SEARCH

SEARCH BY CITATION

Keywords:

  • prostate cancer screening trials;
  • randomized, controlled trial;
  • PSA

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1

Two large-scale randomized screening trials, the Prostate, Lung, Colorectal and Ovary (PLCO) cancer trial in the USA and the European Randomized Screening for Prostate Cancer (ERSPC) trial in Europe are currently under way, aimed at assessing whether screening reduces prostate cancer mortality. Up to the end of 1998, 102,691 men have been randomized to the intervention arm and 115,322 to the control arm (which represents 83% of the target sample size) from 7 European countries and 10 screening centers in the USA. The principal screening method at all centers is determination of serum prostate-specific antigen (PSA). The PLCO trial and some European centers use also digital rectal examination (DRE) as an ancillary screening test. In the core age group (55–69 years), 3,362 of 32,486 men screened (10%) had a serum PSA concentration of 4 ng/ml or greater, which is 1 cut-off for biopsy (performed in 84%). An additional 6% was referred for further assessment based on other criteria, with much less efficiency. Differences in PSA by country are largely attributable to the age structure of the study population. The mean age-specific PSA levels are lower in the PLCO trial (1.64 ng/ml [in the age group 55–59 years], 1.80 [60–64 years] and 2.18 [65–69 years) than in the ERSPC trial (1.28–1.71 [55–59], 1.75–2.87 [60–64] and 2.48–3.06 [65–69 years]). Detection rates at the first screen in the ERSPC trial range from 11 to 42/1,000 men screened and reflect underlying differences in incidence rates and screening procedures. In centers with consent to randomization design, adherence in the screening arm is 91%, but less than half of the men in the target population are enrolled in the trial. In population-based centers in which men were randomized prior to consent, all eligible subjects are enrolled, but only about two-thirds of the men in the intervention arm undergo screening. Considerable progress has been made in both trials. Enrollment will be completed in 2001. A substantial number of early prostate cancers have been detected. The differences between countries seem to reflect both underlying prostate cancer incidence and screening policy. The trials have the power to show definitive results in 2005–2008. © 2002 Wiley-Liss, Inc.

Prostate cancer causes more deaths than any other cancer in men aged 55–74 years in many industrialized countries.1 In the past, one-third of all prostate cancers were diagnosed at a metastatic stage with an unfavorable prognosis.2 Since the introduction of prostate-specific antigen (PSA) as a biologic tumor marker in the 1980s, a substantially higher proportion of prostate cancers has been diagnosed in early stages than before.

Unfortunately, no valid results from randomized treatment trials are available as yet to determine the benefits of radical prostatectomy or radiotherapy. Nevertheless, early lesions are usually treated by surgery or radiotherapy with curative intent or followed up intensively in order to start treatment when needed (watchful waiting). Advanced disease is generally treated palliatively by hormonal treatment and/or radiotherapy. Many believe that treating the cancer by local eradication (surgery/radiotherapy) before it has metastasized may prohibit further progression and thus will lead to reduced mortality from the disease.

Screening healthy men for prostate cancer to detect early cancers has been shown to be feasible and acceptable in large studies.3–5 Whether screening reduces prostate cancer mortality can best be determined on the basis of large randomized screening trials such as those launched in the early nineties in Europe and the USA.6, 7 In the USA, participants in the Prostate, Lung, Colorectal and Ovary cancer (PLCO) trial are recruited at 10 different screening centers. In Europe, the European Randomized Screening for Prostate Cancer (ERSPC) trial recruits men from 7 countries (with large differences in incidence rates).

The PLCO trial in the USA and the ERSPC trial in Europe have joined forces to ensure sound and efficient evaluation of the screening programs. This article presents results concerning design (target populations, screening intervention) and results from the initial screening round (enrollment, PSA distribution, biopsy and detection rate as well as characteristics of screen-detected cancers). The results presented here are essential for the assessment of program performance and eventually for interpretation of the final outcome of the trials.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1

Design

The design and outline of planned evaluations of the PLCO and ERSPC trials have been previously described.6, 7 The PLCO trial has a strict joint protocol followed by all centers, whereas the ERSPC centers share a common core protocol, which allows for variability in the target population and screening procedure (Table I). The age of the participants at enrollment varies by country between 50–74 years.

Table I. Key Characteristics of Study Protocols in ERSPC and PLCO Trials1
 ERSPCPLCO USA (10 sites)
BelgiumFinlandItalyPortugalSpainSwedenThe Netherlands
  • 1

    PSA, prostate-specific antigen; DRE, digital rectal examination; TRUS, transrectal ultrasonography; PopReg, population registry; CaReg, cancer registry; F/T, free-to-total PSA.

  • 2

    The cut-off level for PSA/criteria for application of other screening tests.

