality after 4 years of follow-up, after adjustment for age, Gleason score, and PSA level (adjusted HR, 0.41 [CI, 0.36 to 0.48]) ³¹.

Radiation Therapy

No RCTs compared radiation therapy versus watchful waiting. Five cohort studies (median n=3441 [range, 334 to 30,857]) with follow-up ranging from 4 to 13 years consistently found that radiation therapy (external-beam radiation therapy [EBRT] or unspecified modality) for localized prostate cancer was associated with decreased risk for all-cause mortality (5 studies; median adjusted HR, 0.68 [range, 0.62 to 0.81] ^{31, 35-38}) and prostate cancer–specific mortality (5 studies; median adjusted HR, 0.66 [range, 0.63 to 0.70]) compared with watchful waiting (Appendix Table 3) ^{30, 35-38}. The largest study, a previously described analysis of SEER data, found radiation therapy to be associated with decreased propensity-adjusted risk for all-cause mortality (adjusted HR, 0.81 [CI, 0.78 to 0.85]) ³⁷. A large Swedish cohort study (also described earlier) found radiation therapy to be associated with decreased risk for all-cause mortality (adjusted HR, 0.62 [CI, 0.54 to 0.71]) ³¹.

Key Question 4: What Are the Harms of Treatment of Early-Stage or Screening-Detected Prostate Cancer?

Prostatectomy

Urinary Incontinence and Erectile Dysfunction. Prostatectomy was associated with increased risk for urinary incontinence compared with watchful waiting in 1 RCT (RR, 2.3 [CI, 1.6 to 3.2]) ⁴¹ and 4 cohort studies (median RR, 4.0 [range, 2.0 to 11]) (Appendix Table 4) ^{47, 49, 53, 56}. In the RCT, the absolute increase in risk for urinary incontinence with surgery was 28 percentage points (49% vs. 21%) ⁴¹. In the cohort studies, the median rate of urinary incontinence with watchful waiting was 6% (range, 3% to 10%), with prostatectomy associated with a median increase in absolute risk of 18 percentage points (range, 8 to 40 percentage points) ^{47, 49, 53, 56}.

Prostatectomy was also associated with an increased risk for erectile dysfunction compared with watchful waiting in 1 RCT (RR, 1.8 [CI, 1.5 to 2.2]) ⁴¹ and 5 cohort studies (median RR, 1.5 [range, 1.3 to 2.1]) (Appendix Table 4) ^{47, 49, 53, 56}. In the RCT, the absolute increase in risk for erectile dysfunction with surgery was 36 percentage points (81% vs. 45%) ⁴¹. In the cohort studies, the median rate of erectile dysfunction with watchful waiting was 52% (range, 26% to 68%), with prostatectomy associated with a median increase in absolute risk of 26 percentage points (range, 21 to 29 percentage points) ^{47, 49, 53, 56}.

Differences in study quality, duration of follow-up, or year of publication did not seem to explain differences in estimates across studies. The studies provided few details about the specific surgical procedures evaluated, although open retropubic radical prostatectomy was the dominant procedure when most of the studies were conducted ⁶⁸. One observational study stratified estimates for erectile dysfunction and urinary incontinence by use of a nerve-sparing (n=494; 68% and 9.4%, respectively) versus a non–nerve-sparing (n=476; 87% and 15%, respectively) technique ⁵⁶.

Consistent with the studies reporting dichotomous outcomes, 8 cohort studies that evaluated urinary and sexual function outcomes by using continuous scales found that prostatectomy was associated with worse outcomes compared with watchful waiting (Appendix Table 4 ^{43, 46, 48, 51, 53, 55-57}).

Quality of Life. Eight studies reported generic quality of life ^{43, 46, 48, 50, 51, 53, 55, 56}. Two studies reported very similar Short-Form 36 (SF-36) physical and mental component summary scores after prostatectomy and watchful waiting (Appendix Table 5) ^{43, 56}. On specific SF-36 subscales, prostatectomy was associated with better physical function (6 studies; median difference, 8 points [range, 2 to 16 points]) ^{43, 46, 48, 51, 53, 55} and emotional role function subscale scores (7 studies; median difference, 8 points [range, −5 to 13 points]) ^{43, 46, 48, 50, 51, 53, 55}, with small or no clear differences on other SF-36 subscales.

