Perspectives
Cancer
Volume 198 - Issue 10

The dilemmas of prostate cancer screening

Authors:  Jonas Hugosson and Sigrid V Carlsson

Med J Aust 2013; 198 (10): 528-529. || doi: 10.5694/mja13.10242
Published online: 3 June 2013
It is unquestionable that prostate-specific antigen screening is associated with overdiagnosis. However, the magnitude of this problem still remains uncertain because of the lack of sufficient robust, long-term follow-up data from ongoing trials.
Download PDF (125 KB)
Share: Share on Blue Sky Share on Facebook Share on LinkedIn Share by email

Prostate cancer screening is controversial — the screening decision must be based on high-quality evidence

Screening for cancer, especially screening for prostate cancer with prostate-specific antigen (PSA), is one of the most controversial subjects in medicine today. The main dilemma is that one aims for a beneficial effect on the population level, but on the individual level, some will only be harmed without any gain from participation — such as those in whom the cancer detected is already too advanced at diagnosis, and those diagnosed with a harmless disease that will not surface as a clinical cancer during the lifetime (overdiagnosis). These individuals will unnecessarily be “labelled” as patients for many years, are likely to be treated unnecessarily and suffer from the side effects of treatment (overtreatment) without altering the risk of prostate cancer mortality.

As individuals have various viewpoints of what is an acceptable harm-to-benefit ratio, most professional organisations and guideline groups today emphasise shared decision making and stipulate that “screening must be preceded by a discussion regarding risks and benefits”,1 so that individuals can make the choice they feel is in line with their personal preferences of harms and benefits.

Compared with breast cancer screening, it seems as though prostate cancer screening increases both the benefits (up to 44% relative prostate cancer mortality reduction, with 293 men needed to be invited for screening) and the harm;2 the trade-off in quality of life (QOL) is high, and one-quarter of quality-adjusted life-years (QALYs) is lost.3 To guide men’s decision making in whether to be screened or not, we need more research and longer follow-up of ongoing studies. This issue of the Journal contains three articles dealing with prostate cancer screening.

Evans and colleagues describe patterns of care, as recorded in a prostate cancer registry, for men diagnosed with prostate cancer in one state in Australia.4 Although this registry has a short history and limited participation rate, important findings are reported. The treatment patterns for localised disease are largely in line with data from the Cancer of the Prostate Strategic Urologic Research Endeavor (CAPSURE) registry.5 Promisingly, as many as 40% of men with low-risk disease are managed with initial surveillance, compared with < 10% in the CAPSURE registry5 and 28% in the Göteborg screening trial.2 With longer follow-up and a high recruitment rate, such registries could provide valuable information.

Del Mar and colleagues aim to re-examine the evidence for prostate cancer screening and discuss how men should be informed about screening.6 We agree with the authors that decision aids, as a complement to the discussion with the physician, can help men make a high-quality decision, consistent with their personal preferences and values. However, the re-examination of screening efficacy as performed by Del Mar and colleagues has several limitations. The authors analyse five studies, one of which (the Quebec study) has been criticised for poor quality and had a participation rate of only 24%.7 Further, the Stockholm and the Norrköping studies were small studies that were not designed to evaluate mortality.8,9 In the Stockholm study, men were invited only once, and a minority of men with screen-detected prostate cancer had treatment with curative intent. In the Norrköping study, less than 500 men had two PSA measurements, and none had more than two. In the last publication from the Norrköping study (not included in the present analysis), the results were adjusted for a large difference in age at randomisation between the study groups, thus making the validity of this study questionable.9

The Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial study (an earlier report cited by Del Mar and colleagues) compared organised versus non-organised screening, rather than screening versus no screening. Thus, at 13 years of follow-up (in a later report not cited by Del Mar and colleagues), there was no statistically significant difference in prostate cancer mortality risk between the arms.10 The only large randomised study evaluating the effect of organised repeat screening versus no screening is the European Randomised Study of Screening for Prostate Cancer (ERSPC) trial.11 Del Mar and colleagues have not cited its most recent publication from 2012,11 in which the relative risk in those actually screened was 0.71. This relative risk reduction of 29% is the most accurate measure available today and men should be informed of this.

