Abstract
PURPOSE: To investigate the disease detection rate, diagnostic performance and interobserver agreement of fluciclovine ((18)F) PET-CT and multiparametric magnetic resonance imaging (mpMR) in recurrent prostate cancer. METHODS: Twenty-four patients with biochemical failure after non-prostatectomy definitive therapy, 16/24 of whom had undergone brachytherapy, underwent fluciclovine PET-CT and mpMR with interpretation by expert readers blinded to patient history, PSA and other imaging results. Reference standard was established via a multidisciplinary truth panel utilizing histology and clinical follow-up (22.9 ± 10.5 months) and emphasizing biochemical control. The truth panel was blinded to investigative imaging results. Diagnostic performance and interobserver agreement (kappa) for the prostate and extraprostatic regions were calculated for each of 2 readers for PET-CT (P1 and P2) and 2 different readers for mpMR (M1 and M2). RESULTS: On a whole body basis, the detection rate for fluciclovine PET-CT was 94.7% (both readers), while it ranged from 31.6-36.8% for mpMR. Kappa for fluciclovine PET-CT was 0.90 in the prostate and 1.0 in the extraprostatic regions. For mpMR, kappa was 0.25 and 0.74, respectively. In the prostate, 22/24 patients met the reference standard with 13 malignant and 9 benign results. Sensitivity, specificity and positive predictive value (PPV) were 100.0%, 11.1% and 61.9%, respectively for both PET readers. For mpMR readers, values ranged from 15.4-38.5% for sensitivity, 55.6-77.8% for specificity and 50.0-55.6% for PPV. For extraprostatic disease determination, 18/24 patients met the reference standard. Sensitivity, specificity and PPV were 87.5%, 90.0% and 87.5%, respectively, for fluciclovine PET-CT, while for mpMR, sensitivity ranged from 50 to 75%, specificity 70-80% and PPV 57-75%. CONCLUSION: The disease detection rate for fluciclovine PET-CT in non-prostatectomy patients with biochemical failure was 94.7% versus 31.6-36.8% for mpMR. For extraprostatic disease detection, fluciclovine PET-CT had overall better diagnostic performance than mpMR. For the treated prostate, fluciclovine PET-CT had high sensitivity though low specificity for disease detection, while mpMR had higher specificity, though low sensitivity. Interobserver agreement was also higher with fluciclovine PET-CT compared with mpMR.