Prostate-Specific Antigen–Based Screening for Prostate Cancer: Evidence Report and Systematic Review for the US Preventive Services Task Force

Importance  Prostate cancer is the second leading cause of cancer death among US men.

Objective  To systematically review evidence on prostate-specific antigen (PSA)–based prostate cancer screening, treatments for localized prostate cancer, and prebiopsy risk calculators to inform the US Preventive Services Task Force.

Data Sources  Searches of PubMed, EMBASE, Web of Science, and Cochrane Registries and Databases from July 1, 2011, through July 15, 2017, with a surveillance search on February 1, 2018.

Study Selection  English-language reports of randomized clinical trials (RCTs) of screening; cohort studies reporting harms; RCTs and cohort studies of active localized cancer treatments vs conservative approaches (eg, active surveillance, watchful waiting); external validations of prebiopsy risk calculators to identify aggressive cancers.

Data Extraction and Synthesis  One investigator abstracted data; a second checked accuracy. Two investigators independently rated study quality.

Main Outcomes and Measures  Prostate cancer and all-cause mortality; false-positive screening results, biopsy complications, overdiagnosis; adverse effects of active treatments. Random-effects meta-analyses were conducted for treatment harms.

Results  Sixty-three studies in 104 publications were included (N = 1 904 950). Randomization to PSA screening was not associated with reduced risk of prostate cancer mortality in either a US trial with substantial control group contamination (n = 76 683) or a UK trial with low adherence to a single PSA screen (n = 408 825) but was associated with significantly reduced prostate cancer mortality in a European trial (n = 162 243; relative risk [RR], 0.79 [95% CI, 0.69-0.91]; absolute risk reduction, 1.1 deaths per 10 000 person-years [95% CI, 0.5-1.8]). Of 61 604 men screened in the European trial, 17.8% received false-positive results. In 3 cohorts (n = 15 136), complications requiring hospitalization occurred in 0.5% to 1.6% of men undergoing biopsy after abnormal screening findings. Overdiagnosis was estimated to occur in 20.7% to 50.4% of screen-detected cancers. In an RCT of men with screen-detected prostate cancer (n = 1643), neither radical prostatectomy (hazard ratio [HR], 0.63 [95% CI, 0.21-1.93]) nor radiation therapy (HR, 0.51 [95% CI, 0.15-1.69]) were associated with significantly reduced prostate cancer mortality vs active monitoring, although each was associated with significantly lower risk of metastatic disease. Relative to conservative management, radical prostatectomy was associated with increased risk of urinary incontinence (pooled RR, 2.27 [95% CI, 1.82-2.84]; 3 trials; n = 1796) and erectile dysfunction (pooled RR, 1.82 [95% CI, 1.62-2.04]; 2 trials; n = 883). Relative to conservative management (8 cohort studies; n = 3066), radiation therapy was associated with increased risk of erectile dysfunction (pooled RR, 1.31 [95% CI, 1.20-1.42]).

Conclusions and Relevance  PSA screening may reduce prostate cancer mortality risk but is associated with false-positive results, biopsy complications, and overdiagnosis. Compared with conservative approaches, active treatments for screen-detected prostate cancer have unclear effects on long-term survival but are associated with sexual and urinary difficulties.

Prostate cancer is the most commonly diagnosed cancer in US men and the second leading cause of cancer death.¹ It has been estimated that in 2018, approximately 165 000 US men will be diagnosed with prostate cancer and 29 000 men will die of prostate cancer.² Prostate cancer incidence is 74% greater among African American than white men¹ and is also relatively greater in men with vs men without a family history of prostate cancer.³

US prostate cancer incidence increased sharply with the dissemination of prostate-specific antigen (PSA)–based screening beginning in the late 1980s.⁴ In 2012, the US Preventive Services Task Force (USPSTF) recommended against PSA-based screening for prostate cancer, concluding that there was moderate certainty that the benefits of screening do not outweigh the harms (D recommendation). New evidence has since emerged from screening and treatment trials, and among US men prostate cancer is increasingly managed with active surveillance, in which treatment is deferred indefinitely unless evidence of progression is found during periodic physical examination, PSA–based testing, or repeat biopsy. This systematic review of screening and treatment benefits and harms and whether prebiopsy risk calculators can reliably detect higher-risk prostate cancers, along with a review of decision modeling studies,⁵ was conducted to inform the USPSTF in its update of the 2012 recommendation regarding PSA-based screening for prostate cancer.

This review addressed 5 key questions (KQs) (Figure 1) encompassing the benefits and harms of PSA screening (KQ1 and KQ2), benefits and harms of treatments for localized prostate cancer (KQ3 and KQ4), and the utility of prebiopsy risk calculators to identify men with higher-risk prostate cancers (KQ5). Results addressing primary key questions are summarized here, while results for subquestions 2a, 3a, 4a, 4b, and 5a, as well as additional methodological details regarding search strategies, study inclusion criteria, quality assessment, excluded studies, and data analyses, are publicly available at https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/prostate-cancer-screening1.

