|Home | About | Journals | Submit | Contact Us | Français|
Acute liver failure (ALF) is a dramatic and highly unpredictable clinical syndrome defined by the sudden onset of coagulopathy and encephalopathy. Although many disease processes can cause ALF, acetaminophen overdose is the leading cause in the United States, and has a 66% chance of recovery with early N-acetylcysteine treatment and supportive care. Cerebral edema and infectious complications are notoriously difficult to detect and treat in ALF patients and may lead to irreversible brain damage and multi-organ failure. Emergency liver transplantation is associated with a 70% 1-year patient survival but 20% of listed patients die, highlighting the importance of early referral of ALF patients with a poor prognosis to a liver transplant center.
Acute liver failure (ALF) is an uncommon, but dramatic, clinical syndrome defined by the onset of coagulopathy (INR > 1.5), and mental status changes within 8 to 26 weeks of presentation [1, 2]. The etiology of ALF is usually rapidly established by patient history, laboratory tests, and imaging studies, but the etiology remains unknown in up to 20% of cases . Acetaminophen overdose is the most common cause of ALF in Western countries and its incidence appears to be increasing. Fortunately, most acetaminophen overdose patients recover with early N-acetylcysteine (NAC) therapy and supportive care, but regulatory actions are needed to prevent future cases. Many ALF patients develop infectious, cardiopulmonary, or renal complications that can progress to multi-organ failure. In addition, cerebral edema is notoriously difficult to diagnose and treat and can lead to irreversible brain ischemia and eventual death. Emergency liver transplantation is associated with a 70% 1-year patient survival, but less than 10% of ALF patients are listed and up to 20% of listed patients die awaiting liver transplantation. Therefore, early referral of patients with a poor prognosis to a liver transplant center is essential to optimize clinical outcomes.
The low annual incidence of ALF in the United States, estimated at 2,800 cases per annum, makes it difficult to collect reliable data on the etiologies, risk factors, and outcomes with this clinical syndrome [2, 4]. This low annual incidence can also lead to referral, selection, and ascertainment bias in single center reports. ALF occurs in patients of all ages, but the etiologies and prognosis markedly differ in adults compared to infants and children [2, 5]. In addition, a predominance of female patients have been consistently reported for nearly all etiologies of ALF for reasons that are unclear (Table 1).
The United States Acute Liver Failure Study Group (ALFSG) is a network of 23 tertiary care centers that have been prospectively studying the etiologies and outcomes of ALF since 1998 . A recent analysis of 1,033 consecutive adult patients enrolled through July 2007 demonstrates that acetaminophen overdose accounts for 46% of cases, followed by indeterminate ALF (15%), and idiosyncratic drug reactions (12%) (Figure 1). Over the past 8 years, an increasing frequency of acetaminophen overdose cases and a decreasing frequency of indeterminate and Hepatitis A virus (HAV) cases has been noted [6, 7]. Other identifiable etiologies of ALF include acute HBV infection (7%), autoimmune hepatitis (5%), ischemic hepatitis (4%), and various other causes (5%). Overall survival was 67% at 3 weeks after presentation, with 46% spontaneously improving and 25% requiring emergency liver transplantation . The likelihood of spontaneous recovery was highest in patients with acetaminophen overdose, HAV, and pregnancy (58 to 64%) while patients with Wilson’s disease, indeterminate ALF, and drug reactions had the worst prognosis (0 to 30%).
The diagnosis of ALF is based upon the physical examination (altered mental status) and laboratory findings (INR > 1.5). The initial evaluation should include rapid identification of the underlying etiology, with an emphasis on treatable conditions (Table 2). In addition to serologic testing, a urine toxicology screen and liver imaging, a careful review of all ingested medications is important. ALF is occasionally confused with other clinical entities such as sepsis, systemic disorders with hepatic and brain involvement (e.g. systemic lupus erythematosus, thrombotic thrombocytopenic purpura), and acute decompensation of chronic liver disease. Alcoholic hepatitis or flares of chronic HBV may be mistaken for ALF, but a careful review of the patient’s medical history, laboratory tests, and imaging studies should differentiate between these conditions. Septic patients with intrahepatic cholestasis and disseminated intravascular coagulation typically have low factor VIII levels, while ALF patients typically have normal factor VIII levels but low factor V levels .
ALF usually presents initially with nonspecific symptoms such as nausea, vomiting, and malaise. Severe acute liver injury often leads to impaired elimination of bilirubin, manifesting as jaundice immediately prior to or shortly after presentation. In addition, the depressed synthesis and excessive consumption of clotting factors results in a complex coagulopathy. A diminished synthesis of glucose, increased intracellular lactate production, and reduced hepatic uptake of lactate can lead to hypoglycemia, and metabolic acidosis. Thirty to forty percent of ALF patients present with impaired renal function and associated azotemia and oliguria .
Mental status changes or encephalopathy are a defining criterion of ALF. They are believed due to cerebral edema, particularly in patients with rapid onset ALF, while portosystemic shunting of toxins is implicated in patients with subacute ALF. The complications of ALF, such as hypoglycemia, sepsis, fever, and hypoxia/hypotension, also contribute to neurological abnormalities. The West Haven criteria for encephalopathy are frequently applied to ALF patients, although the Glasgow coma score is more useful for intubated patients. Subjects with grade 1 encephalopathy only have subtle changes in affect, altered sleep patterns, or difficulties in concentration. Subjects with stage 2 encephalopathy have drowsiness, disorientation, and confusion, while stage 3 is marked by somnolence and incoherence. Subjects with stage 4 have frank coma with minimal (4A) or no (4B) responses to noxious stimuli. ALF patients often have asterixis or tremors in stages 1 or 2, and hyperreflexia, clonus, or muscular rigidity in stages 3 or 4. Although worrisome, these upper motor neuron signs are reversible with hepatic recovery. Patients progressing to stage 3 or 4 encephalopathy have a poorer outcome than those with a maximum of stage of 1 or 2 encephalopathy (70% stage I survival versus 20% stage IV survival) .
