Acute liver failure (ALF) occurs following sudden extensive loss of liver cell mass resulting in hepatic encephalopathy and coagulopathy. Severe cases of ALF are associated with a high mortality. Most commonly, death results from cerebral edema, overwhelming sepsis, or multiorgan failure. Previous studies have highlighted the increasing contribution of acetaminophen as a cause of ALF in the United Kingdom, United States, and Europe.1–3 Acetaminophen-induced ALF has a distinct clinical course characterized by rapid clinical deterioration, which contrasts with a relatively good prognosis.4 ALF associated with acute hepatitis A or B is also associated with a relatively good prognosis. In contrast, seronegative hepatitis and other etiologies have a more insidious clinical course than acetaminophen but have higher mortality rates. Epidemiological studies have suggested that liver transplantation (LT) is more commonly used in non-acetaminophen ALF,2 but large longitudinal single-center series are not available.
Improved intensive care and medical management of patients have led to an improved prognosis in patients with ALF. However, emergency LT is life-saving in the group of patients with irreversible ALF. The most widely used criteria for determining when ALF is irreversible and hence transplantation should be considered are the King's College poor prognostic criteria (KCHPPC) and the Clichy criteria.5, 6 The former criteria are in universal use in the United Kingdom to assess prognosis and register patients with ALF for emergency LT. Originally described by O'Grady in 1989, these criteria have been adapted recently to include blood lactate.7 Several recent publications have questioned the validity of the KCHPPC in determining prognosis in ALF.8–11
The aim of this study was to analyze the utilization of LT in the management of patients with acetaminophen and non-acetaminophen ALF in a large cohort of both etiologies and to examine the various decision steps in patient management between admission and transplantation; in other words, we sought to determine whether patients fulfilling the KCHPPC were subsequently considered to have medical or psychiatric contraindications to transplantation and whether they survived on the emergency transplant list to undergo LT.
ALF, acute liver failure; ALT, alanine aminotransferase; GGTP, gamma glutamyl transferase; KCHPPC, King's College poor prognostic criteria; LR+, likelihood ratio positive test; LR−, likelihood ratio negative test; LT, liver transplantation; M/F, male/female; NPV, negative predictive value; NR, normal range; NS, not significant; PPV, positive predictive value; PT, prothrombin time; SLTU, Scottish Liver Transplantation Unit.
PATIENTS AND METHODS
The cohorts retrospectively analyzed were from 469 patients admitted to the Scottish Liver Transplantation Unit (SLTU) between November 12, 1992 and November 11, 2006 with ALF as defined by the presence of coagulopathy and hepatic encephalopathy in the absence of pre-existing liver disease. Following admission, detailed clinical history, examination, and laboratory investigations were performed. Liver imaging studies and biopsy were undertaken when clinically indicated. Laboratory investigations were repeated at daily intervals or more frequently in patients with rapidly progressive liver failure.
The patients admitted to the SLTU are managed with a standard protocol. This protocol is reviewed on an annual basis, but the main goals of therapy have remained similar over the 14 years of this cohort study. Patients with acetaminophen poisoning have a continuous infusion of N-acetylcysteine (6.25 mg/kg/hour) until the international normalized ratio is less than 2. Coagulopathy is not routinely corrected unless there is intracranial or gastrointestinal bleeding or there is a requirement for intracranial pressure monitoring. Hypotension is treated with volume expansion and inotropic support and monitored by frequent hemodynamic assessment. Elective ventilation is initiated in patients with grade III or IV hepatic encephalopathy or when there is progressive respiratory failure. Renal replacement therapy is instituted with continuous venovenous hemofiltration in oliguric patients with a progressive rise in serum creatinine or in those who are anuric. Antibiotic and antifungal treatments are prescribed to all ventilated patients. Nutrition is provided by nasogastric feed. Intracranial pressure monitoring is considered in all ventilated patients who may be transplant candidates. Treatments for cerebral edema include mannitol, therapeutic hypothermia, and thiopentone.
