Cancer detection in the European Randomised Study of Screening for Prostate Cancer (ERSPC)

Abstract

Objectives

To provide a comprehensive analysis regarding outcomes of repeated screening and detection of low‐grade versus high‐grade cancer from the European Randomised Study of Screening for Prostate Cancer (ERSPC).

Materials and Methods

This study analysed men aged 55–69 years at random across the ERSPC trial centres except France. Men in the screening arm were classified as ever‐attenders (who attended at least one screening cycle) or non‐attenders (who did not attend a specific screening cycle). Cancer detection rates (CDRs) were calculated by age group and centre across three screening cycles. A secondary analysis incorporated interval prostate cancer (PCa) cases—those diagnosed outside the screening protocol—among both ever‐attenders and non‐attenders. Incidence data from the control arm were included for comparison.

Results

Based on screen‐detected cases, CDRs were consistently higher for low‐grade than for intermediate‐ and high‐grade cases across all centres and screening cycles. Trends in CDRs for high‐grade cases varied by centre: CDRs decreased from the first to the second cycle and increased again during the third cycle in most centres, whereas CDRs increased in Spain with each successive screening cycle.

Conclusion

The ERSPC data show a decreasing detection rate for low‐grade cancer and substantial variation in the detection rate for intermediate‐ and high‐grade cancer between centres across repeated screening rounds. Increasing interval and non‐attender cancers indicate diminishing effectiveness of uniform screening schedules. These findings support personalised screening approaches and improved strategies to maintain participation.

Abbreviations

CA

control arm

CaP

the Cluster Randomised Trial of PSA Testing for Prostate Cancer

CDRs

cancer detection rates

ERSPC

European Randomised Study of Screening for Prostate Cancer

GG

grade group

ISUP

International Society for Urological Pathology

OPT

Organised Prostate Cancer Testing

PCa

prostate cancer

PLCO

the Prostate, Lung, Colorectal, and Ovarian Cancer Screening

PRAISE‐U

the PRostate cancer Awareness and Initiative for Screening in the European Union

PSA

prostate‐specific antigen

RDs

risk differences

SA

screening arm

Introduction

Screening for prostate cancer (PCa) remains a subject of debate because of conflicting views about the benefits and harms [1]. Several population‐based studies have evaluated the impact of prostate‐specific antigen (PSA) screening on PCa mortality and overdiagnosis. The population‐based European Randomised Study of Screening for Prostate Cancer (ERSPC) provides essential evidence on PSA screening. It has repeatedly been rated the highest‐quality trial in independent evaluations [2, 3, 4]. At 23 years of follow‐up, the absolute reduction in PCa mortality was 0.22 (95% confidence interval [CI] 0.11–0.33), whereas the excess incidence was 27 cases per 1000 men [5].

Two other major population‐based trials have also investigated PSA‐based screening: the Prostate, Lung, Colorectal, and Ovarian Cancer Screening (PLCO) trial in the USA and the Cluster Randomised Trial of PSA Testing for Prostate Cancer (CaP) trial in the UK. However, high contamination and low biopsy compliance in the PLCO and only a single screening round in the CaP trial have complicated the interpretation of their findings [4, 6, 7, 8, 9].

Previously, several efforts have assessed cancer detection rates (CDRs) across different sections of the ERSPC trial, including data from the Finnish, Dutch, and Swedish ERSPC centres [10, 11, 12, 13, 14, 15, 16, 17]. Still, a comprehensive assessment covering all centres and screening rounds has not been carried out.

Recognising the need to develop PCa screening guidelines, the Council of the European Union has endorsed PCa screening, provided there is sufficient evidence of effectiveness, resources are available, and PCa is a national priority [18]. Pilot studies are ongoing in the PRostate cancer Awareness and Initiative for Screening in the European Union (PRAISE‐U) project, and Sweden has implemented Organised Prostate Cancer Testing (OPT) in several regions to curb opportunistic PSA testing [19, 20].

