Is transient elastography valuable for high-risk esophageal varices prediction in patients with hepatitis-B-related cirrhosis?
Version of Record online: 21 FEB 2012
© 2011 Journal of Gastroenterology and Hepatology Foundation and Blackwell Publishing Asia Pty Ltd
Journal of Gastroenterology and Hepatology
Volume 27, Issue 3, pages 533–539, March 2012
How to Cite
Chen, Y. P., Zhang, Q., Dai, L., Liang, X. E., Peng, J. and Hou, J. L. (2012), Is transient elastography valuable for high-risk esophageal varices prediction in patients with hepatitis-B-related cirrhosis?. Journal of Gastroenterology and Hepatology, 27: 533–539. doi: 10.1111/j.1440-1746.2011.06889.x
- Issue online: 21 FEB 2012
- Version of Record online: 21 FEB 2012
- Accepted manuscript online: 22 AUG 2011 03:49PM EST
- Accepted for publication 20 July 2011.
- esophageal varices;
- hepatitis B;
- liver stiffness measurement;
- predicting algorithm
Background and Aim: The aim of this study was to evaluate the clinical value of transient elastography (TE) for high-risk esophageal varices (HREV) prediction in hepatitis-B-related cirrhosis patients.
Methods: A total of 238 patients with hepatitis B cirrhosis were prospectively enrolled. All patients had undergone TE and upper gastrointestinal endoscopy. Diagnostic value was assessed by the area under ROC curve (AUROC), predictive value and likelihood ratio.
Results: The size of esophageal varices correlated with liver stiffness with Kendall's tau_b 0.236 overall and 0.425 in patients with ALT ≥ 5 × upper limit of normal (ULN). The AUROC of TE predicting HREV was 0.73 (95% confidence interval 0.66–0.80) overall and 0.92 (0.82–1.01) for patients with ALT ≥ 5 × ULN. In patients with ALT ≥ 5 × ULN, cut-off 36.1 kPa predicted HREV with a 100% negative predictive value (NPV), an indefinite negative likelihood ratio (NLR), a 72.7% positive predictive value (PPV) and a positive likelihood ratio (PLR) of 9.3. The AUROC of HREV-predicting model, constructed by ultrasonography and TE (USLS), was 0.84 (0.77–0.90) in the training set and 0.85 (0.76–0.94) in the validating set. Cut-off 3.30 excluded HREV with NPV 0.946 and NLR 0.10, and cut-off 5.98 determined HREV with PPV 0.870 and PLR 10.24. Using USLS, nearly 50% of patients could avoid endoscopic screening. The model's predictive values were maintained at similar accuracy in the validation set. Differences of AUROC in USLS, liver stiffness/spleen diameter to platelet ratio score and ultrasonic score were not significant.
Conclusions: TE may predict HREV in patients with ALT ≥ 5 × ULN. Overall, the clinical values of TE and USLS for HREV prediction should be evaluated by further studies.