Acetaminophen overdose is the leading cause of ALF in the United States as well as other Western countries and is recently increasing [3, 6]. There are an estimated 60,000 cases of acetaminophen overdose annually, with most cases being intentional suicide gestures . Nearly 26,000 overdose patients are hospitalized each year, and an estimated 1% develop severe coagulopathy or encephalopathy. The mortality attributed to acetaminophen overdose is 500 per annum, and at least 20% of these deaths occur in patients with non-intentional acetaminophen overdose . Nearly half of acetaminophen-related ALF cases are therapeutic misadventures . The increasing incidence of acetaminophen-induced ALF may, in part, reflect a shift from aspirin to acetaminophen-based products for acute febrile illnesses, the presence of acetaminophen in numerous over-the-counter and prescription medications, and under appreciation of its hepatotoxicity .
Acetaminophen is a dose-dependent hepatotoxin that can cause severe acute hepatocellular injury. The injury leads to a characteristic pattern of pericentral necrosis due to the P-450 mediated oxidative metabolism of acetaminophen to the highly reactive intermediate metabolite, NAPQI (N-acetyl-p-benzoquinone iminine) (Figure 2) . Although there are intracellular mechanisms to detoxify NAPQI, excessive production can deplete intrahepatic glutathione stores and bind to intracellular proteins leading to hepatocellular necrosis. Chronic consumption of alcohol can induce CYP2E1 activity and increase the rate of NAPQI formation with therapeutic dosing . However, short-term studies of therapeutic doses of acetaminophen in recently abstinent alcoholics have not demonstrated hepatotoxicity . Ingestion of other cytochrome P-450 inducers, such as phenytoin and isoniazid, can lower the threshold for acetaminophen hepatotoxicity . Many patients with non-intentional acetaminophen overdose report short-term fasting and/or poor nutritional status immediately preceding the event, but prospective studies of hepatic glutathione stores at presentation with ALF are unavailable .
The hallmark of acetaminophen hepatotoxicity is the presence of elevated serum aminotransferase levels (up to 400 X ULN) with concomitant hypoprothrombinemia, metabolic acidosis, and renal failure. Most patients have normal, or minimally elevated, serum bilirubin levels at presentation due to the acuity of the liver injury. The diagnosis of acetaminophen hepatotoxicity requires a high index of suspicion (Table 3). Some patients present with unexplained nausea and vomiting and are noted to have only mild aminotransferase elevations, metabolic acidosis, or isolated hypoprothrombinemia. Others present with obtundation following a witnessed or unwitnessed overdose. The minimal dose of acetaminophen that produces liver injury varies from 4 to 10 grams. Recent prospective studies demonstrate evidence of mild biochemical liver injury with therapeutic dosing of 1 gram of acetaminophen every 6 hours in healthy volunteers . Acetaminophen hepatotoxicity should, therefore, be considered whenever the dose exceeds 4 grams/day.
Subjects with non-intentional acetaminophen overdose often present with 2 to 3 days of nonspecific symptoms superimposed on the acute or chronic medical condition for which they were taking an acetaminophen-based analgesic [6, 19]. Non-intentional overdose patients generally were exposed over several days, have low or undetectable serum acetaminophen levels, and more advanced encephalopathy at presentation. In addition to a serum and urine toxicology screen for illicit substances, a careful review of all prescription and over-the-counter medications is critical (Table 4).
Serum acetaminophen levels can help estimate the risk of liver injury following a one time ingestion [13, 20]. However, a low level does not exclude significant overdose, and repeated serum samples at 4 to 12 hours may be needed to define the hepatic risks. Serum bilirubin levels exceeding 10-15 mg/dl can lead to false positive acetaminophen levels with some colorimetric assays . Given these limitations, detection of serum APAP-cysteine protein adducts that emanate from the liver may prove to be a more sensitive and specific biomarker [22, 23]. Although the diagnostic and prognostic significance of adduct levels is still evolving, this assay may prove particularly useful in patients unable to provide a medication history or in patients presenting after multiple ingestions over time. Furthermore, detection of adducts in patients with virally-mediated ALF, where acetaminophen may be a toxic cofactor, could permit more rapid administration of NAC.
Standard medical therapy of known or suspected acetaminophen overdose includes induction of emesis via ipecac syrup, gastric lavage of pill fragments, and administration of activated charcoal to reduce absorption (Table 3) . The likelihood of subsequent hepatotoxicity is estimated in patients with a single ingestion by the RUMACK normogram.
Patients with known or suspected intentional acetaminophen overdose should be hospitalized to assess their suicidal risk. Patients with unstable hemodynamics, renal failure, or altered mental status should be monitored in an intensive care unit and transferred to a liver transplant center early, if deemed a potential transplant candidate. Subjects at risk for hepatotoxicity based upon their initial serum acetaminophen level, elevated serum aminotransferase level or INR level, should be immediately administered NAC. Oral NAC is given as a loading dose of 140 mg/kg followed by a maintenance dose of 70 mg/kg for up to 72 hours or until the INR has become < 1.5 . Most patients tolerate oral NAC, with the co-administration of antiemetics. However, an intravenous formulation of NAC (Acetadote®, Cumberland Pharmaceuticals, Nashville, TN) is available for subjects who cannot tolerate oral NAC . Most experts recommend continuous intravenous infusion of NAC until the INR is less than 1.5. This formulation is particularly useful in pregnant women, patients with a short-gut, or patients with an ileus. This drug should be administered in a monitored unit because up to 3% of patients receiving intravenous NAC develop a hypersensitivity reaction. Patients who experience a mild-to-moderate infusion reaction should have the infusion rate decreased by 50% and receive antihistamines and/or corticosteroids.
