Outcomes after prostate brachytherapy are even better than predicted

Abstract

Background: During the first 3 years after prostate cancer treatment with radiation therapy, benign prostate-specific antigen (PSA) bounces are difficult for clinicians to distinguish from a biochemical recurrence, which can result in unnecessary interventions and erroneous predictions of outcomes. The objective of this study was to evaluate a commonly used PSA failure definition in a multinational, multi-institutional study after monotherapy with prostate brachytherapy.

Methods: Participants were selected from 2919 men who underwent permanent prostate brachytherapy at the University Medical Center Utrecht, Princess Margaret Hospital, or Seattle Prostate Institute between 1998 and 2006. Inclusion required not having received androgen-deprivation therapy and having at least 30 months of follow-up. Failure was defined as any post-treatment use of hormone therapy, clinical relapse, or prostogram-defined biochemical (PSA) failure. Cases in which the nomogram predicted biochemical failure were evaluated at each institution to verify biochemical status over time and the actual clinical outcome at 5 years.

Results: The median follow-up for the 1816 patients was 5.2 years. Concordance between the prostogram-predicted and actual outcomes, as measured by the Harrell c statistic, was 0.655 (95% confidence interval [CI], 0.536-0.774; P = .010) for the Princess Margaret group, 0.493 (95% CI, 0.259-0.648; P = .955) for the Seattle group, and 0.696 (95% CI, 0.648-0.744, P < .001) for the Utrecht group. The overall mean difference in biochemical recurrence-free survival at 5 years between actual outcomes and prostogram-defined outcomes was 9.2% (95% CI, 7.7%-10.6%). The total numbers of prostogram-defined and actual biochemical failures were 312 and 157, respectively (P = .001).

Conclusions: The widely used prostogram could not adequately distinguish a benign PSA bounce from a biochemical recurrence after prostate brachytherapy and could not be used to counsel patients about their predicted outcomes after treatment. The authors conclude that, to avoid unnecessary active interventions after treatment, clinicians should monitor PSA levels for at least 3 years and provide reassurance to patients that a PSA rise during this time is common and may not indicate a treatment failure.