Surgical Complications. The largest (n=101,604) study of short-term (≤30-day) complications after prostatectomy reported a 30-day perioperative mortality rate of 0.5% in Medicare claimants ⁶⁰; 3 other large observational studies reported similar findings ^{59, 61, 62}. Advanced age and increased number of serious comorbid conditions were associated with higher perioperative mortality, although absolute rates were less than 1% even in men at higher risk. In the Medicare database study, perioperative rates of serious cardiovascular events were 3% and rates of vascular events (including pulmonary embolism and deep venous thrombosis) were 2% ⁶⁰. In 2 other studies (n=1243 ⁶³ and 11,010 ⁵⁹), rates of cardiovascular events were 0.6% and 3% and rates of vascular events 1% and 2%, respectively. Serious rectal or ureteral injury due to surgery ranged from 0.3% to 0.6% ^{60, 63}.

Other Harms. Five studies (reported in 6 publications) found no clear differences between prostatectomy and watchful waiting in risk for bowel dysfunction ^{41, 42, 46, 47, 49, 56}. One RCT found no difference between prostatectomy and watchful waiting in risk for high levels of anxiety, depression, or worry after 4 years ⁴².

Radiation Therapy

Urinary Incontinence and Erectile Dysfunction. Radiation therapy was associated with increased risk for urinary incontinence compared with watchful waiting in 1 small RCT, but the estimate was very imprecise (RR, 8.3 [CI, 1.1 to 63]) because of small numbers of events (1 in the watchful waiting group) (Appendix Table 4) ³⁹. There was no clear increase in risk in 4 (total n=1910) cohort studies (median RR, 1.1 [range, 0.71 to 2.0]) ^{47, 49, 53, 56}.

Radiation therapy was associated with increased risk for erectile dysfunction compared with watchful waiting in 6 cohort studies, with similar estimates across studies (median RR, 1.3 [range, 1.1 to 1.5]) (Appendix Table 4) ^{47, 49, 53, 54, 56, 58}. Rates of erectile dysfunction ranged from 26% to 68% (median, 50%) with watchful waiting; radiation therapy was associated with a median increase in pooled absolute risk of 14 percentage points (range, 7 to 22 percentage points).

Five of the six studies did not provide details about the type of radiation therapy (for example, EBRT vs. brachytherapy) or dosing regimen. One good-quality cohort study reported a 7.0% rate of urinary incontinence after high-dose brachytherapy (n=47), 5.4% after low-dose brachytherapy (n=58), and 2.7% after EBRT (n=123) ⁵⁶. Rates of erectile dysfunction were 72%, 36%, and 68%, respectively.

Consistent with the studies reporting dichotomous outcomes, 10 studies found radiation therapy to be associated with worse sexual function compared with watchful waiting on the basis of continuous scales, although no clear differences were seen in sexual bother scores and measures of urinary function (Appendix Table 4) ^{40, 43, 46, 48, 51, 53, 55-58}.

Quality of Life. Ten studies reported generic quality of life ^{40, 43, 46, 48, 50, 51, 53, 55, 56, 58}. Three studies found no differences between radiation therapy and watchful waiting in SF-36 physical (median difference, 0 points [range, −3 to 0 points]) or mental (median difference, 0 points [range, −2 to 1 point]) component summary scores (Appendix Table 4) ^{43, 56, 58}. Results favored watchful waiting on the physical role function subscale (7 studies; median difference, −9 points [range, −22 to 1 point]) ^{43, 46, 48, 51, 53, 55, 58}, with no clear differences on other SF-36 subscales.

Other Harms. Six cohort studies consistently found radiation therapy to be associated with worse Prostate Cancer Index bowel bother (median difference, −6 points [range, −10 to −2 points]) and function (median difference, −8 points [range, −15 to −3 points]) than watchful waiting ^{43, 48, 51, 53, 56, 58}. In studies that evaluated bowel function serially, effects seemed to be most pronounced in the first few months after radiation therapy and gradually improved ^{40, 46, 51, 57}. This might help explain the inconsistent results among studies that reported dichotomous outcomes. Although 1 study found radiation therapy to be associated with substantially increased risk for bowel urgency after 2 years (3.2% vs. 0.4%; RR, 7.5 [CI, 1.0 to 56]) ⁴⁷, 2 studies with longer duration of follow-up (5.6 ⁴⁹ and 3 years ⁵⁶) found no increased risk.

One cohort study reported similar effects of EBRT and brachytherapy on Prostate Cancer Index bowel function and bother ⁴³. Another study found low-dose brachytherapy to be associated with smaller effects on bowel bother (about 3-point change from baseline) compared with high-dose brachytherapy (9-point change) or EBRT (8-point change) ⁵⁶.

No study reported effects of radiation therapy versus watchful waiting on anxiety or depression.

