胃间质瘤超声内镜及增强CT特征与病理危险度的相关性研究

Correlation between endoscopic ultrasonography and enhanced CT features and pathological risk of gastric stromal tumors

摘要:

目的 探讨胃间质瘤超声内镜、增强计算机断层成像（computed tomography，CT）图像特征与病理危险度的相关性。方法 回顾性收集2018年1月至2022年11月武汉大学人民医院收治的原发性胃间质瘤病例，胃间质瘤须经内镜或外科手术后病理及免疫组化确诊，术前行超声内镜检查者共计286例（288处病灶），其中70例（71处病灶）有完整术前增强CT检查资料。超声内镜检查者按胃间质瘤病理危险度分成相对良性组（包含极低危和低危，245例247处病灶）和相对恶性组（包含中危和高危，41例41处病灶）；增强CT检查者也按胃间质瘤病理危险度分成相对良性组（38例39处病灶）和相对恶性组（32例32处病灶）。使用单因素分析和多因素logistic回归分析，探寻胃间质瘤病理危险度相对恶性的超声内镜图像特征及增强CT图像特征。结果 （1）单因素分析结果显示，超声内镜下病灶表面有无溃疡、病变长径、边界形态是否规则、内部回声是否均匀、有无囊性变方面组间差异有统计学意义（P<0.05）；logistic回归分析结果显示，超声内镜下病灶表面有溃疡（P=0.010，OR=4.519，95%CI：1.432~14.257）、病变长径（P<0.001，OR=2.398，95%CI：1.688~3.406）是胃间质瘤病理危险度相对恶性的独立危险因素。超声内镜下胃间质瘤长径预测病理危险度相对恶性的受试者工作特征（receive operator characteristic，ROC）曲线分析结果显示，曲线下面积（area under curve，AUC）为0.891（95%CI：0.839~0.943），最大约登指数为0.667，对应的病变长径为1.55 cm，此时预测胃间质瘤病理危险度相对恶性的敏感度、特异度分别为90.2%和76.5%。（2）单因素分析结果显示，增强CT下病变长径、病变形态（圆或类圆形/不规则）、病变边界是否清晰、增强扫描是否均匀强化、有无坏死方面组间差异有统计学意义（P<0.05）；logistic回归分析结果显示，增强CT下病变长径（P=0.010，OR=2.533，95%CI：1.245~5.154）是胃间质瘤病理危险度相对恶性的独立危险因素。增强CT下胃间质瘤长径预测病理危险度相对恶性的ROC曲线分析结果显示，AUC为0.824（95%CI：0.724~0.924），最大约登指数为0.560，对应的病变长径为3.65 cm，此时预测胃间质瘤病理危险度相对恶性的敏感度、特异度分别为68.8%、87.2%。结论 超声内镜及增强CT下胃间质瘤长径与胃间质瘤病理危险度明显相关，可用于预测胃间质瘤病理危险度是否为相对恶性（中危和高危）。

Abstract:

Objective To investigate the correlation between endoscopic ultrasonography (EUS) and enhanced computed tomography (CT) image features and pathological risk of gastric stromal tumors (GSTs). Methods Data of primary GSTs patients treated at Renmin Hospital of Wuhan University from January 2018 to November 2022, confirmed by endoscopy or surgical pathology and immunohistochemistry were retrospectively reviewed. A total of 286 cases (288 lesions) underwent EUS before surgery, of which 70 cases (71 lesions) had complete preoperative enhanced CT examination data. Patients who underwent EUS were divided into relatively benign group (including very low-risk and low-risk, 245 cases of 247 lesions) and relatively malignant group (including intermediate-risk and high-risk, 41 cases of 41 lesions) based on the pathological risk of GSTs. Patients who underwent enhanced CT examination were also divided into relatively benign group (38 cases of 39 lesions) and relatively malignant group (32 cases of 32 lesions) based on the pathological risk of GSTs. Univariate analysis and multivariate logistic regression analysis were used to explore the independent risk factors associated with the malignant potential of GSTs based on EUS and enhanced CT imaging features. Results (1) Univariate analysis showed that there were significant differences among groups in terms of ulceration presence on lesion surface, long diameter, boundary regularity, internal echo uniformity and cystic change under EUS (P<0.05). Logistic regression analysis showed that ulcers on the lesion surface (P=0.010, OR=4.519, 95%CI:1.432-14.257) and lesion long diameter (P<0.001, OR=2.398, 95%CI:1.688-3.406) under EUS were independent risk factors for GSTs relative to malignancy. The result of receiver operator characteristic (ROC) curve analysis of GSTs long diameter predicting pathological risk relative to malignancy under EUS showed the area under curve (AUC) was 0.891 (95%CI: 0.839-0.943), the maximum Youden index was 0.667, and the corresponding long diameter of lesion was 1.55 cm. In this case, the sensitivity and specificity of predicting the pathological risk of GSTs were 90.2% and 76.5%, respectively. (2) Univariate analysis showed that there were significant differences in lesion long diameter, shape (round or quasi-round/irregular), boundary clarity, enhancement pattern, and necrosis under enhanced CT among groups (P<0.05). Logistic regression analysis showed that lesion long diameter under enhanced CT (P=0.010, OR=2.533, 95%CI:1.245-5.154) was an independent risk factor for GSTs relative to malignancy. The ROC curve analysis of GSTs long diameter predicting pathological risk relative to malignancy under enhanced CT showed that AUC was 0.824 (95%CI: 0.724-0.924), the maximum Youden index was 0.560, and the corresponding long diameter of the lesion was 3.65 cm. The sensitivity and specificity of predicting the pathological risk of GSTs were 68.8% and 87.2%, respectively. Conclusion The long diameter of GSTs under EUS and enhanced CT exhibits a significant association with the pathological risk of GSTs, which can be used to predict the relative malignancy of GSTs (medium to high risk).