Assessment of prostate-specific antigen screening: an evidence-based report by the German Institute for Quality and Efficiency in Health Care

Abstract

Context

Prostate-specific antigen (PSA) testing increases prostate cancer diagnoses and reduces long-term disease-specific mortality, but also results in overdiagnoses and treatment-related harms.

Objective

To systematically assess the benefits and harms of population-based PSA screening and the potential net benefit to inform health policy decision-makers in Germany.

Evidence Acquisition

We performed a protocol-guided comprehensive literature search according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement. All steps were performed by one or two investigators; any discrepancies were resolved by consensus. To allow subgroup analyses for identifying the optimal screening parameters, the eight national trials conducted under the umbrella of the European Randomised study of Screening for Prostate Cancer (ERSPC) were included as individual trials.

Evidence Synthesis

We included a total 11 randomised controlled trials (RCTs) with a total of 416 000 study participants. For all-cause mortality, we found neither benefit nor harm. PSA screening was associated with a reduced risk of both prostate cancer mortality and the development of metastases. For the outcomes of health-related quality of life, adverse effects and the consequences of false-negative screening results there was no difference; however, this was due to the lack of eligible RCT data. Finally, PSA screening was associated with large numbers of overdiagnoses with adverse downstream consequences of unnecessary treatment (e.g. incontinence, erectile dysfunction) and large numbers of false-positive PSA tests leading to biopsies associated with a small but not negligible risk of complications. Limitations of this assessment include the clinical heterogeneity and methodological limitations of the underlying studies.

Conclusions

The benefits of PSA-based prostate cancer screening do not outweigh its harms. We failed to identify eligible screening studies of newer biomarkers, PSA derivatives or modern imaging modalities, which may alter the balance of benefit to harm.

Patient Summary

In the present study, we reviewed the evidence on the PSA blood test to screen men without symptoms for prostate cancer. We found that the small benefits experienced by some men do not outweigh the harms to many more men.

Introduction

Updated results of the large trials on PSA screening have become available: the European Randomised study of Screening for Prostate Cancer (ERSPC); the Prostate, Lung, Colorectal, and Ovarian (PLCO) randomised cancer screening trial; and the Cluster Randomized Trial of PSA Testing for Prostate Cancer (CAP) [1-3]. However, their conflicting results make an assessment difficult [4-6]. In Germany, as one of the largest countries in Europe, two prostate cancer screening tests are currently in use: the DRE and the PSA test. While the DRE is reimbursed by statutory health insurance, the PSA test is only available for self-payers. Currently, Germany has a high level of opportunistic PSA screening that, as in other countries, includes men aged >75 years, who appear least likely to benefit [7, 8]. The decision on the reimbursement of the costs of the PSA test for screening purposes in Germany lies with the Federal Joint Committee (Gemeinsamer Bundesausschuss [G-BA]). In December 2018, a patient representative request prompted the G-BA to commission the Institute for Quality and Efficiency in Health Care (Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen [IQWiG]) with the critical assessment of population-based PSA screening in men without clinical suspicion of prostate cancer with regard to patient-relevant outcomes. To allow subgroup analyses for identifying the optimal screening parameters (e.g. PSA threshold for recommending a prostate biopsy, screening interval), the eight national trials conducted under the umbrella of the ERSPC were included as individual trials. In the present review, we present the results and conclusions of the final IQWiG report to inform the G-BA in its decision-making.

Materials and Methods

We performed a systematic review of the beneficial and harmful effects of PSA screening based on an a priori protocol (report plan; Commission No. S19-01) and based on established methods of the IQWiG [9]. We included randomised controlled trials (RCTs) of men at risk of prostate cancer comparing PSA screening with no screening considering the following outcomes: all-cause mortality, prostate cancer mortality, incidence of metastatic prostate cancer, morbidity (e.g. pain due to bone metastases), health-related quality of life, adverse events, direct and indirect screening harms including those due to false screening test results, and overdiagnosis (e.g. complications of subsequent prostate cancer treatment).