The Table shows our summary of the evidence. Screening based on PSA identifies additional cases of prostate cancer, but most trials found no statistically significant effect on prostate cancer–specific mortality. Recent meta-analyses of randomized trials included in this review found no pooled effect of screening on prostate cancer–specific mortality ^69,70. However, the 2 largest and highest-quality trials reported conflicting results ^6,7. The ERSPC trial found PSA screening every 2 to 7 years to be associated with a 20% relative reduction in risk for death from prostate cancer in a prespecified subgroup of men aged 55 to 69 years ⁷, whereas the PLCO trial found no effect ⁶. High rates of previous PSA screening and contamination in the control group of the PLCO trial may have reduced its ability to detect benefits, although these factors do not explain the trend toward increased risk for prostate cancer–specific mortality in the screened group. The proportion of men in the PLCO trial who initially chose active surveillance or expectant management instead of curative treatment was lower than in the ERSPC trial (10% vs. 19%), and the PLCO trial evaluated a shorter screening interval (annual vs. every 2 to 7 years), suggesting that more conservative screening and treatment strategies might be more effective than more aggressive ones. Chance could also explain the apparent discrepancy between the 2 trials because the risk estimate CIs overlapped. Additional follow-up might help resolve the discrepancy, given the long lead time (10 to 15 years) that may be necessary to fully understand the effect of PSA-based screening.

Treatment studies can help inform screening decisions by providing information about potential benefits of interventions once prostate cancer is detected. However, only 1 good-quality randomized trial compared an active treatment for localized prostate cancer with watchful waiting ²³. It found that prostatectomy was associated with decreased risk for all-cause and prostate cancer–specific mortality after 15 years of follow-up, although benefits seemed to be limited to younger men on the basis of subgroup analyses. Because the RCT did not enroll men specifically with screening-detected prostate cancer, its applicability to screening is uncertain. Although cohort studies consistently found prostatectomy and radiation therapy to be associated with decreased risk for all-cause and prostate cancer–specific mortality compared with watchful waiting, estimates are susceptible to residual confounding, even after statistical adjustment.

Screening is associated with potential harms, including serious infections or urine retention in about 1 of 200 men who undergo prostate biopsy as a result of an abnormal screening result. False-positive screening results occurred in 12% to 13% of men randomly assigned to PSA-based screening ^65,66, with 1 trial reporting no prostate cancers in three quarters of screening-triggered biopsies ⁷. Screening also is likely to result in overdiagnosis because of the detection of low-risk cancers that would not have caused morbidity or death during a man's lifetime, and overtreatment of such cancers, which exposes men to unnecessary harms ⁷¹. Over three quarters of men with localized prostate cancer (about 90% of screening-detected cancers are localized) undergo prostatectomy or radiation therapy ^11,12. On the basis of data from the ERSPC trial, the rate of overdiagnosis with screening was estimated to be as high as 50% ⁷², and 48 men received treatment for every prostate cancer–specific death prevented ⁷. Treating approximately 3 men with prostatectomy or 7 with radiation therapy instead of watchful waiting would each result in 1 additional case of erectile dysfunction, and treating approximately 5 men with prostatectomy instead of watchful waiting would result in 1 additional case of urinary incontinence. Prostatectomy and radiation therapy were not associated with worse outcomes on most measures related to general health-related quality of life compared with watchful waiting, suggesting that negative effects related to specific harms may be offset by positive effects (perhaps related to less worry about untreated prostate cancer). Prostatectomy was also associated with perioperative (30-day) mortality (about 0.5%) and cardiovascular events (0.6% to 3%), and radiation therapy was associated with bowel dysfunction.

The evidence on treatment-related harms reviewed for this report seemed to be most applicable to open retropubic radical prostatectomy and EBRT, although details of specific surgical techniques or radiation therapy techniques and dosing regimens were frequently lacking. We found little evidence with which to evaluate newer techniques for prostatectomy (including nerve-sparing approaches that use laparoscopy, either robotic-assisted or freehand) compared with watchful waiting, but found no pattern suggesting that more recent studies reported different risk estimates than older studies. Limited data suggest that low-dose brachytherapy may be associated with fewer harms than high-dose brachytherapy or EBRT ⁵⁶. A potential harm of radiation therapy not addressed in this review is secondary posttreatment carcinogenic effects ^73,74.

Other treatments used for localized prostate cancer are reviewed in the full report, available on the USPSTF Web site ¹⁰. Although androgen deprivation is the next most commonly used therapy for localized prostate cancer after prostatectomy and radiation therapy ¹¹, it is comparatively uncommon, and is not recommended as primary therapy ^75,76 because of evidence suggesting ineffectiveness ³², as well as an association with important adverse events, such as coronary heart disease, myocardial infarction, diabetes, and fractures, when given for more advanced prostate cancer ^77-79.

