CBF tends to vary directly with blood pressure (15). Thus, although excessive CBF contributes to worsening cerebral edema, patients are also especially vulnerable to ischemia if hypotension occurs. Furthermore, despite global increases in CBF, there may be important regional variations, such that blood flow is excessive in some areas and insufficient in others (16). Thus, it is important that CPP be carefully controlled, and extremes avoided.
Prognostic schemes have been developed to predict death without OLT in patients with ALF. The King’s College Criteria (17), retrospectively analyzed outcomes in patients with ALF according to cause and subsequently validated the model in a test population. For patients with APAP-induced ALF, predictors of death included acidosis on admission (arterial pH < 7.30), or azotemia, severe coagulopathy, and high-grade HE (grade 3 or 4). In patients with ALF due to other causes, severe coagulopathy or any three of the criteria listed in Table 127.2 also predicted death. Although the original series using these criteria reported a predictive accuracy for death without OLT of more than 85%, subsequent analyses have suggested that they are less accurate (18). Consequently, other prognostic parameters continue to be applied to individual patients with ALF during the weighty decision of whether to proceed with OLT.
N-acetylcysteine (NAC) has become widely administered to patients with ALF due to APAP on the basis of both laboratory and clinical data (29,30); however, randomized, placebo-controlled studies documenting the efficacy of NAC in APAP overdose have never been performed. Several rules of administration require emphasis. First, although nomograms describing the probability of hepatotoxicity after a single APAP ingestion have been used widely to determine whether NAC should be administered, the time of ingestion frequently cannot be determined accurately, and ingestions are often multiple. Therefore, NAC should be administered whenever there is uncertainty about timing or dose. Second, intravenous NAC should be administered when a patient has higher than grade 1 HE, or in patients who do not tolerate oral dosing (31). Finally, since the administration of NAC, even late after ingestion, appears to confer survival benefit, dosing should continue until evidence of severe liver injury resolves (international normalized ratio [INR] less than 1.5 and resolution of HE) (32).
|TABLE 127.2 Schemes for Predicting Poor Prognosis and the Need for Orthotopic Liver Transplantation in Patients with Acute Liver Failure (ALF)|
Other Etiology-Specific Treatments
Specific medications may be considered in patients with ALF due to non-APAP etiologies, and are outlined in Table 127.3. It should be emphasized that studies to support the use of these therapies do not exist; one using lamivudine for acute hepatitis B found no benefit versus placebo (33). Another randomized, placebo-controlled trial suggested improved transplant-free survival of patients with non-APAP ALF and low-grade (stage I-II) HE who received NAC (34).
|TABLE 127.3 Potential Etiology-Specific Therapy of Patients with Acute Liver Failure|
Management of HE, Cerebral Edema, and Seizures
Considering the central importance of ammonia in the pathogenesis of HE and cerebral edema, clinicians often use therapies aimed at lowering serum ammonia levels in patients with ALF. The most commonly used agents in chronic liver disease, nonabsorbable disaccharides (e.g., lactulose) and oral antibiotics (neomycin, rifaximin, or metronidazole), have not been studied in patients with ALF but probably have little effect on outcome. A computed tomographic (CT) scan of the head should be considered early in the course of ALF to exclude other causes of altered mental status, especially intracerebral hemorrhage. Although a head CT may also detect cerebral edema, a normal scan does not rule out clinically important intracranial hypertension (Fig. 127.2) (42). The specific management of cerebral edema in patients with ALF resembles that of other causes of intracranial hypertension, with the important caveat that hyperemia plays an important additional pathogenic role in the former (Fig. 127.3). Chest physiotherapy and suctioning should be temporarily minimized. The head of the bed should be elevated to at least 30 degrees, as this reduces ICP and decreases the risk of hospital-acquired pneumonia (43). The duration of time that a patient is placed supine or in Trendelenburg position for procedures should be minimized, especially if ICP is not monitored. Endotracheal intubation and mechanical ventilation must be implemented in a timely fashion to avoid the potentially injurious effects of hypoxemia and hypercapnia, while also reducing aspiration risk and facilitating management of intracranial hypertension. Laryngoscopy and intubation may cause transient elevations in ICP and fluctuations in blood pressure. Appropriate measures should be taken to minimize these physiologic derangements, including the appropriate use of sedation and neuromuscular blockade.
