gastric varices are all associated with a greater propensity for High-resolution endoluminal sonography (HRES) was variceal bleeding. 6,7,8 However, several studies have shown used to image and measure esophageal varices in control significant interobserver variability in the interpretation of subjects and patients with portal hypertension and comesophageal variceal endoscopic appearance. [9][10][11][12] Variceal size pared with endoscopic findings. Nine control patients can only be approximated and not precisely measured by and 68 patients with known cirrhosis or noncirrhotic endoscopic appearance.
portal hypertension underwent videotaped HRES andEndoscopic ultrasonography has been shown to be superior videotaped esophagoscopy (EGD). Two blinded investito endoscopy in detecting gastric varices 13,14 ; however, its role gators reviewed the videotapes to determine the presin evaluating esophageal varices is still undefined. The aim ence and size of the largest esophageal varix in each of this study is to evaluate high-resolution endoluminal sonopatient. The largest varix by HRES was measured with graphy's (HRES) potential as a screening modality for dethe esophagus at rest at a point where the varix aptecting early esophageal varices and to compare measurepeared most circular. The largest varix seen on EGD was ment of esophageal variceal size by HRES with grading as graded on a 5-point scale. All nine of the control patients determined by endoscopy. were correctly identified by both EGD and HRES as grade I (no varices). Eight of the 10 patients with no varices seen on EGD had varices identified by HRES.
PATIENTS AND METHODS
The interobserver correlation for HRES was r Å .88 andPatients. Sixty-eight patients with biopsy-proven cirrhosis or for EGD was r Å .79. The correlation between EGD and known noncirrhotic portal hypertension underwent 76 upper endo-HRES was r Å .50. High resolution endoluminal sonogra-scopic and HRES examinations from November, 1993 to December, phy allows quantitative measurement of variceal size, is 1995. All patients were hemodynamically stable at the time of the a more sensitive and reproducible imaging modality study. Nine volunteers known to have normal upper gastrointestinal than esophagoscopy for the detection and sizing of tracts were used as controls. All patients gave informed consent beesophageal varices. There is poor correlation between fore undergoing endoscopy; the study was approved by the Temple University Hospital Investigational Review Board, Phila., PA. with portal hypertension is of great clinical relevance because on the end of a wire core that was connected to a motor drive on the of the significant morbidity, mortality, and hospitalization ultrasound machine. The 1.35-mm diameter transducer core was costs associated with variceal bleeding. 1,2 When esophageal placed into a flexible 2-mm diameter catheter, and rotated to produce varices are detected before an initial bleed, the prophylactic a 360Њ transaxial real-time image. One mL of sterile water was introuse of beta...