Age at entry (yr)55–7455/59/63/6755–7050–7450–7051–6655–7455–74
Screening interval (yr)44414241
Type of trialVolunteerRandomized populationRandomized populationGP consults VolunteerVolunteerRandomized populationVolunteerVolunteer
RandomizationAfter consentBefore consentBefore consentAfter consentAfter consentBefore consentAfter consentAfter consent
Period of initial recruitment1992–991996–991996–991994–991996–991995–961994–20001993–2001
Identification of source populationPopRegPopRegPopRegGPs/Elect. Regist.PopRegPopRegPopRegMail, mass media
Exclusion criteria
 Preexistent prostate cancerYesYesYesYesYesYesYesYes
 Prior PSANoNoNoNoNoNoNo>1 last 3 yr
Target sample size
 Intervention group8,75022,5007,0008,5002,4169,97321,21037,000
 Control group8,75045,0007,0008,5001,8629,97321,16637,000
Screening tests/biopsy2
 PSA (ng/ml)4.04.04.04.04.13.04.0/3.04.0
 DREAllPSA driven 3.0–3.9/ F/TPSA driven 2.5–3.9AllNoNoAll/1.0/NoAll
 TRUSAllNoPSA drivenNoNoNoAll/1.0/NoNo
Source of follow-up information
 IncidenceActiveCaRegCaRegActiveActiveCaRegCaRegActive (annual study update)
 MortalityPopRegPopRegPopRegPopRegActivePopRegPopRegActive
 Death certificateAll, reviewAll, reviewAll, reviewAllAll, reviewAllAll, death reviewAll, review
End points
 Death from prostate cancerYesYesYesYesYesYesYesYes
 Quality of lifePossiblyYesYesPossiblyPossiblyYesYesPlanning
 Cost effectivenessPossiblyYesYesNoPossiblyYesYesPlanning
BiorepositorySerum, prostate tissueSerumNoNoNoSerum, prostate tissueSerum, buffy coatSerum, cells, tissue, buffy, red blood cells
Pilot studyConductedConductedConductedConductedConductedConductedConductedConducted

The core age group targeted by nearly all countries is between 55–69 years; the Swedish center includes men aged 51–66 at the time of the first invitation; the Finnish centers recruit men aged 55, 59, 63 or 67 at the time of the first invitation. The age range in the PLCO trial has been modified from the original 60–74 to include men 55–59 years of age.

In the Portuguese centers, men were initially randomized on visiting their general practitioner (instead of being invited to visit the physician). Now all centers represent either a randomized population-based or a volunteer-based sample. Population-based trials utilize population registers to identify the entire target population, and men are subsequently randomized to either the screening or the control arm of the trial (Finland, Italy and Sweden). In these trials adherence to screening in the intervention group is less than 80–90%, but this should not be a major problem unless compliance approaches 50% or less.

In volunteer-based trials men give their informed consent prior to randomization. Such trials are the first choice where randomization before consent is deemed unethical (USA, The Netherlands), or a sufficiently high participation rate cannot be achieved using a population-based approach (Belgium, Portugal and Spain) or a population register is lacking (USA). Although the volunteers may not necessarily be representative of the whole population, they are likely to resemble the group that would participate if screening was introduced as a routine policy. Several volunteer-based European centers identify the target population from population registers and invite all eligible men to participate. Local and/or national policies determined the approach to obtaining informed consent. Countries randomizing volunteers after obtaining their consent are required to ask for informed consent in both intervention and control groups, whereas those randomizing subjects identified from the population registry prior to consent are not obliged to contact the control group. Written informed consent is obtained from each subject undergoing screening.

All centers had conducted pilot studies before entering the trials and have been approved by a local institutional (medical ethics) review board. All centers apply computerized individual randomization.

Data

A minimum data set has been established for evaluation, with which all countries have agreed to comply.7 The local centers report annually to the Data Monitoring Committee of the ERSPC for process monitoring. In addition, all European centers enter their data into a centralized database, where logical and quality checks are performed. All data from the 10 PLCO centers are transmitted electronically on a monthly basis to a coordinating center. In this article, the most recent data available are reported (until the end of 1998 for most centers) regarding first screens.

Unlike the PLCO trial, the ERSPC trial will be analyzed with stratification by center due to differences in prostate cancer incidence, screening protocol and study population. In Europe, the 3- to 4-fold differences in underlying incidence are crucial for interpreting detection rates, because the 2 may well be correlated. The screening protocol also influences both referral and detection rates. Furthermore, the ERSPC sample size is sufficiently large to allow stratified analysis; the number of men in several European centers is up to 50–80% of the total size of the PLCO. The 3 smaller ERSPC centers (Spain, Portugal and Belgium) demonstrate huge differences in incidence. In the ERSPC trial, the common protocol allows differences in both the screening intervals and the screening tests.

