Background: Prostate cancer is the most commonly diagnosed cancer and second leading cause of cancer death among U.S. men. U.S. prostate cancer incidence increased sharply with the dissemination of prostate-specific antigen (PSA) screening in the late 1980s, although it has been controversial whether the benefits of PSA-based screening outweigh potential harms.
Purpose: To update previous USPSTF systematic reviews regarding the benefits and harms of prostate cancer screening and treatments for screen-detected or localized prostate cancer, and to synthesize evidence on the utility of pre-biopsy risk calculators to identify men with clinically significant prostate cancers that are likely to progress to advanced disease.
Data Sources: We considered all studies included in prior USPSTF reviews, relevant English-language articles identified by searching PubMed, Embase, Web of Science and Cochrane Registries and Databases (through July 2017), and articles referenced in included articles or suggested by experts.
Study Selection: We included randomized controlled trials (RCTs) of PSA-based screening reporting prostate cancer morbidity, prostate cancer mortality, or all-cause mortality. For screening harms, we also considered cohort studies of men undergoing PSA screening and diagnostic followup. For treatment benefits and harms, we included RCTs and cohort studies of men with screen-detected or localized prostate cancer comparing outcomes of active treatments versus conservative management strategies (i.e., active surveillance, watchful waiting). We also included uncontrolled observational studies of treatment harms. For risk calculator studies, we included external validation studies of multivariable risk calculators that used PSA in addition to patient variables routinely available prior to prostate biopsy to predict the presence of prostate cancer (Gleason score ≥7 or stage T2b).
Data Extraction: One investigator abstracted study data, while a second checked data accuracy. Two investigators independently rated study quality based on pre-specified criteria.
Data Synthesis and Results: Fair-quality evidence on the impact of PSA screening on prostate cancer mortality and morbidity derives from two RCTs (the Prostate, Lung, Colorectal, and Ovarian [PLCO] Cancer Screening Trial and the European Randomized Study of Screening for Prostate Cancer [ERSPC] trial). During each year of the PLCO screening phase, approximately 46 percent of control arm participants received PSA screening, so the PLCO has been characterized as trial comparing organized versus opportunistic screening. After median followup of 14.8 and 13.0 years in the PLCO and ERSPC respectively, there was no difference in the risk of prostate cancer mortality in the screening versus control arms in the PLCO (RR, 1.04 [95% CI, 0.87 to 1.24]) but a 21 percent relative reduction in prostate cancer mortality in the ERSPC trial (RR, 0.79 [95% CI, 0.69 to 0.91]). Based on ERSPC incidence and mortality data, an estimated 27 men need to be diagnosed with prostate cancer to avert one prostate cancer death at 13 years of followup (95% CI, 17 to 66). Within the four ERSPC sites that reported it, randomization to screening was associated with 3.1 fewer cases of metastatic prostate cancer per 1,000 men randomized (95% CI, 1.8 to 4.4). In neither trial was screening associated with significantly reduced all-cause mortality.
In the ERSPC trial, there was a high rate of positive screening and biopsy (32.3 positive screens and 27.7 biopsies per 100 men randomized to screening). Biopsy-related harms include moderate to severe pain (7.3% at 35 days [95% CI, 5.7% to 9.1%]), infectious complications (range, 2% to 7%), and hospitalization (approximately 1%). Excess incidence data from the PLCO and ERSPC trials imply that between 20.7 percent and 50.4 percent of screen-detected cancers are overdiagnosed and would not have come to clinical attention in the absence of screening.
In the recently reported Prostate Testing for Cancer and Treatment (ProtecT) trial, prostate cancer survival was approximately 99 percent at 10-year followup among men with screen-detected prostate cancer in each of the three study arms (radical prostatectomy [RP], radiation therapy [RT] with neoadjuvant androgen deprivation therapy [ADT], or active surveillance [AS]), and there were no statistically significant differences in prostate cancer mortality. However, men randomized to active treatment (either RP or RT) were significantly less likely than men assigned to active surveillance to be diagnosed with metastatic disease (2.3% and 2.9% with RP and RT, respectively, vs. 6.0% with AS; NNT, 27 and 33 with RP and RT rather than AS, respectively, to prevent one case of metastatic disease at 10-year followup). Two prior RCTs of RP versus watchful waiting (WW) in localized prostate cancer also observed reduced long-term incidence of metastatic cancer with RP. In cohort studies, RP (7 cohorts) and RT (7 cohorts) were each associated with improved prostate cancer survival among men with localized prostate cancer compared to conservative management, while primary ADT for localized prostate cancer was associated with no significant differences in prostate cancer mortality or overall mortality compared to conservative management in a cohort study using instrumental variable analyses.
Based on pooled meta-analyses of RCT data, approximately 7.9 men would need to be treated with RP rather than conservative management for one additional man to experience urinary incontinence (95% CI, 5.4 to 12.2), and 2.7 men would need to be treated with RP rather than conservative management for one man to experience erectile dysfunction (95% CI, 2.2 to 3.6). In trials and cohort studies, approximately 7 percent of patients undergoing RP experienced major medical or surgical complications, and a median of 0.29 percent died within 30 days of surgery (8 studies). For every 6.9 men undergoing RT (95% CI, 5.1 to 10.7), one man will develop erectile dysfunction; bothersome bowel symptoms are also significantly increased with RT. As compared to conservative management, neither RP nor RT was associated with clinically significant impacts on generic measures of quality of life. In three cohort studies, primary ADT for localized prostate cancer was associated with erectile dysfunction in 73.8 percent to 85.8 percent of men; ADT has been associated with a range of systemic side effects in men with advanced cancer, including osteoporosis.
Two risk calculators have been externally validated for the prediction of significant prostate cancer in multiple biopsy cohorts: the Prostate Cancer Prevention Trial (PCPT) calculator (21 cohorts) and the ERSPC risk calculator (7 cohorts). In nearly all cohorts, the calculators discriminated between men with and without significant cancer better than PSA alone, although discrimination varied across cohorts. When assessed, risk calculator calibration was inconsistent across cohorts.
Limitations: Limitations of the screening trials include a high rate of PSA use in the PLCO control arm, biasing results toward the null, while the ERSPC trial was limited by differences in treatments received by men diagnosed with similarly staged cancers in screening and control arms. Followup duration in trials may be insufficient to detect differences in prostate cancer mortality or to quantify overdiagnosis. Only one treatment trial (ProtecT) exclusively enrolled men with screen-detected prostate cancer, and low event rates in this trial resulted in low power for discerning differences in prostate cancer mortality by study arm. No RCTs have evaluated the benefits and harms of risk calculator use prior to biopsy decisions among men with abnormal PSA screening.
Conclusions: PSA screening for prostate cancer may reduce risk of prostate cancer mortality but is associated with harms including false-positive results, biopsy complications, and overdiagnosis in 20 percent to 50 percent of screen-detected prostate cancers. Early, active treatment for screen-detected prostate cancer may reduce the risk of metastatic disease, although the long-term impact of early, active treatment on prostate cancer mortality remains unclear. Active treatments for prostate cancer are frequently associated with sexual and urinary difficulties.