MEDLINE, PubMed, EMBASE, and the Cochrane Central Register of Controlled Trials were searched to locate studies informing the key questions (eMethods in the Supplement) that were published since the end of the search periods for the 2011 USPSTF reviews^7,8 (July 1, 2011, through July 15, 2017). Database searches were supplemented with expert suggestions and by reviewing reference lists from relevant systematic reviews and prior USPSTF reports. KQ5, on prebiopsy risk calculators, was new to this review; a preliminary search revealed that the earliest article in this field was published in 2006, so databases were searched from January 1, 2006, through October 6, 2016, for KQ5. Since October 2016, ongoing surveillance continued through article alerts and targeted searches of high-impact journals to identify studies published that could affect the conclusions or the related USPSTF recommendation. The last surveillance search was conducted in February 2018; during ongoing surveillance, we identified extended follow-up from a treatment trial,⁹ 2 cohort studies reporting longitudinal treatment outcomes,^10,11 and 3 studies of multivariable risk calculators.^12-14 We also included a recently published large screening trial.¹⁵

For KQ1, randomized clinical trials (RCTs) of asymptomatic men undergoing PSA screening vs no screening were included that assessed cancer incidence, prostate cancer–related morbidity, prostate cancer–specific mortality, or all-cause mortality. For KQ2, RCTs and cohort studies of asymptomatic men undergoing PSA screening or prostate biopsy after abnormal screening results were included that assessed the frequency of false-positive PSA screening, physical or psychological harms of screening or biopsy, or health-related quality of life. As has been performed for breast and lung cancer screening,^16-19 extra-incidence data from trials were used to estimate the percentage of men diagnosed with cancer in the screening groups who were overdiagnosed. In the absence of overdiagnosis, the number of cases in the control group of a screening trial would be expected to eventually catch up with the number of cases diagnosed in the screening group; therefore, an excess of cases in the screening group with extended follow-up implies overdiagnosis.

For KQ3 and KQ4, RCTs and comparative cohort studies of men with localized prostate cancer (stages T1-T2) were included that reported prostate cancer–specific morbidity or mortality, all-cause mortality (KQ3), or physical or psychological harms of treatment, including adverse quality-of-life effects (KQ4). Studies were required to compare outcomes among men receiving active treatments (including radical prostatectomy, radiation therapy, androgen deprivation therapy, cryotherapy, or high-intensity focused ultrasound) and men receiving conservative management (ie, watchful waiting, active surveillance, observation, or no treatment). Because few studies were found for treatments other than radical prostatectomy and radiation therapy, only findings for radical prostatectomy and radiation therapy are summarized in this article; evidence on other treatment modalities is reviewed in the full evidence report. For KQ4 (treatment harms), uncontrolled observational studies with sample sizes of at least 100 men were also included.

For KQ5, external validation studies of multivariable risk calculators were included if they predicted the presence of “significant” prostate cancer using PSA testing in addition to patient variables routinely available before prostate biopsy (eg, patient characteristics, rectal examination results, PSA level). Significant prostate cancers were defined as either high grade (Gleason score ≥7) or clinical stage T2b or higher. Studies of novel serum biomarkers or imaging studies were excluded because results from these studies would not be routinely available before biopsy in most urology practices.

Two reviewers independently appraised the quality of included articles using predefined criteria,^6,20-23 with disagreements resolved by consensus or consultation with a third investigator. One reviewer extracted study-level data into standardized evidence tables; a second checked data accuracy. Included studies were limited to those published in English and were rated as fair or good quality using USPSTF quality rating standards (eTable 1 in the Supplement). Poor-quality studies contained a fatal flaw or multiple significant limitations that could invalidate the results. Three poor-quality RCTs of screening PSA were excluded.^24-26 Limitations of these studies included inadequate statistical power,^24-26 faulty analysis,²⁵ and potentially biased outcomes assessment.^24-26

For each key question, the study designs, population characteristics, screening and treatment details, and overall results were summarized using descriptive statistics. Pooled meta-analyses were not performed for outcomes of screening effectiveness (KQ1), screening harms (KQ2), or treatment effectiveness (KQ3), because data on these outcomes derived from few studies with variable populations and interventions. For KQ4, random-effects meta-analyses were performed using the method of DerSimonian and Laird to estimate pooled relative risks (RRs) of urinary incontinence (ie, daily use of a pad or worse) and erectile dysfunction (ie, erections insufficient for intercourse) among patients receiving radical prostatectomy or radiation therapy vs conservative management.

