AJR:195, November 2010
Chiao-Yun Chen, Twei-Shiun Jaw, Deng-Chyang Wu, Yu-Ting Kuo, Chien-Hung Lee, Wan-Ting Huang, Gin-Chung Liu
OBJECTIVE. The purpose of this study was to evaluate the efficacy of MDCT in the dif-ferentiation of diseases involving giant gastric folds and to identify the features most useful for predicting the presence of malignant gastric disease.
MATERIALS AND METHODS. Blinded reviewers retrospectively analyzed 16-MDCT scans of 64 patients who had endoscopic findings of giant gastric folds. Thirty of these patients had histopathologic results showing scirrhous carcinoma; 15, large B-cell lym-phoma; 14, acute gastric mucosal lesions; and five, Menetrier disease. Unenhanced trans-parent volume-rendered images were assessed to determine the morphologic features of the whole stomach. Contrast-enhanced images were used to analyze wall thickness, stratification, enhancement pattern, and perigastric conditions. We used the Cramer phi-prime correlation coefficient to calculate the overall diagnostic accuracy of MDCT in several gastric disorders. Scatterplot analysis was used for overall diagnostic score, and analysis of the receiver operator characteristic curves of the neighboring two diseases was used to locate the cutoff values for best diagnostic accuracy. We used exact logistic regression to identify which MDCT image features were most predictive of the presence of malignant gastric disease.
RESULTS. The gastric wall was found to be significantly thicker in large B-cell lymphoma than in other disorders (p < 0.001). The overall diagnostic accuracy of MDCT in the four diseases was 100%. Loss of wall stratification was deemed the best MDCT predictor of the presence of malignancy.
CONCLUSION. MDCT may be a reliable means of noninvasive diagnosis in the care of patients with endoscopically detected giant gastric folds and may be useful for differentiating benign from malignant disease.