Screening for Prostate Cancer: A Review of the ERSPC and PLCO Trials

Abstract

The advent of prostate-specific antigen (PSA) testing in the early 1980s revolutionized the diagnosis of prostate cancer. As a result of PSA testing, there has been a surge in the number of prostate cancer diagnoses. This review examines the results of 2 recent landmark trials that studied the effect of screening on prostate cancer mortality: the European Randomized Study of Screening for Prostate Cancer (ERSPC) and the US-based Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial.

Prostate cancer poses a significant problem for men’s health; it has become the most common malignancy and the second most common cause of cancer death in American men. It is estimated that 1 in 6 men will be diagnosed with prostate cancer at some time in their lives, and more than 30,000 men died of the disease in 2002.¹ The advent of prostate-specific antigen (PSA) testing in the early 1980s revolutionized the diagnosis of prostate cancer, and, as a result, there has been a surge in the number of prostate cancer diagnoses.

Similar to other common malignancies, such as breast and cervical cancer, population screening with this effective tumor marker appears enticing, and the American health care model has advocated PSA screening since the early 1990s. This review examines the results of 2 recent landmark trials: the European Randomized Study of Screening for Prostate Cancer (ERSPC)¹ and the US-based Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial.² The results of these trials have contributed significantly to our understanding of the effects and efficacy of prostate cancer screening, and its difficulties. Both trials examined mortality as the endpoint, and both found little effect on mortality from screening.

Prostate Cancer Screening

The ideal screening test is minimally invasive, readily available, easily performed, acceptable to the general population, accurate, and significantly affects the outcome of disease (such as the mortality rate).³ The current standard, PSA testing combined with digital rectal examination (DRE), is minimally invasive and easily available, but does not seem to be ideal in reducing mortality, as the results from the ERSPC and PLCO trials suggest.

ERSPC

The ERSPC study used data from 7 centers in different European countries, with a total of 162,387 men undergoing randomization. Of these, 72,952 men were assigned to the screening group and 89,245 men were assigned to the control group. Randomization was 1:1 in all countries except Finland, where the randomization of the whole birth cohort led to a ratio of 1:1.5 for the screening group to the control group. Slightly different methods and follow-up routines were used; PSA cutoff varied from 3 to 4 ng/mL and serum PSA levels necessitating further testing ranged from 2.5 to 3.9 ng/mL. It is unclear how much screening was present in the control group throughout the study period. Results between the study centers were shown to be generally similar, and no anomalies were found in screening or detection rates. Intervals for the screening group were large-4 years for 87% of patients.

With average and median follow-up times of 8.8 and 9.0 years, respectively, there were 214 prostate cancer deaths in the screening group and 326 in the control group. For the screening group, this results in an unadjusted rate ratio for death of 0.80 (95% confidence interval [CI], 0.67–0.95; P = .01), and an adjusted rate ratio of 0.80 (95% CI, 0.65–0.98; P = .04) (Table 1). In other words, to prevent 1 death from prostate cancer, 1410 (95% CI, 1132–1721) men need to be screened and 48 men treated (Table 2). After adjusting for noncompliance, 1068 need to be treated and the rate ratio after 9 years was 0.73 (95% CI, 0.56–0.90). It was additionally suggested that the population that benefited from screening was restricted to men between the ages of 55 to 69 years, and that other age groups did not show a reduction in mortality through screening.

Table 1.

European Randomized Study of Screening for Prostate Cancer (ERSPC) Results

Study Participants	Cancer Incidence	Cancer Mortality
Total, age 55–69 years	162,243	10,297
Screening group	72,890 (44.9%)	5990 (8.2%)	214	0.80*
Control group	89,353 (55.1%)	4307 (4.8%)	326

^*95% confidence interval, 0.67–0.95; P = .01.

Data from Schröder FH et al.¹

Table 2.

European Randomized Study of Screening for Prostate Cancer (ERSPC) Screening Group Methodology and Outcomes

Positive PSA tests	20,437 (16.2%)
Biopsies	17,543
Biopsy compliance	85.8%
False positive (biopsy after elevated PSA)	13,308 (75.9%)
Men needed to screen to prevent 1 death	1410*
(adjusting for noncompliance)	1068
Men needed to be treated to prevent 1 death	48

^*95% confidence interval, 1142–1721.

PSA, prostate-specific antigen.

Data from Schröder FH et al.¹

PLCO Study

In the PLCO trial, 76,693 men at 10 US study centers were included. The screening group consisted of 38,343 men and the control group consisted of 38,350 men. Randomization was done within blocks of the population stratified according to center, age, and sex. Men in the screening group received annual PSA screenings, whereas those in the control group were not actively screened but sometimes received screening outside of the study, resulting in a contaminated population.

