Prediction of variceal hemorrhage: a prospective study

Prophylactic therapy to prevent the first variceal hemorrhage ceased being recommended for any cirrhotic because risks outweighed benefits. In this study, we identified cirrhotics that have never bled who are most prone to hemorrhage from varices. We developed a numerical grading system which combined size with other endoscopic characteristics, so varices would get an aggregate grade of 1 to 10. Varices were classified as either high grade (scored greater than or equal to 8) or low grade (scored less than or equal to 7). The grading system had a predictive value for bleeding when studied prospectively for a mean duration of 26 months in 52 patients subjected to placebo (15 patients) or various treatments (37 patients) in a randomized trial. High grade varices had a significantly higher association with bleeding than did low grade varices; 73% versus 7% (p less than 0.001) analyzing all 52 patients (of which 21% bled), and 83% versus 11% (p less than 0.05) analyzing the 15 placebo-treated patients (of which 40% bled). An esophagram positive for varices proved to be a sensitive screen for high grade varices, picking up 91% of them. However, it was a poor predictor of bleeding (one-third bled), since only 45% of radiologically evident varices were high grade. Since almost all cirrhotics with high grade varices can be expected to bleed within 2 yr, this group should be the focus of controlled trials to determine whether low risk prophylactic therapies reduce the incidence of first variceal hemorrhage.