Critical Care Management of Patients With Liver Disease

Introduction

Critical illness in patients with liver disease is common and associated with a high mortality rate. Olson et al. noted that 26,000 patients with cirrhosis receive mechanical ventilation and hemodynamic monitoring each year and have a greater than 50% in-hospital mortality. Unfortunately, ICU mortality for these patients has not changed in more than two decades. Other studies report similarly disappointing outcomes with mortalities as high as 100% in the setting of triple organ system failure.

The many chapters in this volume about organ system dysfunction in patients with acute and chronic liver diseases outline current understanding of pathophysiology and available therapeutic options. This chapter is structured to amalgamate current critical care practice with these discussions for five common critical situations seen in patients with acute decompensation of chronic liver disease and acute-on-chronic liver failure, severe sepsis/septic shock, delirium and hepatic encephalopathy, and upper gastrointestinal (UGI) bleeding. Our goal is to offer suggestions for management. Recommendations are limited by a paucity of clinical trials for critically ill patients with liver diseases, thus they reflect current, and in some cases consensus or expert opinion.

Sepsis/Septic Shock

Infection is a common precipitant of hepatic deterioration and multiple organ system failure in cirrhosis and is frequently cited as having the greatest attributable impact on mortality of any comorbidity. In vitro and in vivo studies outline a pathophysiologic model where cirrhotic patients have functional immunologic defects that make them vulnerable to infection and prone to an over-exuberant proinflammatory systemic response. Together these processes lead to a self-perpetuating cycle of worsening of liver function and inflammation that promotes organ system failures. Complicating matters are difficulties diagnosing sepsis in liver disease due to similarities between cirrhosis and acute infection (low blood pressure/vasodilated state). Systematic inflammatory response syndrome (SIRS) criteria such as leukopenia, hypothermia, and tachypnea are often seen in, and are less predictive of sepsis in cirrhotic patients. Biomarkers such as procalcitonin and C-reactive protein have been variably reported to be useful in predicting bacterial infection in patients with liver disease.

Timely goal directed quantitative resuscitation is the state of the practice for severe sepsis and septic shock in all patients. Although recent data challenge fundamental elements of the Surviving Sepsis Campaign (SSC) and a small observational trial questions their utility in patients with cirrhosis, prompt resuscitation to restore oxygen delivery remains the core element of resuscitation of any patient.

As discussed elsewhere in this book cirrhosis is associated with an array of circulatory and neurohumoral abnormalities including increased levels of vasodilators and impaired responsiveness to endogenous and exogenous vasoconstrictors. This physiology introduces challenges in resuscitation of the hemodynamically compromised cirrhotic patient. Current opinion regarding the vascular pathophysiology of cirrhosis is that endothelial-derived vasoactive mediators increase splanchnic capacitance. The resultant arterial pooling of blood and increased blood volume in the splanchnic circulation equates to a reduction in central blood volume and preload and less than expected cardiac output. Reduced cardiac output due to underfilling and possibly cirrhosis-associated cardiomyopathy, lead to a neurohumoral response that causes renal sodium and water retention (reviewed in Chapter 18 ).

Albumin is a mainstay for resuscitation of patients with cirrhosis. Much of the rationale for its use is derived from its beneficial effect in spontaneous bacterial peritonitis (SBP) and for preservation of renal function after large volume paracentesis. Subset analysis of the Saline vs. Albumin Fluid Evaluation (SAFE) study indicated improved outcomes of noncirrhotic septic patients treated with albumin. Primary outcome data to support use of this colloid to resuscitate during sepsis are lacking. Furthermore, the physiologic rationale for its use to improve intravascular colloid oncotic pressure is not supported in the literature regarding noncirrhotic patients. Umgelter et al. infused 400 mL of hyperoncotic albumin (20%) into cirrhotic patients with renal dysfunction and observed increased central blood volume (global end-diastolic blood volume) and cardiac index. A notable finding from this study (and others) was the failure of central venous pressure (CVP) to correlate with changes in cardiac output and an elevated systematic vascular resistance (SVR) as a predictor of response to hyperoncotic albumin. Other studies have failed to describe changes in blood pressure following albumin administration, but do note reduced renin levels, suggesting subclinical increases in effective intravascular volume. These studies suggest that abnormal responses to volume expansion of cirrhotic patients lessen the utility of traditional measures of resuscitation such as blood pressure and heart rate.