Further, the authors assert that “two systematic reviews including all reported trial data concluded that screening is not effective”. We agree that it is valuable to perform secondary analyses of screening trials that adjust for non-attendance; however, pooling heterogeneous studies is a major concern.12,13 For the same reason that incompatible trials should not be combined in formal or informal meta-analyses, we would not pool corrected data as the authors do.14

It is unquestionable that PSA screening is associated with overdiagnosis. However, the magnitude of this problem still remains uncertain because of the lack of sufficiently long follow-up, and estimates come mainly from modelling studies. In computerised simulation models, overdetection rates have been estimated at 15%–50%.15,16 Postmortem examination of men who died from unrelated causes have found prostate cancer in about 30%–50% of men aged between 50 and 70 years.17 However, pathological features of PSA-detected tumours do not exactly resemble those of cancers detected on autopsy.18,19 In fact, even localised tumours have a high potential to progress to fatal, metastatic disease if the patient lives long enough.20 The main challenge is the individual risk assessment, which includes trying not to diagnose low-risk cancers in older men, discouraging PSA testing in asymptomatic men aged > 75 years and recommending active monitoring in men with low-risk cancer. This approach aims at maximising QOL, while still maintaining the focus on early diagnosis and effective treatment in the many men who need it.

Martin and colleagues report a cost-effectiveness analysis to estimate the net benefit and cost of PSA screening versus no screening.21 The authors conclude that focusing on men at very high risk is cost-effective, which seems to be a reasonable conclusion. But the question still remains, how should we best define this high-risk group? It is evident that this calls for a smart screening algorithm that remains to be defined. Several authors now propose risk-stratified screening, taking into account informed consent, baseline PSA and comorbid conditions, as well as avoiding screening elderly men and encouraging active surveillance.1,13,22

Using the Microsimulation Screening Analysis (MISCAN) model, we recently showed a 28% reduction in prostate cancer-specific mortality in favour of screening. However, the benefit (prevented deaths) was mitigated by a loss of QALYs — a decrease of 23% from unadjusted life-years gained. This was mainly a result of the long-term side effects of treatment.3 We currently have very sparse data on long-term effects on QOL after treatment of localised prostate cancer, and the QOL profile of screening may therefore change with further follow-up.

Martin and colleagues used a utility index of 0.95 for the remaining lifetime,21 instead of a duration of 9 years, which will unquestionably affect the outcome.3 The remaining lifetime will exceed 20 years in these patients and the long-term post-recovery period has the greatest impact on QOL. An assumption of lifelong QOL loss is probably incorrect, as positive effects such as a lower risk of benign prostatic hyperplasia and urinary tract infections may, in time, counterbalance treatment-related complications. It is also known that patients may adapt to their new situation and report overall QOL similar to controls after 3 years.23 As the natural course of potentially lethal PSA-detected cancers is not yet fully understood, we believe that it is questionable to conduct cost-effectiveness analyses on theory, and hence that the current evidence for the authors’ clinical applicability and effectiveness is limited.

To give a final recommendation regarding population-based screening, we need robust, long-term follow-up data from ongoing trials. The way forward is to evaluate only high-quality studies, such as the ERSPC, with careful long-term follow-up including QOL and cost-effectiveness evaluations. At this time, caution is required in interpreting modelling studies, as our knowledge of the natural course of screen-detected cancers is still very preliminary. Meanwhile, men who need to make a decision about PSA screening should be given written information about the pros and cons of such testing with the best evidence available today (21%–44% prostate cancer mortality reduction, but with risks of side effects from diagnosis and treatment).2,3,11 This will enable men, in discussion with their physicians, to make a personal choice.

Authors

  • Jonas Hugosson MD, PhD1
  • Sigrid V Carlsson MD, PhD1,2
  • 1. Department of Urology, University of Gothenburg, Gothenburg, Sweden.
  • 2. Department of Surgery, Urology Service, Memorial Sloan-Kettering Cancer Center, New York, NY, USA.

Correspondence: jonas@urol.se

Competing interests

No relevant disclosures.

References

Provenance: Commissioned; externally peer reviewed.

View this article on Wiley Online Library