Predictive Model Of Endoscopic Differentiated Early Gastric Cancer

Objective:To explore the predictive factors of(EGC)in differentiated early gastric cancer from the clinical and endoscopic characteristics,and to establish a Logistic regression prediction model to evaluate the predictive value of the model in judging differentiated EGC.Methods:From January2013 to May 2018,184 patients with early gastric cancer who underwent endoscopic submucosal exfoliation(ESD)in our hospital were selected as the subjects of the study.Univariate and multivariate analysis was performed on age,sex,tumor location,tumor size,microscopic morphology,depth of invasion,degree of differentiation,atrophy,vascular invasion and lymph node metastasis.According to the results of multivariate analysis,the Logistic regression prediction model of differentiated EGC was established,the ROC curve of Logit P predictive differentiated EGC was drawn,and the predictive value of the model for judging differentiated EGC was evaluated.Results:Logistic regression analysis showed that age<44 years,upper gastric lesions and size?2cm were independent predictors of differentiated EGC.According to the above independent predictors and their regression coefficients,a Logistic regression prediction model:Logit P=-5.715-0.112×age(year)+0.307×tumor site(upper stomach=1,middle stomach=2,lower stomach=3)+0.272×tumor size(?2cm=1,>2cm=0)+0.173×lymph node metastasis(yes=0,no=1),the critical value of Logit P is-1.26,the area under the ROC curve is 0.865,and the accuracy is 76.6%.Conclusion:For the first time,our study proposed a predictive model and predictive value for clinical and endoscopic judgment of differentiated EGC,and evaluated the predictive value of the model,and the accuracy of judging differentiated EGC was high.It provides the basis for early diagnosis of differentiated EGC and timely selection of reasonable treatment.