Search Strategy

We performed a comprehensive literature search of multiple databases for English and German language publications without restrictions with regard to study duration or date of publication using a two-step approach. We started with a focussed search for high-quality systematic reviews, published between January 2013 and January 2019 in the Medical Literature Analysis and Retrieval System Online (MEDLINE), the Cochrane Database of Systematic Reviews and Health Technology Assessment (HTA) database. In addition, websites of HTA agencies such as the National Institute for Health and Care Excellence (NICE) and the Agency for Healthcare Research and Quality (AHRQ) were searched for systematic reviews. Thereafter, we performed an update search in MEDLINE, the Excerpta Medica dataBASE (EMBASE) and Cochrane Central Register of Controlled Trials for relevant primary studies published within the period not covered by the systematic reviews (2018 until 20 May 2019). We also systematically screened clinical trial registries. The full search strategies were developed by an experienced information specialist (see Appendix S1). When indicated, we sent study authors queries and included additional relevant information. For the focussed search, one member of the research team performed all screening steps; for the comprehensive search these were done by two independent reviewers; any discrepancies were resolved by consensus.

Data Extraction and Risk of Bias Assessment

All steps of the data extraction and risk of bias assessment procedures were conducted by one person and checked by another using standard IQWiG table templates and methods; disagreements were resolved by consensus. Extracted Information included study characteristics, participants’ inclusion and exclusion criteria, characteristics of the study participants, the screening intervention (including PSA thresholds and screening intervals), adherence, contamination, as well as methodological details to inform the risk of bias assessment. Study results were incorporated in the analyses only if ≥70% of the study participants were taken into account by analysing original or imputed data in the respective study analyses and if adherence to the PSA screening intervention (participation in at least one PSA screening round) was high enough to allow its adequate evaluation.

Data Analysis

Effect measures were reported as incidence rate ratios (IRRs), i.e. the incidence rate in the screening group (number of events divided by the person-time at risk) divided by the incidence rate in the control group, including 95% CIs. If the statistical heterogeneity test [10] was not statistically significant (P ≥ 0.05) no substantial heterogeneity was assumed and a meta-analysis was performed. A pooled effect estimate was calculated using a random-effects model using either the Knapp–Hartung method or the Knapp–Hartung method with ad hoc variance correction [11]. Statistical significance was assumed for a P < 0.05. Clinical heterogeneity was assessed by comparing inclusion and exclusion criteria, as well as baseline characteristics of the study participants. In addition to relative effect measures, absolute effects measures were calculated by application of the IRR from the meta-analysis to the median risk in the control group (baseline risk). Overdiagnosis was estimated using the excess incidence approach and presented relative to the number of randomised men in the screening group (risk of overdiagnosis). We performed pre-planned subgroup analyses based on age, PSA thresholds for recommending a prostate biopsy, duration of screening, number of screening rounds, and screening intervals. Due to differences in the timing between informed consent and randomisation, we additionally performed a subgroup analysis regarding time of consent (before or after randomisation). Subgroup analyses were only performed if more than one study could be assigned to each subgroup after categorisation of the studies with respect to possible effect modifiers. Sensitivity analyses for risk of bias (studies with a low risk of bias versus all studies) were planned, but not performed as the risk of bias of all studies was judged as high.

Assessment of the Trade-Off of Benefits and Harms of PSA Screening

We assessed the relationship of the benefits and harms of PSA screening and the relevant net benefit for men faced with the decision to undergo PSA screening based on the absolute effect size estimates across outcomes.

