The value of enhanced computed tomography-based nomograph model in the differential diagnosis of gastric schwannoma and gastric stromal tumor
10.3760/cma.j.cn311367-20220323-00128
- VernacularTitle:基于增强计算机断层扫描的列线图模型鉴别胃神经鞘瘤与胃间质瘤的价值
- Author:
Xiaohui WANG
1
;
Wei SUN
;
Jingfeng ZHANG
;
Qiaoling DING
;
Risheng YU
Author Information
1. 中国科学院大学宁波华美医院放射科，宁波 315000
- Keywords:
Enhanced CT;
Imaging features;
Nomograms;
Gastric schwannoma;
Gastric stromal tumor
- From:
Chinese Journal of Digestion
2022;42(9):596-603
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct enhanced computed tomography (CT)-based nomograph model, to assist physicians in differentiating gastric schwannoma from gastric stromal tumor.Methods:From January 1, 2012 to January 1, 2022, at the Second Affiliated Hospital of Zhejiang University School of Medicine and Ningbo Hwamei Hospital, University of Chinese Academy of Sciences, 57 patients with gastric schwannoma and 275 patients with gastric stromal tumor confirmed by surgical pathology were retrospectively collected, among whom 39 patients with gastric schwannoma and 201 patients with gastric stromal tumor were enrolled in the training set, and the other 18 patients with gastric schwannoma and 74 patients with gastric stromal tumor were enrolled in the validation set. The contrast-enhanced CT imaging features (tumor size index, arterial phase CT value, venous phase CT value, necrosis, calcification, integrity of mucosal surface, and uniform enhancement, etc.) and clinical data (history of gastritis, carbohydrate antigen 19-9 (CA19-9), carcinoembryonic antigen, and monocyte to lymphocyte ratio (MLR), etc.) were collected. The least absolute shrinkage and selection operator (LASSO) regression analysis was used to screen the independent predictive factors of imaging features in the differential diagnosis of gastric schwannoma and gastric stromal tumor, and a nomograph model was constracted. Logistic regression analysis was used to analyze and screen the independent predictive factors of clinical indicators to distinguish gastric schwannoma from gastric stromal tumor, and a clinical control model was established. The receiver operating characteristic curve(ROC) was used to analyze the area under the curve (AUC) of the nomograph model in the training set and the verification set, and concordance index (CI) and decision curve analysis (DCA) were used to evaluate the predictive efficiency and clinical application value of the nomograph model. DeLong test was used for statistical analysis.Results:The results of LASSO regression analysis showed that tumor size index, arterial phase CT value, venous phase CT value, necrosis, calcification, integrity of mucosal surface, and uniform enhancement were independent predictive factors of imaging features in the differential diagnosis of gastric schwannoma and gastric stromal tumor(all P<0.05). The results of logistic regression analysis indicated that the history of gastritis ( OR=0.280, 95% confidence interval 0.138 to 0.566), CA19-9 ( OR=0.940, 95% confidence interval 0.890 to 0.993), carcinoembryonic antigen ( OR=0.794, 95% confidence interval 0.661 to 0.952), and MLR ( OR=0.087, 95% confidence interval 0.009 to 0.860) were independent predictive factors of clinical indicators in the differential diagnosis of gastric schwannoma and gastric stromal tumor ( P<0.001, =0.028, 0.013 and 0.037). The AUCs of the nomograph model in the training and validation set were 0.881 and 0.850, respectively, and the AUCs of the clinical control model in the training and validation set were 0.814 and 0.772, respectively, and the differences were statistically significant ( Z=2.57 and 1.96, P=0.005 and 0.030). The average CI of the nomograph model was 0.885. The results of DCA analysis showed that the overall benefit of the nomograph model was higher than that of the clinical control model. Conclusion:The enhanced CT-based nomograph model can effectively distinguish gastric schwannoma from gastric stromal tumor, and can help physicians to make precise clinical decisions.