Adequate analgesia and sedation must be administered to manage intracranial hypertension in ALF. In general, shorter-acting agents are preferred, such that patients can be more quickly awakened and re-examined. Increased levels of endogenous benzodiazepine like molecules and GABAergic neurotransmission have been implicated in the pathogenesis of HE (44). Therefore, sedatives with GABAergic properties may exacerbate HE; propofol is preferred over benzodiazepines by some authors (45).
Hyperventilation causes cerebral vasoconstriction and a reduction in cerebral blood volume, effects that can be used therapeutically to reduce ICP (46). Patients with HE often spontaneously hyperventilate, with resultant respiratory alkalosis. If the PaCO2 suddenly normalizes because of sedation or respiratory exhaustion, rebound vasodilatation may occur, with consequent elevated ICP. Thus, initial ventilator settings should probably be set to match the previous, spontaneous minute ventilation of the patient, and either arterial blood gases or end-tidal CO2 should be closely monitored. The major concern with using hyperventilation as a means to lower ICP is that vasoconstriction may be severe enough to cause cerebral ischemia.
Osmotic agents, including mannitol and hypertonic saline (HTS), lower ICP most effectively in the setting of global—rather than unilateral—cerebral edema with an intact blood–brain barrier, which characterizes ALF (47). In small human and animal studies, mannitol effectively lowered ICP and appeared to improve survival (48,49). A general indication for the administration of mannitol includes persistently (more than 10 minutes) elevated ICP (more than 20 mmHg if monitor present; see below in “Controversies”) after ensuring proper calibration of the ICP monitor. The optimal dose of mannitol remains untested, but smaller boluses (e.g., 0.25–0.5 g/kg every 4 to 6 hours) appear to be as effective as larger ones. Apart from potentially causing hypovolemia, the accumulation of mannitol may be nephrotoxic, an important consideration given that many patients with ALF have abnormal renal function. Consider maintaining serum osmolality below 320 mOsm/L, but this threshold is arbitrary, is frequently exceeded without adverse effect in other neurocritical care patients, and does not predict an increased risk of renal failure.
HTS is increasingly being used in various settings as an alternative to mannitol to treat cerebral edema (47). Theoretical advantages over mannitol include the following: (a) the blood–brain barrier is less permeable to HTS, making it a more effective osmotic agent; (b) it is a volume expander rather than a diuretic; and (c) there is no proven nephrotoxicity. HTS can be administered as boluses of 3% to 30% saline every 3 to 4 hours, or as a continuous infusion. In one randomized, controlled trial of patients with ALF and high-grade encephalopathy, an HTS infusion (30% saline, 5 to 20 mL/hr) to maintain serum sodium levels of 145 to 155 mmol/L effectively acted as prophylaxis against intracranial hypertension. The presence of hyponatremia may contribute to the development of cerebral edema early in the course of ALF (50) and is an independent predictor of worse outcome (51). Hyponatremia should therefore be corrected, but serum sodium levels should not be raised more quickly than 0.3 to 0.5 mmol/L/hr to minimize the risk of osmotic demyelination (Central Pontine Myelinolysis).
Nonconvulsive seizures are a potential cause of worsening cerebral edema and secondary brain injury. A small study using two-channel EEG—rather than the usual 16 to 20 channels—suggested that nonconvulsive seizures are relatively common with ALF, although the criteria used for electrographic seizures were not described (52). Given the frequency of nonconvulsive seizures and status epilepticus among comatose patients in general, routinely obtaining an EEG in patients with hepatic encephalopathy, particularly when there is a clinical change in neurologic status, may be useful. However, there is currently insufficient evidence to justify the routine use of prophylactic phenytoin (52,53). Furthermore, sedation with propofol or benzodiazepines is likely to provide more effective antiseizure prophylaxis than phenytoin.