Statistical heterogeneity across studies was estimated using the I² statistic. Meta-analyses were performed using Stata version 14.2 (StataCorp). Pooled RRs were used to estimate the number of patients needed to be treated for 1 patient to be harmed (number needed to harm); in these calculations, the absolute risk of the harm in actively treated patients was estimated as the product of the pooled RR and the absolute risk among all men included in the conservative management control groups of included studies. Statistical tests were 2-sided, with P < .05 indicating statistical significance.

Two reviewers independently assessed 4105 unique citations and 303 full-text articles for inclusion (Figure 2). Overall, 24 articles (3 studies) were included for KQ1, 15 articles (8 studies) were included for KQ2, 23 articles (13 studies) were included for KQ3, 38 articles (31 studies) were included for KQ4, and 14 articles (14 studies) were included for KQ5.

Key Question 1. Is there direct evidence that PSA-based screening for prostate cancer reduces short- or long-term prostate cancer morbidity and mortality and all-cause mortality?

Key Question 1a. Does the effectiveness of PSA-based screening vary by subpopulation or risk factor (eg, age, race/ethnicity, family history, or clinical risk assessment)?

Three fair-quality RCTs (n = 647 906) assessed the effect of PSA screening on prostate cancer morbidity and mortality and all-cause mortality: the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial,²⁷ the European Randomized Study of Screening for Prostate Cancer (ERSPC),²⁸ and the Cluster Randomized Trial of PSA Testing for Prostate Cancer (CAP)¹⁵ (Table 1). Since the previous USPSTF review, the ERSPC³⁰ and PLCO²⁹ trials each reported results of extended follow-up; recent reports from 4 ERSPC sites (Goteborg-Sweden,³² Finland,³⁷ the Netherlands,³⁵ and Spain³⁴) were also included.

In the PLCO trial, 76 683 US men aged 55 to 74 years were recruited from 1993 to 2001 and randomized to either annual PSA screening for 6 years or usual care. When men had PSA levels of 4.0 ng/mL or greater, they and their clinicians were informed; community physicians coordinated diagnostic evaluations and treatments. The PLCO trial has been characterized as comparing the effectiveness of organized vs opportunistic screening,⁴⁰ because during the screening phase of the trial, approximately 46% of control group participants received routine screening PSA testing from community physicians during each year, compared with approximately 85% of men in the screening group.⁴¹ PSA testing was also common in both trial groups during the 7 years after the screening phase.

In the ERSPC trial, 181 999 men aged 50 to 74 years were enrolled at 7 European centers from 1993 to 2003, although primary outcomes were reported among 162 243 men aged 55 to 69 years at randomization. The ERSPC screening interval was 4 years at all sites except in Sweden, which used a 2-year interval; the PSA threshold prompting biopsy recommendation was most commonly 3 ng/mL but varied across sites and over time within some sites. Contamination was not systematically assessed at all ERSPC sites, although 19.4% of control group participants had a screening PSA test during the screening phase at the Netherlands site.³⁶

In the CAP trial, 408 825 UK men aged 50 to 69 years were randomized by primary care practice cluster (n = 573) to invitation (or no invitation) to a single PSA screen conducted from 2001 to 2009.¹⁵ After invitation, 34% of men obtained a valid PSA screening test. Although PSA screening was not routine in UK practice during the study period, from 2002 to 2011, UK men aged 45 to 69 years had a 39.2% cumulative risk over 10 years of receiving at least 1 PSA test for screening or diagnostic purposes.⁴²

Cumulative incidences of prostate cancer in the screening and control groups were 11.1% and 9.9%, respectively, at 13 years of median follow-up in the PLCO trial (RR, 1.12 [95% CI, 1.07-1.17]), 10.2% and 6.0% at 13 years of median follow-up in the ERSPC trial (RR, 1.57 [95% CI, 1.51-1.62]), and 4.3% and 3.6% at 10 years of median follow-up in the CAP trial (RR, 1.19 [95% CI, 1.14-1.25]) (Table 1). Observed risk differences in prostate cancer incidence indicate a number needed to invite of 84 men in the PLCO trial (95% CI, 59-144), 26 men in the ERSPC trial (95% CI, 24-29), and 154 men in the CAP trial (95% CI, 128-192) for 1 additional man to be diagnosed with prostate cancer.^15,30,39

Within 4 ERSPC sites, the cumulative incidence of metastatic cancer (ie, metastases on imaging or a PSA level >100 ng/mL) at a median follow-up of 12.0 years was lower among men randomized to screening compared with control (RR, 0.70 [95% CI, 0.60 to 0.82]), and randomization to screening was associated with an absolute reduction in long-term risk of metastatic prostate cancer of 3.1 cases per 1000 men.⁴³ After accounting for the 55.6% relative increase in prostate cancer incidence in the screening vs control groups at these sites, the number needed to invite to avoid 1 case of metastatic cancer was 323 (95% CI, 227-556), and the number needed to diagnose with prostate cancer through screening to avert 1 case of metastatic cancer was 12 (95% CI, 9-21).⁴³

