Validation of AMMON-OHE Model for Predicting Overt Hepatic Encephalopathy after Transjugular Intrahepatic...

To explore the predictive performance of AMMON-OHE model in transjugular intrahepatic portosystemic shunt (TIPS) cohort and explore the prognostic relevance between overt hepatic encephalopathy (OHE) and all-cause mortality.

This single-center retrospective analysis included 278 cirrhotic patients receiving TIPS creation between 2018 and 2022. Pre-TIPS/before-discharge/1-month AMMON-OHE models were constructed using patients' clinical and biochemical variables at different time points and assessed using integrated Brier scores (IBSs), C-index, and decision curves analysis.

Within follow-up (median, 31.8 months), 91 (32.7%) patients developed OHE and 46 (16.5%) died. The AMMON-OHE models outperformed liver function scores in predicting OHE, with 1-month model performing the best (2-year: IBS, 0.1593; C-index, 0.7462). Age, diabetes, albumin, and ammonia were pre-TIPS independent predictors for OHE. Combining age of 65 years further improved the pre-TIPS AMMON-OHE model's predictive performance (2-year C-index, 0.7124-0.7482) and risk stratification ability, with 2-year OHE-free incidence further stratified in high-risk patients (from 55.1% to 37.0% and 61.0% [age >65 and ≤65 years]; P < .001) and low-risk patients (from 86.0% to 63.0% and 94.7% [age > and ≤65]; P = .021).OHE occurrence was associated with reduced survival rate (hazards ratio, 4.49; 95% CI, 2.44-8.26; P < .001), especially early, nonprecipitated, severe, or recurrent/persistent OHE (P < .001 for all). Such risk stratification was applicable to differentiate death risk, whether combining age (P = .002) or not (P = .019).

For cirrhotic patients treated with TIPS, the AMMON-OHE model can be used to identify high OHE-risk patients before TIPS. Furthermore, the occurrence of post-TIPS OHE is associated with increased mortality.