Document Review

The report plan and the preliminary report were internally reviewed by all team members as well the IQWiG leadership and then made available online (report plan: May 2019; preliminary report: January 2020) for commenting by interested parties including professional organisations, individual medical professionals, patients, industry, and the general public. One organisation submitted comments to the report plan, 14 organisations, and four individuals submitted comments to the preliminary report. Representatives of these organisations and individuals who had provided commentary to the preliminary report were invited to an oral scientific debate held at IQWiG in Cologne, Germany in February 2020. The document was revised to address pertinent comments, but no changes were made to the overall assessment. The document was finalised, approved by IQWiG, and formally submitted to the G-BA in April 2020.

Results

Search Results

Our focussed search identified two recently published high-quality systematic reviews [4-6] that included 11 relevant trials; with the exception of one study (Quebec) these were also captured by the comprehensive search (Fig. 1). We included all 11 trials with a total of 416 000 study participants. These studies consisted of eight national trials conducted under the umbrella of the ERSPC, PLCO trial, as well as the so-called Quebec and Stockholm studies (see Appendix: Table S1). We excluded the CAP trial given that after allocation ~30% of randomised primary care practices in both groups were excluded; mostly because they did not provide a definitive response or pro-actively discontinued after randomisation [3], therefore becoming a major source of selection bias. Thus, the CAP trial was not considered an adequate RCT. Meanwhile, in the other studies with post randomisation consent (e.g. ERSPC Finland and ERSPC Sweden), all study participants were included as allocated in the respective analyses. Despite the high risk of bias in all studies included in our present review, they were therefore considered to be adequate RCTs and included.

Fig. 1

Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flowchart.

Study Characteristics

Follow-up duration ranged between 11 and 20 years (Table 1). Almost all studies included men aged 55–70 years. In all studies, prostate cancer mortality was the primary endpoint. In the PLCO trial, in addition to screening for prostate cancer, participants underwent screening for colon and lung cancer. Studies varied substantially with regard to methods (randomisation, allocation concealment, and consenting process) and screening strategies, in particular the PSA threshold, the use of additional screening tests, the number of screening rounds, and the interval between screening rounds. Studies also differed by the proportion of participants in the screening group who underwent at least one PSA test (adherence). Adherence was particularly low in the French ERSPC study (28%) and the Quebec study (24%), and was judged too low to adequately assess the comparison (see Appendix: Table S2). Therefore, the results of both studies were not incorporated in any analyses. Thus, nine RCTs with a total of 285 000 participants were formally considered, namely the ERSPC studies from Finland, Italy, Sweden, Belgium, the Netherlands, Switzerland, and Spain [1], as well as the PLCO trial [2] and the Stockholm study [12].

Table 1. Characteristics of the included studies.

Study (Number of men)	Country and time of recruitment	Follow-up, years*	Indication for further tests (including prostate biopsy)	Screening rounds, n	Screening interval, years
PSA threshold <4 ng/mL
ERSPC Netherlands (42 368)	Netherlands, 1993–2000	16^†	PSA ≥3 ng/mL^‡	1–5^§	4
ERSPC Sweden (19 911)	Sweden, 1994^¶	18	PSA ≥2.9 ng/mL**	3–10^§	2
ERSPC Switzerland (10 309)	Switzerland, 1998–2003	13^†	PSA ≥3 ng/mL	n.s.	4
ERSPC Spain (3702)	Spain, 1996–1999	16^†	PSA ≥3 ng/mL	1–3^§	4
PSA threshold ≥4 ng/mL
ERSPC Finland (80 379)	Finland, 1996–1999	16^†	PSA ≥4 ng/mL or 3.0–3.9 ng/mL and free PSA <16%^††	2–3^§	4
ERSPC Italy (14 971)	Italy, 1996–2000	15^†	PSA ≥4 ng/mL or 2.5–3.9 ng/mL and abnormal DRE or TRUS	n.s.^§	4
PLCO trial (76 683)	USA, 1993–2001	17^†	PSA ≥4 ng/mL or abnormal DRE	6^‡‡	1
Stockholm (26 602)	Sweden, 1988^¶	20	PSA ≥10 ng/mL or abnormal DRE or TRUS	1	–
Variable PSA thresholds < and >4 ng/mL
ERSPC Belgium (10 359)	Belgium, 1991–2003	16^†	variable PSA thresholds below and above 4 ng/mL^§§	1–3^§	4^¶¶
Studies not incorporated in any analyses due to low adherence
ERSPC France (84 512)	France, 2000–2005	n.s.	PSA ≥3 ng/mL	2	2
Quebec (46 486)	Canada, 1988^¶	11	PSA ≥ 3 ng/mL or abnormal DRE and subsequently abnormal TRUS	n.s.	1

