Sandblom et al reported on the 20- year follow-up of their screening trial in Nörrkoping, Sweden.11 In 1987, all men aged 50-69 were identified and every sixth man was randomized to screening. There were 1,449 men in the screening group and 7,532 men in the control group.
The screening interval was every three years from 1987 to 1996. The screening regimen included DRE for the first two occasions and DRE with a PSA (cutoff 4 ng/mL) for the last two. Depending on the screening occasion, 70%-78% of the invited men were compliant with screening.
After a median follow-up of 75 months, no significant difference was observed in PCa mortality between the screened group and controls (RR 1.16, 95% CI [0.78-1.73]).
The above three trials were considered to be of poor-quality based on a high risk of bias by the USPSTF. Accordingly, the primary focus of the USPSTF report7 was on the PLCO and ERSPC trials.
PLCO Cancer Screening Trial
In the PLCO trial, 576,685 men were randomized to screening or usual medical care. The screening regimen consisted of an annual PSA for six years and an annual DRE for four years. All screened men and their caregivers received PSA results; initially a PSA above 4 ng/mL was considered suspicious and these men were advised to seek further diagnostic evaluation. Patients were recruited across 10 U.S. centers.
After 13 years of follow-up, no difference in PCa mortality (RR 1.09, 95% CI [0.87-1.36]) was observed between the groups.5
There was a large degree of prescreening among men enrolled in both arms and contamination of up to 52% of men in the usual care group undergoing a PSA test at some point during the study. This and the fact that 85% of men in the screened arm underwent testing may have affected the power of the trial, resulting in a decreased ability to detect a beneficial effect from screening.12
A subsequent post-hoc analysis examined PCa-specific mortality in young men with no or minimal comorbidity in the PLCO trial. After a median follow-up of seven years, in men with minimal or no comorbidity, there was a significant 44% reduction in the risk of PCa mortality in the screening arm (hazard ratio [HR], 0.56, 95% CI [0.33-0.95], p=0.03).
Moreover, the number needed to treat (NNT) to prevent one PCa death in this subgroup was five.13 Importantly, in this analysis,13 an expanded comorbidity definition was used and the interaction between screening effect and comorbidity was later found to be sensitive to the comorbidity definition used by PLCO investigators.5
These benefits in a population with minimal or no comorbidity are important to consider given the protracted natural history of screen-detected PCa, which can predate a clinical diagnosis by 6-14 years.14 The majority of these patients have organ confined and potentially indolent disease.15 Of the men treated with curative intent, up to one third can be expected to recur;16 however, even following such a recurrence there can be a prolonged survival before death from PCa.17, 18
In light of the well-documented competing mortality risks in men diagnosed with PCa,19, 20 the men expected to benefit the most from a diagnosis and subsequent treatment of PCa would be those who are young with minimal comorbidities.
The ERSPC trial randomized 182,160 men to a screening arm or control group across seven European countries. The screening regimen was dependent on center and year with most centers using a PSA cutoff of 3 ng/mL for biopsy.
For the entire group there was no mortality benefit to screening; however, in a pre-defined “core group” consisting only of men aged 55-69 years (n=162,243), after a median follow-up of 11 years, a 21% reduction in PCa-specific mortality was seen in the screened group (RR 0.79, 95% CI [0.68-0.91]).
This corresponded to an absolute mortality risk reduction of 1.07 deaths per 1,000 men. Researchers concluded that 1,055 men would need to be invited for screening and 37 cases of PCa would need to be detected in order to prevent one PCa death, which has furthered concerns about overdiagnosis of this disease. No difference in overall mortality was noted (RR 0.99, 95% CI [0.97-1.01]).6 Significant reductions in PCa mortality were seen in only two of the centers (Sweden and Netherlands), and removal of both of these centers would result in no mortality benefit for the entire remaining group.
The PCa-specific mortality in the ERSPC was also affected by noncompliance in the intervention arm as well as contamination in the control group. To adjust for these effects, Roobol et al extrapolated the contamination seen in the Rotterdam arm (15.2% true contamination) to the entire ERSPC cohort.21 After adjusting only for noncompliance, PCa mortality risk reduction further increased to 27% (RR 0.73, 95% CI [0.58-0.93]).
