Individualized Oncologic Surveillance After Radical Prostatectomy for Localized Prostate Cancer.

Abstract

Background: The oncologic surveillance after radical prostatectomy (RP) remains poorly defined.

Objective: To evaluate a model-derived, risk-adapted surveillance approach and estimate when the modeled OCM hazard exceeds the modeled recurrence hazard.

Design, setting, and participants: Retrospective, population-based study including 11,552 patients undergoing RP.

Outcome measurements and statistical analysis: We developed a follow-up schedule based on the per-visit risk of biochemical recurrence (BCR) and high-risk BCR (HR-BCR), according to the EAU and EMBARK criteria, optimizing the number of PSA tests while maintaining the detection rate of the EAU schedule. Weibull regression was used to estimate model-derived time points at which OCM hazard exceeded recurrence hazard.

Results and limitations: Overall, 8,662 (75%) patients had adverse pathological features (pT3-4, pN1, positive surgical margins, and/or ISUP>2). Median follow-up among patients without recurrence was 90 months. The 5-year BCR-, EAU-HR-BCR-, EMBARK-HR-BCR-free survival probabilities were 78%, 91%, and 98% , respectively. Compared to the EAU schedule, adverse-feature patients required intensive follow-up in the first years (quarterly for 2 years, biannually up to 6 years), with de-intensification thereafter. Every 9 months visits were sufficient for favorable-feature patients. A similar pattern was observed for HR-BCR, but with a shorter intensive period (up to 3 years). The follow-up year when the hazard of BCR, EAU-HR-BCR, and EMBARK-HR-BCR became lower than the hazard of OCM ranged from 4-20+, 1-12, and 1-4 years, depending on disease characteristics, age, and comorbidity.

Conclusions: Our results show that surveillance protocols should be individualized. We presented a post-RP surveillance strategy in which the intensity and duration of follow-up are based on the risk of BCR and OCM according to individual recurrence risk and competing mortality. Future studies are needed to validate this strategy before it can be implemented in clinical practice.