At a median follow-up of 14.8 years in the PLCO trial, the prostate cancer–specific mortality rate was 4.8 per 10 000 person-years among men in the intervention group and 4.6 per 10 000 person-years among men in the control group (RR, 1.04 [95% CI, 0.87-1.24]) (Table 1).²⁹ In the CAP trial, at a median follow-up of 10 years, prostate cancer–specific mortality was 3.0 per 10 000 person-years among men invited to screening and 3.1 per 10 000 person-years among men in the control group (RR, 0.96 [95% CI, 0.85-1.08]). Among men in the ERSPC core age group after a median of 13 years of follow-up, prostate cancer–specific mortality was 4.3 per 10 000 person-years in the intervention group and 5.4 per 10 000 person-years in control group (RR, 0.79 [95% CI, 0.69-0.91]; P = .001).³⁰ In the ERSPC trial, the absolute risk reduction in prostate cancer mortality associated with screening was 1.1 deaths per 10 000 person-years (95% CI, 0.5-1.8), or 1.3 fewer prostate cancer deaths per 1000 men; the number needed to invite to prevent 1 prostate cancer death was 781 (95% CI, 490-1929), and the number needed to diagnose was 27 (95% CI, 17-66). Across all ERSPC sites, prostate cancer mortality was statistically significantly reduced only at the sites in the Netherlands and Sweden, although point estimates were in favor of screening at all sites except Switzerland.

Randomization to screening was not associated with statistically significant reductions in prostate cancer mortality among men aged 65 to 74 years at baseline in the PLCO trial (RR, 1.02 [95% CI, 0.77-1.37])²⁹ or among men aged 70 to 74 years at baseline in the ERSPC trial (RR, 1.17 [95% CI, 0.82-1.66]).³⁰ No reports on differences in outcomes by race/ethnicity within the PLCO or ERSPC trials were found, and the sample size of 3370 non-Hispanic black men randomized in the PLCO trial (4.4% of the overall population) was relatively small. Among white men in the PLCO trial who reported of a family history of prostate cancer (n = 4833), the multivariable-adjusted hazard ratio for prostate cancer death among men randomized to screening was lower relative to control participants but was not statistically significant (HR, 0.49 [95% CI, 0.22-1.10]).⁴⁴

After median follow-up periods ranging from 10 years in the CAP trial to 14.8 years in the PLCO trial, randomization to screening (relative to control) was not associated with statistically significantly reduced all-cause mortality in any of the 3 trials (CAP trial: RR, 0.99 [95% CI, 0.94-1.03]; ERSPC trial: RR, 1.00 [95% CI, 0.98-1.02]; PLCO trial: RR, 0.98 [95% CI, 0.95-1.00]) (Table 1).^15,29,30

Key Question 2. What are the harms of PSA-based screening for prostate cancer and diagnostic follow-up?

Evidence on the harms of PSA screening and diagnostic follow-up derives from the PLCO, CAP, and ERSPC trials (n = 647 906), as well as 2 good-quality and 3 fair-quality cohort studies (n = 297 971).^{15,27,29,30,45-55}

Among men who underwent at least 1 PSA screen during the initial 4 (of 6) PLCO screening rounds (n = 32 576), 10.4% received at least 1 false-positive PSA screening result,⁵⁶ compared with 17.8% of men who were screened at least once within 5 ERSPC centers (n = 61 604).⁵⁰ Across all PLCO screening rounds, 12.6% of men randomized to screening underwent 1 or more biopsies, resulting in a total of 6295 biopsies (16.4 biopsies per 100 men randomized to screening). Within ERSPC, the rate of biopsies among men randomized to screening was higher (27.7 biopsies per 100 men). Of biopsies performed in the 3 trials, 67.7%, 75.8%, and 60.6% did not result in a prostate cancer diagnosis in the PLCO, ERSPC, and CAP trials, respectively.^15,30,52

Among men undergoing a single round of PSA screening within the Veterans Affairs health system (n = 295 645), 8.5% had a PSA level of 4 ng/mL or greater; of these, 32.9% underwent subsequent prostate biopsy.⁵⁵ Of biopsies performed, 37.2% did not result in a prostate cancer diagnosis.

Among men who underwent biopsy in the PLCO trial (n = 4861), 2.0% experienced complications consisting predominately of infection, bleeding, or urinary difficulties, while 5.6% of 8313 men undergoing biopsy experienced similar complications in the Veterans Affairs cohort.⁵⁵ Among men undergoing biopsy after abnormal PSA screening results, postbiopsy hospitalization ranged from 0.5% to 1.6% in 1 trial and 2 cohorts.^53,55,57 Within approximately 1 month of biopsy within a UK cohort (n = 1147),⁵³ 7.3% of men experienced moderate or serious pain and 5.5% experienced moderate or severe fevers. Prostate biopsy has not been associated with statistically significantly increased risk of death.^47,52,53 Within 3 cohorts (n = 1179),^46,48,49,51 men who had abnormal PSA screening results followed by normal biopsy results had increased prostate cancer–specific worry up to 1 year after biopsy but no increase in depression or trait anxiety.

