The authors report no commercial relationships that might pose a conflict of interest in connection with this article.
Address reprint requests to Ivo W. Graziadei, M.D., Department of Internal Medicine II, Innsbruck Medical University, Anichstrasse 35, A-6020 Innsbruck, Austria. Telephone: +43-512 504 23401; FAX: +43-512 504 24052; E-mail: email@example.com
Asian Pacific Association for the Study of the Liver
international normalized ratio
Model for End-Stage Liver Disease
Model for End-Stage Liver Disease–sodium
spontaneous bacterial peritonitis
Acute-on-chronic liver failure (ACLF) is an increasingly recognized complication of chronic liver disease characterized by an acute deterioration of liver function mainly in patients with compensated liver cirrhosis. ACLF is initiated by an acute precipitating factor and leads to the deterioration of liver function and other organ systems. Because of rapidly evolving multiorgan failure, ACLF is associated with high short-term mortality rates of 50% to 90%.[1-4] In contrast to chronic hepatic decompensation, which occurs in patients with end-stage cirrhosis as a result of the irreversible progression of the underlying disease, ACLF is potentially reversible if the precipitating event can be controlled. Patients who survive the acute exacerbation leading to ACLF have outcomes similar to those of patients with chronic hepatic decompensation. However, in clinical practice, the identification of the precipitating event is difficult, and exact discrimination between chronic hepatic decompensation and ACLF remains challenging.
ACLF is a poorly defined clinical entity mainly because of the considerable heterogeneity in the mode of presentation. The lack of a uniform definition of ACLF was first addressed in a consensus meeting of the Asian Pacific Association for the Study of the Liver (APASL) in 2009.6 The main criteria for diagnosing ACLF according to the APASL consensus are acute deterioration in a patient with previously well-compensated chronic liver disease manifesting as jaundice (serum bilirubin level > 5 mg/dL) and coagulopathy [international normalized ratio (INR) > 1.5] complicated by ascites and/or encephalopathy within 4 weeks caused by a precipitating event.
Patients with ACLF who do not improve with supportive measures are potential candidates for liver transplantation (LT). However, these patients are at high risk for acquiring bacterial and fungal infections or developing sepsis, which may preclude them from LT. Because clinical deterioration can evolve rapidly, the time period for evaluating patients and assessing them for LT is short, and this suggests an increased risk of wait-list mortality. In contrast to patients with acute liver failure, there is no possibility of putting ACLF patients on a high-urgency list (which implies lifesaving LT within a few days).
Although LT for chronic hepatic decompensation is well established, data concerning LT for patients with ACLF are rare. Results from 1 US study and 3 Asian studies suggest that LT for ACLF is safe and that survival rates are similar to those for other indications.[7-10] However, the etiologies of ACLF and the types of LT differ between Asian and Western countries. Although reactivation of hepatitis B virus is a major cause of ACLF in the Asia-Pacific region, alcoholic hepatitis is a common cause of acute deterioration in Western areas.[12, 13] Living donor LT is preferred in East Asian countries, whereas deceased donor grafts are used in the vast majority of transplants in European and American centers.
So far, almost no data have been available regarding the feasibility and efficacy of LT in patients with ACLF in the Western world based on an intention-to-treat analysis. Therefore, the aims of this study were to describe the natural course of ACLF in a large single-center cohort, to investigate the feasibility of LT for these patients, and to determine the long-term postoperative outcomes.
PATIENTS AND METHODS
This was a single-center study conducted at the Liver Unit of Innsbruck Medical University (Innsbruck, Austria). All patients who were treated between July 2002 and December 2010 were registered in a database collecting biochemical and clinical parameters. To identify eligible patients, we reviewed medical records retrospectively. The study was approved by the local ethics committee.
