Hepatic failure presents with a variety of acute manifestations, most of which will be seen at some point during the ED clinician's career. At the most critical end of the spectrum is the syndrome of acute liver failure (ALF), in which hepatic function is suddenly lost in a person with no previous liver dysfunction. Currently, the most widely accepted definition of ALF is the presence of both coagulopathy (international normalized ratio [INR] > 1.5) and altered mental status consistent with hepatic encephalopathy of less than 26 weeks' duration (a change from the previously defined timeline of 8 weeks).1,2 This disorder is rare, with an annual incidence in the US of 2300 to 2800 cases, and results in 0.1% of all deaths and 6% of all liver-related deaths in the US.3,4 However, this condition can deteriorate rapidly and is associated with high morbidity and mortality. Precipitating etiologies must be quickly identified and disease-specific interventions implemented in order to prevent further decompensation or death. Early transfer to a tertiary care facility with transplant capability may also be necessary.
More common than ALF is chronic liver failure (CLF) with cirrhosis, the 12th leading cause of death in the US.5 While primarily a disorder of long-term outpatient management, acute decompensation of CLF may bring the patient to the ED because of variceal hemorrhage, symptomatic ascites, spontaneous bacterial peritonitis, hepatorenal or hepatopulmonary syndrome, and hepatic encephalopathy. The ED clinician must confront these manifestations and guide management within the broader context of the patient's chronic care.
This issue of Emergency Medicine Practice focuses on the management of ALF and the acutely symptomatic cirrhotic patient.
Click here to download a PDF of the Evidence-Based Practice Recommendations for this issue.
On a typical swing shift, your ED is full of patients with a chief complaint listed as "abdominal pain." You go in to see one such patient, a 40-year-old male, after his nurse tells you he "does not look very good." He states that he has had gradually worsening abdominal pain for over a week, and that today he felt so fatigued he was unable to go to work. Before this, he was otherwise healthy, and no one around him has been sick. On further history, he reveals that 6 weeks ago he was started on isoniazid for a positive PPD (tuberculosis) test in the absence of active chest disease. Other than mild tachycardia (to 112 beats per minute), his vital signs are normal. Physical examination reveals that his sclerae are yellow and detects marked tenderness of the right upper quadrant with mild hepatomegaly. His wife tells you she is concerned because he seems confused.
You are interrupted by a call to the resuscitation bay to evaluate a patient whose blood pressure is 78/37 mm Hg. You recognize him because of his frequent visits to the ED for tense ascites requiring large-volume paracenteses. EMS reports that twice during transit he vomited dark-red blood. He is pale and diaphoretic. This patient is obviously critically ill, and it is clear your next steps will determine his outcome.
Your 40-year-old male patient on isoniazid has developed ALF from an idiosyncratic reaction to the drug — a diagnosis confirmed by his INR of 2.0 and elevated AST, ALT, and bilirubin. He has grade 1 encephalopathy. After moving him to a high-visibility bed in the ED, starting IV fluids via peripheral IV, and evaluating his glucose and electrolyte levels, you contact the transplant center located an hour away and arrange for an advanced life support ambulance to transport the patient.
Your CLF patient has a massive GI bleed, most likely due to varices. After establishing large-bore IV access, administering an initial bolus of crystalloid, and calling for blood, you start treatment with octreotide and ceftriaxone. You arrange for urgent endoscopy and admit the patient to the ICU with an improved blood pressure.
A literature search was performed using the following databases: Ovid MEDLINE® (www.ovid.com) and PubMed (www.pubmed.gov) the Cochrane Database of Systematic Reviews, the National Guideline Clearinghouse, the Agency for Health Care Research and Quality Clinical Guidelines and Evidence Reports, and EBM Online/Evidence-Based Medicine. Searches were limited to the English language and to studies involving adult human subjects. In addition, we examined selected studies drawn from the bibliographies found in the literature. The search yielded many review articles and descriptive studies but few randomized, controlled trials regarding acute hepatic failure. For this reason, the American Association for the Study of Liver Diseases (AASLD), which publishes practice guidelines for many specific hepatologic disorders, presented its recommendations for ALF in the form of a position paper expressly stating that the available data are not sufficient to support a formal practice guideline.6 Significantly more data are available regarding the complications of chronic cirrhosis. Some nationally published guidelines regarding acute hepatic injury are available. Table 1 lists selected guidelines for hepatic injury relevant to the ED setting.
ALF is characterized by severe dysfunction of damaged hepatocytes, which renders these cells less capable of carrying out their synthetic and degradative roles. These effects are manifest as coagulopathy and hypoalbuminemia, as well as the accumulation of harmful metabolites.
The disparate causes of ALF have crucial implications for prognosis as well as intervention. In recent decades, acetaminophen toxicity and idiosyncratic drug reactions have replaced viral hepatitis as the most common causes of ALF in the US.9 When viral hepatitis is responsible, the overwhelming culprits are the hepatitis A and B viruses; hepatitis D rarely causes ALF in patients infected with hepatitis B, and hepatitis E is more likely in developing countries and among pregnant women. Hepatitis C rarely, if ever, causes ALF; although it was not identified in a prospective study of viral hepatitis- related ALF in the US,10 it has been implicated in a small series of studies from Asian countries.11,12Other causes of ALF include inborn errors of metabolism, ischemic insult, Wilson disease,13 malignancy,14 other viruses (herpes simplex, Epstein-Barr, paramyxovirus), nonpharmacologic toxins such as Amanita (mushroom) poisoning15; rarely, acute fatty liver of pregnancy,16 Budd-Chiari syndrome,17 connective tissue disorders,18 and autoimmune hepatitis are responsible for ALF.19 Finally, there have been case reports of ALF occurring secondary to cardiac trauma and seizures.20,21 In approximately 15% of adult cases, no cause is ever found. Figure 1 shows the most common causes of ALF based on a study of 1147 adults by the US ALF Study Group from 1998 to 2007.13
The pathophysiology of ALF clearly varies with the cause; however, in drug-induced hepatic injury the pathophysiology has significant implications for management. Drugs can harm the liver in either a dose-related or a nondose-related (idiosyncratic) fashion, the latter being characterized by latency in effect and rarity of occurrence. Idiosyncratic drug reactions typically occur within 6 months after treatment has begun and rarely occur beyond a year of continuous use.