In the absence of clinical trials, expert opinion recommends goal directed resuscitation of cirrhotic patients using guidelines of the SSC. This includes a target mean arterial pressure of 65 mm Hg, a central venous oxygen saturation (S _cv O ₂ ) greater than or equal to 70%, and a urine output greater than 0.5 mL/kg per hour using fluids, vasopressors, inotropes, and, in some cases, blood. Some caution in adherence to these guidelines is required, including the need to appreciate that cirrhotic patients often have mean arterial pressure (MAP) lower than 65 mmHg, high S _cv O ₂ , elevated lactate concentrations due to abnormal lactate metabolism, and lower urine output than noncirrhotic patients. Furthermore, resuscitation guided by CVPs may be hazardous due to parallel increases in central venous and portal pressures. A prudent approach would be to initiate resuscitation using the SSC guidelines with frequent reevaluation. Clinical improvement without achieving the quantitative resuscitation targets of the SSC should prompt a shift in approach rather than continued resuscitation. In liver disease CVP is not a useful measure of preload, nor has the use of a pulmonary catheter been shown to impact outcome. Volumetric assessments of response to fluids seem a better choice at this time. Use of transpulmonary thermodilution such as the pulse contour cardiac output monitoring (PiCCO) system to assess end-diastolic volume may have some utility. Bedside critical care ultrasound is a readily available, easy to perform measure of central blood volume and/or prediction of volume responsiveness that has gained widespread acceptance despite lack of prospective trials showing improvements in outcome. Whether critical care ultrasonography (US) is useful for guiding resuscitation of patients with cirrhosis is unknown.

Respiratory failure caused by increased metabolic demands (Type 4) may develop in patients with cirrhosis with severe sepsis or septic shock due to the high workload of breathing from acidosis and reduced respiratory system compliance caused by a distended abdomen and elevated diaphragms from ascites. Intubation and mechanical ventilation may be required to reduce oxygen consumption in this setting. Depressed mental status due to sepsis and/or hepatic encephalopathy should prompt early rather than late initiation of mechanical support for respiratory failure. Noninvasive ventilation should be used with caution in patients with cirrhosis, as they may be intolerant due to worsening of abdominal distention caused by positive pressure. Acutely ill patients with liver disease are likely to have altered sensorium, which is also a contraindication for noninvasive ventilation.

Source control is a key aspect of treatment of sepsis. All patients require a thorough survey for evidence and sources of infection. This includes laboratory evidence and blood cultures. All patients with cirrhosis should be evaluated for ascites that, if present, should be sampled when looking for evidence of primary or secondary bacterial peritonitis ( Table 13-1 ). Paracentesis can be performed irrespective of the international normalized ratio (INR) and platelet count provided there is no clinical evidence of fibrinolysis or diffuse intravascular coagulation. Empiric broad-spectrum antibiotics should be selected using a protocol that is sensitive to the local microbiogram. Review of data from an international sepsis database that included 635 patients with cirrhosis and sepsis revealed that use of a single antibiotic, as opposed to two or more appropriate antimicrobials was associated with increased risk of death. Similarly, selection of incorrect antibiotics is associated with a 9.5-fold increase in mortality. Delay of antibiotic administration has been linked to increased mortality in cirrhotic patients, each hour delay increased the adjusted odds ratio for death by 1 : 1. Inferential data supports the notion that early administration of antibiotics translates to improved outcomes from septic shock.

The role of steroids in septic shock is incompletely defined in patients without liver disease. Biochemical evidence for adrenal insufficiency has been reported in up to 83% of critically ill septic patients with cirrhosis. The diagnosis of adrenal insufficiency in all forms of critical illness, including liver disease, is limited by the absence of a gold-standard for adrenal insufficiency. Fernandez et al. reported that hydrocortisone improves survival in critically ill patients with cirrhosis with biochemical evidence of adrenal insufficiency. This small (25 patients), prospective, nonrandomized single center trial that used historical controls should be considered in the context of more recent data showing lack of outcome benefit from steroid supplementation in refractory sepsis. Nevertheless, hydrocortisone (50 mg IV every 8 hours) remains an option for patients with shock who fail to improve with fluids and reasonable doses of norepinephrine.

