Esophageal varices: noninvasive diagnosis with duplex Doppler US in patients with compensated cirrhosis.

Abstract

To prospectively develop and evaluate the accuracy of a duplex Doppler ultrasonographic (US) index for predicting the presence or absence of esophageal varices in patients with compensated cirrhosis (Child-Pugh class A) by using endoscopy as the reference standard. The study had institutional review board approval; all participants gave informed consent. Data in a total of 383 prospectively enrolled patients who underwent duplex Doppler US and screening endoscopy were divided into training (n = 240) and validation (n = 143) sets. Duplex Doppler US indexes, including mean portal vein velocity (PVV), hepatic impedance indexes, splenic impedance indexes, and the splenic index were evaluated with univariate and multivariate logistic regression analyses to find the independent factors predictive of the presence of esophageal varices. Receiver operating characteristic (ROC) curves were constructed for these factors to evaluate diagnostic accuracy in the training set and reproducibility in the validation set. Multivariate logistic regression analysis showed that splenic index and mean PVV were predictive of the presence of esophageal varices in the training set. A splenoportal index (SPI) was calculated as the splenic index divided by mean PVV to amplify the opposite effects on esophageal varices. Areas under ROC curves for SPI were significantly higher than those for the splenic index (0.93 vs 0.90, P = .02) and mean PVV (0.93 vs 0.67, P < .001) in the training set and in the validation set (0.96 vs 0.91 for splenic index, P = .01; 0.93 vs 0.80 for mean PVV, P < .001). An SPI threshold of 3.0 had 92% sensitivity, 93% specificity, 91% positive predictive value, and 94% negative predictive value for esophageal varices. Applying this cutoff value correctly predicted the presence or absence of esophageal varices in 92% of the patients without screening endoscopy. SPI can serve as a useful noninvasive index to predict the presence or absence of esophageal varices.