Kenneth Lin, MD; Robert Lipsitz, MD, MPH; Therese Miller, DrPH; Supriya Janakiraman, MD, MPH
Acknowledgment: The authors thank Caryn McManus at the Agency for Healthcare Research and Quality for her assistance with the literature searches.
Potential Financial Conflicts of Interest: None disclosed.
Requests for Single Reprints: Kenneth Lin, MD, Center for Primary Care, Prevention, and Clinical Partnerships, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; e-mail, email@example.com.
Current Author Addresses: Drs. Lin, Lipsitz, Miller, and Janakiraman: 540 Gaither Road, Rockville, MD 20850.
Lin K, Lipsitz R, Miller T, Janakiraman S. Benefits and Harms of Prostate-Specific Antigen Screening for Prostate Cancer: An Evidence Update for the U.S. Preventive Services Task Force. Ann Intern Med. 2008;149:192-199. doi: 10.7326/0003-4819-149-3-200808050-00009
Download citation file:
Published: Ann Intern Med. 2008;149(3):192-199.
Prostate cancer is the most common nonskin cancer in men in the United States, and prostate cancer screening has increased in recent years. In 2002, the U.S. Preventive Services Task Force concluded that evidence was insufficient to recommend for or against screening for prostate cancer with prostate-specific antigen (PSA) testing.
To examine new evidence on benefits and harms of screening asymptomatic men for prostate cancer with PSA.
English-language articles identified in PubMed and the Cochrane Library (search dates, January 2002 to July 2007), reference lists of retrieved articles, and expert suggestions.
Randomized, controlled trials and meta-analyses of PSA screening and cross-sectional and cohort studies of screening harms and of the natural history of screening-detected cancer were selected to answer the following questions: Does screening for prostate cancer with PSA, as a single-threshold test or as a function of multiple tests over time, decrease morbidity or mortality? What are the magnitude and nature of harms associated with prostate cancer screening, other than overtreatment? What is the natural history of PSA-detected, nonpalpable, localized prostate cancer?
Studies were reviewed, abstracted, and rated for quality by using predefined U.S. Preventive Services Task Force criteria.
No good-quality randomized, controlled trials of screening for prostate cancer have been completed. In 1 cross-sectional and 2 prospective cohort studies of fair to good quality, false-positive PSA screening results caused psychological adverse effects for up to 1 year after the test. The natural history of PSA-detected prostate cancer is poorly understood.
Few eligible studies were identified. Long-term adverse effects of false-positive PSA screening test results are unknown.
Prostate-specific antigen screening is associated with psychological harms, and its potential benefits remain uncertain.
Prostate cancer is the most common nonskin cancer in U.S. men. An estimated 218 890 men received a new diagnosis of prostate cancer in 2007, and 1 in 6 men will receive a diagnosis in their lifetime (1). The American Cancer Society estimates that 27 350 men died of prostate cancer in 2006 (2). After peaking in 1991 (29.4 deaths per 100 000 men), the prostate cancer mortality rate has gradually decreased. Although this positive trend may be related to increased screening for prostate cancer, other factors, including new treatment approaches, could also account for some or all of the observed decline in mortality (3).
The serum prostate-specific antigen (PSA) test was approved by the U.S. Food and Drug Administration in 1986, and its use for prostate cancer screening has increased substantially since the mid-1990s (4). However, PSA testing is not specific to prostate cancer; common conditions, such as benign prostatic hyperplasia and prostatitis, also increase PSA levels. Approximately 1.5 million U.S. men age 40 to 69 years have a PSA level greater than 4.0 μg/L, a widely used cutoff value for a positive screening result (5). Refinements designed to improve the PSA test's sensitivity and specificity for prostate cancer include determination of PSA density, PSA velocity, PSA doubling time, and percentage of free PSA (6–9).
Potential harms from PSA screening include additional medical visits, adverse effects of prostate biopsies, anxiety, and overdiagnosis (the identification of prostate cancer that would never have caused symptoms in the patient's lifetime, leading to unnecessary treatment and associated adverse effects). Much uncertainty surrounds which cases of prostate cancer require treatment and whether earlier detection leads to improvements in duration or quality of life. Two recent systematic reviews of the comparative effectiveness and harms of therapies for localized prostate cancer concluded that no single therapy is superior to all others in all situations (10, 11).
