Abstract: Objective To study on the value of endoscopic ultrasonography (EUS) in evaluating esophageal collateral circulation and predicting esophageal varices (EV) in patients with liver cirrhosis.Methods 76 patients with liver cirrhosis collected from January 2018 to January 2020 were investigated in the study. All patients had received EUS examination and recorded for the diameters and numbers of para-esophageal collateral veins (para-ECV) and peri-esophageal collateral veins (peri-ECV). The patients were divided into observation group (cirrhosis with EV) and control group (cirrhosis without EV) according to whether they were accompanied by EV. Logistic multivariate analysis was used to analyze the independent influence factors of liver cirrhosis with EV. A predictive model was established based on the results of the multi-factor analysis. Receiver operator characteristic (ROC) and the area under the curve (AUC) were used to analysis and verify the value of this model in predicting EV risk. Results Among the 76 patients, 32 cases (42.1%) were accompanied by EV. Within them, 18 cases were mild, 10 cases were moderate, and 4 were severe in EV. The EVs with mild and moderate cirrhosis accounted for 87.5%. The para-ECV and peri-ECV diameters of patients in the observation group (19 and 21 severe cases, respectively) were significantly larger than those in the control group, which were significantly more than the control group (15 and 14 severe cases, respectively) (P<0.05). There were 23 cases in the observation group who had lesions ≥5, which was significantly more than that in the control group (19 cases) (P<0.05). The Logistic multivariate analysis results showed that the antiviral therapy (β=-0.393, 95%CI=0.479-0.951, OR=0.675, P=0.025), para-ECV diameter (β=1.664, 95%CI=3.412-8.174, OR =5.281, P=0.000), peri-ECV diameter (β=0.658, 95%CI=1.454-2.562, OR=1.930, P=0.000) and the number of lesions (β=2.082, 95%CI=1.167-55.034, OR=8.014, P=0.034) were independent factors affecting EV. The ROC analysis results showed that without antiviral therapy (AUC=0.655, 95%CI=0.530-0.780, P=0.022), peri-ECV (AUC=0.693, 95%CI=0.572-0.814, P=0.004), para-ECV (AUC=0.651, 95%CI=0.525-0.776, P=0.026) and the number of lesions (AUC=0.678, 95%CI=0.555-0.801, P=0.009) had certain application value for predicting EV in liver cirrhotic patients. The prediction model based on the logistic multi-factor analysis was then established as: Y=0.465-0.393X₁+0.658X₂+1.664X₃+2.082X₄. It was shown by ROC analysis that the AUC value of the EV was 0.802 (SE=0.050, 95%CI=0.704-0.900,P=0.000), the sensitivity was 0.938 and the specificity was 0.545. The Hosmer-Lemeshow goodness-of-fit test results showed that there was no statistically significant difference between the actual observation results and the expected results of EV (χ²=0.154, P=0.695).Conclusion EUS examination is helpful in assessing the status of esophageal collateral circulation in patients with liver cirrhosis. The prediction model established on EUS examination has high accuracy for the risk of EV.

Key words: Endoscopic ultrasonography, Esophageal collateral veins, Liver cirrhosis, Esophageal varices