With adjustment for both noncompliance and contamination, the PCa mortality risk reduction increased to 29-31% (RR 0.69, 95% CI [0.51-0.92] and RR 0.71, 95% CI [0.55-0.93]),21 which also approached the 44% relative risk reduction seen in the Göteborg trial.10
One concern about ERSPC relates to difference in treatment for men in the screened and control arms. Generally, men in the screened arm were treated at large university centers while control patients were treated in the community. Data have shown statistically significant treatment differences between these groups, even after adjusting for stage differences at the time of diagnosis.22
For example, men in the screened arm with high-risk localized disease were almost twice as likely to undergo radical prostatectomy and half as likely to receive hormonal therapy as men in the control group. Moreover, it is also possible that treatment such as radical prostatectomy (RP) and radiation therapy are better when administered at high-volume (e.g., university) centers as was more common in the screened arm rather than community settings, which predominated in the control arm. These differences may explain some of the mortality difference between the two arms of this study.
In 1994, 19,904 men aged 50-64 in Göteborg, Sweden, were randomized to screening or a control group. Screening was performed with a PSA every two years, with additional evaluation performed for an elevated PSA consisting of DRE and biopsy. The PSA cutoff was noted to have changed over the course of the study. A total of 76% of men in the intervention group attended screening at least once.
After a median follow-up of 14 years, a significant 44% decrease in PCa mortality was noted in the screened group (RR 0.56, 95% CI [0.39-0.82]), with a 56% risk reduction noted in the group of men actually screened (RR 0.44, 95% CI [0.28-0.68]). Based on the risk reduction observed, in order to prevent one PCa death, the NNS was 293 and the NNT was 12.10
Also, it is worth noting that in this very positive trial for screening, it has been estimated that 1,000 men would need to be screened for 14 years to avert five PCa deaths and with that screening about 120 men would have been diagnosed and potentially treated.23
An additional important observation stemming from this trial was that this improvement in mortality was seen in a population where about 30% of men in each arm diagnosed with PCa were managed with active surveillance at last follow-up, suggesting that the risks of overtreatment can be minimized by good clinical management.10
Additional advantages of screening
Certain advantages of screening, which are not addressed by the aforementioned randomized trials, are reviewed below.
Prior to the PSA screening era, a majority of men diagnosed with PCa had advanced or metastatic disease.24 With the introduction of screening, a marked increase in the diagnosis of asymptomatic, clinically localized disease has been observed.15, 25, 26 Catalona et al reported locally advanced tumors in 57% of a comparison group to 29% in a group undergoing serial PSA screening.25
The Cancer of the Prostate Strategic Urological Research Endeavor (CaPSURE) revealed an increase in clinical T1 disease from 23% in 1993-5 to 50% in 2002,27 and SEER data from 2004-5 showed that 94% of newly diagnosed men have localized disease (T1 or T2).15 A decrease in distant or metastatic disease was seen following the introduction of screening,28, 29 with a 40% reduction noted in the screening group in the ERSPC.30
Mortality benefit from epidemiologic studies
A sustained decrease in PCa mortality has been noted since the early 1990s. This decline began as PSA testing was beginning to become widespread. In the prior decade, RP had become widely practiced, modern forms of radiation therapy had been introduced and more men were receiving hormonal therapy.
Since then there have been annual PCa mortality decreases of 4.1% from 1994-2005 and 2.6% from 2005-2007.31 Age-adjusted PCa mortality was also seen to decrease more rapidly (1990-2005) in the U.S., where screening was more prevalent, than in the U.K.32 Similarly, greater improvements in PCa mortality were noted with initiation of population-based screening and treatment in Tyrol, Austria, compared to other regions of Austria without screening.33 Although modeling has attributed 45%-70% of this observed PCa mortality reduction to PSA screening,34 other groups have cautioned that this improvement may result from many additional factors35 such as the contribution of improved treatments36, 37 or attribution bias.38