Potential conflict of interest: Nothing to report.
Data on liver transplantation for patients with alcoholic hepatitis are limited. Using the United Network for Organ Sharing database (2004-2010), adults undergoing liver transplantation for a listing diagnosis of alcoholic hepatitis were matched for age, gender, ethnicity, and model for endstage disease (MELD) score, donor risk index, and year of transplantation with three patients transplanted for a listing diagnosis of alcoholic cirrhosis. Study outcomes of graft and patient survival on follow-up were also analyzed for cohorts based on the diagnosis of the explant (46 alcoholic hepatitis and 138 alcoholic cirrhosis) and diagnosis at both listing as well as of the explant (11 alcoholic hepatitis and 33 alcoholic cirrhosis). Five-year graft and patient survival of alcoholic hepatitis and alcoholic cirrhosis patients were 75% and 73% (P = 0.97) and 80% and 78% (P = 0.90), respectively. Five-year graft and patient survival rates were also similar for cohorts based on diagnosis of the explant and diagnosis at listing as well as explant. Cox proportional regression analysis adjusting for other variables showed no impact of the etiology of liver disease (alcoholic hepatitis versus alcoholic cirrhosis) on the graft and patient survival. The causes of graft loss and patient mortality were similar in the two groups, and were not alcohol-related in any patient. Conclusion: Compared with alcoholic cirrhosis, patients with alcoholic hepatitis have similar posttransplantation graft and patient survival. Based on these preliminary findings, liver transplantation may be considered in a select group of patients with alcoholic hepatitis who fail to improve with medical therapy. Prospective studies are needed to assess the long-term outcome after liver transplantation in patients with alcoholic hepatitis. (HEPATOLOGY 2012)
Alcoholic hepatitis is a distinct clinical entity with a high mortality of about 40% at 6 months among those with severe disease. Despite treatment with corticosteroids and/or pentoxifylline, the mortality remains at about 15%-20%.1 Liver transplantation, a definitive treatment option for patients with alcoholic cirrhosis, requires at least 6 months of alcohol abstinence prior to transplantation.2 Currently, alcoholic hepatitis is considered a contraindication for liver transplantation (OLT) in most transplant centers in the U.S. due to the active drinking status of these patients and their inability to comply with the minimum 6 months of abstinence rule.2 Because of the high short-term mortality at 6 months, this exclusion criterion cannot be applied to patients with alcoholic hepatitis who do not respond to corticosteroids and/or pentoxifylline.3
There is a concern that patients who are actively drinking or are abstinent for less than 6 months prior to transplantation are likely to have a relapse of alcohol use after the transplantation. However, data on relapse of alcohol abuse with a minimum of 6 months of abstinence prior to liver transplantation are conflicting.4-6 Retrospective studies have shown that posttransplantation survival in patients with alcoholic cirrhosis is not affected by the presence of histological changes of alcoholic hepatitis in the explanted liver specimens.7, 8 Further, a recent prospective case-controlled study has provided encouraging data on the outcome after liver transplantation in alcoholic hepatitis patients who failed to respond to corticosteroids, challenging the 6 months rule of alcohol abstinence prior to liver transplantation. However, this study from France had a relatively short posttransplant follow-up of only 2 years.9
Currently, selected patients with alcoholic hepatitis are being randomly transplanted in the U.S. In the present study, we used the United Network for Organ Sharing (UNOS) database to assess the 5-year postliver transplantation graft and patient survival of patients with alcoholic hepatitis and compared the results with patients transplanted for alcoholic cirrhosis.
DRI, donor risk index; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; TIPS, transjugular intrahepatic Porto systemic shunt; UNOS, United Network for Organ Sharing.
Materials and Methods
Source of Data.
We used the Organ Procurement and Transplantation Network dataset for the years 2004-2010 for the present analysis. Developed and maintained since October 1999 by the United Network for Organ sharing (UNOS), the data pertains to patient waiting list, organ matching, and transplantation since 1986. The dataset has information on transplant candidate registration, transplant recipient information, donor information, wait list data, and follow-up information on recipients after the transplant.
Establishing the Study Population.
