Correlation between imaging features and pathological risk of gastric stromal tumors in endoscopic ultrasonography and enhanced computed tomography

BACKGROUND: Endoscopic ultrasonography (EUS) and enhanced computed tomography (CT) can offer more image features of gastric stromal tumors with higher pathological risk, which is of great significance for determining the pathological risk degree of gastric stromal tumors (GSTs) before surgery. AIM: To investigate the correlation between EUS and enhanced CT image features and pathological risk of GSTs. METHODS: Retrospectively collect the cases of primary GSTs. The lesions were divided into four group based on the pathological risk degree. Univariate analysis was used to identify the risk factors for higher pathological risk GSTs. The relatively benign group consisted of the very low-risk group and the low-risk group, while the relatively malignant group consisted of the intermediate-risk group and the high-risk group. Independent risk factors for the relative malignancy of GSTs pathological risk degree were analyzed by multivariate logistic regression. RESULTS: (1) Statistically significant disparities among groups regarding the presence of ulcer on the lesion surface, length, boundary morphology, internal echo uniformity and cystic change (P < 0.05). Lesion length (P < 0.001) and irregular boundary morphology of the lesion (P = 0.003) were independent risk factors. Receiver operating characteristic (ROC) curve analysis for predicting the relative malignancy of GSTs using EUS revealed that the area under the curve (AUC) was 0.876. (2) Statistically significant differences among groups in the lesion length, shape (round or quasi-round/irregular), boundary clarity, enhancement pattern, and necrosis (P < 0.05). Lesion length (P = 0.031) and the irregular lesion shape (P = 0.019) were independent risk factors. ROC curve analysis for predicting the relative malignancy of GSTs using enhanced CT showed that the AUC was 0.795. CONCLUSION: The image features of GSTs under EUS and enhanced CT is significantly correlated with the pathological risk of GSTs, which can be used to predict whether the pathological risk of GSTs is relatively malignant.