Predicting esophageal varices in cirrhosis: diagnostic utility of MR elastography and dynamic contrast-enhanced MRI
3
Issued Date
2025-01-01
Resource Type
ISSN
2366004X
eISSN
23660058
Scopus ID
2-s2.0-105020886597
Pubmed ID
41186713
Journal Title
Abdominal Radiology
Rights Holder(s)
SCOPUS
Bibliographic Citation
Abdominal Radiology (2025)
Suggested Citation
Singwicha T., Chansaengpetch S., Tanpowpong N. Predicting esophageal varices in cirrhosis: diagnostic utility of MR elastography and dynamic contrast-enhanced MRI. Abdominal Radiology (2025). doi:10.1007/s00261-025-05217-y Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/112998
Title
Predicting esophageal varices in cirrhosis: diagnostic utility of MR elastography and dynamic contrast-enhanced MRI
Author(s)
Author's Affiliation
Corresponding Author(s)
Other Contributor(s)
Abstract
Purpose: Liver cirrhosis is often complicated by esophageal varices (EVs) due to portal hypertension. Advances in MRI offer non-invasive methods for EV prediction. This study evaluates the diagnostic performance of liver stiffness (LS) and spleen stiffness (SS) from MR elastography (MRE) and EV visualization on dynamic contrast-enhanced (DCE)-MRI in cirrhotic individuals. Methods: This retrospective cross-sectional study included upper abdominal MRI studies from cirrhotic participants who underwent esophagogastroduodenoscopy (EGD) between February 2022 and July 2023. LS and SS from MRE were measured, while two radiologists independently assessed EV presence on DCE-MRI. Diagnostic performance was evaluated using EGD findings as the reference standard. Optimal LS and SS cut-off values were determined via ROC curve analysis, and sensitivity, specificity, likelihood ratios, predictive values, and accuracy were calculated. Results: Among 117 participants (median age 66, 72.6% male), chronic hepatitis B was the primary etiology (43.6%), and 74.4% had hepatocellular carcinoma (HCC). Optimal EV prediction cut-offs were 5.76 kPa for LS and 9.31 kPa for SS. For detecting any EVs, LS had 76% sensitivity and 76% specificity, while SS had 72% sensitivity and 79% specificity. For high-risk EVs, LS had 65% sensitivity and 77% specificity (AUC = 0.711), whereas SS had 83% sensitivity and 66% specificity (AUC = 0.780). DCE-MRI demonstrated sensitivity of 81–98% and specificity of 75–80%, with substantial inter-rater agreement (Kappa = 0.612). Conclusion: SS by MRE outperformed LS in predicting high-risk EVs, while DCE-MRI demonstrated potential as a non-invasive screening tool. These findings support the integration of noninvasive MRI-based methods for EV screening in cirrhotic populations.