Management of Cardiopulmonary Complications
Patients with ALF frequently develop SIRS, regardless of whether or not their course is complicated by infection (54). Vasoplegia is common in ALF and can be complicated by sedatives, mechanical ventilation, and relative adrenal insufficiency (55). Norepinephrine is currently the preferred vasopressor for vasodilatory shock (56). The goal MAP should be individualized to optimize organ perfusion, rather than choosing an arbitrary number, but is generally kept above 65 mmHg. CPP should be maintained above 50 to 60 mmHg, since CBF autoregulation fails below these levels in most individuals; if an ICP monitor is not placed, clinicians should err on the side of higher blood pressure. Although there are theoretical concerns that vasopressin may exacerbate ICH, it may be used as a second-line vasopressor (57). In patients with relative adrenal insufficiency, treatment with low-dose corticosteroids (e.g., hydrocortisone 50 to 100 mg IV every 6 to 8 hours, or 8 to 10 mg/hr as a continuous IV infusion) reduces vasopressor requirements, although the impact on outcome is uncertain.
As many as 37% of patients with ALF develop acute lung injury (ALI) or acute respiratory distress syndrome (ARDS) (58). With established ALI or ARDS, tidal volumes should be limited to 6 mL/kg of predicted body weight, although increases in PaCO2 cannot be tolerated in the setting of cerebral edema and intracranial hypertension. Given the high risk of developing ARDS, it may be advisable to limit tidal volumes even in the absence of established ALI/ARDS. Positive end-expiratory pressure (PEEP/CPAP) settings should be sufficient to achieve adequate oxygenation, while concomitantly ensuring that ICP, blood pressure, and cardiac output are not compromised.
Management of Acute Kidney Injury
Acute kidney injury (AKI) complicates up to 50% of cases of ALF, and is more prevalent in acetaminophen-induced ALF (59,60). Hypovolemia, hypotension, and the use of nephrotoxins—including aminoglycosides, nonsteroidal anti-inflammatory drugs, and intravenous contrast—should be minimized. Patients with ALF can develop hepatorenal syndrome (HRS) as a consequence of intense renal vasoconstriction (see Acute-on-Chronic Liver Failure, below). The decision to initiate renal replacement therapy (RRT) must consider the magnitude of renal dysfunction, metabolic derangements, and volume overload. Continuous renal replacement therapy (CRRT) is preferred over intermittent hemodialysis (IHD) by many clinicians, largely because of more stable volume management and greater time-averaged dialysis dose. Even transient hypotension is poorly tolerated in patients with cerebral edema; not only does the CPP decrease, but cerebral vasodilatation may increase, with further increases in CBF and ICP (61). Excessively rapid correction of metabolic acidosis with bicarbonate-based dialysate may transiently increase cerebrospinal fluid CO2, reduce central nervous system (CNS) pH, and promote more cerebral vasodilatation (62). Good results have been achieved with CRRT in patients with ALF (63). Regardless of the mode of RRT, an adequate hemofiltration or dialysis dose should be used, while blood pressure—and preferably ICP—are carefully monitored and maintained CRRT. Hemofiltration may also impact serum ammonia levels (64).
Management of Infections
ALF is associated with reticuloendothelial dysfunction and impaired immunity, with reduced complement levels, abnormal opsonization, and ineffective phagocytosis. ALF patients are, therefore, at high risk of nosocomial infections with both bacterial and fungal pathogens, which occur in almost 40% of these patients (65). Early diagnosis can be difficult since patients often have subtle manifestations of infection, but is vital because of the high associated morbidity and mortality. Daily surveillance cultures (urine, blood, sputum) and chest radiography should be considered, as they may improve early diagnosis of infection and guide selection of antimicrobial agents (66). Prophylactic antibiotics (enteral and parenteral) have not shown to consistent impact rates of bacteremia or survival in ALF (67). Nevertheless, many clinicians prefer to use them, especially in patients listed for transplantation. Empiric broad-spectrum antibiotics (including vancomycin and an antifungal agent, as indicated) should be considered in any patient with ALF who develops significant isolates on surveillance cultures, unexplained progression of HE, or signs of SIRS, as these frequently predict sepsis in patients with ALF (14,54).
