Prognostic Value of AI-Assisted Lesion Tracking on End-of-Treatment PSMA PET in mCRPC Patients Treated with 177Lu-PSMA: A Retrospective, Single-Center Study

Abstract

This study aimed to explore the prognostic value of the artificial intelligence–assisted lesion tracking applied to prostate-specific membrane antigen (PSMA) PET in patients with metastatic castration-resistant prostate cancer (mCRPC) treated with PSMA radiopharmaceutical therapy. Methods: TRAQinform IQ was applied to baseline and end-of-treatment ⁶⁸Ga-PSMA-11 PET scans from 20 patients with mCRPC treated with ¹⁷⁷Lu-PSMA-617. Lesion response parameters and the TRAQinform Profile score—designed for early treatment response assessment—were generated. Survival analyses were performed. Results: A higher percentage of “new” lesions was associated with a shorter overall survival (OS) (hazard ratio, 1.03; P = 0.002), whereas a higher percentage of “disappeared” lesions was linked to improved OS (hazard ratio, 0.98; P = 0.028). Patients with a TRAQinform Profile score of 2.7 or greater had a shorter OS than those with a score of less than 2.7 (10.9 mo vs. 44.3 mo; P = 0.027). Conclusion: Artificial intelligence–assisted lesion-tracking analysis of PSMA PET is prognostic for OS in patients with mCRPC undergoing PSMA radiopharmaceutical therapy.