Multicenter External Validation and Optimization of a Proposed Nomogram for Prostate-Specific Membrane Antigen PET/CT Accuracy in Biochemical Recurrence.

INTRODUCTION

Prostate-specific membrane antigen (PSMA) PET/CT has been established as the standard imaging technique after biochemical recurrence (BCR) of prostate cancer (PCa). However, its availability is not widespread, thus, patient selection criteria are necessary. For this reason, a European nomogram was recently developed with the intention of helping to predict and identify those patients with BCR at high risk for a positive PSMA PET/CT. The aim of our study was to test the external validity of this nomogram in a large regional cohort of patients and its impact as a selective tool for patients with BCR who should undergo a PSMA PET/CT.

METHODOLOGY: A multicenter, observational, and retrospective study to validate, calibrate, and readjust the European PSMA PET/CT positivity prediction nomogram in a cohort of patients with BCR after radical treatments for localized PCa. Clinical and demographic data were analyzed. We evaluated the detection rate of PSMA PET/CT, the association of different variables with a positive PSMA PET/CT, and the accuracy of the nomogram, summarized in an ROC curve and a clinical decision curve. The nomogram was then modified and improved for our cohort.

RESULTS

A cohort of 413 patients with BCR undergoing PSMA PET/CT was evaluated. Median age, PSA, and PSAdt were 66 years, 0.52 ng/mL, and 7 months, respectively. Median time to BCR was 34 months and the predominant ISUP was 3 (31%). Most patients underwent radical prostatectomy (88%). PSMA PET/CT was positive in 67% of patients, with pelvic involvement in 32% and 24% positivity outside the pelvis. The independent variables associated with a positive PSMA PET/CT were PSA value (OR: 1.94 (1.2-3.19), with a PSA level ≥ 0.5 ng/dL), and a PSA persistence after primary treatment (OR 2.95 (95% CI 1.37-7.14)). The original nomogram had a low predictive ability, with an AUC of 0.57 (95% CI: 0.52-0.62). It was necessary to adjust and calibrate this to obtain a novel nomogram with an AUC of 0.84 (95% CI 0.70-0.98). The DCA showed a greater net benefit from the use of this nomogram at intermediate threshold levels.

CONCLUSION

The nomogram showed a low predictive ability in our external validation. Nevertheless, our novel nomogram demonstrated a moderate-high predictive ability, which could thus optimize the selection of BCR patients who are candidates for PSMA PET/CT.