A PSA test is the principal screening method at all centers, and a level ≥4 ng/ml is an indication for further assessment in all centers except Spain. All centers use the Hybritech assay, except Sweden, which uses the Wallac assay. Lower PSA cut-off levels are used in several centers. Several centers also use digital rectal examination (DRE) and/or transrectal ultrasound (TRUS) as additional screening tests. Some modifications to the screening procedure have been introduced since the previous report:7 the Finnish centers now use a PSA of 3.0 ng/ml instead of 2.0 ng/ml as a cut-off for further assessment, and 2 new centers, one in Florence (Italy) and one in Getafe (Spain) have joined the ERSPC trial. In Rotterdam (The Netherlands), 3 screening tests (PSA, DRE and TRUS) were performed initially on all participants; later, only those with a PSA of 1.0 ng/ml or higher received further screening tests (DRE and TRUS), and currently only subjects with PSA levels of 3 ng/ml or higher are referred for biopsy.8 TRUS has now been abandoned as a screening test in the ERSPC trial. Most ERSPC centers apply a 4-year screening interval, whereas the PLCO trial maintains a yearly screening interval. The PLCO trial is a multicenter trial with one screening protocol: PSA and DRE.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1

Recruitment

Approximately 218,000 men have thus far been enrolled in the trials (Table II), of whom 102,691 have been allocated to the intervention arms and 115,322 to the control arms. The arms are not of equal size because not all centers apply a 1:1 randomization (see Table I). The Scandinavian centers and The Netherlands have enrolled 80% of the men aged 55–69 years in the ERSPC. The target number of men in the intervention arms is 120,000; in the control arms is this 140,000. The PLCO trial is aiming at an accrual of 74,000 subjects and the ERSPC centers are currently aiming at 185,000. By the end of 1998, approximately 83% of this target had been reached.

Table II. Recruitment in ERSPC/PLCO Trials +Up to 1998–1999
 No. of men allocated to intervention group, all ages% Men screened (compliance rate)No. of men allocated to control group, all agesNo. of new entrants to date, all agesAnticipated target, ages 55–69
All agesAges 55–69 at entry
  • 1

    At the March 2001 Madrid meeting of the ERSPC, it was agreed not to include the Portugese center in the final analysis of the ERSPC trial, since the center had stopped its activities and follow-up data could not be ascertained.

ERSPC trial
 Belgium4,04287.787.94,0068,0486,400
 Finland22,73269.069.035,97358,70575,800
 Italy7,47768.268.57,47214,94914,049
 Portugal1,04690.090.61,0652,1111
 Spain2,416100.0100.01,8624,2781,494
 Sweden9,97358.761.79,97319,94613,200
 The Netherlands21,21094.294.821,16642,37634,771
  Total68,89681,517150,413 (81%)145,714
 PLCO trial (USA)33,79533,80567,600 (91%)
Total to date both trials102,691115,322218,013 (83%)

In The Netherlands, all men identified from the municipality registrations are contacted and asked for consent prior to randomization: 46% of the targeted men aged 55–69 agreed to participate in the trial and were subsequently randomized. In Belgium and Spain, 25–29% participated and were randomized. In centers with consent before randomization, the adherence in the screening arm is 88–100% (on average 91%). In the ERSPC centers with randomization of subjects identified from population registries and a passive follow-up, the control group does not need to be informed. In these centers, the participation rate in the screening arm varies from 59 to 69% (on average 64%).

In the PLCO trial, the study subjects are identified from commercially available databases and advertisements. Hence, little information on the target population is available. Surveys suggest that approximately 3% of eligible men participate in the PLCO trial. All men surviving (without a diagnosed prostate cancer) are invited for rescreening after 1 year; the attendance rate at the second screen for PSA is 92% of participants in the first round and 91% for DRE. The attendance rates at the third screen have so far been 93% and 92%, respectively. In Sweden, the rescreening interval is 2 years, and the attendance rate at the second screening round (including those previously not attending for screening) is currently 56%. Of the screenees in the second round, 90% were being screened for the second time and 10% for the first time, having failed to attend initially.

Age categories at entry

The core age group targeted by nearly all centers is 55–69 years of age at entry. Approximately 83% of all study subjects belong to this age group at randomization. Three ERSPC centers (Sweden, Portugal and Spain) also recruit younger men (50–54 years of age), and 3 ERSPC centers (the Netherlands, Belgium and Portugal) as well as the PLCO trial include men over 70 years of age. No important differences have occurred in the age distribution of men in the study group compared with the control group (results not shown). In Europe, almost half of the subjects are younger than 60 years of age at entry, whereas the largest age group in the PLCO trial is 60–64 years of age (Table III). The age range varies within the ERSPC trial: the proportion of men aged 55–59 years is largest in Finland, Sweden and Spain, and smallest in Belgium. The mean age of men in the intervention arm within the core age group varies by center from 60–63 years.