Our study has limitations. We excluded non–English-language articles, which could result in language bias, although we identified no non–English-language studies that would have met inclusion criteria. We included cohort studies of treatments, which are more susceptible to bias and confounding than well-conducted randomized trials. However, confounding by indication may be less of an issue in studies that evaluate harms ⁸⁰, and analyses stratified by study design did not suggest differential estimates. If patients are selected for a specific prostate cancer treatment in part because of a lower perceived risk for harms, the likely effect on observational studies would be to underestimate risks. For mortality outcomes, which may be more susceptible to confounding by indication, we included only studies that performed statistical adjustment. Finally, studies did not distinguish well between active surveillance and watchful waiting. Active surveillance might be associated with more harms (due to repeated biopsies or subsequent interventions) than watchful waiting, and studies with well-described active surveillance interventions that are consistent with current definitions for this therapy are needed ¹⁴.

In summary, PSA-based screening is associated with detection of more prostate cancers; small to no reduction in prostate cancer–specific mortality after about 10 years; and harms related to false-positive test results, subsequent evaluation, and therapy, including overdiagnosis and overtreatment. If screening is effective, optimal screening intervals and PSA thresholds remain uncertain. The ERSPC trial evaluated longer screening intervals (2 to 7 years) and in some centers lower PSA thresholds (2.5 to 4.0 µg/L) as compared with typical U.S. practice ⁷. When available, results from the Prostate Cancer Intervention Versus Observation Trial, which compared prostatectomy with watchful waiting for screening-detected cancer, may help clarify which patients would benefit from prostatectomy or other active treatments, potentially reducing harms from unnecessary treatment ⁸¹.

Source: This article was first published in Annals of Internal Medicine (Ann Intern Med 2011 Oct 7. [Epub ahead of print]).

Acknowledgment: The authors thank Mary Barton, MD, MPP, Ned Calonge, MD, MPH, and U.S. Preventive Services Task Force Leads Michael LeFevre, MD, MSPH, Rosanne Leipzig, MD, PhD, and Timothy Wilt, MD, MPH, for their contributions to this report.

Grant Support: By the Agency for Healthcare Research and Quality (contract number HHSA-290-2007-10057-I-EPC3, Task Order No. 3), Rockville, Maryland.

Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M11-1085.

Requests for Single Reprints: Roger Chou, MD, Oregon Health & Science University, Mailcode BICC, 3181 SW Sam Jackson Park Road, Portland, OR 97239; e-mail, chour@ohsu.edu.

Current author addresses and author contributions are available at http://www.annals.org.

ERSPC = European Randomized Study of Screening for Prostate Cancer; HR = hazard ratio; KQ = key question; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; PSA = prostate-specific antigen; RCT = randomized, controlled trial; RR = relative risk; SF-36 = Short-Form 36.

Appendix Figure 1. Summary of Literature Search and Selection: Effectiveness and Harms of Screening

Appendix Figure 2. Summary of Literature Search and Selection: Effectiveness and Harms of Treatment

ARR = absolute risk reduction; DRE = digital rectal examination; ERSPC = European Randomized Study of Screening for Prostate Cancer; IRR = incidence rate ratio; NNS = number needed to screen; NNT = number needed to treat; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; PSA = prostate-specific antigen; RR = relative risk; TRUS = transrectal ultrasonography; USPSTF = U.S. Preventive Services Task Force.

ADT = androgen deprivation therapy; EBRT = external-beam radiation therapy; NA = not applicable; NR = not reported; PSA = prostate-specific antigen; RCT = randomized, controlled trial; SEER = Surveillance, Epidemiology, and End Results; WHO = World Health Organization.
* Definition unclear; results not abstracted.
† Conservatively managed patients included those who received ADT.
‡ Results from the hormone therapy group were not abstracted; 32% (57/179) were at stage T3 or higher at baseline

ADT = androgen deprivation therapy; EBRT = external-beam radiation therapy; HR = hazard ratio; NR = not reported; RCT = randomized, controlled trial; RR = relative risk.
* Duration varied by treatment group.

ADT = androgen deprivation therapy; NR = not reported; RCT = randomized, controlled trial; RR = relative risk.

ADT = androgen deprivation therapy; SF-36 = Short-Form 36; UCLA-PCI = University of California, Los Angeles, Prostate Cancer Index.