Predictors of distant metastases at 68Ga-PSMA-11 PET/CT in early-recurrent hormone-sensitive prostate cancer (HSPC) patients

Introduction: Prostate-specific-membrane-antigen/positron-emission-tomography (PSMA-PET) has shown better performance compared to conventional imaging in detecting disease relapse in prostate cancer (Pca) patients with early biochemical recurrence, allowing for more accurate restaging and leading to management changes based on personalized image-guided treatments. A high rate of management changes driven by PSMA-PET is observed in patients with evidence of distant metastatic involvement. These patients should thus be promptly identified and recommended for a PSMA-PET scan.Therefore, the aim of this study was to identify predictors of distant metastases (nodal M1a, skeletal M1b or visceral M1c) at PSMA-PET in early-recurrent hormone-sensitive Pca (HSPC) patients.

Methods: A retrospective analysis was conducted on four hundred forty-three (443) ⁶⁸Ga-PSMA-11 PET/CT scans performed between November 2016 and December 2021 at our institution. According to guidelines, Pca patients undergoing PSMA-PET were hormone-sensitive, hormone-therapy free, previously treated with radical intent (radiotherapy or radical-prostatectomy ± salvage-RT) and had proven biochemical recurrence or persistence (BCR/BCP).For each patient, the collected data included information about disease staging, histopathologic grading, primary and salvage treatments, and kinetics of prostate-specific antigen (PSA). The association between clinical parameters and distant metastases (M1a/b/c) at imaging was analyzed using the Mann-Whitney/Fisher’s exact tests and a complete series of uni-/multi-variate binary logistic regression models.

Results: Patients’ median PSA at the time of PSMA-PET scan was 0.60 [IQR:0.38–1.04] ng/mL. PSMA-PET successfully localized disease recurrence in 42.0% (186/443) of cases. Locoregional relapse (prostate bed or pelvic LNs) was detected in 21.0% (93/443) of cases. Metastatic involvement (M1a/b/c) was present in 17.6% (78/443) of cases: 7.7% (34/443) nodal (M1a), 9.9% (44/443) skeletal (M1b) and 2.5% (11/443) visceral (M1c). 80.8% (63/78) of metastatic cases were oligometastatic (≤ 3 PSMA-positive lesions).A significant association with distant metastases (M1a/b/c) at PSMA-PET was demonstrated for T-stage (≥ 3a, p=0.001), ISUP-grade 5 (p=0.001), clinical-setting (previous salvage-RT or BCP vs first-time BCR, p<0.001), PSA doubling-time (PSAdt, p<0.001) and PSA value at PET-scan (p<0.001).At univariate binary logistic regression analysis, T-stage (≥ 3a), ISUP-grade 5, clinical-setting (previous salvage-RT or BCP) and PSAdt were proven to be significant predictors of metastatic involvement (M1a/b/c). These results were confirmed also by the multivariate model (accuracy: 67.4% - Nagelkerke R²: 17%): T-stage (OR:1.92; 95CI%:1.04-3.58; p=0.039), ISUP-grade (OR:2.14; 95CI%:1.09-4.23; p=0.028), clinical setting (OR:2.82; 95CI%:1.58-5.04; p<0.001), and PSAdt (OR:0.95; 95CI%:0.92-0.99; p=0.012).

Conclusions: T-stage (≥3a), ISUP grade (5), clinical setting (previous salvage-RT or BCP), and PSAdt were found to be significant predictors of distant metastatic involvement (M1a/b/c), with a 5% risk increment for each single-unit decrement of PSAdt. These parameters could help to stratify early-BCR/BCP HSPC-patients according to the likelihood of distant metastases, in order to refine the indication for PSMA-PET.