Table III. Age Subdivisions of Men Aged 55–69 Years at Entry (%) Intervention Group for Different Centers
 55–5960–6465–69Total 55–69Mean age (55–69)Mean age (all ages)
No.%No.%No.%
  • 1

    Center recruits only men aged 55/59/63/67 at entry.

  • 2

    All are 65 at entry.

ERSPC27,1834617,8943013,9792459,05660
 Belgium552171,507471,157363,2166364.6
 Finland113,349584,901224,4822022,73260.260.2
 Italy2,454352,384342,181317,01961.962.4
 Portugal3962750735543381,44662.665
 Spain8135049331305191,61160.257.7
 Sweden3,301502,76442548286,61359.757.3
 The Netherlands6,318385,338314,7632916,45961.963.6
PLCO USA5,769288,484416,4473120,70062.163.5
Total32,96226,37820,42679,756

PSA distribution

Serum PSA concentrations were below 1 ng/ml in 33–50% of screened men (Table IV). Serum PSA concentrations of 4 ng/ml or higher, calling for further assessment in all centers, were detected in 7–15% of screened men. The differences between countries are mainly related to the age distribution of enrolled men. Approximately 40% of all men belong to an intermediate PSA category (1–4 ng/ml) in which the optimal policy for further assessment is relatively unclear and protocols vary between centers.

Table IV. Distribution of PSA Levels and Mean/Median Level First Screens (All Ages) in Different Centers and Mean PSA Levels Per Age Category
CenterPSA level (ng/ml), distribution (%)PSA level (ng/ml), all agesMean age of screened menMean PSA (ng/ml), by age category
<1≥1, <2≥2, <3≥3, <4≥4, <7≥7, <10>10MeanMedian55–5960–6465–69
  • 1

    Exclusion of three outliers (mean PSA = 3.4 if including outliers).

  • 2

    Ages 55/59/63/67 only.

  • 3

    Different PSA assay (Wallac), possibly 10% lower than Hybritech.

  • 4

    Age 65 only.

Belgium133301589332.71.4651.712.122.82
Finland246291155221.91.1611.512.152.77
Italy43301067222.01.1621.581.752.48
Portugal37301387233.31.5651.282.872.68
Spain46301245211.71.0571.701.852.49
Sweden35029955111.81.0571.532.343.064
The Netherlands36311379222.31.363.51.562.062.92
USA (PLCO trial)42301276211.91.263.51.641.802.18

The mean age of screened men is clearly correlated with the mean and median PSA and the percentage of men with PSA levels of 4 ng/ml or higher (columns 9–11). The differences in PSA distribution between the countries are much smaller after stratification by age (Table IV; last 3 columns). Even though the mean age of the men taking part in the PLCO trial is higher than in the ERSPC trial, the age-specific mean PSA levels are lower. The mean PSA concentrations were relatively high in Sweden (where no subjects older than 65 were recruited and a different assay was used) and Portugal (where subjects were randomized on visiting the general practitioner). Very high PSA levels among some men diagnosed with prostate cancer may influence the mean levels.

A screening protocol with additional tests (such as a suspicious DRE or TRUS finding) and a lower PSA cut-off level increase the proportion of screen-positive men. At most centers, an additional 1–7% of the subjects screened are referred for biopsy based on a suspicious DRE, TRUS or lower cut-off level of PSA (Fig. 1). In Rotterdam, where all the participants initially received 3 different screening tests, up to 12% of the subjects with PSA levels below 4 ng/ml are referred for biopsy.

thumbnail image

Figure 1. Percentage of screened men with positive screen result (suspicious for cancer; advice for referral and biopsy) per country and subdivided for PSA level of 4 ng/ml or higher (black bars) and lower levels (white bars) in which other screening tests or cut-offs are being used (all ages).

Download figure to PowerPoint

Biopsies and cancer detection rates

Of the participants in the ERSPC trial, 69–94% with a PSA level ≥4 ng/ml underwent biopsy. However, one center had a substantially lower biopsy rate (38%). Acceptable reasons for nonadherence to the protocol include refusal of the participant to revisit the center, medical contraindications, treatment for other prostate disease or intercurrent death.

