Ambiguities within the ASTRO consensus definition of biochemical failure: never assume all is equal.

Abstract

To quantitate the impact of a number of variables within the American Society of Therapeutic Radiology and Oncology consensus definition (ACD) of biochemical failure (bF) for prostate cancer. The prostate-specific antigen (PSA) data of 1050 men with prostatic adenocarcinoma treated between 1990 and 1997 with external beam radiotherapy monotherapy was examined. A total of 10,872 PSA results were examined for the effect of backdating, stable values, possible nadir date variations, assay resolution, and assay lower limit. Test timing was analyzed according to attendance frequency (mean time between tests [MTBT]) and regularity (using a derived "Irregularity Index" [II]). Median biochemical follow-up was 76 months. Nadir date alterations varied the rate of bF beyond 5 years between 0.8% and 5.3%. The bF rate of all MTBT and II quartiles subgroups differed significantly except between the two most regular and the two most infrequent quartiles. Multivariate analysis showed PSA, MTBT, and II to be statistically significant independent predictors of bF (p < 0.0001 for all). Many variables exist within the ACD that can produce inconsistencies in bF determination or alter the calculated date of bF. Detection bias issues, especially related to test timing, play a significant role in the derived outcomes.