The diagnosis of ACLF was made in accordance with the consensus statement published by the APASL in 2009.6 ACLF was diagnosed if a patient with previously known or unknown compensated chronic liver disease developed liver failure, which was defined as jaundice (serum bilirubin level > 5 mg/dL) and coagulopathy (INR > 1.5) complicated by ascites and/or encephalopathy within 4 weeks after an acute hepatic insult. The following events were considered as possible precipitating events: nonhepatotropic infections/sepsis, variceal or nonvariceal bleeding, surgical interventions, hepatotoxic drugs, acute alcoholic steatohepatitis, and hepatitis B virus flare.
All patients received the standard-of-care treatment, including terlipressin and albumin for patients with hepatorenal syndrome, renal replacement therapy for patients with oliguria or rapidly deteriorating renal function, and mechanical ventilation in case of respiratory failure or prophylactically for comatose patients (to prevent aspiration). Blood, urinary, and oral smear cultures were regularly taken from all patients for microbiological surveillance. Paracentesis was performed in patients with ascites to exclude spontaneous bacterial peritonitis (SBP). Suspected respiratory infections were confirmed by chest X-rays and/or computed tomography scans. SBP was diagnosed if more than 250 × 106 polymorphic cells per liter of ascites were found and/or if cultures were positive. Sepsis was defined according to the American College of Chest Physicians/Society of Critical Care Medicine consensus conference, and multiorgan failure was defined as the development of 1 or more organ failures apart from the primary liver failure. Prophylactic antibiotics (mainly ampicillin/sulbactam or ciprofloxacin) were given to all patients with elevated infectious parameters and with tense ascites in order to prevent SBP. Antibiotic therapies were adopted according to the sensitivity profile of positive cultures.
LT was principally considered for all ACLF patients according to generally accepted guidelines. The transplant evaluation comprised chest/abdominal computed tomography scans, a cardiopulmonary assessment, and the exclusion of extrahepatic malignancies. All patients with alcoholic liver disease underwent an extensive assessment by a psychiatrist experienced in the management of patients with addictive behavior. A 6-month abstinence period was not an absolute prerequisite for listing; however, patients were listed only if their social background was supportive and coexisting psychiatric diseases were absent.
Patients were removed from the waiting list in cases of marked clinical improvement of liver cirrhosis, recurrent alcoholism, and progressive hepatocellular carcinoma beyond extended criteria. Although patients with uncontrolled infections/sepsis remained on the waiting list, they were put on hold until the infectious complications were ameliorated.
The posttransplant outcomes of ACLF patients were compared with the outcomes of a cohort of patients with chronic liver disease who had undergone LT during the same period for indications other than ACLF.
Data are shown as medians and ranges. Patients were followed from the date of admission to our unit until death, or they were censored on the last day on which they were documented to be alive. Transplant-free survival (transplant patients were censored on the date of LT) and overall survival were estimated with the Kaplan-Meier method; differences between subgroups were compared with the log-rank test. To identify predictors of mortality, a multivariate Cox proportional hazards model was performed. Variables with statistical significance in a univariate analysis were entered into a multivariate analysis. A P value ≤ 0.05 was considered statistically significant. All statistical analyses were performed with PASW Statistics 18.0 (SPSS, Inc., Chicago, IL).
Patient Characteristics, Precipitating Events, and Course of the Disease
According to APASL, 144 of 238 patients referred to our center fulfilled the criteria for ACLF. Patients' characteristics are shown in Table 1. A potential precipitating event inducing ACLF was found in 90 patients (62.5%): an apparent infection (n = 40); (non)variceal bleeding (n = 18); transjugular intrahepatic portosystemic shunt placement, radiofrequency ablation, or postoperative decompensation (n = 13); acute alcoholic steatohepatitis (n = 10); hepatitis B flare (n = 2); or another event (including trauma and drug-induced liver failure; n = 7). In 54 patients, no definite precipitating event could be identified. Median Model for End-Stage Liver Disease (MELD) score was higher in patients with a precipitating event as compared to those without (29 versus 27, P = 0.003). Other patient characteristics did not differ between these groups.
Table 1. Demographic and Clinical Characteristics of the ACLF Cohort (n = 144)
The data are presented as medians and ranges.#Includes Wilson's disease, Alpha-1 antitrypsin deficiency and HFE-asscoiated hereditary hemochromatosis.