A similar detection rate (in men with PSA levels ≥4 ng/ml) emerged throughout the ERSPC, i.e., 21–28%, except for Italy (low underlying incidence) and Belgium (low biopsy rate). Comprehensive information on the detection rate was not yet available for the PLCO trial. The detection rates in Portugal and Spain seem relatively high, considering the underlying incidence, although the numbers of cases remain low. The overall detection rates (including all age groups and PSA levels) were between 11 and 42 per 1,000 men. The differences largely reflected underlying incidence rates, which were highest in the USA, followed by Finland, The Netherlands, Sweden and Belgium and lowest in Southern Europe. The low detection rate in Belgium was probably due to a low biopsy rate. The percentage of men biopsied with PSA levels below 4 ng/ml varied by center due to the different protocols but was generally low (6%). Less than 1% of all cancers were detected through other tests (DRE/TRUS) and lower PSA cut-off levels, indicating that these tests were less effective in detecting cancers among men with lower PSA concentrations (lower part of Table V).

Table V. Number of Screened Men, Biopsy Rate and Detection Rate for Men with PSA ≥ 4 (ng/ml) and for Men with PSA < 4 (Ages 55–69) in ERSPC First Screens1
 TotalBelgiumFinlandItalyPortugalSpainSwedenNetherlands
  • 1

    In this table, data are restricted to screens that have complete follow-up (biopsy, biopsy result, cancer detection), so total screens differ from those given in Table II.

  • 2

    Spain cut-off used PSA 4.1.

PSA ≥ 4 (direct indication for biopsy)
 No. of screened men (= also positive screen)3,362 (10%)360 (13%)428 (8%)520 (11%)84 (12%)96 (8%)2354 (9%)1,250 (11%)
 % Biopsies carried out8438937887699493
 % Cancers detected in screened men2011231323212628
PSA < 4 (other test or lower cut-off indication for biopsy)
 No. of screened men29,1242,4334,6254,2936291,1343,70410,161
 % Biopsies carried out, based on positive screen640.90.81n.a.510
 % Cancers detected in screened men0.70.60.20.30.3n.a.0.71
Totals
 No. of men screened32,4862,7935,0534,8137131,2304,05811,411
 No. of prostate cancers detected per 1,000 screened men2719211629112342

Incidence data were available for most centers, allowing assessment of the relationship between the detection rate and the underlying incidence of prostate cancer (prevalence/incidence ratio). The P/I ratio is in the range of 14–20 for most centers (except The Netherlands and Portugal; Table VI). Consent before randomization, wide criteria for diagnostic assessment and the randomization of men visiting general practitioners were associated with higher detection rates. However the P/I ratio also depends on screening performance, natural history of prostate cancer and prior screening history of men enrolled, which may vary by center.

Table VI. Underlying Age-Specific Prostate Cancer Incidence Rates (Per 100,000) Before Screening Center Started, Detection Rate at First Screen (Per 1,000 Men Screened) and Detection Rate at First Screen by Estimated Trial Incidence Rate Before Screening (55–69 Years Old at Entry)
 Country and last years of reporting incidence before start of trial
Spain (ref. 13) '86–'92Italy (Florence) (ref. 13) '88–'91Portugal1 '93Belgium2 '89–91Sweden (ref. 13) '88–'92Netherlands (ref. 13) '90–'94Finland (ref. 22) '95USA SEER (white)3 (ref. 13) '83–'87
  • 1

    Lisbon Cancer Registry; unpublished data.

  • 2

    National Cancer Register and National Institute for Statistics 1989–1991, Antwerp; unpublished data.

  • 3

    SEER, Surveillance, Epidemiology, and End Results Program.

  • 4

    Estimate based on age structure of study population and age-specific incidence data.

  • 5

    Age 55–65 years.

  • 6

    Biopsy rate PSA ≥ 4: 38% (ratio approx. 30 when assuming same probability of detection in nonbiopsied 62% of men).

Underlying incidence rate
 Ages 55–59 yr2123224472556691
 Ages 60–64 yr58727377192141210217
 Ages 65–69 yr126160149194370298417438
Estimated underlying incidence rate in trial study population (55–69)45480881141425152155256
Detection rate first screen (55–69)111629192354221?
Detection rate first screen (55–69) divided by estimated incidence rate before screening (55–69)2020331761652814?

Stage and Gleason score

In the 3 centers where information on stage distribution was available, 70% of screen-detected cancers were clinically localized (cT1c, cT2 without distant metastases; Table VII). Approximately 6% were poorly differentiated (Gleason scores 8–10), 76% moderately (Gleason scores 5–7) and 15% highly differentiated. In the absence of a screening program, approximately 17–25% would have been poorly differentiated tumors.3, 9

Table VII. Number and Clinical Stage Distribution of Prostate Cancers for the Netherlands, Sweden and Finland (Restricted to Core Age Group and First Screen-Detected Cancers)
Gleason scorecT1cNxMocT2NxMoCT3-T4NxMoM1Not recordedTotal (%)
2–4134445413200 (15)
5–7408290128102011,039 (76)
8–1014292281487 (6)
Unknown151220635 (3)
Total (%)571 (42)375 (28)157 (11)22 (2)236 (17)1,361

