To develop a novel tool to increase the number of patients with prostate cancer eligible for active surveillance (AS) without increasing the risk of unfavourable pathological features (i.e., misclassification) at radical prostatectomy (RP).
Overall, 16 049 patients with low- or intermediate-risk prostate cancer treated with RP were identified. Misclassification was defined as non-organ confined or grade group ≥3 disease at RP. The coefficients of a logistic regression model predicting misclassification were used to develop a risk score. We then performed a systematic analysis of different thresholds to discriminate between patients with or without unfavourable disease and we compared it to available AS criteria.
Overall, 5289 (33.0%) patients had unfavourable disease. At multivariable analyses, PSA level, clinical stage, biopsy grade group, the number of positive cores, and PSA density were associated with the risk of unfavourable disease (all P < 0.001). The Prostate Cancer Research International: Active Surveillance (PRIAS) criteria were associated with a lower risk of misclassification (13%) compared to other criteria. Overall, 3303 (20.6%) patients were eligible according to the PRIAS protocol. The adoption of an 18% threshold according to the risk score increased the proportion of eligible patients from 20.6% to 29.4% without increasing the risk of misclassification as compared to the PRIAS criteria.
The use of a novel risk score for AS selection would result in an absolute increase of 10% in the number of patients eligible for this approach without increasing the risk of misclassification.