Management of Coagulopathy
Despite a deficiency of clotting factors, low fibrinogen, thrombocytopenia, and platelet dysfunction, clinically important spontaneous bleeding is relatively infrequent in patients with ALF, being seen in less than 10% of patients (68). Therefore, the routine use of blood products to correct these abnormalities is not justified since they are unnecessary, ineffective, and perhaps most importantly, interfere with the prognostic utility of the INR. Although treatment of coagulopathy may be considered in anticipation of invasive procedures, global hemostasis in patients with ALF is usually normal or even hypercoagulable (69), and transfusion of plasma may exacerbate a prothrombotic state characterized by profoundly high von Willebrand factor and factor VIII, and deficient fibrinolysis (6,70,71). Vitamin K deficiency has been reported to contribute to the coagulopathy of ALF (72) and should be repleted parenterally (10 mg subcutaneously [SC] or slow [over 30 minutes] IV). Coagulopathy and mechanical ventilation are well-established indications for gastrointestinal (GI) stress ulcer prophylaxis, which has been shown to decrease the risk of GI bleeding in ALF patients (73). Deep venous thrombosis prophylaxis is recommended; although there are no published data, low-dose unfractionated heparin or low–molecular-weight heparin can be safely used (Richard T. Stravitz, personal observations).
Management of Metabolic Derangements
ALF is a catabolic state, with increased energy requirements and negative nitrogen balance, which may in turn contribute to immunosuppression (74). Higher-than-usual caloric intake is therefore recommended, with 35 to 40 kcal/kg/day and 0.8 to 1 g/kg/day of protein, preferably provided via the enteral route. Reduced hepatic glycogen stores and impaired gluconeogenesis are responsible for the frequent development of hypoglycemia, which often requires treatment with intravenous dextrose. Conversely, hyperglycemia may contribute to increases in ICP and other complications. Thus, blood glucose values must be closely monitored and maintained within the normal range with intravenous short-acting insulin.
Liver Transplantation for ALF
OLT remains the treatment of last resort for ALF. The decision to list a patient with ALF for OLT requires careful clinical and psychosocial assessment and should be started immediately on recognition of poor prognosis as discussed above. In addition to usual clinical evaluation, patients with ALF due to APAP overdose often present with histories of suicidal ideation or substance abuse, which may preclude their consideration as viable OLT candidates. Because OLT candidates with ALF are generally younger and healthier than their counterparts with chronic liver disease, the pretransplant evaluation can usually be abbreviated to include echocardiography, duplex ultrasonography of the liver, and routine pretransplant laboratories (e.g., total anti-CMV, HIV antibody).
Criteria for listing a patient with ALF for OLT change and current criteria may be found at UNOS.ORG in Policy 3.6. Presently, patients with ALF are given priority to receive a cadaveric organ over all patients with chronic liver disease (Status 1). Candidates must have a life expectancy without OLT of less than 7 days, have onset of HE within 8 weeks of the first symptoms of liver disease, and no history of pre-existing liver disease. In addition, patients must be in the ICU and must fulfill one of the following three criteria: (i) be ventilator dependent; (ii) require dialysis or CRRT; or (iii) have an INR more than 2.0. Patients with acute decompensated Wilson disease may also be listed as Status 1 in consideration with their extremely poor prognosis for spontaneous survival. Before transporting a patient with ALF to the operating room for OLT, a detailed review of the patient’s neurologic status must be made so that OLT is not performed when likelihood of neurologic recovery is poor. Specifically, it has been observed that severe, sustained intracranial hypertension predicts brainstem herniation during OLT or poor neurologic recovery after OLT, and patients with ICP greater than 40 mmHg or CPP less than 40 mmHg for more than 2 hours appear to be particularly vulnerable to these disastrous outcomes (75).
Early (3-month) mortality after OLT for ALF is higher than for patients transplanted for all causes of chronic liver disease—about 20% versus 10%, respectively, reflecting the acuity and severity of disease at the time of transplant. Thereafter, however, 5- and 10-year survival after OLT for ALF approximates 70% and 65%, respectively (76). Well-selected APAP-ALF patients with careful review of psychosocial factors potentially have outcomes similar to patients with chronic liver disease patients (77).
Intracranial Pressure Monitoring
ICP cannot accurately be determined noninvasively and carries important prognostic implications for spontaneous survival and neurologic recovery after OLT, many experts advocate ICP monitor placement in OLT candidates with stage III or IV HE (78). Although there is a perception that ICP monitoring significantly improves the outcome of patients with ALF, no prospective, randomized studies exist to support the practice, and retrospective studies in fact refute this perception (79). The invariable presence of coagulopathy in patients with ALF increases the bleeding risk of ICP monitor placement, and earlier studies reported bleeding complications in up to 20%. However, the clinical significance of many of these bleeding complications was probably negligible (78), and more recent series have found lower bleeding rates (5%) (79). Although placement of ICP monitors into the epidural space may minimize the risk of hemorrhage, the accuracy of this practice, compared with other methods of ICP monitoring, has not been well studied.
