Predictive accuracy and clinical benefit of a nomogram aimed to predict ⁶⁸Ga-PSMA PET/CT positivity in patients with prostate cancer recurrence and PSA < 1 ng/ml external validation on a single institution database.

Abstract

To perform an external validation of a recently published nomogram aimed to predict positive ⁶⁸Ga-PSMA-11 PET/CT in patients with biochemical recurrence (BCR) after radical prostatectomy (RP) by Rauscher et al. (Eur Urol 73(5):656-661, 2018). Overall, 413 PCa patients with BCR after RP (two consecutive PSA ≥ 0.2 ng/ml) and PSA value between 0.2 and 1 ng/ml were included. A multivariable logistic regression model was produced to assess the predictors of positive ⁶⁸Ga-PSMA-11 PET/CT results. The performance characteristics of the model were assessed by quantifying the predictive accuracy, according to model calibration. Yuden's index was used to find the best nomogram's cut-off. Finally, decision curve analysis (DCA) was implemented to quantify the nomogram's clinical value. In the external cohort, the overall detection rate of ⁶⁸Ga-PSMA-11 PET/CT was 44% vs. 64.7% in the original population. At multivariate analysis, PSA at ⁶⁸Ga-PSMA-11 PET/CT (OR: 7.06, p < 0.001) and ongoing ADT at time of ⁶⁸Ga-PSMA-11 PET/CT (OR: 2.07, p = 0.03) were the only independent predictors of PET/CT positivity. The predictive accuracy of nomogram was suboptimal and comparable to that reported in the original model (64% vs. 67%, respectively). The calibration plot indicated suboptimal concordance. The best nomogram's cut-off to predict positive ⁶⁸Ga-PSMA-11 PET/CT was 35% (AUC = 0.61). In DCA, the nomogram revealed clinical net benefit when the threshold probabilities of positive ⁶⁸Ga-PSMA-11 PET/CT is > 35%. We assessed similar suboptimal predictive accuracies in the external cohort compared to the original one. PSA and ongoing ADT were confirmed as positive predictors, and the most informative nomogram cut-off resulted 35%.