Details of the patient history, clinical examination, and laboratory results along with therapeutic interventions including intensive care admission and need for renal replacement or inotropic support were recorded prospectively in the ALF database. Acetaminophen was considered to be the underlying etiology if there was a history of excessive acetaminophen ingestion with or without detectable acetaminophen in the blood. Other etiologies of severe acute liver injury were diagnosed according to positive serological and radiological findings and, in selected cases, transjugular liver biopsy. Seronegative hepatitis was the defined diagnosis when all recognized causes of hepatitis were excluded.
LT was considered in all patients meeting the KCHPPC in conjunction with medical condition and psychological assessment. Both medical and psychiatric contraindications for LT were similar to those applied in other units in the United Kingdom;12 severe coexistent medical disease, progressive inotropic or ventilatory support, irreversible brain stem dysfunction, multiple previous episodes of self-harm (>5) especially if nondrug methods were used, refractory or resistant mental illness, incapacitating dementia or mental retardation, active intravenous drug use or polydrug use in a severe and chaotic fashion, documented persistent noncompliance, a consistently stated wish to die in the absence of established mental illness, and alcohol dependence or excessive alcohol use in a severe and chaotic fashion were all considered contraindications to LT after assessment by appropriate specialists. If accepted as transplant candidates, patients were superurgently registered within the United Kingdom and prioritized for the next available compatible organ. Outcome was defined as spontaneous survival to discharge without transplant, death without transplant, survival with transplant, or death with transplant.
All data were recorded prospectively in the SLTU ALF database. Statistical analysis was performed retrospectively with SPSS 14.0 software (SPSS, Chicago, IL) after November 2006. Data values are presented as medians and ranges or percentages unless otherwise stated. Continuous data were compared with analysis of variance or the Mann-Whitney U test for unpaired data. Categorical data were analyzed with chi-square tests or Fisher's exact test. Actuarial probability curves were constructed by the Kaplan-Meier method and compared with log-rank testing. A 2-sided P value of less than 0.05 was considered statistically significant.
Between November 1992 and October 2006, 469 patients were admitted with ALF. One hundred four (22.2%) patients underwent emergency LT, and 77 transplanted patients (74%) survived 30 days. Acetaminophen was the most common etiology of ALF for patients admitted to the unit over the period of the study (acetaminophen: 326 patients, 69.5%; non-acetaminophen: 143 patients, 30.5%). However, significantly fewer patients admitted with acetaminophen-induced ALF underwent LT in comparison with the non-acetaminophen group (acetaminophen: 45/326 patients transplanted, 13.8%; non-acetaminophen: 59/143 patients transplanted, 41.3%; P < 0.01). Despite this difference in the relative utilization of LT in the acetaminophen cohort compared with the non-acetaminophen cohort, overall survival was similar in the 2 groups (acetaminophen: 165/326 patients survived, 50.6%; non-acetaminophen: 83/143 patients survived, 58.0%; P > 0.05; Fig. 1).
In order to analyze in more detail this discrepancy in the relative utilization of LT in the management of acetaminophen-induced ALF compared with non-acetaminophen ALF, we reviewed the outcomes of the various decision steps in the clinical course of patients before emergency LT; that is, we determined whether patients fulfilling the KCHPPC were subsequently considered to have medical or psychiatric contraindications to transplantation and whether they thereafter survived on the emergency transplant list to undergo LT (Fig. 2). Significantly more patients in the acetaminophen group did not fulfill the KCHPPC versus the non-acetaminophen group (acetaminophen: 161/326 patients, 49.4%; non-acetaminophen: 52/143 patients, 36.4%; P < 0.01). Furthermore, once the KCHPPC were fulfilled, significantly more patients were considered to have contraindications to transplantation in the acetaminophen group compared with the non-acetaminophen group (acetaminophen: 99/165 patients, 60%; non-acetaminophen: 21/91 patients, 23.1%; P < 0.01). Mortality on the superurgent LT waiting list of the acetaminophen patients was significantly increased in comparison with that of the non-acetaminophen patients (acetaminophen: 21/66 patients died, 31.8%; non-acetaminophen: 11/70 patients, 15.7%; P < 0.05). However, following LT, short-term (30 day) posttransplant survival was similar in the acetaminophen and non-acetaminophen groups (acetaminophen: 31/45 patients, 68.9%; non-acetaminophen: 46/59 patients, 78%; P > 0.05). Therefore, at all steps in a patient's course between admission and emergency LT, the decisions favor transplantation in non-acetaminophen patients and nontransplant management in patients with acetaminophen-induced ALF.