The ALFSG recently demonstrated that nearly 50% of acetaminophen related-ALF occurs without an overt suicide intent . In most of these patients, the amount of acetaminophen ingested exceeded the maximal daily recommended dose of 4 grams/day. However, nearly 50% of these patients reported ingesting only 4 to 10 grams/day, and 38% of patients ingested a multitude of products. Contrary to prior reports, these patients were not more likely to be receiving antidepressants or to have a history of alcohol abuse [14, 19]. Nonetheless, non-intentional overdose patients had more advanced encephalopathy at presentation presumably due to frequent narcotic administration. Fortunately, 95% of these patients received NAC, and their spontaneous survival was similar to patients with intentional overdose (64% vs 66%).
Because acetaminophen hepatotoxicity is the leading cause of ALF and yet completely preventable, some experts have recommended regulatory changes regarding the labeling and dispensation of acetaminophen-containing products [12, 26]. In the United Kingdom, blister packaging and restrictions on the dispensation of acetaminophen tablets have led to a reduction in the number of patients with intentional overdose and in those referred for liver transplantation [27, 28]. The Food and Drug Administration (FDA) recently proposed changes in the labeling of all over-the-counter products that contain acetaminophen as well as non-steroidal anti-inflammatory drugs . Additional limitations on the dispensing of prescription acetaminophen-narcotic congeners and reducing or eliminating the acetaminophen component of these products have been suggested but not yet implemented . In the interim, health care providers and pharmacies should be aware of the acetaminophen content in many compound medications.
Severe acute HAV, hepatitis B virus (HBV), and hepatitis E virus (HEV) infection occasionally produces ALF. The diagnosis of HAV-related ALF depends upon the detection of anti-HAV IgM. Young children, subjects over 50 years old, and individuals with underlying liver disease may be more prone to develop severe acute HAV. The overall incidence of ALF from acute HAV infection is < 1% [7,30]. A recent analysis of the UNOS transplant database and the ALFSG confirmed a significant decline in the incidence of fulminant HAV in the United States between 1998 and 2005 . This decline is presumably due to more widespread HAV vaccination and the reduced incidence of sporadic acute infection. Of note, the CDC liberalized recommendations for HAV and HBV vaccination in 2007 so that any individual is now eligible for vaccination .
Fulminant HBV infection occurs in < 1% of acutely infected individuals. It is diagnosed by the presence of detectable HBsAg and/or anti-HBc IgM antibody. However, some patients with chronic HBV may develop transiently detectable anti-HBc IgM during a disease flare [32, 33]. Patients with fulminant HBV occasionally have hepatitis delta virus (HDV) co-infection or superinfection as confirmed by detection of anti-HDV antibodies. Although early studies suggested that pre-core and core-promoter variants of HBV were associated with ALF, recent studies have failed to demonstrate this association [32, 33]. The role of HBV genotypes and host factors in determining susceptibility to HBV-related ALF is unclear. Subjects with HBV-related ALF have only a 30% likelihood of survival [33, 34]. Although the pathogenesis of fulminant HBV is believed due to an overwhelming immune response to infected hepatocytes, the use of oral antiviral agents such as lamivudine or entecavir has been proposed . However, a recent randomized controlled trial of lamivudine in Indian patients with severe acute HBV failed to demonstrate any clinical benefit . A retrospective review of the ALFSG experience from 1995 to 2006 also failed to demonstrate any benefit from antiviral therapy in 76 subjects with HBV-related ALF . Nonetheless, many experts use oral antiviral agents for fulminant HBV because of their relative safety .
Severe acute HEV infection is a leading cause of ALF in tropical countries. It most commonly occurs in pregnant women [37, 38]. It is diagnosed by detection of anti-HEV IgM antibody. The treatment is supportive. Recently, a recombinant protein vaccine for HEV was shown to be safe and effective in preventing acute infection in a high risk population from Nepal . Other non-hepatotrophic viruses including Epstein-Barr virus (EBV), cytomegalovirus (CMV), herpes simplex virus (HSV), varicella zoster virus, human herpes virus-6 (HHV-6), and parvovirus B-19 can rarely cause ALF [40-42]. Whether these rare causes of ALF are due to viral variants or an aberrant host immune response is unclear. Diagnosing ALF due to one of these non-hepatotrophic viruses is frequently difficult and often requires histological confirmation as well as PCR testing. In particular, the majority of patients with HSV-related ALF have no skin lesions at presentation [43, 44]. However, severe acute EBV, CMV, or HSV should be considered because they can be successfully treated with antiviral therapy (Table 2).
Drug induced liver injury (DILI) is a leading cause for the discontinuation of drugs in development and for regulatory actions on previously approved drugs . DILI is rare (1 in 10,000 to 1 in 1,000,000 patient years) and felt to be due to host metabolic idiosyncrasy [46, 47]. Most patients with severe DILI experience acute hepatocellular injury resulting in jaundice, but some patients develop severe DILI from severe cholestatic hepatic injury [48, 49]. Multiple case series demonstrate a preponderance of women with DILI as well as those progressing to ALF [3, 48]. Whether women are more susceptible to idiosyncratic drug-induced ALF due to differences in body weight, drug dosing, or metabolizing/detoxification enzyme activity is unknown.
Idiosyncratic drug reactions are characterized by variable latency after initial administration, but usually occur within 12 months of drug initiation. Genetically determined variability in host toxification, detoxification, and regeneration pathways is implicated in the pathogenesis and outcome of idiosyncratic DILI but supportive data is limited . The role of medication dose, drug-drug interactions, alcohol consumption, the host immune response and other environmental cofactors is largely unknown [50, 51]. The Drug Induced Liver Injury Network should provide insight into the etiologies and mechanisms of DILI by prospectively collecting biological samples from well phenotyped cases . (See http://dilin.dcri.duke.edu/ for additional information).