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1

The trials include both population-based and volunteer-based centers with either randomization-to-consent or consent-to-randomization design. In centers requiring consent prior to randomization, both the screening and the control arm undergo a similar selection process, which ensures comparability between the arms. The adherence is very good, but the participants comprise only a small proportion of the target population. Furthermore, as the subjects are aware of the possibility of screening, the control group may be more prone to contamination. A comparison of all cause mortality and prostate cancer mortality between the study subjects and the entire target population is needed to assess possible selection effect. In population-based randomization prior to consent (commonly obtained only from subjects in the intervention arm), the screening arm is representative (by definition given an intent-to-screen analysis), and contamination may not pose a serious problem. However, the proportion of men in the intervention arm actually screened is relatively low, which may dilute the screening effect and decrease power.

Recently, data have become available from the small, randomized Quebec trial on prostate cancer screening,10, 11 boasting approximately 31,000 men in the intervention arm and 15,000 in the control arm. Given the fact that a minimum sample of 150–200,000 participants is assumed to be required to show a statistically significant effect on prostate cancer mortality reduction, this can be seen as an informative, but low power trial. Tellingly, it is a randomized population-based trial with only a 23% acceptance rate in the intervention arm. The intention-to-screen-analysis revealed no statistically significant result on prostate cancer mortality reduction.12

Obviously, the data from this trial are informative, but too limited for comparison with our trials. For instance, no age-specific data for the group aged 55–69 have been published. The mean age at enrollment was approximately 53.5 (range 45–80), much younger than in our trials; only some 4,000 first screens are limited to the 55–69-year age group (compared with 64,000 to date in our trials). In that trial, PSA levels have not been broken down into the same categories as presented here. The detection rate at first screens has been reported to be either 3011 or 40/1,000.10 Although the province of Quebec has been reported to have high underlying incidence rates,13 the relatively young ages of men enrolled would lead to an estimated rate of 120 (per 100,000). Dividing the detection rate by this estimated incidence rate yields an outcome of 25; a high rate compared with those obtained for the ERSPC/PLCO trials. The size/metastasis stage distributions of screen-detected cancers cannot be compared (due to different staging systems).

Prostate cancer mortality reduction due to PSA testing has not been established so far. Only indirect indications have emerged from ecologic and time series data in the USA before and after widespread PSA testing.14, 15 Introduction of large-scale PSA testing in a largely unscreened (or ineffectively screened) population has been shown to increase prostate cancer incidence rates.16, 17 Other factors, such as changes in treatment policy, can easily mask the effects of screening, because screening will favorably affect only a small proportion of those undergoing screening, which in turn comprises only a small proportion of the total population. Observational data, unlike randomized screening trials, offer several sources of error, including selection bias, lead-time bias and length-biased sampling.

The detection of a substantial number of curable prostate cancers with a corresponding reduction in advanced disease is a prerequisite for achieving a reduction in prostate cancer mortality in the ERSPC and PLCO trials. The high detection rate in the prevalence round with a high prevalence/incidence ratio gives some indication of this target being met. When screening is effective, tumors in their preclinical phase are detected long before they would have become apparent clinically (as an incident case). The detection rate relative to the underlying incidence rate (prevalence/incidence ratio) gives an indication of lead time, i.e., how early these tumors have been found compared with a situation without screening.

These trials, as well as a number of serum bank studies,18–20 suggest that the mean duration of the preclinical detectable stages (related to P/I ratio) may well be 5–10 years, although this is by no means constant and can show substantial variability.21 However, the incidence of prostate cancer rises extremely rapidly with age, and hence our result suggesting a P/I ratio of 14 or more is an overestimate. If the risk for prostate cancer is higher in the study population compared with the general population due to selection bias, a relatively high detection rate can be expected. Low enrollment of the target population increases the selection effect. The difference in detection rates between the 3 European centers (28–33 per 1,000) may reflect a selection effect (with higher incidence rates in the enrolled population than average), better (performance of) screening procedures or a much worse stage distribution in the clinical setting (which might also be related to selection). In Portugal, the selection effect may be enhanced by enrollment based on visits to the general practitioner. The Netherlands and Belgium had the most intensive screening policies of all the European centers; 3 screening tests were called for, which is probably reflected in the higher detection rate than expected. This was counterbalanced in Belgium by the low biopsy rate, without which a very high P/I ratio would have been expected. The Dutch and Swedish screening algorithms are comparable, although the P/I ratio was higher in the Netherlands, possibly due to selection caused by the consent-to-randomization design. Underlying incidence rates were obtained from population-based cancer registries with a high completeness of coverage,13, 22 except for Portugal and Belgium where there may be some underregistration.