• n.s., not stated.
• * From randomisation.
• ^† Median.
• ^‡ Until April 1997 ≥4 ng/mL, until January 1996 abnormal DRE and/or TRUS.
• ^§ The number of screening rounds decreased as the age at randomisation of the participants increased, as a maximum screening age was set close to the upper age at randomisation.
• ^¶ All participants were randomised on the same day.
• ** 1995–1998 ≥3.4 ng/mL, 1999–2004 ≥2.9 ng/mL, from 2005 ≥2.5 ng/mL.
• ^†† Until 1998 abnormal DRE.
• ^‡‡ Four annual PSA screening rounds (T0–T3) for men randomised before May 1994 and five PSA screening rounds (T0–T3 and T5) for men randomised between June 1994 and May 1995.
• ^§§ 1992–1994 ≥10 ng/mL, 1995–1998 ≥4 ng/mL, from 1999 ≥3 ng/mL; abnormal DRE or TRUS in the first screening round, abnormal DRE in the second screening round.
• ^¶¶ The interval between the first and second screening round was 7 years.

Risk of Bias Assessment

The risk of bias on study level was rated as high for all nine RCTs, as it was unclear whether the allocation of men to the intervention and control group was concealed. Furthermore, study participants and personnel were not blinded (see Appendix: Table S3).

Effects of PSA Screening

For all-cause mortality, the ERSPC studies (overall results) and Stockholm study reported IRRs of 0.99 (95% CI 0.98–1.01) and 1.01 (95% CI 0.96–1.07), respectively (Appendix: Fig. S1; due to statistical properties, a pooled effect estimate was not presented). In addition, two ERSPC studies (Finland and Sweden) reported results separately and were pooled with the Stockholm study leading to an IRR of 1.0 (95% CI 0.98–1.02; Appendix: Fig. S2). We were unable to conduct any of the pre-planned secondary analyses.

For prostate cancer mortality, based on available data of seven ERSPC studies (Belgium, Finland, Italy, the Netherlands, Sweden, Switzerland and Spain), PLCO and Stockholm we found a pooled IRR of 0.86 (95% CI 0.75–0.98; Appendix: Fig. S3). Results of a subgroup analysis based on PSA biopsy threshold were notable for a test of interaction <0.001 and therefore suggestive of a subgroup effect favouring the lower PSA threshold (<4 ng/mL: IRR 0.68, 95% CI 0.51–0.89 vs ≥4 ng/mL: IRR 0.95, 95% CI 0.86–1.05; Fig. 2). All other secondary analyses failed to suggest any evidence of subgroup effects (age, duration of screening, number of screening rounds, screening interval, and consent provided before or after randomisation).

Fig. 2

Prostate cancer mortality (subgroup analysis by PSA threshold).

The analysis for metastases was limited to four ERSPC studies (Finland, the Netherlands, Sweden and Switzerland) and found a pooled IRR of 0.67 (95% CI 0.58–0.78; Appendix: Fig. S4). We were unable to conduct any of our pre-planned secondary analyses. We did not find any eligible evidence for the outcomes health-related quality of life and adverse events.

Across all studies the risk of overdiagnosis was 7 (95% CI 3–12) to 60 (95% CI 54–66) per 1000 men invited (Table 2). In the studies with a PSA threshold of <4 ng/mL, the risk of overdiagnosis was 35 (95% CI 13–56) to 60 (95% CI 54–66) per 1000 men invited. In the studies with a PSA threshold of ≥4 ng/mL, the risk of overdiagnosis was 7 (95% CI 3–12) to 16 (95% CI 11–20) per 1000 men invited.