Prognostic Utility of the KCHPPC
Recent studies have suggested that the KCHPPC lack sensitivity and specificity in predicting prognosis. In our cohort of patients, the KCHPPC were specific for both acetaminophen-induced and non–acetaminophen-induced ALF (Table 1). Therefore, very few patients with ALF were further considered for transplantation when recovery was expected. In addition, the sensitivity of these criteria was >80% for both etiologies (Table 1). The correct outcome was predicted by the KCHPPC in 88% of acetaminophen cases and 87% of non-acetaminophen cases. If the transplanted patients are excluded from the analysis, the KCHPPC correctly predicted the outcome in 86.1% of acetaminophen cases and 77.4% of non-acetaminophen cases (Table 1). In summary, these data confirm that the use of the KCHPPC is appropriate in deciding which patients are unlikely to survive without further consideration for emergency LT.
Table 1. Comparison of the Predictive Accuracy of King's College Poor Prognostic Criteria in Acetaminophen and Non-Acetaminophen Cohorts
Outcomes of Patients Who Do Not Fulfill the KCHPPC
There was a trend of reduced survival in patients who did not reach the KCHPPC in the non-acetaminophen group compared with the acetaminophen group; this did not achieve statistical significance (acetaminophen: 33/161 patients died, 20.5%; non-acetaminophen: 17/52 patients, 32.7%; P = 0.055; Fig. 3). The patients who survived ALF in both the acetaminophen and non-acetaminophen groups were significantly younger than those who died without reaching the KCHPPC (Table 2). In addition, the patients who died without reaching the KCHPPC following acetaminophen poisoning had significantly higher creatinine and lactate on admission in comparison with the patients who survived (Table 2). In contrast, serum alanine aminotransferase, albumin, and bicarbonate were significantly lower in patients who died without reaching the KCHPPC in comparison with surviving patients following acetaminophen overdose (Table 2). Apart from age, no other significant differences were observed within the non-acetaminophen cohort (Table 2).
Table 2. Comparison of Patients in Different Etiological Cohorts That Did Not Reach King's College Poor Prognostic Criteria According to Final Outcome
Acetaminophen (n = 161)
Non-Acetaminophen (n = 52)
Died (n = 33)
Survived (n = 128)
Died (n = 17)
Survived (n = 35)
NOTE: The results are medians and ranges for admission unless otherwise stated.
Contraindications to LT in Acetaminophen-Induced and Non–Acetaminophen-Induced ALF
Significantly more patients were rejected for listing for transplantation in the acetaminophen group compared with the non-acetaminophen group because of contraindications to LT (Fig. 2). Twenty patients with non–acetaminophen-induced ALF fulfilling the KCHPPC were rejected for listing because of medical contraindications to LT; 1 patient was rejected because of active and resistant alcohol dependence. In the acetaminophen cohort, 43 patients (43.4%) were excluded because they were medically unfit to survive LT. The other 56 patients (56.6%) in the acetaminophen cohort were excluded for a variety of psychiatric reasons thought to limit long-term graft or patient survival, most commonly active and resistant alcohol dependence (37 patients). Alternative causes were active intravenous drug abuse (8 patients), a documented previous history of noncompliance with medical or psychiatric therapy (5 patients), multiple (>6) episodes of deliberate self-harm (4 patients), and a consistently stated wish to die without clinical psychiatric illness (2 patients). Spontaneous survival in this group of patients with predicted poor prognosis was very poor and was similar in both etiological cohorts (acetaminophen: 6/99 patients survived, 6.1%; non-acetaminophen: 2/21 patients survived, 9.5%; P > 0.05).
