Landmarks in prostate cancer screening
Professor Fritz Schröder, Erasmus University Medical Centre, Rochussenstraat 125, room NH-224, 3000 CA Rotterdam, The Netherlands. e-mail: email@example.com
- • Prostate-specific antigen (PSA) has been widely applied to diagnosis and follow-up of prostate cancer, which led to research on its potential role in the early detection of the disease and its use in screening.
- • The value of PSA screening in reducing disease mortality is controversial and several studies have been conducted to determine the actual benefits. One of the early studies, the Tyrol Screening Study conducted in 1993, showed that during 2004 to 2008 there was a significant reduction in prostate cancer mortality in men aged >60 years compared with the mortality rate during 1989 to 1993.
- • Two studies that showed no benefit of screening in terms of prostate cancer death were conducted in Sweden in 1987 and 1988.
- • The Prostate, Lung, Colorectal, and Ovarian Screening Study conducted in the USA during 1993 to 2001 and involving 76 693 men showed no benefit of screening at 10 years but the trial can be criticised due to excessive contamination of the unscreened group.
- • In contrast, the European Randomized Study of Screening for Prostate Cancer (ERSPC), the largest randomised study with 162 388 participants study, showed that at a median follow-up of 9 years a prostate cancer mortality reduction of 20% resulted (P= 0.04). In an analysis limited to four ERSPC centres with a follow-up of 12.0 years, screening resulted in an overall reduction of metastatic disease of 31%.
- • The arguments against PSA screening include the risks associated with screening tests themselves, e.g. biopsy-related haematuria, urosepsis, and over diagnosis and overtreatment of prostate cancer. The overall evidence points in favour of PSA screening and steps can be taken to avoid overtreatment by offering patients active surveillance.
American Cancer Society
European Randomized Study of Screening for Prostate Cancer
Prostate, Lung, Colorectal, and Ovarian (Screening Study)
randomised controlled trial
USA Preventive Services Task Force
In 2008, the worldwide estimate for new prostate cancer cases diagnosed was 899 000 and for deaths, 258 000 . Prostate cancer is the most common non-skin cancer in men in Europe, with an estimated 382 000 cases occurring in 2008 . Almost 90 000 deaths from prostate cancer were estimated to have occurred in 2008 in Europe, ranking it the third most common cause of cancer death amongst men, after lung and colorectal cancers . Prostate cancer is also the most common cancer in American men . The 2011 American Cancer Society (ACS) estimates for prostate cancer in the USA are that 240 890 new cases of prostate cancer will be diagnosed and 33 720 men will die from the disease . The ACS estimates that one man in six will be diagnosed with prostate cancer during his lifetime . Prostate cancer is the second leading cause of cancer death in American men, behind only lung cancer. About one man in 36 will die from prostate cancer.
ROLE OF PSA IN DIAGNOSIS
PSA is a kallikrein-like serine protease produced almost exclusively by the epithelial cells of the prostate. It is organ-specific, not cancer-specific and serum levels may be elevated in the presence of BPH, prostatitis and other non-malignant conditions. DRE was primarily used in the diagnosis of prostate cancer but the addition of the PSA test has improved the sensitivity, specificity and positive predictive value compared with DRE alone . Increasing levels of PSA are linked to an increased likelihood of prostate cancer; however, there is no specific threshold for distinguishing insignificant cancers, which are not likely to be life-threatening from those that are significant .
The suggested use of PSA in the diagnosis of prostate cancer  led to research on its potential role in the early detection of the disease and its use in screening. The principle aims of screening are to reduce disease mortality and to increase quality of life years. There is considerable controversy concerning PSA screening and its benefits are disputed. This paper will review the evidence from the principle screening studies that have been conducted on PSA screening.
THE TYROL SCREENING STUDY
One of the first studies to examine the benefits of PSA screening was conducted in 1993 in the Tyrol, Austria as a non-randomised register study and involved comparing a screened population from this area with the unscreened area of the rest of Austria . In all, 65 123 men aged between 47 and 75 years were eligible for screening. Of that population, 32.3% actually underwent screening and over two-thirds were tested at least once during the 5-year study period. The age-specific reference ranges for PSA levels for men aged 40–49, 50–59, 60–69 and 70–79 years were 0–1.25, 0–1.75, 0–2.25 and 0–3.25 ng/mL, respectively. Men who were found to have an elevated PSA level underwent a second PSA test 6–12 months later followed by further investigations (DRE, TRUS and biopsy) as required. Biopsy criteria were based on age-referenced PSA levels combined with a percentage free PSA of <22%. These criteria were changed in October 1995 to half of the PSA age-specific reference range plus a percentage free PSA of <18%. An additional biopsy selection parameter of PSA transition zone density was subsequently added.
