Two-Dimensional and Point Shear-Wave Elastography to Predict Esophageal Varices and Clinically Significant Portal Hypertension in Patients with Chronic Liver Disease.

The non-invasive assessment of disease severity remains pivotal in patients with chronic liver disease (CLD) as it has wide implications in predicting liver-related complications or death. Shear-wave elastography (SWE) is an emerging ultrasound-based method to non-invasively measure liver stiffness. The aim of our study was to evaluate two-dimensional (2D) and point (p) SWE to predict the presence of esophageal varices (EV) or clinically significant portal hypertension (CSPH). This was a retrospective analysis of a prospectively performed cohort study of patients with CLD treated in the outpatient clinic of the Frankfurt University Hospital. PSWE using the Hitachi HI Vision ASCENDUS system and the Siemens ACUSON S2000 system or 2D-SWE using the Toshiba APLIO500 system were analyzed at baseline and during follow-up to predict EV or surrogate parameters of CSPH. ROC curves were calculated for pooled liver stiffness measurements (LSMs) using a bootstrap approach. A combined model of SWE and platelet count was created and a mixed-effect logistic regression analysis using log-transformed values was performed. Overall, 511 patients with CLD and 919 consecutive LSMs were included and 315 patients (61.6%) had signs of CSPH. 2D-SWE performed best to predict EV and CSPH, and the addition of platelet count to the predictive model significantly increased test results for EV (AUC 0.83, 95%-CI: 0.76-0.89; difference in AUC 0.11, 95%-CI: 0.03-0.19, = 0.004), but only marginally for CSPH (AUC 0.75, 95%-CI: 0.64-0.85; difference in AUC 0.06, 95%-CI: 0.02-0.14, = 0.150). LSM > 18.5 and >20 kPa were indicative of CSPH and EV, while LSM < 10 kPa and <11 kPa ruled out CSPH and EV, respectively. Our study found that 2D-SWE in combination with platelet count performed best (in comparison to the other SWE methods) to predict EV or CSPH in patients with CLD. Future prospective trials are needed to validate our results.