Nomogram for predicting the risk of gastroesophageal varices after primary endoscopic prophylaxis for variceal hemorrhage in patients with cirrhosis

Abstract

Background and aims: The aim of this study was to identify predictors of non-high-risk gastroesophageal varices and evaluate the probability of the residual high-risk varices in cirrhosis patients after the primary endoscopic treatment.

Patients and methods: Medical records of the patients with cirrhosis admitted for primary endoscopic prophylaxis gastroesophageal varices hemorrhage were retrospectively analyzed. The patients were divided into high-risk varices and non-high-risk varices groups according to the endoscopy. A nomogram was developed based on the results of multivariate Cox analyses. Accuracy of this model was validated by the concordance index (Harrell's c-index) and calibration curve.

Results: Altogether 117 patients were enrolled between March 2014 and April 2018. The multivariate Cox analyses identified spleen length <140 mm [odds ratio (OR) = 2.715; P = 0.037), small or medium size of esophageal varices (OR = 4.412; P = 0.017), unaccompanied with gastric varices (OR = 7.025; P = 0.003) and frequency of endoscopic variceal ligation ≥one time per 4 months (OR = 3.834; P = 0.034) as independent factors of non-high-risk varices. All significant predictors were incorporated into a nomogram to predict the residual high-risk varices, which showed a notable accuracy with the concordance index (0.833).

Conclusion: The nomogram-based prediction of residual high-risk varices can be used for risk stratification in cirrhosis patients with gastroesophageal varices.