Despite extensive research on PSA‐based screening, important knowledge gaps remain, particularly regarding outcomes of repeated screening rounds and the detection of low‐ versus high‐grade PCa. This study provides a comprehensive analysis that integrates data from all ERSPC centres to evaluate PCa detection using the International Society for Urological Pathology (ISUP) grade group (GG) over three screening cycles. We also assessed variation in PCa detection across centres, by age groups, and by screening cycles. The aim is to summarise the ERSPC trial results on cancer detection, with comparisons to the control arm by cycle. Understanding these patterns is essential for planning PCa screening programmes in populations with varying underlying risk levels.

Materials and Methods

This study employed a retrospective analysis of the ERSPC centres in Finland, the Netherlands, Sweden, Italy, Switzerland, Belgium, and Spain. Data from France were excluded because of its non‐randomised design and high contamination in the control group, in which men underwent opportunistic PSA testing during routine medical visits [5]. We examined PCa detection among ever‐attenders (attended at least one screening cycle) and non‐attenders (men who were invited but did not attend a specific screening cycle) in the screening arm (SA), using the control arm (CA) as the comparator. Risk differences (RDs) were calculated for ever‐attenders and non‐attenders relative to the control arm. This analysis included 161,914 men (72,725 in the SA and 89,189 in the CA) from the core age group (55–69 years at randomisation) [21].

The primary screening test was serum PSA determination. The cut‐off value for diagnostic examination in Finland was 4.0 ng/mL, and 3.0 ng/mL at most other centres for most of the study period. More information is provided in Table S1 [22]. In Finland, from 1996 until 2002, sextant biopsies were used. In 2002, a shift was made to use 10–12 core biopsies. The change was made because 10–12 core biopsies had become commonplace in routine clinical practice. This ensured better comparability between the study arms. In Sweden, sextant biopsies were used from 1995 until 2009, when 10‐core biopsies were adopted. In the Netherlands, sextant biopsies were used from 1993 until the end of follow‐up. In other ERSPC centres, a sextant biopsy was recommended and used from 1996 onwards [22, 23].

In the primary analysis, a man's first, second, and third screening attendances were analysed, regardless of the phase of the screening programme. This allowed for variation in age and year of screening participation within the intervention period. Only screen‐detected PCa cases were included in the primary analysis. PCa cases were categorised according to ISUP GG: low‐risk (Gleason <7 or GG 1), intermediate‐risk (Gleason = 7 or GG 2–3) and high‐risk (Gleason >7 or GG 4–5). CDRs per 1000 screened men were calculated to compare detection across screening attendance (based on individual histories, regardless of screening cycle provided), centres, and GGs.

A secondary analysis was conducted to evaluate trends in PCa detection across the first three screening cycles (screening cycle provided in the programme), stratified by age and screening participation. The screening cycle duration was 4 years, except for 2 years in Sweden. Cycle duration was defined as the follow‐up period, during which all PCa cases were identified for incidence estimation. Altogether, the follow‐up covered three cycles, amounting to 12 years, except for only 6 years in Sweden. This difference in cycle durations between Sweden and the other centres was accounted for in the screen‐detected PCa and secondary analyses. In the primary analysis, the differences in cycle durations were not distinguished because screening attendance was based on individual histories, regardless of the screening cycle. Men were classified into ever‐attenders, non‐attenders, and men in the CA. Non‐attenders were analysed from Finland, Sweden, and Italy, as these centres used a pre‐consent randomisation with 70–80% participation. Data on interval PCa cases were obtained from national cancer registries to evaluate the impact of screening on overall PCa detection. Age‐stratified analyses were also performed to assess the influence of age on CDR, with data combined across all centres. CDR was examined within four age groups: 55–59, 60–64, 65–69, and 70–74 years, based on the age of the men at screening attendance during each screening cycle. A supplementary analysis, using the screening cycle durations defined above, assessed the proportion of interval and non‐attender PCa cases relative to all PCa cases detected within the SA. This analysis was stratified by centre and screening cycle. Table S1 shows the screening intervals and calendar years per screening cycle by centre. The screening cycles ran from 1993–1996 to 2007 in most centres, except in Sweden where a 2‐year screening interval was used. Screening also started later in Switzerland, in 1998, continuing until 2010.