Unfortunately, no comprehensive results on detection rate are available yet from the PLCO trial. Assuming a similar P/I ratio, a detection rate of 36–51 per 1,000 may be expected. A lower rate and a lower mean PSA level per age category may indicate a different screening history of the men enrolled compared with the ERSPC/PLCO participants, who are screened for 2 other cancers as well, which may affect the selection effect. PSA levels in volunteers have been shown to be different from PSA levels in asymptomatic blood donors.21 The PLCO trial does enroll volunteer participants who could therefore be different from the general population.

The first results from quality of life studies indicate that the screening tests do not seem to have a clear negative impact on the health-related quality of life of the men involved. Biopsies are bothersome in the short term but have no measurable impact on quality of life.23 The extent of overdiagnosis, i.e., detection of prostate cancers that would never have surfaced clinically, is difficult to evaluate at present. However, given the relatively long lead time and the advanced ages at which men are being screened, this may well be substantial. Earlier diagnosis and treatment inevitably cause adverse effects that may be relatively severe and frequent. However, their significance relative to possible favorable quality of life effects due to the reduction of advanced disease remains unclear. Assessment of quality of life is therefore important for evaluation of the balance between possible favorable and adverse effects of screening.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1

Considerable progress has been made in the recruitment for the ERSPC and PLCO trials: more than 215,000 subjects, or over 80% of the target sample size, have been randomized thus far. Both trials are expected to complete enrollment by the year 2001, including several new screening centers. Combining data from the 2 trials will enable us to reach conclusions sooner and with improved precision. The 2 trials jointly comprise 200,000 male participants and will have sufficient statistical power (90% at the 0.05 one-sided significance level) to detect a 20% difference in prostate cancer mortality between the intervention and control arms in 2008 (when assuming a participation rate of 90%, 10% contamination in the control arm and prostate cancer mortality rates similar to those in The Netherlands).

There are important differences between the centers, but the common design, core age group and key features of the screening intervention will allow joint evaluation. In the USA, the PLCO trial involves an intensive screening program, with yearly screens in a volunteer population, whereas the European trial (ERSPC) has a more conservative screening policy, with longer screening intervals and a more representative study population. Even though the differences between centers may not be optimal for maximizing the statistical power, they provide an opportunity to compare different screening procedures and intervals. Possible small differences in risk of the enrolled men do not hamper the evaluation of prostate cancer mortality.

So far, the possibility of the anticipated 20–30% prostate cancer mortality difference between the study and the control groups (between 2005 and 2008) cannot be assessed. The program performance of 2 large-scale randomized prostate cancer screening trials is acceptable and indicates that obtaining proper evidence of the effects of screening for prostate cancer on mortality and quality of life will be feasible.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1

