Two recently reported large-scale trials conducted in the United States and western Europe have provided evidence that coordinated screening programs will not reduce mortality in countries or regions where prostate-specific antigen (PSA) testing is already highly prevalent, but will reduce mortality in places where PSA testing prevalence is low.
The trials also produce evidence that coordinated screening will cause over-diagnosis and over-treatment.
The instigation of a national screening program should be delayed until a more specific marker for aggressive disease than PSA level becomes available. In the meantime, results of the two trials can be used to inform the development of regional testing policies in Australia.
These policies should encourage regular PSA testing in regions with low testing prevalence, but must also embrace methods of dealing with over-diagnosis and over-treatment.
“Active surveillance” programs (whereby men with early-stage cancers are monitored regularly by PSA testing and digital rectal examinations) and development of counselling services should be encouraged.
Opinion is sharply divided on the value of prostate-specific antigen (PSA) testing for prostate cancer.1 Up to now, the ascendant point of view has been that there is little evidence of benefit from PSA testing but much evidence of morbidity from treatment of slowly progressing cancers.2-4 At the other end of the spectrum is the view that early diagnosis of all cancers that can be cured in their localised stages is beneficial. At times, both sides of the debate have been expressed with a vigour that has left observers thinking they are mutually exclusive. This has resulted in a state of confusion that, we contend, has done much to paralyse the provision of helpful guidance to the public.
In March 2009, the New England Journal of Medicine published the results of two large-scale prostate cancer screening trials that commenced in the early 1990s.5,6 In the Prostate, Lung, Colon and Ovarian (PLCO) Cancer Screening Trial in the United States, 76 693 men aged between 55 and 74 years were randomly allocated to either take part in an annual screening program involving PSA testing and digital rectal examination or to receive routine medical care. In the European Randomized Study of Screening for Prostate Cancer (ERSPC) in western Europe, centres in seven countries pooled data from regional trials involving 182 016 men aged between 55 and 74 years who were randomly assigned to undergo2–4-yearly screening or routine medical care. Although the results of the trials are not yet mature, we believe that they have produced sufficient data to depolarise the PSA-testing debate and provide a framework for helpful public guidelines.
At first sight, it appears that the two trials present conflicting results. After 7 years’ follow-up, results of the PLCO trial indicated that an annual screening program did not reduce the number of advanced cancers detected or the number of deaths due to prostate cancer (Box).5 After a median follow-up period of 9 years, the ERSPC showed that screening at less frequent intervals reduced the number of advanced cancers that were detected by half, and deaths from prostate cancer by 20%, statistically6 (or 31% when adjusted for “contamination” (informal self-screening arrangements).7
How can these different results be reconciled? We believe that the two trials are not producing conflicting results, but rather, among other things, they are reflecting the different “starting positions” of the sampled populations. When a cancer screening test is applied to a large enough number of subjects in an appropriate age range, it will cause an increased number of cancers at all stages of advancement to be detected. If the testing process is repeated within a relatively short period, fewer advanced cancers will be detected by subsequent tests. Therefore, over a given period of time, the total number of advanced cancers detected in the screened population will be lower than the number detected in an unscreened population. In contrast, the number of early cancers detected in the screened population will be higher than in an unscreened population over the same time period. This “stage shift” is exactly what was seen in the ERSPC (Box). In the screening arm, 725 of the 6039 cancers detected (12.0%) were advanced or metastatic (ie, Stage T3, T4 or M1). In the control arm, the corresponding proportion was more than double this figure (1209/4611 [26.2%]).
However, if a well coordinated screening program is commenced in a population that has already been heavily exposed to the screening test on an ad hoc basis,8,9 it is unlikely that a large number of advanced cancers will be newly detected. Thus there will be no large stage shift. This scenario closely reflects what happened in the PLCO trial (Box). In the 3 years before participation in the trial, 44% of men had had one or more PSA tests, and 55% had had a digital rectal examination. As a result, only 135 of the 2940 cancers detected (4.6%) in the control arm of the PLCO trial were advanced or metastatic (ie, Stage 3 or 4) (Box). In addition, half of the men in the “unscreened” control arm of the PLCO trial organised PSA testing for themselves, which weakened the power of the PLCO trial to demonstrate a reduction in mortality. The authors emphasised the point by stating that the cumulative death rate from prostate cancer at 10 years in the intervention and control groups combined was 25% lower in participants who had undergone two or more PSA tests at baseline than in those who had not been tested.