The incidence of death per 10,000 person-years was 2.0 (50 deaths) in the screening group and 1.7 (44 deaths) in the control group (rate ratio, 1.13; 95% CI, 0.75–1.70) (Table 3). All deaths related to any of the PLCO cancers were reviewed by people unaware of the assignments to either screening or control group, thus minimizing possible contamination regarding treatment. The importance of this has been shown by previous research that suggests that the cause of death is less likely to be reported as related to prostate cancer when the subject is receiving attempted curative treatment for the disease.⁴ The majority of cancers found were diagnosed at stage II, were nearly all adenocarcinomas, and more than 50% had a Gleason score of 5 to 6. These findings did not differ between the screening and control group. More advanced-stage cancer diagnoses (stage III or IV) were also similar between the 2 groups, although Gleason scores of 8 to 10 were higher in the control group (341) than in the screening group (289). These results show that, after an average 7 years of follow-up, the mortality did not significantly differ between the 2 groups. Therefore, in this study, screening was not associated with mortality (rate ratio, 1.13).

Table 3.

Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial Results at 10 Years

Study Participants		Cancer Incidence		Cancer Mortality
Total, age 55–74 years	76,693
Screening group	38,343 (50.0%)	3452	1.17*	92	1.11†
Control group	38,350 (50.0%)	2974		82

^*95% confidence interval, 1.11–1.22.

^†95% confidence interval, 0.83–1.50.

Data from Andriole GL et al.²

Discussion

The ERSPC and PLCO trials are extraordinary data sets. Their analysis is highly complex and provides remarkable clarity to the results and conclusions. The magnitude of patients enrolled, 182,000 and 76,693 men in the ERSPC and PLCO trials, respectively, is unprecedented. The randomization of patients was highly sophisticated in both trials, with methodologies resulting in near-perfect distributions of men. This was retested in the ERSPC trial, where age distributions and death rates from all causes were compared in the screening and control groups throughout the study period.

Patient compliance was very high, and is made more impressive by the length of the study period. The PLCO trial had especially high levels of compliance (Table 4): at 7 years vital status was known for 98% of participants, and at 10 years it was known for 67%. The compliance rates for testing were additionally very high, 85% for PSA and 86% for DRE. In the ERSPC trial, 82.2% of men in the screening group were screened at least once, and overall compliance was better in those study centers that had obtained consent before the beginning of the study (88%–100%) than in those that underwent randomization before obtaining consent (62%–68%). The follow-up and compliance data lay the groundwork for an extremely comprehensive population analysis.

Table 4.

Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial Rate of Screening and Compliance

Control Group	Percentage
Rate of PSA screening in control group
Year 1	40
Year 6	52
Rate for those having undergone no more than 1 PSA test at baseline (89%)
Year 1	33
Year 6	46
Rate of DRE screening	41–46
Screening Group
Compliance in the screening group
Vital stats known, 7 years (% baseline population)	98
Vital stats known, 10 years (% baseline population)	67
PSA compliance	85
DRE compliance	86

DRE, digital rectal examination; PSA, prostate-specific antigen.

Data from Andriole GL et al.²

The long follow-up is especially important in the study of prostate cancer, which is often very slow to develop, especially in older men. Follow-up times of 9 and 10 years for the ERSPC and PLCO trials, respectively, are useful and necessary, especially for large populations. The PLCO study will continue to an estimated follow-up of at least 13 years. Extensive follow-up is important due to several factors: (a) the prolonged natural history of prostate cancer, with death of the host from prostate cancer possibly occurring decades from the time of diagnosis; (b) the typically advanced age of men when diagnosed with prostate cancer, with the effect of competing causes of death from other comorbidities; and (c) the detection of indolent tumors in the general population. Albertsen and colleagues⁵ demonstrated that many pre-PSA screening era patients, when followed without treatment, were destined to die of causes other than prostate cancer.

