Definitions of biochemical failure that best predict clinical failure in patients with prostate cancer treated with external beam radiation alone: A multi-institutional pooled analysis

Purpose: Pooled data on 4,839 patients with T1-2 prostate cancer treated with external beam radiation therapy (RT) alone at 9 institutions have previously provided long-term biochemical failure (BF) and clinical outcomes using the American Society for Therapeutic Radiology and Oncology (ASTRO) definition. In this report we determined the sensitivity and specificity of several BF definitions using distant failure (DF) alone or clinical failure (CF), defined as local failure (LF) and/or DF. Materials and Methods: The pooled cohort was treated between 1986 and 1995 with external beam RT (60 Gy or greater) without pre-RT androgen suppression or planned post-RT adjuvant androgen suppression. Median followup was 6.3 years. The sensitivity and specificity of 102 definitions of BF relative to DF and LF were assessed. Results: The BF definitions with higher sensitivity and specificity than the ASTRO definition for DF only and CF are reported. The sensitivity and specificity of the ASTRO definition to predict DF alone was 55% and 68%, respectively. Three definitions had higher sensitivity and specificity, namely prostate specific antigen (PSA) greater than current nadir (lowest PSA prior to current measurement) plus 3 ng/ml (sensitivity 76% and specificity 72%), dated at the call (failure date as the date when the criterion was met), PSA greater than absolute nadir plus 2 ng/ml (sensitivity 72% and specificity 70%, dated at the call, or 2 consecutive increases of at least 0.5 ng/ml, back dated (sensitivity 69% and specificity 73%). The sensitivity and specificity of the ASTRO definition to predict CF was 60% and 72%, respectively. Three definitions had higher sensitivity and specificity, namely PSA greater than current nadir plus 3 ng/ml (sensitivity 66% and specificity 77%), dated at the call, PSA greater than absolute nadir plus 2 ng/ml (sensitivity 64% and specificity 74%), dated at the call, or 2 consecutive increases of at least 0.5 ng/ml, back dated (sensitivity 67% and specificity 78%). Conclusions: Using what is to our knowledge the largest data set of patients with prostate cancer treated with RT alone we correlated multiple definitions of BF with the strict clinical end points of DF alone and CF (DF or local failure). Defining BF as PSA greater than absolute nadir plus 2 ng/ml, dated at the call, PSA greater than current nadir plus 3 ng/ml, dated at the call, or 2 consecutive increases of at least 0.5 ng/ml, back dated, had higher sensitivity and specificity for DF alone or CF compared with the ASTRO definition. This information should contribute to the discussion regarding suggested modifications to the ASTRO definition of biochemical failure.