The prototypical dose-related hepatotoxin is acetaminophen (also known as paracetamol, APAP). When taken in amounts that overwhelm the safe metabolic pathways of glucuronidation and sulfation, acetaminophen is metabolized to the toxic compound N-acetyl-p-benzoquinone-imine (NAPQI) via the cytochrome P-450 system. Drugs and other toxins can also induce hepatic enzymes, leading to secondary hepatotoxicity from other substances. The interaction between ethanol and acetaminophen is the prototypical example of this. Acute ingestion of ethanol with acetaminophen can protect against its toxicity, since ethanol competes for the enzymatic activity of cytochrome P-450 (CYP2E1 isoenzyme), making this pathway less available for the metabolism of acetaminophen. However, ethanol also induces CYP2E1 activity and prevents the degradation of this isoform, thus increasing its availability to metabolize acetaminophen once ethanol is no longer present. The chronic use of ethanol predisposes to acetaminophen toxicity, a fact that has implications for the safety of acetaminophen in alcohol abusers. In a case series of patients who developed hepatic injury from acetaminophen taken for therapeutic relief rather than intentional ingestion, 64% were considered alcoholic. The majority of acetaminophen doses in this group were considered "nontoxic" for the average adult.22
Idiosyncratic reactions to drugs can harm the liver through a variety of mechanisms, including their covalent binding to intracellular proteins, provocation of an autoimmune response, damage to mitochondria, activation of apoptosis, and general circulatory collapse.23 Based on the general mechanism, the pattern of injury with particular drugs may be predominantly hepatocellular, cholestatic, or mixed, which will alter the pattern of abnormality seen on laboratory testing.24 See Table 2 for a list of examples of hepatotoxic compounds.
Over-the-counter herbal supplements have also been associated with hepatotoxicity.25 ALF from drug-related hepatotoxicity is associated with decreased survival without transplantation, so early recognition is critical.
The unifying pathophysiology underlying most complications of chronic liver disease is cirrhosis, the development of fibrous tissue and regenerative nodules in place of normally functioning hepatic tissue. Venous flow into the liver decreases as a consequence of this architectural change, leading to elevated portal pressures. Portal hypertension then leads to splenomegaly, which causes anemia and thrombocytopenia via increased sequestration and destruction of red blood cells and platelets.
Elevated hydrostatic pressure within the portal vasculature causes ascites via the extravasation of extracellular fluid from the portal system into the peritoneal space down a pressure gradient. Hypoalbuminemia from decreased hepatic synthetic capacity lowers the oncotic pressure in the vasculature, worsening fluid egress. Accumulated peritoneal fluid can in turn exert pressure on intra-abdominal organs such as the kidney, decreasing the glomerular filtration rate, which further lowers the excretion of sodium and water via activation of the renin-angiotensin- aldosterone feedback loop and anti-diuretic hormone, which in turn leads to exacerbation of ascites via fluid overload. The resulting intraperitoneal fluid causes uncomfortable abdominal distension.
Hepatorenal syndrome (HRS) occurs when acute renal failure develops in the setting of liver failure in a patient with otherwise normal kidneys. Although this syndrome is more common in patients with advanced cirrhosis, it can also occur in acute hepatic failure. It is caused by extreme renal artery vasoconstriction in the setting of splanchnic vasodilatation, a low effective plasma volume, and insufficient cardiac output. The following criteria developed by the International Ascites Club26,27 define hepatorenal syndrome:
These criteria have been updated to include the new recommendation to use albumin instead of saline for volume expansion and to broaden the criteria to include acutely infected patients without evidence of shock.
Type I hepatorenal syndrome is defined as the rapid loss of renal function over less than 2 weeks (doubling of initial creatinine to Cr > 2.5 mg/dL), whereas type II does not involve such a sudden change. Type I is often provoked by an acute insult such as spontaneous bacterial peritonitis (SBP) and is associated with an extremely high mortality, with 1 series reporting a median survival of 3.3 months.28
Tense ascites can extravasate across the diaphragm to cause pleural effusions that subsequently compress the lungs, causing dyspnea and hypoxia. However, true hepatopulmonary syndrome is a disorder of pulmonary vascular dilatation (precapillary and capillary) and shunting thought to be secondary to vasoactive factors such as nitric oxide.29 This dilatation causes severe arterial hypoxemia through a combination of ventilation-perfusion mismatch, shunting, and impaired diffusion. Diagnostic criteria therefore include the triad of chronic liver disease, arterial deoxygenation, and intrapulmonary vasodilatation, with severity of this syndrome graded according to the alveolar-arterial oxygen gradient or partial pressure of oxygen. Although there is no strict correlation between the degree of liver disease and the presence of this syndrome, it is associated with significantly increased mortality for each stage of liver disease. Overall, it carries a grave prognosis, with a median survival as low as 11 months in 1 cohort,30 and an overall 5-year survival rate of 23% without transplantation.31 For these reasons, the AASLD recommends expedited referral and evaluation for transplantation for patients with hepatopulmonary syndrome32 Orthotopic liver transplantation is the only successful treatment available.
Elevated portal pressures also lead to the development of portosystemic collaterals such as varices, which are dilated vessels within the esophagus, stomach, and colon. The most common cause of death associated with cirrhosis is rupture of gastroesophageal varices and ensuing hemorrhage. The risk of rupture is proportional to variceal wall tension, which is most directly related to the size of the varix.
