Acute liver failure (ALF) can be rapidly fatal because of multiorgan failure, sepsis, or cerebral edema.1, 2 Despite significant advances in critical care and an improved understanding of the pathophysiology of ALF, the mortality rate remains high.3–5 Liver transplantation is the only life-saving treatment available beyond the supportive care of a critical care unit. Most countries have regional or national sharing agreements that assign the highest priority on the waiting list to candidates with ALF. The ability to urgently perform liver transplantation within 48 to 72 hours is crucial to reducing the waiting list mortality rate.6, 7 In countries in which deceased organ donation is uncommon, living donors have been used.8–12 Auxiliary liver transplantation has also been reported, but clinical experience is limited to only a few centers.13–15 The promise of bioartificial hepatic devices has yet to translate into clinical effectiveness.16–18 Thus, orthotopic liver transplantation (OLT) with a whole graft from a deceased donor remains the gold standard treatment for ALF.
It is generally accepted that liver transplantation in the urgent scenario has inferior patient and graft survival rates in comparison with liver transplantation in nonurgent cases.19 The published 1-year survival rates have ranged widely from 58% to 92%.7, 20–22 The long-term outcomes after OLT for ALF have not been reported as extensively. Two large series from the Paul Brousse Hospital in 199523 and the University of California at Los Angeles in 200322 reported 5-year patient survival rates of 61% and 67%, respectively. A 73% survival rate at 5 years was reported for a recent series of emergency OLT procedures for ALF.24
In Canada, liver grafts from deceased donors are allocated nationally at the highest priority to patients with ALF.25 We reviewed our center's experience with emergency liver transplantation for ALF to determine the immediate and long-term outcomes, including patient survival, graft survival, and neurological sequelae.
ABOi, ABO blood group–incompatible; ALF, acute liver failure; CI, confidence interval; CT, computed tomography; ECD, extended criteria donor; HR, hazard ratio; INR, international normalized ratio; KCH, King's College Hospital; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation.
PATIENTS AND METHODS
A retrospective analysis was undertaken of all adult patients listed for liver transplantation at University Hospital, London Health Science Centre, from January 1, 1994 to December 31, 2007. Data collection was undertaken from the hospital medical records, clinic charts, and institutional transplant database. The study was approved by the University of Western Ontario institutional review board (study 13896E). The primary endpoint of this study was patient survival. Secondary endpoints were graft survival and complications. The pretransplantation and posttransplantation data that were collected included recipient demographics, donor factors, operative variables, and postoperative outcomes.
In Canada, the highest priority, 4F, is assigned to ventilated patients in the intensive care unit with ALF. A listing of 3F is assigned to ALF patients who satisfy the King's College Hospital criteria but do not require mechanical ventilation.25 Certain patients with ALF who do not meet the King's College Hospital criteria can be listed with certain indications such as acute Wilson's disease, acute Budd-Chiari syndrome, or progressive clinical deterioration. The general agreement among the transplant centers of Canada states that all grafts, regardless of blood group, are offered first to patients listed as 4F and then to patients listed as 3F.
In this analysis, ALF was defined as the onset of hepatic encephalopathy and coagulopathy within 26 weeks of the onset of symptoms of liver disease and the absence of previously known liver disease.26 Hepatic encephalopathy was graded according to the West Haven criteria.27 All patients were evaluated for emergency liver transplantation by a multidisciplinary team, including transplant surgeons, hepatologists, a social worker, an anesthetist, and other medical services as required, including nephrology, infectious disease, and neurology services. Patients with ALF were excluded from listing because of severe multiorgan failure, florid sepsis, ongoing chronic alcohol or illicit drug abuse, or a high risk of noncompliance.
The treatment of ALF included, as required, mechanical ventilation, broad-spectrum antibiotics with antifungal prophylaxis, renal replacement therapy, and nutritional and inotropic support. Screening for cerebral edema was done with a computed tomography (CT) scan only if there were focal neurological findings or seizures. The cerebral edema grading system used at our institution was as follows:
Mild: Diminished ventricular size and blunting of sulci.
Moderate to severe: Effacement of cisterns or sulci, partial or complete loss of gray-white matter differentiation, and loss of ventricular space.
Invasive intracranial pressure monitoring was not used. Suspected or diagnosed cerebral edema was treated with hyperventilation (target partial pressure of carbon dioxide: 30-35 mm Hg), mannitol, a hypertonic saline infusion, reverse Trendelenburg positioning, topical hypothermia, muscle paralysis, and sedation.