The percentage of detected cancers that were overdiagnosed was estimated using the excess incidence method applied to screening trials data (eTable 2 in the Supplement).¹⁸ When overdiagnosis was estimated as a percentage of all prostate cancers diagnosed, 16.4% of prostate cancers were overdiagnosed in the PLCO trial, 33.2% in the ERSPC trial, and 40.7% in the CAP trial. When estimated as a percentage of cancers detected by screening during the 2 trials reporting such data, 20.7% of cancers were overdiagnosed in the PLCO trial and 50.4% in the ERSPC trial. The extent of overdiagnosis varied across ERSPC sites, with higher rates of overdiagnosis occurring at 2 sites that also reported statistically significantly reduced prostate cancer mortality with screening.^33,35

Key Question 3. Is there evidence that various treatment approaches for early-stage or screen-detected prostate cancer reduce morbidity and mortality?

Three RCTs (n = 2524)^9,58,59 and 7 cohort studies (n = 64 965)^11,60-65 assessed morbidity and mortality with radical prostatectomy compared with conservative management, and 1 RCT (n = 1090)⁵⁸ and 7 cohort studies (n = 60 585)^11,60-65 compared radiation therapy with conservative management (Table 2). Conservative management in these studies consisted of observation without treatment, active surveillance, watchful waiting, or a combination of these approaches. Comparative studies of benefits and harms associated with other treatment modalities are summarized in the full evidence report.

The ProtecT trial randomized men with localized, screen-detected prostate cancer to radical prostatectomy, radiation therapy (with neoadjuvant androgen deprivation therapy), or active surveillance (consisting of periodic PSA monitoring without surveillance biopsy).⁵⁸ During a median follow-up of 10 years, 54.8% of men randomized to active surveillance received treatment, mostly commonly radical prostatectomy; prostate cancer survival was approximately 99% in each group, with no statistically significant differences in prostate cancer mortality or all-cause mortality.

In the Scandinavian Prostate Cancer Group Trial-4 (SPCG-4), which included men with clinically diagnosed, predominately palpable cancers, radical prostatectomy was associated with statistically significantly reduced prostate cancer mortality (RR, 0.56 [95% CI, 0.41-0.77]) and all-cause mortality (RR, 0.71 [95% CI, 0.59-0.86]), compared with watchful waiting.⁵⁹ In the US Prostate Cancer Intervention vs Observation Trial (PIVOT), in which 50% of men had screen-detected rather than clinically detected cancers, radical prostatectomy did not significantly reduce either prostate cancer mortality (RR, 0.65 [95% CI, 0.41-1.03]) or all-cause mortality (RR, 0.92 [95% CI, 0.82-1.02]), compared with conservative management.⁹

In the ProtecT trial, incidence of metastatic disease among men randomized to active surveillance (6.3 cases per 1000 person-years) was higher than among men randomized to radical prostatectomy or radiation therapy (2.4 and 3.0 cases per 1000 person-years; P = .004 for overall difference across groups). Approximately 27 (95% CI, 16-77) and 32 (95% CI, 18-143) men with screen-detected localized cancer would need to be treated with radical prostatectomy and radiation therapy (rather than active surveillance), respectively, to prevent 1 man from progressing to metastatic disease within 10 years. In both the SPCG-4 and PIVOT trials, radical prostatectomy was also associated with reduced progression to metastatic or systemic disease (SPCG-4: RR, 0.57 [95% CI, 0.44-0.75]; PIVOT: HR, 0.64 [95% CI, 0.42-0.97]).^9,59

Compared with conservative management, radical prostatectomy and radiation therapy were associated with reduced prostate cancer mortality and all-cause mortality in several fair- to good-quality cohort studies (Table 2).^60-65 However, cohort study estimates often diverged from RCT estimates, and cohort studies of treatments may be susceptible to residual confounding.

Key Question 4. What are the harms of the various treatment approaches for early-stage or screen-detected prostate cancer?