Hepatic encephalopathy occurs with both acute and chronic liver failure but is more common and rapidly progressive in ALF. It results from hepatocellular damage, which prevents the liver from metabolizing waste products in the blood, including nitrogenous compounds such as ammonia. Although the exact pathophysiology is unknown, it is postulated that these compounds may be structurally similar to neurotransmitters, causing dysfunctional neural transmission with associated altered mental status and motor abnormalities such as asterixis, the repetitive clonic flexion of the wrist when maintained in hyperextension. Table 3 shows the grades of hepatic encephalopathy.
Cerebral edema with elevated intracranial pressure is associated with hepatic encephalopathy (with increasing frequency at higher grades) and is a common cause of death.34 Some evidence suggests that ammonia levels do not correlate with degree of hepatic encephalopathy in general,35 although cerebral edema has been induced by ammonia infusion in a rat model,36 and in 1 human study ammonia levels greater than 200 μg/dL were associated with cerebral herniation.37
The differential diagnosis for ALF should include other potential causes of altered mental status as well as other etiologies of abdominal pain, jaundice, or a clinically evident bleeding diathesis. (See Table 4.) The morbidity of altered mental status varies depending on the cause; some otherwise benign etiologies, such as alcohol intoxication, can be lifethreatening when encountered in the extreme (eg, combined alcohol and benzodiazepine ingestion with depressed respiratory drive). In patients with acute upper gastrointestinal bleeding, alternative diagnoses must be considered if there is no known history of cirrhosis. (See Table 5.) The ED clinician must rapidly consider life-threatening causes of such presentations while simultaneously initiating stabilizing measures.
For most patients with hepatic failure, prehospital care requirements are minimal. For the patient with upper gastrointestinal (GI) bleeding, prehospital care goals include the establishment of large-bore intravenous (IV) access, careful monitoring of the airway in the event of further emesis, and contact with base control if vital signs are unstable so that arrangements with the blood bank can be made in advance of the patient's arrival. Universal precautions should be strictly observed in every case, but the likelihood of viral hepatitis in patients with liver failure further emphasizes the importance of personal protective equipment.
Depending on the presenting complaint, patients with liver failure may be clinically stable but can deteriorate quickly. When appraising vital signs, the ED clinician should pay close attention to the patient's medication regimen, since many patients with cirrhosis will be taking nonselective betablockers that may blunt the normal tachycardic response to bleeding. Patients with a decreased level of consciousness should be evaluated for intubation. Large-bore IV access should be established if it is not already in place, with preference given to peripheral access both to allow rapid resuscitation and to avoid the morbidity associated with central access in patients with potential coagulopathy. If necessary, central access should be established at compressible sites under ultrasound guidance if acuity permits. Patients who report bloody or dark-colored emesis, melena, hematochezia, or altered mental status should be triaged to high-visibility beds with cardiac monitoring capability. Almost all complaints related to liver failure — abdominal pain, nausea or vomiting, bleeding, distention, shortness of breath, altered mental status, jaundice — require laboratory evaluation, so these patients are not candidates for provider- in-triage or other fast-track disposition.
Because acute and chronic liver failure differ significantly in etiologies, complications, and morbidity rates, the approach to the patient suspected of having hepatic failure must focus initially on determining the timing of symptom onset.
If the timing is consistent with ALF, questions should pertain to all the medications the patient is currently taking, including prescription drugs and over-the-counter medications. Particular attention should be paid to the ingestion of acetaminophen to establish quantity and timing, as well as herbal supplements, mushrooms, and plant or tea extracts. In addition, the ED clinician should elicit any history of alcohol and recreational drug use (especially if administered intravenously or by skin popping), recent travel, sexual exposures, needlestick exposures, blood transfusions, exposure to industrial compounds, recent anesthesia or illness that required hospitalization, and any family history of liver disease or autoimmune disease. Information from collateral sources such as family and caregivers should be sought regarding any subtle changes in the patient's mental status consistent with earlystage encephalopathy. A history of light-colored stools or dark urine can be clues to cholestasis in the absence of overt jaundice.
If the onset of symptoms is more consistent with CLF or a known diagnosis of cirrhosis, assessments should include the patient's current management regimen (medications and degree of compliance, endoscopic assessment showing the presence or absence of varices, frequency of follow-up visits), as well as the presence of anorexia and weight loss or pruritus (due to bilirubin deposition), and any history of paracentesis, previous SBP, or renal dysfunction. Further history will be directed at the chief complaint: a complaint of worsening ascites should prompt the ED clinician to ask questions about associated symptoms such as shortness of breath, decreased exercise tolerance, abdominal pain, and fever/chills, as well as potential causes of decompensation such as dietary indiscretion.
Physical examination should begin with an assessment of the patient's vital signs, mental status, and perfusion. In the grossly obtunded patient, the ED clinician should look for signs of impending herniation due to cerebral edema, such as loss of pupillary reflexes or the presence of Cushing's reflex (hypertension with bradycardia and abnormal respirations). Patients with liver failure are susceptible both to intravascular depletion with peripheral vasoconstriction (bleeding, dehydration from vomiting) and to sepsis with distributive physiology and vasodilatation (SBP).
In the stable patient, consider testing for lowgrade encephalopathy with the Quick Confusion Scale, a 6-item, 15-point instrument designed for the rapid assessment of mental status. This scale has been shown to correlate fairly well with results on the Mini-Mental State Examination (MMSE) but requires less time to administer.38
To test for asterixis, ask the patient to raise both arms and hyperextend the wrists. Jaundice may be subtle and is best detected by examining the sclerae and the mucosa under the tongue. In patients with acute hepatitis, the abdominal examination will often reveal tender hepatomegaly, whereas patients with cirrhosis may have splenomegaly but will not have a palpable liver. Tense ascites is typically self-evident, but less overt ascites can be difficult to diagnose on physical examination alone, with a sensitivity between 50% and 90% reported in 1 study.39Because the population is increasingly overweight, it is becoming more difficult to detect ascites. Point-ofcare abdominal ultrasound to detect free fluid can aid in the diagnosis of ascites. Table 6 lists several other findings in CLF.
