Background: Predicting liver-related events (LREs) in stable cirrhosis is crucial; however, traditional scoring methods perform poorly. Ammonia, due to its pathophysiological integration, may prognosticate LRE. This study tests the hypothesis that hyperammonemia can predict LRE and evaluates the AMMON-OHE score in a prospective cohort. Methods: A total of 280 patients with cirrhosis were prospectively enrolled and followed until LRE, liver transplant, or death. Ammonia was measured at enrollment and periodically, along with other parameters. LRE at 1 year was defined as hospitalization for ascites, variceal bleeding, hepatic encephalopathy, or bacterial infection. AMN ≥ULN was calculated by adjusting ammonia to the lab’s upper normal limit. Cox regression and time-dependent analyses were performed. AMMON-OHE was validated using TRIPOD+AI and decision curve analysis. Results: Most patients were male (60.7%), with a mean age of 66.1±10.3 years. MASLD was the commonest etiology (43.7%). Median CTP and MELD scores were 6 (IQR: 5–7) and 8 (IQR: 6–11). Overall, 60 (21.4%) developed LRE at 1 year, and 23 (8.2%) died. In multivariable analysis, AMN ≥ULN was an independent predictor of LRE (HR: 4.92; 95% CI: 2.80–8.64). The AMMON-OHE score predicted LRE with a time-dependent AUROC of 0.859 (CI: 0.810–0.907), outperforming MELD ( p =0.014) but similar to CTP ( p =NS). The AMMON-OHE score underestimated the risk of LRE in our cohort. Conclusions: AMN ≥ULN in stable outpatients with cirrhosis can predict LRE. Although the AMMON-OHE score has a high AUROC, it underestimates the risk of LRE in this cohort.
Ling et al. (Thu,) studied this question.