The prognosis of our ACLF patients was extremely poor: the median overall survival and transplant-free survival times were 54 days [95% confidence interval (CI) = 40-68 days] and 48 days (95% CI = 38-58 days), respectively. Only 10 patients survived without LT with a median follow-up of 1.5 years. Multiorgan failure with or without sepsis was the most common cause of death (90%). Other reasons were hemorrhagic shock due to gastrointestinal bleeding (7%), intracranial hemorrhage (2.5%), and cerebral edema (0.5%). Neither mortality nor the severity of ACLF was related to the presence or absence of a precipitating event.
The poor prognosis of patients with ACLF was mainly caused by infectious complications and sepsis, which led to the acute deterioration of other organ systems. Seventy-six patients developed hepatorenal syndrome (53%), and renal replacement therapy (hemodialysis and/or hemofiltration) was required in 71 cases (49%). Sixty-nine patients (48%) developed pneumonia, and SBP was diagnosed in 23 patients (16%). Positive blood culture results were found for 70 patients (49%), with coagulase-negative staphylococci, Enterococcus faecium, and Candida species being the most common pathogens. Fifty-three patients (37%) developed sepsis. Mechanical ventilation due to respiratory insufficiency was necessary in 57 patients (40%).
LT Evaluation and Listing
Sixty-eight of 144 patients (47%) were evaluated for LT, and 26 patients (18%) were already on the wait list at the time of decompensation (Fig. 1). The most common reasons for excluding patients from the LT assessment were active alcohol/drug abuse (51%), age and comorbidities (16%), and uncontrolled infection/sepsis (12%).
Seventy-one patients (49% of the overall cohort) were listed for LT, and they included the 26 patients already on the wait list. The reasons for not listing patients were psychiatric contraindications (n = 6), uncontrolled sepsis (n = 6), death during the LT evaluation (n = 5), cardiopulmonary contraindications (n = 2), improved liver function (n = 2), patient refusal (n = 1), and a neurological contraindication (n = 1). Two patients were withdrawn from the wait list because of psychiatric complications.
Thirty-three patients (23% of the overall cohort) finally underwent deceased donor LT. The majority of the patients (n = 28) underwent transplantation during their first hospitalization after a median waiting time of 24 days (range = 5-115 days). The waiting time for patients who were already on the wait list was calculated from the date of decompensation. Five patients partially recompensated and were followed on an outpatient basis for a short period of time; they underwent LT between days 130 and 320 on the wait list. Two patients underwent combined kidney-liver transplantation. The wait-list mortality rate was 54%; multiorgan failure with or without sepsis (87%) and hemorrhagic shock (9%) were the main causes of death among the listed patients. The probability of death was 10% after 1 week, 37% after 1 month, and 52% after 3 months on the wait list (Fig. 2).
Outcomes After LT
Twenty-eight of the 33 transplant patients (85%) were alive at the end of the study period after a mean follow-up period of 29 months (range = 2-85 months) after LT. Two patients died because of sepsis, 1 died because of hepatitis C virus recurrence, 1 died because of secondary biliary cirrhosis, and 1 died because of acute graft-versus-host disease 36 to 440 days after LT.
In addition, we compared the posttransplant outcomes of ACLF patients with the outcomes of a cohort of 356 patients who had undergone LT at our center during the same study period for indications other than ACLF (the non-ACLF group) after a median waiting time of 179 days. The main characteristics of both groups are depicted in Table 2.
Table 2. Characteristics of Patients Undergoing LT for ACLF and Patients Undergoing LT for Other Indications (Non-ACLF Group)
Non-ACLF Group (n = 356)
ACLF Group (n = 33)
The data are presented as medians and ranges.
At admission for the ACLF group and before LT for the non-ACLF group.