Patients transplanted for a primary listing diagnosis of alcoholic hepatitis (UNOS code 4217) formed the study population. Because only one patient received liver transplant for the diagnosis of alcoholic hepatitis prior to 2004, for the present analysis we included the dataset starting from 2004 to 2010. Alcoholic liver disease being primarily a disease of adults, the data were designed to include liver transplants performed in people age 18 or above. The diagnosis at explant of 55 cases with a listing diagnosis of alcoholic hepatitis was alcoholic hepatitis in 11, alcoholic cirrhosis in 33, and other diagnoses in 11 cases. With the UNOS database limited by the availability of information on liver biopsy prior to transplantation, we also analyzed another cohort of 46 patients based on primary diagnosis of alcoholic hepatitis on the explant. The listing diagnosis of these 46 cases was alcoholic hepatitis in 11 and alcoholic cirrhosis in 35 cases. We then analyzed a third cohort of 11 cases with the primary listing as well as explant diagnosis of alcoholic hepatitis.
alcoholic hepatitis is a clinical entity with an acute presentation and histological changes of alcoholic hepatitis can persist on explant among patients listed and transplanted for alcoholic cirrhosis.7, 8 We therefore chose to describe our main results from the dataset based on a listing diagnosis of alcoholic hepatitis. Further, choosing the cohort with the listing diagnosis as the main cohort for analysis allowed us to compare patients who did and did not undergo transplantation.
Matching Procedure and Methodology.
Each patient with alcoholic hepatitis was matched to three patients with alcoholic cirrhosis (UNOS code 4215). For each of three cohorts with diagnosis of alcoholic hepatitis at listing in 55 cases (cohort 1), at explant in 46 cases (cohort 2), and at listing as well at explant in 11 cases (cohort 3), we matched with 165, 138, and 33 cases, respectively, with a diagnosis of alcoholic cirrhosis at listing, explant, and listing as well at explant.
Matching was done in sequential steps. In the first step, patients were matched by gender, ethnicity, year of transplantation, age (±5 years), donor risk index (DRI ±0.5), and model for endstage disease (MELD) score (±5 points). Donor risk index was calculated using donor-related factors: age, gender, race, height, cause of donor death, cold ischemia time, partial/split transplant or not, and organ location (local, regional, or national).10 The MELD score has been shown to be fairly accurate for determining the severity and in predicting mortality in patients with alcoholic hepatitis.11, 12 However, being an acute condition and often presenting as acute liver failure, patients with alcoholic hepatitis may have a higher MELD score compared with patients undergoing transplantation for alcoholic cirrhosis. Therefore, apart from matching for other variables, we also decided to match our cases and controls for the MELD score. From the group of patients with alcoholic cirrhosis with the value of continuous variables (age, DRI, MELD) in the matched range, we selected three alcoholic cirrhosis patients with the closest value on these variables. By this method we were able to successfully match 77.6% of alcoholic hepatitis patients with alcoholic cirrhosis patients. Ethnicity was dropped in the further matches. Based on the same criteria as the first step but without ethnicity, we were able to successfully match another 6% cases. The remaining 16.4% of alcoholic cirrhosis patients were matched with alcoholic hepatitis patients in the third and final step using the following less stringent criteria: DRI (±1) and year of transplant (±1). As all the cases (transplanted for alcoholic hepatitis) received deceased donor liver transplants (DDLT), controls (transplanted for alcoholic cirrhosis) were selected only for DDLT.
The diagnosis of alcoholic hepatitis matched at listing and at explant in only 20% (11 of 55) of the cohort with a listing diagnosis and about 25% (11 of 46) of the cohort with diagnosis at explant. In contrast, diagnosis of alcoholic cirrhosis was not contaminated with alcoholic hepatitis, as only about 1% (2 of 165 from the cohort with diagnosis at listing and 2 of 138 from the cohort with diagnosis at explant) of cases with a diagnosis of alcoholic cirrhosis having alcoholic hepatitis.
Our main outcome measures were graft function and patient survival. Both the current status and time to outcome were included as outcome measures. Patients with a functioning graft or those who were alive at their last follow-up visit and/or at the time of data collection were censored in the UNOS dataset. The causes of death or graft failure were obtained from the dataset for patients achieving these outcomes.
The cases (patients transplanted for alcoholic hepatitis) and controls (patients transplanted for alcoholic cirrhosis) were compared for baseline recipient and donor characteristics. Chi-square and Student's t tests were used for comparing categorical and continuous variables, respectively. Kaplan-Meier survival curves were compared with estimate graft and patient outcome on follow-up. Cox proportional hazard models were built for the time to graft loss and time to survival, respectively, to determine the independent effect of the etiology of liver disease (alcoholic hepatitis or alcoholic cirrhosis) on the outcome measures. Variables that were significantly different at baseline comparison as well as those clinically relevant even if similar at baseline were entered into the model. The results were expressed as hazard ratio (HR) with 95% confidence interval (CI). We compared the graft and patient outcomes from the main analysis with the other two cohorts in order to reduce any coding error based on the listing diagnosis alone. The causes of graft loss and patient death were analyzed and compared between cases and controls. Analyses were performed using the Statistical Analysis Software (SAS Institute, Cary, NC).