In conclusion, the significant differences observed between the acetaminophen and non-acetaminophen cohorts at this decision step in patient management are due to the large predominance of psychiatric (mainly alcohol dependence) contraindications in the acetaminophen cohort.
Survival on the Emergency Transplant Waiting List
After the exclusion of patients with medical or psychiatric contraindications to emergency LT, significantly reduced survival on the emergency LT list in the acetaminophen group compared with the non-acetaminophen group was observed.
Median waiting times between listing for emergency LT and subsequent operation were similar in both groups (acetaminophen, 2 days, range 1-4; non-acetaminophen, 2 days, range 1-5 days; P > 0.05). The reduced survival in the acetaminophen group may be explained by the observation that patients with acetaminophen overdose had more organ failure and complications related to ALF than the non-acetaminophen patients. Significantly more patients with acetaminophen-induced ALF on the emergency LT waiting list developed cerebral edema (acetaminophen: 43/66 patients, 65.2%; non-acetaminophen: 18/70 patients, 25.7%; P < 0.01), required inotropic support (acetaminophen: 45/66 patients, 68.2%; non-acetaminophen: 20/70 patients, 28.6%; P < 0.01), and required renal dialysis (acetaminophen: 56/66 patients, 84.8%; non-acetaminophen: 20/70 patients, 28.6%; P < 0.01) prior to LT in comparison with the non-acetaminophen group. Hypoglycemia was also significantly more common in patients with acetaminophen-induced ALF on the emergency LT waiting list (44/66 patients, 66.7%) in comparison with the non-acetaminophen group (16/70 patients, 22.9%).
Temporal Changes in the Management of ALF
The previous data analyzed outcomes in patients with ALF admitted to our unit over a 14-year period. To determine if there had been temporal changes in the utilization of LT over the period of observation, we analyzed the sequential decision steps in the patient management pathway, comparing the initial 4 years of the observation period (1993-1996) with the last 4 years of the observation period (2002-2005). Similar numbers of patients with ALF were admitted in the early (1993-1996) cohort in comparison with the later (2002-2005) cohort, and transplantation was used to a similar degree in the 2 cohorts (1993-1996: 27/133 patients, 20.3%; 2002-2005: 28/130 patients, 21.5%; P > 0.05). However, major differences were observed between the acetaminophen and non-acetaminophen cohorts (Fig. 4A,B) when we compared the initial 4 years of the observation period (1993-1996) with the last 4 years of the observation period (2002-2005).
In the acetaminophen cohort, transplantation was undertaken less frequently in the later cohort compared with the earlier cohort (Fig. 4A). An analysis of the various decision steps in the patient management pathway revealed that significantly more patients had medical or psychiatric contraindications to LT in the 2002-2005 cohort (Fig. 4A). A comparison of the distribution of contraindications in the 2 cohorts suggests that more patients in the later cohort were excluded from transplantation because of active and resistant alcohol dependence. In the earlier (1993-1996) cohort, 13 (52.0%) patients were medically unfit for transplantation, 9 (36%) patients had significant alcohol dependence, 2 (8%) patients had a history of multiple episodes of self-harm, and 1 (4%) patient was an active intravenous drug user. In the later (2002-2005) cohort, 15 (41.7%) patients were medically unfit for transplantation, 16 (44.4%) patients had significant alcohol dependence, 1 (2.8%) patient had a history of multiple episodes of self-harm, 3 (8.3%) patients were active intravenous drug users, and 1 (2.8%) patient consistently expressed a wish to die in the absence of psychiatric illness. In addition, although not reaching statistical significance, more than 50% of cases listed for emergency LT in the later (2002-2005) cohort died awaiting a graft versus 30% in the earlier (1993-1996) cohort.
In the non-acetaminophen cohort, 34 patients were admitted in 1993-1996 versus 49 patients in the latter 4 years (2002-2005). In contrast with the acetaminophen cohort, transplantation was undertaken more frequently in the later cohort compared with the earlier cohort (Fig. 4B). In contrast with the data observed in the acetaminophen cohorts, although not reaching statistical significance, in the non-acetaminophen cohorts, more patients were considered transplant candidates and survived to undergo LT in the later (2002-2005) cohort versus the earlier (1993-1995) cohort (Fig. 4B). Median waiting times between listing for emergency LT and subsequent operation were similar in both groups in the 2 time cohorts analyzed (acetaminophen: 1993-1996, 2 days, range 1-4; 2002-2005, 2.5 days, range 1-4, P > 0.05; non-acetaminophen: 1993-1996, 2.5 days, range 1-4; 2002-2005, 2.0 days, range 2-4, P > 0.05).
