Ultrasonographic findings with bleeding and nonbleeding esophageal varices

Real time ultrasonography was carried out in 49 cirrhotic patients within 10 days of esophagoscopy, and in 20 normal subjects. Among the cirrhotic patients, 42 had varices and 18 of these had hematemesis within 3 months of study. The varices were graded 0 to 4+ endoscopically; the diameter of the portal vein, the splenic vein, the hepatic artery, and the maximal length of the spleen were determined on ultrasound. In addition, portal collateral veins and the sudden amputation of portal vein branches in the liver were identified when present. There was a significant correlation of the diameter of the portal vein and the maximal spleen length with the magnitude of varices on endoscopy; there was no significant relationship between the splenic vein or the hepatic artery diameter. Less than half the patients with varices had sonographically demonstrated collaterals or portal vein branch amputation. A sonoscore was derived allotting one point each for enlarged portal vein (greater than 1.3 cm), enlarged spleen, collaterals or two or more amputated veins. The sonoscore correlated better with the endoscopic grade of varices than any other marker. The sonoscore among the patients with varices who bled was significantly higher (p less than 0.01) than any of the other measures. It is concluded that real time ultrasound can be used to screen for varices and to identify the need for endoscopy.