The primary treatment of drug-induced ALF is to discontinue the suspected drug to avoid further hepatic injury . A recent uncontrolled series suggested a potential role for corticosteroids in some patients with severe DILI, but this approach is controversial . In addition, corticosteroids were not beneficial in large, randomized controlled studies of ALF patients . DILI is notoriously difficult to diagnose because patients usually lack immunologic or allergic features at presentation, often are taking multiple drugs, and a confirmatory laboratory test is unavailable. Liver histology in severe DILI is usually not beneficial except for excluding other treatable causes. Therefore, DILI is a diagnosis of exclusion that uses causality assessment instruments which have substantial limitations [56, 57].
In addition to prescription drugs, a careful history of herbal, complementary, and alternative medicines is needed in patients with unexplained ALF. For example, green tea, ephedra, and various weight loss agents have been associated with ALF [58-60]. Unfortunately, herbal products are not closely regulated during development, manufacturing, or marketing, and the specific hepatotoxic ingredient(s) in many mixtures is impossible to identify.
Development of jaundice in combination with high serum aminotransferase levels in patients with DILI has an estimated mortality rate of 10% (Hy’s rule) . A recent retrospective review of 784 Swedish DILI cases confirmed that serum AST and bilirubin levels at presentation are the most important predictors of mortality or liver transplantation in severe hepatocellular DILI . A recent review of 95 Japanese DILI cases also identified a high serum bilirubin at presentation and a prolonged latency period to be risk factors for mortality . A review of the UNOS liver transplantation database from 1990-2002 highlighted the etiologies and outcomes of 270 adult liver transplant recipients with drug-induced ALF . A striking female predominance was reported (76%), the mean age was 35 years, and acetaminophen was the culprit in 49% of cases. Commonly implicated medications in the idiosyncratic DILI group included isoniazid (17.5%), propylthiouracil (9.5%), phenytoin (7.3%), and valproate (7.3%) . In the US ALFSG, the two-thirds of DILI patients were female, most presented with high serum bilirubin levels (median 22 mg/dl), and had symptoms for an average of 10 days prior to presentation (Table 1). Implicated medications included antituberculosis drugs (20%), sulfa compounds (12%), phenytoin (10%), and various herbs (10%) (WM Lee, MD, personal communication). Overall, idiosyncratic DILI ALF patients had a poor prognosis, with a spontaneous survival rate of only 26% at 3 weeks. Therefore, any patient who develops jaundice with coagulopathy or encephalopathy from suspected DILI should be urgently referred to a liver transplant center.
Autoimmune hepatitis rarely causes ALF . Autoimmune serologies and liver biopsy can aid in the diagnosis, but many of these patients have low titer or undetectable autoantibodies. The benefit of corticosteroids in fulminant autoimmune hepatitis is unclear. However, early identification of ALF due to autoimmune hepatitis is important due to a low spontaneous survival (Figure 1). Acute liver failure is a well known complication of several pregnancy-related liver diseases including acute fatty liver of pregnancy (AFLP) and the syndrome of hemolysis, elevated liver enzyme levels and low platelet count (HELLP) [65-67]. Treatment of these conditions is directed towards prompt delivery of the fetus. The hallmark of AFLP is rapid development of microvesicular steatosis in the third trimester with resultant mitochondrial dysfunction, metabolic acidosis, and coagulopathy, with only mild-to-moderate serum aminotransferase elevations. Mothers with long chain fatty acid metabolic defects (i.e. LCHAD) are at increased risk of developing AFLP, but only 25% of AFLP mothers exhibit an identifiable mutation . Although most women with AFLP or HELLP improve with prompt delivery, some women require emergency liver transplantation.
Severe acute viral hepatitis and HSV hepatitis should also be considered in pregnant women with ALF, particularly in the third trimester, because these conditions are associated with a poor prognosis even with prompt delivery. Sudden hepatic outflow obstruction due to occlusion of all 3 hepatic veins (Budd-Chiari syndrome) is a rare, but potentially treatable, cause of ALF . Patients usually present with recent onset of abdominal pain, hepatomegaly, and ascites. Over 80% of patients have an identifiable thrombophilia which may be treated with anticoagulation but many require liver transplantation . Arterial hypoperfusion to the liver due to cardiogenic shock or hypovolemia can lead to ischemic hepatitis and may progress to ALF . The outcome in these patients is primarily determined by the underlying cardiopulmonary disease, and liver transplantation is rarely required or indicated. Amanita phalloides mushroom poisoning is a rare cause of ALF that often presents with severe gastrointestinal symptoms and diarrhea. Assays for amanita toxin are unavailable. Patients can be successfully treated with IV penicillin G, silymarin, and dialysis, although many require liver transplantation [72, 73].
Wilson’s disease is a hereditary disorder of impaired biliary excretion of copper that presents as ALF in up to 25% of adolescent or young adult patients . Clues to fulminant Wilson’s disease include the presence of Kayser-Fleischer rings on slit-lamp exam in up to 50% of cases, low serum alkaline phosphatase levels, hemolytic anemia with hyperbilirubinemia, and low serum ceruloplasmin levels (but normal in 15%) . Elevated serum and urinary copper levels often occur, but these tests may not be feasible due to the frequent presence of concomitant renal failure. A transjugular liver biopsy can definitively establish the diagnosis by detecting elevated quantitative hepatic copper levels as well as advanced hepatic fibrosis but this biopsy is not always feasible. Since fulminant Wilson’s disease has 100% mortality in the absence of liver transplantation, all of these patients should be quickly listed for transplantation.
