Prediction of large esophageal varices in patients with cirrhosis of the liver using clinical, laboratory and imaging parameters

Abstract

Background: It is currently recommended that all patients with liver cirrhosis undergo upper gastrointestinal endoscopy (UGIE) to identify those who have large esophageal varices (LEVx) that carry a high risk of bleeding and may benefit from prophylactic measures. This approach leads to unnecessary UGIE in those without LEVx. We tried to identify clinical, laboratory and imaging parameters that may predict the presence of LEVx and help select patients for UGIE.

Methods: This prospective study included newly diagnosed patients with cirrhosis and no history of gastrointestinal bleeding scheduled to undergo UGIE. Patients underwent detailed clinical examination, blood tests (hematology, liver function tests) and ultrasonography. Size of esophageal varices was assessed at UGIE; Paquet's grades 0-II were classified as small varices, and III-IV as LEVx. Association of LEVx with qualitative and quantitative parameters was studied using chi(2) and Mann-Whitney U-tests, respectively. Parameters found to be significant were tested in a forward-conditional multivariate logistic regression analysis to identify independent predictors. Receiver operating characteristic curve analysis was used to assess the efficacy of prediction models.

Results: Of the 101 patients (median age 45; range 15-74 years; 87 male; Child-Pugh class: A 18, B 31, C 52), 46 had LEVx. On univariate analysis, five variables were significantly associated with the presence of LEVx. These included pallor (P = 0.026), palpable spleen (P = 0.009), platelet count (P < 0.002), total leukocyte count (P < 0.0004) and liver span on ultrasound (P = 0.031). On multivariate analysis, two of these parameters, namely low platelet count and presence of palpable spleen, were found to be independent predictors of the presence of LEVx. A receiver-operating characteristics curve using the predictor function arrived at from this analysis had an area under the curve of 0.760.

Conclusion: Presence of palpable spleen and low platelet count are independent predictors of presence of LEVx in patients with cirrhosis. Use of these parameters may help identify patients with a low probability of LEVx who may not need UGIE. This may help reduce costs and discomfort for these patients and the burden on endoscopy units.