Virtually all patients with complaints related to liver failure will require laboratory testing. The traditional "liver function tests" are really a combination of tests that reflect hepatocellular injury and cholestasis. The goal of basic testing is to establish the degree of hepatocyte injury and impairment of synthetic function. Other tests are directed toward determining the causal agent of the liver injury, predicting further clinical decline, and detecting infection. The Clinical Pathway For Evaluation Of Patients With Suspected Acute Hepatic Injury presents recommended approaches to laboratory testing, and Table 7 offers interpretations of abnormal results on liver chemistry studies.
Aminotransferase levels and the AST:ALT ratio can also help determine the etiology of ALF. Levels of both AST and ALT may be markedly elevated in acute ischemic injury but improve quickly once the patient's circulatory status has been stabilized. AST may reach extremely high levels (up to 48,000 U/L) in acetaminophen overdose, with 90% of patients with acetaminophen toxicity having values greater than 3000, according to one study.22 An AST level of 3000 was highly associated with acetaminophen injury but not with alcoholic or acute viral hepatitis, although descriptive statistics were not reported. In cases of toxic hepatic injury (including acetaminophen overdose), AST and ALT levels typically decline quickly, with AST declining more rapidly because of its shorter half-life (17 hours vs 47 hours for ALT).40In cases of ALF related to viral hepatitis, an AST:ALT ratio of 0.6 or less was associated with spontaneous survival, especially for hepatitis A patients, though the included numbers were small and the observation empirical. The authors posit that this ratio may reflect a higher ALT level representing more hepatic tissue at baseline with improved reserved capacity, but this observation remains to be explored.10
Other laboratory tests that may assist in the work-up of patients with encephalopathy include arterial blood gas, toxicology screen, ethyl alcohol level, pregnancy test in females, and HIV antibody.
The prognosis for patients with ALF varies depending on the cause. Acetaminophen overdose, HAV infection, "shock liver," and pregnancy-related ALF are associated with the best spontaneous (non-transplanted) survival rates; patients with ALF due to Wilson disease, HBV infection, autoimmune hepatitis, Budd-Chiari syndrome, and malignancy fare worse.10,14 The U.S. Acute Liver Failure Study Group has enrolled 1147 patients at 23 centers. Among all patients in their database, 45% recovered spontaneously, 44% were listed for transplantation, and 25% received a transplant; 10% of the total group and 50% of those listed for transplantation died while awaiting a donor organ. Overall mortality was 30%, compared with over 80% in the era prior to transplantation.14 The most common causes of death in this group were cerebral edema and sepsis, followed by multiorgan system failure, cardiac arrhythmia/ cardiac arrest, and respiratory failure. Coagulopathy in itself is not a frequent cause of death.
Several scoring systems have been developed to assess prognosis in liver failure. The Child-Turcotte- Pugh (CTP) system (see Table 9) was originally developed to assess risk associated with portocaval shunt surgery in variceal hemorrhage, but it is now widely used to assess prognosis in CLF.56,57Point values are associated with different laboratory values as well as the degree of encephalopathy. CTP scores of 5 to 6 result in a Class A diagnosis, 7 to 9 Class B, and 10 to 15 Class C. Mortality increases from class A to C.
More recently, the Model for End-Stage Liver Disease (MELD) score was developed as an adaptation of a scoring system to predict outcome in patients undergoing a transjugular intrahepatic portosystemic shunt (TIPS) procedure. It was subsequently validated in different populations of patients with chronic hepatic failure58 and is now applied to patients with acute hepatic failure as well, although it has not been validated in this population. The MELD score uses the INR, serum creatinine, and total bilirubin to calculate its numerical score, but the actual equation involves logarithmic calculations and is therefore less user-friendly than the CTP system. However, several websites have online calculators available, such as the United Network For Organ Sharing official website (www.unos.org).59,60
In contrast with the above scales for CLF, the King's College Criteria (KCC) were developed to stratify risk in ALF. Several variables were found to be predictive of outcome and were then synthesized into criteria for transplantation.45 For acetaminophen toxicity, these criteria were arterial pH less than 7.3 or the combination of PT longer than 100 seconds (~INR 6.5), serum creatinine greater than 3.5 mg/dL, and grade 3 or 4 encephalopathy. For ALF not caused by acetaminophen, transplantation was recommended for anyone with a PT longer than 100 seconds (~INR 6.5) or any 3 of the following findings: PT longer than 50 seconds (~INR 3.5), serum bilirubin greater than 17.5 mg/ dL, cryptogenic or drug-induced liver failure, age less than 10 or over 40, or jaundice for more than 7 days before the onset of encephalopathy. However, in a more recent prospective trial, these latter 2 criteria were not found to be predictive.9 In a meta-analysis of ALF due to acetaminophen, an arterial pH less than 7.3 was just as specific as the KCC, although neither was particularly sensitive and would therefore miss patients at risk for a poor outcome without transplantation.61 Based on these data, the AASLD does not recommend that the ED clinician rely on these guidelines alone when determining a patient's risk.
ALF often progresses to multiorgan system involvement. Numerous therapies designed to maximize supportive care and prevent the previously discussed causes of mortality in ALF have been studied. Many of these interventions may also be applied to care of the critically ill patient with end-stage CLF.
