OBJECTIVE: The purpose of this study was to describe the CT features of gastroduodenal glomus tumors and determine the CT features differentiating glomus tumors from other subepithelial lesions. MATERIALS AND METHODS: CT images of 89 pathologically proven small (≤ 5 cm) gastroduodenal glomus tumors (n = 11), gastrointestinal stromal tumors (n = 44), schwannomas (n = 15), ectopic pancreata (n = 12), and leiomyomas (n = 7) were retrospectively reviewed. For subjective analysis, location, shape, growth pattern, enhancement pattern, and presence of necrosis, calcification, and surface ulceration were analyzed. For objective analysis, the long and short diameters of the lesion, ratio of long to short diameter, thickness of the overlying layer, CT attenuation of the lesion and aorta, and lesion-to-aorta ratio in the arterial, portal venous, and equilibrium phases were measured. Among these findings, significant CT variables were determined with the chi-square, Fisher exact, and Student t tests. Receiver operating characteristic analysis was used to determine the optimal cutoff values of thickness of the overlying layer and lesion-to-aorta ratio that differentiate glomus tumor from other subepithelial lesions. RESULTS: Typical location (antrum and duodenum), hemangioma-like enhancement pattern, overlying layer thickness 2.6 mm or greater, and portal venous phase lesion-to-aorta ratio 0.86 or greater were found to be significant variables for differentiating glomus tumors from other subepithelial lesions (p < 0.05). Among these CT findings, location and portal venous phase lesion-to-aorta ratio 0.86 or greater were highly sensitive findings for glomus tumor, and a hemangioma-like enhancement pattern was highly specific. All odds ratios were infinity. The odds ratio of overlying layer thickness of 2.6 mm or greater was 12.2 (95% CI, 2.9-51.8). CONCLUSION: A subepithelial lesion with a lesion-to-aorta ratio less than 0.86 in the portal venous phase or not in the gastric antrum or duodenum is never a glomus tumor. On the contrary, a subepithelial lesion with hemangioma-like enhancement during dynamic CT is essentially a glomus tumor. On the basis of these CT findings, glomus tumor can be differentiated from other subepithelial lesions with high diagnostic accuracy.
OBJECTIVE: The purpose of this study was to describe the CT features of gastroduodenal glomus tumors and determine the CT features differentiating glomus tumors from other subepithelial lesions. MATERIALS AND METHODS: CT images of 89 pathologically proven small (≤ 5 cm) gastroduodenal glomus tumors (n = 11), gastrointestinal stromal tumors (n = 44), schwannomas (n = 15), ectopic pancreata (n = 12), and leiomyomas (n = 7) were retrospectively reviewed. For subjective analysis, location, shape, growth pattern, enhancement pattern, and presence of necrosis, calcification, and surface ulceration were analyzed. For objective analysis, the long and short diameters of the lesion, ratio of long to short diameter, thickness of the overlying layer, CT attenuation of the lesion and aorta, and lesion-to-aorta ratio in the arterial, portal venous, and equilibrium phases were measured. Among these findings, significant CT variables were determined with the chi-square, Fisher exact, and Student t tests. Receiver operating characteristic analysis was used to determine the optimal cutoff values of thickness of the overlying layer and lesion-to-aorta ratio that differentiate glomus tumor from other subepithelial lesions. RESULTS: Typical location (antrum and duodenum), hemangioma-like enhancement pattern, overlying layer thickness 2.6 mm or greater, and portal venous phase lesion-to-aorta ratio 0.86 or greater were found to be significant variables for differentiating glomus tumors from other subepithelial lesions (p < 0.05). Among these CT findings, location and portal venous phase lesion-to-aorta ratio 0.86 or greater were highly sensitive findings for glomus tumor, and a hemangioma-like enhancement pattern was highly specific. All odds ratios were infinity. The odds ratio of overlying layer thickness of 2.6 mm or greater was 12.2 (95% CI, 2.9-51.8). CONCLUSION: A subepithelial lesion with a lesion-to-aorta ratio less than 0.86 in the portal venous phase or not in the gastric antrum or duodenum is never a glomus tumor. On the contrary, a subepithelial lesion with hemangioma-like enhancement during dynamic CT is essentially a glomus tumor. On the basis of these CT findings, glomus tumor can be differentiated from other subepithelial lesions with high diagnostic accuracy.
Authors: Danielle M Richman; Sree Harsha Tirumani; Jason L Hornick; Charles S Fuchs; Stephanie Howard; Katherine Krajewski; Nikhil Ramaiya; Michael Rosenthal Journal: Abdom Radiol (NY) Date: 2017-01
Authors: Yang Won Min; Ha Na Park; Byung-Hoon Min; Dongil Choi; Kyoung-Mee Kim; Sung Kim Journal: J Gastrointest Surg Date: 2014-12-04 Impact factor: 3.452