This study was a single-center, retrospective cohort study conducted at the Samsung Medical Center, Seoul, South Korea. We screened 886 patients who were admitted to our liver unit without malignancy, viral hepatitis, autoimmune hepatitis, or primary biliary cirrhosis between January 2014 and December 2018. Among them, we identified 81 severe AH patients who met the diagnostic criteria of definite or probable AH by the National Institute on Alcohol Abuse and Alcoholism [19]. In brief, definite AH is defined for patients meeting the clinical diagnosis criteria of AH with biopsy confirmation. Probable AH is defined for patients meeting the clinical diagnosis criteria of AH in the absence of potential confounding factors without biopsy confirmation. The criteria for the clinical diagnosis of AH are as follows: (1) onset of jaundice (serum total bilirubin >3.0 mg/dL) within the previous 8 weeks; (2) ongoing consumption of >40 g (female) or >60 g (males) of alcohol/day for 6 months or more with less than 60 days of abstinence before the onset of jaundice; (3) aspartate aminotransferase (AST) and alanine aminotransferase (ALT) <400 IU/L, AST >50 IU/L, AST/ALT ratio >1.5; (4) absence of other chronic liver diseases (hepatitis B, hepatitis C, autoimmune, or metabolic liver disease); and (5) absence of sepsis, shock, cocaine use, or recent use (within 30 days) of a drug with the potential for drug-induced liver injury. An MDF score ≥32 was used to define severe AH. The pre-MELD era was from January 2014 to June 2016 and the post-MELD era was from July, 2016 to December, 2018 in this study. In our institution, a physician or surgeon determines the necessity of LT for each patient, and when it is deemed necessary, a multidisciplinary team, which includes transplant surgeons, hepatologists, psychologists, cardiologists, infectious disease specialists, and social workers evaluates each patient. After the multidisciplinary evaluation, the respective physician or surgeon has a discussion with the patient and family members regarding registering on the LT waitlist. The transplant coordinator then lists the patient and registers them to the Korean Network for Organ Sharing. Living donor LT (LDLT) is explained in all situations, and evaluations are performed according to the living liver donor selection criteria if there is a willing donor within the family. In our institution, the selection criteria for living liver donors are as follows: (1) <65 years of age; (2) expected remnant liver volume greater than 30%; and (3) no evidence of chronic liver disease. There are no specific criteria for emergent LDLT in the setting of severe AH. When there is a willing liver donor who fulfills the donor selection criteria, transplant surgeons and hepatologists discuss with the patient and family members whether to proceed with emergent LDLT, wait for DDLT allocation, or wait for spontaneous recovery on a case-by-case basis. If there is no willing live liver donor or no liver donor who fulfills the donor selection criteria, the patient waits for DDLT allocation or spontaneous recovery. When a deceased donor is allocated to a patient, transplant surgeons and hepatologists discuss with the patient and family members whether to proceed with emergent DDLT, wait for spontaneous recovery, or decline the allocation due to high risk of transplant futility.

The primary outcome was overall survival before and after the MELD-based allocation system was implemented. The duration of survival was calculated from the date of initial hospitalization to death or the last date of follow-up. The secondary outcome was transplant-free survival (TFS). For TFS, the duration of survival was calculated from the date of initial hospitalization to death, last date of follow-up, or date of LT, whichever came first. The reference day was May 5, 2019.

The following variables were collected by reviewing the electronic medical records of each patient: age at hospitalization; sex; previous decompensation history; presence of infection; gastrointestinal bleeding; imaging, endoscopic, and laboratory results; medication use; and LT during follow-up. The presence of hepatomegaly, splenomegaly, and ascites were evaluated using computed tomography (CT) or ultrasonography findings. Splenomegaly was defined as the largest dimension >11 cm, and hepatomegaly was defined as a longitudinal axis >16 cm at the midclavicular line on CT or ultrasonography. Infections were defined as follows [20]: (1) spontaneous bacteremia: blood cultures without a source of infection; (2) spontaneous bacterial peritonitis: ascites fluid polymorphonuclear cell count >250/uL; (3) lower respiratory tract infection: new pulmonary infiltration in the presence of (i) at least one respiratory symptom (cough, sputum production, dyspnea, pleuritic pain) with (ii) at least one finding on auscultation (rales or crepitation) or one sign of infection (core body temperature >38°C or <36°C, shivering or leucocyte count >10,000/mm³ or <4,000/mm³) in the absence of antibiotics; (4) skin infection: fever with cellulitis; (5) urinary tract infection: urine white blood cell >15/high-power field with either positive urine Gram stain or culture in symptomatic patients; and (6) other source of infection (e.g., intra-abdominal infection, secondary peritonitis). Gastrointestinal bleeding was defined as evidence of hematemesis/melena/hematochezia, a drop in hemoglobin of 2 g/dL or more from baseline, or the need for a transfusion. Previous decompensation included jaundice, ascites, variceal hemorrhage, or hepatic encephalopathy prior to admission. The presence of varix was identified by reviewing endoscopic findings among patients who underwent upper endoscopy and by reviewing CT findings. The following laboratory variables at initial hospitalization were collected: white blood cell count, platelet count, prothrombin time, bilirubin, ALT, AST, gamma glutamyltransferase, blood urea nitrogen, and creatinine. All patients were assessed using the MELD score (3.78 ln [bilirubin]+11.20 ln [international normalized ratio]+9.57 ln [creatinine]+6.43) [21]. For those who received LT, the MELD score on the day of transplantation, waiting time, and alcohol relapse after LT were collected. For those who received corticosteroid treatments, the Lille score (3.19–0.101×age+0.147×albumin on day 0+0.0165×evolution in bilirubin level–0.206×renal insufficiency–0.0065×bilirubin on day 0–0.0096×prothrombin time was calculated. Steroid non-responders were defined as a Lille score >0.45 [22].