Three RCTs (n = 2524)^66-68 and 11 cohort studies (n = 8809)^10,11,69-78 compared harms among men treated with radical prostatectomy compared with conservative management, and 6 uncontrolled studies of surgical harms related to radical prostatectomy were included.^79-84 Two RCTs (n = 1198)^66,67,85 and 12 cohort studies (n = 4762)^{10,11,69-78,86} compared harms among men treated with radiation therapy compared with conservative management, and 1 uncontrolled observational study of radiation therapy was included (n = 3180).⁸⁷

Both urinary incontinence and erectile dysfunction occurred more commonly in men who underwent radical prostatectomy than in men receiving conservative management (Figure 3 and Figure 4). Pooled relative risks of incontinence with radical prostatectomy were 2.27 (95% CI, 1.82-2.84; I² = 0.0%) in 3 RCTs and 2.75 (95% CI, 1.78-4.23; I² = 63.0%) in 6 cohort studies. Based on risk differences estimated from RCT data, 7.9 men (95% CI, 5.4-12.2) would need to be treated with radical prostatectomy rather than conservative management for 1 additional man to develop incontinence.

In pooled meta-analyses of the 3 RCTs, there was marked statistical heterogeneity in the relative risk of erectile dysfunction with radical prostatectomy (I² = 87.5%), which was attributable to a disparate outcome from the ProtecT trial, in which many men randomized to active surveillance received radical treatments during the comparatively long 6 years of median follow-up for harms (Figure 4).⁶⁶ After excluding the ProtecT trial from the RCT meta-analysis, the pooled relative risk of erectile dysfunction associated with radical prostatectomy vs conservative management was 1.82 (95% CI, 1.62-2.04; I² = 0.0%). Based on estimated risk differences from this pooled analysis, 2.7 men (95% CI, 2.2-3.6) need to be treated with radical prostatectomy rather than conservative management for 1 additional man to develop erectile dysfunction.

In 5 longitudinal studies,^9-11,66,77 mean urinary and sexual function scores decreased to a nadir in the initial year after radical prostatectomy with some longer-term improvement, although urinary and sexual function typically remained statistically significantly lower than baseline up to 6 years after the procedure.

Data on surgical complications and perioperative mortality associated with radical prostatectomy are summarized in eTable 3 in the Supplement. In a cohort of US men undergoing radical prostatectomy, 1.7% experienced major medical complications (most commonly cardiac or pulmonary) and 5.3% experienced major surgical complications (requiring reintervention) within 30 days of surgery.⁸² Across 2 trials and 6 cohort studies, the median perioperative mortality with radical prostatectomy was 0.29% (range, 0.0%-0.52%).

There was marked variability across 8 studies comparing incidence of urinary incontinence with radiation and conservative management, so these were not meta-analyzed (Figure 5). In the ProtecT trial, the prevalence of erectile dysfunction at 6 years of median follow-up was similar among men randomized to radiation therapy vs active surveillance (39.8% vs 36.2%; RR, 0.91 [95% CI, 0.77-1.08]), likely attributable to the substantial crossover to radical treatment of men randomized to active surveillance.⁶⁶ In contrast, in 8 cohort studies, the prevalence of erectile dysfunction was more common in men treated with radiation therapy compared with men receiving conservative management (pooled RR, 1.31 [95% CI, 1.20-1.42]; I² = 22.1%) (Figure 6).^{10,11,69,73,75-77,86} Based on the cohort data, 6.9 men (95% CI, 5.1-10.7) would need to be treated with radiation therapy rather than conservative management for 1 additional man to develop erectile dysfunction. Although 2 US studies observed longitudinal improvement in adverse effects of radiation therapy on sexual function,^10,11 initial decrements in sexual function associated with radiation therapy persisted throughout a 3-year follow-up period in an Australian cohort.⁷⁷

In the ProtecT trial, 10.4% of men randomized to radiation therapy experienced bothersome bowel symptoms (eg, loose stools, fecal incontinence) at 6-month follow-up, which tended to diminish during the 6-year follow-up.⁶⁶ Two US cohort studies suggest a similar pattern,^10,11 in contrast to an Australian cohort study in which adverse bowel effects of radiation therapy persisted during 3 years of follow-up.⁷⁷

In 3 trials comparing radical prostatectomy with conservative management,^9,66,88 randomization to radical prostatectomy was not associated with differences in anxiety, depression, or physical or mental health status. Cohort studies^{11,70-72,74,75,77,78} also found no evidence of an adverse effect of radical prostatectomy on generic quality-of-life measures compared with conservative management. Similarly, 2 trials^66,67 and 9 cohort studies^{11,70-73,75,77,78,86} suggest no substantive adverse effect of radiation therapy on general measures of physical, mental, or global health status compared with conservative management.

Key Question 5. Is there evidence that use of a prebiopsy prostate cancer risk calculator, in combination with PSA-based screening, accurately identifies men with clinically significant prostate cancer compared with PSA-based screening alone?

Fourteen articles (n = 48 234)^12-14,89-99 provided evidence on discrimination or calibration of risk assessment tools for prediction of clinically significant prostate cancer (either Gleason score ≥7 or Stage 2b or higher) among men undergoing prostate biopsy (Table 3). All were external validation studies of either the Prostate Cancer Prevention Trial (PCPT) risk calculator or the ERSPC risk calculator. Most cohorts consisted of men referred to tertiary or academic centers and typically included both symptomatic men and asymptomatic men with abnormal screening results. No RCTs evaluated the clinical effect of risk calculator use on patient outcomes.

