Related Experiment Video
Updated: Jun 24, 2026
Evaluation of the Effectiveness of Longitudinal Incision for Endoscopic Submucosal Excavation of Gastric Subepithelial Lesions
Published on: April 28, 2026
Multidetector row computed tomographic gastrography findings after endoscopic submucosal dissection for early gastric
In Joon Lee1, Se Hyung Kim, Jeong Min Lee
1Department of Radiology, Seoul National University Hospital, Seoul, Korea.
Purpose:
To retrospectively investigate multidetector row computed tomographic (CT) gastrography (CTG) findings after endoscopic submucosal dissection (ESD) of the stomach and to analyze the time evolution and factors for predicting residual tumor.
Materials And Methods:
During an 18-month period, ESD was performed for 92 early gastric cancers (EGCs) in 86 patients. All patients were followed up with CTG, and 6 patients underwent CTG 2times. A total of 98 CTGs were analyzed by 2 radiologists for the presence of mucosal break, overlying enhancing layer and the attenuation of the lesion on 2-dimensional (2D) images, and malignant-looking fold convergence on 3D volume-rendered images. Multidetector row computed tomographic gastrography findings were categorized into 5 types: deep benign ulcer (BU), deformed BU, shallow depressed ulcer, EGC, and advanced gastric cancer (AGC). To analyze the time evolution of CTG findings, the mean time interval between the procedure and CT was calculated and compared among the 5 types using the Kruskal-Wallis test. The chi test or Fisher exact test was used to determine significant CT findings for predicting residual tumor after ESD.
Results:
Tumor involvement was found at the resection margin in 7 lesions (6 radial and 1 deep). Of 98 lesions, 5 (5.1%) were not visualized on either 2D or 3D images. Most lesions had mucosal break (82/93) and showed intermediate low attenuation (67/93) on 2D images. Prominent enhancement of overlying layer above ESD site was seen in 23 lesions. Malignant-looking fold convergence was observed in 11 lesions. Sixty-five lesions appeared as shallow depressed ulcers, 13 as deformed BUs, 8 as EGCs, 4 AGCs, and 3 deep BUs. Among them, deep BUs appeared first (average, 0.2 month after ESD; P < 0.05), followed by deformed BUs (5.3 months), then AGCs (6.0 months), EGCs (7.3 months), and shallow depressed ulcers (7.7 months). However, the only significant time interval difference was between the deep BUs and the other types (P < 0.05). There was no correlation between the presence of marginal tumor involvement and any CT findings.
Conclusions:
Immediately after ESD, lesions appear as deep BUs then evolve over time into shallow depressed ulcers. The presence of malignant-looking fold convergence and enhancing layer above the mucosal break on CT do not indicate tumor involvement at the ESD margin.