Data analysis was performed using Stata statistical software (version 18; StataCorp LLC, College Station, TX, USA). The datasets generated and/or analysed in the current study are not publicly available because of General Data Protection Regulation restrictions and national regulations in Europe that protect the privacy of study participants. However, data summaries are available from the corresponding author upon reasonable request. The aggregated data from the 16 years of follow‐up are freely available at www.erspc.org.

Results

Overall Results for the ERSPC Trial

CDRs for ISUP GG 1 cases decreased across successive screening cycles in all age groups (Fig. 1). A decrease was observed in GG 2–3 and GG 4–5 cases from the first to the second cycle. However, from the second to the third cycle, CDRs for GG 2–3 and GG 4–5 cases increased in all age groups. In terms of biopsy or assessment referrals, 6517 men (11.7% of attenders) were referred in the first cycle, 6278 (15.8%) in the second, and 3940 (17.0%) in the third (Fig. S1). Throughout all three screening cycles, CDRs for GG 1 cases remained above 10 per 1000 screened men in every age group. These findings were primarily driven by the largest contributing centres: Finland, the Netherlands, and Sweden.

Fig. 1.

Combined data from all centres, analysing only screen‐detected prostate cancer (PCa) cases across three screening cycles, expressed as a proportion of the number of men who attended each specific cycle. Cancer detection rates (CDRs) are calculated per 1000 ever‐attendees. The duration of each screening cycle is capped at 4 years, except in Sweden, where a 2‐year interval was used. CDRs are stratified by the age of the men at each screening attendance into four age groups (55–59, 60–64, 65–69, and 70–74 years) and by International Society for Urological Pathology (ISUP) grade groups (1, 2–3 or 4–5).

Cancer Detection by Centre

Based on screen‐detected cases, CDRs were consistently higher for GG 1 cases than for GG 2–3 and GG 4–5 cases across all centres and screening attendances in the primary analysis. The Dutch centre had the highest CDRs for GG 1 cases in all three screening attendances (Table 1).

Table 1.

Primary analysis.

Centre age	ISUP	Screening attendance 1			Screening attendance 2			Screening attendance 3
		Men (n)	PCa cases	CDR	Men (n)	PCa cases	CDR	Men (n)	PCa cases	CDR
Finland	1	23 771	512	21.5	18 044	452	25.1	10 327	214	20.7
	2–3		108	4.54		95	5.26		107	10.4
	4–5		46	1.94		35	1.94		35	3.39
The Netherlands	1	16 502	502	30.4	12 269	419	34.2	7975	287	36.0
	2–3		195	11.8		95	7.74		42	5.27
	4–5		49	2.97		15	1.22		15	1.88
Sweden	1	4484	127	28.3	3772	94	24.9	3136	96	30.6
	2–3		38	8.47		11	2.92		17	5.42
	4–5		5	2.97		2	0.53		4	1.88
Italy	1	5730	56	9.77	4361	40	9.17	2656	27	10.2
	2–3		31	5.41		9	2.06		13	4.89
	4–5		16	2.79		8	1.83		5	1.88
Switzerland	1	4810	133	27.7	4208	101	24.0	3050	74	24.3
	2–3		28	5.82		23	5.47		48	15.7
	4–5		11	2.29		6	1.43		10	3.28
Belgium	1	3908	51	13.1	1965	74	37.7	573	8	14.0
	2–3		15	3.84		11	5.60		2	3.49
	4–5		6	1.54		5	2.54		2	3.49
Spain	1	1056	15	14.8	778	19	24.4	135	5	37.0
	2–3		7	6.63		3	3.86		2	14.8
	4–5		1	0.95		1	1.29		1	7.41

Number of men screened once, twice, and thrice; screen‐detected PCa cases; and CDR per 1000 men invited, stratified by centre and ISUP grade groups (1, 2–3 or 4–5). Screening attendance based on individual histories, regardless of the screening cycle provided.