In 2002, the U.S. Preventive Services Task Force (USPSTF) found insufficient evidence to recommend for or against routine screening for prostate cancer. The USPSTF found good evidence that PSA screening can detect early-stage prostate cancer but found mixed and inconclusive evidence that screening and early detection improve health outcomes. Consequently, the USPSTF was unable to determine the balance between benefits and harms of periodic screening for prostate cancer.
The analytic framework that guided the previous USPSTF evidence review (Figure) (12) included 8 key questions about benefits and harms of prostate cancer screening and treatment. This evidence update focuses on critical gaps in the evidence that the Task Force identified in the previous review: the lack of good-quality studies linking screening to improved health outcomes; limited information about harms of screening; and a paucity of knowledge about the natural history of PSA-detected, nonpalpable, localized prostate cancer (the most common type of prostate cancer detected today). These evidence gaps produced 3 new key questions for this update:
1. Does screening for prostate cancer with PSA, as a single-threshold test or as a function of multiple tests over time, decrease morbidity or mortality?
2. What are the magnitude and nature of harms associated with prostate cancer screening, other than overtreatment?
3. What is the natural history of PSA-detected, nonpalpable, localized prostate cancer?
After consultation with USPSTF liaisons and content experts, we chose a broad definition of PSA screening that included evolving prognostic measures, such as PSA velocity and doubling time. However, a comparison of the performance characteristics of such measures with traditional single-threshold PSA testing is outside the scope of this review.
For evidence on health outcomes associated with PSA screening, we searched PubMed for English-language articles indexed between 1 January 2002 and 12 July 2007 by using combinations of the Medical Subject Heading (MeSH) terms and keywords prostate neoplasms, screening, prostate-specificantigen, earlydiagnosis, PSAvelocity, PSAdoublingtime, and prostatespecificantigendoubling.
For evidence on the harms of screening for prostate cancer, we searched PubMed for English-language articles indexed between 1 January 2002 and 12 July 2007 by using combinations of the MeSH terms and keywords prostate neoplasms; screening; false positive reactions; adverse effects; mass screening/adverse effects; mass screening/psychology; anxiety; quality of life; and health knowledge, attitudes, practice.
For evidence on the natural history of PSA-detected, nonpalpable, localized prostate cancer, we searched PubMed for English-language articles indexed between 1 January 2002 and 23 August 2007 by using combinations of the MeSH terms and keywords prostaticneoplasms, naturalhistory, epidemiology, diseaseprogression, survivalanalysis, watchfulwaiting, activesurveillance, populationsurveillance, expectantmanagement, and conservativemanagement.
We identified additional articles through a search of the Cochrane Library, recommendations of experts, and a hand search of reference lists from major review articles and studies.
Two reviewers independently reviewed the title lists, abstracts, and full articles by using predetermined inclusion and exclusion criteria. Articles selected by at least 1 reviewer advanced to the next stage of review.
For key question 1, eligible studies were randomized, controlled trials (RCTs), meta-analyses, and systematic reviews that compared screening with no screening (or usual care) in general primary care populations and reported morbidity or mortality outcomes. Although the 2002 USPSTF review (12) considered case–control studies and ecological data related to this key question, we excluded these study types from this part of the evidence update to avoid potential sources of confounding that are inherent in nonrandomized studies.
For key question 2, eligible studies were randomized or nonrandomized comparative studies that reported quantitative health or quality-of-life outcomes related to a false-positive screening result. We excluded studies that reported only harms resulting from prostate cancer treatment.
For key question 3, eligible studies were RCTs and cohort studies that reported health outcomes of patients with stage T1c (nonpalpable, localized, PSA-detected) prostate cancer who did not receive active treatment (including patients assigned to watchful waiting or active surveillance protocols). To ensure that we retrieved the most applicable information on natural history, we excluded studies that predominantly involved patients with non–PSA-detected cancer (defined as comprising ≥80% of the study population), were too small to draw reliable conclusions about health outcomes (defined as <50 patients in the watchful waiting or surveillance group), or did not provide separate data on patients with stage T1c prostate cancer.
For all citations that met the initial eligibility criteria, 2 reviewers reviewed the full articles and independently rated their quality by using previously published USPSTF criteria (13). Disagreements between reviewers regarding article inclusion and quality rating were resolved through a consensus process. We assessed the quality of RCTs and cohort studies on the following items: initial assembly and maintenance of comparable groups; absence of important differential loss to follow-up or overall high loss to follow-up; use of equal, valid, and reliable outcome measurements; clear definition of interventions; and appropriateness of outcomes. We evaluated systematic reviews and meta-analyses on the following items: comprehensiveness of sources considered, appropriateness of search strategy, standard appraisal of included studies, validity of conclusions, recency, and relevance. The Appendix Table describes more thoroughly the criteria and definitions for USPSTF quality ratings.
