Interreader agreement of intraprostatic prostate cancer detection, local extension and staging using [18F]PSMA-1007 PET and whole-mount radical prostatectomy specimens.

BACKGROUND: To investigate the diagnostic performance and interreader agreement of intraprostatic prostate cancer (PCa) detection, local extension and staging using [18F]PSMA-1007 PET in primary PCa, with whole-mount radical prostatectomy specimens as the reference standard. This prospective study included 66 patients with biopsy-proven PCa who underwent [18F]PSMA-1007 PET/MRI prior to radical prostatectomy. Four nuclear medicine physicians independently evaluated intraprostatic lesions, T stage, extracapsular extension (ECE), seminal vesicle invasion (SVI) and bilaterality, based on PROMISE V2 criteria and a standardized scoring template. Diagnostic accuracy was calculated using whole-mount histopathology as reference. Interreader agreement was assessed using Fleiss’ and weighted kappa statistics. RESULTS: Accurate lesion-level detection of dominant intraprostatic malignant lesions was achieved across readers with mean sensitivity of 96% (range 96–97%). Mean (range) sensitivity for detecting T3a, T3b, ≥T3, ECE, SVI and bilaterality across readers was 23% (0–50%), 39% (16–53%), 40% (10–63%), 33% (5–59%), 43% (16–58%) and 44% (34–53%), respectively. Mean (range) specificity for detecting T3a, T3b, ≥T3, ECE, SVI and bilaterality across readers was 86% (75–100%), 99% (98–100%), 85% (72–100%), 88% (74–100%), 99% (98–100%) and 86% (71–100%), respectively. Interreader agreement was substantial for bilaterality (kappa 0.66), moderate for SVI (kappa 0.57), fair for T stage (kappa 0.37) and slight for ECE (kappa 0.20). CONCLUSION: Accurate lesion-level detection of [18F]PSMA-1007 PET for dominant intraprostatic malignant lesions can be achieved across multiple readers. However, variable diagnostic performance of [18F]PSMA-1007 PET for local tumor extension was seen between readers, highlighting the need for standardized interpretation guidelines to improve reproducibility and clinical reliability. TRIAL REGISTRATION: ClinicalTrials.gov, NCT03327675. Registered 10 July 2017, https://clinicaltrials.gov/study/NCT03327675?term=NCT03327675%26rank=1 .