NEW PROGNOSTIC MODEL IN HOSPITALIZED PATIENTS WITH NON-SEVERE ALCOHOLIC LIVER DISEASE

Abstract

Background: Several prognostic models have been developed for patients with severe alcoholic liver disease, but few studies have yet investigated prognosis of patients with non-severe alcoholic liver disease (NSALD). In this study, we aimed to develop a new prognostic model for patients with NSALD. Methods: We extracted 364 patients with NSALD, defined as lower than 32 of Maddrey’s discriminant function score from the retrospective KACLiF cohort for the derivation of a new prognostic model (training set), and validated it in 526 patients from the prospective KACLiF cohort (validation set); These cohorts consisted of patients who had chronic liver disease and were hospitalized due to acute decompensations. Prognostic factors for death and liver transplantation were analyzed using Cox proportional hazard model. Final prognostic model was constructed using backward stepwise selection. Results: Twenty-four and 36 patients died or received transplantation within 6 months in training and validation sets. In the training set, the highest area under the curve (AUC) of conventional prognostic models (e.g. MELD-Na score) was 0.744, 0.741, 0.688 for 28-, 90-, 180-days mortality. History of previous decompensation, high neutrophil proportion, presence of large ascites and overt hepatic encephalopathy, low albumin level, high sodium level, and use of vasopressor were significantly associated with the mortality of patient with NSALD. The new model consisted of past decompensation, neutrophil proportion over 70%, large ascites and use of vasopressor. It showed high predictivity for mortality in the training set as well as in the validation set (0.823 and 0.795, 0.816 and 0.784, and 0.805 and 0.755 for 28-, 90-, 180-days mortality in training and validation sets, respectively). In addition, the 6-month transplantation free survivals were 97.9%, 91.7%, and 57.2% of patients with new model score 0-1, 2-3, and 4-6 in the training set (p<0.001); 96.4%, 88.9%, and 58.7% in the validation set (p<0.001). Conclusion: There is the high-risk group even in patients with NSALD, and a new model will help identify them. For these patients, steroid administration can be considered with close monitoring for the need for liver transplantation. Thus, our findings warrant further studies of new treatment strategies for them.