We synthesized the data qualitatively by key question in tabular and narrative formats. Data from the 2002 USPSTF review (12) relevant to key questions 1 and 2 are included to facilitate an overall assessment of the body of evidence. We did not perform quantitative synthesis because of the paucity and heterogeneity of included studies.
The general work of the USPSTF is supported by the Agency for Healthcare Research and Quality. This review did not receive specific funding.
We identified 390 potentially relevant articles on health outcomes associated with PSA screening, 421 potentially relevant articles on harms of prostate cancer screening, and 91 potentially relevant articles on the natural history of PSA-detected prostate cancer. Appendix Figures 1, 2, and 3 contain details of the stages of review and reasons for study exclusion. We obtained 68 articles for full-text review; 10 articles met inclusion criteria for this evidence update.
RCT = randomized, controlled trial.
Does screening for prostate cancer with PSA, as a single-threshold test or as a function of multiple tests over time, decrease morbidity or mortality?
No good- or fair-quality RCTs addressed this question. Two poor-quality RCTs with important flaws in design and analysis (Table 1) do not show a mortality benefit from PSA screening independently or in a meta-analysis. We identified no RCTs that measured health outcomes from PSA screening by means other than single-threshold tests.
In 2002, the USPSTF identified 1 poor-quality RCT of prostate cancer screening by Labrie and colleagues (14) that did not show a mortality benefit from screening when data were reanalyzed by using an intention-to-screen analysis. A 2004 publication described 3 additional years of follow-up of this study (15). By the end of 1999, 23.6% of the screening-invited group and 7.3% of the control group had actually received screening. Comparing all men who were screened with all men who were not, the authors calculated a relative risk (RR) for death from prostate cancer of 0.385 (95% CI, 0.207 to 0.714) in those who were screened. However, when they compared screening-invited men with noninvited men, they found no mortality difference between the 2 groups (RR, 1.085 [CI, 0.822 to 1.433]).
Labrie and colleagues (14) did not report information on the adequacy of randomization, the demographic composition of the 2 groups, or the characteristics of participants who crossed over from screening to no screening or vice versa. Moreover, they did not indicate whether the assessment of outcomes was blinded. Finally, the inappropriate analysis comparing screened with unscreened cohorts did not adjust for potential confounders.
A poor-quality quasi-RCT by Sandblom and colleagues (16) compared total mortality and prostate-cancer–specific mortality in 1494 men who received digital rectal examination and PSA screening with those in 7532 control participants. An intention-to-screen analysis found no statistical difference in total mortality or prostate-cancer–specific mortality between the 2 groups. Sandblom and colleagues did not report information on the comparability of the groups, crossovers from the control group, or how the cause of death was assigned. Also, the study was not adequately powered to detect a statistically significant difference in the outcomes of interest.
A Cochrane meta-analysis (17) combined these 2 studies by using an intention-to-screen analysis and found no difference in prostate cancer mortality between men invited to prostate cancer screening and control groups (RR, 1.01 [CI, 0.80 to 1.29]). The authors assessed both studies as having a high risk for bias because of the methodological problems discussed previously.
What are the magnitude and nature of harms associated with prostate cancer screening, other than overtreatment?
One cross-sectional and 2 prospective cohort studies of fair-to-good quality reported short- and long-term psychological harms from prostate cancer screening (Table 2). Although abnormal screening results did not affect summary measures of anxiety or health-related quality of life, men with false-positive PSA screening test results were more likely to worry specifically about prostate cancer, have a higher perceived risk for prostate cancer, and report problems with sexual function compared with control participants for up to 1 year after the test. In 1 study, 26% of men with false-positive screening results reported moderate-to-severe pain during the prostate biopsy; men with false-positive results were also more likely to undergo repeated PSA testing and additional biopsies.
In 2002, the USPSTF found little evidence on harms associated with prostate cancer screening. Digital rectal examination and prostate biopsy cause discomfort or pain in most men. However, in the initial screening round of an RCT, health-related quality-of-life measures were not negatively affected by false-positive screening test results (18).