CDR, cancer detection rate; ISUP, International Society for Urological Pathology; PCa, prostate cancer.

Trends in CDRs for GG 2–3 cases varied by centre: the CDR increased in Finland and decreased in the Netherlands with each successive screening attendance. In Belgium, the CDR of GG 2–3 cases increased from the first to the second attendance but decreased at the third. In all other centres, the CDRs for GG 2–3 cases decreased from the first to the second attendance but increased from the second to the third (Table 1).

For GG 4–5 cases, CDRs increased in Finland, Belgium and Spain from round to round. In all other centres, the CDRs for GG 4–5 cases decreased from the first to the second attendance but increased at the third (Table 1).

Cancer Detection by Screening Round Attended

At the first screening attendance, the CDRs ranged from 10 to 30 per 1000 men for GG 1 cases, from 5 to 12 per 1000 men for GG 2–3 cases, and from 1 to 4 per 1000 men for GG 4–5 cases. The highest CDRs for all GGs were in Sweden and the Netherlands, and the lowest were in Belgium and Finland (Table 1).

At the second screening attendance, the CDRs for GG 1 cases were comparable to that at the first attendance, except in Belgium and Spain, where CDRs increased. The CDRs for GG 2–3 cases increased in Finland and Belgium and decreased in all other centres. The CDR for GG 4–5 cases increased in Belgium and Spain, stayed the same in Finland, and decreased in all other centres (Table 1).

In the third screening attendance, the CDRs for GG 1 cases remained roughly similar in most centres but decreased in Finland and Belgium and increased in Sweden and Spain. The detection of GG 2–3 cases decreased in the Netherlands and Belgium and increased in all other centres. The CDRs for GG 4–5 cases decreased in Belgium and increased in all other centres (Table 1).

Cancer Incidence During Each Screening Cycle

The secondary analysis incorporated interval PCa cases diagnosed outside the screening protocol among both ever‐attenders and non‐attenders. During the first screening cycle, the RDs for GG 1 cases between ever‐attenders and the control arm exceeded 10 per 1000 men at all centres, reaching 28 per 1000 men in the Netherlands. In the second and third cycles, the RDs remained above 10 per 1000 men in all centres except Italy. Notably, in Italy, the RD of low‐grade cases was higher in the CA than among the ever‐attenders in the third cycle (RD −0.79 per 1000 men) (Table 2).

Table 2.

Secondary analysis.

