Login to Access Video or Poster Abstract: MP43-11
Sources of Funding: Canary Foundation, Department of Defense (PC130355), Institute for Prostate Cancer Research

Introduction

For men on active surveillance (AS) for prostate cancer, utility of PSA kinetics in predicting pathologic reclassification remains controversial. We aimed to develop prediction methods for utilizing serial PSA and evaluate frequency of collection during AS.

Methods

Data were collected from men enrolled in the multicenter Canary PASS with Gleason ≤ 3+4, < 34% of biopsy cores positive, and no 5ARI use. PSA was collected every 3 months, and biopsies were performed at 6-12 months, 24 months, and then every 2 years from diagnosis. PSA velocity (PSAV) was determined by calculating a best linear unbiased predictor at each time point based on a linear mixed effect model (LMEM) that accounted for past and present serial logged PSA levels. The association of diagnostic PSA and/or PSAV with time to reclassification (increase in Gleason grade and/or increase to ≥ 34% of cores positive) was evaluated using Cox proportional hazards models. Models were adjusted for age, prostate size, biopsy parameters, and used serial PSA collected every 3 or every 6 months, where applicable.

Results

908 men met study criteria, of whom 288 (32%) had a reclassification event within 5 years. Median follow up was 2.9 years. After adjusting for age, prostate size, biopsy parameters, and diagnostic PSA, PSAV was a significant predictor of reclassification (HR for each 10% increase in PSAV = 1.7 (95% CI 1.3-2.1, p < .0001). The PSAV model had slightly improved accuracy over diagnostic PSA at 3 year prediction: AUC for diagnostic PSA model was 0.79 (95% CI 0.73-0.83) and for PSAV model was 0.80 (95% CI 0.75-0.84). Model performance was essentially identical using calculations based on q6 month rather than q3 month PSAs.

Conclusions

PSA velocity calculated using LMEM significantly predicts biopsy reclassification. Models that use repeat PSA measurements outperform a model with diagnostic PSA only. Model performance is similar using PSA assessed every 3 or 6 months. These results inform how PSA may be incorporated into active surveillance protocols and risk calculators.

Funding

Canary Foundation, Department of Defense (PC130355), Institute for Prostate Cancer Research