Abstract
591
Aim: We aim to characterize true disease extent in patients with non-metastatic castration-resistant prostate cancer (nmCRPC) using PSMA PET.
Methods: Following a predefined protocol, six high-volume PET centers retrospectively screened 8825 patient files from their PSMA PET databases. 200 nmCRPC patients were recruited. Patients had high risk for development of metastatic disease as defined by PSA doubling time ≤10 months or Gleason Score ≥8. PSMA PET was read locally (unblinded clinical read) and centrally (two independent, blinded readers) following PROMISE criteria. Primary endpoint was detection rate on a patient basis of PSMA PET. Secondary endpoints include tumor location, interobserver agreement and positive predictive value (PPV) validated by histopathology or composite follow-up.
Results: PSMA PET detected prostate cancer in 196 of 200 (98%) patients. 44% had pelvic disease only (24% local recurrence only, 20% pelvic nodal involvement), 55% had M1 disease despite negative conventional imaging (including 24% bone, 6% visceral). In 75 (38%) patients 116 regions were validated, including 30 regions with histopathology. PSMA PET PPV was 96% on a region basis (97% for histopathology validation only). Interobserver agreement for PET interpretation was near-perfect (Fleiss’ κ 0.81 to 0.91). Conclusion: PSMA PET localizes M1-disease with high PPV and reproducibility in more than half of “nm”CRPC patients.