Of a total of 130 patients with alcoholic hepatitis listed for liver transplantation between 2004 and 2010, 59 were transplanted (Fig. 1). The MELD score (mean ± SD) of these 59 cases was higher compared with 71 patients who did not undergo transplantation (23 ± 9 versus 17 ± 11; P = 0.019). The two groups were otherwise similar for patient demographics (age, gender, body mass index), medical condition at orthotopic liver transplantation (OLT) (admitted to the Intensive Care Unit [ICU], on ventilator, on dialysis), proportion with hepatocellular carcinoma (HCC), and those with transjugular intrahepatic Porto systemic shunt (TIPS). Of the 59 cases who were transplanted, 55 cases had an adequate follow-up and were further analyzed (Fig. 1). The yearly distribution of these cases was: 7 in 2004, 18 in 2005, 6 in 2006, 10 in 2007, 8 in 2008, 2 in 2009, and 4 in 2010. Similar numbers for the matched alcoholic cirrhosis patients were 22, 52, 19, 27, 25, 12, and 8, respectively.
The baseline characteristics of alcoholic hepatitis patients (n = 55) and patients with alcoholic cirrhosis (n = 165) are shown in Table 1. Because the two groups were well matched prior to analysis, most recipient and donor characteristics were similar. However, some differences were observed between the groups. For example, patients with alcoholic cirrhosis were more likely to have undergone TIPS placement (12% versus 6%; P = 0.002) prior to transplant, and were more likely to have incidental HCC in the explant (3% versus 0%; P = 0.027). None of the patients with a primary listing diagnosis as well as with explant diagnosis of alcoholic hepatitis had received TIPS prior to transplantation. A higher proportion of patients with alcoholic hepatitis were infected with hepatitis C virus (HCV) infection (26% versus 7%; P = 0.0002). However, when patients with a similar listing and explant diagnosis were compared, this difference was not significant (1 of 11 alcoholic hepatitis patients versus 3 of 33 alcoholic cirrhosis patients). The mean pretransplant waiting period for patients with alcoholic hepatitis tended to be shorter compared with patients with alcoholic cirrhosis (67 ± 86 versus 145 ± 340 days; P = 0.09).
Table 1. Comparison of Patients Transplanted for Alcoholic Hepatitis (AH) and Patients Transplanted for Alcoholic Cirrhosis (AC) for Baseline Recipient and Donor Characteristics
BMI: body mass index; W: white; TIPS: transjugular intrahepatic Porto systemic shunt; SLK TX: simultaneous liver kidney transplant; HCC: hepatocellular carcinoma; CMV: cytomegalovirus; DRI: donor risk index.
On ventilator (%)
On dialysis last 1 week
Region (NE: SW: S: W)
Clinical infection donor
Warm ischemia time
Previous abdominal surgery
Incidental HCC at TX
Days waiting for LT
A total of 14 (25%) alcoholic hepatitis and 44 (27%) alcoholic cirrhosis patients lost their grafts on follow-up. The graft survival of alcoholic hepatitis and alcoholic cirrhosis patients was similar at 1 year (87% versus 84%; Log rank P = 0.58); 3 years (82% versus 77%; P = 0.47); and 5 years (75% versus 73%; P = 0.97) (Fig. 2). A total of 11 (20%) alcoholic hepatitis and 36 (22%) alcoholic cirrhosis patients died. The patient survival at 1, 3, and 5 years were similar in the two groups (93% versus 88%; P = 0.33; 87% versus 81%; P = 0.33; and 80% versus 78%; P = 0.91) (Fig. 3). As the HCV infection rate was significantly higher among patients with a listing diagnosis of alcoholic hepatitis compared with alcoholic cirrhosis (14 of 55 versus 12 of 165; P = 0.0002), we analyzed graft and patient survival at 5 years after excluding HCV-positive cases. In this subgroup also, graft and patient survival were similar comparing patients listed for alcoholic hepatitis as well as those listed for alcoholic cirrhosis (80% versus 75%; P = 0.76 and 82% versus 84%; P = 0.93, respectively).
Cox proportional hazard regression analysis models showed no effect of the indication for transplantation (alcoholic hepatitis versus alcoholic cirrhosis) on the graft or the patient survival (Table 2). Variables that were entered into the model were: geographical region and year of transplantation; recipient characteristics (age, gender, ethnicity, MELD score at listing, abdominal surgery prior to transplant, being on ventilator at the time of transplantation, TIPS prior to transplant, incidental HCC in explant, simultaneous liver kidney transplant, and concomitant HCV infection); and donor characteristics (donor risk index, nonheart-beating donor, and donor infection). Similar results were obtained when the graft and patient survival were analyzed for the datasets based on the primary diagnosis of the explant (46 cases and 138 controls) and based on the primary diagnosis at listing as well as of the explant (11 cases and 33 controls) (Table 3).