Centres	Screening cycle 1						Screening cycle 2						Screening cycle 3
	CDR			RD (vs. control)			CDR			RD (vs. control)			CDR			RD (vs. control)
	1	2–3	4–5	1	2–3	4–5	1	2–3	4–5	1	2–3	4–5	1	2–3	4–5	1	2–3	4–5
Finland
Ever‐attender	26.6	4.18	2.26	16.0	1.37	0.59	32.0	7.02	2.65	16.6	−1.41	−1.61	24.8	13.9	4.45	12.1	1.33	−1.64
Non‐attender	9.72	3.03	2.05	−0.92	0.22	0.38	14.1	6.35	3.36	−1.27	−2.08	−0.90	11.7	9.09	4.66	−0.93	−3.52	−1.43
Control	10.6	2.81	1.67	Ref*	Ref*	Ref*	15.4	8.43	4.26	Ref*	Ref*	Ref*	12.7	12.6	6.09	Ref*	Ref*	Ref*
The Netherlands
Ever‐attender	33.8	12.4	3.03	28.2	8.74	0.15	36.2	8.18	1.69	25.1	1.29	−2.48	39.8	6.66	3.19	25.0	−0.59	−1.84
Control	5.64	3.62	2.88	Ref*	Ref*	Ref*	11.1	6.89	4.17	Ref*	Ref*	Ref*	14.8	7.25	5.03	Ref*	Ref*	Ref*
Sweden
Ever‐attender	24.4	7.67	1.10	23.4	7.17	0.93	23.2	2.46	0.70	18.4	0.91	−0.34	42.2	6.39	1.28	33.8	2.48	−0.50
Non‐attender	1.33	0.88	0.44	0.32	0.38	0.27	8.79	1.95	0.33	3.96	0.40	−0.71	12.9	2.53	0.69	4.53	−1.38	−1.09
Control	1.01	0.50	0.17	Ref*	Ref*	Ref*	4.83	1.55	1.04	Ref*	Ref*	Ref*	8.36	3.91	1.78	Ref*	Ref*	Ref*
Italy
Ever‐attender	14.3	6.08	2.47	10.8	3.60	0.40	13.1	4.47	2.98	6.43	0.54	−1.97	10.4	4.67	2.59	−0.79	0.33	−3.92
Non‐attender	3.96	3.46	2.47	0.51	0.98	0.40	9.66	4.83	5.09	2.96	0.90	0.14	7.47	6.54	4.30	−3.69	2.20	−2.21
Control	3.45	2.48	2.07	Ref*	Ref*	Ref*	6.70	3.93	4.95	Ref*	Ref*	Ref*	11.2	4.34	6.51	Ref*	Ref*	Ref*
Switzerland
Ever‐attender	33.9	7.40	2.75	21.7	3.36	−0.48	33.0	9.91	3.48	17.8	2.62	−0.06	26.9	19.4	7.00	18.3	10.1	2.68
Control	12.11	4.04	3.23	Ref*	Ref*	Ref*	15.2	7.29	3.54	Ref*	Ref*	Ref*	8.64	9.29	4.32	Ref*	Ref*	Ref*
Belgium
Ever‐attender	14.0	3.69	1.32	10.4	2.98	1.08	38.1	5.98	2.72	35.3	5.73	2.22	11.4	0	3.80	9.51	0	2.99
Control	3.53	0.71	0.24	Ref*	Ref*	Ref*	2.74	0.25	0.50	Ref*	Ref*	Ref*	1.90	0	0.81	Ref*	Ref*	Ref*
Spain
Ever‐attender	21.8	8.52	3.79	16.5	5.01	0	24.3	3.47	1.74	20.7	0.75	−3.69	0	31.3	31.3	0	22.8	24.7
Control	5.26	3.51	0	Ref*	Ref*	Ref*	3.62	2.72	5.43	Ref*	Ref*	Ref*	7.54	8.48	6.60	Ref*	Ref*	ref*

Ever‐attenders, non‐attenders and controls by centre across three screening cycles. PCa cases stratified by ISUP grade groups (1, 2–3 or 4–5) per 1000 men, with RDs for ever‐attenders and non‐attenders compared with controls. The screening cycle duration is capped at 4 years (except in Sweden, where it is 2 years). Interval PCa cases are included. Non‐attenders were included for Finland, Sweden, and Italy because of sufficient cases per screening cycle. Ref* = Reference group for RD calculations; RD is calculated as the difference in PCa case rate per 1000 men between ever‐attenders or non‐attenders and the control group.

CDR, cancer detection rate; ISUP, International Society for Urological Pathology; PCa, prostate cancer; RD, risk difference.

In the first screening cycle, the RDs for GG 2–3 cases were higher than 3 per 1000 men across all centres, with the highest value in the Netherlands at 8.74 per 1000 men. Only in Finland was the RD lower, at 1.37 per 1000 men. In the second cycle, the CDRs of GG 2–3 cases were lower in the SA than in the CA only in Finland. The results for the third cycle were similar to those for the second cycle, with more GG 2–3 cases detected in the SA than in the CA in all other centres, except the Netherlands (Table 2).

