| Objective:Delayed bleeding is one of the serious complications of the endoscopic submucosal dissection.In this large cohort and multicenter study,we reviewed the medical records and endoscopic findings of patients who had esophagus and stomach endoscopic submucosal dissection.We aimed to perform a multicenter study to evaluate the incidence and risk factors of delayed bleeding after upper gastrointestinal cancer and precancerous lesions in a large cohort and to develop a risk-scoring model to determine the risk for delayed bleeding after endoscopic submucosal dissectionMethods:We performed a study at 3 centers(General Hospital of western war zone,the Third Military Medical University,Sichuan Mianyang 404 Hospital).Consecutive patients who underwent esophagus and stomach endoscopic submucosal dissection at these centers between August 2017 and August 2019 were enrolled.After screening,367 patients were finally included in the esophagus part and 255 patients in the stomach part.The incidence of delayed bleeding after endoscopic submucosal dissection of esophagus and stomach was statistically analyzed.The patients were divided into bleeding group and non-bleeding group according to whether or not delayed bleeding occurred.In the study,risk factors for delayed bleeding were identified by using the univariate logistic regression model.We developed the scoring model by assigning a weight to each risk factor based on the P-coefficient from the multivariate Logistic regression model for the cohort.The receiver operator characteristic curve was used to test the prediction efficiency of the model.Results:In the esophagus Part,delayed bleeding occurred in 20 of 375 patients(5.4%).In univariate analysis,lesion location,lesion size,lesion infiltration depth,submucosal macrovascular presence,submucosal fibrosis presence and surgical operation time between the two groups had significant differences(P<0.05).In multivariate analysis,The risk factors of delayed bleeding were large submucosal vessels(odds ratio[OR],4.776,95%confidence interval[CI],1.661-13.733,P=0.004)and surgical operation time>60min(odds ratio[OR]=5.480,95%confidence interval[Cl],1.057-28.404,P=0.043).These 2 factors were incorporated into a risk-scoring model for prediction of delayed bleeding.Finally,the prediction model was established:P(Y1=1)=1/(1+exp[-(0.802+1.564 x4+1.701x6)]).The x4 represents hyperplasia of large submucosal vessels,and x6 represents surgical operation time?60min.The Hosmer-Lemeshow test in the cohort suggested that this scoring model was a good fit to these data(?2=9.888,P=0.273).The receiver operator characteristic curve was used to test the prediction efficiency of the model.The results showed that the area under the receiver operator characteristic curve of the model was 0.858(P<0.05,95%confidence interval[CI],0.771-0.945),the prediction sensitivity was 82.7%and specificity was 85.0%,and the prediction accuracy was 94.6%.The cut-off point of the model was screened by the optimal Youden index,and the calculated value greater than 0.927 was associated with the risk of delayed bleeding.In the gastric Part,delayed bleeding occurred in 26 of 255 patients(10.4%).In univariate analysis,the lesion location,lesion size,lesion infiltration depth,submucosal macrovascular presence,submucosal fibrosis presence and operative time between showed significant differences in the two groups(P<0.05).In multivariate analysis,The risk factors of delayed bleeding were large submucosal vessels(odds ratio[OR],2.626,95%confidence interval[CI],1.027-6.715,P=0.044)and submucosal fibrosis presence(odds ratio[OR]=5.261,95%confidence interval[Cl],1.985-13.943,P=0.001).These 2 factors were incorporated into a risk-scoring model for prediction of delayed bleeding.Finally,the prediction model was established:P(Y2=1)=1/(1+exp[-(-3.243+0.965 x4+1.66x5)]).The Hosmer-Lemeshow test in the cohort suggested that this scoring model was a good fit to these data(?2=5.315,P=0.723).The receiver operator characteristic curve was used to test the prediction efficiency of the model.The results showed that the area under the receiver operator characteristic curve of the model was 0.832(95%confidence interval[CI],0.766-0.898,P=0.034),the prediction sensitivity was 79.5%and specificity was 76.9%,and the prediction accuracy was 89%.The cut-off point of the model was screened by the optimal Youden index,and the calculated value greater than 0.890 was associated with the risk of delayed bleeding.Conclusion:1.Endoscopic submucosal dissection is a safe and effective method for the treatment of early esophageal,gastric and precancerous lesions,with high cure rate and low recurrence rate.2.Hyperplasia of large vessels in submucosa and surgical operation time?60min are independent risk factors for delayed bleeding after esophageal endoscopic submucosal dissection.3.The proliferation of large submucosal vessels and the formation of submucosal fibrosis are independent risk factors for delayed bleeding after gastric endoscopic submucosal dissection.4.The risk of delayed bleeding after endoscopic submucosal dissection is significantly increased in patients with submucosal hyperplasia of thick vessels.It is suggested to strengthen coagulation and pruning of visible vessels during the operation,so as to avoid residual vascular remnants and avoid electrocoagulation syndrome.5.The operation experience of surgical surgeons is an important factor that affects the operation time.6.The risk-scoring model of delayed bleeding after early upper gastrointestinal cancer and precancerous lesions endoscopic submucosal dissection can accurately predict delayed bleeding after endoscopic submucosal dissection,and it is believed that through continuous improvement,it can provide scientific basis for prevention and treatment of delayed postoperative bleeding after endoscopic submucosal dissection more effectively. |
PDF Full Text Request