Although neither trial found great differences in mortality, there were results unassociated with the endpoint that are valuable when discussing screening, and the apparent levels of overdiagnosis and overtreatment are an important finding. Reviewing the Surveillance, Epidemiology, and End Results (SEER)⁶ data, the rising gap between the incidence and mortality rates in the PSA screening era can be indicative of increasing rates of overdiagnosis. The declines in mortality are quite small compared with the large number of men diagnosed and treated for prostate cancer. This may imply that even if prostate cancer mortality could be completely eradicated, it would be accomplished at the expense of substantial overtreatment. Recent studies have shown an additional worrying side effect of overdiagnosis of prostate cancer: the effects of diagnosis on the patient’s quality of life. Patients with clear indolent cancers suffer from the diagnosis, and report that the most important reason for seeking and undergoing active treatment is anxiety, not disease progression.⁷ Rather than answering questions, the ERSPC trial has added to the discussion. If 1410 men need to be screened and 48 treated to prevent 1 cancer death, does the benefit of treatment outweigh the risks? This is a question that is not easily answered, and is likely to provide food for thought for patients, urologists, and health care providers for years.

The issue of false-positive results was examined using data from both trials. It was demonstrated that increased prostate screening results in a high rate of false-positive results; 15.0% of DRE and 10.4% of PSA tests resulted in false-positive results based on biopsy.⁸ Prior research has shown that PSA cutoffs are unreliable. It has been shown that a serum PSA level higher than 3 ng/mL is falsely positive for 75% of patients.⁹ Rates of overdiagnosis in the PLCO trial were high, with estimates of diagnosis as high as 50% in men who would not show clinical symptoms during their lifetime.¹⁰ ERSPC trial results showed that sextant biopsies, triggered by an elevated PSA level, did not detect cancer in 3 out of 4 (75%) men. No deaths were directly associated with biopsies during the trial, although previous studies have reported complications with prostate biopsies as well as other screening procedures. Minor complications, such as minor rectal hemorrhage or bleeding from the urethra, were found in around half of biopsied men,¹¹ and a very small number, 0.05%, required hospitalization from complications after a biopsy.¹² Biopsy remains a safe procedure for prostate cancer detection.

Monitoring false-positive results is important because they can also have a psychologic impact on patient health. McNaughton-Collins and colleagues¹³ reported that 49% of men who received a false-positive result and a later, confirmed normal result thought about prostate cancer either “a lot” or “some of the time” compared with only 18% of those with a normal serum PSA level (P < .001). Such results raise a number of interesting questions regarding the impact of diagnosis on a patient’s psychologic well-being. The 13-year follow-up of the PLCO may provide additional answers.

Although the randomization of the population was near perfect and resulted in highly comparable populations, the contamination of the control group is a concern in both studies. The ERSPC report does not describe the control group and its possible screenings. Thus, it is unclear how many patients in the control group were screened and how this unidentified number affected the results. Because there were study centers in different countries, it is possible that the control groups underwent different levels of screening or none at all. It is therefore difficult to assess the level of homogeneity in screening within the control group. The PLCO trial took measures to minimize contamination before it began randomization by excluding men who had had more than 1 PSA test in the 3 years previous to 1995. The trial assesses the screening in the control group by regular surveys, reporting that 9.8% of the control group did in fact have repeated screenings during the study period. An average of these and those who had never been screened was taken to assess contamination of the whole control group. In the PLCO trial, the level of screening in the overall study population was high: 44% of men had at least 1 PSA test and 55% had at least 1 DRE in the past 3 years.

Age at the time of enrollment trials may have further added to the contamination in both the PLCO and ERSPC trials. Recommendations by the American Medical Association¹⁴ state that men above 55 should be screened annually. Because patients up to 75 years of age were enrolled in both trials, the study population was most likely, at least in part, already screened. Carter and colleagues¹⁵ investigated the influence of age on the chance of curable prostate cancer among men with nonpalpable disease. Younger age was found to be associated with greater probability of curable cancer and more likely to lead to a decrease in prostate cancer mortality. Similarly, Smith and colleagues¹⁶ demonstrated that younger age at the time of diagnosis is an independent predictor of better prognosis. The earlier age at diagnosis and stage migration has created a lead-time of at least 3 to 5 years. This lead-time bias is an important consideration in studies demonstrating an improved survival in the PSA screening era. Both studies support the notion that screening is more important among the younger population. The ERSPC trial openly states that “the benefit from screening was restricted to the core age group of subjects who were between the ages of 55 and 69 at the time of randomization,” and the PLCO trial states that their results support “the validity of the recent recommendations of the US Preventative Services Task Force, especially against screening all men over the age of 75 years.” Ethnic variations are not explored in either trial, and the lack of recruitment of African American men, who have a higher risk of prostate disease, may have affected the results of the US-based PLCO trial.