Results from the study showed that the incidence of prostate cancer increased from 1988 to 1993 and subsequently remained constant while the incidence of organ-confined disease (Stages I and II) continued to increase until 1998. The incidence of extraprostatic disease has declined since 1994 as has the incidence of metastatic disease (Stage IV) since 1993. For mortality from prostate cancer, this decreased considerably in the Tyrol between 1993 and 1999 compared with a modest decrease in the rest of Austria. The trend in rates of decrease showed a significant difference (P= 0.006). There were 22 fewer deaths from prostate cancer than expected in Tyrol in the age range 40 to 79 years between 1986 and 1990 and 18 fewer deaths in 1999. The conclusions from the study were that the decline in prostate cancer mortality was probably due to aggressive down staging and successful treatment.
An updated analysis of the study has recently been published with data extending to 2008 . Results showed that from 2004 to 2008 there was a significant reduction in prostate cancer mortality in men aged >60 years (risk ratio 0.70) compared with the mortality rate during 1989 to 1993.
THE QUEBEC SCREENING STUDY
A PSA screening study conducted in Quebec Canada in 1988 has published 11-year follow-up data . In all, 46 486 men aged 45–80 years were randomised to PSA screening (31 133) or no screening (15 353); the upper limit of normal for PSA level was 3.0 ng/mL. DRE was conducted at the first screening visit. TRUS was conducted for elevated PSA levels and biopsy carried out on any hypoechoic areas. The study group in this report has chosen to analyse according to ‘screening received’, which has led to a heavily criticised distortion of the results. The actual number of men screened was 7348 and of these, 10 died from prostate cancer. This compares with 74 prostate cancer deaths in the final group of 14 231 unscreened men. Over the 11-year period, the annual cause-specific death rate incidences were 19.8 and 52.3 per 100 000 man-years in the invited screened and control groups, respectively (P < 0.002). This equates to a 62% reduction in prostate cancer death in the screened group.
This study has been criticised on several methodological points. Due to a high cross-over rate between the two arms, the data were analysed according to whether the participant actually received screening or not, which deviated from the standard randomisation analysis and breaks from the randomisation nature of the trial making it an observational study. In addition, baseline demographic characteristics of the men enrolled are not described, which does not permit comparisons to be made. If the analysis was conducted on an intent-to-treat basis the relative risk would be 1.09, indicating no difference in mortality between the two arms .
SWEDISH SCREENING STUDIES
Two PSA screening studies have been conducted in Sweden. In 1987, every sixth man aged 50–69 years living in Norrköping was invited to undergo screening with DRE and later PSA testing every 3 years . The remaining 7532 men not invited acted as controls and this quasi-randomised pilot study was not powered to detect a significant difference in prostate cancer mortality. Detected were 43 (3%) prostate cancers in the screened group plus 42 interval cancers. This compared with 292 (4%) prostate cancers in the control group. There were no differences in prostate cancer or overall mortality.
The second study involved screening of 1782 men aged 55–70 years, with a single screening involving DRE, PSA testing and TRUS with resultant biopsy for suspected prostate tumours; an unscreened group comprised 27 204 men . Follow-up was 15 years (median 12.9 years). There was no effect of screening on the risk of death from prostate cancer and other causes of death (incidence rate ratio 1.10). Interestingly there was a decrease in any cause mortality rate in the screened population compared with the source population. There are several important limitations to this study. The PSA threshold for biopsy was >10 ng/mL, and the screening approach was a simultaneous use of PSA, DRE and ultrasound, making applicability to current practice challenging. In addition, there is a possibility of misattribution of death bias as it is unclear whether the reviewers were ‘blinded’ to study arm allocation.
THE PROSTATE, LUNG, COLORECTAL, AND OVARIAN (PLCO) SCREENING STUDY
The PLCO Screening Study was a larger screening study conducted in the USA during 1993 to 2001 and involving the randomisation of 76 693 men to annual screening (38 343) or usual care (38 350) . Men in the screening group were offered annual PSA testing for 6 years and DRE for 4 years. Compliance rates for PSA testing were 85% and for DRE 86%. The incidence of prostate cancer per 10 000 person-years after 7 years of screening was 116 (2820 cancers) in the screening group and 95 (2322 cancers) in the control group. There was no difference in prostate cancer mortality between the screened and control populations after 7–10 years follow-up. 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 at 7 years; the data at 10 years were similar. One criticism of the trial was the excessive contamination rate in the control group, which was estimated to be in the range of 53%. In addition, >40% of men had undergone PSA testing before randomisation. Such testing is likely to have reduced the number of aggressive or deadly cancers in both arms, which is reflected in the reported results. Showing an effect of an intervention, in this case PSA-driven screening that was on-going to such a large extent before randomisation, is unlikely. Additionally, the biopsy rate was only ≈40% in men who were positive at screening . Follow-up is stated as 11.2 years, but this was only 98% complete at 7 years. There is no way to reverse these negative effects, even with the planned follow-up to the year 2014. In summary, the PLCO trial did not comply with minimum standards for a trial, i.e. having sufficient power to demonstrate an effect.