Table 2. Cox Proportional Hazard Regression Analysis to Assess Predictors of Graft and Patient Survival
Cohort 1: diagnosis at explant – AH (n=11), AC (n=33), other (n=11) and Cohort 2: diagnosis at listing – AH (n=11), AC (n=35).
Cohort 1: diagnosis at explant – AH (n=2), AC (n=128), other (n=35) and Cohort 2: diagnosis at listing – AH (n=2), AC (n=136).
220 [55 AH, 165 AC]
Diagnosis at listing
184 [46 AH, 138 AC]
Diagnosis at explant
44 [11 AH, 33 AC]
Diagnosis at listing as well as at explant
Independent predictors of both graft and patient survival were: patient on ventilator prior to transplantation and the number of days waiting for transplantation (Table 2). A total of 3 out of 55 (6%) alcoholic hepatitis patients and 6 of 165 (4%) alcoholic cirrhosis patients were on ventilator at the time of transplantation. A patient on ventilator at the time of transplantation was 4 times more likely to lose the graft, and about 7 times more likely to die on follow-up compared with patients not on ventilator. For every 30 days of waiting for liver transplantation, there was a 4.3% increased risk of losing the graft and 4.6% increased risk of patient mortality.
The causes of graft loss were available in 5 of 14 (36%) alcoholic hepatitis patients and 26 of 44 (59%) of alcoholic cirrhosis patients (Table 4). These were similar in the two groups, with the most common causes being primary graft failure and infection. The causes of patient death were available in all the 47 patients who died on follow-up. The common causes were cardiovascular or metabolic problems, infection, and malignancy. The causes of patient mortality were similar in the two groups except for a trend for a higher proportion of deaths due to infection in patients transplanted for alcoholic cirrhosis compared with patients transplanted for alcoholic hepatitis (22% versus 0%; P = 0.09). Malignancy was the cause of death in 6 of 36 (17%) patients with alcoholic cirrhosis and 3 of 11 (27%) patients with alcoholic hepatitis (P = 0.42). Relapse of alcohol use was not listed in any of the patients transplanted for alcoholic hepatitis who lost their graft (n = 14) or died during follow-up (n = 11). Noncompliance of immune suppression treatments on follow-up, a crude surrogate for alcohol relapse, was not encountered in any of the patients in the three analyses. In fact, noncompliance in the whole UNOS dataset was reported in less than 1% cases.
Table 4. Causes of Graft Loss and Patient Mortality Among Patients with Alcoholic Hepatitis and Alcoholic Cirrhosis
Primary graft failure
De novo hepatitis
In summary, our study shows that patients transplanted for alcoholic hepatitis have similar graft and patient survival compared with patients undergoing liver transplantation for alcoholic cirrhosis. The causes of graft loss and patient death were similar in the two groups. Primary graft failure and infection were the most common causes of graft loss, whereas metabolic complications, infection, and malignancy were the most frequent causes of patient mortality.
Patients with alcoholic hepatitis have a poor short-term prognosis. Specific treatment with corticosteroids and/or pentoxifylline provides only 50% survival benefit.1 Options for managing patients who fail to respond to corticosteroids and/or pentoxifylline are limited.13, 14 Liver transplantation is a definitive treatment option for patients with endstage liver disease due to alcoholic cirrhosis. Posttransplant graft and patient outcome for patients with alcoholic cirrhosis has been shown to be better compared with patients transplanted for HCV-related cirrhosis and similar to other causes of endstage liver disease and cirrhosis.14 Most liver transplant centers in the U.S. consider alcoholic hepatitis a contraindication for liver transplantation.15 This is reflected by the fact that only 130 patients with alcoholic hepatitis were listed for transplantation in the U.S. between 2004 and 2010 compared with 8,837 patients with alcoholic cirrhosis. The reasons for not considering alcoholic hepatitis for liver transplantation are manifold, including: (1) the ethical issue of providing an organ to individuals who are actively drinking and whose disease is thought to be self-inflicted16; (2) the rule of a minimum 6 months of abstinence for receiving liver transplant, which is difficult to apply to patients with a 20% mortality at 6 months despite medical treatment15, 17; and (3) the risk of alcohol relapse after liver transplantation.18 However, studies on the association between the duration of pretransplant abstinence and the risk of alcohol relapse have shown conflicting data.19, 20
The present study shows that graft and patient survival of patients transplanted for alcoholic hepatitis are similar to that of patients transplanted for alcoholic cirrhosis. The outcomes were similar irrespective of which analyses were performed: based on the primary diagnosis at the time of listing, diagnosis of the explant, and diagnosis at both listing and explant. However, the effect size on the Cox regression analysis was in a different direction, with a wide confidence interval when the analysis was performed based on the primary diagnosis at listing as well at explant. The small sample size of 44 (11 cases and 33 controls), with a lack of power, probably explains this difference. Previous studies have shown that histological findings of alcoholic hepatitis in the explants of patients undergoing liver transplantation for alcoholic cirrhosis do not influence graft or patient outcome.7, 8 A recent case-controlled prospective study from France showed that liver transplantation was beneficial in alcoholic hepatitis patients with nonresponse to a 1-week course of corticosteroids. In this study, patients who were transplanted had a 2-year survival of 71% ± 9% compared with 23% ± 8% for those who did not receive a transplant.9 In our study the survival rate was 91% at 2 years. Differences in study design, geographic location (Europe versus U.S.), and demographic characteristics of the study population with younger age and more females in the French study compared with our study (47 versus 52 years and 42% versus 24%, respectively) could explain the disparity in the posttransplantation outcome.