Acute Hodgkin’s and non-Hodgkin’s lymphoma, metastatic carcinoma (e.g. lung, breast, melanoma), and several variants of leukemia are rare infiltrative causes of ALF [76, 77]. Although a diagnosis of fulminant malignancy may be suspected based upon history, laboratory tests or imaging, liver biopsy is frequently required for confirmation. These patients have a poor prognosis and are not candidates for liver transplantation .
No etiology is identified in up to 20% of adult patients with ALF and 50% of children with ALF [3, 5, 78]. Prior studies have failed to demonstrate occult infection with HBV, HEV, parvovirus B-19, HSV, and SEN-V in US ALFSG adult patients with indeterminate ALF [33, 43, 79, 80]. Other proposed etiologies include occult autoimmune hepatitis, undiagnosed acetaminophen hepatotoxicity, or DILI . In the ALFSG, 19% of the indeterminate ALF patients had detectable serum APAP-cysteine adducts; these patients tended to have higher serum aminotransferase levels and lower bilirubin levels at presentation than adduct-negative indeterminate ALF patients . However, whether acetaminophen was the primary cause of ALF or merely a cofactor in these cases is unclear. Patients with indeterminate ALF have a poor likelihood of spontaneous recovery and should be rapidly evaluated for liver transplantation.
A key principle in management is the unpredictable and rapid manner in which patients with ALF can deteriorate. Therefore, ALF patients should be monitored in an intensive care unit for frequent neurological and hemodynamic assessment . If the prognosis is poor, early transfer to a liver transplant center is recommended.
A rapid evaluation for treatable causes of ALF allows initiation of specific, appropriate therapy (Table 2). However, except for liver transplantation, no single medical intervention has been shown to be beneficial for all ALF patients. Corticosteroids or intravenous prostaglandin E1 infusions failed to decrease morbidity or mortality in randomized controlled trials [55, 81, 82]. N-acetylcysteine (NAC) is of proven benefit in patients with acetaminophen hepatotoxicity [13, 20]. Some physiologic studies suggest that NAC may be beneficial in non-acetaminophen ALF, possibly due to improved tissue oxygenation [83, 84]. Preliminary results from the ALFSG multicenter, double-blind study of NAC in non-acetaminophen ALF recently demonstrated a survival benefit only in patients with grade 1-2 encephalopathy . Therefore, this simple and widely available therapy may be useful in some ALF patients.
An experienced hepatologist, transplant surgeon, and intensivist should work as a team to direct the management of ALF patients. Evaluation for liver transplantation includes obtaining diagnostic serologies, a chest roentgenogram, a bedside echocardiogram, and psychosocial evaluation. Placement of central venous access and arterial lines can provide for fluid resuscitation, infusion of medications, frequent laboratory monitoring, and titration of acid/base status. Routine laboratory tests including serial lactate, factor V, INR, and liver biochemistries should be obtained at least every 8 to 12 hours, while glucose levels should be monitored hourly and supplemented as needed.
Continuous assessment of neurologic status is critical. Classical signs of intracranial hypertension, such as papilledema, loss of pupillary reflexes and clonus, do not reliably correlate with intracranial pressure measurements or grade of encephalopathy (Table 5). Similarly, head CT findings of cerebral edema frequently occur late and are not adequately sensitive or reliable to detect intracranial hypertension (Figure 3) [86, 87]. Moreover, the scanning time and transportation logistics usually preclude use of magnetic resonance imaging in critically ill ALF patients. The pathogenesis of cerebral edema in ALF patients may involve the “glutamine hypothesis”, wherein detoxification of ammonia by astrocytes leads to the conversion of glutamate to glutamine that can increase tissue osmolarity and cause edema . Alternatively, cerebral edema may develop from failure of intracerebral vascular autoregulation with resultant increases in brain water and brain volume, particularly in patients with advanced encephalopathy [89, 90].
Although invasive, intracranial pressure (ICP) monitoring is the most reliable means to monitor changes in intracranial pressure in ALF patients . Information from an ICP monitor helps guide management decisions regarding the use of mannitol and paralytic agents. However, controversy exists whether ICP monitoring should be utilized only in liver transplant candidates or only in patients enrolled in clinical trials versus all patients with grade 3 or 4 encephalopathy. Sedation should be withheld for at least 2 to 4 hours in intubated ALF patients being considered for ICP monitor placement to assess brain function. A preoperative head CT is recommended to exclude spontaneous hemorrhage. Although parenchymal catheters have a greater risk of intracranial bleeding, they provide more reliable pressure readings [91, 92]. ICP measurements can help intensivists maintain an adequate cerebral perfusion pressure (CPP) (i.e. > 50 mm Hg) via the introduction of vasopressors to raise the mean arterial pressure (MAP) or maneuvers to lower the elevated ICP (i.e. CPP = MAP – ICP).
To prevent exacerbation of cerebral edema, the head of the bed should be elevated at >30° from horizontal in all ALF patients . Vigorous suctioning or other Valsalva maneuvers should be avoided to prevent surges in ICP. Prophylactic intravenous lidocaine may be of value [93, 5]. Mechanical ventilation with high levels of positive end expiratory pressure should be avoided. Cooling blankets can be used to keep the patient’s core temperature <37.0°C. Sedative medications, especially long acting benzodiazepines, narcotics, and diphenhydramine, should be avoided in nonintubated patients because they can obscure neurological changes. If sedation is required for patient comfort and safety, agents with a short half-life, such as midazolam or propofol, are preferred.
If a patient clinically deteriorates or has an ICP exceeding 20 mm Hg for more than 5 to 10 minutes, several measures should be undertaken. Initially, hyperventilation of intubated patients to a PCO2 of 28 to 30 mm Hg is recommended to induce cerebral vasoconstriction . Lactulose can help lower systemic ammonia levels in cirrhotic patients via its osmotic activity and acidification of stool, but lactulose has not been prospectively tested in ALF patients. In addition, lactulose raises concerns about free water depletion and potential abdominal distention with associated bowel ischemia. Nonetheless, many centers utilize lactulose, particularly for patients with subacute liver failure who have evidence of portosystemic shunting .