We thank Mr. P. Smith, chair of the ERSPC Data Monitoring Committee, for comments on the manuscript, F. Alexander, as well as all persons contributing to the trials. Hybritech Inc. (USA) provided a grant to the board of the ERSPC that allowed application of the same PSA assay in nearly all centers. Independence in the execution, analysis and interpretation of the results was guaranteed by the research contract.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1
  • 1
    World Health Organization. World health statistics annual 1995. Geneva: WHO, 1996.
  • 2
    Schmidt JD, Mettlin CJ, Natarajan N, et al. Trends in patterns of care for prostatic cancer, 1974–1983: results of surveys by the American College of Surgeons. J Urol 1986;136: 41621.
  • 3
    Catalona WJ, Richie JP, Ahmann FR. Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol 1994;151: 128390.
  • 4
    Mettlin C, Lee F, Drago J, et al. The American Cancer Society National Prostate Cancer Detection Project. Findings on the detection of early prostate cancer in 2425 men. Cancer 1991;67: 294958.
  • 5
    Carter HB, Epstein JI, Chan DW, et al. Recommended prostate-specific antigen testing intervals for the detection of curable prostate cancer. JAMA 1997;277: 14756.
  • 6
    Kramer BS, Gohagan J, Prorok PC, et al. A National Cancer Institute sponsored screening trial for prostatic, lung, colorectal, and ovarian cancers. Cancer 1993;71: 58993.
  • 7
    Auvinen A, Rietbergen JBW, Denis LJ, et al. International Prostate Cancer Screening Trial Evaluation Group. Prospective evaluation plan for randomized trials of prostate cancer screening. J Med Screening 1996;3: 97104.
  • 8
    Schröder FH, Maas P van der, Beemsterboer P, et al. Evaluation of the digital rectal examination as a screening test for prostate cancer. J Natl Cancer Inst 1998;90: 181723.
  • 9
    Lu-Yao GL, Yao SL. Population-based study of long-term survival in patients with clinically localised prostate cancer. Lancet 1997;349: 90610.
  • 10
    Labrie F, Candas B, Cusan L, et al. Diagnosis of advanced or noncurable prostate cancer can be practically eliminated by prostate-specific antigen. Urology 1996;47: 2127.
  • 11
    Labrie F, Candas B, Dupont A, et al. Screening decreases prostate cancer death: first analysis from the 1988 Quebec prospective randomized controlled trial. Prostate 1999;38: 8391.
  • 12
    Boer R, Schröder FH. Quebec randomized controlled trial on prostate cancer screening shows no evidence for mortality reduction [letter]. Prostate 1999;40: 1301.
  • 13
    Parkin DM. Cancer incidence in five continents, vols. VI, VII. WHO, IARC, Lyon: IARC 1992, 1997.
  • 14
    Stephenson RA, Smart CR, Mineau GP, et al. The fall in incidence of prostate carcinoma. On the down side of a prostate specific antigen induced peak in incidence—data from the Utah Cancer Registry. Cancer 1995;77: 13428.
  • 15
    Smart CR. Prostate cancer facts and fiction [editorial]. J Surg Oncol 1997; 66: 2239.
  • 16
    Boer R, Warmerdam P, Koning Hde, van Oortmarssen G. Extra incidence caused by mammographic screening [letter]. Lancet 1994;343: 979.
  • 17
    Jacobsen SJ, Katusic SK, Bergstralh EJ, et al. Incidence of prostate cancer diagnosis in the eras before and after serum prostate-specific antigen testing. JAMA 1995;274: 14459.
  • 18
    Stenman U-H, Hakama M, Knekt P, et al. Serum concentrations of prostate specific antigen and its complex with α1-antichymotrypsin before diagnosis of prostate cancer. Lancet 1994;344: 15948.
  • 19
    Parkes C, Wald NJ, Murphy P, et al. Prospective observational study to assess value of prostate specific antigen as screening test for prostate cancer. BMJ 1995;311: 3403.
  • 20
    Gann PH, Hennekens CH, Stampfer MJ. A prospective evaluation of plasma prostate-specific antigen for detection of prostatic cancer. JAMA 1995;273: 28994.
  • 21
    Reissigl A, Horninger W, Fink K, et al. Prostate carcinoma screening in the county of Tyrol, Austria. Cancer 1997;80: 181829.
  • 22
    Finnish Cancer Registry—Institute for Statistical and Epidemiological Cancer Research Cancer incidence in Finland 1995. Cancer Statistics of the National Research and Development Center for Welfare and Health. Helsinki: Cancer Society of Finland Publication No. 58; 1997.
  • 23
    Essink-Bot M-L, Koning HJde, Nijs HGT, et al. Short-term effects of population-based screening for prostate cancer on health-related quality of life. J Natl Cancer Inst 1998;90: 92531.

APPENDIX 1

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  9. APPENDIX 1
Additional Author Addresses
Anssi Auvinen
Tampere School of Public Health
University of Tampere
Medisiinarinkatu 3
FIN-33014 Tampere, Finland
Stefano Ciatto
Dept. Diagnostic Medical Imaging
Centro per lo Studio e la Prevenzione Oncologica
Viale Volta 171
I-50131 Florence, Italy
Fernando Calais da Silva
Consultant Urologist
Hospital Desterro
Av. Elias Garcia, 81-6°
1050 Lisbon, Portugal
Louis Denis
Oncology Center Antwerp
Apartment 2, Blok H
Lange Gasthuisstraat 35–37
B-2000 Antwerp, Belgium
John K. Gohagan
Chief, Early Detection Branch
National Coordinator PLCO
National Cancer Institute, NIH
Executive Plaza North, Room 330
6130 Executive Blvd
Bethesda, MD 20852, USA
Matti Hakama
Dept. of Public Health
University of Tampere
PO Box 607
FIN-33101 Tampere 10
Finland
Jonas Hugosson
Ass. Professor
Dept. of Urology
Sahlgrenska University Hospital Östra
S-41345 Göteborg, Sweden
Ries Kranse
Rotterdam Cancer Registry
IKR
PO Box 289
3000 AG Rotterdam, The Netherlands
Vera Nelen
Provinciaal Instituut voor Hygiene
Kronenburgstraat 45
B-2000 Antwerpen 1, Belgium
Philip C. Prorok
Chief, Screening Section
Co-National Coordinator/PLCO
Biometry Branch
National Cancer Institute, NIH
Executive Plaza North, Room 344
6130 Executive Boulevard
Bethesda, MD 20852, USA
A. Berenguer Sanchez
Hospital Universitario de Getafe
Carretera Madrid-Toledo, km 12,5
S-28905 Getafe (Madrid), Spain
Fritz H. Schröder
International coordinator ERSPC
Department of Urology
Erasmus University and Academic Hospital
Rotterdam
Dr. Molewaterplein 40, Room H1074
3015 GD Rotterdam, The Netherlands