Brindle and colleagues (19) administered standardized assessments of anxiety, depression, and mental health to 7344 men who received PSA testing. Of the 855 men with a PSA level greater than an age-specific or numerical threshold, 770 returned for a biopsy and then took the questionnaires again before receiving their biopsy results. Assessment scores did not change in patients with an elevated PSA level. Because some elevated PSA levels were true positive, this study was not able to specifically assess the psychological effect of a false-positive PSA result. It was not clear whether the measures used were sensitive enough to detect changes in mental health related to anxiety specific to prostate cancer. Finally, this study was limited by 2 potential sources of selection bias: Recruited patients were already enrolled in a randomized trial of PSA screening, and more than 20% of participants with abnormal PSA levels were not reevaluated.
Other studies have used prostate-specific measures of harms in addition to generic mental health or quality-of-life scores. Katz and colleagues (20) did a telephone survey of 2 groups of men approximately 2 months after PSA screening. After adjustment for baseline characteristics, men with false-positive screening results were statistically more likely than control participants to worry about getting prostate cancer, have a higher perceived 5-year risk for prostate cancer, and report at least moderate problems with sexual function. This study was limited by the potential of confounding through other sources of psychological differences between the 2 groups (for example, referral vs. primary care patient population) and a lower survey response rate in the control group.
McNaughton-Collins and colleagues (21) compared 167 men who had an abnormal screening result but a benign biopsy specimen with 233 men who had a normal PSA level (defined as <2.5 μg/L). After 6 weeks, 49% of men in the biopsy group reported thinking about prostate cancer “a lot” or “some of the time,” compared with 18% of the control group. In addition, 40% of the biopsy group worried “a lot” or “some of the time” about developing prostate cancer compared with 8% of the control group. A total of 26% of men experienced moderate-to-severe pain from the biopsy. For 25% of men, the most recent benign biopsy was their third biopsy or more. Statistically significant differences between the biopsy and control groups in anxiety related to prostate cancer and perceived prostate cancer risk persisted 6 months and 1 year later (22). After 1 year, more men in the biopsy group than in the control group had at least 1 additional PSA test (73% vs. 42%) and another biopsy (15% vs. 1%).
What is the natural history of PSA-detected, nonpalpable, localized prostate cancer?
Three fair-quality cohort studies with small-to-medium sample sizes, highly self-selected elderly patients, and high drop-out rates show that some men with PSA-detected, nonpalpable, localized (stage T1c) prostate cancer have good health outcomes up to 10 years after diagnosis (Table 3). We did not identify any population-based studies in which patients with stage T1c prostate cancer were followed longitudinally with no intervention in order to determine health outcomes resulting from the natural progression of disease.
The USPSTF did not directly examine this question in 2002. Recent studies have used PSA testing and biopsy-based monitoring protocols to identify groups of men with “favorable-risk” prostate cancer who were candidates for delayed treatment if subsequent testing showed biochemical or histologic evidence of disease progression. Triggers for treatment varied by study, and many men in the monitoring groups eventually opted for treatment without any objective signs of disease progression.
Hardie and colleagues (23) tested the feasibility of a surveillance protocol in 80 men (median age, 70.5 years) with localized prostate cancer (stage T1 to T2) who were referred to a single tertiary care center in the United Kingdom from 1993 to 2002. Delayed treatment was recommended on the basis of serial PSA level testing and life expectancy assessments. After a median of 42 months of follow-up, 64 men remained on surveillance, 11 had received delayed treatment, and 5 had died of causes other than prostate cancer. This study was limited by the self-selected nature of participants (representing only 10% of eligible patients during the study enrollment period) and the absence of a standardized PSA-based threshold (absolute value or rate of increase) for initiating treatment.
Roemeling and colleagues (24) studied 64 men (mean age, 68.4 years) who chose watchful waiting and were part of a larger cohort of 293 men with stage T1c or T2 prostate cancer who met favorable risk criteria. After a mean follow-up of 82.4 months (range, 23.8 to 119.9 months), 37 men were living and untreated, 19 had chosen treatment, and 8 had died of causes other than prostate cancer. The same authors examined health outcomes in 278 men (median age, 69.8 years) who chose an active surveillance protocol (25). After a median follow-up of 3.4 years (range, 1.2 to 6 years), 170 men remained on surveillance, 26 had died of causes other than prostate cancer, and 82 had chosen treatment. Both studies by Roemeling and colleagues were limited by having highly self-selected patient populations and high dropout rates.
We found inconclusive evidence from RCTs about the health benefits of screening for prostate cancer with PSA. Although we excluded nonrandomized studies of PSA screening, several case–control studies have been published since the 2002 USPSTF review. These studies, conducted in a variety of settings and populations, have yielded conflicting results about the relationship between PSA screening and prostate cancer–related morbidity and mortality (26–31).