Patients undergoing transplantation (n = 59) were no different from those not undergoing transplantation (n = 71). A total of 24 patients died before liver transplantation could be done and 12 other patients became too sick to be transplanted (Fig. 1). In contrast, out of the 8,837 patients with alcoholic cirrhosis listed for liver transplantation between 2004 and 2010, 1,026 died and 524 became too sick to be transplanted while on the waiting list. The wait list morbidity and mortality was significantly higher in patients with alcoholic hepatitis compared with patients with alcoholic cirrhosis (27% [36 of 130] versus 17% [1,530 of 8,837]; P = 0.0034). The wait list mortality in patients with alcoholic hepatitis listed for liver transplantation in the French study was only 8% (2 of 26 listed patients). The shorter median waiting time of only 9 days for receiving a transplant in the French study compared with about 20 days in the present analysis explains the difference in the wait list morbidity and mortality. In order to avoid wait list mortality, these acutely sick patients once listed for liver transplantation should receive a priority for liver transplantation similar to that with acute liver failure of any other etiology.
Cardiovascular or metabolic problems and malignancy were common causes of patient mortality on follow-up, contributing to 23% and 20% deaths, respectively. These findings are similar to the common causes of mortality in patients transplanted for alcoholic cirrhosis.21-23
The strength of our study is the inclusion of the large database from the UNOS, which represents the entire transplant population in the U.S. Patient demographics of the study population of patients with alcoholic hepatitis in this analysis is similar compared with patients with alcoholic hepatitis reported from single centers within the U.S. as well as the National Inpatient database, which is a fair representation of the U.S. population.24, 25 Patients with alcoholic hepatitis are acutely sick with a high MELD score (our mean MELD score was 26) and high mortality of 30%-40% at 6 months.1 Because the average MELD score of patients undergoing liver transplantation in the U.S. for alcoholic cirrhosis is around 21 and 22, we matched patients receiving liver transplants for alcoholic cirrhosis with patients receiving liver transplant for alcoholic hepatitis for pretransplant MELD score in addition to matching for other variables.
The present study suffered from some limitations. For example, the lack of availability of criteria for listing these 130 patients among thousands of cases of alcoholic hepatitis seen and managed in the U.S. annually. Being a database study, lack of information on liver biopsy details, use of and response to corticosteroids, and nutritional status of patients limit generalizability of our study results. It is likely that these cases were listed after a careful evaluation to select cases unlikely to relapse to drinking alcohol after the OLT, which is the standard of care for patients undergoing OLT for alcoholic cirrhosis. Further, the dataset lacks information for alcohol relapse on follow-up. We presume this was very low, as the cause of graft loss or patient death was not related to alcohol use in any patient among patients where this cause was known.
In summary, our study shows that the graft and patient outcome after liver transplantation for patients with alcoholic hepatitis are similar to patients transplanted for alcoholic cirrhosis. In view of the nonavailability of better options for treatment of corticosteroid nonresponders, the lack of evidence linking pretransplant abstinence and risk of alcohol relapse, and encouraging data on liver transplantation in patients with alcoholic hepatitis, we propose that liver transplantation may be considered an option in select patients with alcoholic hepatitis. Further, well-designed prospective studies with longer follow-up are needed with the aim of assessing the long-term efficacy of liver transplantation in patients with alcoholic hepatitis and to determine the pretransplant predictors of a patient remaining alcohol abstinent after liver transplantation.