NCCN and PROTEUS high-risk prostate cancer definitions showed comparable predictive accuracy for post-prostatectomy outcomes with AUCs around 0.54-0.64 and no significant differences.
Does the NCCN-based definition of high-risk prostate cancer compare to the PROTEUS definition in predicting oncologic outcomes in patients undergoing radical prostatectomy?
NCCN and PROTEUS definitions of high-risk prostate cancer have comparable predictive accuracy for oncologic outcomes following radical prostatectomy.
344 Background: There are varying definitions of high-risk prostate cancer. The most commonly used is the NCCN definition which defines patients as having high- or very-high-risk disease if they have any of clinical T3 or greater disease, Gleason Grade Group 4 or 5, or PSA > 20ng/mL. Recent studies of androgen receptor signaling inhibitors have employed alternative definitions of baseline risk. Herein, we compared patient distributions and oncologic outcomes for patients undergoing radical prostatectomy according to two definitions of high-risk disease. Methods: Using an institutional database of patients undergoing radical prostatectomy, we characterized patients as having high-risk disease either by using NCCN risk groups (any of Gleason grade group GGG 4 or 5; or Clinical T3 or T4; or PSA >20 ng/mL) or using the definition employed in the PROTEUS trial (overall GGG ≥3 and at least one of the following: GGG5 in at least one core; or GGG4 in at least 2 cores, each with >80% involvement; or GGG3+ in ≥6 systematic cores; or GGG3+ in ≥3 systematic cores and PSA ≥20 ng/mL). Descriptive statistics were used to compare patient distributions. Area under the receiver operating curve (AUC) were compared between NCCN- and PROTEUS-based definitions of high-risk disease for oncologic outcomes including extra-prostatic extension, positive surgical margins, and PSA >0.1 or >0.2 at 12 months post-operatively. Results: Among 424 patients undergoing radical prostatectomy in the dataset, 358 had complete pathological biopsy data allowing for nuanced pre-operative risk stratification. Of these, 62 (17%) were classified as high-risk per NCCN criteria while 44 (12%) were classified as high-risk per PROTEUS criteria. Among 62 patients classified as high-risk per NCCN criteria, 38 met the PROTEUS criteria and 24 did not. Conversely, among 44 patients classified as high-risk per PROTEUS criteria, 38 met the NCCN high-risk criteria and 6 did not. Using AUC, the two definitions were comparably predictive of extra-prostatic extension (NCCN 0.56, 95% CI 0.52-0.60; PROTEUS 0.56, 95% CI 0.53-0.60; p=0.96), positive surgical margins (NCCN 0.54, 95% CI 0.49-0.58; PROTEUS 0.52, 95% CI 0.49-0.56; p=0.44), PSA >0.1 at 12 months post-operatively (NCCN 0.54, 95% CI 0.46-0.62; PROTEUS 0.59, 95% CI 0.51-0.67; p=0.22), and PSA >0.2 at 12 months post-operatively (NCCN 0.59, 95% CI 0.49-0.69; PROTEUS 0.64, 95% CI 0.54-0.75; P=0.32). Conclusions: This comparative analysis based on highly granular data shows that prostate cancer patients defined as high-risk based on NCCN risk groups, or high-risk criteria employed in the PROTEUS clinical trial, have comparable outcomes. Therefore, identifying patients suitable for treatment intensification with peri-operative systemic therapy is critical in this evolving clinical landscape.
Wallis et al. (2026) studied this question. NCCN and PROTEUS high-risk prostate cancer definitions showed comparable predictive accuracy for post-prostatectomy outcomes with AUCs around 0.54-0.64 and no significant differences.