ABSTRACT: : Radium-223 dichloride (Ra223) represents the unique bone-directed treatment option that shows an improvement in overall survival (OS) in metastatic castrate resistant prostate cancer (mCRPC). However, there is an urgent need for the identification of reliable biomarkers to non-invasively determine its efficacy (possibly improving patients' selection or identifying responders' after therapy completion). 18F-Fluorodeoxyglucose (FDG)-avidity is low in naïve prostate cancer, but it is enhanced in advanced and chemotherapy-refractory mCRPC, providing prognostic insights. Moreover, this tool showed high potential for the evaluation of response in cancer patients with bone involvement. For these reasons, FDG Positron Emission Tomography (FDG-PET) might represent an effective tool that is able to provide prognostic stratification (improving patients selection) at baseline and assessing the treatment response to Ra223. We conducted a retrospective analysis of 28 mCRPC patients that were treated with Ra223 and submitted to bone scan and FDG-PET/CT for prognostic purposes at baseline and within two months after therapy completion. The following parameters were measured: number of bone lesions at bone scan, SUVmax of the hottest bone lesion, metabolic tumor volume (MTV), and total lesion glycolysis (TLG). In patients who underwent post-therapy 18F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography (FDG-PET/CT), (20/28), PET Response Criteria in Solid Tumors (PERCIST), and European Organization for Research and Treatment of Cancer (EORTC) criteria were applied to evaluate the metabolic treatment response. The difference between end of therapy and baseline values was also calculated for Metabolic Tumor Volume (MTV), TLG, prostate-specific antigen (PSA), alkaline phosphatase (AP), and lactate dehydrogenase (LDH) (termed deltaMTV, deltaTLG, deltaPSA, deltaAP and deltaLDH, respectively). Predictive power of baseline and post-therapy PET- and biochemical-derived parameters on OS were assessed by Kaplan-Meier, univariate and multivariate analyses. At baseline, PSA, LDH, and MTV significantly predicted OS. However, MTV (but not PSA nor LDH) was able to identify a subgroup of patients with worse prognosis, even after adjusting for the number of lesions at bone scan (which, in turn, was not an independent predictor of OS). After therapy, PERCIST criteria were able to capture the response to Ra223 by demonstrating longer OS in patients with partial metabolic response. Moreover, the biochemical parameters were outperformed by PERCIST in the post-treatment setting, as their variation after therapy was not informative on long term OS. The present study supports the role of FDG-PET as a tool for patient's selection and response assessment in mCRPC patients undergoing Ra223 administration.