Unfortunately, if treatment is not efficacious, all that will be achieved by a stage shift is an apparent prolongation in survival time resulting from earlier diagnosis (a lead-time bias). However, a large body of uncontrolled and randomised controlled clinical data now indicate that survival can be prolonged by radical prostatectomy or radiation10-13 and by the use of adjuvant androgen suppression therapy.14-18
The differences in testing practices between the two continents arose as informal, ad hoc (opportunistic) PSA testing became highly prevalent in the US before the PLCO trial started. In 2002, the age-standardised detection rate of prostate cancer in the US was 124.8 per 100 000 men — much higher than anywhere else in the world,16 including Australia (76.0 per 100 000). The situation in western Europe (and Australia) was quite different. Opportunistic testing rates were generally much lower than in the US, and advanced cancers were detected more frequently.8,9,19,20 Not surprisingly, age-standardised detection rates in 2002 in the countries that contributed to the ERSPC (ranging from 35.9 to 90.9 per 100 000 per annum) were far lower than in the US.21
On the negative side, it must be pointed out that the ERSPC screening programs increased the number of early-stage cancers detected, which, in the opinion of the trial investigators, resulted in an increase in the diagnosis of “harmless” cancers (over-diagnosis). This, in turn, leads to unnecessary treatment (over-treatment). They estimated that, for every life saved by screening, 1410 would have needed to be tested and 48 additional men would have required curative treatment.6 Although more deaths will be prevented and these ratios will fall as time passes, these estimates should be seen in context. For example, Rembold estimated that, in mammography breast cancer screening trials with 9 years of follow-up (ie, similar to the PLCO trial and ERSPC), one life would be saved in 4576 women aged 40–49 years, 1532 aged 50–59 years and 695 aged 60–69 years.22
The results of these two trials make it clear that both sides of the PSA testing debate are correct. Organised PSA population screening programs will result in over-diagnosis and over-treatment and will not save lives in places where PSA testing is already highly prevalent. However, they will save lives in countries where detection rates are low.
In a country like Australia, where survival rates for men with prostate cancer are among the highest in the world,23 the reader will be forgiven for wondering whether the PSA testing debate matters and, more importantly, whether the results of the screening trials are relevant. But substantial regional variations in incidence and mortality rates for prostate cancer provide good reason to answer “yes” to both questions. As Coory and Baade have shown, the prevalence of PSA testing varies significantly between regions.24 Moreover, they found that these variations translated into mortality differences, in much the same way, we contend, that they did in the PLCO trial and ERSPC. For example, 5-year data from the New South Wales Central Cancer Registry between 2002 and 2006 indicate that age-adjusted mortality rates for prostate cancer were about 40% higher in the NSW Cancer Council’s Hunter region than in its Central and Southern Sydney regions (standardised mortality ratios, 16.0 v 11.4 per 100 000; P < 0.01), but that age-adjusted incidence rates were about 7% lower (standardised incidence ratios [SIRs], 69.9 v 75.1 per 100 000). Furthermore, the cancers detected in the Hunter region were more advanced than those in the Central and Southern Sydney regions (SIRs for “localised” cancer, 25.2 v 41.7 per 100 000; SIRs for “distant” cancer, 3.4 v 2.8 per 100 000).25
The implications are that, in some regions, over-diagnosis and over-treatment will be commonplace. In these regions, exhorting all men to get tested annually will achieve no more than was achieved in the screening arm of the PLCO trial. But in regions where belated diagnosis and poor access to specialised medical services are significant problems, discouraging testing will serve to increase belated diagnoses and deaths. Viewed in this way, it is obvious that the PSA-testing debate serves no-one properly.
What can Australia learn from these variations in incidence, mortality and survival rates? We believe that a national population-based screening program using PSA testing is not warranted at the present time. However, if a better screening test becomes available — one that is highly specific for prostate cancers that are likely to cause harm within 15–20 years and thus will reduce over-diagnosis and over-treatment — then the concept of a national screening program will need to be revisited. In the meantime, we need to recognise that regional problems do exist.
A concept that deserves serious consideration is the development of a prostate cancer awareness policy that is designed to meet the needs of different regions. For example, in a region that already has a high prevalence of PSA testing and good access to medical services, the major problems to deal with are over-diagnosis and over-treatment. Even though there are already means of avoiding biopsies (eg, by obtaining confirmatory PSA levels in the ensuing months, one or two of which include free to total PSA ratios) in every man with a PSA level of 4 ng/mL or more, the introduction of more specific tests is a long way off, and over-diagnosis will continue to be a problem. Over-treatment is likely to remain a problem until biomarker staining of biopsy specimens becomes a reliable means of differentiating between aggressive and indolent cancers.26-28
In the meantime, we need to develop other strategies for avoiding over-treatment. The most promising of these is the “active surveillance” concept,29-33 whereby men with early-stage, low-grade cancers are monitored regularly with PSA testing and digital rectal examinations. In some of these programs, biopsies are repeated every 12–24 months. The goal is to delay curative therapy as long as possible (which in many cases will be indefinitely), but not long enough to make cure impossible. This approach differs significantly from the concept of “watchful waiting”, whose goal is to use palliative interventions only when cancer-related symptoms develop. Disadvantages of active surveillance are that it is resource-intensive and that the trigger points for initiating curative interventions are rather arbitrary. In two larger studies,32,33 men with PSA-level doubling times of less than 3 years or with very low free PSA to total PSA ratios underwent earlier interventions than those without.
In a region where PSA testing prevalence is low, where cancers are detected at advanced stages and access to specialist medical services is poor, a well coordinated PSA testing program could reduce the incidence of advanced cancer. Most new cancers diagnosed under such a program would be impalpable, low-grade, indolent cancers that could be managed in active surveillance clinics. The program would need to be accompanied by improved access to specialised monitoring and treatment services.
Finally, we envisage a valuable role for specially trained counsellors. Many of the decisions relating to prostate cancer detection and the need for treatment require lengthy discussions, and usually involve the wife or partner of the man at risk.34 General practitioners and specialists rarely have the time to explain the complexity of important issues such as “who should be tested and how often,” and “who should undergo or avoid treatment.” This is certainly not a novel concept. Breast cancer counsellors have been appointed in most health districts in the past 15 years and play an important role in helping women to understand the complexities of breast cancer treatment.
Staging and prostate cancer deaths among men participating in the PLCO and ERSPC trials5,6