Distinguishing between Thymic Epithelial Tumors and Benign Cysts via Computed Tomography
- Author:
Sang Hyup LEE
1
;
Soon Ho YOON
;
Ju Gang NAM
;
Hyung Jin KIM
;
Su Yeon AHN
;
Hee Kyung KIM
;
Hyun Ju LEE
;
Hwan Hee LEE
;
Gi Jeong CHEON
;
Jin Mo GOO
Author Information
- Publication Type:Original Article
- Keywords: Mediastinum; Cystic mass; Thymic tumor; Computed tomography; FDG-PET/CT
- MeSH: Humans; Logistic Models; Mediastinum; Multivariate Analysis; Pleura; Positron-Emission Tomography; ROC Curve; Thymus Neoplasms; Tomography, X-Ray Computed
- From:Korean Journal of Radiology 2019;20(4):671-682
- CountryRepublic of Korea
- Language:English
- Abstract: OBJECTIVE: To investigate whether computed tomography (CT) and fluorine-18-labeled fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) may be applied to distinguish thymic epithelial tumors (TETs) from benign cysts in the anterior mediastinum. MATERIALS AND METHODS: We included 262 consecutive patients with pathologically proven TETs and benign cysts 5 cm or smaller who underwent preoperative CT scans. In addition to conventional morphological and ancillary CT findings, the relationship between the lesion and the adjacent mediastinal pleura was evaluated qualitatively and quantitatively. Mean lesion attenuation was measured on CT images. The maximum standardized uptake value (SUVmax) was obtained with FDG-PET scans in 40 patients. CT predictors for TETs were identified with multivariate logistic regression analysis. For validation, we assessed the diagnostic accuracy and inter-observer agreement between four radiologists in a size-matched set of 24 cysts and 24 TETs using a receiver operating characteristic curve before and after being informed of the study findings. RESULTS: The multivariate analysis showed that post-contrast attenuation of 60 Hounsfield unit or higher (odds ratio [OR], 12.734; 95% confidence interval [CI], 2.506–64.705; p = 0.002) and the presence of protrusion from the mediastinal pleura (OR, 9.855; 95% CI, 1.749–55.535; p = 0.009) were the strongest CT predictors for TETs. SUVmax was significantly higher in TETs than in cysts (5.3 ± 2.4 vs. 1.1 ± 0.3; p < 0.001). After being informed of the study findings, the readers' area under the curve improved from 0.872–0.955 to 0.949–0.999 (p = 0.066–0.149). Inter-observer kappa values for protrusion were 0.630–0.941. CONCLUSION: Post-contrast CT attenuation, protrusion from the mediastinal pleura, and SUVmax were useful imaging features for distinguishing TETs from cysts in the anterior mediastinum.
