Prognostic Nomogram model for the efficacy of endoscopic treatment in gastric varices caused by liver cirrhosis

Objective To predict the efficacy of endoscopic tissue adhesives in the treatment of gastric varices in patients with liver cirrhosis by Nomogram model. Methods From August 2014 to September 2017, 158 patients with liver cirrhosis caused esophagogastric variceal bleeding and received endoscopic tissue adhesives treatment at Zhongshan Hospital, Fudan University were collected. All patients were followed for 12 months. The primary outcome was rebleeding. The factors of rebleeding after endoscopic treatment of esophagogastric varices were analyzed. Nomogram prognostic model was developed and compared with Child-Pugh grading, computed tomography angiography (CTA) and hepatic venous pressure gradient (HVPG) in prognostic accuracy in rebleeding after endoscopic treatment in liver cirrhosis caused esophagogastric varices. Univariate and multivaricate Cox regression analysis, Kaplan-Meier curve and log-rank test were performed for statistical analysis. Results During the follow-up, rebleading occurred in 18 cases (11.4%), 37 cases (23.4%) and 49 cases (31.0%) at 2, 6, and 12 months after endoscopic treatment. The results of univariate Cox regression analysis showed the risk factors of rebleeding after endoscopic treatment of gastric varices included gender, alcoholic liver cirrhosis, diabetes mellitus, Child-Pugh grade (Grade A vs. B or C), extraluminal vessels on CTA (presence vs. absence) HVPG ( 3 points), injection volume of tissue adhesive (≤ 3 mL vs. > 3 mL) (hazard ratio (HR)=0.575, 2.018, 1.562, 3.433, 2.945, 1.859, 2.743, 0.324, 1.840, 1.477, and 1.716; 95% confidence interval (CI) 0.305 to 1.084, 0.902 to 4.514, 1.753 to 6.724, 1.663 to 5.217, 1.012 to 3.415, 0.852 to 8.830, 0.079 to 1.335, 1.012 to 3.317, 0.839 to 2.602, and 0.935 to 3.152; all P<0.2). The results of multivariate Cox regression analysis indicated that Child-Pugh grade, extraluminal vessels by CTA, and HVPG (HR = 2.095, 95% CI 1.099 to 3.995, P = 0.025) were all independent risk factors of rebleeding after endoscopic treatment of gastric varices (HR=2.665, 2.886, and 2.095; 95% CI 1.339 to 5.300, 1.580 to 5.271, and 1.099 to 3.995; all P<0.05). Kaplan-Meier curves showed that Child-Pugh grade (Grade A vs. B or C), extraluminal vessels on CTA (presence or absent) and HVPG (<16 mmHg vs. ≥16 mmHg) could effectively predict cumulative non-rebleeding rate in one year after endoscopic treatment of gastric varices, and the differences were statistically significant (all P<0.05). Receiver operataring characteristic curve analysis demonstrated that the predictive value of the model combined with Child-Pugh grade, extraluminal vessels on CTA and HVPG was higher than that of Child-Pugh grade and HVPG (AUC=0.746, 0.673 and 0.585; 95% CI 0.662 to 0.829, 0.583 to 0.762, and 0.486 to 0.683; P<0.01, P=0.001 and P=0.089, respectively). Patients were divided into low, medium, and high-risk groups according to the 25th and 75th percentiles of the Nomogram score. The results showed that Nomogram model could effectively distinguish high-risk groups of rebleeding after endoscopic treatment of gastric varices, and the difference was statistically significant (P <0.01). Conclusions Extraluminal vessels on CTA, HVPG and Child-Pugh grade are independent prognostic evaluation indexes of rebleeding after endoscopic treatment of gastric varices. The predictive accuracy of Nomogram model based on these three prognostic factors may be better than Child-Pugh grade and HVPG. Key words: Liver cirrhosis; Prognosis; Nomograms; Gastric varices; Endoscopic treatment