食管内镜黏膜下剥离术后大面积狭窄的危险因素分析

Risk factors of large area stenosis after endoscopic submucosal dissection of esophagus

目的探讨食管内镜黏膜下剥离术(ESD)后大面积狭窄的危险因素。方法将我院行ESD治疗的158例早期食管癌患者作为训练集,根据术后大面积狭窄发生情况将其分为狭窄组和未狭窄组;另选择我院行ESD治疗的76例早期食管癌患者作为验证集。通过Logistic回归分析筛选食管ESD后大面积狭窄发生的独立危险因素,将这些独立危险因素纳入并建立列线图预测模型。分别在训练集和验证集中对预测模型进行内部及外部验证:采用受试者工作特征(ROC)曲线及C-index验证模型区分度;GiViTI校准带及Hosmer-Lemeshow检验验证模型校准度;ROC曲线临界值与决策曲线分析法(DCA)曲线结合验证模型临床有效性。结果训练集患者中发生术后狭窄112例(70.89%)。肿瘤浸润深度为M3~SM1、剥离环周范围≥3/4环周、剥离纵径长度≥50 mm、固有肌层损伤、术中止血夹数量≥6个是食管ESD后大面积狭窄发生的独立危险因素(P<0.05)。建立的列线图模型预测食管ESD后大面积狭窄的发生率为69.74%。通过ROC曲线,训练集和验证集患者AUC分别为0.882(95%CI:0.820~0.955)和0.878(95%CI:0.774~0.928),模型区分度良好;GiViTI校准曲线带的80%~95%置信区间区域均未穿过45°角平分线(P>0.05);Hosmer-Lemeshow拟合优度检验中训练集P=0.581,验证集P=0.716。DCA曲线中阈值概率设定为30.4%,训练集和验证集临床净获益率分别为31%和35%,表明该预测模型具有临床有效性。结论肿瘤浸润深度为M3~SM1、剥离环周范围≥3/4环周、剥离纵径长度≥50 mm、固有肌层损伤、术中止血夹数量≥6个是食管ESD后大面积狭窄发生的独立危险因素,针对此类患者,应提前告知其发生术后狭窄的风险并考虑早期干预。

Objective To explore the risk factors of large area stenosis after endoscopic submucosal dissection(ESD)of esophagus.Methods A total of 158 patients with early esophageal cancer who were treated with ESD in our hospital were taken as the training set,and they were divided into the stenosis group and the non-stenosis group according to the occurrence of large area stenosis after operation.In addition,76 patients with early esophageal cancer who were treated with ESD in our hospital were taken as the verification set.Logistic regression analysis was used to screen the independent risk factors for the occurrence of large area stenosis after ESD of esophagus,and those independent risk factors were included to establish the nomogram prediction model.Internal and external verification of the prediction model in the training set and the verification set were performed,respectively:receiver operating characteristic(ROC)curve and C-index were used to verify the model discrimination degree;GiViTI calibration band and Hosmer-Lemeshow test were used to verify the model calibration degree;the combination of the cut-off value of ROC curve and the decision curve analysis(DCA)curve was used to verify the clinical validity of the model.Results Among the patients in the training set,112 cases(70.89%)had postoperative stenosis.The tumor invasion depth of M3 to SM1,the circumferential extent of dissection≥3/4 circumference,the length of longitudinal diameter of dissection≥50 mm,the injury of lamina propria and the number of intraoperative hemostatic clips≥6 were the independent risk factors for the occurrence of large area stenosis after ESD of esophagus(P<0.05).The incidence of the large area stenosis after ESD of esophagus predicted by the established nomogra m model was 69.74%.Through the ROC curve,the AUC in the training set and the verification set were 0.882(95%CI:0.820 to 0.955)and 0.878(95%CI:0.774 to 0.928),respectively,with good model discrimination degree.The 80%to 95%confidence interval region of the GiViTI calibration curve band did not cross the 45°angle bisector(P>0.05).In the Hosmer-Lemeshow goodness-of-fit test,P=0.581 was for the training set,and P=0.716 was for the verification set.The threshold probability value in the DCA curve was set as 30.4%,and the clinical net benefit rates of the training set and the verification set were 31%and 35%,respectively,which indicated that the prediction model had clinical validity.Conclusion The tumor invasion depth of M3 to SM1,the circumferential extent of dissection≥3/4 circumference,the length of longitudinal diameter of dissection≥50 mm,the injury of lamina propria and the number of intraoperative hemostatic clips≥6 are the independent risk factors for the occurrence of large area stenosis after ESD of esophagus.For such patients,the risk of postoperative stenosis should be informed in advance and early intervention should be considered.