Mannitol 0.5 –1.0 g/kg is a first line therapy for management of ICP surges exceeding 20 mm Hg that do not respond to hyperventilation . Mannitol reduces intracranial volume by drawing fluid into the intravascular space Mannitol infusions should be withheld in patients with renal failure or fluid overload until these problems have been addressed. Monitoring of serum osmolarity is also recommended to avoid a hyperosmolar state. Recent data has suggested that hypertonic saline infusion, with a target serum sodium level of 145 to 155 mmol/l, may reduce the incidence and severity of intracranial hypertension but further studies are needed due to its narrow therapeutic index .
Thiopental and pentobarbital are centrally acting hypnotics that reduce brain oxygen utilization. They represent a second line therapy for severe intracranial hypertension . Pentobarbital, administered as a 100-150 mg bolus over 15 minutes followed by continuous infusion at 1-3 mg/kg/hr, should be monitored to maintain serum drug levels at 20-35 mg/L. Since barbiturate infusions can cause systemic hypotension, dopamine may be required to maintain an adequate CPP. Propofol has been used to reduce ICP. It may be advantageous due to its low risk of systemic hemodynamic effects .
Moderate hypothermia can reduce cerebral hyperemia and decrease intracranial pressure in ALF patients refractory to medical therapy [99-101]. By reducing the core body temperature to 33 or 35°C, cerebral oxygen utilization and blood flow are reduced. Whole body hypothermia can be achieved by external cooling blankets, intravascular cooling devices, and body suits with monitoring of core body temperatures via a rectal or intravascular thermometer. Sedation with a paralytic agent such as atracurium may be needed to prevent reflexive shivering, but propofol or deep sedation may also be effective. The optimal means to safely rewarm hypothermic ALF patients have not been established. Due to the potential risks of hypothermia, including cardiac arrhythmias, worsening coagulopathy, hypotension, and impaired liver regeneration, randomized controlled trials of therapeutic and prophylactic hypothermia with ICP monitoring are needed before this investigational therapy can be routinely recommended.
Seizures in ALF patients may be difficult to detect, particularly in patients receiving deep sedation or barbiturates. In one study of 42 intubated ALF patients, 31% had subclinical seizure activity. The incidence was lower in patients who received prophylactic phenytoin . Some experts recommend continuous or intermittent electroencephalogram (EEG) monitoring for patients with grade 3 or 4 coma, but this practice has not been widely adopted. Hypoglycemia and electrolyte disturbances should be excluded as precipitating factors for seizure development and aggressively treated if detected. Phenytoin, infused as an 18 mg/kg loading dose over 30 minutes followed by 100 mg every 8 hours, is the first-line treatment for seizures, while pentobarbital 3 mg/kg is reserved for refractory seizures .
Infectious complications are common in ALF patients and a leading cause of mortality. Bacterial infections develop in 80% of patients, and fungal infections in 20% to 30% during their hospitalization . Therefore, daily surveillance cultures of blood, urine, and sputum are recommended upon admission to the intensive care unit . A diagnostic paracentesis should be performed on all ALF patients at presentation with ascites, unexplained fever, or leukocytosis. Patients are commonly infected with staphylococcal species, streptococcal species, or gram-negative rods [2, 103]. Coverage with broad spectrum antibiotics should be initiated if the patient develops fever, leukocytosis, or unexplained deterioration in clinical status. A quinolone or third-generation cephalosporin is frequently used. Vancomycin can be added for patients with suspected line infection or further deterioration. Enteral decontamination with poorly absorbed orally administered antibiotics does not appear to alter the outcome of ALF patients who receive parenteral antibiotics. Fluconazole or amphotericin should be added for suspected or proven fungal infection.
Acute renal failure in ALF is usually multifactorial with components of acute tubular necrosis (ATN), hypovolemia, and even hepatorenal syndrome. Renal failure is particularly common in patients with acetaminophen toxicity and portends a poor prognosis . In addition to monitoring of central pressures, a urinalysis and urine electrolytes can help distinguish ATN from hepatorenal or prerenal causes of renal failure. Lactic acidosis is a common complication of ALF that can be worsened by hypovolemia, infection, and poor perfusion pressures . Infusion of normal saline or other colloids may help in the management of hypovolemia. Avoidance of nephrotoxic agents, including aminoglycosides, non-steroidal anti-inflammatory drugs and IV contrast dye, is critical in ALF patients. Enteral feedings are preferred over parenteral nutrition due to the high rate of infectious and metabolic complications with the latter. Hyponatremia is a poor prognostic sign. In addition, serum sodium levels below 125 mmol/L should be avoided because hyponatremia can exacerbate cerebral edema.
If progressive renal failure ensues with oliguria, azotemia or fluid overload, continuous venovenous hemofiltration (CVVH) is preferred over standard hemodialysis due to less dramatic fluid shifts and higher perfusion pressures . Citrate anticoagulation may be preferred to heparin in patients with liver disease, but randomized controlled trials have not been completed.
Acute liver failure is characterized by a hyperdynamic circulation with high cardiac output, low mean arterial pressure (MAP), and low systemic vascular resistance. Following fluid resuscitation, dopamine or norepinephrine may be utilized to maintain an adequate MAP and a CPP > 50 mmHg [2, 88]. Vasopressin and its analogue terlipressin should be avoided because they produce cerebral vasodilation and increased cerebral blood flow leading to worsening intracranial hypertension (107). Placement of a Swan-Ganz catheter is helpful when inotropes or ICP monitors are utilized. Surges in systemic hypertension and bradycardia (Cushing’s reflex) may herald impending uncal herniation. In terminal ALF, patients can become refractory to inotropes and die from circulatory failure. A Tc 99 albumin scan can document the absence of bloodflow in ALF patients with refractory cerebral edema so that these patients can be removed from the liver transplant waiting list.
