Abstract
Dr Takahashi incorrectly argues that current evidence for prostate cancer screening is methodologically inadequate, asserting that prostate cancer screening studies ‘lack sufficient methodological rigor’, ‘lack appropriate control groups’, and resemble ‘case series or feasibility studies’. 1 Randomised controlled trials, including the ERSPC Trial, PLCO Trial, and CAP Trial, were designed with comparator arms, predefined endpoints, and long-term follow-up. Dr Takahashi has previously agreed that these three trials have appropriate control groups 2 and to dismiss them as unsuitable for systematic review is not a defensible methodological position.
Dr Takahashi states that overall survival must serve as the primary endpoint for screening trials. We assume he means overall mortality since survival is subject to lead time and length bias. Disease-specific mortality is the relevant outcome, not overall mortality, which is too insensitive an endpoint; any trial with this as a primary outcome would need to be so large and expensive such that it would be prohibitive as well as unnecessary. The absence of an overall mortality benefit in trials such as ERSPC does not invalidate observed reductions in prostate cancer-specific mortality; rather, it reflects predictable dilution effects.
The dismissal of prostate cancer-specific mortality as a ‘subjective endpoint prone to misclassification’ is overstated. Disease-specific mortality remains a standard and accepted endpoint in cancer screening evaluation, precisely because it captures the outcome that screening is designed to influence.
Our Commentary
3
used results from three sources:
a nested case–control study with long follow-up to evaluate the effectiveness of prostate specific antigen (PSA) in a risk-based screening algorithm with clinical significance for fatal prostate cancer as the outcome.
4
In this study, blood samples were taken before PSA testing was used outside a research setting, so there was no bias introduced from participants undergoing PSA testing before blood collection. the ERSPC trial because this trial provided the best evidence on the effect of PSA testing on prostate cancer mortality; this trial was not subject to the levels of contamination or lack of adherence of other trials.
5
a study of MRI scans following positive risk-based screening using PSA, which reduces false-positives and improves the benefit–harm balance compared with population screening using a single PSA cut-off level.
6
By combining the results of the studies, we estimated that 13 prostate cancer deaths would be prevented per 1000 men screened every 5 years from age 55, with nine deaths prevented for every man treated unnecessarily.
In summary, the response by Dr Takahashi to our paper relies on assertions that disregard the available evidence, which can be used to estimate the effectiveness of prostate cancer screening.