The overall mortality rates did not differ between the 2 groups (15% for the ACLF cohort and 16% for the non-ACLF group). The 3-month, 1-year, and 5-year survival rates were 94%, 87%, and 82%, respectively, in the ACLF cohort and 98%, 93%, and 82%, respectively, in the non-ACLF cohort (Fig. 3); the differences were not statistically significant. The median duration of hospitalization after LT was not significantly different between the 2 cohorts (28 versus 25 days, P = 0.12), but ACLF patients had to stay significantly longer in the intensive care unit (median: 7 versus 6 days, P = 0.02).
Prognostic Factors for Patients With ACLF
To identify predictors of overall survival for patients with ACLF, a Cox proportional hazards model was calculated. Among the various parameters associated with overall survival in the univariate analysis, sepsis [hazard ratio (HR) = 2.70, P < 0.001], mechanical ventilation (HR = 2.15, P = 0.006), INR (HR = 1.68, P < 0.001), and serum sodium (HR = 1.04, P = 0.02) proved to be independent negative predictors of survival, whereas LT was the only strong positive predictor (HR = 0.06, P < 0.001; Table 3).
Table 3. Univariate and Multivariate Cox Regression of Possible Predictive Factors for Overall Survival in the ACLF Cohort (n = 144)
Variables associated signficantly with survival are highlited in bold print.
Renal replacement therapy
Cystatin C (mg/L)
Serum sodium (mmol/L)
C-reactive protein (mg/dL)
The predictors of transplant-free survival were almost the same as those of overall survival. Age (HR = 1.03, P = 0.008) was found to be an additional predictor of transplant-free survival, whereas the serum sodium concentration did not correlate in the multivariate analysis (data not shown).
Patients who successfully underwent LT (n = 33) had better renal function, lower MELD scores, and lower serum C-reactive protein levels at admission. In addition, patients who developed infectious complications (particularly pneumonia and/or sepsis) and patients with renal replacement therapy or mechanical ventilation were less likely to undergo LT (Table 4). No differences between transplant and nontransplant patients were seen with respect to underlying liver disease, causes of decompensation, hepatocellular carcinoma, sex, age, ascites, SBP, hepatorenal syndrome, bilirubin, serum sodium, INR, or albumin.
Table 4. Variables Differing Significantly Between ACLF Patients Undergoing LT and ACLF Patients Not Undergoing LT
ACLF is an increasingly recognized clinical entity; however, an evidence-based definition has not yet been proposed.[1, 3, 4] Despite advances in the treatment of advanced liver diseases, ACLF patients have high short-term mortality rates of 50% to 90%. So far, the possibility of LT has rarely been addressed for these patients; this is particularly true for their eligibility for LT and long-term postoperative outcomes. Therefore, the aims of this study were to evaluate the feasibility and efficacy of LT in patients with ACLF on an intention-to-treat basis and to analyze predictive factors for patient survival.
Our study shows that deceased donor LT in patients with ACLF leads to postoperative outcomes similar to those for patients undergoing LT with other indications. Despite the longer stay in the intensive care unit after LT, which may be attributable to the patients' worse pretransplant condition, the duration of the overall hospitalization and post-LT survival did not differ significantly between the 2 cohorts. These results confirm data from Asian studies demonstrating similar long-term survival rates for ACLF patients and patients undergoing transplantation for decompensated liver cirrhosis.[7-9] In contrast to our study, viral hepatitis B was the main underlying liver disease, and the majority of the patients underwent living donor LT. None of our patients underwent living donor LT. Only 1 study from a Western LT center has investigated the outcomes of ACLF patients. In accordance with our data, ACLF, defined as a rise in the MELD score greater than 5 points within the 4 weeks before LT, was not found to be a negative predictor of posttransplant survival. However, this study did not provide data regarding the overall survival of all ACLF patients treated at the center in an intention-to-treat analysis.