Both calculators usually discriminated better than PSA alone between men with and without clinically significant prostate cancer (median area under the curve [AUC] with PCPT calculator, 0.72 [range across 21 cohorts, 0.51-0.88]; median AUC with ERSPC calculator, 0.74 [range across 7 cohorts, 0.69-0.78]; median AUC with PSA alone, 0.68 [range across 16 cohorts, 0.59-0.82]). However, calibration was mixed with underestimation or overestimation of actual risk in several cohorts for each calculator.

Table 3 summarizes the evidence contained in this review. Direct evidence from 3 fair-quality trials demonstrates that PSA screening increases prostate cancer detection, especially of localized, less aggressive cancers,^30,39 and evidence from 4 ERSPC sites suggests that screening can reduce the long-term incidence of metastatic disease.⁴³ While the PLCO and CAP trials found no association between randomization to screening invitation and reduced prostate cancer mortality, the overall ERSPC trial found a relative risk of 0.79 for prostate cancer mortality with screening.³⁰ Differences in trial outcomes may have been attributable to greater baseline exposure to PSA testing and contamination in the PLCO trial, low adherence to invitation to a single PSA screen in the CAP trial, and higher adherence to screening and biopsy in the ERSPC trial. A recent analysis of individual patient data from both the PLCO and the ERSPC trials suggests that results from both trials are consistent with approximately 25% to 30% relative reduction in the risk of prostate cancer death with screening vs no screening after accounting for differences in baseline risk, screening adherence, contamination, and the intensity of postscreening diagnostic evaluation.¹⁰⁰ In the ERSPC trial, 27 men needed to be diagnosed (and potentially treated) to prevent 1 prostate cancer death, underscoring the potential for overdiagnosis and overtreatment with PSA screening.³⁰ Evidence is limited on the benefit of screening among men older than 70 years or the differential benefits among African American men or men with a family history of prostate cancer.

Trial data demonstrate that abnormal PSA screening test results are common and that most men referred for biopsy after abnormal screening results will not have prostate cancer.^15,30,52 Biopsy harms include pain, bleeding, and infection,⁵³ but perhaps the most serious harm of prostate cancer screening is overdiagnosis, because overdiagnosis burdens men with the potential harms of diagnosis and treatment without improving life expectancy or quality of life. It was estimated that 16.4% to 50.4% of prostate cancers were overdiagnosed during the 3 trials, consistent with estimates based on ecological or modeling studies.^4,101-104

This review assessed both the immediate harms of screening as well as the consequent adverse effects of prostate cancer treatments. Most men undergoing radical prostatectomy will experience long-term sexual difficulties and approximately 17% will experience urinary difficulties, while approximately 36% of men receiving radiation therapy experience erectile problems and many will experience adverse bowel symptoms.⁶⁶ Nevertheless, compared with conservative management, active treatments for localized prostate cancer did not clearly compromise overall quality of life or global physical or mental health status, despite adverse sexual, urinary, and bowel effects.^{11,66,70-72,74,75,77,86}

The ProtecT trial found similarly high prostate cancer survival among men with screen-detected, early-stage prostate cancer randomized to radical prostatectomy, radiation therapy, or active surveillance. Of men assigned to active surveillance, 45.2% remained under surveillance without receiving active treatment during the 10-year follow-up period, although an increase in the incidence of metastatic disease was observed in the active surveillance group (approximately 6% with active surveillance vs 2 to 3% with active treatment). In contrast to the ProtecT active surveillance protocol that consisted of periodic PSA monitoring, many active surveillance protocols include surveillance biopsy or imaging, which might reduce metastatic disease risk, albeit with added harms associated with repeated biopsy. In contemporary case series, prostate cancer–specific survival estimates for patients receiving active surveillance have been as high as 99% at 10 years of follow-up, although since relatively few men in these series have had extended follow-up, uncertainty remains about long-term outcomes.¹⁰⁵

This review demonstrates that prebiopsy risk calculators can discriminate between men with and without high-risk cancer better than PSA screening alone, but net clinical benefit of routine calculator use in biopsy decisions is not established by existing evidence. Diagnostic or surveillance strategies based on serum or urine tests or multiparametric magnetic resonance imaging are under study, and some have been recommended by the National Comprehensive Cancer Network.¹⁰⁶ Further research is needed to elucidate whether the use of adjunctive tests can improve the balance of benefits and harms of PSA-based prostate cancer screening in community settings.