THE EUROPEAN RANDOMIZED STUDY OF SCREENING FOR PROSTATE CANCER (ERSPC)
The ERSPC was initiated in 1991 for men aged 50–74 years from seven European countries . In all, 72 952 men were randomised to PSA screening at a 2- or 4-year interval and 89 435 to no screening (control group). These numbers and all outcome data refer to the common core age group of 55–69 years. Most centres used a PSA threshold value of 3.0 ng/mL as an indication for biopsy, except Finland, where the PSA threshold value of ≥4.0 ng/mL was used with ancillary testing in the PSA range of 3–3.9 ng/mL. During a median follow-up of 9 years, the cumulative incidence of prostate cancer was 8.2% in the screened group and 4.8% in the control group. With an average follow-up time of 8.8 years there were 214 prostate-cancer deaths in the screening group and 326 in the control group. The rate ratio for death from prostate cancer in the screening group, as compared with the control group, was 0.80 (95% CI 0.65–0.98; adjusted P= 0.04) indicating a significant 20% relative risk reduction of death from prostate cancer in relation to the control group. The absolute risk difference was 0.71 deaths per 1000 men. From these data, also taking into account the difference in incidence of prostate cancer between the two arms, it was found that in order to prevent one death from prostate cancer, 1410 men would need to be screened and 48 cases of prostate cancer would need to be treated. This finding was shown to be sensitive to the duration of follow-up, and these unfavourable numbers are expected to decrease with time .
Four of the ERSPC centres: Finland (Tampere), Sweden, Switzerland and The Netherlands, have complete follow-up data available on the occurrence of metastatic disease in the screen and control arms and evaluated the effect of screening on the prevention of metastatic prostate cancer (F. Schröder, personal communication). Patients entered into the study were aged 55–69 years and were randomised to no screening or screening at an interval of 4 years in The Netherlands, Finland, and Switzerland and 2 years in Sweden. Metastatic disease was defined according to the following criteria: as a positive bone scan with or without X-ray confirmation, or positive CT or MRI evidence, and a PSA level of >100 ng/mL. At a median follow-up of 12.0 years, screening resulted in an overall risk reduction of metastatic disease of 31%.
THE GÖTEBORG RANDOMISED SCREENING TRIAL
The Göteborg screening trial forms part of the ERSPC and included 20 000 men aged 50–64 years randomised to PSA screening at 2 year intervals or to a control group (not screened) . The primary endpoint of the study was prostate-cancer specific mortality. In men randomised to screening, 7578 (76%) of 9952 attended at least once. Because of different legal regulations in Sweden, all participants could be randomised without informed consent on one day in 1994. During the resulting median follow-up of 14 years, 1138 men (12.7%) in the screening group and 718 (8.2%) in the control group were diagnosed with prostate cancer, (hazard ratio 1.64; 95% CI 1.50–1.80; P < 0.001). The absolute cumulative risk reduction of death from prostate cancer at 14 years was 0.40%. The rate ratio for prostate cancer death was 0.56 translating into a 44% relative reduction of the risk of death from the disease in the screening arm of the study (P= 0.002). After adjustment for non-participation, the rate ratio was 0.44, which is compatible with a significant relative reduction of the risk of death from prostate cancer in men actually screened of 56%. Overall, 293 (95% CI 177–799) men needed to be invited for screening and 12 to be diagnosed to prevent one prostate cancer death. As in the ERSPC study as a whole, follow-up will continue and regular updates published. In addition, in line with the agreements dating to 1994, all centres may publish their own results independently after the ‘first final analysis’ was published in 2009. A separate report of the Rotterdam data is pending.
A formal meta-analysis was conducted of randomised controlled trials (RCTs) that compared screening by PSA with or without DRE and no screening . The analysis of six RCTs involving 387 286 men showed that screening increased the probability of prostate cancer diagnosis but had no significant effect on death from prostate cancer or overall mortality. Limitations of this meta-analysis were the methodological issues with each of the studies, including biases during and after randomisation, contamination of the unscreened arm due to men undergoing PSA testing before randomisation, the short duration of follow-up and other methodological shortcomings, which may have made it impossible to detect a difference in mortality and made the underlying power calculation invalid. More recently, a Cochrane systematic review of RCTs of PSA screening vs no screening was completed . This study also identified five studies involving 341 351 men; as in the meta-analysis provided in the previous meta-analysis , three of the studies had methodological flaws that could represent a high risk of bias. The analysis also showed that the number of men diagnosed with prostate cancer was significantly greater in men randomised to screening but no statistically significant difference in prostate cancer-specific mortality was found. In addition to these two systematic reviews, an evidence review was carried out in preparation for the 2011 revision of the USA Preventive Services Task Force (USPSTF) recommendations . This review used the same information as the two previous citations, but differentiated between high and low quality RCTs of screening. It acknowledged the significant differences found in the ERSPC and Göteborg randomised trials but concluded that ‘after about 10 years, PSA-based screening results in small or no reduction in prostate cancer-specific mortality and is associated with harms related to subsequent evaluation and treatments, some of which may be unnecessary’. The draft of the USPSTF report of 7 December 2011, which recommends against PSA-driven screening, has evoked heavy discussion and criticism. The final version is still pending.