As discussed earlier, hyperammonemia is thought to play a role in hepatic encephalopathy, which in turn is related to cerebral edema, intracranial hypertension, and hypoxic brain injury. Nonabsorbable disaccharides such as lactulose are frequently administered to treat hepatic encephalopathy. In one retrospective case-control study by the U.S. Acute Liver Failure Study Group, lactulose therapy was associated with increased survival time but not with improvement in neurologic status or long-term outcome; this report suggested that lactulose may offer a bridge to transplantation.62 In contrast, a recent Cochrane review concluded that nonabsorbable disaccharides (lactulose and lactitol) have no effect on mortality but may have an effect on encephalopathy. It also suggests that antibiotics such as neomycin might be superior to disaccharide treatment, although more data are required to confirm this suggestion.63 A Cochrane meta-analysis of antibiotic treatment for hepatic encephalopathy is now under way. Data have failed to support treatment with branched-chain amino acids or dopaminergic agonists.64,65 A transient improvement in degree of encephalopathy may be seen with flumazenil, but this treatment had no effect on ultimate recovery or survival; a Cochrane review concluded that more data are needed to recommend its routine use.66
In contrast with normal seizure management in the ED, the AASLD recommends that seizures be controlled with phenytoin primarily to prevent benzodiazepine oversedation due to decreased hepatic metabolism. However, a randomized, controlled trial of patients with ALF found that prophylactic phenytoin did not prevent seizures, cerebral edema, or the requirement for mechanical ventilation, nor did it improve survival.67 Newer agents that undergo predominantly nonhepatic metabolism, such as levotiracetam, have not yet been studied.
Intracranial hypertension (as evident from neurologic signs) should be managed by elevating the head of the bed to 30° and a mannitol bolus (0.5 to 1.0 g/kg); its use has been shown to decrease ICP and improve survival in ALF patients.68 Short-term hyperventilation may also be useful in critical situations to prevent acute herniation. One study demonstrated a positive effect on restoring cerebral autoregulation (survival or neurologic outcomes not evaluated),69but another trial failed to show survival benefit with continuous hyperventilation.70 Prophylactic hyperventilation is not recommended. The induction of hypernatremia (sodium 145-155 mmol/L) by the administration of 30% hypertonic saline may prevent the development of intracranial hypertension.71Several small studies have shown that hypothermia to 32° to 33ºC (89.6° to 91.4°F) will reduce intracranial pressure and can be used safely as a bridge until orthotopic liver transplantation can be performed72,73; however, no randomized, controlled trials have been carried out to evaluate this intervention. Corticosteroid treatment is not helpful.68
Despite the often profound coagulopathy associated with ALF, a well-designed randomized, controlled trial has shown that prophylactic normalization of the INR is not necessary unless a procedure other than paracentesis is planned.74 The AASLD supports the use of vitamin K administration, although this therapy has not been specifically studied. Thrombocytopenia also occurs in patients with ALF. Data from the oncology literature support a platelet transfusion threshold of 10,000/μL for asymptomatic patients75,76and 50,000 to 70,000/μL for invasive procedures. Patients with active hemorrhage should receive platelets and fresh frozen plasma by transfusion. The use of recombinant factor VIIa in conjunction with fresh frozen plasma may be more effective in reversing coagulopathy77 but requires further study.
Patients with liver damage are inherently prone to hypoglycemia and should be monitored frequently for this complication. Electrolyte derangement secondary to fluid shifts and multiorgan dysfunction is also common and should be monitored by means of serial laboratory analysis.
Acetaminophen overdose is now the most common cause of ALF in the US, as had previously been the case in Europe. It is the only truly reversible etiology of ALF, and it is the job of the ED clinician to evaluate the patient for possible acetaminophen toxicity and, if it is present, to begin therapy with N-acetylcysteine (NAC) as soon as possible. Table 10 presents the American College of Emergency Physicians (ACEP) evidence-based clinical policy recommendations for treating patients with acetaminophen overdose.78
The decision to administer activated charcoal must be individualized based on likely time of ingestion and mental status; however, according to one randomized, controlled trial, administering activated charcoal just before treatment with NAC does not reduce its efficacy.79 When the time and amount of ingestion are known, the indication for NAC can be determined using the Rumack-Matthew Nomogram.50 However, this nomogram does not address ingestions that occurred longer than 24 hours before presentation, potential toxicity from chronic overdose, or therapeutic misadventure in the alcoholic or malnourished patient.
Although the efficacy of NAC (when it is administered within 8 hours of an acute ingestion) is well supported,80 data also support its effect when given more than 10 hours after overdose (median delay 17 hours). In 1 retrospective study of 100 cases of acetaminophen-induced ALF, mortality was significantly reduced and progression to grade 3 or 4 encephalopathy decreased among patients who had been given NAC compared with those who were not.81 NAC may be administered orally or intravenously depending on the patient's mental status and likelihood of compliance; there is no difference in efficacy between these 2 routes of administration.82
Based on the safety of NAC and its high rate of efficacy, AASLD guidelines recommend that NAC be given to all patients for whom the quantity of acetaminophen ingested, the serum acetaminophen level, or a rise in aminotransferase levels indicates progressive liver injury and to those for whom acetaminophen ingestion is a possibility or the circumstances of ingestion are unknown. While the qualifying quantity of acetaminophen ingested is not specified in this statement, a practice guideline by the American Association of Poison Control Centers for out-of-hospital management of nonsuicidal ingestion states that adults should be referred to the ED if there is an "acute, single, unintentional ingestion" of more than 10g or 200 mg/kg (whichever is lower) or when the amount is unknown. Patients with "repeated supratherapeutic ingestion (RSTI) of acetaminophen" should be referred to the ED if they have ingested more than 10g or 200 mg/kg (whichever is lower) over a 24-hour period OR if they have ingested more than 6g or 150 mg/kg (whichever is lower) per 24-hour period during the preceding 48 hours or more. The threshold for patients at risk for acetaminophen hepatotoxicity at baseline (such as alcoholism) is more than 4 g or 100 mg/kg (whichever is lower) per day.83 The Poison Control Center should be contacted with questions regarding treatment in specific cases.
Ascites is the most common major complication of cirrhosis.86 The ED clinician most frequently encounters symptomatic or refractory ascites in the patient known to have cirrhosis; other causes include rightsided heart failure, nephrotic syndrome, pancreatitis, and malignancy.