Therapeutic Hypothermia in ALF
Fever increases ICP and is an independent predictor of worse outcome in brain-injured patients, such that hyperthermia should be avoided (80). The induction of mild hypothermia may interfere with several steps in the pathogenesis of cerebral edema. Specifically, hypothermia attenuates the osmotic gradient created by increased astrocytic glutamine, normalizes extracellular glutamate and lactate, decreases CBF, restores autoregulation, and reduces ICP (81). Temperatures of 32° to 33°C have been used to control intracranial hypertension in patients with ALF refractory to standard care. However, a recent controlled study failed to show significant benefit with the potential exception of younger acetaminophen patients (82). Important potential adverse effects of hypothermia in the setting of ALF include interference with coagulation, an increased risk of infection, and cardiac dysrhythmias.
ACUTE-ON-CHRONIC LIVER FAILURE
Patients with cirrhosis, the fibro-inflammatory alteration of hepatic architecture resulting in portal hypertension, usually die from acute hepatic decompensation (AD), which triggers MOSF. The term AD is often invoked in a patient with cirrhosis who develops ascites, hepatic encephalopathy, GI hemorrhage, and/or bacterial infection, without progression to MOSF (83). By contrast, acute-on-chronic liver failure (ACLF) has recently been defined as AD with MOSF (83,84). The natural history and clinical manifestations of ACLF have been recently described in order to study the entity, and a prognostic score for ACLF was later developed and validated, which included entries for the number of organs failed, age, and white blood cell (WBC) count (85,86) (available online at www.clifconsortium.com). A North American consortium subsequently proposed a second definition of ACLF, including grade III/IV hepatic encephalopathy, hypotension despite adequate volume repletion, and/or a requirement for mechanical ventilation or RRT (87). Transplant-free mortality at 28 and 90 days is 33% and 51%, respectively, in patients with ACLF, but only 2% and 10%, respectively, in those with AD (86). The most common extrahepatic manifestation of ACLF is renal failure, which also portends the worst prognosis.
According to the Scientific Registry of Transplant Recipients, the most common cause of end-stage liver disease in the United States is chronic hepatitis C, with or without a contribution from alcohol abuse (seen in about 40% of cases). Patients with alcoholic cirrhosis and indeterminate (cryptogenic) causes, many of whom have nonalcoholic steatohepatitis (NASH), occupy second and third most frequent causes, respectively. Other less common etiologies include chronic hepatitis B, immune-mediated liver diseases—autoimmune hepatitis, primary biliary cirrhosis, primary sclerosing cholangitis—and hereditary liver diseases—hemochromatosis, alpha-1-antitrypsin deficiency.
Patients with cirrhosis decompensate as a result of contributions from two basic pathogenic mechanisms: portal hypertension and hepatocellular insufficiency. Complications of portal hypertension include hemodynamic alterations, functional renal failure, ascites, and GI bleeding, most commonly from variceal hemorrhage. Complications of hepatocellular insufficiency include coagulopathy and hepatic encephalopathy, although it should be appreciated that the latter also occurs as a result of portosystemic shunting.
The pathophysiology of ACLF is the subject of intense ongoing research; however, the topic remains poorly understood largely because the syndrome required formal definition and recognition as a clinical entity. Although the definitions of ACLF noted above were not used in most of the studies exploring pathogenesis, a consensus of experts has summarized the available data (86). ACLF usually occurs after a triggering event in a patient with stable cirrhosis, such as alcoholic hepatitis, superimposed viral hepatitis, drug-induced liver injury, or a complication of cirrhosis itself, most often infection, in turn, often due to bacterial translocation from the gut.
