Prediction of esophageal varices in hepatic cirrhosis by noninvasive markers.

OBJECTIVE: To determine whether Model for End-stage Liver Disease (MELD) Child-Turcotte-Pugh (CTP) classification, AST to platelet ratio index (APRI), and laboratory tests could predict the presence of esophageal varices (EV) or varices which need prophylactic therapy (medium or large size EV).

METHODS: Three hundred patients with cirrhosis (193 men; mean age 53.1 years; majority with chronic C hepatitis) were prospectively analyzed. The presence of EV (any size and medium or large EV) was correlated with patients' characteristics (MELD, CTP classification, APRI, platelets count, and liver tests).

RESULTS: One hundred and seventy-one patients (57%) had EV, of whom 35% (105) had varices which need prophylactic therapy (VPT). The distribution of EV according to CTP classification was as follows: A, 49%; B, 75.3% and C, 80%. Independent predictors of EV were: MELD higher than 8 (P=0.02); APRI higher than 1.64 (P=0.01); platelet count lower than 93,000/mm³ (P<0.01); aspartate aminotransferase higher than 1.34 × UNL (P=0.01), and total bilirubin higher than 1 mg/dl (P=0.04). MELD higher than 8 had the highest discriminative value for presence of EV (sensitivity=80.1%; specificity=51.2%; area under receiver operating characteristics=0.68). Factors independently associated with VPT were: thrombocytopenia (<92,000/mm³; P<0.01) and aspartate aminotransferase higher than 1.47 × UNL (P=0.03). Platelet count lower than 92,000/mm³ had sensitivity of 65.7%, specificity of 57.9%, and an area under receiver operating characteristics of 0.62 for the presence of VPT.

CONCLUSION: High values on MELD are associated with EV and thrombocytopenia, with varices which need prophylactic therapy. As a result of their low sensitivity and specificity, it is suggested to maintain the recommendation of upper gastrointestinal endoscopy for all patients with cirhosis.