Grafts from extended criteria donors (ECDs) were defined as follows: age > 59 years, body mass index > 35 kg/m2, and cold ischemia time > 12 hours. Nationally shared grafts were defined as grafts imported from outside the provinces of Ontario and Manitoba, which represent the procurement region covered by the London Health Science Centre. The furthest procurement regions are British Columbia to the west (3300 km) and Nova Scotia to the east (1300 km). Venovenous bypass was used selectively in patients who were judged to be intolerant of caval occlusion. The standard immunosuppressive regimen since 1997 has included tacrolimus, mycophenolate mofetil, and prednisone. ABO blood group–incompatible (ABOi) grafts were used only in dire circumstances. The immunosuppressive protocol for ABOi grafts included, in addition to the standard regimen, antibody induction (Thymoglobulin, Genzyme Transplant, Cambridge, MA), cyclophosphamide, and plasmapheresis if the anti-A/B immunoglobulin titers were >1.8 U/cm3. Patients were followed in the transplant clinic regularly, and in the case of out-of-region recipients, the local transplant hepatologist was contacted for follow-up data.
Statistical analysis was performed with SPSS version 16.0. Patient survival and graft survival were calculated with Kaplan-Meier estimates. Univariate analyses were performed by Cox regression analysis.
Etiology of ALF and Patient Characteristics
Of 919 liver transplants performed at this institution from 1994 to 2008, 60 were performed in patients for ALF. In all, 64 patients with ALF were listed for transplantation. The patient demographics are shown in Table 1. The highest listing status, 4F, was assigned to 45 patients (70.3%), of whom 15 had a status upgrade from 3F to 4F because of clinical deterioration. The most common etiology was cryptogenic (28%); that is, there was no clinical evidence of viral infection, autoantibodies, drug exposure, metabolic disease, or genetic disease. The category of other diagnoses included acute Budd-Chiari syndrome (2 patients), uncontrollable hemorrhage from an unresectable hepatic hemangioma (1 patient), and ALF of pregnancy (1 patient).
Table 1. Demographics of Patients with Acute Liver Failure Listed for Emergency Liver Transplantation (1994-2007)
Patients listed, n
Median age (range), years
Female gender, n (%)
Final listing status, n (%)
4F (mechanical ventilation)
Idiosyncratic drug reaction
The waiting list mortality rate was 6.3% (4 patients), and the associated diagnoses were cryptogenic (n = 3) and acute fatty liver of pregnancy (n = 1). The complications leading to death were cerebral edema in 3 patients and sepsis in the fourth. The time from wait listing to death was 1 day in 3 patients and 8 days in 1 patient. No patient listed for transplantation had a spontaneous recovery.
The pretransplant clinical status of the ALF cohort (n = 64) is shown in Table 2. The King's College Hospital criteria for ALF were met in 37 patients (57.8%), as this was not mandatory for the listing of ventilated patients with ALF in Canada. In addition, 9 of 19 patients listed at 3F did not meet the King's College Hospital criteria and were listed for other clinical indications. The indication for transplantation in these cases included acute Budd-Chiari syndrome (n = 2), acute Wilson's disease (n = 2), cryptogenic (n = 2), acetaminophen overdose (n = 1), sodium valproate toxicity (n = 1), and Epstein-Barr virus hepatitis (n = 1).
Table 2. Pretransplant Clinical and Laboratory Data
Abbreviations: INR, international normalized ratio; MELD, Model for End-Stage Liver Disease.
Therapeutic interventions, n (%)
Renal replacement therapy
Laboratory, mean values
Serum bilirubin, μmol/L
Serum creatinine, μmol/L
King's College Hospital criteria, n (%)
Grade of hepatic encephalopathy, n (%)
Radiological evidence of cerebral edema, n (%)
Moderate to severe
Donor Graft and Operative Characteristics
All transplants were performed with whole grafts from brain-dead donors. No partial grafts or donation after cardiac death grafts were used. The time from listing to transplantation was less than 3 days in 41 patients (69%). The median cold ischemia time was 8 hours 28 minutes with a range of 3 hours 53 minutes to 32 hours 06 minutes. An ECD graft was used in 20 patients. The ECD risk factors included older donor age (n = 12), a prolonged cold ischemia time (n = 5), and a body mass index > 35 kg/m2 (n = 5). Two grafts had more than 1 of the defining ECD risk factors. The median cold ischemia time of grafts procured within the region was 8.0 hours, and for grafts from outside the region, it was 10.3 hours. A nationally shared graft was required in 16 recipients, and only 2 of these grafts exceeded a 12-hour cold ischemia time. An ABOi graft was required in 6 patients. The operative characteristics are shown in Table 3.
