Elastography, spleen size, and platelet count identify portal hypertension in patients with compensated cirrhosis

Abstract

BACKGROUND & AIMS: Noninvasive methods are needed to identify clinically significant portal hypertension (CSPH) and esophageal varices (EVs) in patients with compensated cirrhosis. We looked for markers of the presence of CSPH and EVs in patients with cirrhosis. METHODS: We performed a cross-sectional study that included a training set of 117 patients with compensated cirrhosis, confirmed by histology, from a tertiary referral center. Spleen diameter was measured by ultrasound, and liver stiffness (LS) was measured by transient elastography; endoscopy was used as the standard for detection of EVs, and measurements of hepatic venous pressure gradient were used as the standard for identifying CSPH. We assessed the ability of platelet count, spleen diameter, LS, and combinations of these factors (ie, ratio of platelet count to spleen size, and LS × spleen size/platelet count [LSPS]) to identify patients with CSPH and EV. The analysis included 2 new statistical models: the PH risk score and the varices risk score. Results were validated using an independent series of 56 patients with compensated patients from another center. RESULTS: LS was the best single noninvasive variable for identifying patients with CSPH (area under the receiver operating characteristic, 0.883; 95% confidence interval [CI], 0.824-0.943; P < .0001). The area under the receiver operating characteristic value increased when LS was combined with platelet count and spleen size, either as LSPS (0.918; 95% CI, 0.872-0.965; P < .0001) or PH risk score (0.935; 95% CI, 0.893-0.977; P < .0001). More than 80% of patients were accurately classified using LSPS and PH risk score. Analyses of the varices risk score and LSPS were superior to all other noninvasive tests for identifying patients with EVs (area under the receiver operating characteristic, 0.909; 95% CI, 0.841-0.954 and 0.882; 95% CI, 0.810-0.935, respectively); they correctly classified 85% of patients in the training set and 75% in the validation set. CONCLUSIONS: Combined data on LS, spleen diameter, and platelet count can be used to identify patients with compensated cirrhosis most likely to have CSPH and EV.