The prognostic stratification using acute-on-chronic liver failure scoring system for predicting short-term mortality in patients with alcoholic hepatitis

Abstract

Background: To compare the existing various prognostic scoring models and newly proposed scores for acuteon-chronic liver failure (ACLF) and evaluate usefulness of stratification for the prediction of short-term morality in patients with alcoholic hepatitis (AH). Methods: A total of 705 clinical AH patients, enrolled in the APASL-ACLF Research Consortium (AARC) with 90-day follow-up, were analyzed. AARC-ACLF score, Maddrey discrimination function (DF) score, age, bilirubin, international normalized ratio and creatinine score (ABIC), Glasgow Alcoholic Hepatitis Score (GAHS), Child-Turcott-Pugh (CTP) score, model for endstage liver disease (MELD), and MELD-Sodium (Na) scores were used to compare the performance for predicting 30-day and 90-day mortality. AARC-ACLF scores were categorized into three grades (Gr I: 5-7; II: 8-10; and III: 11-15 points) and survival curves by the Kaplan-Meier method were created and compared using log-rank test. Results: Of 708 patients, 286 (40.4%) and 363 (51.3%) patients died within 30 days and 90 days, respectively. The area under receiver operating characteristics curve (AUC) of AARC-ACLF, DF, ABIC, GAHS, CTP, MELD, and MELD-Na was 0.752 (0.705-0.799), 0.630 (0.575-0.685), 0.658 (0.604-0.711), 0.577 (0.523-0.631), 0.641 (0.589-0.694), 0.705 (0.653-0.756), 0.703 (0.651- 0.755), respectively, for 30-day mortality. The AUC of various prognostic scores for the prediction of 90-day mortality is similar. The performance of AARC-ACLF was superior to that of DF, ABIC, GAHS, CTP, while comparable to that of MELD and MELD-Na in predicting short-term mortality. According to AARC-ACLF grades, short-term cumulative survivals was statistically different (30-day, 82.4, 70.4, and 35.3%, p <0.001; 90-day, 76.9, 56.0, and 26.4%, p <0.001). Conclusion: Compared to the previous AH prognostic scores, AARCACLF score and grades are simple and useful for predicting the short-term mortality in patients with AH. Further studies are needed to confirm these implications.