For the ever‐attenders, in the first screening cycle, the RDs for GG 4–5 cases were higher in all other centres, except Switzerland. In the second cycle, the RDs were lower in all other centres, except Belgium. In the third cycle, the RDs were lower in Finland, the Netherlands, Sweden, and Italy (Table 2).

Among the non‐attenders, the RDs of GG 1 cases during the first cycle were slightly higher in the SA than in the CA in Sweden and Italy, whereas in Finland, the detection rate was lower. For the GG 2–3 and 4–5 cases, the detection rates were higher in the SA in all three centres. In the second cycle, the RD for all three GGs was lower in the SA than in the CA in Finland, whereas, in Italy, the RD was higher in the SA for all three GGs. In Sweden, the RDs for GG 1 and GG 2–3 cases were higher in the SA, whereas for GG 4–5 cases, the RD was lower in the SA. In the third cycle, the RDs for all three GGs were lower in the SA in Finland and Italy. In Sweden, the RD for GG 1 cases was higher in the SA, but was lower for GG 2–3 and GG 4–5 cases (Table 2).

Overall, fewer differences between the centres were observed in the CA incidence rates than in the SA rates, with a steady, moderate increase in the CA rates over time in all centres. During the first screening cycle, the CDRs for GG 1 cases in the CA ranged from 1 to 12 per 1000 men, whereas those for GG 2–3 cases ranged from 0.50 to 4 per 1000 men. For GG 4–5 cases, the CDRs ranged from 0 to 3 per 1000 men. During the second screening cycle, the CDRs ranged from 3 to 15 per 1000 men for GG 1 cases in the CA, from 2 to 8 per 1000 men for GG 2–3 cases (except in Belgium, where the CDR was 0.25), and from 0.5 to 5 per 1000 men for GG 4–5 cases. During the third screening cycle, the CDRs in the CA ranged from 2 to 15 per 1000 men for GG 1 cases, from 4 to 13 per 1000 men for GG 2–3 cases, and from 1 to 7 per 1000 men for GG 4–5 cases. The Belgian CA did not have any cases in the GG 2–3 (Table 2).

The proportion of interval and non‐attender PCa cases among all PCa cases in the SA increased at all centres with repeated screening. By the third screening cycle, more than half of GG 1, GG 2–3 and GG 4–5 cancers were interval or non‐attender cases in nearly all centres (Table S2).

Discussion

This study examined screening attendance and the detection of high‐, intermediate‐ and low‐grade cancers across three screening cycles in the ERSPC to provide a comprehensive summary of the screening results. We analysed PCa incidence uniformly and comprehensively, including cases detected both through screening and outside screening, in both arms of the ERSPC trial over 12 years. The incidence of GG 1 cases consistently exceeded that of GG 2–3 and GG 4–5 cases. The highest CDRs for GG 1 cases were in the first screening round in all age groups. Repeated screening showed decreasing detection rates for low‐grade PCa in all age groups, whereas detection of intermediate‐ and high‐grade cases showed no consistent trend. Some centres reported increases, and others decreased. There were notable differences in detection rates between centres, for example in Italy the CDR for GG 1 cases was 9.77 per 1000 men at the first screening attendance, whereas in Sweden and the Netherlands the CDRs were close to 30 per 1000 men. At the third screening attendance the CDRs for GG 4–5 cases were 1.88 per 1000 men in the Netherlands, Sweden and Italy, whereas in all other centres the CDRs were over 3.00 per 1000 men. In the secondary analysis, which also included interval cases, the excess of low‐grade cases among ever‐attenders remained above 10 per 1000 men across the three screening cycles. Conversely, for high‐grade cases, the excess incidence among ever‐attenders decreased in most centres with repeated screening. Ever‐attenders were more likely than non‐attenders and controls to be diagnosed with low‐grade PCa. However, repeated screening did not unequivocally increase the detection of low‐grade PCa. Among ever‐attenders, there was an increase in CDRs for GG 4–5 cases in almost all centres, except in Italy, from the second cycle to the third. For GG 2–3 cases, there was also an increase in CDRs in almost all centres, except for the Netherlands and Belgium. For GG 1 cases, CDRs increased only in the Netherlands and Sweden from the second to the third cycle. The RDs for GG 1 cases were significantly larger in all three screening rounds, except for the third cycle in Italy. The proportion of mostly GG 2–3 cases increased with screening, and a smaller effect was observed for GG 4–5 cases. This upward grade migration might have been caused by the switch to 10–12 core biopsies from sextant biopsies. However, high‐grade PCa was less common than in the CA after the initial screening cycle, suggesting a potential benefit of repeated screening in reducing high‐risk PCa incidence through grade shift. For non‐attenders, the RDs for GG 1 cases were consistently lower across all three screening cycles only in Finland, referring to low rates of contamination.