Table 2. Prostate cancer diagnoses and risk of overdiagnosis (by PSA threshold).

Studies (age in years; median follow-up in years)	Screening				Control				Risk of overdiagnosis
	N S	Men with events			N C	Men with events			(S2/NS–C2/NC), %* [95% CI*]
		S2	%	Person years		C2	%	Person years
Studies with a PSA threshold <4 ng/mL
ERSPC Netherlands (55–69; 16)	17 443	2376	13.6*	n/a	17 390	1325	7.6*	n/a	6.0 [5.4–6.6]
ERSPC Sweden (50–64; 18)	9950	1396	14.0	143 776	9949	962	9.7	149 129	4.4 [3.5–5.3]
ERSPC Switzerland (55–69; 13)	4948	620	12.5*	n/a	4955	364	7.3*	n/a	5.2 [4.0–6.4]
ERSPC Spain (50–69; 16)	1056	92	8.7*	n/a	1141	60	5.3*	n/a	3.5 [1.3–5.6]
Studies with a PSA threshold ≥4 ng/mL
ERSPC Finland (55–67; 16)	31 970	3500	10.9*	n/a	48 409	4546	9.4*	n/a	1.6 [1.1–2.0]
ERSPC Italy (55–69; 15)	7265	560	7.7*	n/a	7250	452	6.2*	n/a	1.5 [0.6–2.3]
PLCO (55–74; 17)	38 340	5574	14.5*	n/a	38 343	5287	13.8*	n/a	0.7 [0.3–1.2]
Stockholm (55–70; n/a^†)	2400	292	12.2*	n/a	25 081	2727	10.9	n/a	1.3 [−0.1– 2.7]
Studies with a variable PSA threshold < or >4 ng/mL
ERSPC Belgium (55–69; 16)	4307	482	11.2*	n/a	4255	393	9.2*	n/a	2.0 [0.7–3.2]

• C2, number of men diagnosed with prostate cancer in the control group at follow-up; n/a, not available; N_C, number of men in the control group; N_S, number of men in the screening group; S2, number of men diagnosed with prostate cancer in the screening group at follow-up.
• * Own calculation.
• ^† Mean follow-up (range): 15.4 (0.2–20.8) years.

Across all studies, 68–82% of the positive screening results per screening round proved to be false, and 4–19% of all screening participants had a false-positive screening result. Also, 11–26% of all screening participants had at least one false-positive screening result after three or more screening rounds.

The percentage of false-positive screening results per screening round did not differ between studies by PSA threshold. However, both, the percentage of screening participants with a false-positive screening result per screening round and the percentage of screening participants with at least one false-positive screening result after three or more screening rounds was double in the studies with a PSA threshold of <4 ng/mL compared to studies with a PSA threshold of ≥4 ng/mL.

Only the PLCO trial provided relevant results regarding complications related to false-positive screening results. Biopsy complications overall, infectious and non-infectious complications were reported in 22.6, 7.7 and 15.2 per 1000 participants with a false-positive screening result, respectively [13]. We did not find any eligible evidence for the consequences of false-negative screening results.

