Caution in interpreting biochemical control rates after treatment for prostate cancer: Length of follow-up influences results

Objectives. Prostate-specific antigen (PSA) based end points are commonly used to report outcomes after treatment for prostate cancer. This study examines the influence of follow-up length on biochemical control (bNED) rates. Methods. We reviewed 437 patients with clinically localized prostate cancer treated with conformal radiotherapy without neoadjuvant androgen deprivation. Biochemical failure was defined as three consecutive PSA increases or an increase large enough to prompt androgen deprivation therapy. The failure date was projected back to the midpoint between the PSA nadir and the first PSA increase (or between the nadir and the initiation of androgen deprivation therapy). The analysis was performed by censoring patients with longer follow-up in a stepwise fashion, thus creating smaller subgroups with shorter follow-up intervals. Subgroup 1 (n = 191) and subgroup 2 (n = 273) were defined to include those patients monitored for up to 2 years and up to 3 years, respectively. Results. The median follow-up intervals for subgroup 1, subgroup 2, and the original study population were 1.1, 1.5, and 2.5 years. No significant differences were seen in pretreatment prognostic factors among the three groups. The 2-year bNED of subgroup 1, subgroup 2, and the original population was 86%, 77%, and 73%, respectively. Although subgroup 1 had a superior bNED compared with the original population (P = 0.04), no differences in clinical recurrence rates were seen among any of the three groups. Conclusions. Because of projecting the biochemical failure dates back according to commonly used bNED definitions, control rates are highly dependent on the length of follow-up. UROLOGY 54: 875-879, 1999. (C) 1999, Elsevier Science Inc.