The control group and overall population prescreening data make the 2 trials more difficult to compare. The PLCO trial regards those men in the control group as contaminants only if they had repeated screenings, defined as at least twice in 7 or 10 years of follow-up. This may lead to a control group in the PLCO trial that is similar to those in the screening arm of the ERSPC trial. This does not diminish their individual importance, but does make it more difficult to compare the 2 studies. Differences in the serum PSA levels defined as a positive result within and between trials could also cloud comparisons. ERSPC used different levels in the different centers in their study and PSA cutoff varied from 3 to 4 ng/mL, with rates necessitating further testing ranging from 2.5 to 3.9 ng/mL; the PLCO study used 4 ng/mL as the uniform cutoff rate. Studies in the PSA screening era have demonstrated a fall in serum PSA level at the time of diagnosis: it has decreased from 11.8 ng/mL in 1990 to 6.3 ng/mL in 1998.^17,18 Generally, a lower PSA value at diagnosis has been associated with better pathologic outcomes and disease-free survival.¹⁹ Yet, the high levels of overdiagnosis in these trials show that this association may not be as clear as previously thought.

Stage migration has been one of the most significant changes in the PSA screening era. Catalona and colleagues²⁰ first reported a decrease in advanced prostate cancer in the screened population in 1993. In comparison with the referred population from the same institute, 70% of the prostate cancers detected with PSA screening were pathologically organ confined, in contrast to 51% in the referred population. According to the Center for Prostate Disease Research (CPDR) database, the percentage of patients presenting with metastatic disease decreased from 19.8% in 1989 to 3.3% in 1998.²¹ Concurrently, data from other American registries also documented falling age-adjusted incidence rates of distant metastasis by approximately 50%.²² Roehl and colleagues²³ found that more than 60% of the prostate cancers in the PSA screening era were clinical T1c tumors, compared with more than 70% of the clinical T2 tumors in the pre-PSA screening era. The results from the ERSPC and PLCO trials muddy these waters. The PLCO trial (Table 5) found that the characteristics of patients were similar in both groups, and that regardless of screening or control group or mode of detection, the majority of tumors were stage II at diagnosis. The study, moreover, found little difference between the detection at other stages, with screening and control groups showing similar results. The complication at this point is the contamination of the control group, as it cannot be determined whether there was actually no benefit to those screened in terms of tumor stage or if the control group was screened to an extent where the effects of screening rivaled those of the annual tests.

Table 5.

Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial Mortality According to Tumor Stage and Screening Rate

Parameter	Screening Group	Control Group	Rate Ratio (95% CI)
Cancer incidence	3452	2974	1.17 (1.11–1.22)
Cancer mortality	92	82	1.11 (0.83–1.50)
Death according to tumor stage
Stage I or II (%)	60	52
Stage III (%)	2	4
Stage IV (%)	36	39
Death according to rate of screening
≦ 1 PSA test at baseline, at 7 years	48	41	1.16 (0.76–1.76)
≦ 1 PSA test at baseline, at 10 years	83	75	1.09 (0.80–1.50)
2 or more PSA tests in previous 3 years at baseline, 7 years	2	3	0.70 (0.12–4.17)
2 or more PSA tests in previous 3 years at baseline, 10 years	9	7	1.34 (0.50–3.59)

Data from Andriole GL et al.²

Conclusions

The findings regarding the risk of overdiagnosis and overtreatment remain the most intriguing aspect of the current prostate cancer screening discussion. Both the ERSPC and PLCO authors mention the need for further studies that assess the relationship between prostate cancer screening, treatment, and quality of life (QoL). These are especially important if results continue to show little impact on mortality as well as increasing stress placed on the patient through overdiagnosis and overtreatment. It has been shown that there is a difference in QoL between different treatments for prostate cancer. For example, with retropubic radical prostatectomy (RRP) and permanent brachytherapy (BT), RRP patients scored better in overall QoL than those receiving BT, except in the months following surgery.²⁴ Such studies could be used as a starting point for future screening studies evaluating QoL during screening and diagnosis.

As the results are ambiguous concerning mortality, the question of how to screen and treat to prevent mortality remains. The PLCO trial suggests that contamination of the control group through DRE is less problematic than contamination through PSA; only 25% of control group patients have had DRE compared with 48% with screened serum PSA levels. Consequently, DRE may be a worthwhile test for future examination.

Main Points.

• The European Randomized Study of Screening for Prostate Cancer (ERSPC) and the US-based Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial recently reported on the mortality benefit of prostate-specific antigen screening.

• The decline in mortality rates are quite small compared with the large number of men diagnosed and treated for prostate cancer.

• Both studies mention the need for further investigations that assess the relationship between prostate cancer screening, treatment, and quality of life. This is especially important if results continue to show little impact on mortality and increasing stress placed on the patient through overdiagnosis and overtreatment.