Serial INR and factor V levels provide useful prognostic information. Routine correction of elevated INR levels with fresh frozen plasma (FFP) is not recommended unless there is evidence of active bleeding or an invasive procedure is planned (Table 6). A therapeutic trial of vitamin K 10 mg subcutaneously for three consecutive days is recommended at presentation because many ALF patients are vitamin K deficient. Prior to an invasive procedure, such as central line placement, ICP monitor or liver biopsy, FFP is infused in an attempt to decrease the INR below 1.5. However, the volume of infused FFP needs to be carefully monitored to avoid exacerbation of fluid overload and cerebral edema. Similarly, the platelet count should be maintained above 50,000 platelets/ml immediately prior to invasive procedures via platelet infusions . Cryoprecipitate can be administered if the fibrinogen level is less than 100 mg/dL. Acid suppression with a proton pump inhibitor is used to prevent upper gastrointestinal bleeding in intubated ALF patients, rather than sucralfate or histamine-2 receptor blockers [2, 109].
Recombinant activated factor VII (rFVIIa) (NovoSeven®, NovoNordisk, Copenhagen, Denmark) has been used for patients with severe coagulopathy prior to invasive procedures, but it is not FDA-approved for this indication [110, 111]. The goal of rFVIIa infusion is to promote localized clot formation in areas of tissue factor release. Due to its cost and risks, most centers reserve rFVIIa infusion for patients with an INR > 1.5 despite infusion of at least 4 units of FFP who require an invasive procedure. Typically a single dose of 80 ug/kg is rapidly infused to enhance clot formation and to normalize the INR for 2 to 12 hours. Contraindications include Budd-Chiari syndrome; known or suspected malignancy; history of deep venous thrombosis, pulmonary embolism, or thrombophilia; pregnancy; and hypersensitivity to vitamin K. The medication should be administered immediately before invasive procedures. Repeating coagulation parameters immediately thereafter is not recommended due to its short half-life. It is unclear whether additional doses or continuous infusions of rFVIIa prevents spontaneous bleeding in ALF patients.
Prior to the widespread availability of liver transplantation, the reported survival of ALF patients was 3% to 18% [2, 112]. Later studies reported survival of 14% to 25% without liver transplantation and 41% to 49% with liver transplantation . Among transplant recipients, 1-year patient survival now varies between 60% and 80% [114, 115]. The severity of encephalopathy and coagulopathy correlate inversely with survival [116, 117]. Numerous prognostic scales have been proposed to identify patients with the greatest need for liver transplantation.
The King’s College criteria were developed from a retrospective cohort of 588 medically managed ALF patients and then prospectively validated in an additional 175 ALF patients . Readily obtained clinical and laboratory parameters were selected to enhance their clinical usefulness for patients with acetaminophen and non-acetaminophen-related ALF. In the acetaminophen cohort, an arterial pH <7.3 or INR >6.5, serum creatinine > 3.4 mg/dl, and grade III or IV encephalopathy had prognostic significance. In the non-acetaminophen cohort, an INR >6.5 or three or more of the following five parameters were independent predictors of poor outcome: unfavorable causes (non-A, non-B hepatitis, DILI), jaundice for >7 days before encephalopathy, age <10 or >40 years, INR >3.5, and serum bilirubin >17.5 mg/dL . The positive predictive value of these criteria for mortality was 84% in the acetaminophen cohort and 98% in the non-acetaminophen cohort, while the negative predictive values was 86% and 82%, respectively . This prognostic model has been tested in other patient cohorts with lower PPV and NPV values [119, 120]. Recently, the value of early arterial lactic acid levels in conjunction with the standard King’s College criteria have been studied in patients with acetaminophen-induced ALF. A post-resuscitation arterial lactate level >3.0 mmol/L and an “early” level >3.5 mmol/L had a negative predictive value of 97% and 99%, respectively, but had a positive predictive value of only 79% and 74%, respectively .
The Model for End Stage Liver Disease (MELD), consisting of serum creatinine, total bilirubin and INR, was shown to predict 3-month survival better than the Child-Turcotte Pugh score in liver transplant candidates with cirrhosis . The MELD score has become the means by which liver allografts are allocated to cirrhotic patients in the United States. In non-acetaminophen ALF patients, MELD scores identify patients with the worst prognosis, while other patients had a high rate of spontaneous recovery that was independent of MELD score .
Global assessment models have been proposed to predict outcome . Larson et al. recently showed that admission APACHE II scores were superior to the King’s College criteria or MELD scores in predicting outcomes in patients with acetaminophen-induced ALF . Other potential prognostic laboratory markers include serum phosphate levels,l which decline in patients with rapid hepatic regeneration . Similarly, serum alpha-fetoprotein levels increase in patients undergoing rapid liver regeneration and increasing levels indicate a better prognosis [125, 126]. However, these biochemical parameters likely have inadequate predictive power by themselves, and may be clinically useful in combination with other prognostic variables. Abdominal CT scanning, to assess liver volume, and liver biopsy, to assess hepatic histopathology, have also been proposed but both methods have limited sensitivity, specificity, and feasibility [127, 128]. Since etiology is an important, consistent predictor of outcome, disease-specific prognostic models may prove useful for non-acetaminophen as well as acetaminophen-related ALF .
Emergency liver transplantation is the only intervention with known survival benefit in ALF patients with a poor prognosis . Outcome after liver transplantation is closely linked to the severity of the pre-transplant illness and the nature of the graft utilized. Currently, the 1-year survival of patients transplanted for ALF is lower than that of patients transplanted for chronic liver failure (70% vs. 85%) probably due to the emergent nature of the surgery, concomitant organ failure, and higher incidence of immunologically-mediated graft dysfunction.
