Reconciling primary outcomes and statistical power: comment on a randomised controlled trial on prostate biopsy approaches

For three decades, transrectal biopsy for prostate cancer (PCa) detection has been the standard of care. However, in recent years there has been renewed enthusiasm for the transperineal approach due to observational evidence suggesting lower rates of infection with potentially higher cancer detection rates. Therefore, the randomised control trial by Schieda et al. [1] on this topic is of great interest. While the trial has several merits, there are notable issues related to the handling of the primary outcome, which undermine trust in the results. The most concerning issue lies in the unacknowledged discrepancies between the clinical registry (ClinicalTrials.gov identifier: NCT03936127) and the publication regarding the primary outcome and intended sample size [1]. The original primary outcome was post-procedural infection rates, with 360 patients intended to be recruited. However, within the publication, the former secondary outcome of clinically significant (cs)PCa detection was reported as the primary outcome instead. This change is not addressed, raising significant concerns of post hoc outcome switching [2], whether in response to lower-than-expected recruitment or infection rates. While interim analyses are an accepted part of clinical trials and can lead to significant redesign, these need to be pre-planned and fully accounted for. To restore trust in the reliability of the results, a transparent and full outline of the decision process within this trial would be appropriate, in line with Consolidated Standards of Reporting Trials (CONSORT) guidelines. Furthermore, unfortunately the modified sample size calculation provided is incorrect based on the information provided. The authors intended to power this superiority trial to detect an increase in the csPCa detection rate with transperineal biopsy to 51% from 34% at 80% power and 5% significance (two-sided). Although they state that a minimum of 105 patients per arm (n = 210) would be required, this corresponds to a one-sided significance only. Instead, 132 patients per arm are required (n = 264), with post hoc power calculations showing it remains notably underpowered (74%) for even its new primary outcome (Supporting Information in Data S1). Therefore, the main conclusion, that there is no clinically significant difference between the biopsy approaches, is not valid based on the evidence presented. However, it is important to consider trials in the context of the work that has come prior. Trial sequential analysis (TSA) can be used to quantify the number of participants needed in a meta-analysis to detect or reject the intervention effect [3]. Using the trials identified by Schieda et al. [1] and the contemporaneous TRANSLATE trial (NCT05179694) [4], a TSA was performed (Supporting Information in Data S1) using the median detection rate of transrectal biopsy (52%, range = 47–55%), and assumed an even more conservative difference between approaches (5%). This consistently crossed the TSA boundaries into the ‘futility’ area (Fig. S1), indicating further superiority trials examining these approaches are unlikely to change the conclusion of no clinically significant difference in overall detection rates. Therefore, Schieda et al. [1] appropriately highlight future trials should focus on different primary outcomes or subgroups. Nevertheless, the need for research must be balanced against the need to conduct and report studies in a robust fashion, ensuring adequate statistical power to fully address the research question. The authors have no disclosures. Fig. S1. Trial sequential analysis graph examining MRI-ultrasonography fusion transperineal vs transrectal biopsy to detect clinically significant prostate cancer. Data S1. Supporting Information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.