To differentiate between causes of ascites and to treat symptomatic ascites, paracentesis should be performed. Indications for paracentesis in patients with cirrhosis include abdominal pain and shortness of breath from presumed compression. Paracentesis is a safe procedure with a low rate of serious complications (< 0.1%).87 There is no absolute level of coagulopathy (INR or platelet level) beyond which paracentesis is contraindicated, nor are there data demonstrating benefit from the prophylactic transfusion of blood products prior to paracentesis; the AASLD does not recommend this practice.7 The serum-ascites albumin gradient (SAAG) should be calculated from the fluid analysis. A SAAG of 1.1 g/dL or higher supports a diagnosis of portal hypertension with 97% accuracy.88 More-specialized tests, such as cytologic analysis, should be reserved for cases in which a nonhepatic etiology is likely. Large-volume paracentesis for tense ascites may lead to the syndrome of paracentesis-induced circulatory dysfunction (PICD), a largely silent activation of the renin-angiotensinaldosterone system that nonetheless is associated with renal dysfunction, earlier recurrence of ascites, and higher mortality.89 PICD may be prevented with plasma volume expansion; a recent trial demonstrated no increased benefit from albumin administration over that of normal saline in the prevention of PICD if the total volume of ascitic fluid removed was less than 6 L.90 Frank hypotension is unusual unless the total volume of fluid removed is very high; 1 small study demonstrated no hemodynamic disturbances after a 5-L paracentesis,91 whereas a total paracentesis protocol with an average volume of 8.2 L removed resulted in hypotension in 26% of patients. These data support a general threshold value for ascitic fluid removal of 6 L.92
Hepatorenal syndrome (HRS) is traditionally managed with plasma expansion with albumin as well as hemodialysis as a bridge to orthotopic liver transplantation. Albumin may improve aspects of circulatory function such as peripheral vascular resistance, thereby lowering plasma renin activity.93 Several vasoactive agents have also been studied in small trials. In 1 study, a protocol of albumin 20 to 40 g/d, along with octreotide and midodrine adjusted to produce an increase in mean arterial pressure of 15 mm Hg (with a starting dose of octreotide of 100 μg tid up to 200 μg tid and a starting dose of midodrine of 7.5 mg tid up to 12.5 mg tid), safely improved renal function.94 Other treatments include the addition of splanchnic doses of dopamine or norepinephrine (0.5-3 mg/h IV)95 to albumin.
Terlipressin, a vasopressin analogue, is not currently available in the US but is frequently studied in the treatment of HRS. A recent Cochrane review of terlipressin in HRS (irrespective of the use of additional agents such as albumin) concluded that terlipressin reduced mortality and improved renal function, although available data are still not sufficient to formulate treatment recommendations.96 A more recent small, randomized, controlled trial demonstrated that terlipressin significantly improved renal function, but the results showed no survival benefit at 3 months.97 Two open-label, randomized, controlled trials that compared norepinephrine (0.1- 3.0 mg/h IV) to terlipressin (0.5-2 mg) reported similar outcomes,98,99 suggesting that norepinephrine may be a viable alternative in US patients. Finally, a recent small, nonrandomized trial has suggested that paracentesis in volume-resuscitated patients with HRS improves renal function, although the effect is transient.100 TIPS placement may be indicated in refractory cases.
Spontaneous bacterial peritonitis (SBP), an infection of ascitic fluid without a clear source such as perforation or abscess (secondary peritonitis), is diagnosed via an ascitic fluid polymorphonuclear cell count of 250 cells/mm3 or higher or a positive fluid culture. Typical responsible organisms are Klebsiella pneumoniae, Escherichia coli, and Streptoccocus pneumoniae. Detection of SBP is important even in the absence of overt symptoms, since it is associated with increased mortality. While recent analyses of asymptomatic patients undergoing outpatient routine paracentesis have shown a low rate of occult SBP (0-3%),101,102analysis of serial samples from inpatients (who may be more similar to a symptomatic ED population) demonstrated a much higher rate (21%).103 It is therefore recommended that ascitic fluid from all paracenteses be sent for cell count and differential counts. Culture bottles should be inoculated at the bedside for highest yield.104 If secondary peritonitis is suspected, ascites fluid may also be analyzed to determine levels of glucose, lactate dehydrogenase (LDH), carcinoembryonic antigen (CEA), and alkaline phosphatase (AP); fluid levels consistent with secondary peritonitis are glucose less than 50 mg/dL, LDH greater than the upper limit of normal for serum levels, CEA greater than 5 ng/mL, and AP greater than 240 U/L, or total protein exceeding 1 g/L.105,106Bedside testing for SBP using a leukocyte esterase reagent strip is becoming more common, especially in more austere settings. However, this method has not yet been standardized. A recent systematic review of 17 prospective trials showed wide variability in the characteristics of test strips - ie, sensitivity 45% to 100%, specificity 81% to 100%, positive predictive value 42% to 100%, and negative predictive value 87% to 100%.107 The high negative predictive values reported in most analyses indicate the possible role of such testing as a means of ruling out SBP, but the method is still under investigation.
When the neutrophil count in ascitic fluid is 250 cell/mm3 or higher, the patient should be treated empirically with antibiotics. Some patients have infection in the absence of neutrocytic fluid - a condition termed bacterascites. Those patients who progress to SBP will have signs of infection at the time of paracentesis108; outcomes are similar for both types of SBP patients. Therefore, patients with cirrhosis and ascites with clinical SBP (fever, leukocytosis, abdominal pain) but less than 250 neutrophils/mm3 should be treated until culture results are known. Antibiotic regimens include cefotaxime (2 g IV q8h), ampicillin (2 g IV q4h) combined with tobramycin (1.75 mg/kg PO q8h), or oral ofloxacin (400 mg PO bid) for less severe illness in patients who can take medication orally. Although 1 older study suggested that cefotaxime was superior to ampicillin/tobramycin,109 a 2001 Cochrane Review concluded that there was insufficient evidence to support a particular antibiotic regimen based on 9 available randomized, controlled trials comparing antibiotic regimens (no placebo trials) for SBP.110 Cefotaxime with albumin (1.5 g/kg IV on day 1 and 1.0 g/kg on day 3 of treatment) was shown in 1 study to be superior to cefotaxime alone,111 although this finding has not been replicated.