Table 3. Operative Characteristics and Posttransplantation Complications
Median patient survival after re-LT, years (range)
Morbidity and Mortality
Hepatic artery thrombosis occurred in 3 patients, who all underwent laparotomy and arterioplasty. Subsequently, 1 patient was retransplanted for hepatic artery thrombosis, another died of sepsis 3 months after transplantation, and the remaining patient had good graft function with 14 years of follow-up. Primary nonfunction occurred in 1 patient. Bile leaks occurred in 2 cases, and they were converted to a Roux-en-Y hepaticojejunostomy. Four patients who received ABOi grafts developed diffuse biliary strictures. There were 9 perioperative deaths (15%). In the immediate postoperative period (< 24 hours), this was due to cerebral edema (n = 2) and cardiac arrest (n = 1). The remaining deaths occurred from 8 to 108 days after the operation and were secondary to sepsis and multiorgan failure (n = 5) and cardiac arrest (n = 1).
Preoperative Cerebral Edema of ALF and Posttransplant Neurological Complications
A radiological investigation of clinical neurological changes for cerebral edema was performed by a CT scan in 31 patients, of whom 11 had positive findings (Table 2). The perioperative mortality rate associated with cerebral edema was 64%. A moderate to severe grade of cerebral edema pre-transplant resulted in 3 deaths after transplantation and 1 recipient with severe cognitive deficits, impaired speech, and motor dysfunction requiring long-term assisted living. Moderate to severe cerebral edema (n = 4) and mild cerebral edema (n = 7) were associated with 1-year survival rates of 25% and 44%, respectively. There were no deaths within the first year among patients without signs of cerebral edema on imaging (n = 20).
Major posttransplantation neurological morbidities occurred in 6 recipients, including anoxic injury, central pontine myelinolysis, and stroke. These were associated with 4 deaths from concomitant sepsis and, in the remaining 2 recipients, with chronic neurological deficits requiring assisted living. Six of the 8 patients with neurological complications or death had radiological evidence of pretransplant cerebral edema. One recipient with a neurological complication did not have imaging pre-transplant, and another had had a normal CT scan.
Long-Term Patient and Graft Outcomes
The overall 1-, 5-, and 10-year patient and graft survival rates by Kaplan-Meier estimates were 80%, 76%, and 69% and 73%, 65%, and 59%, respectively (Fig. 1). In long-term follow-up, 5 patients died from rejection due to noncompliance (3.4 years), a ruptured abdominal aortic aneurysm (4.9 years), recurrent hepatitis B (5.6 years), metastatic prostate cancer (9.8 years), and chronic rejection (11.8 years). The mean long-term survival of patients who received ABOi grafts was 6.5 years, and 2 of these patients died from acute rejection due to noncompliance and chronic rejection, as described previously. Five of the 6 ABOi grafts failed, with retransplantation required in each patient. By univariate analysis, 2 factors reached statistical significance with long-term patient survival: pretransplant cerebral edema and the use of an extended criteria graft (Table 4). However, when nationally shared grafts were included in the definition of ECD, it was no longer statistically significant.
Table 4. Univariate Analysis of Overall Survival by Cox Regression Analysis
Retransplantation was performed in the 4 patients with biliary strictures in ABOi grafts and in 4 other patients for primary nonfunction (n = 1), hepatic artery thrombosis (n = 1), and hyperacute rejection (n = 2), 1 of which was associated with an ABOi graft.
In this series of emergency OLT for ALF, the 5-year patient and graft survival rates were 76% and 65%, respectively, comparing favorably with the most recent reports for emergency liver transplantation, in which the 5-year patient survival rate was 73%24 and the graft survival rate was 57%.22 Successful outcomes in these patients are determined by the balance between urgent need and graft quality, whereas the tendency is to compromise on the latter because of the critical status of the candidates on the waiting list.
Cerebral edema is a leading cause of death in patients with ALF3, 28 and, in this series, accounted for 5 of the 7 deaths in the pretransplant and immediate (<24 hours) posttransplant periods. Pretransplant cerebral edema was associated with worse long-term survival. In most published studies of ALF, a CT scan is used as a screening tool and to rule out intracranial hemorrhaging and other mass lesions.29 Cerebral edema is usually noted, but a grading system is not reported. The grading system of cerebral edema in this series has been used clinically at our institution but never validated as a prognostic indicator of outcome in ALF. The poor posttransplant outcomes associated with a severe grade could reflect the high risk of irreversible brain injury, and it should be considered a relative contraindication to transplantation. The 1-year survival rate of 44% associated with mild cerebral edema with focal neurological findings is also a troubling finding. Although the findings of CT scans have not been shown to correlate with intracranial pressure,30, 31 radiological evidence of macroscopic brain injury should still be of clinical concern. Even though cases of survival after severe intracranial hypertension have been documented and are possible, judicious use of grafts is warranted because of limited donor organ availability. Further prospective study of this grading system is necessary to confirm the prognostic value, and we should also seek to determine the pretransplant predictors of successful posttransplant outcomes in patients with ALF complicated by cerebral edema. In our institutional practice, only patients with a change in the neurological examination were imaged. Thus, we do not know the true incidence of cerebral edema in this cohort. It may be that subclinical cerebral edema, at least in the mild form, was a more common finding than we have shown. However, cerebral edema coupled with neurological findings was associated with poor survival and should also be a cause for grave concern in the pretransplant period. Clearly, more clinical research on ALF-associated cerebral edema will be required in the future to improve posttransplant neurological outcomes.