This study is the first to present a direct comparison of the utilization of LT in the management of patients admitted to a single center where a standard ALF protocol is applied and King's College criteria are used when decisions are made with respect to LT. Large numbers of patients were admitted with either acetaminophen (326 patients) or non-acetaminophen (143 patients) etiologies, allowing meaningful comparisons to be made. Previous studies have presented reports from multiple centers in which decisions regarding prognosis and eligibility for transplantation may vary2 or have reported the outcomes of a single etiology, most commonly acetaminophen.12–14 Alternatively, some reports are from transplant centers where patients preselected as potential transplant candidates are admitted.9 Lastly, others have reported experiences for centers where either acetaminophen or non-acetaminophen cases predominate, and this precludes a meaningful direct comparison.15–18
Although acetaminophen was the most common etiology of patients admitted to the unit over the period of study, the numbers of patients transplanted for ALF were significantly higher in the non-acetaminophen group. Our analysis identified significantly improved transplant-free survival in the acetaminophen group compared with the non-acetaminophen group. This confirms data from other groups in the United Kingdom, Europe, and the United States.2, 4, 12 Several studies have shown that although patients with acetaminophen-induced ALF have more severe derangement of liver functions than non-acetaminophen cases, a significantly higher spontaneous or transplant-free survival is to be expected in this group even in the presence of deep encephalopathy.2, 4, 12 We have also shown that the predominance of transplantation in the non-acetaminophen cohort compared with the acetaminophen cohort is due to a significant exclusion of patients in the latter cohort for psychiatric contraindications combined with reduced survival on the LT waiting list. Previous analysis of registrations for emergency LT and transplantation rates for acetaminophen poisoning shows that our unit is similar in these respects to others in the United Kingdom.19
LT is a long-established treatment for selected patients with ALF. In our series, 104 patients underwent LT with short-term posttransplant survival of 74% (77 patients). The posttransplant survival was similar in the acetaminophen cohort (69%, 30-day survival) and the non-acetaminophen cohort (78%, 30-day survival). Others have observed similar posttransplantation survival in acetaminophen and non-acetaminophen ALF.20–22 Our data show that mortality on the LT waiting list in the acetaminophen cohort (31.8%) was double the mortality of the non-acetaminophen cohort (15.7%). The median waiting times on the urgent transplant list were similar in the 2 cohorts. The patients with acetaminophen-induced ALF listed for emergency LT had an increased incidence of organ failure with cerebral edema, inotropic support, and renal dialysis in comparison with the non-acetaminophen patients. Given these observations of transplant list survival, it may be worth considering giving priority to patients with acetaminophen-induced ALF over those with non-acetaminophen etiologies for the available liver grafts.
Crucial to the decision about the appropriate utilization of LT in patients with ALF is an accurate prediction of prognosis. In a landmark article, O'Grady and colleagues5 described the etiologically specific prognostic criteria for acetaminophen-induced and non–acetaminophen-induced ALF. These are the transplant criteria in use throughout the United Kingdom and have recently been modified to include the circulating lactate concentration.7 Although we found that the KCHPPC were specific and had good positive predictive value, these criteria in both the acetaminophen and non-acetaminophen cohorts had reduced sensitivity and negative predictive value. Therefore, a significant number of patients with ALF due to acetaminophen or non-acetaminophen causes die without reaching the KCHPPC. Given this limitation of the KCHPPC, several other studies have investigated other potential prognostic indicators in ALF. The Model for End-Stage Liver Disease formula is predictive of outcome in both acetaminophen and non-acetaminophen ALF.10, 11, 23, 24 However, the use of the Model for End-Stage Liver Disease in decisions regarding transplantation in these groups of patients remains controversial.25 Alternative parameters such as Gc concentrations, monocyte DR expression, and metabolic profiles26–29 lack ready availability and have not been generally adopted into clinical practice.
