Prediction of the First Variceal Hemorrhage in Patients with Cirrhosis of the Liver and Esophageal Varices

TLDR

The study aimed to determine whether clinical and endoscopic features could predict first variceal hemorrhage in cirrhotic patients with esophageal varices. A prospective cohort of 321 patients was evaluated, with varices graded per Japanese guidelines and a prognostic index constructed from modified Child class, varice size, and red wale markings, then prospectively validated in 75 patients. During a median 23‑month follow‑up, 26.5% of patients bled, and the index identified a high‑risk group with a one‑year bleeding incidence exceeding 65%, ranging from 6% to 76% across risk strata. N Engl J Med 1988; 319:983–9.

Abstract

We conducted a prospective study of 321 patients with cirrhosis of the liver and esophageal varices with no history of bleeding to see whether a comprehensive analysis of their clinical features and of the endoscopic appearances of their varices could help to identify those at highest risk for bleeding. Varices were classified endoscopically as suggested by the Japanese Research Society for Portal Hypertension. Patients were followed for 1 to 38 months (median, 23), during which 85 patients (26.5 percent) bled. Multiple regression analysis (Cox's model) revealed that the risk of bleeding was significantly related to the patient's modified Child class (an index of liver dysfunction based on serum albumin concentration, bilirubin level, prothrombin time, and the presence of ascites and encephalopathy), the size of the varices, and the presence of red wale markings (longitudinal dilated venules resembling whip marks) on the varices. A prognostic index based on these variables was devised that enabled us to identify a subset of patients with a one-year incidence of bleeding exceeding 65 percent. The index was prospectively validated on an independent sample of 75 patients with varices and no history of bleeding. We conclude that our prognostic index, which identifies groups of patients with one-year probabilities of bleeding ranging from 6 to 76 percent, can be used to identify candidates for prophylactic treatment. (N Engl J Med 1988; 319:983–9.)