A fundamental feature of cirrhosis critical to the pathogenesis of AD and ACLF is an abnormal hemodynamic state characterized by low systemic vascular resistance (SVR), systemic hypotension, and splanchnic vasodilation. Consequently, arterial underfilling of critical regulatory vascular beds in the renal arterial and hypothalamic circulations result in the elaboration of compensatory neurohumoral effectors, such as renin, angiotensin, vasopressin, and norepinephrine. The primary mechanism underlying low SVR includes release of vasodilatory mediators such as endothelins and nitric oxide by the portal endothelium, which is exacerbated by triggers of ACLF (86). The normal compensatory mechanisms which occur in response to low SVR underlie the formation of ascites, the functional renal failure of cirrhosis—HRS—and hyponatremia. HRS and cirrhotic ascites have the same basic pathogenesis (Fig. 127.4). As outlined above, renal arterial constriction occurs in normal compensation for systemic hypotension. Poor renal perfusion results in sodium retention, plasma volume expansion, and, in the presence of hepatic sinusoidal hypertension, the transudation of lymph across the Glisson capsule as low-protein ascites (88). HRS can be considered an exaggeration of this renal vasoconstriction, often in the setting of cardiac hypocontractility (89).
In addition to the circulatory and neurohumoral features of AD and ACLF discussed above, patients with decompensated cirrhosis have impaired cardiac contractility, particularly after infection or GI bleeding (90). Myocardial failure was formerly ascribed to the ethanol toxicity or iron deposition in myocardium; however, cirrhotic cardiomyopathy, depressed myocardial contractility as a complication of cirrhosis per se, has been recognized as a distinct clinical entity. Diagnostic criteria of cirrhotic cardiomyopathy include blunted inotropic and chronotropic responses to stress, diastolic dysfunction, and prolonged QT interval on electrocardiogram (ECG). A pathogenic role of cirrhotic cardiomyopathy has been documented in patients with HRS, particularly in the setting of infection, and in the circulatory dysfunction after large volume paracentesis without adequate plasma expansion (89,91,92).
AKI in patients with decompensated cirrhosis is an independent predictor of death in the ICU (93), frequently signals the onset of infection (94), and is an integral component of ACLF (86,95). The definition of AKI in cirrhosis is based upon the modified Risk, Injury, Failure, Loss, End-stage renal disease (RIFLE) criteria rather than a fixed serum creatinine (96). The differential diagnosis of AKI in patients with cirrhosis includes prerenal azotemia, HRS, and acute tubular necrosis (ATN). Analysis of urine sediment and sodium differentiate the above possibilities: the former two diagnoses present with normal urine sediment and low (<10 mEq/L) urine sodium, and the latter with renal tubular cell debris and high urine sodium. The distinction of these causes of renal failure remains paramount, since in its late stages, HRS portends a very poor prognosis and is generally irreversible without OLT, in contrast to prerenal azotemia and ATN. In practical terms, the diagnosis of HRS is often made after the exclusion of septic shock, intrinsic renal disease, obstructive uropathy, and most important, prerenal azotemia, the latter after a 1.5-L IV fluid challenge—normal saline with or without colloid (Table 127.4) (97).
Bacterial infections represent the most common trigger for ACLF and remain one of the two primary causes of death (84,95). Risk factors for bacterial infections in hospitalized patients with cirrhosis include ICU admission and GI bleeding (98). Patients with cirrhosis are relatively immunocompromised as a result of portal hypertension and immune dysfunction. Portal hypertension results in the formation of a low-protein ascites, which is susceptible to infection because of its low complement concentration and, thus, low opsonic activity (99). In addition, gut congestion from portal hypertension increases the likelihood of bacterial translocation into blood, which seeds the ascites secondarily, the so-called spontaneous bacterial peritonitis (SBP) (100).
Most studies of bacterial infections in patients with cirrhosis were performed in the 1980s, during which community-acquired, gram-negative infections (urinary tract infections and SBP) predominated. More recent studies, however, have documented an evolution of the epidemiology of infection in patients with cirrhosis. SBP remains the most common bacterial infection in patients admitted to the ICU, but a shift toward gram-positive infections has occurred. In one major hepatic disease ICU, 77% of isolates were gram-positive, which was ascribed to the widespread use of prophylactic fluoroquinolones in cirrhotic patients with low-protein ascites (101), and to the frequent use of invasive procedures, including IV catheter insertion and variceal band ligation (102).
|TABLE 127.4 Diagnostic Criteria of Hepatorenal Syndrome|