Abstract
PURPOSE: Studies have indicated that PSMA-positive ganglia represent a diagnostic pitfall for nuclear medicine physicians. No studies have described choline and FDG uptake in ganglia, which may be a source of misdiagnosis. Herein, we described the percentage and uptake pattern of (68)Ga-PSMA, (11)C-choline and (18)F-FDG PET/CT in ganglia and evaluated the heterogeneous metabolic patterns of ganglia to differentiate from lymph node metastases (LNM). METHODS: Thirty-nine patients who underwent (11)C-choline PET/CT and 120 patients who underwent (68)Ga-PSMA PET/CT and (18)F-FDG PET/CT were retrospectively analyzed. The prevalence of PSMA-positive, choline-positive and FDG-positive ganglia was determined, the SUVmax of ganglia in different locations were measured, and the configuration was described. The SUVmax cutoff of PSMA-PET, choline-PET and FDG-PET was determined by ROC curve analysis to differentiate ganglia from LNM. RESULTS: 329 PSMA-positive ganglia were identified in 120 patients, 95 choline-positive ganglia were identified in 39 patients, and 39 FDG-positive ganglia were identified in 34 patients. PSMA-positive uptake was observed in 98.3%, 95.8%, and 80.0% of cervical, coeliac, and sacral ganglia, respectively. Choline-positive uptake was observed in 84.6%, 97.4%, and 61.5% of cervical, coeliac, and sacral ganglia, respectively. FDG-positive uptake was observed in 16.7%, 13.3%, and 2.5% of cervical, coeliac, and sacral ganglia, respectively. Cervical and coeliac ganglia had a higher rate of PSMA-positive uptake than sacral ganglia. Choline uptake was highest in coeliac ganglia followed by cervical and sacral ganglia. PSMA, choline or FDG uptake in LNM was all significantly higher than ganglia. ROC curve analysis revealed that at a 4.1 SUVmax cutoff of PSMA-PET, the sensitivity, specificity and accuracy of LNM identification was 88.4%, 97.9% and 96.2%, respectively. ROC curve analysis revealed that at a 2.35 SUVmax cutoff for choline-PET, the sensitivity, specificity, and accuracy of LNM identification was 95.0%, 92.6% and 93.0%, respectively. ROC curve analysis revealed that at a 2.55 SUVmax cutoff for FDG-PET, the sensitivity, specificity, and accuracy of LNM identification was 77.3%, 87.2%, and 81.9%, respectively. PSMA-, Choline- and FDG-positive ganglia are mainly band-shaped; most LNMs exhibited nodular and teardrop-shaped configuration. CONCLUSION: (68)Ga-PSMA and (11)C-choline uptake in ganglia was common, and FDG-positive ganglia were observed at lower frequency. Using (68)Ga-PSMA, (11)C-choline and (18)F-FDG uptake and anatomic location and configuration, the differentiation of ganglia from adjacent LNM is feasible.