The analysis of the utilization of LT over the period of our observational study revealed that more patients are currently treated with LT in the non-acetaminophen cohort (24 patients in a total cohort of 49 patients, 49.0%, 2002-2005) compared with the earlier cohort (11 patients in a total cohort of 34 patients, 32.4%, 1993-1996). In comparison, LT was used much less frequently in the current acetaminophen cohort (4 patients in a total cohort of 81 patients, 4.9%, 2002-2005) compared with the earlier cohort (13 patients in a total cohort of 99 patients, 13.1%, 1993-1996). These differences appear to be a direct result of the increased number of patients in the acetaminophen cohort that fulfilled the KCHPPC but were considered to have contraindications to successful long-term transplant survival. In 1993-1996, 52.1% of cases were considered to have contraindications to LT versus 80% of cases in 2002-2005. Further analysis identified a significant increase in the numbers of cases excluded for further consideration for LT because of active and resistant alcohol dependence. Interestingly, this parallels the increased incidence of alcoholic liver disease in Scotland over the last 15 years.30 An analysis of trends in acetaminophen overdose in Scotland over the period of 1996-2004 observed little change in the number of cases of adults (ages 20-69) admitted to Scottish hospitals following acetaminophen overdose: approximately 600 cases per quarter.31 There were 327 in-hospital deaths over the period of 1995-2003 attributed to acetaminophen overdose; the number of deaths ranged from 27 to 53 deaths per year.31 In an analysis of in-hospital deaths related to acetaminophen-induced ALF (1994-1995), 50% of cases in Scotland were not transferred to the SLTU, mainly because of poor medical condition or alcohol and/or drug use.32 These data suggest that the population admitted to our unit with ALF due to acetaminophen overdose is preselected into a population in which LT may be more applicable and that the number of patients with medical or psychiatric contraindications to transplantation is higher in the general population than we have reported. The retrospective nature of our study results in some other unavoidable limitations or bias, especially when we are comparing the data between the early and late cohorts. The sizes of the cohorts, especially in the non-acetaminophen group, are relatively small. Since the initial cohorts were managed, there have been a number of clinical developments that may have positively influenced clinical survival, such as cooling in the treatment of cerebral edema33; moreover, lactate has been introduced into the listing criteria for patients with acetaminophen poisoning,7 and the measurements of several important clinical variables, such as the prothrombin time, have changed.34
Our data highlight the clinical need for an effective artificial liver support system to eradicate toxins either not removed by or produced by the injured liver and replace important plasma proteins synthesized by the damaged liver. Plasmapheresis has been studied for many years in patients with ALF. Improvements in both cerebral and systemic hemodynamics have been reported with this treatment, but the effect on the survival of patients with ALF is less convincing.35–37 Many different artificial and bioartificial liver support systems have been described; however, after initially encouraging results, randomized clinical trials have shown no survival benefit.38
In summary, we have presented a retrospective analysis of the relative utilization of LT in ALF, reporting large numbers of both acetaminophen and non-acetaminophen cases from a single center. We have found that although acetaminophen is the most common etiology of ALF in our cohort, LT is used significantly more frequently in the non-acetaminophen cohort. The KCHPPC correctly predicted outcome in 88% of acetaminophen cases and in 87% of non-acetaminophen cases. An analysis of the sequential decision steps in the management of patients with ALF revealed that at all steps in patient management, LT was favored in the non-acetaminophen cohort compared with the acetaminophen cohort. These differences in the relative utilization of LT appear to have further diverged over the 14 years of the study, with more patients excluded from LT because of active and resistant alcohol dependence. These data confirm that the majority of patients with ALF are managed without LT and as such suggest that significant improvements in survival in patients with ALF will depend on improved medical and intensive therapy and liver support devices.
We are grateful for the support of the other members of the medical and nursing team in the management of these patients and the database managers, Janice Davidson and Kirsty Martin.