Prognostic performance of Child-Pugh and MELD scores in critically ill patients with hepatocellular carcinoma: A MIMIC-IV cohort study.

Abstract

e16235 Background: Hepatocellular carcinoma predominantly arises in cirrhotic patients, creating dual disease burden that complicates intensive care unit (ICU) prognostication. We evaluated the comparative performance of liver-specific prognostic scores in predicting ICU mortality among HCC patients. Methods: This retrospective cohort study utilized the MIMIC-IV database (2008-2022) to identify adult HCC patients admitted to ICU. We calculated Child-Pugh scores and Model for End-Stage Liver Disease (MELD) scores from admission laboratory values. Primary outcome was hospital mortality. We performed multivariable logistic regression to identify independent mortality predictors and compared prognostic discrimination using area under the receiver operating characteristic curve (AUROC). Results: Among 1,771 HCC patients, 698 (39.4%) had cirrhosis. Paradoxically, cirrhotic patients demonstrated lower mortality than non-cirrhotic patients (6.4% vs 8.9%). Among cirrhotic patients, Child-Pugh class stratified mortality risk: class A 2.3%, class B 6.4%, class C 14.8% (p < 0.001). Child-Pugh score showed superior discrimination compared to MELD (AUC 0.702 vs 0.644). Independent mortality predictors included age (OR 1.04, 95% CI 1.01-1.08), MELD score (OR 1.06, 95% CI 1.02-1.10), and vasopressor use (OR 8.54, 95% CI 3.45-21.14). Vasopressor requirement demonstrated the strongest mortality association, exceeding liver dysfunction severity. A combined liver-ICU prognostic score provided only marginal improvement over Child-Pugh alone (AUC 0.707 vs 0.702). Conclusions: In critically ill HCC patients, Child-Pugh score outperforms MELD for short-term mortality prediction, likely reflecting its capture of acute hepatic decompensation. Vasopressor requirement dominates prognostic importance, emphasizing that hemodynamic failure supersedes chronic liver disease severity in ICU settings.