To our knowledge, the present study is the first to address the eligibility of ACLF patients for LT on an intention-to-treat basis. In our cohort, not even half of the patients could be listed for LT, and even when ACLF patients were listed for LT, their mortality rate remained high. Mortality rapidly increased during the first 2 months after listing, and this emphasizes the limited timeframe for LT for these patients. Although the high MELD scores of the ACLF patients led to significantly shorter waiting times in comparison with the non-ACLF patients, more than 50% of the listed ACLF patients died on the waiting list, whereas only 16% of the patients listed for other indications did. In contrast to patients with acute liver failure, the current allocation systems do not grant a high-urgency status to ACLF patients. Because of the short median transplant-free survival time of 48 days, our results suggest an extra priority status for ACLF also. This might prevent the development of irreversible multiorgan failure, which was the most common cause of mortality in our ACLF population.
A second possibility for reducing wait-list mortality is the increased use of living donation for these patients, which is the preferred LT modality in Asian centers. Unfortunately, no data regarding wait-list mortality are provided in the aforementioned Asian studies.[7-9]
A prospectively validated and generally accepted definition of ACLF is a requirement for establishing a priority status for ACLF patients on the wait list. In our study, we used the APASL criteria, which were the first available criteria for defining ACLF. However, the APASL criteria are not validated and were developed predominantly in an Asian population in which reactivation of hepatitis B was one of the major causes of ACLF. This limits their general validity in a Western cohort such as ours, which differs significantly with respect to the underlying liver diseases and causes of decompensation. Recently, prospectively evaluated diagnostic criteria for ACLF have been proposed by the European Association for the Study of the Liver and Chronic Liver Failure Consortium. These criteria comprise acute decompensation, which is defined as the acute development of 1 major complication of liver disease (ie, ascites, encephalopathy, gastrointestinal hemorrhaging, or a bacterial infection); the presence of organ failure; and a 28-day mortality rate greater than 15%. Our results were comparable to the results of that study: our cohort was characterized by a high risk of short-term mortality and the development of multiorgan failure in a substantial proportion of the patients.
In order to identify those ACLF patients with the highest chance of undergoing LT, we analyzed several patient characteristics and laboratory parameters. Our data show that those patients who presented with elevated serum C-reactive protein and creatinine levels at admission were less likely to undergo LT. In addition, the development of pneumonia and the need for renal replacement therapy and/or mechanical ventilation during hospitalization were associated with a reduced possibility of receiving a liver graft. These findings are not surprising because several other studies have shown that the admission of patients with cirrhosis to the intensive care unit because of infectious complications leading to multiorgan failure is associated with an extremely poor outcome.[19, 20] Moreover, sepsis and mechanical ventilation were the strongest negative predictors of overall and transplant-free survival in our cohort.
Inflammation and neutrophil dysfunction are of major importance for the pathophysiology of ACLF. In particular, there is increasing evidence that systemic inflammatory response syndrome, characterized by a predominantly proinflammatory cytokine profile (interleukin-6 and tumor necrosis factor α), causes the transition from stable cirrhosis to ACLF.[5, 21-23] The importance of infection as a precipitating event in ACLF is strengthened by the finding that an apparent infection was the most common reason for decompensation in our cohort. This confirms earlier studies demonstrating that (non)hepatic infections are the most important triggers of ACLF.[6, 24, 25]
No identifiable precipitating event could be identified in approximately 40% of the patients, and the survival rate did not differ between patients with a precipitating event and patients without one. This agrees with recent prospective studies[3, 18] and emphasizes that an identifiable precipitating factor is not a prerequisite for the diagnosis of ACLF. Because an overt infection developed in a number of cases after admission and bacteremia was found in 49% of our patients during the hospital stay, an occult infection at admission may be common in ACLF patients without an obvious precipitating event. This highlights the difficult distinction between chronic hepatic decompensation and ACLF in clinical practice. A clear discrimination between these 2 entities would be a premise for establishing a high-priority listing for patients with ACLF.
In conclusion, our study demonstrates that LT remains the only definite therapeutic option for the vast majority of patients with ACLF. LT provides good survival rates with posttransplant outcomes comparable to those for other indications. The wait-list mortality rate of more than 50% highlights that the time to transplantation is crucial and that the window of opportunity is small. High-urgency allocation and the possibility of living donor LT should be considered for ACLF patients.