Assessment of the Trade-Off of the Benefits and Harms of PSA Screening

Based on the findings of the present review, it is uncertain whether PSA screening would reduce all-cause mortality. On the side of potential benefits, we estimated for prostate cancer mortality that PSA screening would reduce death from prostate cancer by approximately 3 (95% CI 1–5) in 1000 men over a 16-year period, whereas 6 (95% CI 5–8) in 1000 men would die from prostate cancer despite screening. Progression to metastatic disease would be avoided in approximately 3 (95% CI: 2–4) in 1000 men over a 12-year period, whereas 6 (95% CI: 5–7) in 1000 men would develop metastases despite screening. We found no comparative evidence on quality of life and adverse events. On the side of potential harms, we estimated the risk of overdiagnosis at about 35–60 in 1000 men. These men would be given a diagnosis of prostate cancer and likely experience some form of treatment-related complications yet without benefit. Also, 223–261 men per 1000 screening participants would undergo at least one prostate biopsy due to a false-positive PSA result and about 5–6 men per 1000 screening participants would experience biopsy complications (2% of biopsied men experienced biopsy complications). Therefore, we determined PSA screening harms significantly more men through overdiagnosis and false-positive results than it benefits men (Table 3). Moreover, many screening harms occur much earlier than the benefits and often have a life-long impact. In summary, we therefore conclude that the benefits of PSA screening do not outweigh the harms.

Table 3. Assessment of the trade-off of benefits and harms of population-based PSA screening.

Outcomes	Basic risk* per 1000 men	Absolute effect per 1000 invited men [95% CI]	Interpretation
Benefits ^†
Prostate cancer mortality	9	3^‡ [1–5]	Over a 16-year period, screening with the PSA test having a PSA threshold <4 ng/mL would prevent death by prostate cancer in about 3 men per 1000 invited men. It is unclear whether PSA screening prolongs life significantly considering the high number of competing causes of death in the relevant age group
Prostate cancer metastases	9	3^‡ [2–4]	Over a 12-year period, screening with the PSA test would prevent metastases in about 3 men per 1000 invited men
Harms ^†
Consequences of overdiagnosis	–	35 [13–56]^§ to 60 [54–66]^¶	About 35–60 men per 1000 invited men would receive a prostate cancer diagnosis, yet without benefit. These diagnoses might lead to serious and long-lasting complications due to prostate cancer therapy
Consequences of false-positive screening results	–	223** to 261**	About 223–261 men per 1000 screened men would receive at least one false-positive screening result. Rarely, a subsequent prostate biopsy might result in a serious adverse event (e.g. sepsis)

• * Median risk of control group.
• ^† For patient-relevant outcomes not shown in this table, there was no difference; however, this was due to the lack of eligible RCT data.
• ^‡ It appears likely that men dying from prostate cancer are a subset of men with prostate cancer who developed metastases at an earlier point in time.
• ^§ ERSPC Spain.
• ^¶ ERSPC Netherlands.
• ** The denominator is screened men (instead of invited men).

Stakeholder Discussion of Report Findings

Issues raised by stakeholder representatives both in writing, as well as at the face-to-face meeting, focussed on methodological aspects such as the appropriateness of considering all-cause mortality as a relevant outcome in this setting and consideration of the high level of contamination in the PLCO trial. It was also raised that an increased uptake of active surveillance in Germany was not adequately reflected (between 2013 and 2017 from 13% to 27% [14]) and, therefore, the burden of overtreatment was overestimated, and that the project team did not consider novel risk-adapted strategies of PSA screening in its deliberation. Moreover, it was stated that the evaluation did not enable clinicians to advise the individual man as to whether PSA screening would benefit him. As to the latter, the hearing’s Chair clarified that the charge by the G-BA to IQWiG was to assess the benefits and harms of population-based PSA screening.

Discussion

Statement of Principal Findings

The central overall finding of the present review was that the benefits of PSA-based prostate cancer screening do not outweigh its harms. For all-cause mortality, there was neither benefit nor harm; PSA screening was associated with a reduced risk of both prostate cancer mortality and the development of metastases. For the outcomes of health-related quality of life, adverse effects and the consequences of false-negative screening results there was no difference; however, this was due to the lack of eligible RCT data. Finally, PSA screening was associated with large numbers of overdiagnoses with adverse downstream consequences of unnecessary treatment (e.g. incontinence, erectile dysfunction) and large numbers of false-positive PSA tests leading to biopsies associated with a small but not negligible risk of complications (e.g. sepsis). These conclusions only apply to the type of screening used in the included studies. The benefit–harm ratio may in future be shifted in favour of PSA screening by novel risk-adapted strategies of PSA screening, the higher uptake of active surveillance, and improved treatment options and biopsy techniques with fewer and less severe complications. However, the present review failed to identify eligible screening studies that included newer biomarkers, PSA derivatives or modern imaging technology (MRI), which have been suggested to improve the detection of clinically relevant prostate cancer and/or to reduce the number of overdiagnoses. Two RCTs are ongoing, but will not yield results for another decade [15, 16].