Baseline characteristics were described using frequency (percentage) and median (interquartile range). Continuous variables were analyzed using Student t-test or the Mann-Whitney U-test. Categorical variables were analyzed using the chi-square test or Fisher’s exact test. For the MELD score, patients were divided into two groups according to MELD score (<25 and ≥25). The cutoff point was chosen based on the area under the receiver operating characteristic analysis. The survival curves were plotted using the Kaplan-Meier method, and the difference between the curves was tested using the log-rank test. Multivariate Cox regression analysis was performed with all variables with a P-value <0.10 in univariate analysis to evaluate predictors of overall survival and TFS. Statistical significance was set at P<0.05.

The baseline characteristics of the patients are summarized in Table 1. There were no differences in baseline characteristics before and after the MELD-based allocation system was implemented. The proportion of patients receiving LT increased from 25% (10/40) in the pre-MELD era to 65% (27/41) in the post-MELD era (P<0.001). The proportion of patients receiving DDLT increased from 17% (7/40) to 51% (21/41) (P=0.001) in the pre- and post-MELD era, respectively. Although not statistically significant (P=0.30), the proportion of patients receiving LDLT increased from 7% (3/40) to 14% (6/41) in the pre- and post-MELD eras, respectively. When stratified according to MELD score, the DDLT rate did not change significantly (18% to 25%, pre- and post-MELD eras, P=0.53) for those with low MELD scores (<25), while the DDLT rate increased from 17% (pre-MELD era) to 68% (post-MELD era, P<0.001) for those with high MELD scores (≥25) (Fig. 1).

During the follow-up period of a median of 8.4 months (range, 0.1–64.0 months), mortality was observed in 30 patients (37%). Of the 81 patients, 22 (27%) used corticosteroids, while 59 (73%) did not. The proportion of patients who recovered without LT, received LT, and died without LT were 36%, 36%, and 27%, respectively, among the 22 patients who used steroids, and were 20%, 49%, and 30%, respectively, among the 59 patients who did not use steroids (Fig. 2). Overall survival improved in patients with severe AH managed after the MELD-based allocation system compared to before (80% vs. 50% at 12 months, P=0.005) (Fig. 3). In the univariate analysis, LT and the MELD-based allocation system era were factors associated with overall survival (Table 2). Age (per year), male sex, and hepatomegaly showed a marginal association with overall survival. In the multivariate analysis, age (hazard ratio [HR], 1.04; 95% confidence interval [CI], 1.00–1.08; P=0.028) and LT (HR, 0.16; 95%, CI, 0.06–0.52; P<0.001) were independent factors affecting overall survival (Table 2). Patients who received LT showed better overall survival than patients who did not receive LT (88% vs. 44% at 1 year, P<0.001) (Supplementary Fig. 1).

The TFS rates were 42.3% and 32.7% at 3 and 6 months, respectively. In the multivariable analysis, response to corticosteroids (compared to non-users; HR, 0.25; 95% CI, 0.09–0.71; P=0.009), corticosteroid non-response (compared to non-users; HR, 2.07; 95% CI, 1.00–4.30; P=0.050), and MELD score ≥25 at initial hospitalization (compared to <25; HR, 2.79; 95% CI, 1.60–4.88; P<0.001) were independently associated with TFS (Table 3). When stratified according to MELD score and steroid response, the 3-month TFS was 70% and 24% for MELD scores <25 and ≥25, respectively (P<0.001) (Supplementary Fig. 2A), and 38%, 88%, and 16% for steroid non-users, steroid responders, and steroid non-responders, respectively (P=0.005) (Supplementary Fig. 2B).

Among the 37 patients who received LT, 9 received LDLT, and 28 received DDLT. The median time from LT listing was not significantly different between the two groups (LDLT, 14 days vs. DDLT, 9 days; P=0.24). The DDLT group had higher MELD scores at transplantation than the LDLT group (40 vs. 29, P=0.035). After LT, alcohol relapse was identified in 5 of the 37 patients (13%). Six deaths occurred after LT. The cause of mortality was sepsis in four patients, graft rejection in one patient, and liver failure related to recidivism in one patient (Supplementary Table 1).