Department of Nuclear Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea.
Abstract
In this retrospective study, we investigated whether the (18)F-FDG uptake pattern and CT findings improved the accuracy over the standardized uptake value (SUV) for differentiating benign from malignant focal thyroid lesions incidentally found on (18)F-FDG PET/CT. We also defined the prevalence of these lesions and their risk for cancer.
METHODS: (18)F-FDG PET/CT was performed on 1,763 subjects without a previous history of thyroid cancer from May 2003 to June 2004. Two nuclear medicine physicians and 1 radiologist interpreted PET/CT images, concentrating on the presence of focal thyroid lesions, the maximum SUV of the thyroid lesion, the pattern of background thyroid (18)F-FDG uptake, and the CT attenuation pattern of the thyroid lesion.
RESULTS: The prevalence of focal thyroid lesions on PET/CT was 4.0% (70/1,763). Diagnostic confirmation was done on 44 subjects by ultrasonography (US)-guided fine-needle aspiration (n = 29) or US with clinical follow-up (n = 15). Among 49 focal thyroid lesions in these 44 subjects, 18 focal thyroid lesions of 17 subjects were histologically proven to be malignant (papillary cancer in 16, metastasis from esophageal cancer in 1, non-Hodgkin's lymphoma in 1). Therefore, the cancer risk of focal thyroid lesions was 36.7% on a lesion-by-lesion basis or 38.6% on a subject-by-subject basis. The maximum SUV of malignant thyroid lesions was significantly higher than that of benign lesions (6.7 +/- 5.5 vs. 10.7 +/- 7.8; P < 0.05). When only the maximum SUV was applied to differentiate benign from malignant focal thyroid lesions for the receiver-operating-characteristic curve analysis, the area under the curve (AUC) of PET was 0.701. All 16 focal thyroid lesions with very low attenuation or nonlocalization on CT images, or with accompanying diffusely increased thyroid (18)F-FDG uptake, were benign. When those lesions were regarded as benign lesions, irrespective of the maximum SUV, the AUC of PET/CT was significantly improved to 0.878 (P < 0.01).
CONCLUSION: Focal thyroid lesions incidentally found on (18)F-FDG PET/CT have a high risk of thyroid malignancy. Image interpretation that includes (18)F-FDG uptake and the CT attenuation pattern, along with the SUV, significantly improves the accuracy of PET/CT for differentiating benign from malignant focal thyroid lesions.