Preoperative encephalopathy has been shown to be a strong predictor of postoperative neurological morbidity.32 Pretransplant cerebral edema may also be an important predictor and was present pre-transplant in 4 of 6 patients with neurological morbidities (excluding seizures). In this series, patients with serious neurological morbidities post-transplant either died within the first year because of sepsis and multiorgan failure or had permanent cognitive, gait, and speech deficits. The identification of patients at a high risk of neurological complications is important to continue improving the long-term outcomes of transplantation for ALF. The consequences of permanent neurological sequelae, including loss of quality of life, institutionalization, failure to return to gainful employment, health care resource usage, and the burden on the family, are not captured in patient or graft survival. Disappointment for the patient, family, and health care team when life-saving transplants are met with unacceptable quality of life is real and needs to be acknowledged.
The quality of the donor graft, as defined by ECD criteria, had a statistically significant association with long-term recipient survival. In earlier reports of emergency liver transplantation, graft quality was sacrificed to avoid waiting list mortality,22, 23 yet the practice of using the first available graft was complicated by a primary nonfunction rate as high as 13%.22 Higher risk grafts, such as severely steatotic or partial grafts, have also been shown to result in worse survival in ALF.23 In this series, an ECD graft was defined by older age, donor obesity, and a long cold ischemia time and was a significant risk factor for long-term survival. However, when the definition of ECD was broadened to include nationally shared organs, it was no longer significant. In Canada, organ sharing across the continent has proven to be feasible, but it imposes logistical challenges to minimize the cold injury in the face of long-distance transport and dovetailing donor and recipient operations. Only 2 of 16 nationally shared grafts had a cold ischemia time over 12 hours. Thus, the donor pool is not restricted by transcontinental distance but requires tight coordination of the procurement, transport, and recipient operation to minimize cold ischemic injury. It has been our desire to avoid the use of questionable grafts in critically ill recipients who can least afford the consequences of immediate poor graft function. We have felt that the use of ECD grafts in our experience is conservative and has minimized the rate of primary nonfunction to 2%. Only 2 grafts had more than 1 ECD risk factor, and no grafts from non–heart beating, split, or living donors were used. The luxury of waiting for a whole graft of good quality requires a sufficient pool of shared grafts.
The use of ABOi grafts has been most frequently reported for emergency liver transplantation.33 The usage has ranged from 13% to 29%22, 23, 34 and is associated with worse outcomes, including mortality,7, 23, 33 graft survival,7 and complications.35 More recently, some centers have had equivocal short-term results in comparison with ABO-compatible grafts,21, 34, 36 but long-term outcomes are unreported. The most common long-term complication has been diffuse biliary strictures.37 In our experience, ABOi grafts were associated with poor graft survival. There were high rates of chronic biliary strictures (67%) and the need for retransplantation (83%). Long-term patient survival was still good in this group, but only because of the use of a second compatible graft and the commitment of resources that are always greater in patients who undergo repeat transplantation.
If mortality on the waiting list is to be avoided in patients with ALF, organ sharing agreements have to effectively keep the waiting list times to a few days at most. National organ sharing has allowed our center to perform OLT in the majority within 3 days of listing and has contributed to a low waiting list mortality rate. It has been previously documented that the implementation of regional sharing agreements in the United Network for Organ Sharing system led to dramatic improvements in the waiting list mortality rate.6 A short waiting period, particularly one less than 3 days, has been associated with improved posttransplant survival.7, 38 In this series, only 1 waiting list death, on the eighth day, might have been averted if a suitable graft had been made available in a timely manner. The remaining 3 waiting list deaths occurred within 24 hours of listing and were likely inevitable because of the overwhelming progression of cerebral edema and sepsis.
The long-term outcomes of emergency liver transplantation are still inferior to the standard of 80% 5-year survival in nonurgent liver transplantation, but they are getting closer. The keys to an optimal outcome include a short waiting time, a large donor pool, good quality grafts, and a dedicated multidisciplinary transplant team. Severe cerebral edema will still be the greatest pretransplant variable that determines outcome, and our experience suggests that it should be considered a relative contraindication for transplantation. ABOi grafts can be used in emergency situations but are associated with a high graft failure rate.