Strengths and Weaknesses of the Review

The main limitations of the present review relate to issues regarding the quality of the individual RCTs that constitute the body of evidence that informed this assessment of related benefits and harms. All included studies were rated at high risk of bias. Included studies were marked by both methodological heterogeneity (as it relates to adherence and contamination rates) and clinical heterogeneity (in terms of screening intervals and biopsy thresholds). This included studies that informed the two groups of the subgroup analysis based on PSA threshold that also differed in other important characteristics. The relatively small number of studies contributing to any of the analyses also limited the ability for secondary sensitivity and subgroup analyses. Moreover, subgroup analyses regarding other variables were not possible due to a lack of established cut-offs (e.g. median follow-up times) and of appropriate data (e.g. time of recruitment to each trial: recruitment lasted several years in most trials and overlapped widely between trials). Finally, several sources of heterogeneity unknown before randomisation, such as adherence and biopsy rate, were not investigated in subgroup analyses. In these cases, differences between subgroups might be due to imbalances of prognostic variables between screening and control groups within each subgroup [17]. A meta-regression analysis investigating the impact of adherence on prostate cancer mortality was not conducted, as there are studies showing similar effects on prostate cancer mortality and at the same time differing widely regarding adherence (e.g. ERSPC Netherlands and ERSPC Sweden) and conversely, there are studies showing similar adherence and differing widely regarding prostate cancer mortality (e.g. ERSPC Sweden and ERSPC Finland); such a post hoc analysis was therefore unlikely to provide meaningful results.

There was a lack of direct evidence on downstream consequences of screening related to treatment complications (including reduced urinary and sexual quality of life).

Strengths and Weaknesses in Relation to Other Systematic Reviews

The results of the two recently published high-quality systematic reviews [4-6] are broadly consistent with those of the present review. There is agreement that PSA screening most likely does not affect all-cause mortality and that there is an effect in favour of PSA screening in terms of prostate cancer mortality in the presence of a high risk of overdiagnosis and a high number of false-positive screening results. While the results for prostate cancer mortality and false-positive screening findings are largely consistent, estimates of overdiagnosis differ significantly from the estimate in the present review. In Ilic et al. [6] 2018, the risk of overdiagnosis was estimated at ~7 per 1000 invited men, in Fenton et al. [4, 5] 2018 at ~32 per 1000 invited men, and in the present review at ~35–60 per 1000 invited men. In both systematic reviews, only the joint evaluations of all ERSPC studies were used, which is why subgroup analyses such as those in our present review could not be performed. In Ilic et al. [6] 2018, the much lower estimate is due to the inclusion of the CAP study [3], in which only a few overdiagnoses occurred due to low adherence in the screening group (38%).