Rupture of gastroesophageal varices is the most common fatal complication of cirrhosis. For the ED clinician, the diagnosis of likely variceal bleeding is made in the context of upper gastrointestinal (GI) bleeding in the patient with known or clinically apparent cirrhosis. In the case of gross hematemesis, the diagnosis of upper GI bleeding is clear. A report of coffee-ground emesis or melena is also likely to indicate an upper GI source.
For the diagnosis and assessment of bleeding severity, a nasogastric tube (NGT) may be indicated. When deciding whether to place an NGT, the ED clinician must balance the value of the information yielded against the substantial discomfort caused by the procedure, as well as the risk of damage to surrounding structures should the tube be improperly positioned. There is no literature to support the theoretical risk that NGT placement will further disrupt varices, so the procedure is generally considered to be safe in these circumstances.
In the patient with hematemesis, NGT placement does not yield additional information about the location of the bleed, but it may reveal the rate of ongoing bleeding. In addition, the NGT serves to clear gastric contents prior to endoscopy; however, since gastric aspiration may be performed at the time of endoscopy, and if vomiting does not threaten airway patency, such advance clearance may not be necessary. One retrospective study found that an uncleared fundal pool of blood at endoscopy was associated with increased rebleeding, number of units transfused, need for emergent surgery, and death112; however, these patients had residual fundal blood despite lavage and positioning, not because they did not have an NGT placed prior to endoscopy.
Intravenous erythromycin has been shown in 2 randomized, controlled trials to improve gastric emptying prior to esophagogastroduodenoscopy (EGD),113,114 thus providing another means for achieving this goal. Other variables such as a drop in hemoglobin levels, tachycardia, and a reduction in blood pressure may serve as alternate indicators of the rate of blood loss. Even though results of randomized, controlled trials to address this issue are lacking, NGT placement is probably most useful in patients with hematemesis who are hemodynamically unstable, who have recurrent vomiting, and for whom endoscopy is likely to be delayed. In patients without hematemesis, the NGT serves to localize the site of bleeding proximal to the pylorus, thus supporting the need for EGD. In a patient with known varices and clinical evidence of upper GI bleeding without hematochezia (melena, an elevated blood urea nitrogen:creatinine ratio), the need for endoscopy to evaluate a variceal source based on history alone may obviate a diagnostic NGT. A final indication for NGT placement in cases of bright-red or coffee-ground emesis and possible ongoing bleeding would be to protect against the aspiration of gastric contents should a decline in mental status require intubation, but this issue has not been subjected to rigorous study.
For patients with variceal hemorrhage, aggressive primary resuscitation measures such as airway protection and large-bore IV access should be initiated immediately. Although crystalloids should be administered as needed for hypotension, blood transfusion should be arranged early to prevent significant hemodilution; replacement of the total volume of blood lost is not necessary and may be detrimental. In a Cochrane review that incorporated 1 multicenter randomized, controlled trial, recombinant factor VIIa was not shown to be of benefit.115,116 Nonetheless, early consideration should be given to reversing coagulopathy with fresh frozen plasma and platelets based on the INR and platelet count. In addition, transfusion of fresh frozen plasma should be considered independent of the INR for those patients who receive multiple units of factor-poor packed red cells.
Pharmacologic and endoscopic treatments are available for patients with bleeding varices. Medical therapy is based on acutely lowering the portal pressure by vasoconstriction of splanchnic arterial inflow and inhibition of local vasodilatory peptides, as well as systemic venodilation. Available regimens in the US include the somatostatin analogue octreotide and vasopressin accompanied by intravenous nitroglycerin to counteract any ischemic effects; it should be noted that this vasopressin regimen can be used for only 24 hours. (See Table 11) A Cochrane meta-analysis of the effects of somatostatin analogues demonstrated no survival benefit, but failure of initial hemostasis was reduced and the amount of blood transfused was slightly decreased.117
In a Cochrane meta-analysis of 15 randomized, controlled trials comparing all types of medical therapy with endoscopic sclerotherapy, sclerotherapy was not superior to medical therapy for a variety of outcomes, including mortality after the initial treatment of variceal hemorrhage.118 Although this analysis supports initial medical therapy, the AASLD recommends that EGD be performed within 12 hours to definitively locate the source of hemorrhage.8 EGD can help identify gastric-only varices, which differ from esophageal varices in terms of their endoscopic treatment. If endoscopy is to be performed, medical therapy should be continued, since it has been shown on meta-analysis to improve initial and 5-day hemostasis, although no effect on mortality has been demonstrated.119
Rescue techniques for uncontrolled or early, recurrent bleeding include TIPS or balloon tamponade. Balloon tamponade is a temporizing measure until a more definitive procedure such as TIPS can be performed. If balloon tamponade is performed, the patient should be intubated.
Severe bacterial infections are common in cirrhosis with variceal bleeding. According to 1 meta-analysis, prophylactic antibiotic administration decreases the risk of infection and increases survival,120 so such prophylaxis should be initiated in the ED. (See Table 12.) Although the optimal antibiotic regimen remains to be determined, the AASLD recommends oral norfloxacin or intravenous quinolones for the majority of patients,7 with ceftriaxone recommended for patients with more advanced cirrhosis (ie, Child- Turcotte-Pugh, Class B or C).121
Airway management in the liver failure patient requires special attention to the cerebral edema and elevated intracranial pressure (ICP) associated with hepatic encephalopathy. Care should be taken to avoid further increases in ICP from airway stimulation. The traditional methods for blunting increased ICP utilized in head trauma patients may be employed here. Historically, ketamine was considered to be contraindicated in patients at risk for intracranial hypertension (and therefore for patients with liver failure) because of concerns about worsening ICP. However, small, nonrandomized studies have indicated that ketamine does not increase ICP in sedated patients with traumatic brain injury, brain tumor, or aneurysm — settings in which intracranial hypertension is likely.122-124 In fact, these studies showed that ketamine lowered ICP while cerebral blood flow was not significantly affected. These studies do not address nonsedated patients (ie, those undergoing rapid-sequence intubation in which ketamine is the primary induction agent); however, these early results are promising. If larger studies demonstrate the same effect, ketamine may become the agent of choice for these patients, since the majority will be hypotensive and thus require an agent that can provide hemodynamic support.
