Multicenter External Validation of a Nomogram for Predicting Positive Prostate-specific Membrane Antigen/Positron Emission Tomography Scan in Patients with Prostate Cancer Recurrence

Background: A nomogram has recently been developed to predict 68Ga-labeled prostate-specific membrane antigen (PSMA)-11 positron emission tomography (PET)/computed tomography (PSMA-PET) results in recurrent prostate cancer (PCa) patients. Objective: To perform external validation of the original nomogram in a multicentric setting. Design, setting, and participants: A total of 1639 patients who underwent PSMA-PET for prostate-specific antigen (PSA) relapse after radical therapy were retrospectively included from six high-volume PET centers. The external cohort was stratified according to clinical setting categories: group 1: first-time biochemical recurrence (n = 774); group 2: PSA relapse after salvage therapy (n = 499); group-3: biochemical persistence after radical prostatectomy (n = 210); and group-4: advanced-stage PCa before second-line systemic therapies (n = 124). Intervention: PSMA-PET in recurrent PCa. Outcome measurements and statistical analysis: PSMA-PET detection rate was assessed in the overall population and in each subgroup. A multivariable logistic regression model was produced to evaluate the predictors of a positive scan. The performance characteristics of the model were assessed by quantifying the predictive accuracy (PA) according to model calibration. The Youden's index was used to find the best nomogram's cutoff. Decision curve analysis (DCA) was implemented to quantify the nomogram's clinical net benefit. Results and limitations: In the external cohort, the overall detection rate was 53.8% versus 51.2% in the original population. At multivariate analysis, International Society of Urological Pathology grade group, PSA, PSA doubling time, and clinical setting were independent predictors of a positive scan (all p ≤ 0.02). The PA of the nomogram was identical to the original model (82.0%); the model showed an optimal calibration curve. The best nomogram's cutoff was 55%. In the DCA, the nomogram revealed clinical net benefit when the threshold nomogram probabilities were ≥20%. The retrospective design is a major limitation. Conclusions: The original nomogram exhibited excellent characteristics on external validation. The incidence of a false negative scan can be reduced if PSMA-PET is performed when the predicted probability is ≥20%. Patient summary: A nomogram has been developed to predict prostate-specific membrane antigen/positron emission tomography (PSMA-PET) results for recurrent prostate cancer (PCa). The nomogram represents an easy tool in the decision-making process of recurrent PCa.