Comparison to Other Assessments of PSA Screening

A number of guidelines that meet minimal requirements by the National Academy of Medicine (formerly: Institute of Medicine) for trustworthy guidelines exist. For Germany, the so-called ‘S-3 guideline’ on prostate cancer is the most relevant and was led by the German Society of Urology (DGU) [18]. Based on a level of evidence of 4 (which refers to expert consensus) this document recommends counselling of men aged ≥45 years about the pro and cons of prostate cancer screening. Whereas, in the same guideline, primary care physicians (who would likely be the main implementers of any nationwide screening programme) recommend to not proactively raise this issue with their patients unless these inquire about screening. Meanwhile, guidelines by the European Association of Urology (EAU) suggest offering well-informed men with a life-expectancy of at least 10–15 years a risk-adapted strategy for early detection [19]. This risk-adapted strategy is based on age, family history of prostate cancer, ethnic background as well as (added recently) carriers of the BReast CAncer type 2 (BRCA2) mutation, and baseline PSA level. Meanwhile, the third guideline by a professional organisation of urologists, the AUA, makes a strong recommendation for shared decision-making for men aged 55–69 years that are considering PSA screening, and then proceeding based on a man’s values and preferences. Meanwhile, guidelines by the United States Preventive Services Task Force (USPSTF) state that it ‘does not recommend screening for prostate cancer unless men express a preference for screening after being informed of and understanding the benefits and risks’. It then goes on to emphasise the role of each man’s own values about the potential benefits and harms in arriving at a determination [20, 21]. Similarly, the American Academy of Family Physicians and the Canadian Task Force on Preventive Health Care both recommend against population-based PSA screening, while once again emphasising shared decision-making [22, 23]. Meanwhile, an independent group recently published a rapid recommendation triggered by publication of the CAP trial, making a conditional recommendation against PSA screening [24]. Therefore, there is considerable controversy on this issue, which may have less to do with the body of evidence, but differing underlying assumptions about how an improvement in disease-specific mortality and risk of metastatic progression experienced by a relatively small number of men should be weighed against the risk of overdiagnosis and overtreatment in a much larger group of men who will not benefit. It should be noted that despite the major emphasis that most guidelines now place on shared decision-making, many documents remain vague as to how this should take place. A recent systematic review on the role of decision-aids for making the prostate cancer screening choice found that existing trials lacked compelling evidence to warrant their use [25].

G-BA Decision

In December 2020 the G-BA decided against the introduction of a population-based PSA screening and recommended awaiting the results of the ongoing studies on risk-adapted strategies of PSA screening.

Unanswered Questions and Future Research

One of the central challenges of the prostate cancer screening debate is the very long time it takes for RCTs to provide meaningful results for patient important outcomes, in particular disease-specific mortality. Nevertheless, these trials will continue to be needed to avoid the well-known issues of lead-time and length-time bias of non-randomised studies. One of the areas of greatest promise for prostate cancer screening lies in the use of multiparametric MRI imaging [26, 27]. Two RCTs are ongoing but will not yield results for another decade [15, 16]. Additional, serum- and urine-based biomarkers such as the four-kallikrein panel (4K)score, Prostate Health Index and prostate cancer antigen 3 (PCA3) have also become commercially available, but high-quality evidence to support their use is lacking [28, 29]. Large, high-quality trials should find strong support by all involved stakeholders in Germany to be impacted by the G-BA’s decision. Additional directions of future research would be to develop better approaches to shared decision-making, as well as an even greater uptake of active surveillance in men with lower risk prostate cancer, in accordance with current evidence-based guidelines thereby reducing the burden of overtreatment.

Conclusions

The central overall finding of the present review is that the benefits of PSA-based prostate cancer screening do not outweigh its harms. Various measures to reduce PSA screening harms show promise to shift the benefit–harm ratio in favour of PSA screening. However, valid results in this regard are still being awaited.

Disclosures of Interest

None declared.

Acknowledgment

Open access funding enabled and organized by Projekt DEAL.

Abbreviations

CAP

Cluster Randomized Trial of PSA Testing for Prostate Cancer

ERSPC

European Randomised study of Screening for Prostate Cancer

G-BA

Gemeinsamer Bundesausschuss

HTA

Health Technology Assessment

IQWiG

Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen

IRR

incidence rate ratio; MEDLINE, Medical Literature Analysis and Retrieval System Online

PLCO

Prostate, Lung, Colorectal, and Ovarian (randomised cancer screening trial)

RCT

randomised controlled trial