None of the other medications typically employed for intubation are contraindicated in liver failure. However, results of a small study showing propofol to be useful for controlling elevated ICP in liver failure in the ICU setting suggest that this agent may be considered for induction of anesthesia, particularly since hepatic failure does not affect its metabolism.125
Patients with liver failure and variceal hemorrhage are at increased risk for aspiration during intubation due to the cathartic effect of blood in the stomach as well as the ongoing volume entering the stomach during active hemorrhage. This risk may be reduced by maintaining the patient in an upright position until just prior to intubation and by gastric evacuation via NGT.
Pregnant patients are more likely to develop serious sequelae of hepatitis E infection, with up to 20% mortality.126 Pregnant patients are at increased risk of intrahepatic cholestasis, particularly in the third trimester. These patients present with pruritus and jaundice without other evidence of hepatic failure. They are at increased for complications such as preterm delivery and intrauterine demise.127 Acute fatty liver of pregnancy (AFLP) and the related HELLP (hemolysis, elevated liver enzymes, low platelets) syndrome are life-threatening complications of third-trimester pregnancy manifest as acute liver failure, for which the treatment is rapid delivery. Although delivery will improve outcome, postpartum transplantation may rarely be indicated.16,128
As noted in this article, many interventions for ALF are available but are still the subject of some controversy because evidence to support them is lacking. Two areas of ongoing study and controversy have been selected for further discussion below.
The use of NAC in acetaminophen toxicity is wellsupported. In addition, because of its effectiveness in scavenging oxygen free radicals and improving glutathione stores, NAC has been suggested as a "liver tonic" for patients with liver damage. Some studies have suggested that NAC provides circulatory support by increasing oxygen delivery and consumption and improving cardiac index and mean arterial pressure in ALF due to acetaminophen (ie, those for whom an earlier NAC treatment window had passed) or ALF due to other causes.129 NAC has also been shown to increase the cardiac index and hepatic blood flow in patients with septic shock without liver failure.130 However, another study using different measurements of oxygen delivery did not demonstrate improvements in hemodynamics or in oxygen delivery or consumption with NAC.131
A systematic review of published articles on the use of NAC in ALF not related to acetaminophen concluded that the data were not sufficient to recommend its routine employment in this setting.132 Since the efficacy of NAC in nonacetaminophen-related ALF is still under investigation, it may eventually be borne out by the results of future studies.
Many review articles over the years have mentioned bioartificial liver support devices as an "up-andcoming" potential bridge to transplantation. However, data demonstrating a survival benefit with these devices are scarce, and the methodologies involved are still under investigation. A Cochrane review of bioartificial and artificial devices has concluded that they may offer a survival benefit in acute-on-chronic liver failure but not in ALF; further investigation was advised before such approaches can be routinely implemented.133 Studies to evaluate these support systems are ongoing.
Any patient with ALF should be admitted to the hospital. As noted earlier, the AASLD recommends admission to the ICU for all patients with ALF based on their risk of rapid deterioration. Early contact with a transplant center should also be established (in cases of grade 1 or 2 encephalopathy) to enable triage for transplantation as well as early transfer before the patient requires airway protection or becomes otherwise unstable. Patients who present with variceal bleeding should also be admitted, with their level of care determined by their overall clinical picture. Patients with symptomatic ascites may be discharged with close follow-up after peritoneal fluid analysis rules out spontaneous bacterial peritonitis, provided that their symptoms improve after paracentesis and they have no signs of encephalopathy. Collateral information regarding the patient's baseline mental status is critical to detecting subtle signs of encephalopathy.
ALF is a life-threatening disorder that requires immediate recognition by the ED clinician so that supportive measures, treatment of reversible causes, and possible transfer to a transplant facility can be initiated. ALF patients have the potential to decompensate quickly, which will require intubation, management of intracranial hypertension, and treatments for coagulopathy, hypoglycemia, and seizures. Acute complications in patients known to have cirrhosis are frequently seen in the ED; patients with respiratory symptoms or other discomfort from tense ascites should undergo paracentesis and surveillance for spontaneous bacterial peritonitis. Variceal bleeding is a lethal complication of CLF and should be treated aggressively with medical and endoscopic therapy as well as antibiotics. Swift intervention in these ill patients can stabilize them long enough to allow recovery or evaluation for transplantation.
Your 40-year-old male patient on isoniazid has developed ALF from an idiosyncratic reaction to the drug — a diagnosis confirmed by his INR of 2.0 and elevated AST, ALT, and bilirubin. He has grade 1 encephalopathy. After moving him to a high-visibility bed in the ED, starting IV fluids via peripheral IV, and evaluating his glucose and electrolyte levels, you contact the transplant center located an hour away and arrange for an advanced life support ambulance to transport the patient.
Your CLF patient has a massive GI bleed, most likely due to varices. After establishing large-bore IV access, administering an initial bolus of crystalloid, and calling for blood, you start treatment with octreotide and ceftriaxone. You arrange for urgent endoscopy and admit the patient to the ICU with an improved blood pressure.
Evidence-based medicine requires a critical appraisal of the literature based upon study methodology and number of subjects. Not all references are equally robust. The findings of a large, prospective, randomized, and blinded trial should carry more weight than a case report.
To help the reader judge the strength of each reference, pertinent information about the study, such as the type of study and the number of patients in the study, will be included in bold type following the reference, where available. In addition, the most informative references cited in this paper, as determined by the authors, will be noted by an asterisk (*) next to the number of the reference.
Caitlin Bailey; H. Gene Hern, Jr.
April 2, 2010