Reports from individual ERSPC centres have been published [10, 11, 12, 13, 14, 15, 17]; however, these findings are not easily comparable because of a lack of consistency. This paper provides a comprehensive assessment of the entire ERSPC trial, incorporating data from all participating centres. The findings are based on the largest trial, with a screening programme covering repeated rounds over a 12‐year period and cover cancer detection at screening and outside in both trial arms.

In the PLCO trial, PCa diagnoses during the 13‐year follow‐up were classified by clinical stage (I–IV) rather than grade. The SA showed a 1.1‐fold excess relative to the CA for both T1 and T2–4 cases, with similar absolute excess for both groups (7 vs 5 per 1000 men) [24]. Our results demonstrate a larger difference between the SA and CA than in the PLCO trial when considering all three screening cycles. However, the results are not fully comparable because we evaluated CDRs by GG rather than by stage.

The CaP trial included only a single screening round over a 14‐year follow‐up. It reported PCa diagnoses by clinical stage rather than grade. In this SA, CaP reported an absolute excess of 6.97 per 1000 men relative to CA for stage cT1–T2 cases. For cT3 cases, there was no difference between the SA and CA, whereas for cT4 cases, there were fewer cases in the SA than the CA (absolute difference −0.91 per 1000 men) [7]. As with the PLCO trial, direct comparability with our findings is limited because we assessed CDRs by GG rather than stage.

The limitations of our study include smaller sample sizes in some centres, particularly Belgium and Spain, whereas the large centres dominated the results. Variations in screening protocols reduced comparability across centres. Non‐attenders were included only for Finland, Sweden, and Italy because of pre‐consent randomisation in these centres. Inter‐observer variability in Gleason grading introduced inconsistency in the results. The redefinition of Gleason scores in 2005 and 2007 did not materially affect the screening outcomes, as screening was largely completed by 2007. Nevertheless, the grade shift results should be interpreted cautiously because of grade inflation [25, 26]. When analysing GG 1 cases across age groups, the CDRs did not materially increase with repeated screening. However, changes in pathology criteria lowered the threshold for high‐grade cancer classification. Consequently, a substantial proportion of cases diagnosed as GG 1 in the 1990s were reclassified as GG 2–3 in the 2000s. This probably obscured the true effect of repeated screening. Correspondingly, the CDRs for GG 2–3 cases declined only marginally, whereas with consistent grading criteria, the decline would likely have been steeper. The Gleason grading data were collected during the trial follow‐up and did not comply with the current ISUP GG classification. In particular, we were unable to subdivide Gleason 7 cases into grade groups 2 and 3. Furthermore, we lacked data on PSA testing conducted outside the trial. Information on family history was also unavailable. In Finland and Sweden, the increase in biopsy core counts from sextant biopsies to 10–12 cores during the study period may also have contributed to upward grade migration. Finally, as the screening was based on PSA testing only, the applicability of our findings to current diagnostic pathways—incorporating multiparametric magnetic resonance imaging and targeted biopsies—remains uncertain [27, 28]. A key difference is the detection of fewer low‐grade cases with the current diagnostic pathway. However, our findings on changes in cancer detection from initial to repeat screening and in cancer incidence outside screening may inform current policies.

