Predicting Biopsy Outcomes During Active Surveillance for Prostate Cancer: External Validation of the Canary Prostate Active Surveillance Study Risk Calculators in Five Large Active Surveillance Cohorts.

BACKGROUND: Men with prostate cancer (PCa) on active surveillance (AS) are followed through regular prostate biopsies, a burdensome and often unnecessary intervention, not without risks. Identifying men with at a low risk of disease reclassification may help reduce the number of biopsies. OBJECTIVE: To assess the external validity of two Canary Prostate Active Surveillance Study Risk Calculators (PASS-RCs), which estimate the probability of reclassification (Gleason grade ≥7 with or without >34% of biopsy cores positive for PCa) on a surveillance biopsy, using a mix of months since last biopsy, age, body mass index, prostate-specific antigen, prostate volume, number of prior negative biopsies, and percentage (or ratio) of positive cores on last biopsy. DESIGN, SETTING, AND PARTICIPANTS: We used data up to November 2017 from the Movember Foundation's Global Action Plan (GAP3) consortium, a global collaboration between AS studies. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: External validity of the PASS-RCs for estimating reclassification on biopsy was assessed by calibration, discrimination, and decision curve analyses. RESULTS AND LIMITATIONS: Five validation cohorts (Prostate Cancer Research International: Active Surveillance, Johns Hopkins, Toronto, Memorial Sloan Kettering Cancer Center, and University of California San Francisco), comprising 5105 men on AS, were eligible for analysis. The individual cohorts comprised 429-2416 men, with a median follow-up between 36 and 84 mo, in both community and academic practices mainly from western countries. Abilities of the PASS-RCs to discriminate between men with and without reclassification on biopsy were reasonably good (area under the receiver operating characteristic curve values 0.68 and 0.65). The PASS-RCs were moderately well calibrated, and had a greater net benefit than most default strategies between a predicted 10% and 30% risk of reclassification. CONCLUSIONS: Both PASS-RCs improved the balance between detecting reclassification and performing surveillance biopsies by reducing unnecessary biopsies. Recalibration to the local setting will increase their clinical usefulness and is therefore required before implementation. PATIENT SUMMARY: Unnecessary prostate biopsies while on active surveillance (AS) should be avoided as much as possible. The ability of two calculators to selectively identify men at risk of progression was tested in a large cohort of men with low-risk prostate cancer on AS. The calculators were able to prevent unnecessary biopsies in some men. Usefulness of the calculators can be increased by adjusting them to the characteristics of the population of the clinic in which the calculators will be used.