Because the lead time for prostate cancer may be very long, current screening trials (with <15 years of median follow-up) might underestimate mortality benefits. Limited follow-up duration from trials may also exaggerate estimates of overdiagnosis based on extra incidence, although estimates based on longer-term follow-up may be influenced by posttrial PSA testing. In the PLCO trial, contamination among control group participants would be expected to bias trial results toward the null. The ERSPC trial, in contrast, was limited by unexplained stage-adjusted differences in prostate cancer treatments by study group that may have biased results in favor of screening.¹⁰⁷ There was also variation across ERSPC sites in recruitment methods, screening intervals, use of ancillary testing, and PSA thresholds for biopsy referral. The CAP trial was limited by the low adherence of intervention participants to the invitation to a single PSA screen. Across all studies, relatively few men older than 70 years were enrolled, and there is limited evidence about the differential benefits or harms of screening for men at higher risk.

Of 4 randomized trials comparing the effectiveness and harms of treatments for localized prostate cancer, only the ProtecT trial exclusively enrolled men with screen-detected cancer, and prostate cancer–specific and all-cause mortality were extremely low in that study.⁵⁸ Other treatment RCTs enrolled many or mostly men with clinically detected prostate cancer.^59,68,85 Thus, uncertainty remains about the generalizability of treatment trial results to US men with early-stage prostate cancer detected by PSA screening.

This review may be limited by language or publication biases. Aside from uncontrolled studies of treatment harms, few or no studies were found on comparative effectiveness of new or novel treatment modalities, such as alternative surgical approaches (eg, nerve-sparing or robotic surgery), cryotherapy, or high-intensity focused ultrasound. Because of the limited use and variable definitions of active surveillance during the time periods of most included studies, active surveillance and watchful waiting were grouped in analyses, although outcomes may differ between these conservative approaches.

PSA screening may reduce prostate cancer mortality risk but is associated with false-positive results, biopsy complications, and overdiagnosis. Compared with conservative approaches, active treatments for screen-detected prostate cancer have unclear effects on long-term survival but are associated with sexual and urinary difficulties.

Corresponding Author: Joshua J. Fenton, MD, MPH, Department of Family and Community Medicine, University of California, Davis Medical Center, 4860 Y St, Ste 2300, Sacramento, CA 95817 (jjfenton@ucdavis.edu).

Accepted for Publication: March 23, 2018.

Author Contributions: Dr Fenton had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Fenton, Weyrich, Melnikow.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Fenton, Weyrich, Durbin.

Critical revision of the manuscript for important intellectual content: Fenton, Durbin, Liu, Bang, Melnikow.

Statistical analysis: Fenton, Liu, Bang.

Obtained funding: Melnikow.

Administrative, technical, or material support: Weyrich, Melnikow.

Supervision: Fenton, Weyrich, Bang, Melnikow.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This research was funded under contract HHSA-290-2012-00015-I, Task Order 6, from the Agency for Healthcare Research and Quality (AHRQ), US Department of Health and Human Services, under a contract to support the USPSTF.

Role of the Funder/Sponsor: Investigators worked with USPSTF members and AHRQ staff to develop the scope, analytic framework, and key questions for this review. AHRQ had no role in study selection, quality assessment, or synthesis. AHRQ staff provided project oversight, reviewed the report to ensure that the analysis met methodological standards, and distributed the draft for peer review. Otherwise, AHRQ had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript findings. The opinions expressed in this document are those of the authors and do not reflect the official position of AHRQ or the US Department of Health and Human Services.

Additional Contributions: We gratefully acknowledge the following individuals for their contributions to this project: David Meyers, MD, and Tracy Wolff, MD, MPH (Agency for Healthcare Research and Quality); current and former members of the US Preventive Services Task Force who contributed to topic deliberations; and Jennifer S. Lin, MD, MCR (Kaiser Permanente Research Affiliates Evidence-based Practice Center, Portland, Oregon), Russell Harris, MD, MPH (University of North Carolina), and Bruce Abbott, MLS (University of California, Davis). USPSTF members, peer reviewers, and federal partner reviewers did not receive financial compensation for their contributions.

Additional Information: A draft version of this evidence report underwent external peer review from 6 content experts (Otis Brawley, MD, American Cancer Society; Peter Carroll, MD, University of California, San Francisco; Phillip Dahm, MD, University of Minnesota; David Penson, MD, Vanderbilt-Ingram Cancer Center; Ian Thompson, MD, University of Texas Health Science Center at San Antonio; Andrew Vickers, DPhil, Memorial Sloan Kettering Cancer Center) and 8 federal partners from the National Institute on Minority Health and Health Disparities, National Cancer Institute (NCI), National Institute of Nursing Research, and the Centers for Disease Control and Prevention. Comments from reviewers were presented to the USPSTF during its deliberation of the evidence and were considered in preparing the final evidence review.

Editorial Disclaimer: This evidence report is presented as a document in support of the accompanying USPSTF Recommendation Statement. It did not undergo additional peer review after submission to JAMA.