Prostate-specific antigen (PSA) velocity and benign prostate hypertrophy predict for PSA spikes following prostate brachytherapy

Purpose To evaluate if variants of serum PSA or benign prostatic hypertrophy correlate with the development of a PSA spike following permanent prostate brachytherapy. Methods and Materials Two-hundred-eighteen hormone-naı̈ve patients with clinical T1b-T3a (1997 AJCC) prostate cancer who were treated with brachytherapy between August 1995 and November 1999, with or without supplemental external beam radiation therapy, and who remained free of biochemical failure were analyzed. The median follow-up was 46 months. A PSA spike was defined as a rise ≥0.2 ng/mL followed by a durable decline. Biochemical disease-free survival was defined by the ASTRO Consensus Definition with the additional constraint that the most recent PSA be ≤1.0 ng/mL. In addition, none of the patients possessed equivocal biochemical results (1 or 2 consecutive PSA rises or a declining PSA >1.0 ng/mL). In addition to previously reported clinical, treatment, and dosimetric parameters evaluated for spike, PSA density, transition zone (TZ) PSA density, percent free PSA, PSA velocity, PSA doubling time, TZ volume, and transition zone index (TZI) were included. The PSA kinetics of 18 hormone naı̈ve patients who were implanted during the same time period and subsequently failed were also evaluated. Results Fifty-two (23.9%) developed a PSA spike. Of the demographic and preimplant clinical parameters, patient age, TZ volume, TZI, TZ PSA density, and 125I were statistically significant predictors for a PSA spike. Of the postimplant parameters, V200, follow-up, first postimplant PSA, and most recent PSA predicted for a PSA spike. In multivariate Cox regression analysis, PSA nadir, TZI, follow-up, age, months to PSA nadir and preimplant PSA velocity were significant predictors for spike. However, when variables only determinable after a PSA spike were included in the multivariate analysis, TZI, age, PSA velocity, and first postimplant PSA were predictors for a spike. Using categorical cutpoints of TZI >0.25, age at implant <62 years, and first postimplant PSA >1.0 ng/mL in the regression analysis, a positive likelihood ratio for a PSA spike of >1.8 was noted for each variable. Patients with PSA progression displayed significantly different PSA kinetics than those with a spike. Conclusions In multivariate analysis, PSA nadir, TZI, patient age, months to PSA nadir, follow-up, and preimplant PSA velocity were predictive of a PSA spike. However, when only variables identifiable prior to a spike were evaluated, TZI, patient age, preimplant PSA velocity, and first postimplant PSA were the strongest predictors for a PSA spike.