Incremental value of binary PSMA PET nodal status when added to Briganti-2019 and MSKCC models for predicting pathologic lymph node metastasis in prostate cancer.

Abstract

Objective: To evaluate whether adding PSMA PET-detected lymph node involvement to established nomograms (Briganti-2019 and MSKCC) improves prediction of pathologic lymph node metastasis(LNM) in prostate cancer.

Materials and methods: Retrospective, registry-based, multi-center cohort coordinated by the Turkish Urooncology Association, with data from 10 centers. We included 301 men undergoing radical prostatectomy with extended pelvic lymph node dissection (ePLND) and preoperative 68Ga-PSMA PET/CT. For each nomogram, we fitted a baseline model and a model with the addition of PSMA PET nodal involvement (positive/negative). Discrimination (AUC) was compared using the DeLong test; model fit with the likelihood-ratio (LR) test; clinical utility with decision curve analysis (DCA).

Results: For Briganti-2019, AUC numerically increased from 0.744 to 0.828 (ΔAUC +0.084; DeLong P = .342), with improved fit (Δ-2LL 12.633; P < .001). For MSKCC, AUC numerically increased from 0.813 to 0.830 (ΔAUC +0.017; DeLong P = .620), with better fit (Δ-2LL 9.306; P = .002). DCA showed a consistent net-benefit gain for Briganti-2019 with PSMA PET across 5%-20% thresholds (largest around 15%), whereas gains for MSKCC were modest and confined to higher thresholds.

Conclusion: Adding PSMA PET/CT nodal status improved model fit and suggested potential incremental clinical utility, particularly for Briganti-2019; however, statistically significant improvement in discrimination was not demonstrated by DeLong testing. PSMA PET/CT and clinical nomograms may be considered complementary tools for selecting candidates for ePLND.