The key strength of our study is the high‐quality data from the large, randomised ERSPC trial, which provides comprehensive, detailed cancer incidence data and long‐term follow‐up.

The current findings underline the complexity of PSA‐based screening for PCa, highlighting notable differences in detection across the ERSPC centres and screening cycles. Although repeated screening consistently identified more low‐grade than high‐grade cases, its impact on the detection of intermediate‐ and high‐risk PCa was inconsistent. Early adaptation of lower PSA cut‐off values during the first screening round in Finland, the Netherlands, Sweden and Belgium may have slightly reduced cancer detection at the first round. In Finland and Sweden, the increased number of biopsy cores may also have increased detection of high‐grade cancers.

In this multicentre analysis of the ERSPC trial, cancer detection patterns varied substantially across screening cycles, GGs, and centres, highlighting both the strengths and the limitations of repeated PSA‐based screening. Detection of low‐grade PCa decreased after the initial screening round. However, assessment of changes across screening rounds was complicated by contemporaneous changes in grading practices, which introduced grade inflation and resulted in migration of cases previously classified as low‐grade into intermediate‐grade categories. Patterns of cancer detection by GG also varied across centres, likely reflecting differences in screening protocols and diagnostic practices. Increasing proportions of interval and non‐attender cancers, particularly by the third cycle, suggest diminishing marginal returns of uniform repeated screening and underscore the influence of adherence and screening fatigue on outcomes. Centre‐specific differences—especially the opposing trends in Finland and the Netherlands—demonstrate that local screening organisation, biopsy thresholds, developments in diagnostic procedures and population characteristics materially affect detection patterns. Altogether, the findings emphasise the need for refined, risk‐adapted screening protocols that can maintain detection of clinically significant disease while reducing unnecessary diagnosis of low‐grade cancers.

Author Contributions

Niko Pasanen: data analyses and presentation, preparation of the initial draft manuscript, and manuscript revision. Kirsi Talala, Luis Gonzàlez, and Sebastiaan Remmers: data curation, manuscript review and commenting. Teuvo Tammela, Kimmo Taari, Elly Den Hond, Vera Nelen, Stephen Wyler, Maciej Kwiatkowski, Donella Puliti, Giuseppe Gorini, Marcos Lujan, Jonas Hugosson, Monique Roobol, Rebecka Arnsrud Godtman, and Chris Bangma: research process, data collection, resource acquisition, project administration, coordination of study conduct, manuscript review and commenting. Anssi Auvinen: study conceptualisation, methodology, resource acquisition, project administration, supervision, manuscript review and commenting.

Disclosure of Interests

Niko Pasanen, Kirsi Talala, Sebastiaan Remmers, Chris Bangma, Kimmo Taari, Jonas Hugosson, Teuvo Tammela, Elly Den Hond, Vera Nelen, Stephen Wyler, Maciej Kwiatkowski, Donella Puliti, Giuseppe Gorini, Luis Gonzàlez, Marcos Lujan, and Monique Roobol have no competing interests to declare. Rebecka Arnsrud Godtman has received lecture fees and travel honorarium from Bayer, IPSEN and Astellas, and Anssi Auvinen has received a lecture fee and travel support from IPSEN.

Supporting information

Table S1. Screening intervals, calendar years per screening cycle and biopsy techniques by centre.

Table S2. Proportion of interval and non‐attender prostate cancer (PCa) cases relative to all detected cases within the screening arm per centre and screening cycle, divided into International Society for Urological Pathology (ISUP) grade groups (1, 2–3, or 4–5). The screening cycle duration was limited to 4 years at all centres, except in Sweden, where a 2‐year cycle was implemented.

Fig. S1. Combined data from all centres, analysing only ever‐attenders across three screening cycles. The figure shows the number and percentage of men referred for biopsy or assessment among ever‐attenders in each screening cycle. Attendance dates are individual and not restricted by the duration of the screening cycles.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available because of privacy or ethical restrictions.