Altered Mental Status

Acutely altered mental status (delirium) is a catchall for impaired cognition, reduced attention, reduced awareness and/or altered level of consciousness. Regardless of the etiology, delirium conveys an adverse prognosis such that patients with structural causes for delirium have an odds ratio for death of 43 as compared with patients with cirrhosis without delirium. In patients with cirrhosis, altered mental status is often routinely attributed to hepatic encephalopathy (HE); however, other causes, especially infection, must be considered. Table 13-1 lists the findings from a retrospective review of 1218 patients with cirrhosis and delirium. Whereas HE was the most common cause of delirium, it accounted for less than half of cases of delirium in this population.

Brain computed tomography (CT) is commonly ordered in search of coagulopathy-related, spontaneous intracranial hemorrhage (ICH) in patients with liver disease. Donovan et al. have shown that patients with cirrhosis without trauma and with a nonfocal neurologic exam are very unlikely to have ICH, regardless of platelet count or INR. The number needed to scan to find a single ICH in the absence of a focal neurologic exam in this study was 293, for those with a focal neurologic exam or trauma it was 9 and 20, respectively. This is consistent with data from studies of other critically ill medical ICU patients. Stated otherwise, head CT is of limited utility in patients without an exam or history that predicts intracranial pathology.

Although rare, meningitis should be considered in febrile patients with cirrhosis with otherwise unexplained altered mental status. The incidence of meningitis may be increased as much as 10-fold in patients with cirrhosis compared to the normal population, it nevertheless accounts for less than 1% of bacterial infections in patients with cirrhosis and remains uncommon. In cirrhotic patients, the clinical presentation of meningitis, as compared to noncirrhotic patients, is often atypical; meningismus may be delayed (up to 24 hours), symptoms prior to diagnosis are often prolonged (4 days or more), there is a greater incidence of relapse, and the mortality rate is very high. The bacterial milieu in patients with cirrhosis differs from normal adults and includes Gram-negative enteric organisms (E. coli) and Listeria . Confirmation of meningitis requires lumbar puncture (LP). Unlike liver biopsy, paracentesis, and central lines there is no consensus about safe coagulation parameters required to perform a lumbar puncture. The clinical practice guidelines of the Vascular and Interventional Radiology Society of Europe provide the following reasonable empiric recommendations for procedures deemed moderate risk for bleeding; platelet count greater than 50,000/µL, INR less than 1.5, withholding anticoagulation for durations that respect each agents functional half-life, and carefully considering the need for LP in the presence of active bleeding or hyperfibrinolysis. Thus, the clinician must balance the risks and benefits of empiric antibiotics with confirmation of an uncommon diagnosis in patients with coagulopathy.

Upper GI Bleeding

There are multiple causes of UGI bleeding in patients with cirrhosis. However, data from small studies suggests that patients with gastroesophageal varices almost always bleed from their varices. In an older, small study of 40 patients with Child-Pugh criteria for cirrhosis who presented with UGI bleeding, 11 had varices, and 17 had varices plus another lesion (Mallory-Weiss tears, gastritis, esophagitis). Of these 17 patients, only half bled from varices. The others had Mallory Weiss tears, gastritis and gastric ulcers. Child-Pugh scores in these patients did not predict the risk of variceal bleeding.

The need for emergent endoscopic evaluation and treatment of massive UGI bleeding frequently raises the question of prophylactic intubation for airway protection and prevention of cardiovascular complications. Intubation of patients with severe liver disease introduces a variety of ICU-related issues, including use of hepatically metabolized sedatives that can lead to prolonged mechanical ventilation resulting in a cycle of events that may promulgate or worsen hepatic decompensation or potentially trigger acute-on-chronic liver disease. Several retrospective studies note that elective intubation does not reduce postprocedure pulmonary infiltrates or mortality. The results of a recent meta-analysis of four controlled studies reported a significant increase in pneumonias at 48 hours postprocedure. A retrospective study of prophylactic intubation of patients with variceal bleeding who had no greater than Grade II encephalopathy and no preprocedure pulmonary infiltrates on chest X-ray reported that intubated patients had significantly more infiltrates (17% vs. 0%, p = 0.01) and a trend toward higher morality (21% vs. 5%, p = NS). In the absence of prospective studies to define indications for intubation during endoscopy for suspicion of varices, it seems reasonable to recommend that, in patients with no respiratory issues (no infiltrates, chronic lung disease or baseline hypoxemia) and who do not have severe hepatic encephalopathy (defined as Grade I or II), prophylactic intubation is not necessary, provided the patient is responding to resuscitation. If intubation is necessary we recommend avoiding potent, hepatically metabolized anesthetics such as propofol.

General Critical Care Issues