In 2007, Aus and colleagues (32) reported interim results from the ERSPC (European Randomized Study of Screening for Prostate Cancer), an ongoing trial that is designed to detect a mortality difference in men randomly assigned to biennial PSA screening or usual care. In the 19 945 men in this subsection of the study, the authors observed a 49% reduction in the risk for metastatic prostate cancer in the screening group (24 cases compared with 47 cases in the control group) after 10 years of follow-up (32).
We determined that this study did not meet inclusion criteria but brought it to the attention of the USPSTF. The primary outcome, metastatic prostate cancer, is an uncertain surrogate for mortality because of high initial response rates to androgen deprivation therapy and competing causes of death. Also, the criterion used for testing to ascertain this outcome may have resulted in unequal attention to the 2 groups, thereby biasing the results. In the absence of symptoms, bone scans were obtained only in men with a PSA level greater than 20 μg/L. Because patients in the control group with prostate cancer had a higher mean PSA level (90.4 μg/L) than did patients in the screening group (19.8 μg/L), the reported difference in metastatic disease may have been exaggerated in favor of the screening group.
Although we found some evidence that false-positive PSA test results are associated with adverse psychological effects, we cannot determine from the existing studies the precise magnitude of psychological harms of prostate cancer screening. Because the populations studied have almost exclusively consisted of college-educated white men, these results may not be generalizable to men with less formal education, or to ethnic or racial minorities. No studies of the effects of false-positive PSA test results have included many black men, who have a higher risk for diagnosis of and death from prostate cancer. Studies excluded during our review that were performed in black patient samples involved strategies for increasing rates of prostate cancer screening in these patients, with the benefit of such screening being assumed.
Short-term monitoring studies of highly selected older men with PSA-detected, nonpalpable, localized prostate cancer do not suggest that delayed or no treatment leads to poor health outcomes. Larger, longer-term studies are urgently needed. A recent report of a population-based cohort of men with untreated, early-stage prostate cancer found a sharp decline in prostate cancer–specific survival after 15 years of follow-up (33). None of the participants in this study received their diagnosis through screening, and most cancer cases were detected when clinically palpable. However, these results, in addition to those from a retrospective cohort study that did not find an increase in prostate cancer mortality rates over a similar period (34), suggest that decades of follow-up may be required to determine the safety and effectiveness of current monitoring protocols.
Although we did not examine new evidence on the harms of treatment in this focused update, the 2002 USPSTF review found that prostate cancer treatments cause clinically significant harms, including erectile dysfunction and urinary incontinence, in many patients (12). Still, many physicians continue to believe that the benefits of immediately treating PSA-detected prostate cancer outweigh the risks of delayed or no treatment. In this context, a study that was excluded from this review merits mention.
In 2005, Bill-Axelson and colleagues (35) reported the results of a trial of 695 men with localized prostate cancer who were randomly assigned to receive radical prostatectomy or watchful waiting. The study did not meet inclusion criteria because only 5.2% of the population had prostate cancer diagnosed through screening and 77.8% of the treatment group had stage T2 (palpable) cancer. After a median of 8.2 years, 14.4% of men in the control group and 8.6% of men in the treatment group had died of prostate cancer. An analysis of prostate cancer–specific mortality stratified by age and intervention suggested that the men younger than 65 years were much more likely to benefit from radical prostatectomy than men 65 years of age or older. In the latter subgroup, the cumulative incidence of death from prostate cancer after 10 years was comparable in the watchful waiting and prostatectomy groups.
Two large RCTs of PSA screening are currently under way. The ERSPC randomly assigned 190 000 men between ages 50 and 75 years to screening with PSA, digital rectal examination, and transrectal ultrasonography or usual care; the intervention was later changed to PSA screening alone (36). Biopsies were performed in patients with PSA levels greater than 3.0 μg/L, and positive biopsy results led to treatment outlined by a standardized protocol. The prostate component of the U.S. National Cancer Institute's Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial is evaluating the effect of annual screening with PSA and digital rectal examination on prostate cancer–specific mortality in 76 705 men (37). Abnormal results were provided to the patient's primary care physician of record, and further diagnostic work-up and treatment were based on individual patient and physician preferences.
These trials may provide valuable and complementary information about the health outcomes associated with PSA screening in the general primary care population. Even if 1 or both ultimately demonstrates a population-level mortality benefit, however, individual screening decisions will still need to be made by weighing the benefits and harms of prostate cancer screening and treatment summarized in the previous USPSTF review (12) and this focused evidence update.
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only