RISK ASSOCIATED WITH SCREENING
The risks associated with PSA screening can be attributed to the screening tests themselves or from the subsequent diagnostic evaluations and treatment. In the Rotterdam section of the ERSPC, 5802 transrectal sextant biopsies were evaluated for complications . Haematuria lasting for >3 days and haematospermia were present after 22.6% and 50.4% of the procedures, respectively. Fever developed after biopsy in 200 patients (3.5%) and urinary retention on 20 occasions (0.4%); 27 patients (0.5%) required hospitalisation primarily for signs of prostatitis and/or urosepsis. Major complications were rare. A recent publication has examined further the complications of prostate needle biopsy as a potential risk of prostate cancer screening in the ERSPC . Between 1993 and 2011, 10 474 prostate needle biopsies were carried out as part of the study. Complications assessed by questionnaires at 2 weeks after biopsies indicated that <5% of biopsies resulted in febrile complications; significant risk factors were diabetes and prostatic enlargement. Infection remained the leading cause of hospital admission after biopsy, although the absolute frequency of hospital admissions remained low (<1%).
Over diagnosis and overtreatment are probably the most important adverse effects of prostate cancer screening. Other aspects that have to be balanced against a potential reduction in prostate cancer mortality rates are quality of life, costs and cost-effectiveness of screening. PSA testing cannot distinguish lethal prostate cancer from indolent disease before biopsy and establishment of the diagnosis of prostate cancer. Unfortunately, we do not have a screening tool that can distinguish potentially aggressive cancer in an early, curative stage. In the ERSPC, the rate of over diagnosis of prostate cancer, i.e. the diagnosis of men who would not have clinical symptoms during their lifetime, has been estimated at ≈50% in the screening group . Steps are being taken to address the issue of unnecessary prostate biopsies after PSA detection. Roobol et al.  have developed a risk based strategy using the ERSPC/SWOP risk calculator (http://www.prostatecancer-riskcalculator.com) that incorporates data on ultrasound, prostate volume, DRE plus PSA as indicators for biopsy. With proper clinical application large numbers of potentially unnecessary biopsies and the diagnosis of many potentially ‘indolent’ prostate cancers can be avoided.
Other ways of reducing the costs of screening are to adapt the PSA screening interval or increase the PSA threshold level. However, it has been shown that even a PSA threshold of 3.0 ng/mL misses 67.8% of biopsy detectable prostate cancers and 42.4% of potentially aggressive ones . These findings indicate that high-grade cancer can be present with low levels of PSA. The unnecessary treatment that is used as an argument against PSA screening can be reduced by offering patients active surveillance, which has been selected as an option by 25% of patients in the ERSPC.
PSA SCREENING GUIDELINES
Differences exist between the screening recommendations of various organisations. The AUA revised its guidelines on PSA screening in April 2009 and now recommends that baseline PSA screening be considered in men at age 40 years instead at 50 years . The ACS recently updated its guidance to recommend that asymptomatic men with a life expectancy of ≥10 years be given an opportunity to select PSA screening as part of an informed decision-making process . The National Comprehensive Cancer Network recommends a risk-based screening algorithm, including family history, race and age . As mentioned above, the USPSTF has drafted revised recommendations recently to a policy that recommends against PSA screening mainly balanced on the stated imbalance between benefits and harms of screening . It should be noted that the USPSTF based their recommendations on a meta-analysis that had significant shortcomings, as previously discussed. The EAU guidelines consider population-based screening as premature pending more solid information of its effect on prostate cancer mortality and quality-of-life effects. However, consistent with most other guidelines, PSA driven early diagnosis is recommended for well-informed men who wish to be screened. Testing men aged <50 years is considered premature .
Prostate cancer is an important healthcare problem and it has been established through several studies that PSA screening is effective in reducing the relative risk of metastatic disease and prostate cancer death. Steps need to be taken to address the issue of over diagnosis and overtreatment of the disease before a population-based screening programme can be enacted.
CONFLICT OF INTEREST