# Development and evaluation of a predictive model based on multi-frequency magnetic resonance elastography for high-risk esophagogastric varices in patients with cirrhotic portal hypertension

## TL;DR

A new non-invasive model using magnetic resonance elastography effectively identifies high-risk esophagogastric varices in cirrhotic patients, aiding early intervention.

## Contribution

A predictive model combining spleen stiffness and portal vein diameter from multi-frequency MRE for high-risk esophagogastric varices is developed and validated.

## Key findings

- Spleen stiffness and portal vein diameter are independent predictors of high-risk esophagogastric varices.
- The predictive model outperforms individual indicators in identifying high-risk patients.
- The model's robustness was confirmed through bootstrap resampling and sensitivity analyses.

## Abstract

This study aimed to assess the severity of esophagogastric varices (EGV) in patients with cirrhotic portal hypertension by measuring liver and spleen stiffness with multi-frequency magnetic resonance elastography (MRE), and to develop a predictive model for identifying high-risk EGV.

Fifty-four patients with cirrhosis and portal hypertension were enrolled. Clinical and imaging parameters were analyzed, and Bland–Altman analysis and intraclass correlation coefficient (ICCs) evaluated interobserver agreement. Independent-sample t-tests were used to analyze the differences in liver and spleen stiffness between high-risk and low-risk groups. The correlations between variables and the endoscopic EGV severity were analyzed by Spearman’s correlation analysis and univariate logistic analysis. Multivariate logistic regression identified independent predictors, and receiver operating characteristic (ROC) and decision curve analyses assessed model performance and clinical utility. Bootstrap resampling and sensitivity analyses validated model robustness.

Bland–Altman analysis and ICC analysis showed high consistency. There were significant differences in liver and spleen stiffness between low-risk and high-risk patients. Platelet count and plateletcrit were negatively correlated with variceal severity, whereas bilirubin, INR, liver/spleen stiffness, spleen volume, and portal/splenic vein diameters showed positive correlations. Multivariate analysis identified spleen stiffness and portal vein diameter as independent predictors. The predictive model y = −17.50 + 2.77× spleen stiffness +0.57× portal vein diameter demonstrated superior predictive performance compared to individual indicators. Bootstrap resampling and sensitivity analyses demonstrated the robustness and stability of the model’s predictive performance.

The model effectively identifies high-risk EGV, highlighting the potential of noninvasive MRE-based assessment to guide early intervention and reduce variceal bleeding risk.

The predictive model based on multi-frequency MRE and MRI parameters (spleen stiffness + portal vein diameter) can effectively screen out high-risk EGV, which is helpful for guiding timely clinical intervention, so as to reduce the risk of EGV bleeding in these patients, and improve their prognoses.

Multifrequency magnetic resonance elastography (MRE) can indicate the stiffness of the liver and spleen.The splenic stiffness is a reliable predictor of high-risk EGV.The predictive model can effectively screen out high-risk EGV.

Multifrequency magnetic resonance elastography (MRE) can indicate the stiffness of the liver and spleen.

The splenic stiffness is a reliable predictor of high-risk EGV.

The predictive model can effectively screen out high-risk EGV.