Rapid medical and surgical evaluation is required for all ALF transplant candidates prior to listing to exclude significant cardiopulmonary disease, malignancy, or other conditions which negatively impact patient outcomes . In addition, a comprehensive psychosocial evaluation of patient compliance, family support, and substance abuse is extremely important in patients with acetaminophen overdose. Ongoing evaluation of the need and suitability of listed ALF patients for transplantation is necessary due to the unstable nature of this patient population and the potential development of contraindications. The frequent delay in securing a suitable donor liver renders medical decisions complex and difficult. Most centers consider refractory systemic hypotension or intracranial hypertension, uncontrolled sepsis, or progressive multiorgan failure as contraindications to transplantation.
To facilitate rapid distribution of livers for life-saving transplantation, UNOS developed a special Status 1designation for patients with a high short-term risk of death. Patients eligible for status 1 designation include ALF patients with illness onset in the prior 8 weeks, fulminant Wilson’s disease patients, and transplant recipients with primary graft non-function (PNF) or early hepatic artery thromboses (HAT). Status 1 patients move ahead of all other listed patients with chronic liver failure. Grafts are allocated to status 1 patients based upon blood type, geography, and waiting time [122, 129]. In August 2005, the Status 1 category was modified to include specific clinical and laboratory criteria for fulminant hepatic failure (FHF), HAT, and PNF. Donor livers are initially offered to status 1A adults or children, while a Status 1B category was developed for pediatric patients with chronic liver disease requiring intensive care, non-metastatic hepatoblastoma, or metabolic disease. In calendar years 2004 and 2005, 1,529 patients were listed as UNOS Status 1 . Outcomes 15 days after listing included 54.2% of patients underwent transplantation, 16.1% were dying or too sick to transplant, 11.6% had recovered, 8.7% were still listed for transplantation, and 9.1% were de-listed. Extrapolating these data to the overall population, less than 10% of the 2,800 ALF patients receive a liver transplant each year in the United States.
Artificial and bioartificial liver support devices are under development for patients with acute and acute on chronic liver failure. These devices may be ideally suited for patients with ALF to bridge these patients to spontaneous recovery during native liver regeneration. However, clinical trial design is difficult due to variable spontaneous recovery rates and variable availability of liver transplantation. The ideal liver replacement device should perform normal hepatocyte functions including detoxification, metabolism, and synthesis of critical proteins. Early attempts at artificial liver detoxification included hemodialysis, hemofiltration, exchange transfusion, plasma exchange, and resin hemoperfusion, but none of these interventions improved the outcome [131, 132]. Newer artificial detoxification devices, such as the Molecular Absorbent Recirculating System (MARS), utilize charcoal or other adherent particles in an extracorporeal circuit [132, 133]. However, these artificial devices only provide filtration function. The need for arterial and venous cannulation, anticoagulation, and extracorporeal perfusion can potentially cause complications.
Bioartificial liver support devices utilize human or other mammalian-derived hepatocytes in an extracorporeal circuit . These systems can theoretically synthesize proteins and metabolize xenobiotics in addition to performing filtration and detoxification. However, maintaining viable, sterile hepatocytes for continuous extracorporeal use is a formidable challenge. Concerns have been raised regarding transmission of hepatocytes to the host, transmission of zoonoses, activation of the clotting cascade, and immunological reactions with development of xenoantibodies . In the largest randomized, controlled trial, 171 patients with fulminant or subfulminant liver failure or with PNF following liver transplantation were randomized to receive a daily 6 hour treatment with the HepatAssist device, a device that contains 100 grams of porcine hepatocytes loaded in a dialysis cartridge in series with charcoal filters, versus standard care . Overall, 30-day survival was similar in both treatment groups (71% HepatAssist versus 62% control, p=0.26). This landmark study highlights the difficulties of performing clinical trials in ALF and the need for appropriate patient inclusion criteria and clinically relevant endpoints. Further refinements of device components and perfusion circuitry, and an improved understanding of liver regeneration are needed to improve these devices for use in ALF patients.
Acute liver failure (ALF) remains a dramatic and highly unpredictable clinical syndrome. Studies of its etiologies and natural history are hampered by its low incidence, variable terminology, and variable clinical management. In the United States, acetaminophen is the leading cause of ALF and the incidence of non-intentional acetaminophen overdose appears to be increasing. ALF is a clinical syndrome of coagulopathy and encephalopathy ensuing from a multitude of infectious, immunological, vascular, infiltrative, and metabolic diseases (Table 2). Proven treatments for specific causes of ALF include NAC for acetaminophen overdose and delivery in pregnancy related ALF. In addition, antivirals are frequently recommended for HBV and HSV-related ALF due to the generally safe medication profile. Intravenous NAC for non-acetaminophen-related ALF appears promising, but corticosteroids should not be used in indeterminate ALF or DILI based upon the available data. Cerebral edema is a hallmark of ALF that requires specialized management by a group of experienced intensivists, hepatologists, and transplant surgeons. A low threshold for broad spectrum antibiotics is recommended in ALF patients at high risk of bacterial and fungal infections. Subjects with advanced encephalopathy or an otherwise unfavorable prognosis should be promptly referred to a liver transplant center for further evaluation. Emergency liver transplantation can be associated with favorable outcomes, but requires coordinated intensive care and constant reassessment.
RJF was supported in part by NIH grant (UO1 DK058389-07) as a participant in the Acute Liver Failure Study Group. The author would like to acknowledge the contributions and mentorship provided by Dr. William Lee of the University of Texas Southwestern who is the principal investigator of the Acute Liver Failure Study Group.